article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
stellar - mass black holes ( bhs ) are mostly detected through dramatic x - ray outbursts exhibited by transient x - ray binaries ( sxts ; @xcite ) .
about 60 bh candidates have been identified in the 50 year lifetime of x - ray astronomy @xcite , although only 17 of these have been confirmed by dynamical studies i.e. they possess a mass function @xmath15 in excess of 3 m@xmath14 , where @xmath16 is the mass ratio of the companion star to the compact object .
the reason for the low rate of confirmed bhs lies in the difficulty of detecting the companion star for quiescent optical magnitudes fainter than @xmath1722 .
time - resolved spectroscopy with signal - to - noise ratio @xmath18 is typically needed to detect the weak absorption features and trace their orbital motion .
determining the binary mass ratio is even more challenging as it also requires resolving powers @xmath19 to measure the rotational broadening of the absorption lines ( e.g. @xcite ) .
in addition , moderately short integration times are essential to avoid significant orbital smearing .
it should be mentioned that alternative methods based on the radial velocities of the disc emission lines are uncertain and prone to large systematic effects ( see e.g. @xcite ) .
a critical review on the determination of system parameters in bh sxts can be found in @xcite . in @xcite ( henceforth paper i )
we showed that the full - width - half maximum ( @xmath2 ) of the disc h@xmath3 line in quiescent bh sxts scales with the velocity semi - amplitude of the companion star .
we here present the discovery of a another correlation between @xmath0 and the ratio of the double peak separation to the line width .
both relations open the door to constrain fundamental parameters and perform dynamical studies in much fainter samples of quiescent bh candidates than is currently possible .
we have selected a sample of quiescent spectra of nine dynamical bh and two neutron star ( ns ) sxts from paper i , all with reliable mass ratio determinations .
only mass ratios obtained by resolving the rotational broadening of the companion star ( the @xmath20 technique ) have been considered .
table 1 presents our sample and the associated references .
full observational details for every target are given in paper i. for the case of xte j1118 + 480 , we decided to include also 34 additional spectra obtained with the 10.4 m gran telescopio canarias ( gtc ) and reported in @xcite . note that gro j0422 + 320 is not included despite having a @xmath0 determination through the @xmath20 technique .
this is because the resolution of the only three quiescent spectra available to us is too poor ( @xmath21% of the double peak separation ) for a realistic determination of the double peak separation . for comparison
we have also chosen a sub - sample of quiescent cataclysmic variables ( cvs ) from paper i. from the 24 possible cvs with reliable @xmath0 determinations , 13 have been rejected because the line core is dominated by a strong s - wave component which prevents us from measuring the double peak separation with the simple model outlined below . in any case , these are all cvs with large mass ratios @xmath22 which , as we will show later , are not relevant to our analysis . from the remaining 11 cvs , three have @xmath0 values based on the @xmath20 technique ( gk per , ip peg and ctcv j1300 - 3052 ) and another two on a direct determination of the radial velocity curves of both the white dwarf and the donor star ( u gem and wz sge ) . in the remaining six the companion star is not directly observed but robust @xmath0 values are available through modeling the eclipse of the white dwarf and the hot spot ( see e.g. @xcite ) . our cv sample is listed in table 2 .
double peak separations ( @xmath1 ) were obtained by fitting a symmetric 2-gaussian model to the average h@xmath3 profile in every sxt and cv . in the case of eclipsing cvs
we excluded those spectra obtained within @xmath130.05 phases from the time of the central eclipse .
the fitted model consists of a constant plus two gaussians of identical width and height .
the continuum rectified spectra were fitted in a window of @xmath1310000 km s@xmath23 , centered on the h@xmath3 line after masking the neighboring hei line at @xmath246678 . prior to the fit
, the 2-gaussian model was degraded to the resolution of the data by convolution with the instrumental profile% ) on the ratio @xmath7 because both quantities are almost equally affected .
] we adopted 1-@xmath25 formal errors on the fitted parameter as derived through @xmath26 minimization .
1 displays some fit examples using our 2-gaussian model .
in addition , @xmath2 values were extracted from single gaussian fits to the same average h@xmath3 profiles following paper i. tables 1 and 2 list the parameter @xmath7 and its propagated 1-@xmath25 error as derived from our gaussian model fits .
the evolution of @xmath7 with @xmath0 is presented in fig .
the figure shows that @xmath7 varies very rapidly for small @xmath0 values , with a 10% increase for @xmath0 under 0.2 . in order to understand this behaviour
, we follow on from paper i and start by assuming that the @xmath2 of the h@xmath3 line is determined by gas with keplerian velocity at a characteristic radius @xmath27 , with @xmath28 and @xmath29 the roche lobe of the compact star , i.e. @xmath30 on the other hand , the double peak separation is set by the velocity of the outer disc , whose radius @xmath31 is truncated by the tidal forces of the companion star @xcite .
there is ample evidence for the outer disc velocities to be sub - keplerian ( e.g. @xcite ) and thus we decided to adopt @xmath32 where the parameter @xmath33 accounts for the fraction by which the outer disc material is sub - keplerian . for extreme mass ratios
@xmath34 the disc is effectively truncated at the resonance radius of the ( j=3 , k=2 ) commensurability or 3:1 resonance radius i.e. the radius at which the disc angular velocity is three times the angular velocity of the companion star ( see @xcite ) .
therefore , @xmath35 where @xmath36 the binary separation . if we now bring eq . 3 into eq . 2 and use eggleton s relation @xcite to remove @xmath37
we find @xmath38 where @xmath39 is the same expression as in eq .
6 of paper i , i.e. @xmath40 by computing the ratio between the double peak separation and the line width we have managed to cancel out the dependence on compact object mass and binary inclination .
interestingly , in contrast with the @xmath12 correlation presented in paper i , eq .
4 is very sensitive to the mass ratio for @xmath34 i.e. the typical values achieved by bh sxts . for the sake of comparison
, we also plot eq . 4 in fig . 2 for @xmath41 ( adopted from paper i ) and @xmath42 .
although not intended to be a fit , the alignment of the data with the model indicates that eq . 4 provides a good description of the observations
this endorses our interpretation that the strong dependence of @xmath7 with @xmath0 is driven by the truncation of the outer disc radius caused by the 3:1 resonance tide .
the bottom panel in fig .
2 also displays the evolution of @xmath7 for the cv sample , with the model for @xmath41 and @xmath43 superimposed .
the comparison of the sxt and cv data with the model indicates that the velocities at the outer accretion disc are sub - keplerian by @xmath620% , in good agreement with other studies ( e.g. @xcite ) .
note that for @xmath44 the 3:1 resonance lies beyond @xmath60.9 @xmath29 and the outer disc radius is then limited by the largest non - intersecting orbits allowed by three - body interactions @xcite . under these circumstances ,
the dependence of the outer disc radius on @xmath0 is very weak @xcite and , therefore , a nearly constant evolution of @xmath7 versus @xmath0 is expected for @xmath44 . for practical purposes , we display in fig .
3 the variation of @xmath7 versus @xmath0 in logarithmic units . a least - squares linear fit yields @xmath45 witha pearson correlation coefficient @xmath46 . in order to estimate the error in @xmath0 implied by this relation we have computed the difference with respect to the true observed values for our 11 sxts .
the distribution of differences can be approximated by a normal function with @xmath47 .
this indicates that mass ratios can be realistically obtained from eq .
6 with a typical @xmath48% uncertainty .
the spectra that we have used to produce the @xmath8 correlation are orbital averages .
this is party because individual spectra rarely possess enough signal - to - noise for the technique to be applicable . but also , because orbital means average out possible asymmetries in individual spectra from , for example , hot - spots or disc eccentricitiies which could potentially bias the determination of @xmath0 . at this point
we decided to explore the impact of line asymmetries in the results of our technique . since this can only be tested on data with sufficient signal - to - noise we have focused on the 154 high quality gtc spectra of xte j1118 + 480 obtained along four different orbits over two years .
we have performed gaussian fits on every individual spectrum and computed mass ratios using eq .
the distribution of @xmath0 values is found to peak at @xmath49 ( bottom panel in fig .
4 ) , with 68% of the values contained between @xmath50 and 0.045 .
this indicates that , if @xmath0 were to be obtained from a single individual spectrum , the typical uncertainty would be about @xmath1756% . in order to trace the effect of line asymmetries we also extracted the @xmath51 parameter from each spectrum , with @xmath51 defined as the ratio of equivalent widths between the blue and the red part of the h@xmath3 profile
the two halves of the line are set from the rest wavelength till @xmath132500 km s@xmath23 .
for example , @xmath52 indicates a line with the red part stronger by 20% while @xmath53 implies a symmetric profile . a plot of @xmath0 versus @xmath51 ( see top panel in fig .
4 ) seems to show a trend , with asymmetric profiles preferring slightly lower @xmath0 values , although the large scatter prevents from drawing a firm conclusion .
however , as we have mentioned above , the technique outlined in this paper is most useful on phase averaged spectra because individual spectra typically have very limited signal - to - noise .
consequently , we have extracted @xmath0 values by fitting gaussians to the four orbital averages of the 154 individual spectra .
the distribution of @xmath0 values has a mean at 0.026 and a standard deviation of 0.005 , indicating that the typical error on phase averaged spectra is reduced to @xmath1719% i.e. smaller than the 25% uncertainty drawn from the correlation .
we therefore conclude that the uncertainty expected from the application of our technique to phase averaged spectra is about 32% .
in paper i we showed that the @xmath2 of the h@xmath3 line in quiescent sxts and cvs is formed at @xmath5442% of @xmath29 .
furthermore , it is tightly correlated with the projected velocity of the donor star @xmath55 and thus , the quantity @xmath56 can be used to extract dynamical information from single epoch low resolution ( @xmath57 ) spectroscopy .
in addition , we showed that @xmath56 is weakly dependent on @xmath0 , resulting in a @xmath1727% flatter slope for ( long - period ) cvs .
we here now present a new method to estimate the binary mass ratio in quiescent bh sxts from the properties of the h@xmath3 line .
we have proved that the quantity @xmath7 is strongly dependent on @xmath0 , with a 10% variation for @xmath34 .
the reason behind this is the efficient truncation of the outer disc radius by the 3:1 tidal resonance of the donor star .
the correlation of @xmath7 with @xmath0 , therefore , opens a new avenue to measure mass ratios in quiescent bh sxts .
the double peak separation can be solidly measured by fitting a symmetric double - gaussian model to phase averaged h@xmath3 profiles .
we estimate that instrumental resolution better than 25% of the double peak separation is required to resolve the latter .
this typically demands resolving powers of only @xmath581000 i.e a factor @xmath59 lower than required to measure @xmath0 using the @xmath20 technique .
more significantly for observational feasibility , this method makes use of the disc h@xmath3 line which , with a typical @xmath60 , is much stronger that the weak atmospheric features of the donor star .
the relations presented in this paper and in paper i thus allow for a reasonably accurate estimation of the system parameters in very faint sxts which otherwise can not be tackled with current instrumentation and standard techniques . as an example , we have applied our method to the bh sxts xte j1650 - 500 , xte j1859 + 226 and swift j1357 - 0933 .
they all have @xmath61 and none of them has yet a mass ratio determination .
we have produced averaged spectra for xte j1650 - 500 and xte j1859 + 226 using the data presented in table 1 of paper i. regarding swift j1357 - 0933 we have used a more recent and extended database reported in @xcite .
fig.5 displays the averaged spectra of the three bhs together with the best double - gaussian model fits , which result in @xmath62 , [email protected] and [email protected] for xte j1650 - 500 , xte j1859 + 226 and swift j1357 - 0933 respectively .
the mass ratios implied by eq . 6 are @xmath9 , 0.049@xmath10 and 0.040@xmath11 , respectively .
the quoted uncertainties correspond to 68% confidence regions and have been computed using a monte carlo simulation with @xmath63 realizations .
we note in passing that our mass ratio for swift j1357 - 0933 is in excellent agreement with an independent estimate based on the radial velocity curve of the wings of the h@xmath3 line @xcite .
regarding swift j1357 - 0933 we are now in a position to present a credible bh mass based on our scaling relations . by combining our mass ratio with the mass function obtained by means of the @xmath56 correlation @xcite and a conservative estimate of the inclination angle @xmath64 @xcite we find @xmath65 m@xmath14 .
this result confirms swift j1357 - 0933 as one of the most massive bh low - mass x - ray binaries in the galaxy , only rivaled by grs 1915 + 105 @xcite . to conclude we would like to stress that the relations presented here and in paper i will help deepen the search for new bh transients to substantially fainter limits .
they will also prove very useful in extracting fundamental parameters from large numbers of data to be delivered by new spectroscopic surveys such as gaia or weave .
we would like to acknowledge the hospitality of the department of physics of the university of oxford , where this work was performed during a sabbatical visit .
we also thank the anonymous referee for valuable comments which helped to improve the quality of the paper .
we are grateful to serena repetto for bringing to our attention a mass ratio determination based on the @xmath20 technique for n. vel 93 ( = grs 1009 - 45 ) .
we further thank p. charles , t. maccarone and t. muoz - darias for useful comments on the manuscript of this paper an on that of paper i. this paper makes use of data obtained from the isaac newton group archive which is maintained as part of the casu astronomical data centre at the institute of astronomy , cambridge .
partially based on observations made with the gtc operated on the island of la palma by the instituto de astrofsica de canarias in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
this work is supported by dgi of the spanish ministerio de educacin , cultura y deporte under grants aya2013 - 42627 and pr2015 - 00397 .
baptista , r. , catalan , m.s .
, horne , k. & zilli , d. 1998 , , 300 , 233 beekman , g. , somers , m. , naylor , t. & hellier , c. 2000 , , 318 , 9 calvelo , d.e . , vrtilek , s.d . , steeghs , d. , torres , m.a.p . ,
neilsen , j. , filippenko , a.v . &
gonzlez hernndez , j.i .
2009 , , 399 , 539 casares , j. 2015 , , 808 , 80 casares , j. 1996 , astrophysics and space science library , proc . of the 158th coll . of iau ,
a. evans and janet h. wood ed .
, dordrecht : kluwer academic publishers , vol .
208 , p.395 casares , j. , charles , p.a . ,
jones , d.h.p . ,
rutten , r.g.m . & callanan , p.j .
1991 , , 250 , 712 casares , j. & charles , p.a .
1994 , , 271 , l5 casares , j. et al . 2009a , , 181 , 238 casares , j. & jonker , p.g .
2014 , , 183 , 223 corral - santana , j.m . , casares , j. , muoz - darias , t. , rodrguez - gil , p. , shahbaz , t. , torres , m.a.p . , zurita , c. & tyndall , a.a .
2013 , science , 339 , 1048 corral - santana , j.m . , casares , j. , muoz - darias , t. , bauer , f.b , martnez - pais , i.g . & russell , d.m .
2015 , , 587 , a61 eggleton , p.p . 1983 , , 268 , 368 frank , j. , king , a.r . &
raine , d.j .
2002 , , 3rd edn . ,
cambridge university press friend , m.t . , martin , j.s . , connon smith , r. & jones , d.h.p . 1990 , , 246 , 637 gonzlez hernndez , j.i
, rebolo , r. & casares , j. 2014 , , 438 , l21 harlaftis , e.t . ,
horne , k. & filippenko , a.v .
1996 , , 108 , 762 harlaftis , e.t . , steeghs , d. , horne , k. & filippenko , a.v .
1997 , , 114 , 1170 hirose , m. & osaki , y. 1990 , , 42 , 135 littlefair , s.p .
2008 , , 388 , 1582 long , k.s . & gilliland , r.l . 1999 , , 511 , l916 macias , p. et al . 2011 ,
43 , 2011 marsh , t.r . ,
robinson , e.l . & wood , j.h .
1994 , , 266 , 137 mata snchez , d. , muoz - darias , t. , casares , j. , corral - santana , j.m .
& shahbaz , t. 2015 , , 454 , 2199 morales - rueda , l. , still , m.d . ,
roche , p. , wood , j.h .
& lockley , j.j .
2002 , , 329 , 597 north , r.c . , marsh , t.r . ,
kolb , u. , dhillon , v.s .
& moran , c.k.j .
2002 , , 313 , 383 orosz , j.a .
2011 , , 730 , 75 paczyski , b. 1977 , , 216 , 822 papaloizou , j. & pringle , j.e . 1977 , , 181 , 441 reid , m.j . ,
mcclintock , j.e . ,
steiner , j.f . , steeghs , d. , remillard , r.a . , dhawan , v. & narayan , r. 2014 , , 796 , 2 savoury , c.d.j . ,
littlefair , s.p . ,
marsh , t.r .
, dhillon , v.s . , parsons , s.g . ,
copperwheat , c.m . &
steeghs , d. 2012 , , 422 , 469 shahbaz , t. , watson , c.a . & dhillon , v.s .
2014 , , 440 , 504 southworth , j. , hickman , r.d.g . , marsh , t.r .
, rebassa - mansergas , a. , gnsicke , b.t . , copperwheat , c. m. & rodrguez - gil , p. 2009 , , 507 , 929 steeghs , d. , howell , s.b .
, knigge . , c. , gnsicke , b.t . ,
sion , e.m . &
welsh , w. 2007 , , 667 , 442 tanaka , y. & shibazaki , n. 1996 , , 34 , 607 tomsick , j.a . , heindl , w.a . , chakrabarty , d. & kaaret , p. 2002
, , 581 , 570 wade , r.a . & horne , k. 1988 , , 324 , 411 wood , j. & horne , k. 1990 , , 242 , 606 wu , j. et al .
2015 , , 806 , 92 parameter with mass ratio @xmath0 for sxts .
blue solid circles indicate bhs while nss are marked by open circles .
the solid line represents eq . 4 for @xmath41 and @xmath42 .
bottom : same as above for cvs .
the solid line depicts eq . 4 for @xmath41 and @xmath43 . [ fig2 ] ] lccc v404 cyg & 0.067 @xmath13 0.005 & 0.5636 @xmath13 0.0024 & 1 + bw cir & 0.12 @xmath13 0.03 & 0.5534 @xmath13 0.0032 & 2 + xte j1550 -564 & 0.033 @xmath13 0.008 & 0.5705 @xmath13 0.0039 & 3 + n. oph 77 & 0.014 @xmath13 0.016 & 0.5991 @xmath13 0.0119 & 4 + n. mus 91 & 0.079 @xmath13 0.007 & 0.5679 @xmath13 0.0037 & 5 + gs 2000 + 25 & 0.042 @xmath13 0.012 & 0.5801 @xmath13 0.0094 & 6 + a0620 - 00 & 0.067 @xmath13 0.010 & 0.5766 @xmath13 0.0013 & 7 + n vel 93 & 0.055 @xmath13 0.010 & 0.5689 @xmath13 0.0038 & 8 + xte j1118 + 480 & 0.024 @xmath13 0.009 & 0.5873 @xmath13 0.0010 & 9 + cen x-4 & 0.176 @xmath13 0.003 & 0.5446 @xmath13
0.0001 & 10 + xte j2123 - 058 & 0.37 @xmath13 0.15 & 0.5298 @xmath13 0.0234 & 11 + lccc gk per & 0.55 @xmath13 0.21 & 0.5551 @xmath13 0.0007 & 1 + sdss j100658.40 + 233724 & 0.51 @xmath13 0.08 & 0.5402 @xmath13 0.0013 & 2 + u gem & 0.359 @xmath13 0.013 & 0.5663 @xmath13 0.0013 & 3,4 + ip peg & 0.32 @xmath13 0.08 & 0.5760 @xmath13 0.0035 & 5 + ctcv j1300 - 3052 & 0.25 @xmath13 0.03 & 0.5681 @xmath13 0.0007 & 6 + ht cas & 0.150 @xmath13 0.015 & 0.5825 @xmath13 0.0027 & 7 + oy car & 0.102 @xmath13 0.003 & 0.6100 @xmath13 0.0003 & 8 + v2051 oph & 0.19 @xmath13 0.03 & 0.5737 @xmath13 0.0043 & 9 + sdss 103533.02 + 055158.3 & 0.055 @xmath13 0.002 & 0.6311 @xmath13 0.0016 & 8 + wz sge & 0.088 @xmath13 0.013 & 0.6150 @xmath13 0.0007 & 10 + sdss j143317.78 + 101123.3 & 0.069 @xmath13 0.003 & 0.5777 @xmath13 0.0025 & 8 + | we find that the mass ratio @xmath0 in quiescent black hole ( bh ) x - ray transients is tightly correlated with the ratio of the double peak separation ( @xmath1 ) to the full - width - half maximum ( @xmath2 ) of the h@xmath3 emission line , @xmath4 .
the correlation is explained through the efficient truncation of the outer disc radius by the 3:1 resonance with the companion star .
this is the dominant tidal interaction for extreme mass ratios @xmath5 , the realm of bh ( and some neutron star ) x - ray transients .
mass ratios can thus be estimated with a typical uncertainty of @xmath632% , provided that the h@xmath3 profile used to measure @xmath7 is an orbital phase average .
we apply the @xmath8 relation to the three faint bh transients xte j1650 - 500 , xte j1859 + 226 and swift j1357 - 0933 and predict @xmath9 , 0.049@xmath10 and 0.040@xmath11 , respectively .
this new relation , together with the @xmath12 correlation presented in paper i @xcite allows the extraction of fundamental parameters from very faint targets and , therefore , the extension of dynamical bh studies to much deeper limits than was previously possible . as an example , we combine our mass ratio determination for swift j1357 - 0933 with previous results to yield a bh mass of [email protected] m@xmath14 .
this confirms swift j1357 - 0933 as one of the most massive bh low - mass x - ray binaries in the galaxy . |
the generalization of mesonic chiral perturbation theory @xcite to the one - nucleon sector originally turned out to be problematic due to the fact that higher - order loops contribute to lower - order calculations @xcite ( for a recent review see , e.g. , ref .
this problem was first overcome in the framework of the so - called heavy - baryon approach to chiral perturbation theory @xcite .
more recently , it has been realized that the power counting can also be restored in the original manifestly lorentz - invariant formulation @xcite . in this context , the infrared ( ir ) regularization of becher and leutwyler @xcite ( based on ideas of ref .
@xcite ) is the most widely used renormalization scheme for one - loop diagrams @xcite . in ref .
@xcite it has been suggested that the ir renormalization can also be generalized to multiloop diagrams . in a recent paper @xcite ,
we have reformulated the ir renormalization in a form analogous to the extended on - mass - shell ( eoms ) scheme of ref . @xcite . in its new formulation
the ir renormalization can also be directly applied to diagrams involving resonances as well as multiloop diagrams . in ref .
@xcite the application of the eoms scheme to two - loop diagrams has been illustrated for the nucleon self - energy rainbow " diagram in the chiral limit .
the aim of this work is to demonstrate the application of the reformulated ir renormalization @xcite to two - loop diagrams . for comparison the results of the eoms renormalization
are also given .
in order to simplify the calculations and without loss of generality to make the renormalization procedure more transparent , we suppress the complicated spin and chiral structure of baryon chiral perturbation theory and restrict ourselves to the toy model lagrangian @xmath0 where @xmath1 and @xmath2 are scalar particles with masses @xmath3 and @xmath4 , respectively , @xmath5 , and @xmath6 .
the lagrangian @xmath7 contains all interaction terms which are consistent with lorentz invariance and the invariance under the transformations @xmath8 and @xmath9 .
both the ir and the eoms renormalization are constructed such that , after subtraction , feynman diagrams have certain `` chiral '' orders @xmath10 , which are determined by the following power counting : if @xmath11 stands for small quantities such as the @xmath1 mass @xmath3 , small external four momenta of @xmath1 or small external three - momenta of @xmath2 , then the @xmath12 interaction explicitly shown in eq .
( [ lagrange ] ) counts as @xmath13 , the @xmath2 propagator as @xmath14 , the @xmath1 propagator as @xmath15 , and a loop integration in @xmath16 dimensions as @xmath17 , respectively .
as a specific example of a two - loop diagram , we consider the @xmath2 self - energy diagram of fig .
[ fse : fig ] ( a ) , @xmath18 where @xmath19 is a symmetry factor and @xmath20}\ ] ] with @xmath16 denoting the number of space - time dimensions . using the above power counting
, we assign the order @xmath21 to the diagram of fig .
[ fse : fig ] ( a ) .
a dimensional counting analysis @xcite ( see the appendix for an illustration ) suggests that , if first @xmath22 and then @xmath23 , the self - energy @xmath24 can be written as @xmath25 the functions @xmath26 , @xmath27 , and @xmath28 can be expanded in nonnegative integer powers of @xmath29 and , in the following , will be analyzed in detail .
first , we consider @xmath26 .
the taylor expansion of @xmath30 in @xmath29 is obtained by expanding the integrand of eq .
( [ orint ] ) in @xmath29 and interchanging summation and integration : @xmath31}.\ ] ] the integrals in eq .
( [ f ] ) can be written as @xcite @xmath32}=\nonumber\\ & & \sum_{l=0}^{\infty}(p^2-m^2)^l f_{ij , l}^{(1)}(m^2,n)+(p^2-m^2)^{2n-5 - 2i-2j}\sum_{l=0}^{\infty } ( p^2-m^2)^lf_{ij , l}^{(2)}(m^2,n).\end{aligned}\ ] ] inserting eq .
( [ fcoeff ] ) into eq .
( [ f ] ) and taking into account that @xmath33 counts as order @xmath11 , we see that the part which is proportional to noninteger powers of @xmath33 ( for noninteger @xmath16 ) is of order @xmath21 and therefore satisfies the power counting .
the other part is a taylor expansion in @xmath29 and @xmath34 and contains terms which violate the power counting .
this taylor series can be obtained by formally expanding the integrand of the original integral of eq .
( [ orint ] ) in @xmath29 and @xmath33 @xcite and interchanging summation and integration .
those terms in the taylor series which do not violate the power counting contain ir divergences . in the ir renormalization , all terms of the so obtained taylor series must be subtracted from the original integral .
[ these subtractions are generated by the counterterm diagrams of fig .
[ fse : fig ] ( c ) ] .
however , before the subtraction all ir divergences need to be removed from the subtraction terms @xcite .
this is necessary to ensure that the subtracted expression does not contain ir divergences . in the following
, the omission of ir divergences from the subtraction terms in the ir renormalization is always implied . in the eoms scheme
we only need to subtract those terms which violate the power counting ( these terms are free of infrared divergences)in the present case all terms of order @xmath35 or less @xcite . up to and including order @xmath35 the subtraction terms of the eoms renormalization scheme
are given by @xmath36 where the coefficients are given by @xmath37 the corresponding subtraction terms of the ir renormalization scheme read @xmath38 with @xmath39 next , let us investigate @xmath27 which is identified from eq .
( [ orint ] ) by rescaling @xmath40 and @xmath41 .
both cases produce an overall factor of @xmath42 .
one then expands the remaining integrands in @xmath3 , interchanges summation and integration , and adds up the two contributions . since the integral in eq .
( [ orint ] ) is invariant under the interchange of @xmath43 and @xmath44 , we only need to perform one of the above manipulations and then multiply the result with a factor of @xmath45 . as a result
we obtain @xmath46 where @xmath47 is a binomial coefficient and the integrals @xmath48 are defined as @xmath49}^{1+j}}.\ ] ] for @xmath50 odd , the loop integral of eq .
( [ loopint ] ) vanishes , because in that case the integrand is an odd function of @xmath43 . from this one
concludes that the nonvanishing terms in eq .
( [ g ] ) are proportional to nonnegative integer powers of @xmath29 . using a dimensional counting analysis @xcite , it can be shown that @xmath48 is of the form @xmath51 combining this result with eqs .
( [ intsplit ] ) and ( [ g ] ) , we see that the term nonanalytic in @xmath33 is of order @xmath52 and therefore does not violate the power counting . on the other hand , the first part of eq .
( [ loopanalysis ] ) contains terms that are analytic in @xmath33 which , when combined with eqs .
( [ intsplit ] ) and ( [ g ] ) , give rise to contributions which are nonanalytic in @xmath3 : @xmath53\frac{\gamma(1-n/2)}{(4\pi)^{n/2 } } + \cdots.\ ] ] note that the first term in eq .
( [ g1 ] ) violates the power counting and , since it is nonanalytic in @xmath3 , can not be directly absorbed by a counterterm . in solving this apparent puzzle , one has to keep in mind that , in order to consistently renormalize the diagram of fig .
[ fse : fig ] ( a ) , all relevant contributions from counterterms in @xmath7 of eq .
( [ lagrange ] ) must be taken into account . in the present case ,
these countertems originate in the renormalization of the one - loop diagrams of fig .
[ scatt : fig ] ( a ) and ( b ) with the corresponding expressions @xmath54}\ ] ] and @xmath55},\ ] ] respectively . both diagrams are assigned the order @xmath56 .
evaluating eqs .
( [ sub1 ] ) and ( [ sub2 ] ) we find that @xmath57 and @xmath58 contain contributions that violate the power counting .
to renormalize these diagrams we follow refs.@xcite and , up to order @xmath11 , find the subtraction terms as @xmath59 and @xmath60\right\},\ ] ] where @xmath61 the counterterms are then given by @xmath62 for the eoms scheme and by @xmath63\ ] ] in the ir renormalization . in a two - loop calculation
these counterterms give a contribution to the self - energy as shown in fig .
[ fse : fig ] ( b ) .
the corresponding expressions read @xmath64 and @xmath65i_{\pi},\end{aligned}\ ] ] where @xmath66 in fact , the last term in eq .
( [ ctselfir ] ) is of order @xmath67 , @xmath68 and can therefore be neglected in a calculation up to and including order @xmath69 .
noting that @xmath70 we see that in the ir renormalization the self - energy contribution of eq .
( [ ctselfir ] ) , which stems from the renormalization of one - loop diagrams , exactly cancels the contributions of eq .
( [ g1 ] ) which are explicitly shown , including the part which violates the power counting . in the eoms scheme
no terms that satisfy the power counting are subtracted .
let us finally consider the last term of eq .
( [ intsplit ] ) . the function @xmath28 is obtained from eq .
( [ orint ] ) by simultaneously rescaling @xmath71 and @xmath72 , extracting a factor of @xmath73 , expanding the remaining integrand in @xmath3 , and interchanging integration and summation , yielding @xmath74^{i - j}[m(k_1+k_2)^2]^j}{(k_1 ^ 2 - 1+i0^+ ) ( k_2 ^ 2 - 1+i0^+)}.\end{aligned}\ ] ] noting that @xmath75 , it is easy to see that eq .
( [ h ] ) in combination with the factor @xmath73 of eq .
( [ intsplit ] ) satisfies the power counting .
furthermore , only nonnegative integer powers of @xmath29 survive in @xmath76 . combining the results above
, all terms violating the power counting are canceled in the sum of the diagrams in fig .
[ fse : fig ] in both the ir and the eoms renormalization .
finally , we have also studied the case where first @xmath23 and then @xmath22 , and have explicitly verified that the renormalization procedure remains exactly the same , i.e. , the counterterms are the same and the renormalized diagram satisfies the power counting .
in conclusion , we have demonstrated that the application of both the reformulated ir renormalization and the eoms scheme to two - loop diagrams leads to a consistent power counting . in this context
, the subtraction of one - loop sub - diagrams plays an important role in the renormalization of two - loop diagrams .
the procedure can be applied iteratively to multiloop diagrams .
calculations using the lagrangian of baryon chiral perturbation theory are more involved , but the general features of the renormalization program do not change . j. gegelia acknowledges the support of the alexander von humboldt foundation .
in this appendix we provide an illustration of the dimensional counting analysis of ref .
@xcite in terms of a specific example . to that end
let us consider the integral of eq .
( [ sub1 ] ) for @xmath77 : @xmath78}.\ ] ] one would like to know how the integral behaves for small values of @xmath3 and/or @xmath33 as a function of @xmath16 . if we consider , for fixed @xmath79 , the limit @xmath80 , the integral @xmath81 can be represented as @xmath82 where the functions @xmath83 are analytic in @xmath29 and are obtained as follows .
first , one rewrites the variable of the loop integration as @xmath84 , where @xmath85 is an arbitrary nonnegative real number .
next , one isolates the overall factor of @xmath86 so that the remaining integrand can be expanded in positive powers of @xmath3 and interchanges the integration and summation .
the resulting series represents the expansion of @xmath87 in powers of @xmath29 .
the summation over @xmath88 includes all values of @xmath88 for which @xmath83 is nontrivial . to be specific ,
we obtain for @xmath81 : @xmath89[\tilde k^2 m^{2 a_i}+2 p\cdot \tilde k m^{a_i}+ p^2-m^2+i0^+]}. \label{ippsichv}\ ] ] from eq .
( [ ippsichv ] ) we see that the second propagator does not contribute to the overall factor @xmath86 for any @xmath85 and has to be expanded in ( positive ) powers of @xmath90 unless @xmath91 . for @xmath92 , we rewrite the first propagator as @xmath93 and expand the second factor in eq .
( [ 1pr ] ) in positive powers of @xmath94 . on the other hand ,
if @xmath95 we write the first propagator as @xmath96 and expand the second factor in eq .
( [ 1pr2 ] ) in positive powers of @xmath97 . in both cases
one obtains integrals of the type @xmath98 as the coefficients of the expansion .
however , such integrals vanish in dimensional regularization .
therefore the only nontrivial terms in the sum of eq .
( [ ippsiexp ] ) correspond to either @xmath91 or @xmath99 .
thus we obtain @xmath100 where @xmath101^{1+i}[k^2 + 2 p\cdot k+ p^2-m^2+i0^+ ] } , \label{ippsi2}\ ] ] and @xmath102}. \label{ippsi1}\ ] ] on the other hand , the integral @xmath81 can be calculated directly without applying the dimensional counting technique .
doing so and comparing with the results given above one sees that the dimensional counting method leads to the correct expressions . while the loop integrals of eq .
( [ ippsi1 ] ) have a simple analytic structure in @xmath33 , the technique sketched above can be directly applied to the loop integrals of eq .
( [ ippsi2 ] ) when @xmath103 , now using the change of variable @xmath104 with arbitrary nonnegative real numbers @xmath105 .
s. weinberg , physica a * 96 * , 327 ( 1979 ) .
j. gasser and h. leutwyler , ann .
( n.y . ) * 158 * , 142 ( 1984 ) .
j. gasser and h. leutwyler , nucl .
b250 * , 465 ( 1985 ) . j. gasser , m. e. sainio , and a. varc , nucl .
* b307 * , 779 ( 1988 ) . s. scherer , in _ advances in nuclear physics , vol . 27 _ , edited by j. w. negele and e. w. vogt ( kluwer academic / plenum publishers , new york , 2003 ) .
e. jenkins and a. v. manohar , phys .
b * 255 * , 558 ( 1991 ) .
v. bernard , n. kaiser , j. kambor , and u .- g .
meiner , nucl .
* b388 * , 315 ( 1992 ) . h. tang , hep - ph/9607436 .
p. j. ellis and h. tang , phys .
c * 57 * , 3356 ( 1998 ) .
t. becher and h. leutwyler , eur .
j. c * 9 * , 643 ( 1999 ) .
j. gegelia and g. japaridze , phys .
d * 60 * , 114038 ( 1999 ) .
m. lutz , nucl .
* a677 * , 241 ( 2000 ) . m. f. lutz and e. e. kolomeitsev , nucl .
phys . * a700 * , 193 ( 2002 ) . t. fuchs , j. gegelia , g. japaridze , and s. scherer , phys . rev .
d * 68 * , 056005 ( 2003 ) .
t. fuchs , m. r. schindler , j. gegelia , and s. scherer , hep - ph/0308006 , to appear in phys .
b. t. fuchs , j. gegelia , and s. scherer , hep - ph/0309234 .
p. j. ellis and k. torikoshi , phys .
c * 61 * , 015205 ( 2000 ) .
b. kubis and u .- g .
meiner , nucl .
* a679 * , 698 ( 2001 ) . s. l. zhu , s. puglia , and m. j. ramsey - musolf , phys .
d * 63 * , 034002 ( 2001 ) .
b. kubis and u .- g .
meiner , eur .
j. c * 18 * , 747 ( 2001 )
. j. l. goity , d. lehmann , g. prezeau , and j. saez , phys .
b * 504 * , 21 ( 2001 ) .
t. becher and h. leutwyler , j. high energy phys .
* 0106 * , 017 ( 2001 )
. b. borasoy and s. wetzel , phys .
d * 63 * , 074019 ( 2001 ) .
n. beisert and b. borasoy , eur .
j. a * 11 * , 329 ( 2001 ) .
s. l. zhu , g. sacco , and m. j. ramsey - musolf , phys .
d * 66 * , 034021 ( 2002 ) .
n. beisert and b. borasoy , nucl .
* a705 * , 433 ( 2002 ) . v. bernard , t. r. hemmert , and u .- g .
meiner , phys .
b * 545 * , 105 ( 2002 ) .
j. gasser , m. a. ivanov , e. lipartia , m. moji , and a. rusetsky , eur .
j. c * 26 * , 13 ( 2002 ) .
k. torikoshi and p. j. ellis , phys . rev .
c * 67 * , 015208 ( 2003 ) .
v. bernard , t. r. hemmert , and u .- g .
meiner , phys .
b * 565 * , 137 ( 2003 ) .
r. higa and m. r. robilotta , phys .
c * 68 * , 024004 ( 2003 ) .
r. higa , m. r. robilotta , and c. a. da rocha , nucl - th/0310011 .
d. lehmann and g. prezeau , phys .
d * 65 * , 016001 ( 2002 ) .
m. r. schindler , j. gegelia , and s. scherer , hep - ph/0309005 .
j. gegelia , g. japaridze , and x. q. wang , j. phys .
g * 29 * , 2303 ( 2003 ) .
j. gegelia , g. s. japaridze , and k. s. turashvili , theor .
* 101 * , 1313 ( 1994 ) [ teor
. mat .
225 ( 1994 ) ] . | using a toy model lagrangian we demonstrate the application of both infrared and extended on - mass - shell renormalization schemes to multiloop diagrams by considering as an example a two - loop self - energy diagram .
we show that in both cases the renormalized diagrams satisfy a straightforward power counting . |
for a large class of nonlinear evolutions the size of the resolvent has been proposed as an explanation of instability for spectrally stable problems .
celebrated examples include the plane couette flow , plane poiseuille flow and plane flow
see trefethen - embree ( * ? ? ?
* chapter 20 ) for discussion and references .
motivated by this we consider the mathematical question of evolution involving a small parameter @xmath3 ( in fluid dynamics problem we can think of @xmath3 as the reciprocal of the reynolds number ) in which the linearized operator has the spectrum lying in @xmath4 , uniformly in @xmath5 , yet the the solutions of the nonlinear equation blow up at time @xmath6 for data of size @xmath7 .
we know of one rigorous example of such a phenomenon given by sandstede - scheel @xcite .
they considered @xmath8 on @xmath9 $ ] with dirichlet boundary conditions , and showed that blow up occurs with arbitrarily small initial data as @xmath10 . in that problem @xmath11
the paper @xcite is our starting point and we use its maximum principle approach to obtain results for suitable operators in any dimension .
in addition , we emphasize the connection with the semiclassical pseudospectrum and provide some numerical comparisons .
we consider a semiclassical nonlinear evolution equation @xmath12 where @xmath13 is the following semiclassical differential operator @xmath14 here , @xmath15 is a potential function with @xmath16 for some @xmath17 . also , @xmath18 has the properties @xmath19 .
finally , @xmath20 .
* remark : * all of our results hold for weaker assumptions on the growth of @xmath21 and @xmath22 , however and are convenient for our purposes .
we will show in section [ sec : spectra ] that for @xmath23 and @xmath22 as in ( [ eqn : potentialprop ] ) and ( [ eqn : vfieldprop ] ) respectively , the linearized problem is spectrally stable , that is , the spectrum is bounded away from @xmath24 uniformly in @xmath3 .
yet , we also show that has an unstable equilibrium at @xmath25 for all potentials @xmath23 satisfying ( [ eqn : potentialprop ] ) and all @xmath22 satisfying ( [ eqn : vfieldprop ] ) .
specifically , we show [ thm : main ] fix @xmath20 .
then , for each @xmath26 where @xmath27 is small enough , there exists @xmath28 such that the solution to with @xmath29 , satisfies @xmath30 where @xmath31 a nice example for which our assumptions hold is ( [ eqn : l ] ) with @xmath32 , @xmath33 , and @xmath34 .
that is @xmath35 it is easy to see ( and will be described in section [ sec : spectra ] ) that @xmath36 for @xmath37 the spectrum intersects the right half plane and thus instability of the linear problem follows .
we are interested in the range @xmath38 , where we will relate the instability of @xmath39 to the presence of pseudospectrum in the right half plane . for more about see ( * ? ? ?
* chapter 12 ) .
in particular , cossu - chomaz @xcite relate it to the linearized ginzburg - landau equation and analyze resolvent of and the norm of the semigroup @xmath40 numerically .
the operator is also closely related to the advection - diffusion operators mentioned above , @xmath41 , on @xmath9 $ ] , with , say , dirichlet boundary conditions ; see ( * ? ? ?
* chapter 12 ) for a discussion and references .
when rescaled using @xmath42 , @xmath43 the operator becomes the semiclassical operator @xmath44 on @xmath45 $ ] .
when the domain is extended to @xmath46 , the potential @xmath47 is added to @xmath48 to produce a confinement similar to a boundary .
we relate the blow - up of solutions to to the presence of pseudospectrum of in the right half plane .
however , because estimates on semigroups for ( [ eqn : main ] ) with quasimode initial data are poor , we are unable to exhibit blow - up starting from a quasimode . instead , we present a simple and explicit construction of quasimodes for @xmath49 ( for a more general setting see @xcite ) .
we then use these quasimodes as initial data in numerical simulations and observe that , although in some cases the ansatz solution blows up more quickly , the solutions with quasimode initial data behave similarly to what is expected from a pure eigenvalue for with positive real part . the plot shows numerical simulations of the evolution of ( [ eqn : main ] ) with @xmath50 and two initial data . the evolution with initial data a real valued @xmath51 error quasimode with eigenvalue @xmath52 is shown in the top two graphs and that with the ansatz constructed in the proof of lemma [ lem : o(1)ndim ] as initial data is shown in the bottom two graphs .
we observe that , when the initial data is a quasimode , blowup occurs in time @xmath53 , while for ansatz initial data , blow - up occurs in time @xmath54 .
however , as would be expected from eigenfunction initial data , we see that the solution with quasimode initial data exhibits little transport to the left . on the other hand ,
the ansatz transports left significantly.,width=624 ] the paper is organized as follows . in section [ sec : spectra ] we review the definitions of spectra and pseudospectra and discuss them for our class of operators . in section [ sec : odq ] we give a construction of quasimodes for one dimensional problems .
although the results are known , ( see @xcite,@xcite,@xcite ) a self - contained presentation is useful since we need the quasimodes for our numerical experiments . also , there is no reference in which analytic potentials ( for which quasimodes have @xmath55 accuracy ) is treated by elementary methods in one dimension .
section [ sec : i ] is devoted to the proof of theorem [ thm : main ] using heat equation methods .
finally , in section [ sec : numerical ] we report on some numerical experiments which suggest that quasimode initial data gives more natural blow - up and that blow - up occurs at complex energies . acknowledgemnts .
the author would like to thank maciej zworski for suggesting the problem and for valuable discussion , guidance , and advice .
thanks also to laurent demanet and trever potter for allowimg him to use their matlab codes , justin holmer for informing him of the paper by sandstede and scheel , and hung tran for comments on the maximum principle in lemma [ lem : o(1)ndim ] .
the author is grateful to the national science foundation for partial support under grant dms-0654436 and under the national science foundation graduate research fellowship grant no .
dge 1106400 .
we do not use the results of this section to prove theorem [ thm : main ] .
instead , we present them to emphasize the connection of the size of the resolvent with instability . we believe that instability based on quasimodes would be more natural and allow for proof of instability at complex energies .
we illustrate this with numerics in section [ sec : numerical ] . to describe the spectrum of @xmath56 , we observe that latexmath:[\[e^{\frac{\rho(x)}{2h}}p(x , hd)e^{-\frac{\rho(x)}{2h}}=-\left[(hd)^2+v(x)+{\frac{1}{4 } } of @xmath56 is given by that of a schrdinger operator with potential @xmath58 . since @xmath23 and @xmath22 have the properties given in ( [ eqn : potentialprop ] ) and ( [ eqn : vfieldprop ] ) respectively , @xmath49 has a discrete spectrum that has real part bounded above by @xmath59 ( see for instance ( * ? ? ? * section 6.3 ) ) .
we now examine the pseudospectral properties of ( [ eqn : l ] ) .
* definition .
* let @xmath60 be a second order semiclassical differential operator .
then , @xmath61 if and only if @xmath62 @xmath63 such that @xmath64 and @xmath65 we say @xmath66 is in the semiclassical _ pseudospectrum _ of @xmath67 if @xmath68 . * remark .
* we note that for @xmath69 , @xmath70 , for any @xmath71 .
this relates our definition to the more standard defintions of pseudospectra in terms of the resolvent . for discussion and generalizations
see dencker @xcite and pravda - starov @xcite .
the criterion for @xmath69 is based on hrmander s bracket condition ( see zworski @xcite and dencker - sjstrand - zworski @xcite ) : @xmath72 then @xmath73 . we use this condition to show that the pseudospectrum of @xmath49 nontrivially interesects the right half plane . specifically , for @xmath49 given by ( [ eqn : l ] ) , @xmath74 $ ] .
first , observe that @xmath75 we have assumed @xmath76 .
therefore , we need only show that , for a dense subset @xmath77 $ ] , @xmath78 implies that there exists @xmath79 such that ( [ eqn : inpseudo ] ) holds for the symbol @xmath80 , at @xmath81 with @xmath82 .
we proceed by contradiction .
suppose there is no such @xmath83 .
then , there exists @xmath84 open such that for all @xmath85 , @xmath86 .
let @xmath87 be the integral flow of @xmath88 and @xmath89 have @xmath90 .
. then @xmath92 .
suppose that @xmath93 escapes every compact set as @xmath94 increases .
then implies that @xmath95 as @xmath94 increases .
let @xmath96 and @xmath97 .
then @xmath98 is finite since @xmath99 and @xmath95 .
together , @xmath100 and @xmath101 open imply the existance of @xmath102 such that for @xmath103 , @xmath104 .
but , @xmath104 implies @xmath105 .
therefore , @xmath106 for @xmath107 and thus , since @xmath95 , there exists @xmath108 such that @xmath109 , a contradiction .
we have shown that there is a dense subset @xmath77 $ ] with @xmath110 .
hence @xmath111\subset \overline{\lambda ( p)}$ ] . next , observe that @xmath112 and thus , @xmath74 $ ] as desired . to finish the proof , we need only show that @xmath93 escapes every compact set .
suppose the flow at @xmath113 exists for all @xmath114 .
define @xmath115 . then @xmath116 and
we have that @xmath117 as @xmath118 .
but , @xmath18 and is therefore bounded on every compact set .
thus , @xmath93 escapes every compact set as @xmath119 . now , suppose the flow at @xmath113 is not global .
then , @xmath93 is an integral curve of @xmath120 with @xmath121 domain a proper subset of @xmath46 .
thus , as proved in ( * ? ? ?
* lemma 17.10 ) , @xmath93 escapes every compact set . putting this together with our discussion of the spectrum of @xmath49
, we have that for @xmath122 and @xmath22 as in , although @xmath123 is bounded away from @xmath24 , @xmath124 nontrivially intersects @xmath24 .
we see that the spectrum of ( [ eqn : l1d ] ) ( blue dots ) is bounded away from @xmath125 , while the pseudospectrum ( shaded region ) enters the right half plane .
the region for which we prove blow - up corresponds to the dashed red line .
, width=288 ] for the specific case , @xmath126 , and @xmath127 constant with @xmath128 , the above argument gives us that @xmath129 in addition , the pseudospectrum is given by , @xmath130 we see that for @xmath37 , the spectrum interesects the right half plane and so instability of @xmath25 is a classical result .
however , for @xmath38 , the spectrum is bounded away from the @xmath24 and only the pseudospectrum enters the right half plane . yet , in the regime @xmath38 , we will show that @xmath25 is unstable and , moreover , for arbitrarily small initial data , the solution blows up in finite time .
we proceed by constructing quasimodes for operators in the one dimensional case with @xmath131 .
we implement wkb expansion for the quasimode following the method used in @xcite .
let @xmath132 where @xmath15 and @xmath21 may be complex .
+ * remark . *
the following theorem is a special case of general theorems about quasimodes ( * ? ? ?
* theorems 2 and 2 ) . for the reader
s convenience we present a direct proof in the spirit of davies @xcite . [
prop : constructquasi ] suppose that @xmath133 is given by and that @xmath134 where @xmath135 satisfies the condition that @xmath136 there exists an @xmath137dependent function @xmath138 , such that @xmath139 and @xmath140 in addition @xmath141 is microlocalized to @xmath142 in the sense that for every @xmath143 vanishing in a neighbourhood of @xmath144 , @xmath145 when @xmath146 is real analytic than we can find @xmath147 such that @xmath148 let @xmath149 with @xmath150 if @xmath151 and @xmath152 if @xmath153 where @xmath154 will be determined below .
define @xmath155 where @xmath156 finally , let @xmath157 for all @xmath32 .
we will find appropriate @xmath158 and @xmath159 in what follows .
first , by a simple computation @xmath160 where @xmath161 are inductively defined by @xmath162 where we use the convention that @xmath163 for @xmath164 or @xmath165 .
now , we set @xmath166 for @xmath167 .
given that @xmath168 is small enough , this will enable us to determine all @xmath158 as well as @xmath159 .
observe that , using the condition , @xmath169 , we obtain @xmath170 now , letting @xmath171 , we have a complex eikonal equation @xmath172 then , letting @xmath173 , we have @xmath174 and hence @xmath175 now , we have assumed that @xmath176 .
therefore there exists @xmath177 such that @xmath178 for all small enough @xmath79 and @xmath179 .
also , for @xmath79 and @xmath179 small enough @xmath180 satisifies @xmath181 .
we choose @xmath182 small enough so that these conditions both hold for @xmath183 and @xmath184 .
the condition @xmath185 , implies @xmath186 with the convention that @xmath187 and initial conditions , @xmath188 putting @xmath189 we obtain @xmath190 and @xmath191 before proceeding to show exponential error for @xmath21 analytic , we show @xmath192 error for arbitrary @xmath21 . to complete the proof of @xmath192 quasimodes
, we need to estimate @xmath193 let @xmath194 denote various positive constants that are independent of @xmath179 and @xmath79 .
then , @xmath195 next , we compute @xmath196 thus , we need to estimate each of the norms .
note that @xmath197 and @xmath198 have support in @xmath199 .
thus , we have @xmath200 similarly , @xmath201 next , observe that @xmath202 now , @xmath203 on @xmath204 , uniformly for @xmath205 .
therefore , combining with inequalities , , , and , gives @xmath192 quasimodes for arbitrary @xmath206 .
we then normalize to obtain @xmath207 .
we will now assume that @xmath23 is real analytic and prove exponential smallness of the error .
[ l : ana ] let @xmath208 $ ] where @xmath209 is a small constant .
suppose that @xmath210 , @xmath211 , @xmath212 , and @xmath213 are holomorphic functions of @xmath214 and @xmath215 for some @xmath216 where @xmath217 is a neighborhood of @xmath218 in @xmath219 .
if @xmath220 is defined inductively by @xmath221 then for some @xmath222 , @xmath223 and @xmath224\subset\tilde{\omega}\subset \omega$ ] , @xmath225 using integration by parts , we obtain that @xmath226 then , since @xmath212 is holomorphic in @xmath227 and @xmath228 for some @xmath216 , we make the conformal change of variables @xmath229 where @xmath230 then , letting @xmath231 , we have @xmath232 where @xmath233 and @xmath234 put @xmath235
. then , since the change of variables was conformal , and @xmath236 , @xmath237 , are holomorphic , we have that there exists @xmath238 such that @xmath239 where we define @xmath240 for a function @xmath241 defined on @xmath227 .
we claim that for some @xmath242 , @xmath243 , @xmath244 we prove the claim by induction .
the holomorphy of @xmath245 gives us the base case .
we now prove the inductive case . by the inductive hypothesis
, we have that @xmath246 . by leibniz rule
, we have that @xmath247 we claim that for @xmath248 , @xmath249 . to see this
, we write this inequality as @xmath250 putting @xmath251 and @xmath252 , the inequality is equivalent to @xmath253 which follows from the monotonicity of the function @xmath254 next , observe that , @xmath255 , @xmath256 where we use @xmath257 . then , since for @xmath258 , @xmath249 , we have @xmath259 such that @xmath260 such that @xmath261 next , choose @xmath262 then , combining , , , and , we have @xmath263 then , since @xmath264 is a change of variables independent of @xmath265 which maps @xmath266 and @xmath267 , we have @xmath268 now , choose @xmath269 with @xmath270 . then ,
since @xmath271 are holomorphic , we apply cauchy estimates to obtain @xmath272 we apply the lemma with @xmath273 , @xmath274 , @xmath275 , and @xmath276 where @xmath277 and @xmath278 are given above .
analyticity of @xmath146 implies that @xmath279 , @xmath280 , and @xmath159 are holomorphic in a neighbourhood of @xmath281 .
then , putting @xmath282 , using lemma [ l : ana ] and that @xmath159 is real analytic , we have @xmath283}|h^n\phi_n|&\leq & h^nc(c_2c_1^{n}(n-1)^{n-1}+2c_2 ^ 2c_1^{n}(n-1)^{n-1}+6c_2 ^ 3c_1^n(n-2)^{n-2}){\nonumber\\}&\leq & c h^nc_1^{n}n^{n}\leq c ( h^nc_1^{n}\left({\frac{1}{ec_1h}}\right)^n)= c e^{-n}= ce^{-{\frac{1}{ec_1h}}}\label{eqn : expsmall}\end{aligned}\ ] ] where @xmath194 denotes various postive constants that are independent of @xmath206 . finally , combining with inequalities , , , , and gives @xmath284 quasimodes .
we then normalize @xmath285 to obtain @xmath207 .
now , applying this result to @xmath286 , we obtain @xmath287 for @xmath288 and @xmath289
. we set @xmath290 and see that the difference between the solution to ( [ eqn : main ] ) with initial data a quasimode with error @xmath291 ( red line ) and the solution with initial data a quasimode with error @xmath51 ( blue dots ) is negligible .
thus , by using @xmath51 error quasimodes , we have not introduced large error into our numerical calculations.,width=576 ]
we show a numerical simulation @xmath292 of the evolution of with a quasimode at imaginary energy as initial data .
specifically , we set @xmath50 and use a quasimode with eigenvalue @xmath293 . the real part is shown in the top graph and the imaginary part in the bottom graph .
we see that although subsolution methods do not apply to these quasimondes , blow up still occurs in time @xmath294.,width=576 ] our approach to obtaining blow - up of ( [ eqn : main ] ) will follow that used by sandstede and scheel in @xcite . we will first demonstrate that , from small initial data , we obtain a solution that is @xmath295 on a deformed ball in time @xmath296 .
we will then use the fact that the solution is @xmath295 on this region to demonstrate that after an additional @xmath297 , the solution to the equation blows up .
first , we prove that there exists initial data so that the solution to ( [ eqn : main ] ) is @xmath295 in time @xmath298 .
recall that @xmath299 denotes the flow of @xmath120 .
[ lem : o(1)ndim ] fix @xmath20 , @xmath300 , @xmath301 , and @xmath302 such that both @xmath303 $ ] for @xmath304 and @xmath299 is defined on @xmath305 for @xmath306 .
then , for each @xmath307 where @xmath308 is small enough , there exists @xmath309 and @xmath310 so that the solution to ( [ eqn : main ] ) with initial data @xmath311 satisfies @xmath312 on @xmath313 .
let @xmath314 solve @xmath315 let @xmath316 and define @xmath317 .
we make the following assumptions on @xmath318 , @xmath319 @xmath320 @xmath321 where @xmath322 are smoothly extendible functions on @xmath323 .
we will construct such a function at the end of the proof .
define @xmath324 by @xmath325 since supp @xmath326 and @xmath299 is defined on @xmath327 , @xmath328 is continuous .
we proceed by showing that @xmath328 is a viscosity subsolution of @xmath329 in the sense of crandall , ishi , and lions @xcite .
first , we show that @xmath328 is a subsolution on @xmath330 for @xmath331 .
@xmath332 now , by taylor s formula , @xmath333 .
hence @xmath334=-\delta w_0(x)+o(t)$ ] .
we have @xmath331 , and @xmath335 on @xmath323 .
therefore , for @xmath168 small enough , @xmath336 .
hence , for @xmath179 small enough independent of @xmath337 , @xmath338 now , since for some @xmath339 and @xmath331 , @xmath340 $ ] we have that @xmath341 for @xmath179 small enough .
thus , @xmath328 is a subsolution on @xmath342 for @xmath331 .
next , we observe that on @xmath343 , @xmath344 and hence is a subsolution of ( [ eqn : linprob ] ) . finally , we consider @xmath345 .
we have that @xmath346 is smooth .
if @xmath347 and @xmath328 is twice differentiable at @xmath348 , then @xmath349 and @xmath328 is clearly a subsolution to at @xmath348 .
let @xmath347 be a point where @xmath328 is not twice differentiable .
suppose that @xmath350 such that @xmath351 has a maximum at @xmath348 .
we take paths through @xmath348 to reduce to a one dimensional problem .
for any path @xmath352 with @xmath353 , define @xmath354 and @xmath355 .
since @xmath328 is nonnegative , continuous , smooth on @xmath342 , and extends smoothly from @xmath342 to a function on @xmath356 for all @xmath331 , @xmath357 and @xmath358 exist .
simlarly , @xmath359 and @xmath360 exist . therefore , since @xmath351 is maximized at @xmath348 , @xmath361 .
now , since @xmath328 is not twice differentiable at @xmath348 , either there exists @xmath362 such that @xmath363 is not differentiable at @xmath364 or @xmath328 is differentiable at @xmath348 , but not twice differentiable .
case 1 : : @xmath365 is a path through @xmath348 for which @xmath363 is not differentiable
. + then @xmath366 and there exists no such @xmath367 .
case 2 : : @xmath328 is differentiable at @xmath113 but not twice differentiable .
+ then , for all @xmath362 through @xmath348 , @xmath368 and @xmath369 .
now , let @xmath370 be the coordinate paths through @xmath113 with @xmath371 for @xmath372 . then ,
since on @xmath373 , @xmath328 is a subsolution of the linearized problem ( [ eqn : linprob ] ) , we have @xmath374 thus , we have that @xmath328 is a subsolution on @xmath346 . putting this together with the arguments above , we have that @xmath328 is a viscosity subsolution for ( [ eqn : linprob ] ) on @xmath331 .
now , by an adaptation of the maximum principle found in ( * ? ? ?
* section 3 ) to parabolic equations , any solution , @xmath314 to ( [ eqn : linprob ] ) with initial data @xmath375 will have @xmath376 for @xmath331 .
but , since @xmath377 , @xmath378 and hence the solution @xmath379 to ( [ eqn : main ] ) with initial data @xmath380 will have @xmath381 for @xmath331 .
now , since for @xmath382 , @xmath383 on @xmath384 , we have the result .
we now construct a function @xmath385 satsifying the assumptions , ( [ eqn : wassume1 ] ) , , and .
let @xmath386 be the ground state solution of the dirichlet laplacian on @xmath387 i.e. @xmath388 then , @xmath386 extends smoothly off of @xmath389 and has @xmath390 inside @xmath389 ( see for instance ( * ? ? ?
* section 6.5 ) ) .
let @xmath391 , @xmath392 with @xmath393 on @xmath394 and supp @xmath395 .
then , define @xmath396 ( |x|^2 - 1)$ ] where @xmath397 is large enough so that @xmath398 .
there exists such an @xmath397 since @xmath399 in @xmath389 and @xmath400 imply that @xmath401 and hence , by hopf s lemma , @xmath402 on @xmath403 where @xmath404 is the outward normal vector to @xmath403 ( see for instance ( * ? ? ?
* section 6.4 ) ) . for @xmath405 , there exists @xmath216 such that , @xmath406(|x|^2 - 1)-4{\langle}\partial \chi , x{\rangle}-2\chi d\leq \lambda mv_1+c.\nonumber\ ] ] for @xmath405 , @xmath407 .
thus , by increasing @xmath397 if necessary , we obtain @xmath408 such that @xmath409 now , for @xmath410 , @xmath411 thus , for @xmath339 small enough , there exists @xmath412 such that @xmath413 finally , @xmath62 @xmath414 , @xmath89 , and @xmath415 constants so that @xmath416 satisfies the conditions on @xmath318 . * remark 1 . *
if a shorter time is desired , one may use initial data of @xmath417 to obtain the same result in time @xmath418 . * remark 2 . *
notice that to obtain a growing subsolution it was critical that @xmath20 .
this corresponds precisely with the movement of the pseudospectrum of @xmath56 into the right half plane .
now , we will demonstrate finite time blow - up using the fact that in time @xmath298 the solution to ( [ eqn : main ] ) is @xmath295 on an open region .
the proof of theorem [ thm : main ] follows we show simulations of solutions to the equation @xmath419 with @xmath50 .
the solution using a quasimode @xmath311 with eigenvalue @xmath52 and error @xmath51 as initial data is shown in the blue solid linse .
the red dotted lines show @xmath420 .
we see that the solution to the linearized problem with quasimode initial data closesly approximates the exponential until @xmath421,width=576 ] let @xmath422 and @xmath423 be the initial data and time found in lemma [ lem : o(1)ndim ] with @xmath424 such that @xmath299 is defined on @xmath425 , @xmath426 $ ] for @xmath427 $ ] , and @xmath428 .
then , @xmath312 on @xmath429 .
now , let @xmath430 be a smooth bump function with @xmath431 on @xmath432 , @xmath433 , supp @xmath434 , and @xmath435 ( one such function is given by @xmath436 for @xmath437 ) .
define @xmath438 by @xmath439 next , let @xmath440 and let @xmath441 then , we have that @xmath442 finally , define the operations , @xmath443 $ ] and @xmath444 $ ] by @xmath445:={{\int\negthickspace\negthickspace\negthickspace\negthinspace -}}_{b(0,a)}f(y)dy\quad \text { and } \quad [ f , g ] : = { { \int\negthickspace\negthickspace\negthickspace\negthinspace -}}_{b(0,a)}{\langle}f(y),g(y){\rangle}dy.\ ] ] then , we have that @xmath446_t&=&h^2[\delta v]+\mu [ v]+[v^3]-h^2\left[\delta \chi , u\right ] -2h^2\left [ \nabla\chi , \nabla u\right ] + \left [ \chi -\chi ^3,u^3\right ] -\left [ v(y'),v\right ] { \nonumber\\}&\geq & \mu [ v]+[v^3]+h^2\left [ \delta \chi , u\right ] + \left [ \chi -\chi ^3,u^3\right]-\left [ v(y'),v\right ] \label{eqn : intpartsnd}\end{aligned}\ ] ] here , ( [ eqn : intpartsnd ] ) follows from integration by parts , the fact that @xmath447 at @xmath448 and that @xmath449 we will later need that @xmath450\geq [ v]^3 $ ] . to see this use hlder s inequality with @xmath451 , @xmath452 , @xmath453 , and @xmath454 to obtain @xmath455=\int_{b(0,a)}\frac{v^3c_d}{a^d}d\xi\geq \left(\int_{b(0,a)}\frac{vc_d}{a^d}d\xi \right)^3=[v]^3.\ ] ] next , we will estimate @xmath456 $ ] . observe that @xmath457 = \left|{{\int\negthickspace\negthickspace\negthickspace\negthinspace -}}\delta \chi u\right|&\leq & \left|{\frac{1}{c_d}}\int_0^a\left(\frac{2a^{-1}(d-1)}{r}\phi ' ( 2a^{-1}r ) + 4a^{-2}\phi ' ' ( 2a^{-1}r)\right ) \int_{\partial b(0,r ) } u(r,\phi ) dsdr\right| { \nonumber\\}&\leq & c\int_0^a\left(\phi^{1/2}+\phi^{1/3}\right)\int_{\partial b(0,a)}u(r,\phi)dsdr\label{eqn : chigreat0}\\ & \leq & c{{\int\negthickspace\negthickspace\negthickspace\negthinspace -}}\chi ^{1/3}u\leq c{{\int\negthickspace\negthickspace\negthickspace\negthinspace -}}(1+\chi u^3)\leq c'+c{{\int\negthickspace\negthickspace\negthickspace\negthinspace -}}\chi u^3\label{eqn : cubicnd}\end{aligned}\ ] ] where @xmath458 and @xmath194 do not depend on @xmath179 . here
( [ eqn : chigreat0 ] ) follows from the fact that for any function @xmath459 , @xmath460 and that @xmath461 near @xmath462 .
( [ eqn : cubicnd ] ) follows from @xmath463 .
now , we have @xmath446_t&=&\mu [ v]+[v^3]+h^2\left [ \delta \chi , u\right ] + \left [ \chi -\chi ^3,u^3\right]-\left [ v(y'),v\right]{\nonumber\\}&\geq & \mu [ v]+[v^3]-o(h^2)+\left [ ( 1- o(h^2))\chi -\chi^3,u^3\right ] -\left [ v(y'),v\right]{\nonumber\\}&\geq & \mu [ v]+[v^3]-o(h^2)-o(h^2)[v^3 ] -\left [ v(y'),v\right]\label{eqn : chicubesubnd}\\ & \geq&\mu [ v ] + ( 1-o(h^2))[v]^3-o(h^2)-\left [ v(y'),v\right]\label{eqn : holder}\end{aligned}\ ] ] here , ( [ eqn : chicubesubnd ] ) follows from the fact that @xmath464 and ( [ eqn : holder ] ) follows for @xmath465 since @xmath450\geq [ v]^3 $ ] .
now , on @xmath466 , we have @xmath467 .
thus , for @xmath372 , @xmath468_t\geq \frac{\mu}{2}[v ] + ( 1-o(h^2))[v]^3-o(h^2).\ ] ] we have that @xmath469(t_1)\geq 1/4 $ ] and @xmath20 .
therefore there exists @xmath177 such that , for @xmath179 small enough and @xmath470 , @xmath468_t\geq \frac{\mu}{4}[v]+{\frac{1}{2}}[v]^3.\ ] ] but , the solution to this equation with initial data @xmath469(0)\geq 1/4 $ ] blows up in time @xmath297 .
hence , so long as @xmath471 and @xmath179 is small enough , @xmath469 $ ] blows up in time @xmath472 .
observe that since @xmath428 , @xmath473 for @xmath179 small enough .
thus , the solution to [ eqn : main ] blows up in time @xmath298 .
* remark . *
a similar result holds for polynomially small data with blow up in time @xmath418 .
we expect that the instability of ( [ eqn : main ] ) is related to the presence of pseudospectrum in the right half plane .
in fact , using numerical simulations for ( [ eqn : linprob ] ) based on code from @xcite with @xmath474 as in ( [ eqn : l1d ] ) , ( see figure [ f:4 ] ) we are able to demonstrate that the the solution with a quasimode for a positive eigenvalue as initial data closely approximates an exponential . based on these results
we expect that a proof of blow - up using quasimodes will allow the results of theorem [ thm : main ] to be extended to complex energies and wider classes of operators .
evans and m. zworski .
_ semiclassical analysis , _ graduate studies in mathematics , ams , to appear .
a. kassam and l. n. trefethen , _ fourth - order time stepping for stiff pdes _ , siam j. sci .
comput . 26
( 2005 ) , 1214 - 1233 . | we consider a nonlinear evolution problem with an asymptotic parameter and construct examples in which the linearized operator has spectrum uniformly bounded away from @xmath0 ( that is , the problem is spectrally stable ) , yet the nonlinear evolution blows up in short times for arbitrarily small initial data .
we interpret the results in terms of semiclassical pseudospectrum of the linearized operator : despite having the spectrum in @xmath1 , the resolvent of the linearized operator grows very quickly in parts of the region @xmath2 .
we also illustrate the results numerically . |
the efar collaboration ( wegner et al . 1996 ; paper i ) has collected photometric ( saglia et al .
1997a , 1997b ; paper iii and iv ) and spectroscopic ( wegner et al .
1999 ; paper ii ) data for galaxies in clusters with the primary goal of using the tight correlations between the global properties of early - types ( the fundamental plane , hereafter fp , and the relation , djorgovski & davis 1987 , dressler et al .
1987 ) to measure the peculiar motions and the mass distribution on large scales .
colless et al .
( 2000 , paper vii ) discuss the results of these investigations and the interpretation of the peculiar motions in the context of cosmological models .
this paper presents the methodology adopted to derive these results and the monte carlo simulations performed to test them and estimate the uncertainties and residual systematic biases .
colless et al .
( 1999 , paper v ) discuss the implications of the efar relation .
an abundant literature is dedicated to the problem of how to best determine regression lines from incomplete datasets subject to errors and explicit selection criteria .
regression algorithms involving two variables are discussed by isobe et al .
( 1990 ) and feigelson & babu ( 1992 ) , who provide least - squares fits and their uncertainties .
akritas & bershady ( 1996 ) address the problem of linear regressions with errors and intrinsic scatter , by considering the variance of the variables involved .
la barbera , busarello and capaccioli ( 2000 ) extend this work to the analysis of the fp equation .
the canonical methods used to find the coefficients of the fp ( or of correlations between two variables ) are of the least - squares type : one minimizes the orthogonal ( absolute or squared ) residuals from the plane ( jrgensen et al . 1996 , jrgensen 1997 , pahre et al .
1997 , scodeggio et al . 1997 ) or the ( squared ) residuals from one of the variables ( hudson et al .
principal component analysis has also been used ( carvalho & djorgovski 1992 ) .
ample discussion of problems related to selection is given by willick ( 1994 ) and teerikorpi ( 1997 ) , in the context of the determination of peculiar velocity fields .
however , none of these approaches is able to deal effectively with the multiple problems that the efar and similar datasets pose : ( i ) the factor two in redshift spanned , ( ii ) the strong selection effects , ( iii ) the non - negligible and widely varying measurement errors , ( iv ) the presence of defined selection criteria , ( v ) the intrinsic scatter of the relation . in this paper we demonstrate that `` classical '' methods fail when confronted with the efar dataset , and we quantify the biases they produce .
we construct an algorithm based on maximum likelihood ( ml ) gaussian modelling and test it with monte carlo simulations .
we demonstrate that it solves the problems posed by the dataset , producing nearly unbiased estimates of the parameters involved .
the paper is organized as follows . [ mlalgo ] discusses critically the results of linear regression analysis and presents the ml algorithm for the one- , two- and three - dimensional cases .
[ montecarlo ] describes the method to generate mock catalogues of the efar database , the tests of the `` canonical '' least - square methods and of the ml gaussian algorithm , and the estimates of the errors on the derived parameters .
simulations backing the results presented in papers ii , v and vii are discussed for the , , relations , the fundamental plane and the cluster peculiar velocities .
conclusions are drawn in [ conclusions ] .
the appendices give the analytical solution of the n - dimensional gaussian maximum likelihood problem in the presence of small errors and some results used in the paper .
after reviewing the results of the standard least - squares analysis in [ linear ] , we describe the properties of the efar database in [ efarsample ] to argue that a ml approach is needed to study the sample .
we set the equations of the problem for the general case in [ general ] and address the simplified cases when one ( [ onedim ] ) , two ( [ twodim ] ) or three ( [ threedim ] ) variables are considered . before discussing the specific problems related to the efar database and the solutions we have developed to address them
, it is worth considering the more general question of how to model the correlations existing between the structural parameters of galaxies . as usual in astrophysics ,
logarithmic quantities are considered in order to avoid scaling problems . in the simplest approach ,
one focuses on pairs of datapoints @xmath0 , looking for a linear relation @xmath1 between them .
table 1 of isobe et al .
( 1990 ) ( see also eqs . [ eqyx]-[eqao ] below ) summarizes the estimates of the slopes @xmath2 and zeropoints @xmath3 , and their variances , derived using five methods ( y - x and x - y regressions , bisector , reduced major axis and orthogonal regression ) , when measurement errors are negligible .
briefly , we recall that the y - x regression derives the slope @xmath4 and zero point @xmath5 of the line @xmath6 by minimizing the quantity : @xmath7 and the x - y regression by minimizing the analogous @xmath8 .
the bisector solution gives the slope and zeropoint of the line passing in between the y - x and x - y lines .
the reduced major axis slope is the geometric mean of the y - x and x - y slopes .
the orthogonal regression derives @xmath9 and @xmath10 by minimizing the orthogonal residuals from the line : @xmath11 if the differences in slopes and zeropoints between the methods are smaller than the expected variance , a well - defined answer is found .
unfortunately , this is rarely the case in astronomy .
therefore , isobe et al . (
1990 ) recommend astronomers use the y - x regression when it is clear that y is the variable to be `` predicted '' , and the bisector method when the functional relation between the variables has to be investigated .
they discourage the use of the orthogonal regression , since its slope has greater dispersion .
the problem is complicated when errors ( on one or both variables ) are present . in particular , feigelson and babu ( 1992 )
conclude that `` models incorporating both measurement errors and intrinsic scatter are complex and not yet fully developed '' .
akritas and bershady ( 1996 ) improve on this aspect , presenting an updated version of table 1 of isobe et al .
( 1990 ) , where corrections for measurement errors are taken into account .
isobe et al . (
1990 ) , feigelson and babu ( 1992 ) and akritas and bershady ( 1996 ) stress that `` there is no such thing as true slope '' , and that the regression method should be chosen according to the problem to be solved .
these considerations become more transparent when the underlying bivariate probability distributions @xmath12 are examined . assuming normality , @xmath12
can always be written as : @xmath13 \right\ } , \ ] ] where @xmath14 and @xmath15 are the mean values of @xmath16 and @xmath17 , @xmath18 and @xmath19 the standard deviation ( rms scatter ) , and @xmath20 the correlation coefficient defined as @xmath21 .
calling @xmath22 and @xmath23 , eq .
[ eqbivariate ] can be cast as : @xmath24 \right\ } , \ ] ] where @xmath25 , @xmath26^{-1}$ ] and @xmath27 is the covariance matrix : @xmath28 eqs .
[ eqbivariate ] and [ eqbivlambda ] are equivalent to the three following formulae : @xmath29 , \ ] ] @xmath30 , \ ] ] @xmath31
\frac{1}{\sqrt{2\pi}\sigma_2}\exp\left[-\frac{(a_o\hat y+\hat x)^2}{2(1+a_o^2)\sigma_2 ^ 2}\right ] , \ ] ] where : @xmath32 , \ ] ] @xmath33 @xmath34 eq . [ eqyx ] shows that a gaussian bivariate probability distribution can be generated extracting first the @xmath16 variable around its mean with rms @xmath18 , then the residual @xmath35 around the @xmath36 line with slope equal to the regression of @xmath37 on @xmath38 and rms @xmath39 .
similarly , eq . [ eqxy ] shows that the same distribution can be generated first extracting the @xmath17 variable and then the residual @xmath40 around the @xmath36 line with slope equal to the regression of @xmath38 on @xmath37 .
finally , eq . [ eqortho ] generates the distribution extracting the residuals around the eigenvectors of the covariance matrix .
the direction of the eigenvector with the largest eigenvalue is the slope of the orthogonal regression .
the zero points @xmath3 of the relevant linear relations of slope @xmath2 are obtained from the mean values as : @xmath41 do we therefore conclude , as akritas and bershady ( 1996 ) , that `` there is no such thing as true slope '' ?
in fact , the key of the problem is the covariance matrix , and therefore the natural slope is the direction of the principal axes ( given by eq . [ eqao ] ) with the largest eigenvalue ( given by eq .
[ eqsigma1 ] ; eq .
[ eqsigma2 ] is the second eigenvalue of the covariance matrix ) . as a consequence ,
the orthogonal regression determines the `` true slope '' , not the bisector , which isobe et al .
( 1990 ) prefer for its lower variance . however , what about distance determination problems , where eq .
[ eqyx ] should be preferred according to akritas and bershady , if @xmath17 is the distance - dependent quantity ? indeed , if one seeks for the distance shift @xmath42 of the datapoint @xmath43 , the most probable value when eq .
[ eqbivariate ] is given is : @xmath44 which involves the @xmath36 regression coefficient @xmath45 .
further complications arise when the datapoints @xmath0 are affected by errors comparable to the rms spreads @xmath18 and @xmath19 , or when only a subset of the possible data volume is available due to incompleteness .
these affect the estimation of the covariance matrix @xmath27 and the mean values of the variables .
for example , we estimate in appendix [ twodimlim ] that the orthogonal regression solution ignoring the errors underestimates the true slope if the error in the @xmath38 direction is larger than the one in @xmath37 .
these effects might be included in the regression approach by careful treatement ( see references given in feigelson and babu 1992 ) , but it is clear from the discussion above that a maximum likelihood approach is superior , allowing the natural and simultaneous solution of all these difficulties .
the efar database comprises photometric and spectroscopic information for a set of cluster galaxies spanning a factor of two in redshift , with non - negligible measurement errors and selection effects .
the size of the errors can be quantified comparing the rms spread of the variables with the mean of the errors . in this sense
photometric errors ( @xmath46 mag , see paper iii and iv ) are small in the efar database .
unfortunately , this is not the case for the spectroscopic data ( see paper ii ) . for the case of the velocity dispersions
@xmath47 , we find that the mean value ( 25 km / s ) of the measurement errors on @xmath47 is 36% of the rms spread of the values ( 71 km / s ) . for the mg@xmath3 index , the mean value ( 0.39 ) of the measurement errors is 48% of the rms spread of the values ( 0.83 ) .
in addition , we can not measure central velocity dispersions smaller than 100 km / s for lack of spectral resolution .
selection effects are also severe .
the galaxy sample is more than 50% complete only for galaxies with @xmath48 kpc ( see paper i ) , with 46% of the galaxies having selection probabilities larger than 50% .
the efar galaxy selection @xmath49 is described in paper i and is a function of the @xmath50 diameter measured in arcsec for galaxy @xmath51 : @xmath52\right\ } , \ ] ] where @xmath53 is the cluster index , @xmath54 is the midpoint and @xmath55 the width of the cutoff in the selection function .
the diameter @xmath50 correlates with the @xmath56 diameter of the sample as @xmath57 with 0.09 dex scatter in @xmath56 ( see paper iii ) .
therefore we compute the selection probabilities @xmath58 from the @xmath59 diameters derived from the very accurate @xmath56 instead of using the values of @xmath50 given in paper i. in the following we shall indicate with @xmath60 the selection probability for the galaxy @xmath51 , dropping the cluster index , and with @xmath61 the selection weight . due to the large spread in redshift , the efar clusters have been sampled down to different limiting @xmath62(kpc ) .
however , 95% of the early - types of the sample have @xmath63 kpc , so that nearly unbiased estimates of the cluster peculiar velocities can be obtained just by dropping from the sample objects with @xmath64 . to our knowledge ,
no algorithm based on linear regressions has been developed to model all these features at the same time . in the following
we describe a parametric algorithm based on gaussian modelling able with to deal with ( i ) a spread of selection weights , ( ii ) sizeable measurement errors with a large spread , ( iii ) possible explicit selection limits . in the general case , we have n ( logarithmic ) data @xmath65 for each galaxy @xmath51 .
we indicate with @xmath66 the related @xmath67 error matrix .
if the errors are uncorrelated , @xmath66 is a diagonal matrix with @xmath68 , where @xmath69 if @xmath70 and 0 otherwise .
each galaxy has been selected according to the value of a selection diameter , which possibly could be expressed as a linear combination of a subsample of @xmath71 , giving a selection weight @xmath61 .
one variable @xmath72 may be distance dependent .
we want to estimate the vector of the mean values of the data @xmath73 , the covariance matrix @xmath27 and , possibly , the vector @xmath74 of the m cluster peculiar velocity shifts . in this case
we need an additional constraint to fix @xmath75 , such as @xmath76 .
other options are discussed in paper vii .
the direction of the eigenvector with the smallest eigenvalue of the matrix v defines the minimum variance hyperplane describing the data distribution , or the slope of the linear correlation in the two dimensional case .
we solve the problem assuming that both the distributions of the galaxy parameters and of their errors are normal , with covariance matrix @xmath27 and the error matrices @xmath66 respectively . the first assumption can be verified _ a posteriori _ , estimating the likelihood of the best - fitting gaussian model ( see figs .
[ figmg2sig ] , [ figalllike ] and [ alllike ] ) .
the second is based on the error analysis performed in papers ii and iii for the efar database .
once the error convolution is performed , the probability density of the vector @xmath77 is : @xmath78
\hat{\theta}(a\hat{\vec x_i}-\vec{x}_{cut})\ ] ] where @xmath79 . the function @xmath80 , where @xmath81 if @xmath82 and 0 otherwise , takes into account that parts of the parameter space might not be accessible because of selection effects or explicit cuts .
for simplicity we assume that these cuts are applied to linear combinations of the variables , described by the appropriate matrix @xmath83 .
the normalization factor @xmath84 is such that @xmath85 .
following eadie et al .
( 1971 ) , we write the likelihood of the observed sample as : @xmath86 counting @xmath61 times galaxies with selection probability @xmath60 we determine @xmath87 , @xmath27 and , possibly , @xmath88 by minimizing @xmath89 . in the following sections we describe how this general scheme is implemented in the different cases of interest . in all cases
the minimization is performed numerically , using the simplex algorithm ( press et al .
appendix [ derivation ] discusses the analytical solution of eq .
[ eqlikelihood ] for small errors , no explicit cuts and no peculiar velocities .
the first simple application of the scheme described above is the determination of the mean value @xmath90 and the ( intrinsic ) rms @xmath47 of a set of data @xmath91 with errors @xmath92 , selection weights @xmath93 and subject to the censoring @xmath94 .
mean values and intrinsic spread of the efar photometric ( logre , , , @xmath95 ) or spectroscopic ( , , ) parameters , or of the measured cluster peculiar velocities can be estimated in this way .
the simplified version of eq . [ eqlikelihood ] reads now : @xmath96\theta(x_i - x_{cut})\right)^{w_i},\end{aligned}\ ] ] errors are taken into account by convolving the intrinsic distribution with the appropriate gaussian error distribution .
selection effects are taken into account in eq .
[ eqonedim ] by counting @xmath97 times galaxies with selection probability @xmath60 .
the presence of explicit cuts @xmath98 is taken into account ( for mild cases ; see below ) by the normalization correction : @xmath99 we expect our approach to work in the presence of _ mild _ cuts with @xmath100 , so that the cut is well away from the mean value of the distribution ( see [ onedimtest ] ) .
appendix [ onedimlim ] gives the analytical solution of the problem for the limiting case of small errors .
we turn now to the two - dimensional case , which allows us to investigate correlations between pairs of variables , such as examined in paper ii and the and discussed in paper v. other relations that we shall not explicitly consider here are the @xmath101 ( where @xmath102 with @xmath103 ) , the @xmath104 , @xmath105 and @xmath106 relations , the `` kormendy '' relation @xmath107 , or the - relation .
note that the efar database measures the photometric parameters in the r band .
[ eqprobndim ] reads in this case : @xmath108 where the normalization factor @xmath109 is @xmath110 here @xmath111 and @xmath112 is the bivariate probability integral ( abramovitz and stegun 1971 ) with @xmath113 , @xmath114 and @xmath115 .
we set @xmath116 , @xmath117 ( and therefore @xmath118 ) when considering the the relation , @xmath116 , @xmath119 , when studying the and relations , and @xmath120 , @xmath119 , when examining the @xmath121 relation . in this case
we derive using its relation with @xmath122 : @xmath123 ( see paper iii ) .
finally , note that in every case we compute @xmath124 using the simplifying assumption that the error matrix is diagonal with diagonal terms given by the estimated total errors .
for the spectroscopic data these involve two terms , the uncorrelated statistical errors and the correlated errors coming from run - to - run corrections ( see paper ii ) .
appendix [ twodimlim ] gives the analytical solution for the limiting case of small errors , no peculiar velocities and no cuts .
the three - dimensional case allows us to study the efar fundamental plane .
as is well known , early - type galaxies do not fill the three - dimensional space defined by the coordinates @xmath125 , but rather occupy a narrow region around the fundamental plane defined by the equation : @xmath126 following [ general ] , we determine the mean values , and , the covariance matrix @xmath27 and the cluster peculiar velocity shifts @xmath127 ( subject to the constraint @xmath128 , see [ general ] ) by maximizing the likelihood @xmath129 : @xmath130,\nonumber\end{aligned}\ ] ] where @xmath131 is the fraction of galaxies with @xmath132 and @xmath133 . the integral ( eq .
[ eqnorm ] ) is performed in appendix [ norm3dim ] .
the offset @xmath127 between true mean galaxy size , and the mean galaxy size observed for cluster @xmath53 , @xmath134 , is related to the peculiar velocity of the cluster . in particular , the ratio of the true angular diameter distance of a cluster , @xmath135 , to the angular diameter distance corresponding to its redshift , @xmath136 , is : @xmath137 the relation between angular diameter distance and redshift ( weinberg 1972 ) is given by @xmath138 we assume @xmath139=50kms@xmath140mpc , @xmath141=0.5 , and compute all redshifts and peculiar velocities in the cmb frame of reference .
the peculiar velocity of the cluster , @xmath142 , is then obtained as @xmath143 where @xmath144 is the redshift corresponding to the true distance @xmath135 through the inverse of eq .
[ eqn : daz ] .
the error matrix @xmath66 is given by : @xmath145 where @xmath146 is the error on @xmath147 , @xmath148 is the error on the combined quantity @xmath149 , with @xmath150 , and @xmath151 is the photometric zero - point error ( see paper iii and iv ) .
note that in principle @xmath3 might be different from @xmath152 .
the quantity @xmath153 is the error on the effective surface brightness coming from the fitting .
see appendix [ photerrmatrix ] for a derivation of these formulae .
the error @xmath154 on combines the uncorrelated and correlated errors ( see paper ii ) .
it is worth noting that the eigenvectors @xmath155 , @xmath156 , @xmath157 of the v matrix can be written to a close approximation as a function of the parameters @xmath2 and @xmath3 defining the fundamental plane of eq .
[ eqfp ] : @xmath158 where @xmath159 measures the displacement from the fp , and @xmath156 and @xmath157 are two orthogonal vectors in the fp .
the absence of a @xmath160 component in @xmath161 is remarkable . in paper vii ( eq .
12 ) we fit for the coefficient of in the second principal component and find that it is small . in
[ secvariant ] ( eq .
[ eqthird ] ) we show that it is compatible with zero within the statistical uncertainties . in the following
we indicate the rms spread around the vectors @xmath159 , @xmath161 and @xmath162 with @xmath163 , @xmath164 , @xmath165 , respectively .
the algorithms described in the previous section have been extensively tested on mock catalogues of the efar database . after describing how these catalogues are generated ( [ mock ] ) , we proceed to test cases of increasing complexity .
we start with the one - dimensional @xmath47 distribution ( [ onedimtest ] ) , as an example of the influence of hard cuts on the mean and rms of the sample .
then we consider the relation ( [ mg2mgbp ] ) , where distance - independent quantities not subject to explicit cuts are involved .
we turn to the relations ( [ mgsigma ] ) , where no datapoints with central velocity dispersions smaller than the resolution limit can be present . finally , we examine the three - dimensional case of the fundamental plane ( [ fp ] ) , where one distance - dependent quantity ( ) is correlated against two distance - independent quantities ( and ) , with distance - dependent selection limits and the presence of cluster peculiar velocities .
we generate mock catalogues of the efar database using the gaussian distribution functions that are reconstructed from the data ( see paper ii , paper v , paper vii and case 0 of table [ tabsimumean ] ) .
we specify the values of , , , , , the coefficients of the fp and the slopes of the and relations , and the dispersions of the distributions in the orthogonal directions .
we generate one entry in the mock catalogue for each early - type galaxy present in the database . as a first step
we extract the triplet ( , , ) , using eq .
[ eqrot ] to specify the directions of the principal components of the distribution in the ( , , ) space .
we determine the corresponding value of the @xmath166 diameter using the relation ( a second order approximation to the equation 4 of van albada , bertin and stiavelli 1993 ) : @xmath167 where @xmath168 matches the surface brightness used to determine the efar @xmath166 diameters ( see paper iii ) .
only galaxies with a dominant exponential component in the luminosity profile deviate strongly from eq .
( [ eqdn ] ) .
we simulate the effects of cluster peculiar velocities by adding cluster - dependent shifts to the extracted values of and .
we convert into @xmath169 using the relation @xmath170 ( see paper iii ) and add the 0.09 dex rms random scatter of the relation .
we compute the corresponding selection probability , according to the selection parameters of the cluster being extracted and its ( redshift ) distance and test if the extracted datapoints should be in the sample .
if this is the case , the values of and are extracted following the ( two - dimensional ) gaussian distribution computed at @xmath171 .
measurement errors are assigned following the efar catalogue , so extracted
object n has the total errors of object n of the database , taking into account the correlated terms of the total errors ( see paper ii ) properly .
therefore , the error on @xmath47 , and for the extracted object n is computed considering the spectroscopic runs where object n of the database was observed , generating proper run correction errors and determining the final error taking into account the run weights .
finally , three additional sets of simulations ( cases 45 , 46 and 47 of table [ tabsimumean ] ) are generated using uniform instead of gaussian distributions , and zero input peculiar velocity fields .
following the procedure described above , we consider the values of , , , @xmath172 and of the coefficients of the fp as case 0 of table [ tabsimumean ] , and eq .
[ eqrot ] to specify the directions of the principal components of the distribution in the ( , , ) space . for case 45
we add gaussian errors generated as above , but we assume a uniform distribution of the fp parameters , extending @xmath173 , @xmath174 , @xmath175 in the @xmath176 directions respectively from the mean values .
this choice preserves the covariance matrix of the gaussian case . for case 46
we use gaussian distributions of the fp parameters , but measurements errors drawn using a uniform distribution extending @xmath177 the estimated rms . for case 47
we use uniform distributions for both the fp parameters and the errors , with the @xmath178 scaling as above . as a first example we discuss the use of the 1 dimensional algorithm of [ onedim ] .
we generated 99 samples of the efar database ( see paper v ) and considered the @xmath47 distribution .
[ figmlone ] shows the results as a function of the cut applied to the data . for cuts well below the true mean of the distribution
, the ml algorithm with or without normalization correction estimates the mean and rms of the distribution with small bias .
the simple mean , which does not take into account the selection weights of the data , slightly overestimates the true value , because objects with small @xmath47 tend to have small selection probabilities .
the simple rms , corrected for measurement errors as suggested by arkitas and bershadi ( 1996 ) , underestimates slightly the true values , not taking into account the selection weights .
as the cut is increased to values similar to the true mean of the distribution , the ml algorithm with normalization correction estimates the mean and rms with small bias , with the statistical errors ( for the single sample @xmath179 larger than the error bars shown ) becoming increasingly large . with cuts one sigma or more larger than the mean ,
the ml algorithm starts to fail , biasing the mean low and the sigma high .
however , the bias is always smaller than the random errors .
both the ml algorithm without normalization correction and the simple estimates bias the mean to larger values and the rms to smaller values .
= as it is well known ( press et al . 1986 ) , the likelihood analysis offers in principle a simple recipe to compute confidence intervals for the fitted parameters . the @xmath180 one - dimensional confidence interval is fixed by the contours of the likelihood function where : @xmath181 figure [ figlikeone ] shows four contours of constant likelihood in the @xmath182 plane for one of the simulations of figure [ figmlone ] analyzed by the ml algorithm with normalization correction @xmath183 km / s respectively .
the thick errorbars mark the position of @xmath90 and @xmath47 , averaged over the 99 simulations of fig .
[ figmlone ] , and show their rms .
a straight application of eq . [ eqlikeone ] would produce @xmath184 one - dimensional confidence intervals for @xmath90 and @xmath47 a factor @xmath185 smaller than the rms measured by the simulations .
inspection of eq .
[ eqlikelihood ] reveals that since the mean weighting of each galaxy is @xmath186 , the @xmath180 one - dimensional confidence interval is reached at a @xmath187 distance from the maximum likelihood value .
this factor is taken into account in fig .
[ figmlone ] , where the scaled @xmath184 one - dimensional confidence intervals match approximately the rms measured by the simulations .
the two - dimensional @xmath188 confidence region , set by @xmath189 , is slightly larger .
however , fig .
[ figalllike ] shows that the mean weight fluctuates in the range @xmath190 , making the estimate of confidence intervals from the likelihood function uncertain for a given dataset .
in addition , this analysis does not allow to study the small residual biases in the ml estimates .
therefore , in the following we will estimate confidence intervals and residual biases using monte carlo simulations only .
= the simplest test of the 2-dimensional algorithm is performed considering the relation .
we generated 99 samples without taking into account the efar selection function .
the input parameters are the ones derived in paper ii .
two cases are examined , one with a realistic error distribution and one with a factor 5 smaller errors .
figure [ figmg2mgb](d ) shows one of the realizations with realistic errors .
figure [ figmg2mgb](g ) shows the results averaged over the 99 simulations . in both the cases of small and realistic errors
the ml algorithm retrieves the input parameters with small biases and with small errors .
the @xmath36 , @xmath191 and bisector regressions fail to derive the correct slopes and zeropoints , because of the finite orthogonal scatter and the non - negligible errors .
when the method of akritas and bershady ( 1996 ) is used to correct for the errors ( dotted cross ) , the @xmath36 result is less biased , but with large random errors .
note that the method can not be used when selection weights are present .
the orthogonal residual fit gives a slightly biased answer as expected from eq .
[ eqafirst ] , with a nearly perfect result when the errors are small . as expected from eq .
[ eqzeropoint ] , the errors on @xmath2 and @xmath3 are perfectly anticorrelated . figs .
[ figmg2mgb](a ) , ( b ) , ( c ) , ( e ) and ( f ) show the histograms of the results of the ml parameters ( for the case of realistic errors )
. not only are the slope and the mean values determined with high precision , but so also is the intrinsic scatter ( orthogonal and parallel to the relation ) , despite the non - negligible measurement errors ( the observed scatter orthogonal to the relation is more than a factor two larger than the intrinsic one ) .
= the tests on the and relations were performed using the efar mock catalogues with realistic error distributions .
the input parameters are the ones derived in paper v. figure [ figmg2mgbsig](a ) shows one of these realizations .
data at small @xmath47 or mg@xmath192 have larger selection weights .
[ figmg2mgbsig](b ) shows that the spread in the errors is large .
the distribution of the selection probabilities ( fig .
[ figmg2mgbsig](c ) ) has a peak at the completeness value ( @xmath193 ) with a prominent tail down to the selection probability cut ( @xmath194 ) .
the effective number of datapoints per selection weight bin ( fig .
[ figmg2mgbsig](d ) ) is approximately constant .
the simulations ( full lines ) reproduce closely the observed dataset ( dotted lines ) analyzed in paper v. = fig .
[ figmg2mgbsig](e ) shows the slopes and zero - points derived from the various regressions and the ml algorithm , considering points with selection probabilities larger than 0.1 for the relation . as already seen in [ mg2mgbp ]
, the regressions give biased results due to the errors ( as discussed in [ linear ] ) , while the ml algorithm is nearly free of biases and accurate .
figure [ figmg2sig ] illustrates the effects of the selection weighting and of the cut in @xmath47 .
when the analysis is performed including all the available datapoints to the largest selection weights and without any likelihood cut ( open circle ) , the parallel and orthogonal scatter are derived best . however , the rms on all the parameters are rather large , because the presence of some datapoints with very high weights can bias the analysis of some realizations .
it is therefore prudent to limit the analysis to datapoints with weights not larger than 10 ( e.g. , @xmath195 ) and clip points with low likelihood ( @xmath196 , triangular stars ) .
this allows us to determine the slope and the zero - point of the relation with small biases and with small variance .
the price is a slight underestimate of the orthogonal and parallel spread , @xmath197 and @xmath198 .
fitting points with weights not larger than 5 ( filled triangles ) and the same likelihood clipping , one observes that the mean quantities and are biased to systematically larger values , because the small galaxies have on average larger selection weights ( see fig .
[ figmg2sig ] ) .
the largest bias in the mean quantities is obtained weighting all datapoints equally , independently of the selection ( open square ) .
these tests justify the procedure adopted in paper v to study the and relations , where we consider galaxies with @xmath195 and apply likelihood clipping .
= to illustrate the effect of the cut in @xmath47 , we examine the extreme case where only galaxies with @xmath199 km / s are considered in the analysis . when no correction is applied (
filled square ) , the slope and the zeropoint of the relation are slightly biased , the mean quantities and are biased to larger values , while the parallel spread is biased low ( similar to what is seen in figure [ figmlone ] ) .
the biases are fully corrected for when we apply the normalization corrections appropriate for @xmath200 km / s ( open star ) , with a slight increase of the variance .
the cut applied to the efar database ( @xmath201 km / s ) is in any case low compared to the mean value of the sample .
finally , plot ( d ) of figure [ figmg2sig ] shows the mean values and rms ( over the simulations ) of the mean ( over the datapoints ) logarithmic likelihood and selection probability .
the cross shows the values derived for the efar sample examined in paper v , with errorbars equal to their rms divided by the square root of number of the galaxies considered .
the simulations match the mean likelihood of the efar sample within the estimated errors , justifying _ a posteriori _ the assumption of a gaussian distribution .
the mean selection probability of the simulations ( see figure [ figmg2sig ] ) is only slightly larger than the ones of the real sample .
similar results are obtained when considering the relation .
figure [ figmg2mgbsig](f ) shows again that the linear regressions give biased results .
the @xmath36 and orthogonal fitting are very similar , because of the shallow slope .
the ml with probability cut @xmath195 and likelihood clipping gives nearly unbiased results .
histograms similar to fig .
[ figmg2mgb](a)-(f ) for the and relations have been already presented in paper v. in this section we describe the tests of the ml algorithm used to derive the fp solution of paper vii and the peculiar velocities of the efar sample . in [ metcomp ] we assess the superiority of the ml algorithm with respect to the linear regression methods , when the @xmath2 and @xmath3 parameters of the fp are determined . in [ secvariant ] we study the precision of the fp parameters derived using the ml algorithm , justifying the strategy adopted in paper vii . [ alllike ] justifies a posteriori the assumption of a gaussian distribution of galaxies in the ( , , ) space .
[ resbias ] tests the bias correction scheme for the peculiar velocities adopted in paper vii . finally , [ meanmotions ] investigates how well coherent and random motions can be measured from the efar data sample . as we have done above , we start comparing the performances of the `` classical approaches '' ( linear regressions and orthogonal fitting ) and the ml algorithm , focusing on mock catalogues generated using parameters close to those derived for the efar data sample ( see paper vii and case 0 of table [ tabsimumean ] ) .
we consider the 29 best clusters of the efar sample ( see paper vii ) and we do not add peculiar velocities ( i.e. , the input peculiar velocity field is zero , @xmath202 ) . in each of the four methods considered we solve simultaneously for the coefficients @xmath2 and @xmath3 ( see eq .
[ eqfp ] ) and the vector @xmath127 .
the shifts @xmath127 are determined by assuming that the mean of @xmath127 over the 29 clusters is zero , which fixes the value of the parameter @xmath203 . objects with @xmath204 km / s , or with selection diameter @xmath205 less than 12.6 kpc ( i.e. , @xmath206 ) , or with selection probabilities less than 0.1 , are excluded from the fits . therefore
, this procedure mimics closely the `` fiducial '' solution of paper vii ( `` case 1 '' of table 4 ) .
[ figfp ] shows the mean values and rms ( over 99 simulations ) of the fundamental plane coefficients @xmath2 and @xmath3 as determined using regressions on , , , orthogonal fitting and ml .
as can be expected from the previous discussion , linear regressions and orthogonal fitting are inadequate to determine the coefficients of the fundamental plane .
all five methods determine @xmath3 quite accurately , with the and regressions giving the largest systematic deviations ( @xmath207 ) .
only the ml algorithm , however , is able to estimate @xmath2 with a bias smaller than the expected statistical uncertainty .
as expected from appendix [ twodimlim ] , the and regressions and ( less severely ) the orthogonal fitting underestimate @xmath2 , while the regression overestimates it . as pointed out by isobe et al .
( 1990 ) , the regression gives the smallest rms .
the ml algorithm determines @xmath2 to 0.09 and @xmath3 to 0.013 statistical accuracy .
= figure [ figpec ] shows how well the five methods measure the input peculiar velocity shifts @xmath127 ( all equal to zero ) of the efar clusters .
similarly to the procedure described in [ resbias ] for the ml algorithm ,
the parameters @xmath2 , @xmath3 and @xmath203 of the fp determined from the regressions on the subset of 29 best clusters for each separate simulation are used to compute the peculiar velocities of the remaining clusters .
let us first focus on the clusters with @xmath208 .
these are the clusters that statistically have at least one galaxy with @xmath209 and selection probability larger than 0.1 . in this case
, the biasing influence of errors and selection effects is small for all the methods .
the absolute values of the mean @xmath127 are typically smaller than 0.02 dex .
the rms over the efar clusters of the mean @xmath127 over the 99 simulations is @xmath210 .
the regression gives slightly poorer results .
the statistical errors on the single cluster measurements are in any case larger , in the range 0.02 - 0.04 dex for the best - populated clusters , and up to 0.15 dex for the poorer populated clusters . the mean over the 99 simulations of the standard deviation ( over the efar clusters ) of the @xmath127 values differs for the various methods .
the regression gives the smallest value ( 0.043 dex , or a typical 10.4% accuracy on the determination of the distance of a single clusters ) , similar to the regression ( 0.045 dex , 10.9% distance error ) .
the ml algorithm and the orthogonal regression perform only slightly worse ( 0.049 dex , or 11.9% distance error ) .
the regression gives the largest ( 0.065 dex , or 16% distance error ) random error .
the ml algorithm ( and in order of increasingly importance the orthogonal , and regressions ) derives @xmath127 biased to low values when @xmath211 .
the clusters where this happens are so distant that their smallest galaxies in the efar sample have @xmath212 .
[ figfpsim ] shows that these are clusters where we sample less than half of the galaxy distribution . from what discussed in [ onedimtest ]
we expect the ml algorithm to progressively fail in this regime . in [ resbias ] we correct for this residual bias due to the difference in the way the fp galaxy distribution is sampled in different clusters through simulations . in any case
, the standard deviations ( over the efar clusters ) of the @xmath127 values do not change , remaining @xmath213 times larger that the typical systematic errors .
therefore we conclude that overall the ml method provides the best solution , keeping low the systematic and random errors on both the fp parameters and the @xmath127 values .
in addition , it is more robust against outliers ( since we can identify and remove the extreme low - likelihood objects , see paper vii ) .
= = in this section we study the precision of the fp parameters derived using the ml algorithm , to justify the strategy adopted in paper vii .
table [ tabsimumean ] summarizes the tests performed , giving the mean values ( over 99 simulations ) of the number of clusters and galaxies ( rounded to integer numbers ) and the parameters of the fundamental plane derived for 28 cases .
table [ tabsimurms ] lists the corresponding rms .
figure [ figparvar ] shows the results graphically .
= case 0 gives the parameters of the fundamental plane as derived in paper vii for the standard fit .
these are the input parameters used to generate the simulations considered here .
case 1 is the standard fit discussed above .
2 of paper vii presents the histograms of 1000 simulations for this case .
the mean values derived from the 99 simulations used here do not differ by more than 1.5% from those derived from the 1000 simulations . as discussed in paper vii , there are small residual biases in the fitted parameters : a is biased low by 6% ( i.e. , since the mean value of @xmath2 recovered from the simulations is smaller than the input value , we infer that the true value of @xmath2 for the efar sample must be larger by @xmath214% than what found in case 0 ) , b is biased low by 2% ; c is biased high by 4% ; , and are all biased high , by 0.036 dex , 0.07 dex and 0.05 mag respectively ; the scatter about the fp is under - estimated by 0.006 dex , or 1.4% ; the widths and of the galaxy distribution in the fp are biased by 0.049 ( low ) and 0.009 ( high ) dex respectively .
these biases are all less than or comparable to the rms width of the distribution , so that although they are statistically significant ( i.e. much greater than the standard error in the mean ) , they do not dominate the random error in the fitted parameters . therefore in paper vii we do not correct for these biases , since they are small and have negligible impact on the derived distances and peculiar velocities ( see [ resbias ] ) .
case 2 is similar to case 1 : we do not add peculiar velocities and we do not fit objects with @xmath204 km / s , or with selection diameter @xmath205 less than 12.6 kpc ( i.e. , @xmath206 ) , or with selection probabilities less than 0.1 .
however , in each simulation the list of fitted clusters is not fixed to the 29 clusters considered in case 1 , rather it is restricted to the clusters that have 6 or more galaxies .
the number of fitted clusters fluctuates from 20 to 29 from simulation to simulation .
the mean fp parameters do not show differences from case 1 .
cases 3 to 8 examine fits performed on the subset of the clusters having 10 or more galaxies ( case 3 ) to clusters having 3 or more galaxies ( case 8) .
the mean fp parameters derived for case 3 are the ones with the smallest biases .
however , only 5 - 6 clusters for @xmath215 galaxies are fit . as a consequence ,
the statistical error on the parameters is nearly a factor two larger than for case 1 . on the contrary ,
case 8 fits a larger number of galaxies and clusters than case 1 , and therefore gives slightly smaller ( @xmath216% ) statistical errors .
however , the mean values of the fp parameters are biased more than for case 1 . in particular , the value of is spuriously small , as offsetting the fp with spurious peculiar velocity suppresses the apparent scatter . we conclude that the list of clusters adopted in paper vii is a reasonable compromise between the need of obtaining bias - free parameters and minimizing the statistical errors .
cases 9 to 12 explore the effects of different .
low give slightly smaller @xmath2 coefficients ; in addition , is too large , since essentially no normalization correction is applied .
however , the statistical errors are smaller , because the sample size is maximized .
in contrast , large give larger @xmath2 coefficients and too small .
in addition , the statistical errors are larger , because the sample size is reduced .
we conclude that again case 1 is the reasonable compromise both in terms of bias and statistical errors .
this choice , however , forces us to apply additional corrections when deriving the peculiar velocities of the whole cluster sample ( see [ resbias ] ) .
case 13 ignores selection probabilities altogether and applies a uniform weight to all galaxies , resulting in an effective over - weighting of the larger galaxies . as a consequence , the mean values of , , and are biased to higher values .
a similar effect , but of reduced amplitude , is observed for case 14 , where galaxies with selection probabilities lower than 0.2 are excluded . on the contrary ,
case 15 includes essentially all galaxies in the fit , irrespective of their selection probabilities .
the biases in the recovered parameters are reduced , but the statistical errors increase , because of the presence of a few ( deviant ) points with large weights . to conclude , weighting the data points with their inverse selection probability , and excluding galaxies with selection probabilities lower than 0.1 is a reasonable compromise that minimizes the biases and the statistical errors of the recovered parameters .
case 16 shows that excluding the ( few ) galaxies with large errors on @xmath47 does not affect the results of the fits , both the mean and rms values of the parameters . case 17 excludes not only the galaxies rejected from the standard fit , but also galaxies with low likelihoods ( @xmath217 ) ; this results in a highly biased fit , with low @xmath2 coefficient , an artificially lowered fp scatter and a substantially narrower distribution in the fp .
case 18 replaces the individual error estimates for all measured quantities with uniform ( average ) errors .
this has little effect on the derived parameters , but the statistical errors are larger .
case 19 allows an extra degree of freedom by permitting the orientation of the major axis of the galaxy distribution within the fp to be fitted , rather than specified a priori .
the vectors @xmath161 and @xmath162 of eq .
[ eqrot ] read now : @xmath218 the only parameters affected are and which are biased high and low respectively .
[ figthird ] shows the histogram of the recovered second and third component of vector @xmath161 , and first and second component of vector @xmath162 .
the input parameters ( derived from the input @xmath2 and @xmath3 values and eq . [ eqrot ] ) are recovered with residual biases that are smaller than the statistical error . therefore in paper vii we use the simplifying approximation of eq .
[ eqrot ] .
= cases 20 to 23 explore the effects of errors on the parameters of the selection functions determined in paper i. case 20 considers mock catalogues where the selection probabilities used in the fitting procedure have been computed adding + 0.1 dex to the true . as a consequence ,
the selection weights are systematically higher than the true ones ( see fig .
[ figalllike ] ) and the mean number of selected galaxies ( i.e. with selection probabilities larger than 0.1 ) is systematically smaller ( see table [ tabsimumean ] ) . in case 21 we added -0.1 dex to the true , getting selection weights systematically smaller than the true ones ( see fig .
[ figalllike ] ) and therefore a larger mean number of selected galaxies .
cases 22 and 23 use selection weights distorted by adding + 0.1 dex ( case 22 ) and -0.1 dex ( case 23 ) to the selection probability widths @xmath219 .
the largest systematic effects are observed for case 20 , where systematically smaller values of @xmath2 , @xmath3 , , and systematically larger values of @xmath203 and ( because more small galaxies are excluded from the fit ) are obtained .
see [ alllike ] and fig .
[ figalllike ] for a discussion of the effects on the distribution of likelihood probabilities .
cases 24 to 28 refer to mock catalogues of the efar sample with peculiar velocity fields .
case 24 has a random peculiar velocity field ( always the same for the performed 99 simulations ) of 0.05 dex rms amplitude .
we considered this particular field ( see fig .
[ figpecinput ] , where we plot the peculiar velocities of all clusters , the 29 fp plus the remaining ones ) because its mean bulk motions of the clusters at positive ( the hercules corona borealis sample , hcb ) and negative galactic longitudes ( the pisces perseus cetus sample , ppc ) are slightly non - zero and of opposite sign , reminiscent of what measured in paper vii . in addition , its mean velocities projected along the lp and smac dipole directions ( see fig .
[ figlpsmac ] for the peculiar velocity field of all clusters ) are also similar to what seen in paper vii .
case 25 has a random field ( always the same for the performed 99 simulations ) of the same rms amplitude as case 24 ( 0.05 dex ) plus the bulk flow motions determined by lauer and postman ( 1994 , hereafter lp , see fig .
[ figlpsmacinput ] for the peculiar velocity field of all clusters ) .
lp measured the bulk motion of the abell clusters with @xmath220 km / s with respect to the cosmic microwave background to be 764 km / s in the direction of @xmath221 ( reanalysis by colless 1995 ) , using the brightest cluster galaxies @xmath222 relation .
the resulting peculiar velocity field has a prominent asymmetry of the mean bulk motions of the hcb and ppc clusters ( see fig .
[ figlpsmacinput ] ) , which gives an apparently larger dipole motion in the lp direction ( see fig .
[ figlpsmac ] ) .
case 26 adds on top of a random field of 0.05 rms amplitude ( always the same for the performed 99 simulations ) the bulk motion determined by hudson et al .
( 1999 , hereafter smac ) .
smac determined the motions of the abell clusters with @xmath223 km / s using the fp , finding @xmath224 km / s towards @xmath225 .
the resulting peculiar velocity field has small mean bulk motions of the hcb and ppc clusters ( see fig .
[ figlpsmacinput ] for the peculiar velocity field of all clusters ) , and an approximately zero dipole motion in the smac direction ( see fig . [ figlpsmac ] ) .
case 27 has a pure lp bulk motion without a random component .
case 28 has a pure smac bulk motion without a random component . in all five cases
the fp parameters are recovered without additional biases with respect to case 1 .
the statistical errors are slightly larger , especially for the parameter .
the mean number of galaxies varies slightly , as galaxies near the selection cut might get included or not in the sample accordingly to the positive or negative peculiar velocity of the clusters they belong to ( see sect .
[ mock ] ) .
we conclude that the ml algorithm recovers the fp parameters correctly also when peculiar velocity fields are present .
= = the last three sets of simulations ( cases 45 , 46 and 47 ) refer to input distributions uniform in the fp parameters , in the errors , or both ( see sect . [ mock ] ) , with zero peculiar velocity fields .
inspection of tables [ tabsimumean ] and [ tabsimurms ] and fig .
[ figparvar ] shows that also in these cases the input parameters are recovered with small biases .
the largest effects are seen for @xmath198 , where the ml gaussian algorithm overestimates the effects of the uniform error distribution , biasing @xmath198 low .
the rms values are similar to the ones derived for the previous cases : @xmath226 are marginally larger , @xmath227 marginally smaller . summarizing , with only a few exceptions ( noted above ) , the systematic differences in the fits derived for different cases are comparable to the random errors in the determination of the parameters for the standard case .
therefore , the uncertainties in our best - fit fp parameters are dominated by the random errors and not by systematic effects from the fitting method . in addition , the systematic effects explored in the simulations ( tables [ tabsimumean ] and [ tabsimurms ] ) reproduce , at least qualitatively , the trends observed in the fits to the actual efar data ( table 4 of paper vii ) . [ cols= " > , > , > , > , > , > , > , > , > , > , > , > , < " , ] [ tabsimurms ] we now investigate whether the assumption of a gaussian distribution of galaxies in the fp space is compatible with the efar data sample .
[ figalllike ] ( left panel ) shows the cumulative distributions of the mean ( i.e. normalized to the number of galaxies fitted ) likelihoods of the simulations of cases 1 to 8 , 18 to 28 , and 45 to 47 ( top row , only galaxies of clusters used to determine the fp parameters ) and cases 1 and 20 to 47 ( bottom row , all galaxies of the clusters for which peculiar velocities have been computed ) .
note that cases 29 to 44 , not listed in tables [ tabsimumean ] and [ tabsimurms ] , are discussed in detail in [ resbias ] .
they are perturbed versions of case 1 . except for cases 20 and 21 , all simulations ( including the uniform distribution cases 45 , 46 and 47 ) give similar cumulative distributions of the mean likelihoods .
depending on the exact case , a mean likelihood equal to or larger than that of the efar sample is observed in up to 30% of the simulations .
case 45 ( a uniform distribution of the fp parameters and gaussian errors ) gives only slightly smaller mean likelihoods .
case 20 gives systematically higher mean likelihoods , and case 21 systematically lower , than that of the efar sample .
inspection of the central and left panel plots of fig .
[ figalllike ] clarifies that the cumulative distribution of the mean unweighted likelihoods are similar for all the simulations with gaussian error distributions , predicting that values larger than that of the efar sample are observed in 70 to 95% of the cases .
the cases 46 and 47 , where the error distributions are uniform , produce systematically 20 - 30% larger values , because the ml gaussian algorithm is overstressing the role of errors . as noticed before
( see fig .
[ figmg2sig ] ) , the mean selection weights of all simulations , except case 20 , are smaller than that of the efar sample . therefore , we conclude that the gaussian modeling is a reasonable description of the distribution of the efar galaxies in the ( , , ) space and of the error distribution .
the efar data set does not allow to discriminate between a gaussian or a uniform distribution of the fp parameters .
error distributions with tails slightly stronger than gaussians are hinted .
values of slightly larger ( by @xmath228 dex ) than that measured in paper i are needed to match the mean selection weight of the efar sample .
= once the parameters of the best - fitting gaussian distribution are determined using a subset of clusters ( hereafter , the fp clusters ) , the ml algorithm is used with these parameters fixed , to compute the peculiar velocities of the remaining clusters .
tables [ tabdelta ] , [ tabdelta2938 ] and [ tabdelta3947 ] give the average residuals ( over 99 simulations ) from the input peculiar velocities @xmath229 for a number of cases .
table [ tabdelta ] considers cases 1 and 20 to 28 of table [ tabsimumean ] , listing also the values of given in table 7 of paper vii , based on 1000 simulations of case 1 , and the statistical precision reached for case 1 and 99 simulations .
tables [ tabdelta2938 ] and [ tabdelta3947 ] give the results for 19 additional cases , from 29 to 47 .
the 16 cases from 29 to 44 are obtained by perturbing one of the fp parameters by plus or minus one sigma , keeping the others fixed to the best solution value .
cases 29 and 30 perturb the @xmath2 coefficient , 31 and 32 @xmath3 , 33 and 34 , 35 and 36 , 37 and 38 , 39 and 40 , 41 and 42 , 43 and 44 , respectively .
cases 45 , 46 and 47 refer to the simulations with uniform distributions in the fp parameters , the errors or both . fig .
[ figpecvar ] shows -@xmath230 as a function of . here
@xmath230 is the average residual over the fp clusters .
it is zero by construction for cases 1 , 20 to 28 and 45 to 47 .
the general trend of increasingly negative residuals at large values of is easily explained .
the clusters at higher redshifts ( large ) lack small galaxies and therefore have mean values of larger than , or @xmath231 progressively more negative .
once the zeropoint offset ( given by the residual averaged over the fp clusters ) is subtracted off , the simulations 29 to 44 give residuals compatible with case 1 within the statistical errors .
therefore , the residual bias corrections are robust against small errors in the fp solution . the largest variations of the zeropoint are observed when the values of , and are perturbed .
as expected , one finds @xmath232 , when is perturbed , @xmath233 , when is perturbed , and @xmath234 when is perturbed .
errors in affect the selection weighting scheme and therefore the determination of for the clusters ( at higher redshifts ) that have the largest . in some cases the errors in are larger than the statistical precision of case 1 .
positive errors ( @xmath235 , case 20 ) produce too negative , since more small galaxies than with the correct values of are excluded from the fits .
the contrary happens for negative errors ( @xmath236 , case 21 ) .
errors in @xmath219 affect also , especially at higher redshifts . in this case negative errors ( -d@xmath237 , case 23 ) produce too negative , because a sharper ( i.e. , a smaller @xmath219 ) selection function eliminates more small galaxies . positive errors ( d@xmath237 , case 22 ) produce the reverse .
the presence of a peculiar velocity field ( cases 24 to 28 ) can also affect , because redshift distances are used to convert the measured selection cuts from arcsec to kpc .
in particular , clusters sampled by a small number of galaxies are prone to small number statistics fluctuations , as galaxies near the selection cutoff might not be always included in the simulation catalogues .
uniform distributions of the fp parameters , of the errors or both do not affect systematically , within the statistical errors .
= finally , fig .
[ figpecdw ] illustrates the effect of the value on the required peculiar velocities .
when no is applied ( case 9 ) , the residual biases at high are larger than when is equal to 1.1 ( case 1 ) .
the biases are reduced if a larger ( 1.2 , case 12 ) is chosen , but the statistical errors increase . in addition , as discussed in
[ secvariant ] , this choice gives larger systematic and statistical errors for the fp parameters .
we conclude that the peculiar velocity biases are much smaller than the random errors in the peculiar velocities .
the uncertainties in the peculiar velocity bias corrections as computed from the 1000 simulations of case 1 used in paper vii do not increase the random errors in the peculiar velocities . however , residual systematic effects at the level of 0.01 dex might still be present , especially for the clusters at higher redshifts , due to the uncertainties in the determination of the selection function .
= given the large mean distance of the efar clusters , and the precision of the fp distance estimator ( 20% for the single galaxy , see paper vii ) , the present dataset does not allow the determination of the peculiar motions of single clusters .
nevertheless , the mean motions of the efar clusters can be constrained . in the following
we investigate this point using the subsample of clusters identified in paper vii for the study of the bulk motions .
to this belong the 50 efar clusters with 3 or more galaxies ( hereafter peculiar velocity , pv , clusters ) , redshifts less than 15000 km / s , and peculiar velocity errors less than 1800 km / s .
we compute the average mean velocities without applying the bias corrections investigated in [ resbias ] ( although this is done in paper vii ) , to stress that our results are not affected much by these residual systematic uncertainties , and that random errors dominate the noise .
we begin by examining the mean peculiar velocities @xmath238 , @xmath239 and @xmath240 of the sample .
these are the mean over the pv clusters , over the pv clusters with positive ( the hercules corona borealis , hcb , sample ) , and over the ones with negative galactic longitude ( the perseus pisces cetus , ppc , sample ) .
[ figvmean ] shows that these average quantities are recovered with statistical errors of about 200 km / s . in general these errors
are larger than the systematic differences from the input values .
in particular , errors on the selection function parameters do not affect the result much ( cases 20 to 23 ) and uniform distributions of the fp parameters , of the errors or both do not change this conclusion ( cases 45 to 47 ) .
the largest systematic difference is observed for cases 24 to 26 and @xmath241 , where the input values are ill - defined , since the rms of the cluster peculiar velocities is large ( see figs .
[ figpecinput ] and [ figlpsmacinput ] ) .
= as a second test we consider the supergalactic x , y and z components of the mean peculiar velocities in radial shells .
[ figxyz ] shows the averages of 99 simulations obtained for case 1 and 24 of table [ tabsimumean ] .
again , the input values are recovered with systematic differences smaller than the statistical errors .
= we determine now how well the efar cluster subsample can constrain the bulk flow motions determined by lauer and postman ( 1994 ) and hudson et al .
we fitted the peculiar velocities of the 50 clusters used in paper vii with a bulk flow model @xmath242 , where @xmath243 is the angle between the dipole and cluster direction .
determining @xmath244 in a least squares sense .
15 of paper vii shows these fits to the efar data .
[ figlpsmac ] shows that when simulations with no peculiar velocities are considered ( cases 1 and 20 to 23 of table [ tabsimumean ] ) , the recovered mean @xmath244 is compatible with zero , with statistical errors of about 160 km / s .
errors on the selection function parameters do not affect the result much ( cases 20 to 23 ) , as uniform distributions of the fp parameters , of the errors or both ( cases 45 to 47 ) .
similarly , when the simulations with pure lp or smac dipole motions are considered ( cases 27 and 28 respectively ) , the recovered mean @xmath244 is consistent with the input values within the statistical uncertainties .
16 of paper vii shows the histograms of the recovered bulk flow amplitudes for 500 simulations .
only one out of these simulations of the lp flow , and none of smac yields a @xmath244 less than the value measured from the efar data in paper vii .
in contrast , the simulations with a random peculiar velocity component ( case 24 ) give mean @xmath244 values similar to the observed values , both for the lp and the smac directions .
simulations with a random component on top of the lp and smac bulk flows ( cases 25 and 26 ) produce large mean @xmath244 along the lp direction , and little net motion along the smac direction .
= finally , we investigate whether not only the mean motions , but also their global intrinsic fluctuations could be determined with the efar cluster sample . to this purpose
we use the one - dimensional ml algorithm of [ onedim ] : we consider the peculiar velocities derived for the 99 simulations of case 24 of table [ tabsimumean ] and their projections along the supergalactic x , y and z coordinates with their estimated errors , separately for the whole sample of pv clusters , and the northern hcb and southern ppc subsets . for each of these datasets
the ml algorithm estimates the intrinsic ( gaussian ) dispersion .
[ figrms ] shows the results .
the derived rms are systematically underestimated , up to more than a factor two , when the input values are of the order of @xmath245 km / s , and by @xmath246% if larger input values are considered .
we conclude that cluster sample is too small to allow an accurate determination of these quantities .
= to summarize , the efar cluster sample allows to measure the mean motions of the hcb and ppc regions with 200 km / s precision , and to constrain the bulk flow along the lp dipole direction strongly .
it is less suited to constrain the smac result and does not allow the accurate determination of the random peculiar motion component .
we have described the maximum likelihood gaussian algorithm we developed to investigate the correlations between the parameters of the efar database ( the relation of paper ii , the and relations of paper v , the fundamental plane of paper vii ) and determine the cluster peculiar velocities of paper vii .
we performed extensive testing based on mock catalogues of the efar sample .
we find that `` canonical '' methods based on a least - squares approach can not cope with the challenge of a sample with a spread of a factor of two in redshift , with sizable selection effects , non - negligible and non - uniform errors , and explicit cuts .
we quantify the size of the systematic biases these methods introduce .
in contrast , the maximum likelihood gaussian algorithm takes into account errors , selection effects and the presence of explicit cuts , determining nearly - unbiased estimates of the slopes of the correlations and their intrinsic and parallel spread .
ten to thirty percent of the analyzed simulations have mean likelihoods larger than that of the efar sample , justifying the use of gaussian modeling .
we derive the analytical solution of the maximum likelihood gaussian problem in n dimensions in the presence of small errors .
we show that the residual systematic biases are always smaller than the statistical errors .
we investigate in detail the effects of cluster sample selection , errors in the selection function parameters , and selection cuts .
we explore the cases of uniform distributions of parameters and errors .
we conclude that the mean peculiar motions of the efar clusters can be determined reliably .
in particular , the large amplitude of the dipole motion measured by lauer and postman ( 1994 ) can be strongly constrained .
rps acknowledges the support by dfg grant sfb 375 .
gw is grateful to the serc and wadham college for a year s stay in oxford , and to the alexander von humboldt - stiftung for making possible a visit to the ruhr - universitt in bochum .
mmc acknowledges the support of a lindemann fellowship , a dist collaborative research grant and an australian academy of science / royal society exchange program fellowship .
this work was partially supported by nsf grant ast90 - 16930 to db , ast90 - 17048 and ast93 - 47714 to gw , ast90 - 20864 to rkm .
the entire collaboration benefitted from nato collaborative research grant 900159 and from the hospitality and monetary support of dartmouth college , oxford university , the university of durham and arizona state university .
support was also received from pparc visitors grants to oxford and durham universities and a pparc rolling grant : `` extragalactic astronomy and cosmology in durham 1994 - 98 '' .
abramowitz , m. , stegun , a. , 1971 , handbook of mathematical functions , dover , new york akritas m.g .
, bershady m.a . , 1996 ,
apj 470 , 706 carvalho r.r . , djorgovski s. , 1992 , apj , 389 , l49 colless , m.m .
, 1995 , aj , 109 , 1937 colless , m. , burstein d. , davies r.l . , mcmahan r. , saglia r.p . ,
wegner g. , 1999 , mnras , 303 , 813 ( paper v ) colless , m. , saglia r.p . , burstein d. , davies r.l . , mcmahan r. , wegner g. , 2000 , mnras , in press ( paper vii ) djorgovski s. , davis m. , 1987 , apj , 313 , 59 dressler a. , lynden - bell d. , burstein d. , davies r.l . , faber s.m . , terlevich r.j . ,
wegner g. , 1987 , apj 313 , 42 eadie , w.t . , dryard , d. , james , f.e .
, roos , m. , sadoulet , b. , 1971 , statistical methods in experimental physics , north - holland publishing company , amsterdam feigelson e.d . ,
babu g.j . , 1992 , apj , 397 , 55 isobe t. , feigelson e.d .
, akritas m.g . ,
babu g.j . , 1990 ,
apj , 364 , 104 hudson m.j . ,
smith , r.j . ,
lucey , j.r . ,
schlegel , d.j . , davies , r.l . , 1999 , apj , 512 , 79l hudson m.j . , lucey j.r . ,
smith r.j . , steel j. , 1997 , mnras , 291 , 461 jrgensen i. , franx m. , kjrgaard p. , 1996 ,
mnras , 280 , 167 jrgensen i. , 1997 , mnras , 288 , 161 la barbera , f. , busarello , g. , capaccioli , m. , 2000 , a&a , 362 , 851 lauer , t.r . , postman , m. , 1994 , apj , 425 , 418 pahre m.a .
, djorgovski s.g .
, de carvalho r.r . , 1997 , proc . of the eso workshop on `` galaxy scaling relations : origins , evolution and applications '' , eds .
l.n . da costa & a. renzini , p. 197
press w.h . ,
flannery b.p . ,
teukolsky s.a . ,
vetterling w.t .
, 1986 , numerical recipes , cambridge univ . press , cambridge saglia r.p . , burstein d. , baggley g. , bertschinger e. , colless m.m .
, davies r.l . , mcmahan r. , wegner g. , 1997a , mnras 292 , 499 ( paper iii ) saglia r.p . ,
bertschinger e. , baggley g. , burstein d. , colless m.m . , davies r.l .
, mcmahan r. , wegner g. , 1997b , apjs , 109 , 79 ( paper iv ) scodeggio m. , giovanelli r. , haynes m.p . , 1997 , aj , 113 , 101 teerikorpi , p. , 1997
, araa , 35 , 101 van albada t.s . , bertin g. , stiavelli m. , 1993 , mnras , 265 , 627 wegner g. , colless m. , baggley g. , davies , r.l .
, bertschinger e. , burstein d. , mcmahan r. , saglia r.p . , 1996 , apjs , 106 , 1 ( paper i ) wegner g. , colless m. , burstein d. , saglia r.p . , mcmahan r. , davies , r.l . , baggley g. , 1999 , mnras , 305 , 259 ( paper ii ) weinberg s. , 1972 , gravitation and cosmology , wiley , new york willick , j.a
. , 1994 , apjs , 92 , 1
neglecting the presence of cuts , the maximization of eq .
( [ eqonedim ] ) leads to : @xmath247 @xmath248 here we solve eqs .
[ eqonemean ] and [ eqonesigma ] in the limiting case of small measurement errors ( @xmath249 ) .
we get : @xmath250 @xmath251 where @xmath252 is the selection weighted mean , @xmath253 is the error and selection weighted mean , @xmath254 is the selection weighted mean square error , @xmath255 the selection weighted rms and @xmath256 the selection and error weighted rms .
therefore the mean and rms are obtained by correcting the usual formulae with a term taking into account the spread in errors .
here we present the solution which minimizes eq . [ eqlikelihood ] in the limiting case of small errors ( with a diagonal error matrix ) , no peculiar velocities and no cuts , considering expansions to first order in the errors , for the two dimensional case . in appendix [ derivation ]
we derive the general result in the n dimensional case . to simplify the notation in the following we call the components @xmath257 as @xmath258 .
for the mean values we find : @xmath259 @xmath260 where @xmath261 , @xmath262 , @xmath263 , @xmath264 , @xmath265 , @xmath266 , @xmath267 , @xmath268 , @xmath269 , @xmath270 , @xmath271 , @xmath272 , and @xmath273 .
the covariance matrix @xmath27 is : @xmath274 @xmath275 @xmath276 where @xmath277 , @xmath278 , @xmath279 , @xmath280 . as in the 1-dim case , we recognise
the usual ( error corrected ) zeroth - order plus a term taking into account the spread in the errors . with no spread in the errors we have @xmath281 and we recover the _ ansatz _ of akritas and bershady ( 1996 ) . to first order eq .
[ eqao ] then reads : @xmath282 where @xmath283 is eq .
[ eqao ] evaluated using @xmath284 . with no spread in the errors : @xmath285.\ ] ] therefore @xmath283 _ underestimates _ the true slope if the error in the @xmath16 direction is larger than the one in @xmath17 .
let @xmath27 be the intrinsic covariance matrix and @xmath286 its inverse , @xmath66 the error matrix of each point , @xmath287 the observed covariance matrix of each point , with @xmath288 .
then we have : @xmath289^{-1}=(i+\lambda e_i)^{-1}\lambda \sim ( i-\lambda e_i)\lambda=\lambda-\lambda e_i\lambda,\ ] ] where the first order approximation @xmath290 was used .
we get : @xmath291= det v [ 1+\sigma_{jk } \lambda_{jk } e_{i , jk}],\ ] ] where the first order approximation @xmath292 was used . therefore to first order we get : @xmath293.\ ] ] the equation for the likelihood reads : @xmath294\right \}^{1/s_i},\ ] ] where @xmath295 , @xmath296 are the n - dimensional m vectors of datapoints , and @xmath87 the vector of the mean values . taking the logarithm
we get : @xmath297\ ] ] the equation for the vector of the mean values reads : @xmath298 setting @xmath299 , the zero - th order term is : @xmath300 the first order term is then : @xmath301 and therefore : @xmath302 which is equivalent to eqs . [ eqxmeantwo]-[eqymeantwo ] for @xmath66 diagonal and taking the zeroth order for @xmath286 ( see below ) .
let now consider the derivatives with respect to the @xmath286 components : @xmath303 let first note that : @xmath304 where @xmath305 is the matrix obtained eliminating the row @xmath306 and the column @xmath307 . since @xmath286 is symmetric we get : @xmath308 if @xmath309 , and : @xmath310 otherwise .
similarly , we get @xmath311 if @xmath309 and @xmath312 .
finally , we derive : @xmath313 if @xmath309 , and : @xmath314 otherwise . writing the covariance matrix as @xmath315 ,
the zeroth order reads @xmath316 .
the first order comes from the terms of eq .
[ eqdlambda ] linear in @xmath66 : @xmath317 @xmath318 where @xmath319 and @xmath320 .
note that @xmath321 if there is no spread in the errors ( i.e. , @xmath322 ) .
the photometric part of the error matrix @xmath66 of eq . [ eqerror ] can be derived as follows .
paper iv shows that the errors in the half - luminosity radii @xmath146 and average surface brightness @xmath323 derived from the fits to the photometric profiles are correlated , with @xmath324 and @xmath150 , and calibrates the values of @xmath146 and @xmath148 in terms of the quality parameter q. therefore , we consider the rotated coordinates @xmath325 , where @xmath326 and @xmath327 , and @xmath328 the corresponding rotation matrix . in these
coordinates the error matrix @xmath329 is diagonal , with @xmath330 and @xmath331 . from the definition of @xmath17 it follows @xmath332 . from @xmath333 and
the fact that the errors on @xmath17 and @xmath334 are uncorrelated , we get @xmath335 and finally @xmath336 .
the matrix @xmath337 gives then : @xmath338 @xmath339 @xmath340
the integral of eq . [ eqnorm ] can be performed by defining the set of rotated coordinates @xmath341 : @xmath342 writing @xmath343 , integration over @xmath344 leads to the following integral : @xmath345 where the @xmath346 matrix @xmath347 is @xmath348 and @xmath349 , where @xmath350 ( paper iii , @xmath351 ) , @xmath352 and @xmath353 .
finally one gets : @xmath354 where @xmath112 is the bivariate probability integral ( abramovitz & stegun 1971 ) , @xmath355 , @xmath356 and @xmath357 . | the efar project is designed to measure the properties and peculiar motions of early - type galaxies in two distant regions . here
we describe the maximum likelihood algorithm we developed to investigate the correlations between the parameters of the efar database .
one- , two- , and three - dimensional gaussian models are constructed to determine the mean value and intrinsic spread of the parameters , and the slopes and intrinsic parallel and orthogonal spread of the , , relations , and the fundamental plane . in the latter case , the cluster
peculiar velocities are also determined .
we show that this method is superior to `` canonical '' approaches of least - squares type , which give biased slopes and biased peculiar velocities .
we test the algorithm with monte carlo simulations of mock efar catalogues and derive the systematic and random errors on the estimated parameters .
we find that random errors are always dominant .
we estimate the influence of systematic errors due to the way clusters were selected and the hard limits and uncertainties in the selection function parameters for the galaxies .
we explore the influence of uniform distributions in the fundamental plane parameters and the errors .
we conclude that the mean peculiar motions of the efar clusters can be determined reliably . in particular , the placement of the two efar sample regions relative to the lauer & postman dipole allows us to strongly constrain the amplitude of the bulk motion in this direction .
we justify a posteriori the use of a gaussian modeling for the galaxy distribution in the fundamental plane space , by showing that the mean likelihood of the efar sample is obtained in 10 to 30% of our simulations .
we derive the analytical solution for the maximum likelihood gaussian problem in n dimensions in the presence of small errors .
galaxies : clusters - galaxies : elliptical and lenticular , cd - galaxies : fundamental parameters - - galaxies : distances and redshift - cosmology : large scale structure of universe |
in the orthodox way of unification , such as the standard model and the grand unified theories , fundamental fermions belonging individually to the irreducible representations of the external lorentz group are organized into definite multiplets of the internal symmetry group . here
, we propose an unorthodox scheme of unification where a family of fundamental fermions is described by a master spinor field obeying to a generalized dirac equation with coefficients belonging to an algebra , called a _ triplet algebra _ , which consists of all the triple - direct - products of the dirac matrices . in accordance with the coleman - mandula theorem @xcite insisting that spacetime and internal symmetries can not be combined in any but a trivial way , we decompose the triplet algebra into mutually - commutative subalgebras .
the subalgebra related with internal property of fundamental fermions , _ quarks and leptons _ , is made up of two subalgebras which can be related , respectively , to repetitional family structure and color symmetry .
the master spinor field , named a _ triplet field _ , is proved to include threefold ( fourfold ) repetitional modes of spin @xmath1 component fields with su(3 ) ( su(4 ) ) color symmetry .
we construct explicitly the triplet algebra and its subalgebra for the external spacetime symmetry in 2 and define the triplet field forming a multiple representation of the lorentz group in 3 .
characteristic of the subalgebra for internal properties of fundamental fermions is closely examined in 4 and 5 .
first , chirality operators acting on the triplet field are decomposed further into the sums of the projection operators which can be interpreted to discriminate repetitional degrees of freedom for family structure .
second , we find the algebra for color symmetry in the set of the elements which are commutative both with the chirality operators and all elements of the algebra for external symmetry . in the present scheme ,
the triplet field has no degrees of freedom for the symmetry of electroweak interaction . in 6 , we reformulate the standard model by introducing the set of chiral triplet fields for the left - handed doublet and the right - handed singlets of the weinberg - salam symmetry .
it turns out possible to qualify the yukawa interaction and make a new interpretation on the yukawa coupling constants naturally in an intrinsic mechanism of the triplet field formalism . as an explicit illustration ,
the dirac mass matrices with quasi - democratic structure are derived .
possible applications of the present scheme to the systems of fundamental fermions with different attributes and symmetries are discussed in 7 .
the dirac s @xmath0-matrices @xmath2 satisfying the clifford anticommutation relations @xmath3 with @xmath4 are introduced solely as the mathematical quantities which carry no direct physical meaning .
hermite conjugate of @xmath5 is defined by @xmath6 .
the @xmath0-matrices generate the 16 dimensional manifold , named a _ dirac algebra _ , as @xmath7 where @xmath8 and @xmath9 .
let us call the triple - direct - product of the bases @xmath10 and @xmath11 a _ primitive triplet _ and the triple - direct - product of arbitrary elements of @xmath12 a _
triplet_. then , we define the _ triplet algebra _ @xmath13 as the @xmath14 dimensional manifold spanned by all the linear combinations of triplets . in other words ,
the triplet algebra @xmath13 is generated in terms of the 12 primitive triplets @xmath15 and @xmath16 as follows : @xmath17 the primitive triplets @xmath18 and @xmath19 obey the multiplication rule @xmath20 and the triplets satisfy the distribution law @xmath21 where @xmath22 , @xmath23 and @xmath24 are the bases of @xmath25 in ( [ diracalg ] ) , and @xmath26 , @xmath27 and @xmath28 are complex numbers .
the transpose ( hermite conjugate ) of the primitive triplet is defined by the triple - direct - product of its transposed ( hermite conjugate ) components of @xmath25 as @xmath29 and the trace of the triplet is set to be @xmath30 the triplet algebra @xmath13 is large enough to form a variety of realizations of the lorentz group representation .
to obtain such kind of multiple spinor representation that describes exclusively the spin @xmath1 states , we note that the primitive triplets @xmath31 satisfy the clifford relations @xcite @xmath32 where @xmath33 .
hermite conjugate of @xmath34 is defined by @xmath35 and the traces are calculated to be @xmath36 with these new elements , we are now able to construct a subalgebra of @xmath13 being isomorphic to the dirac algebra @xmath25 as follows : @xmath37 where @xmath38 and @xmath39 the set of elements which are commutative with the subalgebra @xmath40 , _
i.e. _ , @xmath41 = 0\,\}\ ] ] forms a subalgebra of @xmath13 , called a _ centralizer _ of @xmath40 in @xmath13 . the primitive triplet in @xmath42 is proved to have even numbers of components @xmath5-matrices for all @xmath43 @xcite .
therefore , the composition of @xmath42 is determined to be @xmath44 the triplet algebra has a non - intersecting decomposition @xmath45 .
note that all of the antisymmetric operators @xmath46 form a subalgebra @xmath47 we identify this subalgebra @xmath48 with the _ external algebra _ which generates the lorentz group in our spacetime .
it is readily proved that the operators @xmath49 are subject to the commutation relations of the lie algebra for the orthogonal group o(1,3 ) as @xmath50 = -i\eta_{\kappa\mu } m_{\lambda\nu } + i\eta_{\kappa\nu } m_{\lambda\mu } -i\eta_{\lambda\nu } m_{\kappa\mu } + i\eta_{\lambda\mu } m_{\kappa\nu}\ ] ] and @xmath51 and @xmath52 satisfy the relations @xmath53
= i\eta_{\lambda\nu } \gm_{\mu } -i\eta_{\lambda\mu } \gm_{\nu } .\ ] ] accordingly , it is possible to postulate that the operators @xmath51 generate the lorentz transformations and that the subscripts of operators consisting of @xmath34 refer directly to the superscripts of the spacetime coordinates @xmath54 of the 4 dimensional world where we exist as observers .
therefore , it is natural to define the raising and lowering operations of index of @xmath34 by the metric @xmath55 as follows : @xmath56 as is evident from the construction so far , the component @xmath0-matrices carry no direct physical meaning and regarded as playing only the role of _ alphabets_. it is the elements of the triplet algebra @xmath13 that play the role of _ codons _ carrying the direct physical meanings .
at this stage , we are allowed to introduce the triplet field , @xmath57 , on the spacetime point @xmath58 , which spans a @xmath59 dimensional vector space for multiple spinor representations of the lorentz group .
the triplet field is proved to possess @xmath60 component fields with spin @xmath61 carrying different internal attributes . for the triplet field @xmath57 and its adjoint field
@xmath62 the scalar product is defined as @xmath63 under the proper lorentz transformation @xmath64 where @xmath65 and @xmath66 , the triplet field and its adjoint are presumed to transform as @xmath67 where the transformation matrix @xmath68 belongs to the external algebra @xmath48 . for the bilinear form @xmath69 to behave as the vector under the lorentz transformation
, @xmath68 must satisfy the conditions @xmath70 which determine its explicit form to be @xmath71 for the spacetime rotations with angles
@xmath72 in the @xmath43-@xmath73 planes . for the discrete spacetime transformations such as the space inversion ,
the time reversal and the charge conjugation , the present scheme retains exactly the same structure as the ordinary dirac theory .
the internal attributes of fundamental fermions should be fixed independently of the inertial frame of reference in which observations are made .
this means that the generators specifying their internal attributes must be commutative with the elements of the external algebra @xmath48 for the lorentz transformations .
accordingly , we are led naturally to postulate that the algebra for internal symmetry should be related to the centralizer of @xmath48 defined by @xmath74 = 0 \}.\ ] ] the primitive triplet of the centralizer @xmath75 is proved to consist of either an even number or an odd number of component @xmath5-matrix for all @xmath43 @xcite .
consequently , the composition of the centralizer @xmath75 is determined as follows : @xmath76 it is now recognized that the triplet algebra has another non - intersecting decomposition @xmath77 .
we have to examine detailed structures of @xmath75 , which is identified with _ internal algebra _ , and to inquire what sorts of internal attributes of the fundamental fermions can be inscribed on the triplet field @xmath57 forming the multiple spinor representation of the lorentz group .
the subalgebras of the triplet algebra @xmath13 defined so far , @xmath40 , @xmath48 , @xmath42 and @xmath75 , obey the inclusion relations @xmath78 connectedly , the element @xmath79 belongs to the subalgebras @xmath40 and @xmath75 as follows : @xmath80
equation ( [ whereisgm5 ] ) shows that the chirality operators @xmath81 belong to both of the subalgebras @xmath40 and @xmath75 .
it is straightforward to verify that these operators have the following subdivisions in the internal algebra @xmath75 as @xmath82 and @xmath83 where @xmath84 note that this subdivision can not be realized in the algebra @xmath40 . to give independent physical roles and meanings to those subdivided terms in ( [ loperator ] ) and ( [ roperator ] ) ,
let us introduce the following operators in @xmath75 as @xmath85 which are chiral projection operators obeying the relations @xmath86 for @xmath87 and @xmath88 .
next , we define the non - chiral projection operators @xmath89 satisfying the relations @xmath90 all of these operators form the subalgebra @xmath91 of the internal algebra @xmath75 . here , we make interpretation that the projected modes of triplet field given by @xmath92 and @xmath93 represent the repetitional family structures of quarks and leptons . to apply this scheme to the three generations of quarks and leptons observed in the low energy regime , we impose the constraint condition , a posteriori , @xmath94 on the triplet field to suppress the fourth mode .
henceforth , we describe this three generation model in the present paper .
note , however , that it is possible also to describe the model including the fourth mode whose existence could be confirmed in sufficiently high energy regime .
in the previous section , one aspect of the internal algebra @xmath75 was clarified by examining its subalgebra @xmath95 .
consequently , it is necessary and reasonable to investigate the structure of the centralizer @xmath96 of @xmath95 in @xmath75 . for its purpose ,
let us notice that the dirac algebra @xmath12 has the three elements @xmath97 which satisfy the multiplication rules of the pauli algebra , _
i.e. _ , @xmath98 . taking the triple - direct - products of these elements ,
we are able to construct the bases for a 16 dimensional submanifold included exclusively in the centralizer @xmath96 as follows : @xmath99 the 15 operators @xmath100 ( @xmath101 ) satisfy the commutation relations @xmath102
= 2f_{jkl}\lam_l \label{comsu4}\ ] ] and the anticommutation relations @xmath103 of the lie algebra a@xmath104 , where @xmath105 and @xmath106 are the symmetric and antisymmetric constants characterizing the algebra .
the operators @xmath100 are self - adjoint and have the traces @xmath107 we postulate , here , that the lie group generated by the algebra @xmath108 describes an extended color su@xmath109(4 ) symmetry . evidently , this is one realization of the pati - salam symmetry in which the leptons are interpreted as the 4-th _ lilac _ color mode .
the projection operators to the quark and lepton sectors are constructed , respectively , by @xmath110 and @xmath111 which are subject to the relations @xmath112 for @xmath113 .
accordingly , the operator for _ baryon number minus lepton number _ takes the form @xmath114 it is straightforward to verify that the operator @xmath115 has the subdivision as @xmath116 where @xmath117 are the projection operators into tricolor quark modes .
the operators @xmath118 satisfy the relations @xmath119 for @xmath120 . the extended color symmetry su@xmath109(4 )
is presumed to be broken spontaneously at an appropriate energy scale down to the symmetry of semi - simple group su@xmath109(3)@xmath121u@xmath122(1 ) . to see how su@xmath109(3 )
is embedded in su@xmath109(4 ) , it is necessary to note the following property of the constant @xmath106 of su(4 ) group as @xmath123 for @xmath124 , and the relation between @xmath125 and @xmath115 in ( [ bl15 ] ) .
for the eight operators @xmath100 @xmath126 , we obtain the commutation relations in ( [ comsu4 ] ) with the structure constants @xmath105 of the su(3 ) group and anticommutation relations @xmath127 note that the projection operator @xmath115 appears in the right hand side of this relation . without it ,
this relation runs into contradiction , when applied to the operator @xmath128 from the right
. the identities @xmath129 which hold for @xmath130 and @xmath131 imply that @xmath132 are simultaneous eigenvectors of @xmath115 and @xmath128 with respective eigenvalues 1 and 0 , and that the operators @xmath132 annihilate the leptonic mode .
these results prove that the 9 dimensional submanifold of the extended color algebra @xmath133 defined by @xmath134 generates the color su@xmath109(3 ) group .
the u@xmath122(1 ) group is generated by the operator @xmath135 given in ( [ bl15 ] ) . from the triplet spinor field @xmath57 ,
the quark and lepton fields are projected out by @xmath136 respectively , for @xmath137 , and four color modes of su@xmath109(4 ) symmetry are given by @xmath138 for @xmath139 .
to separate the component fields of internal modes of the triplet field , it is convenient to introduce the braket symbols for the projection operators @xmath118 and @xmath140 by @xmath141 then , the decomposition of the bilinear scalar and vector forms of the triplet fields can be achieved as follows : @xmath142 and @xmath143 where @xmath144 implies that @xmath145 and @xmath146 .
the chiral component fields @xmath147 and @xmath148 are defined by @xmath149 and @xmath150 we have extracted maximally @xmath60 component spinor fields with spin @xmath61 , @xmath151 , from the original triplet spinor field @xmath57 . in the present formalism , the triplet field has no degrees of freedom for the weinberg - salam symmetry su@xmath152u@xmath153(1 ) .
therefore , to describe the actual system of quarks and leptons in this algebraic scheme , it is required to prepare a set of triplet fields to make up for the deficit degrees of freedom for electroweak interaction . here
we incorporate the weinberg - salam symmetry into our formalism by introducing the set of chiral triplet fields so that the left - handed triplets constitute the doublet @xmath154 and the right - handed triplets form the singlets @xmath155 of the electroweak isospin , where @xmath156 and @xmath157 signify the up and down states .
the quark and lepton parts of the electroweak doublet @xmath158 have the component modes expressed schematically as @xmath159 and @xmath160 similarly , the electroweak singlets @xmath161 and @xmath162 possess the quark and lepton parts with contents of the component modes as follows : @xmath163 @xmath164 @xmath165 and @xmath166 the chiral triplets in ( [ chiraldoublet ] ) and ( [ chiralsinglets ] ) enable us to express the lagrangian density of the kinetic part including the gauge interactions of quarks and leptons in the generic form @xmath167 with the covariant derivatives @xmath168 .
in the low energy regime where the standard model can be applied , the covariant derivatives act on the triplet fields as follows : @xmath169 @xmath170 and @xmath171 where @xmath172 and @xmath173 are the gauge fields which interact , respectively , to the electroweak isospin @xmath174 of su@xmath175 and the hypercharge of u@xmath153(1 ) , and @xmath176 is the gauge field of color symmetry .
this density @xmath177 is readily proved to reproduce the gauge interactions of the standard model with right - handed neutrino species . in the present scheme ,
the chiral triplet fields include the repetitional degrees of freedom for family structure .
this feature enables us to formulate the yukawa interactions in an intrinsic mechanism without bringing in so many unknown parameters from outside . in terms of chiral triplet fields , the lagrangian density of the yukawa interactions
can be expressed to be @xmath178 where @xmath179 are the higgs field and its conjugate of the standard model , and the kernel operator @xmath180 works to induce mixings among repetitional modes . to be commutative with the generators of color symmetry
, the operator must be composed of triple - direct - products of the dirac matrices @xmath181 and @xmath11 . to frame its generic construction
, we impose the conditions that it is self - adjoint and it brings forth the quasi - democratic mass matrices in the low energy regime . as a natural candidate for @xmath182 , we adopt @xmath183 with real parameters @xmath184 and @xmath185 . the quark part of the lagrangian density in ( [ yukawaint ] ) is readily decomposed into the yukawa interactions among component quark fields as @xmath186\phi \psi_{jr}^{(a)}.\ ] ] this result and the similar decomposition of the lepton part lead us to the interpretation that the quantities @xmath187 are the yukawa coupling constants of the @xmath188 sector @xmath189 . in the low energy regime where the higgs field takes the vacuum expectation value @xmath190 and , consequently , the weinberg - salam symmetry is broken , the yukawa interaction with the kernel in ( [ kernel ] ) results in the dirac mass matrices @xmath191 where @xmath192 , @xmath193 and @xmath194 ( cyclic ) .
this type of the hermitian quasi - democratic matrices are recognized to explain the hierarchical structures of mass spectra and the weak mixing matrix for the quark sector @xcite .
in this way , we have formulated an algebraic theory for unified description of external and internal symmetries of quarks and leptons .
the basic ingredients of the theory are the _ triplet algebra _ consisting of all the triple - direct - products of the dirac matrices and the _ triplet field _ forming the multiple spinor representation of the lorentz group .
the triplet field has the @xmath59 component fields which afford the freedom for the repetitional family structure as well as the external spacetime symmetry and the internal color symmetry . by postulating the triplet fields of the electroweak doublet and singlets to exist , the standard model with three generations
is successfully reproduced so that the yukawa interactions are formulated , without bringing in many unknown coupling constants from outside , and the dirac mass matrices with quasi - democratic structure are naturally derived .
the present formalism is flexible enough to be applicable to various field theories suitable for descriptions for different energy scales .
removing of the constraint in ( [ constraint ] ) enables us formulate the four generation theory of quarks and leptons and introduction of the right - handed electroweak doublet in place of the singlets in ( [ chiralsinglets ] ) allows us to obtain the gauge theory of left - right symmetry .
further , we can bring in the gauge symmetry for repetitional family modes .
note that the three dirac matrices @xmath195 , @xmath196 and @xmath197 generate the pauli algebra . taking the triple - direct - products of these elements just as in ( [ gencolorsym ] ) , we can construct the 16 dimensional submanifold which generates the su(4 ) group for family symmetry .
therefore , it is possible to formulate the gauge field theory with the symmetry su@xmath198su@xmath152su@xmath199su@xmath200(4 ) in the triplet field formalism . 9
coleman s and mandula j 1967 _ phys .
rev . _ * 159 * 1251 weinberg s 2000 _ the quantum field theory of fields _ vol iii(cambridge university press ) chapter 24 sogami i s 1987 _ prog .
* 78 * 487 branco g c , silva - marcos j i and rebelo m n 1990 _ phys .
lett . _ * b237 * 446 sogami i s , nishida k , tanaka h and shinohara t 1998 _ prog .
phys . _ * 99 * 281 | an unorthodox unified theory is developed to describe external and internal attributes and symmetries of fundamental fermions , _ quarks and leptons_. basic ingredients of the theory are an algebra which consists of all the triple - direct - products of dirac @xmath0-matrices and a triple - spinor - field , called a _ triplet field _ , defined on the algebra .
the algebra possesses three commutative subalgebras which describe , respectively , the external spacetime symmetry , the family structure and the internal color symmetry of quarks and leptons .
the triplet field includes threefold ( fourfold ) repetitional modes of spin @xmath1 component fields with su(3 ) ( su(4 ) ) color symmetry .
it is possible to qualify the yukawa interaction and to make a new interpretation of its coupling constants naturally in an intrinsic mechanism of the triplet field formalism .
the dirac mass matrices with quasi - democratic structure are derived as an illustration . |
an oriented smooth surface @xmath2 is called _ translating soliton _ of the mean curvature flow ( _ translator _ for short ) if its mean curvature vector field @xmath3 satisfies the differential equation @xmath4 where @xmath5 is a fixed vector of unit length and @xmath6 stands for the orthogonal projection of @xmath7 to the normal bundle of the immersion @xmath8 .
if @xmath9 is the outer unit normal of @xmath8 , then the translating property can be expressed in terms of scalar quantities as @xmath10 where @xmath11 is the scalar mean curvature of @xmath8 .
translators are important in the singularity theory of the mean curvature flow since they often occur as type - ii singularities .
an interesting example of a translator is the _
canonical grim reaper cylinder _ @xmath12 which can be represented parametrically via the embedding @xmath13 given by @xmath14 any translator in the direction of @xmath7 which is an euclidean product of a planar curve and @xmath15 is either a plane containing @xmath7 or can be obtained by a suitable combination of a rotation and a dilation of the canonical grim reaper cylinder .
the latter examples will be called _ grim reaper cylinders_. note that the canonical grim reaper cylinder @xmath12 is translating with respect to the direction @xmath16 for simplicity we will assume that all translators to be considered here are translating in the direction @xmath17 . before stating the main theorem
let us set up the notation and provide some definitions .
[ asymp-1 ] let @xmath18 be an open half - plane in @xmath0 and @xmath19 the unit inward pointing normal of @xmath20 .
for a fixed positive number @xmath21 , denote by @xmath22 the set given by @xmath23 1 .
we say that a smooth surface @xmath24 is @xmath25-asymptotic to the open half - plane @xmath18 if @xmath24 can be represented as the graph of a @xmath25-function @xmath26 such that for every @xmath27 there exists @xmath28 so that for any @xmath29 it holds @xmath30 2 . a smooth surface @xmath24 is called @xmath25-asymptotic outside a cylinder to two half - planes @xmath31 and @xmath32 if there exists a solid cylinder @xmath33 such that : 1 .
the solid cylinder @xmath33 contains the boundaries of the half - planes @xmath31 and @xmath32 , 2 .
the set @xmath34 consists of two connected components @xmath35 and @xmath36 that are @xmath1-asymptotic to @xmath31 and @xmath32 , respectively . for example
the canonical grim reaper cylinder @xmath12 is asymptotic to the parallel half - planes @xmath37 and @xmath38 outside the solid cylinder @xmath39 where here @xmath40 is a positive real constant .
let us now state our main result .
let @xmath2 be a connected , properly embedded has no self - intersections .
] translating soliton with uniformly bounded genus on compact sets of @xmath0 and @xmath33 be a solid cylinder whose axis is perpendicular to the direction of translation of @xmath41 .
assume that @xmath24 is @xmath42-asymptotic outside the cylinder @xmath33 to two half - planes whose boundaries belongs on @xmath43
. then either 1 . both half - planes
are contained in the same vertical plane @xmath44 and @xmath45 , or 2 .
the half - planes are included in different parallel planes and @xmath24 coincides with a grim reaper cylinder .
let us make here some remarks concerning our main theorem . 1 .
notice that in the above theorem infinite genus a priori could be possible .
the assumption that @xmath24 has uniformly bounded genus on compact sets of @xmath0 means that for any positive @xmath46 there exists @xmath47 such that for any @xmath48 it holds @xmath49 where @xmath50 is the ball of radius @xmath46 in @xmath0 centered at the point @xmath51 . roughly speaking
, the above condition says that as we approach infinity the size of the holes of @xmath24 is not becoming arbitrary small and furthermore they are not getting arbitrary close to each other .
2 . we would like to mention here that nguyen @xcite constructed examples of complete embedded translating solitons in the euclidean space @xmath0 with infinite genus . outside a cylinder , these examples look like a family of parallel half - planes .
this means that the hypothesis about the number of half - planes is sharp .
very recently , dvila , del pino & nguyen @xcite and , independently , smith @xcite constructed examples of complete embedded translators with finite non - trivial topology . for an exposition of examples of translators
see also ( * ? ? ?
* subsection 2.2 ) .
ilmanen constructed a one - parameter family of complete convex translators , defined on strips , connecting the grim reaper cylinder with the bowl soliton @xcite .
note that the level sets of these translators are closed curves .
this means that our hypothesis of being asymptotic to two planes outside a cylinder is natural and can not be removed .
let us describe now the general idea and the steps of the proof .
as already mentioned , we will assume that @xmath17 .
without loss of generality we can choose the @xmath52-axis as the axis of rotation of @xmath33 .
first we show that the half - planes must be parallel to each other , they should be also parallel to the translating direction and that both wings of @xmath24 outside the cylinder must point in the direction of @xmath7 . then , after a translation in the direction of the @xmath53-axis , if necessary , we prove that the asymptotic half - planes @xmath31 and @xmath32 are subsets of the parallel planes @xmath54 and @xmath55 respectively , and that @xmath24 is contained in the slab between the planes @xmath56 and @xmath57 .
to prove this claim we study the @xmath53-coordinate function of @xmath24 in order to control its range . by the strong maximum principle
we conclude that the @xmath53-coordinate function can not attain local maxima or minima . to prove that @xmath58 we perform a blow - down " argument based on a compactness theorem of white @xcite for sequences of properly embedded minimal surfaces in riemannian @xmath59-manifolds .
the next step is to show that @xmath24 is a bi - graph over @xmath57 and that the plane @xmath60 is a plane of symmetry for @xmath24 . to prove this claim we use alexandrov s method of moving planes . in the sequel
we show that @xmath24 must be a graph over a slab of the @xmath61-plane .
thus , @xmath24 must have zero genus and it must be strictly mean convex . to achieve this goal we carefully investigate the set of the local maxima and minima of the profile curve @xmath62 performing again a blow - down " argument along the ends of the curve @xmath63 we deduce that @xmath24 looks like a grim reaper cylinder at infinity .
to finish the proof , we consider the function @xmath64 which measures the @xmath52-coordinate of the gau map @xmath9 of @xmath24 .
then , by applying the strong maximum principle to @xmath65 , we deduce that @xmath64 is identically zero .
this implies that the gau curvature of @xmath24 is zero and so @xmath24 must coincide with a grim reaper cylinder ( see ( * ? ? ?
* theorem b ) ) .
the structure of the paper is as follows . in section
[ sec : main_tools ] we introduce the tangency principle , the compactness and the strong barrier principle of white @xcite . in section 3
we present a lemma that will play a crucial role in the proof of our theorem .
this lemma ( lemma [ pro : limit_process ] ) asserts that every complete , properly embedded translating soliton in @xmath0 with the asymptotic behavior of two half - planes has a surprising amount of internal dynamical periodicity .
the main theorem is proved in section [ sec : proof_of_the_main_theorem ] .
we will introduce here the main tools that we will use in the proofs . according to this maximum principle ( see ( * ? ? ?
* theorem 2.1 ) ) , two different translators can not `` touch '' each other at one interior or boundary point . more precisely : [ thm : tangency_principle ] let @xmath66 and @xmath67 be embedded connected translators in @xmath0 with boundaries @xmath68 and @xmath69 . 1 .
@xmath70*interior principle*@xmath71 suppose that there exists a common point @xmath72 in the interior of @xmath66 and @xmath67 where the corresponding tangent planes coincide and such that @xmath73 lies at one side of @xmath74
. then @xmath66 coincides with @xmath67 .
@xmath70*boundary principle*@xmath71 suppose that the boundaries @xmath68 and @xmath69 lie in the same plane @xmath44 and that the intersection of @xmath73 , @xmath74 with @xmath44 is transversal .
assume that @xmath73 lies at one side of @xmath74 and that there exists a common point of @xmath68 and @xmath69 where the surfaces @xmath66 and @xmath67 have the same tangent plane .
then @xmath66 coincides with @xmath67 .
let @xmath75 be a surface in a @xmath59-manifold @xmath76 .
given @xmath77 and @xmath78 we denote by @xmath79 the tangent disc of radius @xmath46 .
consider now @xmath80 as a vector subspace of @xmath81 and let @xmath82 be the unit normal vector of @xmath80 in @xmath81 . fix a sufficiently small @xmath27 and denote by @xmath83 the solid cylinder around @xmath51 , that is latexmath:[\[w_{r,\varepsilon}(p):=\big\{\exp_p(q+t\nu_q):q\in d_r(p)\,\,\text{and}\,\ , exponential map of the ambient riemannian @xmath59-manifold @xmath76 . given a function @xmath85 ,
the set @xmath86 is called the graph of @xmath87 over @xmath88 .
let @xmath76 be a riemannian @xmath59-manifold and @xmath89 a sequence of connected embedded surfaces .
the sequence @xmath89 converges in the @xmath90-topology with finite multiplicity to a smooth embedded surface @xmath91 if : 1 .
@xmath91 consists of accumulation points of @xmath89 , that is for each @xmath92 there exists a sequence of points @xmath93 such that @xmath94 , for each @xmath95 , and @xmath96 .
2 . for all @xmath92 there exist @xmath97 such that @xmath98 can be represented as the graph of a function @xmath87 over @xmath88 .
3 . for all large @xmath95 ,
the set @xmath99 consists of a finite number @xmath100 , independent of @xmath101 , of graphs of functions @xmath102 over @xmath88 which converge smoothly to @xmath87 .
the multiplicity of a given point @xmath92 is defined to be the number of graphs in @xmath99 , for @xmath101 large enough .
note that although each surface of the sequence @xmath89 is connected the limiting surface @xmath91 is not necessarily connected .
however , the multiplicity remains constant on each connected component @xmath75 of @xmath91 .
for more details we refer to @xcite .
let @xmath89 be a sequence of embedded surfaces in a riemannian @xmath59-manifold @xmath76 . 1 .
we say that @xmath89 has uniformly bounded area on compact subsets of @xmath103 if @xmath104 for any compact subset @xmath105 of @xmath103 .
we say that @xmath89 has uniformly bounded genus on compact subsets of @xmath103 if @xmath106 for any compact subset @xmath105 of @xmath103 .
[ thm : compactness_theorem_brian_white ] let @xmath76 be an arbitrary riemannian @xmath59-manifold .
suppose that @xmath107 is a sequence of connected properly embedded minimal surfaces .
assume that the area and the genus of @xmath107 are uniformly bounded on compact subsets of @xmath103 .
then , after passing to a subsequence , @xmath107 converges to a smooth properly embedded minimal surface @xmath108 .
the convergence is smooth away from a discrete set denoted by @xmath109 .
moreover , for each connected component @xmath75 of @xmath91 , either 1 .
the convergence to @xmath75 is smooth everywhere with multiplicity @xmath110 , or 2 .
the convergence is smooth , with some multiplicity greater than one , away from @xmath111 .
now suppose that @xmath103 is an open subset of @xmath0 while the metric @xmath112 is not necessarily flat .
if @xmath113 , @xmath95 , converges to @xmath92 then , after passing to a further subsequence , either @xmath114 or there exists a sequence of real number @xmath115 tending to @xmath116 such that the sequence of surfaces @xmath117 , where @xmath118 converge smoothly and with multiplicity @xmath110 to a non - flat , complete and properly embedded minimal surface @xmath119 of finite total curvature and with ends parallel to @xmath120 .
a crucial assumption in the compactness theorem of white is that the sequence has uniformly bounded area on compact subsets of @xmath103 .
let us denote by @xmath121 the set where the area blows up .
clearly @xmath122 is a closed set
. it will be useful to have conditions that will imply that the set @xmath122 is empty . in this direction ,
white ( * ? ? ?
* theorem 2.6 and theorem 7.4 ) shows that under some natural conditions the set @xmath122 satisfies the same maximum principle as properly embedded minimal surfaces without boundary .
[ thm : controlling_area - blowup ] let @xmath76 be a riemannian @xmath59-manifold and @xmath107 a sequence of properly embedded minimal surfaces , with boundaries @xmath123 in @xmath76 .
suppose that : 1 .
the lengths of @xmath123 are uniformly bounded on compact subsets of @xmath103 , that is @xmath124 for any relatively compact subset @xmath105 of @xmath103 .
2 . the set @xmath122 of @xmath107 is contained in a closed region @xmath125 of @xmath103 with smooth , connected boundary @xmath126 such that @xmath127 at every point of @xmath126 , where @xmath128 is the mean curvature vector of @xmath126 at @xmath51 and @xmath129 is the unit normal at @xmath51 to the surface @xmath126 that points into @xmath125 .
if the set @xmath122 contains any point of @xmath126 , then it contains all of @xmath126 .
the above theorem is a sub - case of a more general result of white .
in fact the strong barrier principle of white holds for sequences of embedded hypersurfaces of @xmath130-dimensional riemannian manifolds which are not necessarily minimal but they have uniformly bounded mean curvatures . for more details
we refer to @xcite . due to a result of ilmanen @xcite
there is a duality between translators in the euclidean space @xmath0 and minimal surfaces in @xmath131 , where @xmath132 is the conformally flat riemannian metric @xmath133 and @xmath134 stands for the euclidean inner product of @xmath0 .
the metric @xmath132 will be called ilmanen s metric . in particular , every translator in the euclidean space @xmath0 is a minimal surface in @xmath131 and vice - versa .
the levi - civita connection @xmath135 of @xmath132 is related to the levi - civita connection @xmath136 of the euclidean space via the relation @xmath137 one can check that parallel transports and rotations with respect to the euclidean metric that preserve @xmath7 preserve the geodesics of @xmath131 .
moreover , one can easily verify that vertical straight lines and grim - reaper - type " curves , i.e. , images of smooth curves @xmath138 of the form @xmath139 are geodesics with respect to the ilmanen s metric . using the above mentioned transformations
we can construct all the geodesics of @xmath131 .
let now @xmath21 be a sufficiently small positive number and @xmath140 a point in @xmath0 such that @xmath141 and @xmath142 .
let us denote by @xmath143 the distance of @xmath51 from the plane @xmath144 with respect to the ilmanen s metric and by @xmath145 the euclidean distance of the point @xmath51 from the plane @xmath146 .
the distance @xmath143 is given as the length with respect to the ilmanen s metric of the smooth curve @xmath147 given by @xmath148 a direct computation shows that @xmath149 from the above formula we immediately obtain the following result which will be very useful in the last step of the proof of our theorem . [ asymptotic - lemma ] suppose that @xmath24 , regarded as a minimal surface in @xmath131 , is @xmath90-asymptotic to two parallel vertical half - planes @xmath31 and @xmath32 outside the cylinder @xmath33 . then the translator @xmath24 is also smoothly asymptotic to the above mentioned half - planes outside @xmath33 with respect to the euclidean metric .
the translating property is preserved if we act on @xmath24 via isometries of @xmath0 which preserves the translating direction . therefore ,
if @xmath150 is a vector of @xmath0 then the surface @xmath151 is again a translator .
based on white s compactness theorem , we can prove a convergence result for some special sequences of translating solitons .
more precisely , we show the following : [ pro : limit_process ] let @xmath24 be a surface as in our theorem .
suppose that @xmath152 is a sequence of real numbers and let @xmath89 be the sequence of surfaces given by @xmath153 then , after passing to a subsequence , @xmath107 converges smoothly with multiplicity one to a properly embedded connected translating soliton @xmath91 which has the same asymptotic behavior as @xmath24 .
recall that any translator @xmath154 can be regarded as a minimal surface of @xmath155 where @xmath132 is the ilmanen s metric .
notice that each element of the sequence @xmath107 has the same asymptotic behavior as @xmath24 .
without loss of generality , we can arrange the coordinate system such that @xmath156 by assumption our surface @xmath24 is @xmath1-asymptotic outside @xmath33 to two half - planes @xmath31 , @xmath32 ( see fig .
[ tilted_planes ] ) .
let now @xmath157 , @xmath158 be the unit inward pointing vectors of @xmath159 , @xmath160 , respectively . for
any @xmath28 consider the closed half - planes @xmath161 for @xmath162 and denote by @xmath163 , @xmath162 , the closed half - space of @xmath0 containing @xmath164 and with boundary containing @xmath165 and being perpendicular to @xmath166 .
moreover , consider the closed half - spaces @xmath167 for any @xmath162 . in the case where the sequence @xmath168 is bounded
, we can consider a subsequence such that @xmath169 .
then obviously @xmath89 converges smoothly with multiplicity one to the properly embedded translating soliton @xmath170 clearly @xmath91 has the same asymptotic behavior with @xmath24 .
let us examine now the case where the sequence @xmath168 is not bounded .
split each surface @xmath171 of the surface into the parts @xmath172 * claim 1 . * _ the sequences @xmath173 and @xmath174 have uniformly bounded area on compact sets .
_ _ proof of the claim .
_ let @xmath105 be a compact subset of @xmath103 and @xmath175 a ball of radius @xmath46 centered at the origin of @xmath0 containing @xmath105 .
denote by @xmath176 the projection of the surface @xmath177 to the closed half - plane @xmath178 .
hence we can parametrize @xmath179 by a map @xmath180 of the form @xmath181 where @xmath95 , @xmath182 is the standard basis of @xmath0 , @xmath183 is the angle between the vectors @xmath184 and @xmath157 and @xmath185 is a fixed point on @xmath186 . by taking @xmath21 very large
we can make sure that @xmath187 and @xmath188 are bounded by a universal constant @xmath189 .
hence , for any index @xmath95 we have that @xmath190 where @xmath191 is a constant depending on @xmath46 and @xmath192 is the euclidean area of @xmath176 .
note that @xmath192 is less or equal than the euclidean area of the projection of @xmath105 to the plane containing @xmath178 .
thus there exists a number @xmath193 depending only on @xmath105 such that @xmath194 consequently , @xmath195 has uniformly bounded area . similarly , we show that @xmath196 has uniformly bounded area and this concludes the proof of the claim .
* claim 2 . * _ the sequence of surfaces @xmath197 has uniformly bounded area on compact sets .
_ _ proof of the claim .
_ let us show a first that the sequence @xmath198 has uniformly bounded length on compact sets .
following the notation introduced in the above claim , each connected component of @xmath199 can be represented as the image of the curve @xmath200 given by @xmath201 for any index @xmath95 .
let @xmath105 be a compact set of @xmath103 , @xmath175 a ball of radius @xmath46 centered at the origin and containing @xmath105 .
denote by @xmath202 the projection of @xmath203 to @xmath204 .
estimating as in claim 1 , we get that @xmath205 where @xmath191 is a constant depending on @xmath46 .
thus , there exists a constant @xmath206 depending only on the compact set @xmath105 such that @xmath207 hence , the sequence @xmath198 has uniformly bounded length on compact sets .
recall now that the set @xmath122 is closed . from claim 1
it follows that @xmath122 is contained inside a cylinder .
consider now a translating paraboloid and translate it in the direction of the @xmath208-axis until it has no common point with @xmath122 .
then move back the translating paraboloid until it intersects for the first time the set @xmath122 ( see fig .
[ areablow ] ) . ] from the strong barrier principle of white ( theorem [ thm : controlling_area - blowup ] ) , the translating paraboloid is contained in @xmath122 .
but this leads to a contradiction , because now the area blow - up set @xmath122 is not contained inside a cylinder .
thus , @xmath122 must be empty and consequently @xmath209 has uniformly bounded area . since the parts @xmath210 , @xmath211 , @xmath212 have uniformly bounded area , we see that the whole sequence @xmath213 has uniformly bounded area . from our assumptions , also
the genus of the sequence is uniformly bounded .
the convergence to a smooth properly embedded translator @xmath91 follows from theorem [ thm : compactness_theorem_brian_white ] of white .
since each @xmath214 , @xmath162 , is a graph and each @xmath171 is connected , we deduce that the multiplicity is one everywhere . thus , the convergence is smooth .
moreover , observe that each component of @xmath215 , @xmath162 , can be represented as the graph of a smooth function @xmath216 which is the limit of the sequence of graphs generated by the smooth functions @xmath217 for any @xmath95 .
thus , the limiting surface @xmath91 has the same asymptotic behavior as @xmath24 .
the limiting surface @xmath91 must be connected since otherwise there should exist a properly embedded connected component @xmath75 of @xmath24 lying inside @xmath33 .
but then , the @xmath208-coordinate function of @xmath75 must be bounded from above , which is absurd .
this concludes the proof . as a first application of the above compactness result
we show that the half - planes @xmath31 and @xmath32 must be parallel to each other .
let @xmath24 be a translating soliton as in our theorem .
then , the half - planes @xmath31 and @xmath32 must be parallel to the translating direction .
moreover , if @xmath31 and @xmath32 are parts of the same plane @xmath44 , then @xmath24 should coincide with @xmath44 .
we follow the notation introduced in the last lemma .
assume to the contrary that the half - plane @xmath218 is not parallel to the translating direction @xmath7 .
let us suppose at first that the cosine of angle between the unit inward pointing normal @xmath219 of @xmath159 and @xmath184 is positive .
consider the strip @xmath220 given by @xmath221 for sufficiently large @xmath222 this slab does not intersects the cylinder @xmath33 . for fixed real numbers
@xmath223 let @xmath224 be the grim reaper cylinder @xmath225 by our assumptions , as @xmath21 becomes larger the wing @xmath226 of @xmath24 is getting closer to @xmath31 . by the asymptotic behavior of @xmath24 to two half - planes
, there exists @xmath227 large enough such that @xmath228 does not intersect @xmath229 .
then translate this grim reaper cylinder in the direction of @xmath230 . since @xmath31 is not parallel to @xmath7 ,
after some finite time @xmath231 either there will be a first interior point of contact between the surface @xmath229 and @xmath232 or there will exist a sequence of points @xmath233 in the interior of @xmath229 , with @xmath234 bounded and @xmath235 unbounded , such that @xmath236 the first possibility contradicts the asymptotic behavior of @xmath24 .
so let us examine the second possibility .
consider the sequence of surfaces @xmath89 given by @xmath237 for any @xmath95 . by lemma [ pro : limit_process ] , after passing to a subsequence , @xmath89 converges smoothly to a connected and properly embedded translator @xmath91 which has the same asymptotic behavior as @xmath24 .
but now there exists an interior point of contact between @xmath91 and @xmath232 , which is absurd .
similarly we treat the case where the cosine of the angle between @xmath219 and @xmath238 is negative .
hence both half - planes must be parallel to the translating direction @xmath7 .
suppose now that the half - planes @xmath31 and @xmath32 are contained in the same vertical plane @xmath44 .
without loss of generality we may assume that @xmath239 .
suppose to the contrary that the translator @xmath24 does not coincide with @xmath44 .
observe that in this case the @xmath53-coordinate function attains a non - zero supremum or a non - zero infimum along a sequence @xmath233 in the interior of @xmath24 , with @xmath234 bounded and @xmath235 unbounded . performing a limiting process as in the previous case we arrive to a contradiction .
therefore , the @xmath53-coordinate function must be zero constant and thus @xmath24 must be planar .
another application of the above compactness result is the following strong maximum principle .
[ max ] let @xmath24 be a translating soliton as in our theorem and assume that the half - planes @xmath31 and @xmath32 are distinct .
consider a portion @xmath75 of @xmath24 ( not necessarily compact ) with non - empty boundary @xmath240 such that the @xmath208-coordinate function of @xmath75 is bounded .
then the supremum and the infimum of the @xmath53-coordinate function of @xmath75 are reached along the boundary of @xmath75 i.e. , there exists no sequence @xmath93 in the interior of @xmath75 such that @xmath241 and @xmath242 or @xmath243 recall that from the above lemma the half - planes @xmath31 and @xmath32 must be parallel to each other and to the direction @xmath7 of translation . from our assumptions the @xmath53-coordinate function of the surface @xmath24 is bounded .
moreover , the extrema of @xmath53 can not be attained at an interior point of @xmath75 , since otherwise from the tangency principle @xmath75 should be a plane .
this would imply that @xmath24 is a plane , something that contradicts the asymptotic assumptions .
so , let us suppose that there exists a sequence of points @xmath233 in the interior of @xmath75 such that @xmath244 and @xmath245 is tending to its supremum or infimum .
then , consider the sequence of surfaces @xmath89 given by @xmath237 for any @xmath95 . by lemma [ pro : limit_process ] , after passing to a subsequence , @xmath89 converges smoothly to a connected and properly embedded translator @xmath91 which has the same asymptotic behavior as @xmath24 .
but now there exists a point in @xmath91 where its @xmath53-coordinate function reaches its local extremum , which is absurd .
the @xmath53-coordinate function of @xmath24 satisfies the partial differential equation @xmath246 however , lemma [ max ] is not a direct consequence of the strong maximum principle for elliptic pde s because in general @xmath75 is not bounded .
we have to deal only with the case where @xmath31 and @xmath32 are distinct and parallel to @xmath7 .
we can arrange the coordinates such that @xmath17 and such that the @xmath52-axis is the axis of rotation of our cylinder @xmath247 following the setting in @xcite let us define the family of planes @xmath248 , given by @xmath249 moreover , given a subset @xmath250 of @xmath0 , for any @xmath251 we define the sets @xmath252 * step 1 : * we claim that both parts of @xmath24 outside the cylinder point in the direction of @xmath7 .
we argue indirectly .
let us suppose that one part of @xmath34 is asymptotic to @xmath258 and the other part is asymptotic to @xmath259 for some @xmath28 ( see fig .
[ comp - left ] ) .
fix real numbers @xmath223 and let @xmath224 be the grim reaper cylinder @xmath225 the idea is to obtain a contradiction by comparing the surface @xmath24 with an appropriate grim reaper cylinder @xmath224 .
let us start with the grim reaper cylinder @xmath260 note that @xmath261 lies outside the strip @xmath262 and it is asymptotic to two half - planes contained in @xmath263 and @xmath264 .
fix @xmath265 .
because outside a cylinder the grim reaper cylinder @xmath261 is asymptotic to two half - planes , there exists @xmath266 , depending on @xmath189 , such that @xmath267 is inside the region @xmath268 moreover , there exists @xmath269 , depending on @xmath189 , such that @xmath270 is inside the region @xmath271 consider now the grim reaper cylinder @xmath272 and choose @xmath273 large enough so that @xmath274 translate the above grim reaper cylinder in the direction of @xmath275 . since @xmath276 , we see that after some finite time @xmath222 either there will be a first interior point of contact between @xmath24 and @xmath277 or there will exist a sequence @xmath233 of points in @xmath24 , with @xmath234 bounded and @xmath235 unbounded , such that @xmath278 as in lemma [ max ] , we deduce that both cases contradict the asymptotic behavior of @xmath24 .
therefore , both parts of @xmath34 must point in the direction of @xmath7 .
the asymptotic assumptions on @xmath24 imply that the @xmath208-coordinate of @xmath75 is bounded .
therefore , due to lemma [ max ] , @xmath282 but since @xmath283 we have that @xmath284 for any @xmath285 , which is absurd .
thus @xmath286 .
observe that if equality holds , then a contradiction is reached comparing @xmath24 and the plane @xmath263 using the tangency principle .
hence @xmath287 .
similarly , we can prove that @xmath288 consequently , @xmath24 should lie inside the slab @xmath289 .
* step 3 : * using the same arguments we will prove now that @xmath290 . indeed , suppose at first that @xmath291 .
we can then place a grim reaper cylinder @xmath292 inside the slab @xmath289 , by taking @xmath293 sufficiently large , so that @xmath294 ( see fig .
[ comp-1 ] ) .
consider now the set @xmath295 let @xmath296 .
assume at first that @xmath297 . because @xmath298 it follows that there is an interior point of contact between @xmath24 and @xmath299 .
but then @xmath300 which leads to a contradiction with the asymptotic assumptions on @xmath24 .
let us treat now the case where @xmath301 . in this case
@xmath302 therefore , there exists a sequence of points @xmath303 in @xmath24 such that @xmath304 and @xmath305 consider the sequence @xmath306 by lemma [ pro : limit_process ] we know that after passing to a subsequence , @xmath89 converges to a connected properly embedded translator @xmath91 which has the same asymptotic behavior as @xmath24 . on the other hand
@xmath91 has an interior point of contact with @xmath299 and thus they must coincide . but this contradicts again the assumption on the asymptotic behavior of @xmath24
. thus @xmath307 must be less or equal than @xmath308 .
we exclude also the case where @xmath309 by comparing @xmath24 with a grim reaper cylinder from outside ( see fig . [ comp-2 ] ) .
_ proof of the claim .
_ recall that @xmath313 hence , from lemma 3.2 , we have that @xmath314 suppose now to the contrary that @xmath315 then , there exists a sequence @xmath316 of points of @xmath317 such that @xmath318 consider the sequence of surfaces @xmath319 . from lemma 3.1 we know that @xmath89 converges to a connected properly embedded translator @xmath91 which has the same asymptotic behavior as @xmath24 . on the other hand , there is an interior point of contact between @xmath91 and @xmath320 , which is a contradiction .
thus , @xmath321 which implies that @xmath322 in the same way , we can prove that @xmath323 this completes the proof of the claim .
[ graph-1 ] there exists a sufficiently large number @xmath273 such that the parts of @xmath324 are graphs over the @xmath61-plane , and there exists a sufficiently small @xmath28 such that @xmath325 is a graph over the @xmath61-plane .
_ proof of the claim . _ from step 3 we know that @xmath24 lies inside the slab @xmath326 since @xmath12 and @xmath34 are @xmath1-asymptotic to @xmath327 , we can represent each wing of @xmath34 as a graph over @xmath12 . fix a sufficiently small positive number @xmath328
then , there exists @xmath28 such that the interior of the right wing @xmath329 of @xmath34 can be parametrized by a smooth map @xmath330 given by @xmath331 where the map @xmath332 describes the position vector of @xmath12 , @xmath333 is the outer unit normal of @xmath87 and @xmath334 is a smooth function such that @xmath335 a straightforward computation shows that the outer unit normal @xmath9 of @xmath8 is given by the formula @xmath336 because @xmath8 is a translator , we deduce that its mean curvature is @xmath337 consequently , @xmath338 .
thus , each point of @xmath329 has an open neighborhood that can be represented as a graph over the @xmath61-plane .
due to lemma [ max ] , the surface @xmath329 must be connected .
indeed , assume to the contrary that @xmath329 has more than one connected component .
let @xmath75 be a connected component different from the one whose @xmath208-coordinate function is not bounded ( there is at least one by assumption ) . then due to lemma [ max ]
the infimum and the supremum of the @xmath53-coordinate function of @xmath75 are reached along the boundary , that is , @xmath75 is an open piece of the plane @xmath339 , so the whole surface @xmath24 must coincide with this plane , which is a contradiction .
moreover , its projection to the @xmath61-plane must be the simply connected set @xmath340 .
thus , @xmath329 must be a global graph over the subset @xmath340 of the @xmath61-plane .
similarly , we prove that also the left hand side wing of @xmath34 is graphical .
this completes the proof of the claim because by the hypothesis on the asymptotic behavior of @xmath24 , there exists a sufficiently large number @xmath273 such that @xmath341 .
* step 5 : * we shall prove now that @xmath24 is symmetric with respect to @xmath60 and that @xmath24 is a bi - graph over this plane .
the main tool used in the proof is the method of moving planes of alexandrov ( see @xcite ) .
let us define @xmath342 recall from ( * ? ? ?
* definition 3.1 ) that the relation @xmath343 means that @xmath344 is on the right hand side of @xmath345 .
we will prove that @xmath346 . in this case
we have that @xmath347 . by a symmetric argument
we can show that @xmath348 .
thus @xmath349 and the proof of this step will be completed .
the steps of the proof are the same as in ( * ? ? ?
* proof of theorem a ) with the difference that here we have to control the behavior of the gau map in the direction of the @xmath52-axis . _ proof of the claim . _ due to claim [ graph-1 ] it follows that given a sufficiently small number @xmath189 , there exists a positive number @xmath273 such that the surface @xmath353 can be represented as a graph over @xmath146 as well as a graph over the @xmath61-plane .
hence one can easily show that @xmath350 is a non - empty set .
following the same arguments as in ( * ? ? ?
* section 3 , proof of theorem a ) , we can show that @xmath350 is a closed subset of @xmath354 .
moreover if @xmath355 , then @xmath356 .
suppose now that @xmath357 .
then we will get at a contradiction , i.e. , we will show that there exists a positive number @xmath189 such that @xmath358 . _
condition @xmath110 _ : we will show at first that there exists a positive constant @xmath359 such that @xmath360 is a graph over the plane @xmath146 .
take a positive number @xmath183 and consider the sets @xmath361 @xmath362 and @xmath363 @xmath364 since @xmath365 is a graph over @xmath146 , there exists @xmath183 large enough such that @xmath366>0.\ ] ] we fix such an @xmath183 . from it
follows that there exists @xmath367 such that @xmath368 can be represented as a graph over the plane @xmath146 and furthermore @xmath369 let us now investigate the lower part of our surface @xmath370 .
because @xmath371 , we can represent @xmath372 as a graph over the plane @xmath146 .
note that there is no point in @xmath373 with normal vector included in the plane @xmath146 since otherwise @xmath373 and its reflection with respect to @xmath374 would have the same tangent plane at that point so by the tangency principle at the boundary @xmath24 would have been symmetric to a plane parallel to @xmath146 .
but this contradicts the asymptotic behavior of @xmath24 .
consequently , @xmath375 * assertion .
* _ there exists @xmath376 $ ] such that , for all @xmath377 $ ] , _
@xmath378 _ proof of the assertion .
_ suppose to the contrary that such @xmath379 does not exist .
this implies that for all @xmath95 there exists @xmath380 $ ] such that @xmath381 then there exists a sequence @xmath382 such that @xmath383 .
only two situations can occur , namely either the sequence @xmath384 is bounded or it is unbounded .
we will show that both cases lead to a contradiction .
if @xmath385 is bounded , then it should have a convergent subsequence that we do not relabel for simplicity .
denote its limit by @xmath386 .
note that @xmath386 belongs to the closure of @xmath373 .
hence , by the continuity of the gau map @xmath387 then @xmath388 which contradicts the relation .
let us now examine the case where the sequence @xmath389 is not bounded .
the first coordinate @xmath390 of @xmath391 is bounded .
the last coordinate @xmath392 of @xmath385 is also bounded .
therefore , the second coordinate @xmath393 of the sequence must be unbounded .
consider now the sequence @xmath394 .
due to lemma [ pro : limit_process ] , we have that after passing to a subsequence , @xmath395 converges smoothly to a properly embedded connected translator @xmath91 which has the same asymptotic behavior as @xmath24 . furthermore , the limiting surface @xmath91 has the following additional properties : 1 .
the surface @xmath396 can be represented as a graph over the plane @xmath146 .
the inequality @xmath397 holds true .
there exists a point in @xmath91 in which the gau map belongs to the plane @xmath146 . applying the tangency principle at the boundary of @xmath398 and @xmath399
we deduce that @xmath400 is a plane of symmetry for @xmath91 , something that contradicts the asymptotic behavior of @xmath91 .
this completes the proof of our assertion .
the relation implies that , for every @xmath377 $ ] , the surface @xmath401 can be represented as a graph over @xmath146 .
consequently , @xmath402 is a graph over @xmath146 for all @xmath403 .
hence the first condition in the definition of the set @xmath350 is verified . for a fixed real number @xmath273 consider the half - grim reaper cylinder " ( see fig .
[ comp - half - grim ] ) given by @xmath415 define now the set @xmath416 obviously , @xmath417 is a non - empty set .
moreover , if @xmath418 then @xmath419 let @xmath420 we claim that @xmath421 .
suppose this is not true .
if @xmath422 , then there would be an interior point of contact ( notice that the boundaries of both surfaces do not touch when @xmath423 ) .
this implies that @xmath424 , which contradicts the assumption on the asymptotic behavior of @xmath24 .
let us consider now the case where @xmath425 .
in this case there exists a divergent sequence @xmath426 such that @xmath427 because the asymptotic behavior of @xmath428 and @xmath429 is different and the distance between their boundaries is positive , then one can find constants @xmath430 and @xmath431 such that @xmath432 , for all @xmath433 .
so , @xmath434 tends to infinity .
now we can apply lemma [ pro : limit_process ] in order to deduce that the limit of the sequence @xmath89 , given by @xmath435 exists and has the same asymptotic behavior as @xmath24 .
let us call this limit @xmath436 .
but now @xmath436 and @xmath428 have an interior point of contact and thus they must coincide .
this leads again to a contradiction because @xmath436 and @xmath428 do not have the same asymptotic behavior .
hence , @xmath421 .
consequently , @xmath437 and @xmath429 have a boundary contact at @xmath51 .
observe that the tangent plane at @xmath51 of both surfaces is horizontal by step @xmath438 , and therefore by the boundary tangency principle they must coincide . _ case b : _ suppose now that there exists @xmath439 such that @xmath440 in this case , we compare @xmath429 with the family of half - grim reaper cylinders " @xmath441 and we proceed exactly as in the proof of case a. _ proof of the claim : _ recall that the profile curve @xmath442 lies inside the cylinder @xmath33 .
let @xmath443 take sequences @xmath444 and @xmath445 along the curve @xmath63 such that @xmath446 and define the sequences of translators @xmath447 , @xmath448 given by @xmath449 from lemma [ pro : limit_process ] we deduce that @xmath450 where @xmath451 and @xmath452 are connected properly embedded translators with the same asymptotic behavior as our surface @xmath24 .
consider the points @xmath453 and @xmath454 taking into account the way in which we have constructed our limits , we have that @xmath455 at this point , we can use claim [ cl : maxmin ] to conclude that the limits @xmath456 and @xmath457 are grim reaper cylinders , possibly displayed at different heights . from the definition of the limit and the second part of theorem [ thm : compactness_theorem_brian_white ]
, it follows that for large enough values @xmath458 there exist : 1 . strictly increasing sequences of positive numbers @xmath459 , @xmath460 , @xmath461 and @xmath462 satisfying @xmath463 for every @xmath458 , 2 . real smooth functions @xmath464 and @xmath465 satisfying the conditions @xmath466 for any @xmath458 , such that the surfaces @xmath467 and @xmath468 can be represented as graphs over grim reaper cylinders that are generated by the functions @xmath469 and @xmath470 , respectively . from the formula , by taking larger @xmath471 if necessary , we deduce that the strips @xmath472 and @xmath473 are strictly mean convex and so their outer unit normals are nowhere perpendicular to @xmath17 .
hence each point has a neighborhood that can be represented as a graph over the @xmath61-plane . because the strips @xmath474 , @xmath475 under consideration are smoothly asymptotic to strips of the corresponding grim reaper cylinders and because for the grim reaper cylinders it holds @xmath476
, we deduce that the projections of @xmath474 , @xmath475 to the @xmath61-plane are simply connected sets .
therefore , they can be represented globally as graphs over rectangles of the @xmath61-plane .
the boundary of each @xmath480 is piecewise smooth and consists of two lateral curves that converge to grim reapers and two top curves that converge to two parallel horizontal lines .
observe that in a strip @xmath481 around @xmath482 ( see again fig .
[ fig-9 ] ) the surface @xmath483 is a graph over the @xmath61-plane .
the proof will be concluded if we prove that there exists @xmath484 such that each @xmath480 is a graph over the @xmath61-plane , for any @xmath485 . indeed , at first fix a large height @xmath222 such that @xmath486 is a graph over the @xmath61-plane .
from claim [ wm ] we know that @xmath487 from the asymptotic behavior of @xmath24 we know that there exists a number @xmath488 such that @xmath489 now fix an integer @xmath490 , and suppose to the contrary that there is @xmath485 such that @xmath480 is not a graph over the @xmath61-plane . we will derive a contradiction .
let @xmath491 be the translation of @xmath480 in direction of @xmath7 .
take a number @xmath407 such that @xmath492 start to move back @xmath493 in the direction of @xmath230
. then there exists @xmath494 where @xmath495 intersects @xmath480 . from the choice of @xmath496
we see that the intersection points must be interior points of contact .
but then , from the tangency principle , it follows that @xmath497 , which is a contradiction .
therefore , for each @xmath498 the surface @xmath480 must be a graph over the @xmath61-plane . because @xmath499 is a compact exhaustion of @xmath24 we deduce that @xmath24 itself must be a graph over the @xmath61-plane .
in particular , @xmath500 . * step 7 : * from claim [ v - graph ] we see that our surface @xmath24 must be strictly mean convex .
consider now the @xmath52-coordinate of the gau map , i.e. , the smooth function @xmath501 given by @xmath502 where here @xmath503 . by a straightforward computation
( see for example the paper ( * ? ? ?
* lemma 2.1 ) ) we deduce that @xmath64 and @xmath11 satisfy the following partial differential equations @xmath504 and @xmath505 where @xmath506 stands for the squared norm of the second fundamental form of @xmath24 .
define now the function @xmath507 . combining the equations and we deduce that @xmath508 satisfies the following differential equation @xmath509 _ proof of the claim .
_ consider the sequence @xmath89 given by @xmath510 for any @xmath95 .
one can readily see that for any compact set @xmath105 of @xmath0 , it holds @xmath511 from the compactness theorem of white , the sequence of surfaces @xmath89 converges smoothly ( with respect to the ilmanen s metric ) to the union @xmath512 .
hence , due to lemma [ asymptotic - lemma ] , the wings of the translator @xmath24 outside the cylinder must be smoothly asymptotic to the corresponding wings of the grim reaper cylinder .
this completes the proof of the claim .
_ proof of the claim .
_ consider the compact exhaustion @xmath514 defined in the step 6 .
the boundary of each @xmath480 consists of four parts , namely : @xmath515 bearing in mind the asymptotic behavior of @xmath24 , we deduce that around each boundary curve line there exists a tubular neighborhood that can be represented as the graph of a smooth function over a slab of the grim reaper cylinder .
if @xmath516 is such a function then , from the equations and , we can represent @xmath508 in the form @xmath517 let us examine at first the behavior of @xmath508 along @xmath518 .
note that these curves belong to the wings of @xmath24 outside the cylinder .
fix a sufficiently small @xmath27 .
then , there exists @xmath269 and large enough index @xmath519 such that @xmath520 can be written as the graph over the grim reaper cylinder of a smooth function @xmath516 defined in the domain @xmath521 satisfying @xmath522 because for any fixed @xmath52 we have @xmath523 we get @xmath524 hence , for any @xmath525 , from equation we see @xmath526 because of the symmetry we immediately get that @xmath527 on the other hand , recall that the strips @xmath474 and @xmath475 are getting @xmath1-close to the corresponding grim reaper cylinders . hence , there exists an index @xmath528 such that for @xmath529 we can represent @xmath530 as the graph over a grim reaper cylinder of a smooth function @xmath469 defined in a slab of the form @xmath531 , where here @xmath532 depends only on @xmath533 , satisfying the properties @xmath534 exactly the same estimate can be obtained along the strips @xmath475 .
note that in this case the @xmath53-coordinate is not tending to @xmath535 and so @xmath536 is bounded from below by a positive number .
going now back to equation we obtain that for @xmath529 we have @xmath537 therefore @xmath538 becomes arbitrary small as @xmath101 tends to infinity .
this completes the proof of the claim . from claim [ cl : h_at_infinity ]
, there exists an interior point where @xmath508 attains a local maximum or a local minimum . from the strong maximum principle of hopf
we deduce that @xmath508 must be identically zero .
consequently , @xmath539 and thus @xmath503 is a tangent vector of @xmath24 . differentiating the equation @xmath540 , we deduce that @xmath541 .
thus , @xmath542 and so @xmath543 .
but then , from ( * ? ? ?
* theorem b ) , we deduce that @xmath24 should be a grim reaper cylinder .
* acknowledgments : * the authors would like to thank brian white and antonio ros for many stimulating and helpful conversations .
moreover , the authors would also like to thank oliver schnrer and leonor ferrer for plenty of useful discussions .
finally we would like to thank the referee for the valuable comments on the content of the manuscript and the suggestions for improving the paper . | in this article we prove that a connected and properly embedded translating soliton in @xmath0 with uniformly bounded genus on compact sets which is @xmath1-asymptotic to two planes outside a cylinder , either is flat or coincides with the grim reaper cylinder . |
the coherent dynamics of electrons in heterostructures have been thoroughly examined during the past decade for the case of the interband ultrafast excitation by a near - infrared ( ir ) pulse ( see ref . for review ) .
recently , a mid - ir pump have been also employed for the treatment of the coherent dynamics of electrons under the intersubband excitation 2 .
for example , a coherent transfer of electrons between tunnel - uncoupled states of a double quantum well ( dqw ) to the common excited state under mid - ir pump was considered in ref . .
moreover , a new type of semiconductor unipolar laser operating in the mid - infrared spectral region was demonstrated .
this type of device is based on a three - bound - state coupled dqw with a single - excited level and two coupled lower levels @xcite .
thus , an investigation of the coherent dynamics in the tunnel - coupled dqws under ultrafast mid - ir pump is now appropriate . in the present work
we carry out the theory of the ultrafast response on the intersubband excitation between the tunnel - coupled states and the single level , which can be ground or excited .
the study we will fulfill next is based on the quantum kinetic equation for the density matrix averaged over the pump frequency ( see evaluation in ref . ) .
we will discuss the effects of the intersubband transition peculiarities by means of the intersubband generation rate . with this purpose
we take into account the peculiarities of the intersubband excitation for two cases : @xmath0 when the electron transition occurs between the single - ground and the tunnel - coupled excited states , or @xmath1 when the transition takes place from the tunnel - coupled states to the single - excited state . to illustrate these scenarios
we have represented in fig.1 the band diagrams and the dispersion laws for two dqw samples of @xmath2 , with the layer widths of 150/130/40 @xmath3 and 150/20/120 @xmath4 , corresponding to the cases @xmath0 and @xmath1 , respectively .
we have chosen the dqw structures in such a way that the energy separation between the coupled sub - levels , @xmath5 , is about @xmath6 mev for both cases . in this context
the population relaxation is controlled by the @xmath7 phonon emission @xcite , while the dephasing of the tunnel - coupled states for the case @xmath1 is determined by the quasi - elastic scattering .
since the interwell re - distribution of the charge appears under a relatively low pump intensity , we have considered both the second order response and the nonlinear regime of oscillations .
moreover , we will compare the present results with the corresponding ones to the interband excitation case .
the paper is organized as follows . in sec .
ii we derive the balance equations , which describe the coherent response of electrons in dqws under the ultrafast intersubband excitation . in sec .
iii we discuss the emerging quantum beats and the peculiarities of the coherent response under the finite duration excitation , stressing the differences between the cases of intersubband and interband excitation .
iv contains the description of the nonlinear response .
the conclusions and discussion of the approximations used are done in the last section .
the coherent dynamics of the electrons , when photoexcited by an ultra - short pulse , is described below in the framework of the second order response on the intersubband excitation . performing the average over the period of the radiation we obtain the quantum kinetic equation for the density matrix , @xmath8 , in the following form ( see refs . ):
@xmath9=\hat{g}_{t}+\hat{i}_{sc},\]]where @xmath10 is the hamiltonian of the dqws under consideration , @xmath11 is the collision integral , and @xmath12 is the intersubband generation rate .
when the electrons are excited by a transverse electric field @xmath13 , with a frequency @xmath14 and a form - factor @xmath15 , the generation rate is given by @xmath16 e^{i\hat{h}\tau /\hbar } , \widehat{\delta h}% _ { t}^{+}\right ] + h.c.\]]here @xmath17 , the perturbation operator , @xmath18 , is written through the transverse velocity operator @xmath19 and @xmath20 is the equilibrium density matrix when the second - order contributions to the response are taken into account . neglecting the non - resonant mixing between the single and the tunnel - coupled levels
we describe the system by the scalar distribution function , @xmath21 , where @xmath22 correspond to the single electron state [ ground @xmath23 or excited @xmath24 state for the cases @xmath0 or @xmath1 , respectively ] , and by the @xmath25 matrix function @xmath26 which describes the tunnel - coupled states @xmath27 and @xmath28 ( upper and lower , respectively ) . within the framework of the momentum representation , with the in - plane momentum @xmath29 , eq .
( 1 ) is transformed into : @xmath30 & = & \hat{g}_{\mathbf{p}t}+\hat{% i}_{sc}(\hat{f}_{t}|\mathbf{p}),\end{aligned}\]]where @xmath31 is the matrix hamiltonian of the tunnel - coupled states , @xmath32 is the interlevel splitting energy , @xmath33 is the tunnel matrix element , and @xmath34 are the pauli matrices . here
the generation rates are different for the cases @xmath0 and @xmath1 . neglecting the overlap between @xmath35 , and doing the straightforward calculations of eq .
( 2 ) , we obtain for the case @xmath0 : @xmath36where @xmath37 is the ground state equilibrium distribution for the zero temperature case , @xmath38 is the fermi energy , and @xmath39 is the kinetic energy with the effective mass @xmath40 .
the dephasing time , @xmath41 , is introduced here instead of the @xmath42-parameter of eq .
( 2 ) with the aim of describing a finite broadening of the intersubband transitions . for the case
@xmath1 we use @xmath43 and the generation rate takes form : @xmath44where @xmath45 is the equilibrium density matrix of the tunnel - coupled levels . the detuning frequency in eqs .
( 4,5 ) , @xmath46 , is evaluated through the energy difference between single and tunnel - coupled levels , @xmath47 ( see fig.1 ) .
the remaining matrix elements in eqs .
( 4,5 ) are calculated by using the matrix equalities : @xmath48here @xmath49 is the frequency of oscillations due to transitions between tunnel - coupled levels , @xmath50 and @xmath51 /2 $ ] . when doing the summation over the 2d momenta we introduce the population of the single level , @xmath52 , and the @xmath25 matrix of concentration @xmath53 which is written through the scalar and vector components of the concentration , @xmath54 and @xmath55 . due to the particle conservation law , @xmath56 with the total 2d concentration @xmath57 , the system ( 3 )
is transformed into the balance equations : @xmath58 + \mbox{\boldmath $ \sigma$}% ( t)=\mathbf{g}(t),\]]where @xmath59 for the case @xmath0 or @xmath60 for the case @xmath1 and @xmath61 . the vector @xmath62 is defined as @xmath63 [ case @xmath0 ] or @xmath65 [ case @xmath1 ] . here
@xmath67 stands for the population relaxation time between single level and tunnel - coupled states , while the vector @xmath68 describes the dynamic properties of the tunnel - coupled electronic states .
the relaxation matrix in the case @xmath69 @xmath70 is determined by the non - zero components @xmath71 , where the dephasing relaxation time , @xmath72 , was introduced in ref . 9 for the case of elastic scattering in dqws .
the generation rates @xmath73 and @xmath74 $ ] are obtained from eqs .
( 4 - 6 ) in the form : @xmath75 & = & \frac{2t}{\delta _ { \scriptscriptstyle t}}\frac{\mathcal{n}w_{t}}{% \pi } \int_{-\infty } ^{0}\frac{d\tau } { \tau _ { p}^{2}}w_{t+\tau } e^{\tau /\tau _ { 2 } } \\ & & \times \left\ { a_{+}\left [ \begin{array}{l } -\cos ( \delta \omega + \omega _ { \scriptscriptstyle t}/2)\tau \\
\sin ( \delta \omega + \omega _ { \scriptscriptstyle
t}/2)\tau% \end{array}% \right ] -a_{-}\left [ \begin{array}{l } -\cos ( \delta \omega -\omega _ { \scriptscriptstyle t}/2)\tau \\ \sin ( \delta \omega -\omega _ { \scriptscriptstyle t}/2)\tau% \end{array}% \right ] \right\ } , \nonumber\end{aligned}\]]@xmath76 , \]]the photoinduced concentration in eqs .
( 8,9 ) is determined as : @xmath77with the characteristic pulse duration @xmath78 and the characteristic velocities @xmath79 equal to @xmath80 or @xmath81 for the cases @xmath0 or @xmath69 respectively .
the coefficients @xmath82 in eq .
( 8) are given by : @xmath83 [ moreover @xmath84 for the dqw @xmath0 ] while , in eq .
( 9 ) , @xmath85 for the case @xmath0 and @xmath86 for the dqw @xmath1 , where @xmath87 .
next , taking into account the coulomb renormalization of the tunnel - coupled levels , we have to replace @xmath88 in the matrix equation ( 3 ) by the hartree - fock hamiltonian , @xmath89 , written in the form ( see refs .
10 and 11 ) : @xmath90 .\]]here @xmath91 is the 3d wave vector , @xmath92 is the coulomb matrix element , and @xmath93 is the fourier transform of the electron density .
further transformations lead to the balance equation ( 7 ) with the renormalized vector @xmath94 written through the level splitting energy @xmath95where @xmath96 is the distance between the centers of @xmath97- and @xmath98-qws and @xmath99 is the dielectric permittivity supposed to be uniform across the dqws .
the signs @xmath100 and @xmath101 in eq .
( 12 ) correspond to the cases @xmath0 and @xmath102 respectively .
the evaluation of @xmath103 coincides with that done for the dqw @xmath0 in ref .
in this section we present a solution of the linear system of balance equations ( 7 ) , neglecting the second addendum in eq .
( 12 ) , for the cases of short and finite pulse duration . respecting the short - pulse approximation , if the pulse duration @xmath104 , the generation rates [ eqs .
( 8) and ( 9 ) ] take the forms : @xmath105 , @xmath106 , and @xmath107 with the @xmath108-like function : @xmath109 .
thus , the photoinduced redistribution of the concentration can be written as the step - like function : @xmath110 which is proportional to the step function @xmath111 if @xmath112 .
since the photoinduced dipole moment is expressed through @xmath113 , we obtain the @xmath114-component of @xmath115 in the form : @xmath116 + \cos \left ( \omega _ { % \scriptscriptstyle t}t\right ) \right\ } .\ ] ] for the short - pulse approximation , the differences between the above - presented results and those corresponding to the case of the interband excitation ( as considered in ref . )
are mainly attributable to the different characteristic concentrations and to the strong damping caused by the interband relaxation .
comparing eq .
( 10 ) with the characteristic concentration for the interband excitation , @xmath117 [ given by the eq . ( 18 ) in ref . ]
, we obtain @xmath118where the interband excitation is characterized by the kane velocity @xmath119 , the gap @xmath120 , the reduced density of states @xmath121 , and the field strength @xmath122 .
if @xmath123 , and the pulse is not too short ( @xmath1241ps ) , the ratio ( 14 ) is about 16 [ case @xmath125 and 26 [ case @xmath126 for the @xmath127-based structures with a total 2d - concentration @xmath128 @xmath129 @xmath130 and the dimensions used in fig.1 .
thus , the intersubband excitation appears to be more effective than the interband one .
the response seems to be more complicated for the finite pulse duration case due to the peculiarities of the relaxation processes .
we have used below the gaussian form - factor , @xmath131 $ ] , a semiempirical value of the damping @xmath132 ps @xcite , a dephasing time caused by the finite broadening of the intersubband transition @xmath133 ps 13,14 and an interband relaxation time due to lo phonons @xmath134 ps @xcite .
we consider first the evolution of the concentration .
fig.2 shows the evolution of @xmath54 with the increase of the pulse duration @xmath135 , for three detuning frequencies @xmath136 , @xmath137 , and @xmath138 [ figs .
2(@xmath139-@xmath140 ) , respectively ] and for the dqw @xmath0 . for dqw
@xmath141 the only difference is that the amplitude of the concentration @xmath54 is half of the corresponding to the structure @xmath142 because , initially , there are two occupied levels in dqw @xmath141 . therefore , we will pass by its interpretation , restricting ourselves to the case @xmath142 .
one can see a new non - monotonic behavior in contrast to the one of the interband excitation case @xcite . for @xmath143 @xmath144
behaves like in the interband case ( corresponding to the short pulse context ) with some type of oscillations superimposed . for @xmath145 @xmath144 these oscillations
are strongly amplified around @xmath146 for @xmath147 and @xmath148 , when the excited sublevel(s ) is(are ) not syntonized , before decaying .
it should be noted that the excitation pulse is centered at @xmath146 .
the number of oscillations depends on the pulse duration @xmath149 , as figs .
2@xmath150 display .
it is important to note that these oscillations have a period @xmath151 twice the @xmath113 quantum beats period because such oscillations are controlled by the term @xmath152 and strongly influence the initial stages of @xmath153 an exception takes place when one of the levels is syntonized , e.g. , @xmath154 .
then , the concentration shows a monotonous behavior with a growth rate similar to that of the interband pump [ fig . 2@xmath155 .
also visible in fig .
( 2 ) is the exponential damping of the photoexcited electrons caused by the dephasing time , @xmath41 .
figs.3 and 4 illustrate temporal evolution of the dipole moment , which is proportional to @xmath113 , for different regions of parameters , @xmath78 , @xmath156 , and @xmath157 . figs .
3(@xmath139 ) , 4(@xmath139 ) stand for the sample @xmath142 and figs .
3(@xmath158 ) , 4(@xmath158 ) for the sample @xmath141 , respectively .
the main difference between the finite pulse excitation and the short pulse excitation is the existence of two different regimes in the former event .
when @xmath147 and @xmath159 [ upper panels of figs .
3@xmath160 the finite duration pulse produces a transition from a regime in which the electron density is mainly located in a well to two - well oscillations .
this transition occurs when the pulse is switched off .
the dipole moment exhibits the biggest oscillation amplitude while the pulse holds , then decaying due to relaxation until reaching the equilibrium after switching off the pulse .
the balance situation is different for the two samples studied . in the first one the electronic redistribution between both wells
are quickly reached , because the photoexcited electrons of the coupled levels decay to the ground state by means of the lo phonon emission .
we must keep in mind that we are representing here the distribution @xmath113 corresponding to the coupled excited levels .
on the contrary , in the second sample , @xmath161 one can see the non - excited coupled levels . for this reason ,
the oscillations stay during some time until the electronic balance redistribution between the wells is reached because of the inter - subband dephasing relaxation .
the time @xmath72 for the last process is longer than that for the interband relaxation , @xmath67 ( see numerical values above ) .
figs . 3 and 4 show these features of the dipole moment in the cases of zero - phase shift ( @xmath162 ) and @xmath163-phase shift ( @xmath164 ) as indicated in figure captions .
3 has been calculated for @xmath159 , when the two tunnel - coupled states resonate and it corresponds to applied electric fields of 7 @xmath165 ( dqw @xmath142 ) and 2 @xmath165 ( dqw @xmath141 ) , respectively .
4 has been calculated for @xmath166 , out of the resonance of the tunnel - coupled levels .
in this situation the electronic concentration mainly occupies the left well and the oscillation amplitude becomes quenched . the influence of the detuning frequency when @xmath159 can be explained as follows .
if @xmath147 ( upper panel of fig .
3 ) , a fast transfer of the electron density from the well in which electrons were initially created to the other well occurs . for @xmath167 ( lower panel of figs .
3 ) , the electron density oscillates between coupled levels from the beginning of the excitation . out of the resonance between the tunnel - coupled levels ( @xmath168 , fig .
4 ) most of the electron density remains in the left well and the transfer does nt become effective because of level decoupling .
it is specially striking the practical disappearance of the oscillations when @xmath169 .
now we turn to the description of the nonlinear response . in order to do this
we will take into account the coulomb renormalization of the level splitting energy , when @xmath55 is governed by the nonlinear system of eqs .
( 7 ) , and @xmath94 is determined through eq .
the characteristic concentration , @xmath170 directly related to the pulse excitation density , is responsible for the nonlinearity . in order to get an effective coulomb renormalization we have used @xmath171 @xmath130 ( corresponding to an excitation energy density of about @xmath172 ) when the nonlinear response becomes noticeable .
5@xmath173 show the evolution of the dipole moment , @xmath174 corresponding to a characteristic concentration of @xmath175 at the coupled - level resonance ( @xmath176 , zero - phase shift , and for structures @xmath142 and @xmath1 , respectively .
we should always keep in mind that @xmath177 depends on @xmath178 .
thus , for a fixed excitation energy , we have a different @xmath179 values for each pulse duration .
the main result we can observe is that the oscillation period decreases and this is caused by a high @xmath179 value .
this period also depends on the detuning frequency . as a consequence of this dependency
, a slight coulomb - induced dephasing appears between different @xmath78 and @xmath157 cases .
this behavior is more noticeable in the structure @xmath141 than in @xmath0 because of the relation @xmath180 , which mainly determines coulomb effects in eq .
( 7 ) ( see ref . ) , is greater in the former case for the same characteristic concentration because of the different values of @xmath181 .
another feature induced by the coulomb interaction occurs while the excitation pulse is acting on the samples .
the term @xmath182 , which initially controls @xmath54 ( and dipole moment oscillations ) , loses part of its importance and the masking of the intersubband oscillations diminishes . by comparing fig . 5 with fig . 3
one can see a slight displacement of the electronic concentration to the left qw caused by the above mentioned coulomb renormalization when @xmath147 ( upper panels ) .
once again the detuning frequency plays the main role in the oscillatory behavior , leading to a concentration , which is located in the left well , one order of magnitude higher for @xmath137 than for @xmath147 .
such a bearing is common for both samples studied .
we have already shown ( fig .
4 ) that , being out of the resonance condition ( e.g. @xmath183 ) , differences produced by the detuning frequency are small and this kind of behavior remains when the coulomb renormalization is introduced [ figs .
6@xmath184 $ ] .
however , there is a clear dissimilarity between structures @xmath0 and @xmath1 . in the first sample
the electronic concentration oscillates between the two wells from when the excitation pulse is switched on [ fig . 6@xmath185 .
such behavior is caused by a new situation of resonance at @xmath168 .
to understand this point we must underline that the @xmath186-values corresponding to resonance and off - resonance are strictly defined for the linear response .
when the level renormalization is included resonance conditions vary and , hence , the electric fields to get them will also vary . in the other case , and for the same reason , electrons always prefer to stay mainly in the left qw [ fig .
these different behaviors are caused by the opposite sign in the expression for the coulomb level splitting renormalization ( eq .
finally , one can observe as a general bearing that the dipole moment oscillations are weak in the structure @xmath142 .
furthermore , for both structures , the temporal evolution of the dipole moment loses its oscillatory behavior almost completely when @xmath169 , the evolution depending essentially on the total concentration of excited electrons .
summarizing , we have described the coherent dynamics of electrons in dqws taking into account the peculiarities of the intersubband excitation and relaxation for transitions between single and tunnel - coupled states .
the temporal dependencies of the photoinduced concentration and the dipole moment are obtained both for the second order response and the nonlinear regime , when the splitting energy is renormalized by the photoexcited charge .
furthermore , we discuss the assumptions made .
both the tight - binding approximation for the description of the tunnel - coupled states and the use of the parabolic dispersion laws are valid for the dqws under consideration .
the simple relaxation time approach is also widely used for the description of similar structures . applying the single - particle description of the high - frequency response we have neglected the coulomb renormalization of the intersubband transitions due to depolarization and exchange effects , so that the nonlinear regime of the response under a not very low pump intensity may take place if @xmath188 is not very big . on the other hand
, we do not consider here the high - intensity pump case restricting ourselves to the inequality @xmath189 when there is no rabi oscillations .
all these conditions are satisfied for the concentrations and intensities used in sects .
iii and iv . to conclude , the peculiarities of coherent dynamics under the intersubband transitions of electrons described in sections iii and iv are interesting in order to select effective conditions both for the thz emission , observed only under the interband excitation , and for the photoinduced concentration redistribution ( see recent mid - ir measurements in a single qw @xcite ) .
it would also be interesting to verify scattering mechanisms by the use of this approach and to study the high - intensity pump , when an interplay between the nonlinear dynamics and rabi oscillations appears .
this case requires a special consideration .
f. rossi and t. kuhn , rev . of mod .
phys . , * 74 * , 895 ( 2002 ) . t. elsaesser and m. woerner 1999 , phys . rep . * 321 * , 254 ( 1999 ) .
p. tamborenea and h. metiu , phys .
a * 240 * 265 , ( 1998 ) ; m. rufenacht , s. tsujino , s.j .
allen , w. schoendeld , and p. petroff , physica status solidi b * 221 * , 407 ( 2000 ) ; s. tsujino , m. rufenacht , p. miranda , s.j .
allen , p. tamborenea , w. schoendeld , g. herold , g. lupke , t. lundstrom , p. petroff , h. metiu , and d. moses , physica status solidi b * 221 * , 391 ( 2000 ) .
julien , o. gauthier - lafaye , ph .
boucaud , s. sauvage , j - m .
lourtioz , v. thierry - mieg , and r. planel , in _ intersubband transitions in quantum wells _ , s.s . li and y - k .
kluwer academic pub .. ( boston , 1998 ) ; o. gauthier - lafaye , s. sauvage , p. baucaud , f.h .
julien , r. prazeres , f. glotin , j - m .
ortega , v. thierry - mieg , r. planel , j - p .
leburton , and v. berger , appl .
* 70 * , 3197 ( 1997 ) .
vasko , o.e .
raichev , phys .
b * 51 * , 16965 ( 1995 ) . f. t. vasko and a. kuznetsov , _ electronic states and optical transitions in semiconductor heterostructures .
_ springer , 1998 . c. gmachl , f. capasso , d.l .
sivco , a.y .
cho , reports on progr . in phys . *
64 * , 1533 ( 2001 ) .
note , that the same @xmath190 appears in the population balance equation and here because we neglect the overlap between @xmath191 and @xmath192 states .
vasko , o.e .
raichev , _ jetp _ * 81 * , 1146 ( 1995 ) .
raichev , phys .
b * 51 * , 17713 ( 1995 ) ; o.e .
raichev , f.t .
vasko , a. hernndez - cabrera , and p. aceituno , phys .
b * 56 * , 4802 ( 1997 ) . f.t .
vasko , _ jetp _ * 93 * , 1279 ( 2001 ) .
n. sekine , k. hirakawa , and y. arakawa , jpn .
. phys . * 37 * , 1643 ( 1998 ) ; f. wolter , h.g .
roskos , p.h .
bolivar , g. bartels , h. kurz , k. khler , h.t .
grahn , and r. hey , phys .
solidi b * 204 * , 83 ( 1997 ) .
y. lavon , a. saar , f. julien , j .-
leburton , and r. planel , in _ intersubband transitions in quantum wells _ , s.s .
li and y - k .
kluwer academic pub.(boston , 1998 ) .
unuma , t. takahashi , t. noda , and m. yoshita , appl .
. lett . * 78 * , 3448 ( 2001 ) .
m. woerner , r.a .
kaindl , f. eickemeyer , k. reimann , t. elsaesser , a.m. weiner , r. hey , and k.h .
ploog , physica b * 314 * , 244 ( 2002 ) . for different pulse duration values and for the structure @xmath0 .
pulse duration times are indicated by arrows and @xmath194.,title="fig : " ] for different pulse duration values and for the structure @xmath0 .
pulse duration times are indicated by arrows and @xmath194.,title="fig : " ] for different pulse duration values and for the structure @xmath0 .
pulse duration times are indicated by arrows and @xmath194.,title="fig : " ] for @xmath195 .. figures 3@xmath139 and 3@xmath158 correspond to structures ( @xmath196 ) and ( @xmath193 ) , respectively .
solid and dashed curves are plotted for @xmath197 ( zero - phase shift ) and for @xmath198 ( @xmath199-phase shift ) .
upper and lower panels correspond to @xmath200 and @xmath201 .
, title="fig : " ] for @xmath195 .. figures 3@xmath139 and 3@xmath158 correspond to structures ( @xmath196 ) and ( @xmath193 ) , respectively .
solid and dashed curves are plotted for @xmath197 ( zero - phase shift ) and for @xmath198 ( @xmath199-phase shift ) .
upper and lower panels correspond to @xmath200 and @xmath201 .
, title="fig : " ] for @xmath195 .
figures 5@xmath139 and 5@xmath158 correspond to dqws ( @xmath196 ) and @xmath1 , respectively .
solid and dashed curves are plotted for @xmath197 ( zero - phase shift ) and for @xmath198 ( @xmath199-phase shift ) .
upper and lower panels correspond to @xmath200 and @xmath201.,title="fig : " ] for @xmath195 .
figures 5@xmath139 and 5@xmath158 correspond to dqws ( @xmath196 ) and @xmath1 , respectively .
solid and dashed curves are plotted for @xmath197 ( zero - phase shift ) and for @xmath198 ( @xmath199-phase shift ) .
upper and lower panels correspond to @xmath200 and @xmath201.,title="fig : " ] | ultrafast intersubband excitation of electrons in tunnell - coupled wells is studied depending on the structure parameters , the duration of the infrared pump and the detuning frequency .
the temporal dependencies of the photoinduced concentration and dipole moment are obtained for two cases of transitions : from the single ground state to the tunnel - coupled excited states and from the tunnel - coupled states to the single excited state .
the peculiarities of dephasing and population relaxation processes are also taken into account .
the nonlinear regime of the response is also considered when the splitting energy between the tunnel - coupled levels is renormalized by the photoexcited electron concentration .
the dependencies of the period and the amplitude of oscillations on the excitation pulse are presented with a description of the nonlinear oscillations damping . |
blazars constitute one of the most extreme classes of active galactic nuclei ( agn ) , distinguished by their high luminosity , rapid variability , high ( @xmath9% ) optical polarization , radio core dominance , and apparent superluminal speeds @xcite . the broad band emission in these objects , which extends from the radio to the gamma ray band , appears to be dominated by non thermal processes from the heart of the agn , often undiluted by the thermal emission present in other agn .
therefore , blazars represent the ideal class to study to further our understanding of non thermal emission in agn .
the blazar class includes bl lacertae objects , characterized by an almost complete lack of emission lines , and a subclass of radio quasars ( which by definition display broad emission lines ) which have been variously called highly polarized quasars ( hpq ) , optically violently variable quasars ( ovv ) , core - dominated quasars ( cdq ) .
one of their observational properties which is easier to define is the flat spectrum radio emission ( @xmath10 ) and so we will refer to them as flat
spectrum radio quasars ( fsrq ) . given the lack of prominent emission lines in bl lacs , more than 95% of all known such objects have been discovered either in radio or x ray surveys .
up work on radio and x ray selected samples has shown that the two selection methods yield objects with somewhat different properties . the energy output of most radio selected bl lacs peaks in the ir / optical band @xcite ; such objects are now referred to as lbl ( low - energy peaked bl lacs ) .
by contrast , the energy output of most x ray selected bl lacs ( referred to as hbl : high - energy peaked bl lacs ) peaks at uv / x ray energies .
@xcite and @xcite have demonstrated that the difference in broad - band peaks for hbl and lbl is not simply phenomenological .
rather , it represents a fundamental difference between the two sub classes .
the location of the broadband peaks also suggests a different origin for the x ray emission of the two classes .
namely , an extension of the synchrotron emission likely responsible for the lower energy continuum in hbl , which typically display steep ( energy index @xmath6 ) x ray spectra , and inverse compton emission in lbl , which have harder ( @xmath11 ) spectra @xcite .
_ bepposax _ observations of bl lacs are confirming this picture @xcite . in this respect , one could expect to find a similar range in peak frequencies in the fsrq class for which , until recently , no evidence existed .
indeed , it was suggested by some authors @xcite , based upon the similarities of the optical x ray broad - band spectral characteristics of lbl and fsrq , that no fsrq with synchrotron peak emission in the uv / x ray band should exist .
two studies have drastically changed this picture : 1 .
the multifrequency catalog of @xcite identified more than 50 fsrq ( @xmath12% of the fsrq in their catalog ) spilling into the region of parameter space once exclusively populated by hbl ; 2 . about 30% of fsrq found in the deep x
ray radio blazar survey ( dxrbs ) @xcite were found to have x ray to radio luminosity ratios , @xmath13 , typical of hbl ( @xmath14 or @xmath15 ) , but broad ( fwhm @xmath16 ) and luminous ( @xmath17 ) emission lines typical of fsrq .
the discovery of a large population of `` x ray strong '' fsrq ( labeled hfsrq by @xcite to parallel the hbl moniker ) represents a fundamental change in our perception of the broadband emission of fsrq .
x ray observations of these objects play a fundamental role in finding their place within the blazar class .
for example , if the x ray spectra were found to be relatively steep , one could infer a dominance of synchrotron emission , as observed in hbl .
flatter x ray spectra , with corroborating evidence from the whole broad - band emission @xcite , would instead suggest inverse compton emission . in the latter case ,
the simple equations @xmath18 ( where by lfsrq we mean the `` typical '' fsrq with low - energy synchrotron peak ) and @xmath19 would not be valid , and some more complicated explanation for the existence of this class should be sought .
our previous knowledge of the x ray spectra of this new class of objects is scanty and mostly based on low signal to noise ratio ( s / n ) rosat data ( padovani et al .
1997 ; padovani et al .
, in preparation ) and is therefore also limited to the relatively narrow @xmath20 kev band .
the _ bepposax _ satellite @xcite , with its broad - band x ray ( @xmath21 kev ) spectral capabilities , is particularly well suited for a detailed analysis of the individual x
ray spectra of these sources .
in this paper we present _ bepposax _ observations of four hfsrq candidates , selected as described below . in
2 we present our sample , 3 discusses the observations and the data analysis , while
4 describes the results of our spectral fits to the _ bepposax _ data .
5 discusses the rosat data , 6 presents the spectral energy distributions and synchrotron - inverse compton fits to the data ,
7 discusses our results , while
8 summarizes our conclusions . throughout this paper spectral indices
are written @xmath22 and the values @xmath23 km s@xmath24 mpc@xmath24 and @xmath25 have been adopted .
readers wanting to skip the details of the data reduction and go directly to our results can read 2 , the summaries of our _ bepposax _ and rosat results in
[ sax_sum ] and [ rosat_sum ] , and then go straight to 6 .
the sample selection was done in two separate steps , as our objects were observed in the _ bepposax _ cycle 2 and 3 . for both cycles we defined as hfsrq flat
spectrum quasars with @xmath0 , @xmath1 , and @xmath26 values within @xmath27 from the mean values of the hbl in the multifrequency agn catalog of @xcite .
these are the usual effective spectral indices defined between the rest - frame frequencies of 5 ghz , 5,000 , and 1 kev .
x ray and optical fluxes have been corrected for galactic absorption .
the effective spectral indices have been k - corrected using the appropriate radio and x ray spectral indices , available for most sources ( otherwise mean values were assumed ) .
optical fluxes at 5,000 have been derived and k - corrected as described in padovani et al .
( in preparation ) .
we have chosen this definition so that our selection of hfsrq matches the traditional `` hbl box '' as detailed in , e.g. , @xcite .
.2 in lrrlllrrcccl name & ra(j2000 ) & dec(j2000 ) & @xmath28 & o & e & f@xmath29 & @xmath30 & @xmath0 & @xmath1 & @xmath26 & galactic n@xmath31 + & & & & & & mjy & & & & @xmath32 @xmath33 + wga j0546.6@xmath346415 & 05 46 41.8 & @xmath3464 15 22&0.323&16.0 & 14.7 & 287 & @xmath35&0.39&1.17&0.66 & 4.54 + rgb j1629@xmath364008 & 16 29 01.3 & @xmath3640 08 00&0.272&17.6 & 17.1 & 20 & @xmath37&0.34&0.93&0.54 & 0.85 + rgb j1722@xmath362436 & 17 22 41.2 & @xmath3624 36 19&0.175&16.5 & 15.1 & 35 & @xmath38&0.26&1.34&0.63 & 4.95 + s5 2116 + 81 & 21 14 01.2 & @xmath3682 04 48&0.084&14.6 & 13.3 & 376 & @xmath39&0.31&1.41&0.69 & 7.41 + lrcrcrcc name & lecs & lecs & mecs & mecs & pds & pds & observing date + & exp .
( s ) & count rate ( cts / s ) & exp .
( s ) & count rate ( cts / s ) & exp .
( s ) & count rate ( cts / s ) & + wga j0546.6@xmath346415 & 18564 & @xmath40 & 47234 & @xmath41 & 20004 & @xmath42 & 1998 oct 1 - 2 + rgb j1629@xmath364008 & 20512 & @xmath43 & 44759 & @xmath44 & 22798 & @xmath45 & 1999 aug 11 - 12 + rgb j1722@xmath362436 & 12175 & @xmath46 & 43993 & @xmath41 & 19732 & ... & 2000 feb 13 - 14 + s5 2116@xmath3681 & 13171 & @xmath47 & 28826 & @xmath48 & 13296 & @xmath49 & 1998 apr 29 + s5 2116@xmath3681 & 5575 & @xmath50 & 19674 & @xmath51 & 10488 & @xmath52 & 1998 oct 12 - 13 + in january 1998 , for cycle 2 , we selected the four x ray brightest hfsrq candidates then known .
three of these came from the agn catalog of @xcite , while the fourth one was the x ray brightest hfsrq in the dxrbs list of @xcite .
we were granted _
bepposax _ observing time for two of these sources . in january 1999 , for cycle 3 , we were able to select even more extreme candidates by using the rosat - green bank survey ( rgb ) @xcite .
being based on the rosat all - sky survey ( rass ) the rgb sample has an x ray flux limit higher than dxrbs ( @xmath53 erg @xmath33 s@xmath24 vs. @xmath54 erg @xmath33 s@xmath24 ) while it reaches slightly lower 5 ghz radio fluxes ( @xmath55 mjy vs. @xmath56 mjy ) .
rgb sources have then , by selection , higher @xmath57 ratios , that is lower @xmath26 , as needed to get more extreme ( with synchrotron peaks , @xmath58 , at higher energies ) hfsrq .
the identification of fsrq from the rgb sample is based on a cross - correlation with the nrao vla ( very large array ) sky survey ( nvss ) @xcite at 1.4 ghz and is described in detail in padovani et al .
( in preparation ) . amongst all fsrq in the hbl region with @xmath59 which , based on hbl spectral energy distributions ( seds ) , would correspond to @xmath60 hz @xcite , we again selected the four x ray brightest hfsrq candidates and , again , were granted _
bepposax _ observing time for two out of four sources .
the object list and basic characteristics are given in table 1 , which presents the source name , position , redshift , @xmath61 ( blue ) and @xmath62 ( red ) magnitudes from the automatic plate measuring ( apm ) @xcite , 5 ghz radio flux , radio spectral index , @xmath0 , @xmath1 , and @xmath26 values , and galactic n@xmath31 .
figure 1 shows the @xmath63 plane for fsrq and hbl and the location of our sources , well within the region which includes @xmath64% of hbl and clearly offset from where most fsrq are located .
note that , in fact , our objects have @xmath65 . for comparison , the fsrq with x ray data belonging to the 2 jy sample ( @xmath66% of the total ) , which include all `` classical '' fsrq ( e.g. , 3c 273 , 3c 279
, 3c 345 , 3c 454.3 : @xcite ) , have @xmath67 , which corresponds to an x ray to radio flux ratio @xmath2 times smaller .
we anticipate that our sources have powers which , although within the fsrq range , are relatively low , as discussed in [ powers ] and [ interpr ] . -2truecm -3.3truecm .2 in
-0.1truecm lllllllll name & n@xmath68 & @xmath69 & @xmath70 & @xmath71}$ ] & @xmath72}$ ] & norm & @xmath73/dof & f test , notes + & @xmath32 @xmath33 & & @xmath74jy & erg @xmath33 s@xmath24 & erg @xmath33 s@xmath24 & ( l / m ) & & fixed - free n@xmath31 + wga j0546.6@xmath346415 & 4.54 fixed & @xmath75 & @xmath76 & 3.84e-12 & 4.14e-12 & 0.82 & 1.07/41 & + & @xmath77 & @xmath78 & @xmath79 & 3.84e-12 & 4.15e-12 & 0.82 & 1.09/40 & 4% + rgb j1629 + 4008 & 0.85 fixed & @xmath80 & @xmath81 & 1.25e-12 & 7.96e-12 & 0.67 & 0.90/28 & + & @xmath82 & @xmath83 & @xmath84 & 1.27e-12 & 7.19e-12 & 0.67 & 0.88/27 & 79% + rgb j1722 + 2436 & 4.95 fixed & @xmath85 & @xmath86 & 1.07e-12 & 9.77e-13 & 0.81 & 0.63/18 & + & @xmath87 ( @xmath88 ) & @xmath89 & @xmath90 & 1.08e-12 & 8.85e-13 & 0.79 & 0.55/17 & 92% + s5 2116 + 81 & 7.41 fixed & @xmath91 & @xmath92 & 1.61e-11 & 1.77e-11 & 0.77 & 0.96/61 & 1998 apr 29 + & @xmath93 & @xmath94 & @xmath95 & 1.60e-11 & 1.87e-11 & 0.78 & 0.91/60 & 96% + & 7.41 fixed & @xmath96 & @xmath97 & 1.19e-11 & 1.43e-11 & 0.67 & 0.98/46 & 1998 oct 12 - 13 + & @xmath98 & @xmath99 & @xmath100 & 1.19e-11 & 1.44e-11 & 0.67 & 1.00/45 & 16% + & 7.41 fixed & @xmath101 & @xmath102 & & & 0.70 & 1.01/63 & sum + & @xmath103 & @xmath104 & @xmath105 & & & 0.70 & 0.97/62 & 92% + -0.3truecm llllllllll name & n@xmath68 & @xmath106 & @xmath107 & e@xmath108 & @xmath70 & @xmath71}$ ] & norm & @xmath73/dof & f
test + & @xmath32 @xmath33 & & & kev & @xmath74jy & erg @xmath33s@xmath24 & ( l / m ) & & + wga j0546.6@xmath346415 & @xmath109 ( @xmath110 ) & @xmath111 & @xmath112 & @xmath113 & @xmath114 & 3.91e-12 & 0.75 & 0.99/38 & 88% , 94% + rgb j1629 + 4008 & 0.85 fixed & @xmath115 & @xmath116 & @xmath117 & @xmath118 & 1.22e-12 & 0.67 & 0.94/26 & 32% + rgb j1722 + 2436 & 4.95 fixed & @xmath119(@xmath120 ) & @xmath121 & @xmath122(@xmath123 ) & @xmath124 & 1.08e-12 & 0.80 & 0.59/16 & 75% + s5 2116 + 81 & 7.41 fixed & @xmath125 & @xmath126 & @xmath127 & @xmath128 & 1.60e-11 & 0.77 & 0.92/59 & 92% , 4/29/98 + a complete description of the _ bepposax _ mission is given by @xcite .
the relevant instruments for our observations are the coaligned narrow field instruments ( nfi ) , which include one low energy concentrator spectrometer ( lecs ) @xcite sensitive in the 0.1 10 kev band ; three ( two after may 1997 ) identical medium energy concentrator spectrometers ( mecs ) @xcite , covering the 1.5 10 kev band ; and the phoswich detector system ( pds ) @xcite , sensitive in the @xmath129 kev band , coaligned with the lecs and the mecs .
a journal of the observations is given in table 2 .
the data analysis was based on the linearized , cleaned event files obtained from the _ bepposax _ science data center ( sdc ) on - line archive @xcite .
the data from the two mecs instruments were merged in one single event file by the sdc , based on sky coordinates .
the event file was then screened with a time filter given by sdc to exclude those intervals related to events without attitude solution ( i.e. , conversion from detector to sky coordinates ; see @xcite ) .
this was done to avoid an artificial decrease in the flux .
as recommended by the sdc , the channels @xmath130 and above 4 kev for the lecs and @xmath131 and @xmath132 for the mecs were excluded from the spectral analysis , due to residual calibration uncertainties .
the spectral analysis was performed using the matrices and blank - sky background files released in november 1998 by the sdc , with the blank - sky files extracted in the same coordinate frame as the source file , as described in @xcite . because of the importance of the band below 1 kev to assess the presence of extra absorption or soft excess ( indicative of a double power
law spectrum ) , we have also checked the lecs data for differences in the cosmic background between local and blank - sky field observations , comparing spectra extracted from the same areas on the detector ( namely , two circular regions outside the @xmath133 radius central region , located at the opposite corners with respect to the two on - board radioactive calibration sources ) . we looked for time variability in every observation , binning the data in intervals from 500 to 4000 s , with null results except for rgb j1629@xmath364008 .
this source , in fact , clearly varied during the observation ( @xmath134-test probability of constancy @xmath135 , see fig . [ fig2 ] ) , doubling its mecs flux in about @xmath136 hours , with a more rapid decay of a factor of @xmath137 in about 7 hours at the end of the observation . the same pattern is present in the single mecs units light curves , and the background showed no significant flux variations . in spite of the flux variation ,
no appreciable spectral changes seem to be present .
the hardness ratio , defined as the ratio between the count rates in the 410 kev and 1.54 kev bands , is constant ( with a @xmath134-test probability @xmath138% ) .
we have also looked for possible differences extracting the spectra in the `` high '' and `` low '' states ( defined as @xmath139 or @xmath140 0.03 cts / s , respectively ) , finding no significant differences between the spectral indices .
the lecs data also show evidence of variability , although less significant than the mecs one ( @xmath134-test probability of constancy @xmath141 , depending on the bin size ) . we stress that our only variable source is also the only one whose x ray band
is clearly dominated by synchrotron emission ( see next section ) . -1.3truecm
the spectral analysis was performed with the xspec 10.0 package . using the program grppha in heasoft
, the spectra were rebinned with more than 20 counts in every new bin , using the rebinning files provided by the sdc .
various checks using different rebinning strategies have shown that our results are independent of the adopted rebinning within the uncertainties .
the lecs / mecs normalization factor was left free to vary in the range 0.651.0 , as suggested by the sdc @xcite .
we fitted the combined lecs and mecs data both with single and broken power law models , with galactic and free absorption .
the absorbing column density was parameterized in terms of n@xmath31 , the hi column density , with heavier elements fixed at solar abundances and cross sections taken from @xcite .
the galactic value was derived from the nh program at heasarc ( based on @xcite ) .
the n@xmath31 parameter was set free to vary for all sources to check for internal absorption and/or indications of a `` soft excess '' .
the main results of our single power law fits are reported in tab .
[ singpowbl ] .
the @xmath142-test probability quantifies how significant is the decrease in @xmath143 due to the addition of a new parameter ( free n@xmath31 ) .
the errors quoted on the fit parameters are the 90% uncertainties for one and two interesting parameters , for galactic and free n@xmath31 respectively .
the errors on the 1 kev flux reflect the statistical errors only and not the model uncertainties .
the results for the broken power
law fit to the data are presented in tab .
[ bknbl ] .
@xmath144 and @xmath145 are the soft and hard energy indices respectively . in this case
the @xmath142-test probability quantifies the decrease in @xmath143 due to the addition of two parameters ( from a single power
law fit to a broken power
law fit ) .
the errors quoted in this case are the 90% uncertainties for two interesting parameters .
we now discuss the details and results of the analysis for each objects .
this source is located near the large magellanic cloud ( lmc ) at @xmath146 from lmc x3 . towards the edges of the lecs and mecs images ( in the region at right ascension @xmath140 05@xmath147
46@xmath148 ) , apparently diffuse emission is visible , consistent with the emission of some sources in the lmc already catalogued through rosat pspc observations ( see @xcite ; hereafter hp99 ) .
the most intense source is at @xmath149 from the quasar , with a count rate @xmath150% that of the quasar in the mecs images .
it is spatially coincident with the source [ hp99]0049 , identified as a foreground star in hp99 . due to the presence of serendipitous sources we have used an extraction region of 4@xmath151 also for the lecs instrument .
the data are well fitted by a single power law model with galactic absorption .
the best fit plots and values are reported in fig .
[ fig3 ] and tab .
[ singpowbl ] . the obtained spectral index is flat ( @xmath152 ) , thus indicating an inverse compton power law origin of the emission , given the sed characteristics of this quasar ( see 6 ) .
this result at first glance does not seem to confirm the `` hbl like '' nature of this object , suggested by its location in the @xmath0 @xmath34@xmath1 plane .
we note , however , that a slightly better fit , even if not significantly improved , can be obtained with a broken power law model , allowing for a column density higher than galactic ( n@xmath31@xmath153 @xmath154 , @xmath142-test @xmath155% , where the two values refer to a comparison with a single power law with free and galactic n@xmath31 respectively ; see tab .
[ bknbl ] ) . in this case
, the steeper component below @xmath156 kev may indicate the presence of the tail of the synchrotron emission , as characteristic of `` intermediate '' bl lacs ( e.g. , giommi et al . 1999 ) .
variability could also play a role in this ( see [ 0546_rosat ] ) . -2truecm
-3.3truecm significant flux was detected also in the pds instrument up to @xmath157 kev , but with a value not compatible with that expected by a simple extrapolation of the lecs+mecs spectrum ( see fig .
[ fig3 ] , right ) . due to the high number of x ray sources in the field of view , we consider contamination the most likely explanation for this detection .
in fact , there are 9 [ hp99 ] sources within @xmath158 from the quasar , with pks 0552@xmath34640 , an object classified as agn in ned , at @xmath158 , and lmc - x3 , a well known and bright x ray binary source , at @xmath146 .
given the known x ray spectrum of the latter @xcite , we checked its possible contamination level using the published best fit model , after accounting for the off - axis efficiency of the instrument .
indeed , the hardest component , a power law of index @xmath159 @xmath160 @xcite , although steep , is quite strong ( @xmath161 at 1 kev ) , and provides a non - negligible contribution up to @xmath162 kev .
this source is located 37@xmath151 from the center of the well known cluster abell ( a ) 2199 .
the x ray emission from the hot gas of the cluster is clearly visible at the edges of the lecs and mecs images . to avoid contamination from the cluster ,
therefore , an extraction region of 6@xmath151 has been used for the lecs .
the best fit results are reported in fig .
[ fig4 ] and tab .
[ singpowbl ] .
the spectrum is well fitted by a single power
law with galactic values for the absorbing column density .
the resulting spectral index is quite steep ( @xmath159 @xmath163 ) , as typical for hbl objects .
this source therefore confirms its `` hbl like '' properties also in the x ray spectrum .
we stress that rgb j1629 + 4008 is the first fsrq ever found with synchrotron emission extending all the way to the x ray band .
the pds detected this source up to @xmath138 kev but the flux is more than an order of magnitude higher than expected from the lecs+mecs extrapolation ( see fig .
[ fig4 ] , right ) .
this is likely due to the cluster emission .
a2199 is a very x ray bright source ( t @xmath164 kev , f@xmath165}=1.2 \times 10^{-10}$ ] erg s@xmath24 , emission integrated over a region of radius 20@xmath151 ; @xcite and private communication ) and , being at 37@xmath151 from the quasar , is entirely within the pds field of view and near where the pds efficiency is still @xmath166% ( 42@xmath151 ) .
although its emission decreases exponentially towards higher energies , its high brightness still gives a flux in the 1540 kev band @xmath167 erg s@xmath24 , higher than the quasar lecs+mecs extrapolation ( @xmath168 erg s@xmath24 ) . as a check
, we therefore fitted the lecs+mecs+pds data including in the pds the best fit model for the a2199 spectrum @xcite , after accounting for the off - axis efficiency .
in such a case the pds data are well accounted for , with a total @xmath169@xmath170 ( as compared to 1.38 without the cluster contribution ) .
note that for completeness we also fitted the data for this source with a broken power
law with absorption fixed at the galactic value ( see tab . [ bknbl ] ) .
no improvement with respect to the single power
law fit case was obtained . -2.3truecm -3.8truecm
-2truecm -3.3truecm this source is the weakest of the four observed in the soft x ray range ( see the lecs count rates in tab . [ tab2 ] ) . to enhance the s / n ,
therefore , we used a 6@xmath151 extraction region for the lecs , as suggested by the sdc @xcite .
the source is not detected by the pds .
the data are well fitted by a flat ( @xmath171 ) single power law model with galactic absorption ( see fig .
[ fig5 ] and tab .
[ singpowbl ] ) .
the residuals , however , show the presence of some soft excess below [email protected] kev .
this is confirmed by the free n@xmath31 fit , which gives values below the galactic one . due to the low s / n and the narrow band affected , the significance is only marginal ( @xmath142-test @xmath172% , last column in tab .
[ singpowbl ] ) .
a fit with a broken power
law with the absorption fixed at the galactic value yields a steep low - energy component with @xmath159 @xmath173 , though unconstrained ( see tab .
[ bknbl ] ) .
the improvement with respect to the single power law fit and galactic absorption is however not significant ( @xmath174% ) the flatness of the x ray spectrum , as for wga j0546.6 + 6415 , does not confirm the hbl like " nature of this objects as suggested by the broad band indices .
however , the possible steeper component at soft x ray energies , if confirmed , may be attributed to the tail of the synchrotron emission , thus locating the observed x
ray band in the valley " between synchrotron and compton emissions ( see , in fact , fig .
[ fig7 ] ) . in this case
, the quasar could be the fsrq counterpart of intermediate " bl lacs .
again , variability could have also played a significant role .
this object was observed by _ bepposax _
twice , six months apart ( see tab .
[ tab2 ] ) . in both observations
the source was detected in the pds up to @xmath175 kev .
the source varied between the two observations ( @xmath176% ) , with the highest state in the first observation .
both datasets are well fitted by a flat single power law model ( see fig .
[ fig6 ] and tab .
[ singpowbl ] ) , with very similar spectral indices ( within the errors ) .
the pds data agree , within the errors , with the extrapolation of the lecs+mecs spectrum .
given the similarity of the spectral properties , we have also added the event files to obtain a higher s / n spectrum , whose fit is also reported in tab .
[ singpowbl ] .
the free n@xmath31 value agrees with the galactic one in the second dataset ( within the large errors ) , while there is a marginal indication of possible excess absorption in the first and the summed datasets , according to the @xmath142-test ( @xmath177% and @xmath178% , respectively ) .
a broken power law model with galactic absorption provides an equivalent fit ( giving roughly the same @xmath169 of the free n@xmath31 fit ) , with a flatter index ( @xmath159 @xmath179 ) below 1.1 kev ; however the significance of the improvement is slightly less ( @xmath142-test @xmath180% ; see tab .
[ bknbl ] ) , due to the higher number of parameters .
the flat spectral indices in both observations ( i.e. , flux states ) classify this source as a standard fsrq . to summarize our _ bepposax _ spectral results , the fitted energy indices are flat , @xmath181 , for all but one source .
the mean value is @xmath182 and the weighted mean is @xmath183 .
this latter value is heavily weighted towards rgb j1629 + 4008 , the only object with a steep ( @xmath6 ) slope . excluding this source we derive a mean value @xmath184 and a weighted mean @xmath185 .
three of our sources may have a spectrum which is more complex than a single power
law , as also illustrated by the residuals to the single power
law fits . a double power law model fitted to their spectra indicates , within the rather large errors , a low - energy ( @xmath186 kev ) excess for wga j0546.6 + 6415 and rgb j1722 + 2436 , with a flatter component emerging at higher energies . in fact , the spectra are concave with quite a large spectral change , with @xmath187 , and energy breaks around @xmath188 kev .
the hard x ray spectral index is @xmath189 , while @xmath190 .
although the fit is improved by using a double power law model , an @xmath142-test shows that the improvement is more suggestive than significant , with probabilities @xmath191% .
we do include rgb j1722 + 2436 here , despite its relative low @xmath142-test probability , as we believe that a best fit @xmath192 value a factor @xmath193 below the galactic one is highly unlikely and strongly suggestive of a soft excess .
we also note that by adding up the @xmath143 values in tab .
[ bknbl ] , an @xmath142-test shows that the improvement in the fit provided by a double power
law model _ for the two sources together _ is significant at the @xmath194% level .
s5 2116 + 81 , on the other hand , shows a hint of a _
component at low energies ( @xmath186 kev ) , with @xmath195 .
lccrrl obs .
i d & instrument & off axis & exposure time & net count rate & observing date + & & ( arcmin ) & ( sec . ) & ( cts / s ) & + rp140008n00 & pspc - c & 37.63 & 1334 & @xmath196 & 1990 jul 9 + rp140638n00 & pspc - c & 37.63 & 1500 & @xmath197 & 1990 jul 10 + rp140009n00 & pspc - c & 21.98 & 1543 & @xmath198 & 1990 jul 9 + rp140639n00 & pspc - c & 21.98 & 1628 & @xmath199 & 1990 jul 10 + rp400078n00 & pspc - b & 51.66 & 7490 & @xmath200 & 1993 apr 10 + llllll obs .
& n@xmath31 & @xmath159 & @xmath70 & @xmath72}$ ] & @xmath73/dof + & @xmath32 @xmath33 & & @xmath74jy & erg @xmath33 s@xmath24 & + pspc - b & 4.54 fixed & @xmath201 & @xmath202 & 8.69e-12 & 0.61/31 + pspc - c sum8 & 4.54 fixed & @xmath203 & @xmath204 & 3.06e-12 & 1.09/16 + pspc - c sum9 & 4.54 fixed & @xmath205 & @xmath206 & 3.34e-12 & 1.01/14 + pspc - c fit total & 4.54 fixed & @xmath207 & @xmath208 & 3.22e-12 & 1.00/32 + in short , one of our sources appear to show an `` hbl like '' x - ray spectrum , while the other three are characterized by flat `` lbl like '' spectra .
one , or possibly two , of these sources , however , show hints of concave x - ray spectra typical of `` intermediate '' bl lacs .
in order to study possible variations of the x - ray spectral properties ( and synchrotron peak ) in these objects and have a better understanding of the underlying emission processes , we have also looked for other x ray observations available , in particular from the rosat archive , to take advantage of the higher resolution and collecting area at low energies .
we found position sensitive proportional counter ( pspc ) observations for two sources , namely wga j0546.6@xmath346415 and rgb j1629 + 4008 .
these objects are located near two famous and well observed x
ray sources , i.e. , lmc x3 and a2199 respectively , and thanks to the wide field of view of the rosat satellite we could find observations where our targets were present as serendipitous sources .
the journals of the rosat observations are reported in tab .
[ 0546rlog ] and tab .
[ 1629rlog ] . a third object , rgb j1722@xmath362436 , was detected in the rass @xcite .
we now present the results of the rosat analysis , along with a comparison with the _ bepposax _ results .
many pspc observations of the large magellanic cloud region are present in the rosat public archive , both for b and c detectors , but most have short exposures ( @xmath209 s ) and/or with the quasar at high off - axis angles .
we therefore decided to analyze the best ones , i.e. , the longest observation ( 7490 s , pspc - b , pointed on lmc - x3 ) , and those with the lowest off - axis displacement ( pspc - c ) .
table [ 0546rlog ] reports the basic information , namely the observation sequence i d code , the detector used , the off - axis position of our quasar in the field of view , the exposure time , the net count rate full band and date of observation .
the pspc - c observations took place in the same epoch ( 1990 july 9 and 10 ) , with the satellite often changing position between the two pointings ( thus the two off - axis positions in the same day ) .
the many short observations have been then reduced and summed together in a single dataset by the rosat data center .
we fitted all the pspc - c datasets independently , with a single power
law model plus galactic absorption , obtaining good fits with similar values for the spectral indices and normalizations , within the errors . to increase the s / n
, therefore , we summed the event files for the two observations with the same source detector coordinates ( i.e. , same off - axis angle ) .
the fit results are reported in tab .
[ 0546rfit ] , where `` sum8 '' refers to the summed event files .. 008n00 and .. 638n00 , and `` sum9 '' refers to the summed event files .. 009n00 and .. 639n00 . given the similarity of the values ,
we have also fitted the two datasets together ( sum8 and sum9 ) , using the same model parameters for both ( including normalization ) .
the resulting fit is good ( @xmath169@xmath210 ) thus confirming that both datasets have compatible spectral properties .
llllll obs .
& n@xmath31 & @xmath159 & @xmath70 & @xmath72}$ ] & @xmath73/dof + & @xmath32 @xmath33 & & @xmath74jy & erg @xmath33 s@xmath24 & + rp150083n00 & 0.85 fixed & @xmath211 & @xmath212 & 8.15e-12 & 2.16/53 + & @xmath213 & @xmath214 & @xmath215 & 7.43e-12 & 2.17/52 + rp800644n00 & 0.85 fixed & @xmath216 & @xmath217 & 9.68e-12 & 2.12/52 + & @xmath218 & @xmath219 & @xmath220 & 9.39e-12 & 2.14/51 + rp701507n00 & 0.85 fixed & @xmath221 & @xmath222 & 9.48e-12 & 1.37/43 + & @xmath223 & @xmath224 & @xmath215 & 6.74e-12 & 0.95/42 + all datasets analyzed are well fitted by single power
law models with galactic column density .
the free n@xmath31 fits provide values around the galactic ones , and other absorbing models and/or a broken power law model do not improve the fits .
the results seem to indicate a variation of the spectral properties for this source : the 1993 pspc - b spectrum is characterized by a steep `` hbl like '' slope ( @xmath225 ) , contrary to what obtained by both the 1990 pspc - c and 1998 _ bepposax _ observations , which agree with each other reporting a flat and roughly equal spectral index ( @xmath226 and @xmath227 respectively ) .
the 1990 and 1998 flux levels are similar , too , with the _ bepposax _ flux about 30% higher than the rosat one . compared to these , instead , in 1993 the quasar was in a high state , with a 0.12.4 kev flux more than 2.5 times higher than the 1990 values .
the x ray spectrum of this object , therefore , seems to have a `` steeper when brighter '' behavior , which may indicate a shift towards higher energies of the synchrotron peak during high states , when synchrotron emission dominates over inverse compton in the x ray band @xcite .
it must be noted , however , that these considerations are based on the 1993 pspc - b observation , which may be affected by possible calibration problems .
in fact , it has the highest off - axis angle ( it is near the edge of the field of view ) , and these regions are usually not well studied and calibrated .
in addition , also possible pspc - b spectral miscalibrations , as reported in @xcite and @xcite , must be taken into account , as these might steepen the rosat slopes by @xmath228 @xcite .
as for the previous object , we looked in the rosat public archive for the best observations possible , i.e. , the longest ones and/or those with the lower off - axis angle .
we found two very long observations of the cluster a2199 taken in 1990 and 1993 with the pspc - c and b instruments , respectively .
these included our target in the field of view with an offset of @xmath229 .
we found also a pointed observation , taken 5 days after the long exposure on a2199 .
the basic information for these datasets is reported in tab .
[ 1629rlog ] . lllllllll name & n@xmath68 & @xmath106 & @xmath107 & e@xmath108 & @xmath70 & @xmath72}$ ] & @xmath73/dof & f
test + & @xmath32 @xmath33 & & & kev & @xmath74jy & erg @xmath33 s@xmath24 & & + rp150083n00 & 0.85 fixed & @xmath230 & @xmath231 & @xmath213 & @xmath232 & 8.70e-12 & 1.47/51 & @xmath233% + rp800644n00 & 0.85 fixed & @xmath234 & @xmath235 & @xmath236 & @xmath237 & 9.95e-12 & 1.43/50 & @xmath233% + rp701507n00 & 0.85 fixed & @xmath238 & @xmath239 & @xmath240 & @xmath241 & 1.08e-11 & 0.91/41 & @xmath233% + & & n@xmath31 @xmath242 & @xmath243 & e@xmath244 & @xmath70 & @xmath72}$ ] & @xmath73/dof & f - test + & & @xmath245 & & kev & @xmath74jy & erg @xmath33 s@xmath24 & & + rp150083n00 & 0.85 fixed & @xmath246 & @xmath247 & @xmath248 & @xmath249 & 8.77e-12 & 1.17/51 & @xmath233% + rp800644n00 & 0.85 fixed & @xmath250 & @xmath251 & @xmath252 & @xmath253 & 9.78e-12 & 1.27/50 & @xmath233% + rp701507n00 & 0.85 fixed & @xmath254 & @xmath255 & @xmath256 & @xmath257 & 1.10e-11 & 1.04/41 & @xmath25899.8% + at first we fitted all datasets with a single power law model , both with galactic and free absorption , but these did not provide an adequate fit ( see @xmath169 in tab .
[ s1629 ] ) .
the presence of a more complex spectral shape is also indicated by the free n@xmath31 fit of the on - axis observation ( rp701507n00 ) , which provided a good @xmath169 only with absorption values below the galactic ones .
we therefore tried alternative models , reported in tab .
[ b1629 ] .
the fits were significantly improved both adopting a broken power law model with galactic absorption ( top three rows in tab .
[ b1629 ] ) , and adding a warm absorption component , at the redshift of the a2199 cluster , to the single power law model ( bottom three rows in tab .
[ b1629 ] ) .
this under the hypothesis that , at @xmath229 from the center , there is still a not negligible amount of cluster gas , likely partially ionized .
the model used for the warm absorber is zwndabs in xspec , an approximation to the warm absorber using the photoelectric cross - sections in @xcite .
the results are a bit puzzling and do not settle clearly the question of which model should be preferred .
the f - test does not help here , given that the number of model parameters is the same .
the off - axis observations seem to be better fitted by the warm absorber model , according to the @xmath134-test ( 15 - 20% against 2% , see the @xmath169 in tab . [ b1629 ] ) , but this is not confirmed by the on - axis observation , which has the best @xmath169 for the broken power law model .
moreover , the obtained values for the extra ( warm ) absorption are different ( at the 90% level ) , a fact which would imply a variable absorption with timescales less than five days ( the time interval between the last two observations , i.e. , rp800644 and rp701507 ) , quite unlikely .
for this reason , and possible feature calibration problems for the off - axis observations ( outside the pspc central rib ring ) , we consider the broken power law model a better representation of these data . in both cases , however , the spectral properties of the continuum for this object confirm its `` hbl like '' character showed in the _ bepposax _ observations , displaying steep spectra in all three datasets .
this result is confirmed also accounting for the possible rosat miscalibration ( i.e. , subtracting a maximum @xmath259 ) .
the broken power law fits , which show a flattening of the spectrum towards higher energies , may indicate that the inverse compton contribution is becoming not negligible , i.e. , that the region where the synchrotron and compton emissions cross was only slightly beyond the rosat band . this source is included in the rass .
@xcite give a @xmath20 kev flux @xmath260 erg @xmath33 s@xmath24 , i.e. , a factor @xmath137 larger than the _ bepposax _ value , with a spectral index derived from the hardness ratio @xmath261 , assuming galactic absorption .
this is steeper than the _ bepposax _ value and again suggests a `` steeper when brighter '' behavior . to summarize our rosat spectral results , the fitted energy indices are relatively steep , @xmath262 , steeper than the _ bepposax _ values by @xmath263 .
possible rosat miscalibrations are supposed to account for a steepening of the order @xmath264 , so we regard this difference as significant . in two cases the source was also brighter , suggesting a `` steeper when brighter '' behavior , indicative of a shift of the synchrotron peak towards higher energies during high states .
overall , this indicates a non - negligible synchrotron component in the x - ray band , which actually dominates for rgb j1629 + 4008 .
to address the relevance of our _ bepposax _ data in terms of emission processes we have assembled multifrequency data for all our sources .
the main source of information was ned , and the data are not simultaneous with our _ bepposax _ observations
. we also looked for near infrared observations in the two micron all sky survey ( 2mass ) second incremental release @xcite , finding data only for wga j0546.6@xmath346415 . given that our fits are mostly based on x - ray and optical data and that the latter are particularly important to determine @xmath58
, we need to address the non - simultaneity of the optical and x - ray observations .
the optical fluxes of fsrq are in fact known to show significant variability , which could influence our results .
we have then investigated the range of optical variability of our sources , with the following results .
* wga j0546.6@xmath346415 .
this is a newly identified source from the dxrbs ( perlman et al .
nevertheless , we did find variability information using supercosmos sky survey @xcite data , spanning the period 1975 1986 .
in particular , the uk schmidt telescope ( ukst ) and european southern observatory ( eso ) red observations , taken at the end of 1985/beginning of 1986 and separated by 50 days , agree within 0.01 magnitude , and give @xmath265 .
this is 0.5 magnitudes fainter than the @xmath266 magnitude given in the usno a2.0 catalogue , which refers to mid 1980 .
the guide star catalogue - ii ( gsc - ii ) , available at http://www-gsss.stsci.edu/ , gives @xmath267 and @xmath268 for 1989 , which indicates a brightening of @xmath269 magnitudes compared to the ukst blue value of 16.5 for 1975 .
* rgb j1629@xmath364008 .
this is a faint images of the radio sky at twenty centimeters ( first ) @xcite quasar . @xcite
have studied the long - term optical variability of first quasars .
they found that in 1953 palomar observatory sky survey ( poss i ) observations this object was @xmath270 magnitudes fainter in @xmath271 and @xmath269 magnitudes fainter in @xmath272 than in 1996 ccd observations .
the gsc - ii gives @xmath273 and @xmath274 for 1993 .
this is perfectly consistent with the ccd @xmath271 and @xmath272 values given by @xcite .
* rgb j1722@xmath362436 .
this source was identified as a quasar by @xcite due to its optical variability .
these authors quote a range of photographic magnitude for this source between 15.7 and 16.7 , with long `` still - stands '' between brief periods of activity .
the gsc - ii gives @xmath275 and @xmath276 for 1996 , in good agreement with the @xmath61 and @xmath62 apm values given in tab .
* s5 2116@xmath3681 .
the gsc - ii gives @xmath277 for 1993 , consistent with the apm @xmath61 value given in tab .
in summary , within the available optical data , our sources display maximum variability amplitude @xmath278 magnitudes .
we have then conservatively assigned to the optical fluxes an error of 0.2 dex , which corresponds to @xmath279 magnitudes ( see fig .
[ fig7 ] ) .
the seds for our sources are shown in figure [ fig7 ] .
the _ bepposax _
data have been converted to @xmath280 units using the xspec unfolded spectra after correcting for absorption .
rosat data are shown by a bow tie that represents the spectral index range . to derive the intrinsic physical parameters which could account for the observed data we have fitted the sed of our sources with a homogeneous , one
zone synchrotron inverse compton model as developed in @xcite .
this model is very similar to the one described in detail in spada et al .
( 2001 ; it is the `` one zone '' version of it ) , and is characterized by a finite injection timescale , of the order of the light crossing time of the emitting region ( as occurs , for example , in the internal shocks scenario , where the dissipation takes place during the collision of two shells of fluid moving at different speeds ) . in this model ,
the main emission comes from a single zone and a single population of electrons , with the particle energy distribution determined at the time @xmath281 , i.e. , at the end of the injection , which is the time when the emitted luminosity is maximized .
details of the model can be found in @xcite , who have applied it successfully to both low power bl lacs and powerful fsrq .
here we summarize its main characteristics .
the source is assumed cylindrical , of radius @xmath272 and width @xmath282 ( in the comoving frame , where @xmath283 is the bulk lorentz factor ) .
the particle distribution @xmath284 is assumed to have the slope @xmath285 [ @xmath286 above the random lorentz factor @xmath287 , for which the radiative cooling time equals the injection time .
the latter is assumed to be equal to the light crossing time ( i.e. , @xmath288 ) . the electron distribution is assumed to cut off abruptly at @xmath289 .
below @xmath290 there can be two cases , depending on the values of @xmath290 and @xmath291 : \i ) if @xmath292 , we have @xmath293 between @xmath291 and @xmath287 and @xmath294 below @xmath291 .
\ii ) alternatively , if @xmath295 , then @xmath296 between @xmath287 and @xmath291 and @xmath297 below @xmath287 . according to these assumptions , the random lorentz factor @xmath298 of the electrons emitting most of the radiation ( i.e. , emitting at the peaks of the seds ) is determined by the relative importance of radiative losses and can assume values in the range @xmath291 to @xmath299 .
our sources have broad lines and therefore the contribution of the disk to the sed could be relevant .
furthermore , photons produced in the broad line region could contribute to the seed photon distribution for the inverse compton scattering .
we accounted for this by assuming that a fraction @xmath300 10% of the disk luminosity @xmath301 is reprocessed into line emission by the broad line region ( blr ) , @xmath302 , assumed to be located at @xmath303 .
@xmath302 was estimated following @xcite from the fluxes of the strongest broad lines .
these were available for all our sources from our own spectrum ( wga j0546.66415 ; @xcite ) , from published spectra kindly made available to us by sally laurent
muehleisen ( rgb j1629@xmath364008 and rgb j1722@xmath362436 ) and from @xcite ( s5 2116@xmath3681 ) .
@xmath303 is assumed to scale as @xmath304 , following @xcite .
disk emission is assumed to be a simple black body peaking at @xmath305 hz . within these assumptions
, we can fix two important parameters for our modeling , and we can reliably estimate the importance of the seed photons produced externally to the jet for the compton scattering process .
the source is assumed to emit an intrinsic luminosity @xmath306 and to be observed with the viewing angle @xmath307 .
the model parameters are listed in table [ tab_sed ] , which gives the name of the source in column ( 1 ) , @xmath306 in column ( 2 ) , @xmath301 in column ( 3 ) , @xmath303 in column ( 4 ) , the magnetic field @xmath271 in column ( 5 ) , the size of the region @xmath272 in column ( 6 ) , the lorentz factor @xmath283 in column ( 7 ) , the viewing angle @xmath307 in column ( 8) , the slope of the particle distribution @xmath285 in column ( 9 ) , the minimum lorentz factor of the injected electrons @xmath291 in column ( 10 ) , and @xmath298 and @xmath308 in columns ( 11 ) and ( 12 ) respectively .
column ( 13 ) gives the magnetic field @xmath309 derived assuming equipartition of the energy content of the magnetic field with the electron energy .
note that @xmath306 , @xmath271 , @xmath272 , @xmath283 , @xmath307 , @xmath291 and @xmath285 are varied during the fitting procedure , @xmath301 and @xmath302 are fixed by observations , while @xmath298 , @xmath308 , and @xmath309 are derived quantities . in the case of a pure synchrotron self
compton model , all the above parameters are constrained in sources for which : 1 ) we have an estimate of the minimum timescale of variability ; 2 ) both the synchrotron and the self
compton peak are well defined ; 3 ) the spectral slopes below and above the peaks are known ; 4 ) the redshift is known .
as discussed in @xcite , this suffices to fix the values of the magnetic field , the intrinsic power of the source , the slopes of the emitting electron distribution , the relativistic doppler factor , and the dimension of the source .
when the radiation produced externally to the jet is important there is one unconstrained unknown , but the superluminal motion of the radio knots observed in blazars indicates values of the bulk lorentz factor in the range 1015 on average , and we therefore use these values for our fits ( see , e.g. , ghisellini et al .
1998 ) .
we do not have information about the high energy ( compton ) peak for the sources in our sample so we lack the direct determination of @xmath310 .
but in the model we use here the finite time of injection of particles plays a crucial role , and we have an additional constraint with respect to the simplest synchrotron inverse compton model .
namely , the peak of the synchrotron emission is either due to the electrons injected with @xmath311 , or it is due to the electrons which were able to radiatively cool in the timescale @xmath281 ( i.e. , electrons with lorentz factor @xmath290 ) . in the latter case ( which is verified for all the sources studied in this paper but rgb j1629@xmath364008 ) , the value of @xmath290 depends on the total energy density ( magnetic plus radiative ) as seen in the comoving frame . since the `` external '' radiation energy density is known once we specify the bulk lorentz factor ( it is assumed to be produced by the blr ) , @xmath290 depends on the magnetic field .
the break in the particle distribution at @xmath290 corresponds to a break of the emitted synchrotron spectrum ( and can often correspond to the peak of the synchrotron spectrum ) and in this way we can reliably estimate the value of the magnetic field . -0.1truecm
-2truecm -2truecm lllllllllllll name & @xmath306 & @xmath312 & @xmath313 & @xmath271 & @xmath272 & @xmath283 & @xmath307 & @xmath285 & @xmath314 & @xmath315 & @xmath316 & @xmath309 + & erg s@xmath24 & erg s@xmath24 & cm & g & cm & & & & & & hz & g + wga j0546.6@xmath346415 & 2.5e42 & 1.4e46 & 2.7e18 & 1.0 & 8.0e15 & 12 & 3.5 & 3.6 & 3.0e1 & 1.0e4 & 5.8e15 & 3.2 + rgb j1629@xmath364008 & 3.0e41 & 7.2e44 & 3.5e17 & 2.5 & 1.0e16 & 13 & 3.0 & 4.2 & 1.0e4 & 1.0e4 & 1.7e16 & 0.9 + rgb j1722@xmath362436 & 1.3e42 & 9.3e44 & 4.0e17 & 4.0 & 7.0e15 & 13 & 5.0 & 3.5 & 1.0e2 & 1.5e3 & 3.8e14 & 2.7 + s5 2116@xmath3681 & 2.0e42 & 1.9e45 & 7.0e17 & 1.0 & 1.0e16 & 11 & 5.0 & 3.5 & 2.0e1 & 3.0e3 & 3.9e14 & 2.3 + we are aware of the fact that for three of our objects the shape , sampling , and lack of simultaneity of the sed do not allow us to firmly constrain the peak of the synchrotron emission ( the exception being rgb j1629 + 4008 ) , also because of the non - negligible contribution of the thermal disk component to the optical flux ( especially for wga j0546.6@xmath346415 and s5 2116@xmath3681 ; see fig .
[ fig7 ] ) .
however , some reasonable arguments can be made .
for wga j0546.6@xmath346415 and rgb j1722@xmath362436 the presence of a steeper ( and variable ) component at soft x - ray energies ( [ sax_sum ] and [ rosat_sum ] ) may be attributed to the tail of the synchrotron emission ( see also sect .
[ bump ] ) , thus suggesting a synchrotron peak which can not be at too low energies ( @xmath317 hz ) .
this is corroborated also by the ir and optical data for rgb j1722@xmath362436 , whose optical flux is not dominated by the disk emission . in the case of wga j0546.6@xmath346415 , which has a stronger disk component , the synchrotron peak could reach @xmath318 hz only if one neglects the evidence for an upturn at low x - ray energies .
2116@xmath3681 is the less constrained source , but it is also the case for which we have no indication of a steep , soft x - ray component .
the radio to ir data suggest a peak above @xmath319 hz and no set of parameters was found to be compatible with the ir and _ bepposax _ data and with a @xmath308 less than @xmath320 hz .
for all three objects , however , the synchrotron peak frequency has to be @xmath321 hz , since otherwise the x - ray band would have a strong and obvious steep synchrotron emission ( as for rgb j1629@xmath364008 ) , contrary to what observed . in summary , we have modelled our objects mainly on the basis of the _ bepposax _ ( and partly rosat ) data , trying to fit the other data as far as they could be compatible with the _ bepposax _ ones .
the derived parameters , then , apart from rgb j1629 + 4008 , should be regarded as tentative .
the model fits are shown in fig .
[ fig7 ] as solid lines .
the applied model is aimed at reproducing the spectrum originating in a limited part of the jet , thought to be responsible for most of the emission .
this region is necessarily compact , since it must account for the fast variability shown by all blazars , especially at high frequencies .
the radio emission from this compact region is strongly self absorbed , and the model can not account for the observed radio flux .
this explains why the radio data are systematically above the model fits in the figure .
our model fails also to reproduce the optical flux of s5 2116 + 81 .
this is the source with the lowest redshift ( @xmath322 ) and we therefore expect a non negligible contamination from the host galaxy , which is clearly detected in the digitized sky survey ( dss ) plates .
note that the model fits predict that the high energy inverse compton emission has a luminosity comparable to the synchrotron one in all sources but one , s5 2116 + 81 .
the shape of the inverse compton component is sometimes complex , because of the contributions of the synchrotron self compton component , especially in the x ray range , and the external component , especially at higher energies .
as shown in table [ tab_sed ] , the intrinsic luminosities , the source dimensions , the bulk lorentz factors and the viewing angles are quite similar for all sources .
the fact that rgb j1629 + 4008 has a steep x ray ( synchrotron ) spectrum is explained by our model by the very large value of @xmath291 , the minimum lorentz factor of the injected electrons which in this case corresponds to @xmath298 .
the inferred values of the magnetic field are not very far from the equipartition values , with an average ratio @xmath323 .
our source selection requires the objects to be in the hbl region and therefore to have relatively low @xmath0 values .
given also their relatively low radio fluxes ( and powers ; see
[ interpr ] ) , before discussing our results we need to assess if our sources are of the same kind as the more powerful and more studied blazars .
we do this in the following subsection .
two different definitions of radio loud quasars have been used in the literature ( see , e.g. , stocke et al .
the first is based on the rest
frame ratio @xmath272 of radio ( typically 5 ghz ) to optical ( typically 4,400 ) flux density , @xmath324 ( independent of redshift if @xmath325 is assumed ) .
loud sources are taken to have @xmath326 which translates , in our notation , to @xmath327 .
the other definition is based on radio power and appears to be redshift - dependent , with a dividing line ranging from @xmath328 w hz@xmath24 for the palomar green ( pg ) sample ( optically bright and at relatively low - redshift ) up to @xmath329 w hz@xmath24 for the large bright quasar survey ( lbqs ) sample ( optically fainter and at higher redshift ) .
the two definitions overlap somewhat @xcite . as shown in fig .
[ fig1 ] and tab .
1 , all our sources are firmly in the radio - loud regime as far as the first definition is concerned .
all of them have @xmath330 , with @xmath331 ( or @xmath332 ; typical radio - quiet sources [ i.e. , optically selected quasars ] have @xmath333 or @xmath334 ) .
the 5 ghz radio powers of our sources are in the range @xmath335 w hz@xmath24 , with @xmath336 w hz@xmath24 . given their relatively low redshift ( @xmath337 ) , these values also classify our sources as radio - loud .
as described in the introduction , blazars are characterized by a variety of properties , which have also led to a proliferation of names ( hpq , ovv , cdq , fsrq ) .
the flat spectrum radio quasar definition is the easiest one to apply , and it appears that the majority of powerful fsrq indeed show rapid variability , high polarization , and radio structures dominated by compact radio cores , and vice versa ( urry & padovani 1995 and references therein ) . as our sources are less powerful than the fsrq commonly studied ( see [ powers ] ) , it is important to check , as far as possible , that we are dealing with the same type of sources .
we do this in the following .
our objects have , by selection , a flat radio spectrum .
in fact , @xmath338 , that is their typical radio spectrum is not only flat but inverted .
three of our sources are included in the nvss and therefore low - resolution ( 45@xmath339 ) radio maps at 1.4 ghz are available .
rgb j1629@xmath364008 and rgb j1722@xmath362436 are compact and unresolved .
rgb j1629@xmath364008 is also included in the first @xcite radio survey , which has a higher resolution ( 5@xmath339 ) than the nvss .
this source is still compact and unresolved in the first map , with a 1.4 ghz radio flux of @xmath340 mjy , as compared to @xmath341 mjy in the nvss .
the fact that the higher - resolution first flux is similar to ( actually larger than ) the nvss flux suggests that the source is core - dominated ( modulo variability effects ) .
s5 2116@xmath3681 has a complex nvss map , with four radio sources within 3@xmath151 from the source position .
however , dedicated radio observations @xcite show that the vla and vlba ( very long baseline array ) 5 ghz core fluxes of the stronger source are basically the same , again indicating a very compact core .
finally , our own 4.8 and 8.6 ghz australia telescope compact array ( atca ) maps of wga j0546.6@xmath346415 ( 2@xmath339 and 1@xmath339 resolution , respectively ) indicate a compact , unresolved source , with a core - dominance parameter at 4.8 ghz @xmath342 ( see landt et al .
2001 for details of the atca observations ) .
the fact that the higher - resolution atca flux at 5 ghz is similar to ( actually larger than ) the lower - resolution parkes - mit - nrao ( pmn ) @xcite flux ( 339 mjy vs. 287 mjy ) again suggests that the source is core - dominated ( modulo variability effects ) .
optical polarization data are available only for two of our sources , rgb j1722@xmath362436 and s5 2116@xmath3681 .
the former has @xmath343% @xcite while the latter has @xmath344% @xcite .
we note that although these values are below the `` canonical '' division between high - polarized ( hpq ) and low - polarized ( lpq ) quasars ( and are indeed not inconsistent with possible scattering by the interstellar medium of our own galaxy ) , many flat
spectrum lpq are also thought to be seen with their jets at small angles w.r.t .
the line of sight ( e.g. , ghisellini et al .
1993 ) , and therefore classify as `` blazars '' according to our definition . furthermore ,
if indeed these objects are the equivalent to hbl , as discussed in 7.2 , we would expect their duty - cycle for high polarization to be lower than their low - frequency peaked relatives , as was found for hbl and lbl by @xcite . finally , the relative prominent thermal emission , especially in s5 2116@xmath3681 ( see fig .
[ fig7 ] ) , would also explain a low optical polarization .
as regards variability , at least three of our sources appear to vary in the x ray band , based on our _ bepposax _ and rosat comparison , with rgb j1629 + 4008 decreasing in flux by a factor of 4 in 7 hours in our mecs and lecs observations .
the purpose of this paper was to study the x ray properties and , more generally , the sed of a class of flat spectrum radio quasars with relatively large x ray to radio flux ratios and effective spectral indices typical of hbl .
we now address the question if these sources are really the strong , broad - lined counterparts of hbl .
the sed of rgb j1629 + 4008 is hbl like , with the steep x ray spectrum clearly attributable to synchrotron emission .
for wga j0546.6@xmath346415 and rgb j1722 + 2436 the situation is less clear .
our fits and the seds are not inconsistent with the fact that the _ bepposax _ band might be close to the `` valley '' between synchrotron and inverse compton emission , which would classify them as akin to `` intermediate '' bl lacs .
finally , s5 2116 + 81 appears to be lbl like . a good indicator of the dominant emission process in the x ray band is the x ray spectral index
synchrotron peak frequency plot @xcite .
@xcite found a strong anti
correlation between the rosat @xmath345 and @xmath58 for hbl ( i.e. , the higher the peak frequency , the flatter the spectrum ) , while basically no correlation was found for lbl .
this was interpreted as due to the tail of the synchrotron component becoming increasingly dominant in the rosat band as @xmath58 moves closer to the x ray band ( see fig .
7 of padovani & giommi 1996 ) .
the _ bepposax _ version of this dependence was studied by padovani et al .
( 2001 ; see their fig .
they confirmed the rosat findings , namely a strong anti - correlation between @xmath159 and @xmath58 for hbl and no correlation for lbl , with an initial increase in @xmath159 going from lbl to hbl ( see padovani et al .
2001 for details ) . 0.2 in fig .
[ fig8 ] shows an updated version of the @xmath346 plot for bl lacs from @xcite , including the hbl studied by @xcite .
we have also included in this plot our sources , represented by filled points .
we notice that rgb j1629 + 4008 falls right in the region occupied by hbl , again confirming our interpretation that its _ bepposax _ band is dominated by synchrotron emission .
this is further supported by the detection of rapid variability , which has been seen in the synchrotron emission of several bl lacs observed with _
bepposax _ but never in the inverse compton component , which varies on much longer time scales ( e.g , @xcite and references therein ) .
the other three sources have values which , although quite uncertain , are probably in the range @xmath347 hz , which overlap with the values typical of `` intermediate '' bl lacs , but display relatively flat @xmath159 . we notice that the scatter in the diagram is larger in the @xmath58 range @xmath348 hz , which is also where the hbl / lbl division becomes blurrier .
we interpret this as due to the fact that in this range of @xmath349 objects can be both synchrotron and inverse compton dominated , depending on variability state .
for comparison , note that 3c 279 , a classical fsrq , has a _ bepposax _ spectral index @xmath350 and @xmath351 hz @xcite , i.e. , well into the lbl region . fig .
[ fig7 ] shows that the thermal component is non negligible in most of our sources , with a ratio of black body to synchrotron emission at @xmath352 ranging from @xmath353 for wga j0546.6@xmath346415 and s5 2116@xmath3681 to @xmath354 for rgb j1722@xmath362436 . while these ratios are model
dependent , especially for the former sources for which the optical band is close to the synchrotron cut
off , it is instructive to check the effect of the thermal component on the position of our objects in the @xmath355 plane . by subtracting the thermal component and therefore using only the synchrotron flux in the optical band to derive effective spectral indices , we can see where the objects would fall ( we assume that the blue bump has no effect at 1 kev ; but see below ) . as @xmath26 stays constant while @xmath0 and @xmath1 get steeper and flatter respectively , the sources move along diagonal lines up and to the left in the diagram .
more specifically , the `` synchrotron only '' values are as follows : wga j0546.6@xmath346415 , @xmath356 , @xmath357 , rgb j1629@xmath364008 , @xmath358 , @xmath359 , s5 2116@xmath3681 , @xmath358 , @xmath360 ( practically no change is required for rgb j1722@xmath362436 ) .
it then follows that even by subtracting the effect of the thermal component our sources would still be well within the hbl region in the @xmath63 plane ( fig .
[ fig1 ] ) . actually , the effect of a strong blue bump , by moving sources to the right and down in the diagram , is that of `` pushing '' objects off the hbl region and into the radio - quiet zone .
even with some freedom in the modelling of the synchrotron component , the location of our sources in the hbl region results quite robust , and can not be ascribed to the accretion disk .
that relatively high values of @xmath58 are not due to the accretion disk is also indicated by the low , `` hbl like '' values of @xmath26 , a parameter well correlated with @xmath58 ( e.g. , padovani & giommi 1996 ) and not affected by the optical contamination .
note that the @xmath58 values given in tab .
[ tab_sed ] and shown in fig .
[ fig8 ] are those relative only to the synchrotron component and therefore do not include the thermal component . in seyfert galaxies and radio quiet quasars the x
ray emission is thought to be produced by a hot thermal corona sandwiching a relatively cold accretion disk .
one could then ask if the same component can contribute also in our objects , two of which indeed have a relatively large disk component in the optical uv band . while it is certainly possible that some fraction of the x ray emission ( especially at low energies ) is produced by a hot corona ,
we believe that the non thermal components in these objects are dominant , due to the x - ray variability indicated by the comparison between _ bepposax _ and rosat data , particularly the rapid lecs / mecs variability seen in rgb j1629 + 4008 , and to the lack of any indication of iron line features and hardening of the spectrum due to the so called compton hump ( which are instead common in seyfert galaxies ) . a possible way to decrease the importance of the iron line emission and compton hump in the spectrum could be to assume a dynamic corona , but in such case the corona emission would be characterized by harder x - ray spectra ( @xmath171 ; see malzac et al .
2001 ) , and thus could not account for soft excesses or steep spectra .
as discussed above , the 5 ghz radio powers of our sources are in the range @xmath335 w hz@xmath24 , with @xmath361 w hz@xmath24 .
their 1 kev x ray powers are in the range @xmath362 w hz@xmath24 , with @xmath363 w hz@xmath24 .
these values , although not very different from the mean values of the emss bl lacs ( @xmath364 w hz@xmath24 and @xmath365 w hz@xmath24 ) , are still in the fsrq range , albeit on the low side ( see , for example , fig . 5 of perlman et al .
1998 and padovani et al .
, in preparation ) .
their radio powers put our sources near the low
luminosity end of the dxrbs fsrq radio luminosity function @xcite .
we have found three examples of strong , broad lined blazars with an sed typical of hbl ( one source ) and intermediate bl lacs ( two sources ) .
while the first case is quite strong , the other two are more tentative .
these are the first such sources , as most broad lined blazars known so far have sed typical of lbl , i.e. , with synchrotron peaks at ir / optical energies .
our fsrq , however , have relatively low powers , more typical of bl lacs than quasars .
this ties in with the suggestion by @xcite and @xcite that blazars form a sequence , controlled mainly by their bolometric luminosity which in turn controls the amount of cooling suffered by the emitting electrons .
in powerful sources , where cooling is severe ( high radiative ( @xmath366 ) plus magnetic field ( @xmath367 ) energy densities ) , electrons of all energies cool rapidly ( i.e. , in a time shorter than the injection time ) , making the particle distribution to have a break at @xmath368 , which in this case becomes @xmath298 .
these sources are typically lbl like , and are those located in the lower part of the @xmath369 plot ( see fig .
[ fig9 ] ) . in less powerful sources , instead
, only the highest energy electrons can cool in the injection time , and consequently the particle distribution will have a break at @xmath370 , for which @xmath371 . in this case
@xmath372 ( and @xmath373 ; see details in @xcite ) .
[ fig9 ] shows that the fsrq studied in this paper , which have relatively high @xmath298 ( and therefore @xmath58 ) values , appear to fit within this scenario , as they follow the suggested sequence .
( note that , given the scatter in the plot , this conclusion stands even in the case of a decrease of @xmath300 one order of magnitude in the @xmath310 values for the three sources with relatively uncertain estimates of this parameter . ) moreover , in agreement with their relatively low powers , they are positioned at the low - end of the fsrq region and actually within the bl lac region . in other words ,
we still have not found a case of high@xmath58 ( and therefore high@xmath310 ) _ and _ high power blazar , which would invalidate this scenario .
-0.8truecm 0.2 in another explanation for the relatively low power of our sources is related to a selection effect .
[ fig7 ] shows that our only `` true '' hbl like fsrq , rgb j1629@xmath364008 , has a synchrotron peak at @xmath5 hz , well above the peak of the thermal emission .
its non - thermal power is also low enough ( as implied by its radio power @xmath374 w hz@xmath24 ) that the thermal component , and its associated broad lines , is non negligible in the optical / uv band . given that the equivalent width of its strongest emission line is @xmath66 , the same object with a non - thermal power a factor of @xmath150 higher , corresponding to @xmath375 w hz@xmath24 , still rather modest for an fsrq , would be classified as a bl lac object , since its equivalent width would go down to @xmath137 .
an even stronger non - thermal component would completely wash out the emission lines , making the redshift determination of this object impossible , as there are no galaxy absorption features in its spectrum .
it then follows that a broad - lined source with the same sed as rgb j1629@xmath364008 but a much larger power would not even be classified as an fsrq but as an hbl without redshift .
this could also explain why high power , high@xmath58 bl lacs have not been yet found .
a key ingredient of this argument is the frequency of the synchrotron peak . assuming a typical @xmath58 value for a `` classical '' fsrq of @xmath376 hz , which implies a much weaker optical synchrotron component , an object with a radio power @xmath4 times larger than rgb j1629@xmath364008 will have the same optical ( non - thermal ) power and therefore will still be recognized as a quasar ( assuming similar thermal powers ) .
it follows that high power , low@xmath58 fsrq can be identified as quasars up to very high luminosities .
conversely , high@xmath58 blazars would be classified as hbl without redshift already at moderate radio luminosities .
one problem with this explanation for the lack of high@xmath58 high power blazars is that there should be intermediate cases , i.e. , hbl with broad ( but weak ) lines .
this does not seem apparent , for example , from the optical spectra of the emss bl lacs ( e.g. , rector et al .
2000 ) . as the selection of our sources was done @xmath377 years ago , one could ask if our sources are still really extreme in terms of their x ray to radio flux ratios . out of the 199 dxrbs fsrq known to date
, wga j0546.6@xmath346415 is still the second most extreme .
the fsrq rgb sample , given its higher x ray flux limit and slightly lower radio flux limit , is even better suited to find sources with high @xmath57 ratios ( lower @xmath26 ; see padovani et al .
, in preparation ) .
sorting the known rgb fsrq in order of ascending @xmath26 value , rgb j1629@xmath364008 is at the top of the list while rgb j1722@xmath362436 is number four ( rgb j1413 + 4339 , which is number two , is a borderline quasar ; see laurent - muehleisen et al .
1998 ) . finally , s5 2116@xmath3681 is number 15 out of the @xmath378 fsrq in the multifrequency catalog of @xcite with x ray data sorted in order of ascending @xmath26 value .
it was selected because it is the x ray brightest .
we find it significant that , having selected some of the few known fsrq with effective spectral indices typical of hbl and therefore relatively large x ray to radio flux ratios , we found only one case where the x ray band is clearly dominated by synchrotron emission , with a @xmath379 hz , while the other three sources have @xmath380 hz . as shown , for example , in fig .
[ fig8 ] , hbl can reach much higher values , up to @xmath381 hz , albeit in cases of extreme variability ( e.g. , mkn 501 ; pian et al .
a more meaningful comparison is with the hbl in the multifrequency catalog of @xcite having radio flux @xmath382 mjy , the minimum value for our sources . in that case
the @xmath58 distribution has a _ typical _ value @xmath383 hz and reaches @xmath384 hz .
a discussion of the @xmath58 distribution for fsrq and bl lacs belonging to the _ same _ sample will be presented by padovani et al .
( in preparation ) .
@xcite reported on _ asca _ observations of four fsrq characterized by steep rosat spectra ( @xmath385 ) .
the sources were all found to have flat hard x
ray spectra , with @xmath386 .
sambruna et al .
discuss their results in terms of relatively high synchrotron peaks and thermal emission extending into the x ray band .
we stress , however , that their sources sample a region of parameter space widely different from ours . their effective spectral indices , in fact ,
place their fsrq firmly in the lbl region , unlike ours , so that their four objects should not have been expected to show high @xmath58 values and steep _ asca _ spectra .
@xcite used the fact that these four fsrq happen to have relatively low core - dominance parameters ( @xmath387 ) compared to two `` intermediate '' bl lacs to suggest that these sources could be seen at larger angles than more typical fsrq .
we stress that , as discussed in
[ blazars ? ] , our sources , and especially rgb j1629 + 4008 , appear to be quite compact so that his interpretation might not apply to our objects .
we have presented new _ bepposax _ observations of four flat spectrum radio quasars selected to have effective spectral indices ( @xmath0 , @xmath388 ) typical of high - energy peaked bl lacs . the main purpose of the paper was to see if these sources are indeed the broad - lined counterparts of bl lacs with the synchrotron peak at uv / x ray energies ( hbl ) .
our objects are quite extreme in terms of their x ray to radio flux ratios ( a factor @xmath2 higher than `` classical '' fsrq ) , qualify as radio - loud sources both in terms of their radio - to - optical flux ratio and power , and are variable and radio - compact .
our main results can be summarized as follows : 1 .
we have discovered the first fsrq ( rgb j1629 + 4008 ) whose x ray emission is dominated by synchrotron radiation , as clearly shown by our _ bepposax _ observations ( @xmath6 in the @xmath389 kev band ) , in which the source was also rapidly variable .
this object is therefore the first confirmed member of a newly established class of hbl like fsrq .
this result is fully consistent with earlier rosat observations which detected a steep soft x ray spectrum ( @xmath390 in the @xmath20 kev band ) .
the combination of the x ray data with archival radio and optical data gives a spectral energy distribution typical of hbl sources , with a synchrotron @xmath379 hz . the derived values follow very well the @xmath391 correlation seen for hbl sources ( fig .
[ fig8 ] ) .
2 . the other three sources show a relatively flat ( @xmath392 ) _ bepposax _ spectrum .
however , we have found for two of them ( wga j0546@xmath346415 and rgb j1722 + 2436 ) some indication of steepening at low - energies .
this is based on _ bepposax _ and rosat data and on their ( sparsely sampled ) seds .
we interpret this as the tail of synchrotron emission , which is not strongly constrained but would peak around @xmath8 hz , as typical of `` intermediate '' bl lacs for which the synchrotron and inverse compton components meet within the _ bepposax _ band .
the last source ( s5 2116 + 81 ) turned out to be a more typical fsrq , with the _ bepposax _ band fully dominated by inverse compton emission and a synchrotron @xmath393 hz .
3 . by fitting a synchrotron inverse compton model which includes the contribution of an accretion disk , whose power we estimate from the broad - line region luminosity , to the spectral energy distributions ,
we have derived physical parameters ( e.g. , intrinsic power , magnetic field , etc . ) for our sources .
the thermal component was found to be non - negligible in three out of four objects , with ratios of thermal to synchrotron emission in the range @xmath394 at 5,000 .
although the original selection was based mostly on the effective broad - band ( radio through x ray ) colors , which are in first approximation independent of luminosity , all four candidates are at relatively low powers , more typical of bl lacs than of fsrq and in any case close to the low - luminosity end of the fsrq radio luminosity function @xcite .
we interpret this as due to two , possibly concurrent , causes : an inverse dependence of @xmath298 ( and therefore @xmath395 ) , the lorentz factor of the electrons emitting most of the radiation , on bolometric power , due to the more sever cooling at work in more powerful sources ; and a selection effect , namely the fact that a high - power , hbl like quasar will have such a strong optical / uv non - thermal component that its emission lines will be swamped and the object will be classified as a featureless bl lac .
we thank sally laurent - muehleisen for providing us with the optical spectra of the two rgb sources .
lc acknowledges the stsci visitor program .
ep acknowledges support from nasa grants nag5 - 9995 and nag5 - 10109 ( adp ) and nag5 - 9997 ( ltsa ) .
this research has made use of the usnofs image and catalogue archive operated by the united states naval observatory , flagstaff station ( http://www.nofs.navy.mil/data/fchpix/ ) , 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 , of the stsci digitized sky survey , and of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
the guide star catalogue - ii is a joint project of the space telescope science institute and the osservatorio astronomico di torino .
space telescope science institute is operated by the association of universities for research in astronomy , for the national aeronautics and space administration under contract nas5 - 26555 .
the participation of the osservatorio astronomico di torino is supported by the italian council for research in astronomy .
additional support is provided by european southern observatory , space telescope european coordinating facility , the international gemini project and the european space agency astrophysics division .
dummy ballo , l. , et al .
2002 , , 567 , 50 balucinska - church , m. , & mccammon , d. 1992 , , 400 , 699 becker , r. h. , white , r. l. , & helfand , d. j. 1995 , , 450 , 559 beckmann , v. , wolter , a. , celotti , a. , costamante , l. , ghisellini , g. , maccacaro , t. , & tagliaferri , g. 2002 , , 383 , 410 boella , g. , et al .
1997a , a&as , 122 , 299 boella , g. , et al .
1997b , a&as , 122 , 327 bond , h. e. , kron , r. g. , & spinrad , h. 1977 , , 213 , 1 brinkmann , w. , et al .
1997 , , 323 , 739 celotti , a. , padovani , p. , & ghisellini .
g. 1997 , , 286 , 415 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , , 115 , 169 cutri , r. m. , et al .
2000 , explanatory supplement to the 2mass second incremental data release , available at : http://www.ipac.caltech.edu/2mass/releases/second/doc/explsup.html de grandi , s. & molendi , s. 2002 , , 567 , 163 dickey , j. m. & lockman , f. j. 1990 , , 28 , 215 fiore , f. , et al .
1999 , handbook for nfi spectral analysis , available at : http://www.asdc.asi.it/bepposax/software/ fossati , g. , maraschi , l. , celotti , a. , comastri , a. , & ghisellini , g. 1998 , , 299 , 433 frontera , f. , et al .
1997 , a&as , 122 , 357 georganopoulos , m. 2000 , , 543 , l15 ghisellini , g. , celotti , a. , fossati , g. , maraschi , l. , & comastri , a. 1998 , , 301 , 451 ghisellini , g. , celotti , a. , & costamante , l. 2002 , , 386 , 763 giommi , p. , & fiore , f. 1997 , in proc .
5th international workshop on data analysis in astronomy ( singapore : world scientific ) , 93 giommi , p. , & padovani , p. 1994
, , 268 , l51 giommi , p. , et al . 1999 , , 351 , 59 giommi , p. , padovani , p. , & perlman , e. 2000 , , 317 , 743 haardt , f. , et al . 2001 , , 133 , 187 haberl , f. , & pietsch , w. 1999 , a&as , 139 , 277 hambly , n. c. et al .
2001 , , 326 , 1279 helfand , d. j. , stone , r. p. s. , willman , b. , white , r. l. , becker , r. h. , price , t. , gregg , m. d. , & mcmahon , r. g. 2001 , , 121 , 1872 irwin , m. , maddox , s. , & mcmahon , r. 1994 , spectrum , 2 , 14 iwasawa , k. , fabian , a. c. , & nandra , k. 1999 , , 307 , 611 jannuzi , b. t. , smith , p. s. , & elston r. 1994 , , 428 , 130 kaspi , s. , smith , p. s. , netzer , h. , maoz , d. , jannuzi , b. t. , & giveon , u. 2000 , , 533 , 631 kollgaard , r. i. 1994 , vistas astron . , 38 , 29 lamer , g. , brunner , h. , & staubert , r. 1996 , , 311 , 384 landt , h. , padovani , p. , perlman , e. s. , giommi , p. , bignall , h. , & tzioumis , a. 2001 , , 323 , 757 laurent - muehleisen , s. a. , kollgaard , r. i. , ciardullo , r. , feigelson , e. d. , brinkmann , w. , & siebert , j. 1998 , , 118 , 127 malzac , j. , beloborodov , a.m. , & poutanen , j. 2001 , , 326 , 417 march , m. j. m. , browne , i. w. a. , impey , c. d. , & smith , p. s. 1996 , , 281 , 425 mineo , t. , et al .
2000 , , 359 , 471 morrison , r. , & mccammon , d. 1983 , , 270 , 119 padovani , p. 1992
, , 257 , 404 padovani , p. 1993
, , 263 , 461 padovani , p. , & giommi , p. 1995
, , 444 , 567 padovani , p. , & giommi , p. 1996
, , 279 , 526 padovani , p. , giommi , p. , & fiore , f. 1997 , in from the micro- to the mega - parsec , ed .
a. comastri , t. venturi , & m. bellazzini , , 68 , 147 padovani , p. 2001 , in asp conf .
227 , blazar demographics and physics , ed .
p. padovani , & c. m. urry , 163 padovani , p. , et al .
2001 , , 328 , 931 parmar , a. , et al .
1997 , a&as , 122 , 309 perlman , e. s. , padovani , p. , giommi , p. , sambruna , r. , jones , l. r. , tzioumis , a. , & reynolds , j. 1998 , , 115 , 1253 perlman , e. s. , stocke , j. t. , morris , s. l. , & wang , q. d. 1996 , , 456 , 451 pian , e. , et al .
1998 , , 492 , l17 ravasio , m. et al .
2002 , , 383 , 763 rector , t. a. , stocke , j. t. , perlman , e. s. , morris , s. l. , & gioia , i. m. 2000 , , 120 , 1626 sambruna , r. , maraschi , l. , & urry , c. m. 1996 , , 463 , 444 sambruna , r. , chou , l. , & urry , c. m. 2000 , , 533 , 650 spada , m. , ghisellini , g. , lazzati , d. , & celotti , a. 2001 , , 325 , 1559 stickel , m. , khr , h. , & fried , j. w. 1993 , a&as , 97 , 483 stocke , j. t. , morris , s. l. , weymann , r. j. , & foltz , c. b. 1992 , , 396 , 487 stockman , h. s. , moore , r. l. , & angel , j. r. p. 1984
, , 279 , 485 taylor , g. b. , vermeulen , r. c. , readhead , a. c. s. , pearson , t. j. , henstock , d. r. , & wilkinson , p. n. 1996 , , 107 , 37 tavecchio , f. , maraschi , l. , & ghisellini , g. 1998 , , 509 , 608 urry , c. m. , & padovani , p. 1995
, , 107 , 803 urry , c. m. , sambruna , r. m. , worrall , d. m. , kollgaard , r. i. , feigelson , e. d. , perlman , e. s. , & stocke , j. t. 1996 , , 463 , 424 voges , w. , et al . 1999 , , 349 , 389 wolter , a. , et al .
1998 , , 335 , 899 wright , a. e. , griffith , m. r. , burke , b. f. , & ekers , r. d. 1994 , , 91 , 111 yuan , w. , brinkmann , w. , siebert , j. , & voges , w. 1998 , , 330 , 108 | we present new _ bepposax _ lecs , mecs , and pds observations of four flat spectrum radio quasars ( fsrq ) having effective spectral indices @xmath0 and @xmath1 typical of high - energy peaked bl lacs .
our sources have x ray to radio flux ratios on average @xmath2 times larger than `` classical '' fsrq and lie at the extreme end of the fsrq x ray to radio flux ratio distribution .
the collected data cover the energy range @xmath3 kev ( observer s frame ) , reaching @xmath4 kev for one object .
the _ bepposax _ band in one of our sources , rgb j1629 + 4008 , is dominated by synchrotron emission peaking at @xmath5 hz , as also shown by its steep ( energy index @xmath6 ) spectrum .
this makes this object the _ first _ known fsrq whose x ray emission is not due to inverse compton radiation .
two other sources display a flat _ bepposax _ spectrum ( @xmath7 ) , with weak indications of steepening at low x ray energies . the combination of _ bepposax _ and rosat observations , ( non - simultaneous ) multifrequency data , and a synchrotron inverse compton model suggest synchrotron peak frequencies @xmath8 hz , although a better coverage of their spectral energy distributions is needed to provide firmer values .
if confirmed , these values would be typical of `` intermediate '' bl lacs for which the synchrotron and inverse compton components overlap in the _ bepposax _ band .
our sources , although firmly in the radio loud regime , have powers more typical of high energy peaked bl lacs than of fsrq , and indeed their radio powers put them near the low
luminosity end of the fsrq luminosity function .
we discuss this in terms of an anti - correlation between synchrotron peak frequency and total power , based on physical arguments , and also as possibly due to a selection effect . |
the so - called `` uv - upturn '' phenomenon @xcite , i.e. , the rising ultraviolet emission shortward of 2000 , sometimes featuring in the spectral energy distribution ( sed ) of elliptical galaxies and the bulges of spirals , has been for long a puzzling problem for old galaxy environments dominated by stars of mass comparable to the sun .
in fact , the implied existence of an important contribution of ( long - lived ) b stars , hotter than @xmath9 k and providing up to about 2% of the galaxy bolometric luminosity @xcite , has been alternately identified with different evolutionary stages .
such stages include binaries @xcite , blue stragglers @xcite , blue horizontal - branch ( hb ) stars @xcite , asymptotic giant branch ( agb ) _ manqu _ stars @xcite , and post - agb nuclei of planetary nebulae ( pne ) @xcite ( see * ? ? ?
* , for an exhaustive review , and a more recent update by @xcite ) resolved color magnitude ( c - m ) diagrams of stellar populations in m32 @xcite have definitely shown that even its relatively poor uv emission almost entirely arises from a fraction of hot hb stars , further complemented by a minority contribution from post - agb pn nuclei . still , facing the established interpretative scenario ,
one is left with at least three important issues that need to be assessed to understand the real nature of the uv - upturn phenomenon . _
( i ) _ the canonical evolutionary framework experienced in galactic globular clusters naturally predicts a blue hb morphology only for old , metal - poor stellar populations @xcite .
if this is the case for ellipticals too , then uv stars should represent the @xmath10 tail of a ( supposedly ) broad metallicity distribution seen to peak at much higher values , around solar abundance .
clearly , a more composite picture might be envisaged once one admits non - standard models ( i.e. , including the effects of stellar rotation , helium mixing , differential mass loss , etc . ) to account , in particular , for the well known `` second parameter '' dilemma @xcite .
however , this unconventional approach still suffers from a somewhat arbitrary fine - tuning of the key physical assumptions . _
( ii ) _ hot hb stars might , nonetheless , also be naturally predicted among super metal - rich stellar models , as far as metal abundance ( and the linked helium content ) exceeds some critical threshold .
presumably , in this case mass - loss allows stars to reach the hb phase with a conveniently low external envelope , compared to the helium core mass @xcite .
such `` extreme hb '' stars ( ehb ) have actually been observed , for example , in @xmath11 cen @xcite , ngc 6388 and ngc 6441 @xcite , and in some old galactic open clusters as well , like ngc 6791 @xcite ; they clearly remain the favorite candidates to explain the evolutionary framework of uv - enhanced elliptical galaxies @xcite .
this hypothesis implies , however , a direct relationship between chemical abundance ( modulating the helium core mass at the hb onset ) and mass - loss efficiency ( to suitably `` peel off '' the stellar envelope along the rgb ) . as a consequence
, one has to expect the uv - to - optical color to be , eventually , one of the most quickly evolving features in the sed of elliptical galaxies @xcite . in theory ,
the uv - upturn can fade by several magnitudes as the lookback time increases by a few gyr , although the effect is still detectable at intermediate redshift ( @xmath12 ) @xcite .
unfortunately , the evolutionary details are extremely model - dependent , and a strong uv excess could be triggered at ages as early as @xmath13 gyr @xcite or as late as @xmath1415 gyr @xcite . _ ( iii ) _ an established correlation seems to be in place between pn luminosity - specific rate and @xmath15 color for elliptical galaxies in the virgo and fornax clusters , and in the leo group @xcite .
the sense is that reddest metal - rich systems display , at the same time , a stronger uv - upturn @xcite _ and _ a poorer pn population per unit galaxy luminosity @xcite .
if the pn event is the final fate for agb stars at the end of their thermal pulsing phase @xcite , then the relative deficiency of nebulae might be evidence of an incomplete ( or fully inhibited ) agb evolution of low - mass stars under special environment conditions of the parent galaxy . as a central issue in this discussion , _ it is clear therefore that a preeminent connection should exist between uv excess and agb distinctive properties of stellar populations in early - type galaxies . _ on account of the fuel consumption theorem @xcite , a 1 m@xmath2 star of solar metallicity enters its core he - burning phase with , at most , the equivalent of 0.43 m@xmath2 of h to be spent as nuclear fuel .
star with solar abundance @xmath16 starts its hb evolution with a total he amount of roughly @xmath17 m@xmath2 , of which @xmath18 m@xmath2 are confined in the core @xcite and @xmath19 m@xmath2 reside in the envelope .
metals amount to roughly @xmath20m@xmath2 and , accordingly , fresh h is 0.37 m@xmath2 .
taking into account the nuclear rates ( e.g. , * ? ? ?
* ) , the h+he fuel provides at most the equivalent of @xmath21 m@xmath2 of hydrogen . ]
this means that , under quite general conditions , post - rgb evolution alone could easily account , in principle , for up to 3/4 of the total bolometric luminosity of a galaxy stellar population @xcite . whether this energy is eventually reduced ( if stars loose their fuel before they burn it ) , or whether it is finally released in the form of ultraviolet or infrared photons , crucially depends on mass - loss and its impact along the entire red - giant evolution .
hence , it is of special pertinence to constrain the relevant physical conditions that affect agb evolution in favor of an earlier transition of hb stars towards high temperature and enhanced ultraviolet emission . in this paper
we would like to draw the reader s attention to a possibly new and powerful approach to the problem , that can find straightforward applications even to distant galaxies .
as explained in sec . 2 , the method relies on surface - brightness fluctuation theory to safely tie infrared effective magnitudes ( that _ can _ be determined for unresolved stellar populations ) , to stellar luminosity at the agb tip ( that _ can not _ be directly observed in distant galaxies )
. we will show , in sec . 3 , that these results tightly correlate with the ultraviolet properties of `` uv upturn '' elliptical galaxies , allowing a self - consistent physical picture and a quite accurate diagnostic of the post - rgb evolution of their underlying stellar populations , including hb morphology and agb deployment .
our results will be finally summarized and discussed in sec .
@xcite and @xcite first realized the potentially useful information about the composing stars hidden in the surface brightness fluctuations ( sbfs ) of galaxies with unresolved stellar populations .
the problem has since received a more complete theoretical assessment by @xcite , and @xcite .
briefly , the basic relationship upon which the theory relies is @xmath22 the l.h .
side of the equation links an observable quantity ( namely , the relative variance of the galaxy surface brightness ) with the theoretical second - order statistical moment of the composing stars .
this quantity , in turn , traces the distribution of stellar luminosity ( @xmath23 ) for the whole population ; it is also a natural output of any population synthesis code , and can easily be computed for different photometric bands and distinctive evolutionary phases of simple stellar populations ( ssps ; see , e.g. , the current theoretical databases of * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the derived `` effective '' stellar luminosity , @xmath24 , in eq .
( [ eq : fluc ] ) has some important properties . _ ( a ) _ it can be derived from a fully observational procedure , without any supplementary theoretical assumptions .
_ ( b ) _ it is an intrinsic distinctive parameter of the stellar aggregate , and its empirical measurement does not require any individual star to be ( fully or partially ) resolved .
_ ( c ) _ it identifies an `` effective '' magnitude ( i.e. @xmath25 ) in a completely similar way and with the same photometric zero points as `` standard '' magnitudes , and it scales accordingly with distance and foreground screen reddening .
[ cmd ] [ kk ] being statistically representative of the stellar system as a whole , the effective magnitude can not be physically associated to any specific star or stellar groups along the c - m diagram of a stellar aggregate ; nonetheless , it can be instructive to identify stars with @xmath26 luminosity as the `` prevailing '' group tracing the whole population in the different photometric bands .
this is shown in fig . 1 for the illustrative case of a 15 gyr ssp with solar metallicity and salpeter imf . because of the quadratic @xmath23 dependence of the summation in the numerator of eq .
( [ eq : fluc ] ) , the effective magnitude at any optical / infrared band is highly sensitive to the giant stars and only marginally responds to a shift in the ssp `` mass leverage '' caused by a change in the imf slope @xcite . likewise , as probe of the brightest stars in a population at a given wavelength ,
the effective magnitude is relatively insensitive to any underlying older component in the case of composite stellar systems .
it is evident from fig .
1 that near - infrared ( near - ir ) magnitudes closely trace the bright - end tail of the ssp luminosity function .
in particular , one can notice that @xmath0 is potentially the best tracer of the ssp tip stellar luminosity ( @xmath27 ) , as _ both quantities are expected to depend in quite the same way on the overall distinctive parameters of the stellar population , including age , metallicity , imf , and mass - loss .
_ `` effective contributors '' ( in the statistical definition of @xcite ) to the @xmath28-band sbf for a galaxy of @xmath29 m@xmath2 .
this number further increases to about @xmath30 for @xmath31-band contributors , and to @xmath32 for the @xmath33 band sbf .
] this is shown in fig . 2 , where we compare the @xmath0 vs. @xmath34 relationship for a full collection of ssp models from the @xcite and the updated charlot & bruzual ( cb07 * ? ? ?
* ) synthesis codes .
band is a variant of the standard @xmath28 filter , sometimes preferred for its reduced background noise ( see * ? ? ?
* for specific details ) .
both filters have similar effective wavelength and photometric zero points , so that @xmath35 within a typical @xmath36 mag uncertainty , as we directly verified in our tests.]@xmath37 ) from the original isochrone set , while @xmath34 is just read as the absolute magnitude of the brightest stellar type in the agb+rgb phases . ]
the correlation appears quite robust , in spite of the extremely wide range explored for the population parameters , and the different input physics adopted by the two sets of theoretical models .
formal fits to the whole sets of models yield : @xmath38 a comparison with real stellar populations extending across the widest range of evolutionary parameters would clearly be the ideal test for the envisaged theoretical picture .
the magellanic clouds ( mcs ) stand out as striking candidates , in this regard , with the age of their star clusters spanning over four orders of magnitude , from a few myr up to @xmath39 yr .
in addition , their relatively close distance allows _ both _ fluctuation luminosity _ and _ agb / rgb luminosity tip to be directly measured on resolved c - m diagrams .
near - ir sbfs for a sample of 191 mc star clusters have been obtained by @xcite using @xmath40-band data from the two micron all sky survey ( 2mass , * ? ? ?
* ) . to reduce stochastic effects due to small - number stellar statistics along fast evolutionary phases , in their study these authors assembled seven `` super - clusters '' , by homogeneously co - adding several objects for each age class according to the ( * ?
* , swb ) classification scheme .
the procedure to obtain the sbfs for these data has been extensively described before @xcite ; briefly , direct summation of bright stars provided the numerator of the r.h .
fraction of eq .
( [ eq : fluc ] ) , while total luminosity ( at the denominator of the equation ) was safely estimated from integrated photometry of each supercluster mosaic , to include the contribution of faint , unresolved stars .
following @xcite , for the present work we have determined the tip luminosity of the agb+rgb by convolving the luminosity function of each supercluster with a sobel edge - detection filter ( the zero - sum kernel [ -2,0,+2 ] ) .
the tip luminosity was identified with the peak of the filter response function , after checking that there is indeed an important count discontinuity at that location in the luminosity function ; the error in the measurement equals the width of the histogram bin .
we list in table [ tab_mc ] the @xmath40-band absolute fluctuation magnitude , @xmath40 absolute tip magnitude , age , and metallicity of the mc superclusters .
lcccc pre & [email protected]&8.7 @xmath41 0.2 & 6.4 @xmath41 0.3 & 0.3 @xmath41 0.2
+ i & [email protected]&9.8 @xmath41 0.2 & 7.0 @xmath41 0.3 & 0.3 @xmath41 0.2 + ii & [email protected]&8.9 @xmath41 0.2 & 7.5 @xmath41 0.3 & 0.3 @xmath41 0.2 + iii & [email protected]&8.3 @xmath41 0.2 & 8.0 @xmath41 0.3 & 0.1 @xmath41 0.2 + iv & [email protected]&8.4 @xmath41 0.2 & 8.5 @xmath41 0.3 & 0.8 @xmath41 0.2 + v & [email protected]&8.0 @xmath41 0.2 & 9.0 @xmath41 0.3 & 0.6 @xmath41 0.2 + vi & [email protected]&7.6 @xmath41 0.2 & 9.5 @xmath41 0.3 & 1.0 @xmath41 0.2
+ vii & [email protected]&7.2 @xmath41 0.2 & 9.9 @xmath41 0.3 & 1.4 @xmath41 0.2 + [ tab_mc ] [ mcfig ] again , a tight relationship between @xmath34 and effective @xmath0 magnitudes is in place , as shown in fig . 3 , along the entire age range ( and the corresponding metallicity drift , see column 5 in table [ tab_mc ] ) .
a fit to the mc data ( dashed line in the figure ) provides : @xmath42 with a data scatter @xmath43 mag across the fitting line .
note that this perfectly compares , within the statistical uncertainty of the observations , with both eqs .
( [ eq : ddb ] ) ; this is true even beyond the nominal age range of the theoretical relations since , for instance , the pre - swb and the swb type i mc superclusters are too young to even sport any standard agb or rgb phase ! the link between @xmath0 and @xmath34 can be regarded as a _ much more deeply intrinsic property of ssps _ , not exclusively related to age ( or metallicity ) evolution .
of course , our method can not remove , by itself , any age / metallicity degeneracy , that makes integrated photometric properties of young metal - rich stellar systems closely resemble those of older metal - poor ones ( e.g. * ? ? ? * ; * ? ? ? * ) .
however , as far as old stellar populations are concerned , the unbiased and model - independent estimate of the tip stellar luminosity , even in unresolved stellar populations , provides an additional and powerful tool to quantitatively size up important effects , like the mass - loss impact on red - giant evolution .
[ mloss ] relying on the @xcite standard @xmath44 parameterization , fig .
4 gives an instructive example of how mass - loss can modulate both @xmath34 and @xmath0 . in the figure we explore the behavior of an old ( 15 gyr ) ssp reference model ( big dots in the plot ) , along a full variety of mass - loss scenarios , spanning from a virtually vanishing stellar wind ( i.e. @xmath45 ) , up to a violent mass - loss rate ( @xmath46 ) capable of fully wiping out the external envelope of stars while on the rgb phase . similarly to fig .
3 , we still report the same clean relationship between @xmath34 and @xmath0 , with old agb - enhanced models ( @xmath47 ) closely resembling , all the way , young ( 2 - 5 gyr ) standard ( @xmath48 ) populations in the upper right corner of the panel . again , according to the trend of the 15 gyr model sequence vs. @xmath44 we can propose the following parametric equations : @xmath49 rcccccccc 221 & 1.90 @xmath41 0.03 & 0.198 @xmath41 0.007 & 0.88 & 6.77@xmath50 & 4.50 @xmath41 0.17 & 5.28 @xmath41 0.10 & 5.95 @xmath41 0.14 & 2.30 @xmath41 0.05 + 224 & 2.27 @xmath41 0.03 & 0.324 @xmath41 0.007 & 0.95 & 6.94@xmath51 & 3.51 @xmath41 0.17 & 4.46 @xmath41 0.09 & 5.69 @xmath41 0.14 & 1.66 @xmath41 0.07 + 1316 & 2.38 @xmath41 0.08 & 0.265 @xmath41 0.023 & 0.87 & 7.50@xmath52 & 5.0 @xmath41 0.2 & 5.39 @xmath41 0.19 & ... & 2.20 @xmath41 0.07 + 1379 & 2.11 @xmath41 0.13 & 0.269 @xmath41 0.006 & 0.88 & ... & 3.01 @xmath41 0.10 & 5.11 @xmath41 0.19 & 5.75 @xmath41 0.12 & 1.70 @xmath41 0.09 + 1387 & ... & ... & 0.98 & ... & 2.16 @xmath41 0.05 & 5.4 @xmath41 0.8 & 5.77 @xmath41 0.10 & ...
+ 1389 & 2.12 @xmath41 0.06 & 0.236 @xmath41 0.002 & 0.90 & ... & 3.38 @xmath41 0.09 & 5.16 @xmath41 0.20 & 6.35 @xmath41 0.13 & ... + 1399 & 2.52 @xmath41 0.03 & 0.357 @xmath41 0.007 & 0.95 & 7.30@xmath52 & 2.05 @xmath41 0.17 & 4.55 @xmath41 0.16 & 5.28 @xmath41 0.15 & 1.41 @xmath41 0.08 + 1404 & 2.39 @xmath41 0.03 & 0.344 @xmath41 0.007 & 0.95 & ... & 3.30 @xmath41 0.17 & 4.76 @xmath41 0.21 & 5.53 @xmath41 0.10 & 1.58 @xmath41 0.08 + 3379 & 2.33 @xmath41 0.03 & 0.329 @xmath41 0.007 & 0.94 & 6.77@xmath53 & 3.86 @xmath41 0.17 & 4.70 @xmath41 0.14 & 5.43 @xmath41 0.17 & 1.46 @xmath41 0.16 + 3384 & 2.20 @xmath41 0.11 & 0.296 @xmath41 0.014 & 0.91 & 6.42@xmath54 & 3.9 @xmath41 0.2 & 4.82 @xmath41 0.22 & ... & 2.05 @xmath41 0.11 + 4278 & 2.45 @xmath41 0.03 & 0.293 @xmath41 0.007 & 0.90 & ... & 2.88 @xmath41 0.17 & 4.49 @xmath41 0.22 & ... & 1.37 @xmath41 0.03 + 4374 & 2.48 @xmath41 0.03 & 0.323 @xmath41 0.007 & 0.94 & 6.77@xmath54 & 3.55 @xmath41 0.17 & ... & 5.71 @xmath41 0.25 & 1.70 @xmath41 0.04 + 4406 & 2.42 @xmath41 0.03 & 0.330 @xmath41 0.007 & 0.90 & 6.89@xmath54 & 3.72 @xmath41 0.17 & ... & 5.74 @xmath41 0.12 & 1.61@xmath41 0.16 + 4472 & 2.49 @xmath41 0.03 & 0.331 @xmath41 0.007 & 0.95 &
7.16@xmath54 & 3.42 @xmath41 0.17 & 4.64 @xmath41 0.11 & 5.64 @xmath41 0.13 & 1.52 @xmath41 0.14 + 4486 & 2.60 @xmath41 0.03 & 0.303 @xmath41 0.007 & 0.93 & 7.10@xmath55 & 2.04 @xmath41 0.17 & ... & ... & 1.38 @xmath41 0.04 + 4552 & 2.44 @xmath41 0.03 & 0.346 @xmath41 0.007 & 0.94 & ... & 2.35 @xmath41 0.17 & ... & 5.86 @xmath41 0.13 & 1.65@xmath41 0.04 + 4621 & 2.41 @xmath41 0.03 & 0.355 @xmath41 0.007 & 0.92 & ... & 3.19 @xmath41 0.17 & ... & 5.59 @xmath41 0.21 & 1.43 @xmath41 0.11 + 4649 & 2.56 @xmath41 0.03 & 0.360 @xmath41 0.007 & 0.95 & 7.22@xmath55 & 2.24 @xmath41 0.17 & ... & ... & 1.26 @xmath41 0.04
+ [ tab_tot ] curiously enough , note from the equation set that @xmath34 seems slightly less sensitive to @xmath44 than @xmath0 . as we will see in a moment , such a smaller variation range for @xmath34 stems from the `` bottoming out '' effect of the rgb tip , that eventually replaces the agb in providing the brightest stars in a ssp when mass - loss increases to @xmath56 .
the now established importance of mass - loss in old stellar populations dominated by stars with @xmath57 m@xmath2 was first emphasized in a series of important theoretical contributions in the early seventies @xcite ; this was urged by the intervening observational evidence that a mass spread was needed to reproduce hb and agb morphology in galactic globular clusters @xcite . to consistently match the observed agb tip in local globular clusters , for instance
, @xcite suggested a fine - tuning value of @xmath58 .
this calibration for old pop ii stars might however not so straightforwardly be extended , _ a priori _ , to more metal - rich environments , as one could realistically expect mass - loss efficiency to depend on ( increase with ? ) metallicity .
depending on mass - loss strength , standard stellar evolution theory basically features three characteristic scenarios that constrain post - rgb evolution of low - mass stars ( see fig . 5 ) .
under different physical conditions and to a different extent , each one of these cases eventually leads to the formation of hot stars , thus potentially supplying an important contribution to the integrated ultraviolet luminosity of a galaxy . _
( a ) _ for @xmath59 , models tell us that at the end of rgb evolution , low - mass stars ignite helium in a degenerate core ( the so - called `` helium flash '' ) .
a bright agb evolution has to be expected , reaching or even surpassing the rgb tip luminosity .
if thermal pulses are set on ( @xmath60 ) , stars end up loosing the external envelope and generating a planetary nebula @xcite . in this case ,
hot post - agb nuclei of nebulae easily exceed an effective temperature of 30000 k @xcite .
[ sketch ] _ ( b ) _ if @xmath61 , the pn event is aborted @xcite , and agb evolution is partially or fully inhibited @xcite . in this case , the brightest stars ( both in bolometric luminosity and in the infrared ) belong to the rgb tip , and they settle on the hb with a he core mass close or slightly exceeding 0.47 m@xmath2 ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . for slightly increasing mass - loss efficiency
, the stellar envelope becomes thinner and thinner , and the hot internal core is unveiled .
if this is the case , stars likely settle at high effective temperature ( @xmath62 k ) and originate a blue hb morphology . _
( c ) _ as a final case , for an even stronger mass - loss ( @xmath63 ) , the stellar envelope might be wiped out well before the full completion of the rgb phase . as a consequence , standard hb and agb evolution
would be fully inhibited ; stars would be left `` midway '' along the rgb , with their rapidly exhausting h - shell emerging at ever shallower stellar layers .
such `` hot flasher '' stars would therefore postpone the he ignition by moving straight to the high - temperature region of the c - m diagram , as is likely observed , indeed , in the cluster ngc 2808 @xcite .
[ ell2 ] in terms of uv contribution , cases _ ( a ) , ( b ) _ , and _ ( c ) _ represent a sequence of increasing energy budget , as a prevailing fraction of stellar nuclear fuel is more efficiently burnt at high effective temperature .
the outlined scenario has , of course , to be regarded as a general scheme , where the @xmath44 thresholds discriminating among the different evolutionary regimes slightly depend on ssp age ( through the ms turn off mass ) . while the illustrative case of fig .
5 refers to a 15 gyr ssp , one could easily verify that the overall trend for @xmath28 luminosity is maintained , for instance for a 10 gyr population , by just `` shrinking '' the @xmath44-scale of the plot by a factor of @xmath64 ; this means , for example , that _ agb - manqu _ stars are produced in a 10 gyr scenario for @xmath65 .
in order to assess the relevance of the previous theoretical framework to the appearence and strength of the `` uv - upturn '' phenomenon in ellipticals , it is convenient to parameterize our analysis in terms of the @xmath66 color , that is a measure of the galaxy emission around the 1550 region vs. the johnson visual band , as originally defined by @xcite . to this end , the @xcite iue galaxy sample has been taken as a reference ; we supply in table [ tab_tot ] , for each object , the infrared effective magnitudes from @xcite , @xcite , @xcite , and @xcite and , when available , further relevant pieces of information , like stellar velocity dispersion , integrated @xmath15 color , h@xmath4 and mg@xmath67 lick indices , and the specific pn rate per unit galaxy luminosity ( from * ? ? ?
* ) . as a first relevant clue in our analysis , fig . 6 is suggestive of a trend among most ellipticals , with uv - enhanced galaxies being about one mag fainter in @xmath40 effective magnitude ( some 0.7 mag in the inferred @xmath34 ) , compared to uv - poor systems .- band effective luminosities based on the @xmath68 photometry only assuming , empirically from table 2 , @xmath69 , with a conservative error bar of @xmath70 mag .
the extrapolated @xmath0-mag data are displayed with a different marker on the plots.[nota ] ] on the plot one should report , however , a couple of outliers , displaying a brighter @xmath0 magnitude ( or , alternatively , a `` bluer '' @xmath66 color ) .
these include ngc 1387 and ngc 1389 in the fornax cluster , and ngc 4552 in virgo
. we will be back to these objects in the next section , for a brief discussion .
in addition , we have to mention that the `` merger '' radio - galaxy ngc 1316 ( fornax a ) exhibits visible dust lanes that , when accounted for , may lead to an intrinsically `` bluer '' ( 1550 - @xmath71 ) color , although with possibly negligible effects on the @xmath28 sbf magnitude . after excluding these controversial objects ,
the data tend to display a fairly good correlation ( @xmath72 ) that , also recalling eq .
( [ eq : eta ] ) , we can approximate with @xmath73 if we entirely ascribe the dimming in red - giant tip luminosity suggested by fig .
6 to a mass - loss effect , then eq . ( [ eq : eta ] ) implies a spread @xmath74 among the galaxy population , with uv - enhanced ellipticals requiring a reimers parameter @xmath48 - 0.4 , very similar to what @xcite derived , empirically , for the galactic globular clusters . according to fig . 5
, this confirms that , overall , the agb is deployed in luminosity slightly above the rgb tip , so that the brightest stars in ellipticals should indeed always be genuine agb stars .
[ fig201 ] [ fig200 ] there is an obvious caveat , though , in this simplified picture .
if a standard monolithic scenario is assumed for early - type galaxy formation , then gravitational collapse of the primeval gas clouds should have proceeded over a free - fall timescale @xmath75 ( e.g. , * ? ? ?
* ) , leading to a mass age
metallicity relation for early - type galaxies , where more massive galaxies are both older and more metal - rich than less massive ones ( e.g. , * ? ? ?
the combined action of age and metallicity plays an important role to modulate the red - giant luminosity tip , as the agb is `` naturally '' brighter among young or metal - rich ssps ( see , again , fig .
the complex interplay of the different galaxy physical properties is well depicted in figs . 7 and 8 .
for one , as @xcite had already shown , the uv upturn seems to appear among the most massive and metal rich ellipticals ( i.e. , there is a positive correlation with @xmath76 and the mg@xmath67 index , respectively ) .
figure 7 indicates for these galaxies also a less extended agb , as for a lower stellar core - mass distribution .
this feature is likewise expected to directly affect the pn rate per unit galaxy luminosity ( the so - called `` @xmath77 '' index , see fig .
8) ; indeed , there is a scantier pn stellar population among old / metal - rich ( uv - enhanced ) ellipticals @xcite .
[ hbeta ] observations of local star samples show that the h@xmath4 index reaches its maximum strength for a - type stars in the temperature range t@xmath78 k @xcite . given this selected sensitivity , synthesis models of old ssps @xcite predict the integrated index to be strongly enhanced ( by roughly 0.8 or more ) in the presence of a broad color - extended hb , as observed for most metal - poor galactic globular clusters
. this important piece of information could therefore usefully complement the more extreme @xmath66 color in the analysis of hb morphology for unresolved stellar populations in distant galaxies .
the situation is summarized in fig . 9 , where elliptical galaxy data are compared with three illustrative cases of 15 gyr ssps with slightly supersolar metallicity ( [ fe / h]=+0.3 ) , and different ranges of hb temperature distribution .
in particular , in the figure we account for _ ( i ) _ a red hb ( rhb ) morphology , that is , a clump of red stars very close to the rgb location , mimicking the real case of metal - rich galactic globular clusters , like 47 tuc ; _ ( ii ) _ an intermediate hb ( ihb ) morphology , with a broad ( roughly bell - shaped ) temperature distribution peaked about the a - stars temperature range and extending up to @xmath79 k ; _ ( iii ) _ a blue hb ( bhb ) morphology , peaked at @xmath80 k and with a tail of hot sdb stars , up to a temperature of @xmath81 k ( see * ? ? ?
* , for further details ) .
interestingly enough , the comparison with the observations shows that `` uv - upturn '' ellipticals _ can not _ be compatible with any bhb morphology alone .
too many hot stars would make the 1550 galaxy luminosity exceedingly bright compared to the energy bulk released at the @xmath71 band . on the other hand , neither an ihb is viable , given the catastrophic impact of its a - star component on the integrated @xmath82 index of the population ( e.g. * ? ? ?
definitely , an optimum match to these systems requires a mixed contribution , where the bulk of the galaxy luminosity ( over 80% ) is provided by `` standard '' ( metal - rich ) stellar populations with a red hb , upon which a 20% , or less , in terms of total optical contribution , bhb component is superposed .
figure 10 provides an illustrative picture of this case , showing the synthetic c - m diagram and the integrated sed for the composite stellar population corresponding to the big triangle in fig .
[ cmdhb ] so , in full agreement with @xcite conclusions , a somewhat dichotomic scenario seems to emerge for giant ellipticals , where the occurrence of the `` uv - upturn '' phenomenon is intimately related to the appearence , among the galaxy hb population , of a hot and very low - mass stellar component ( m@xmath8 m@xmath2 , e.g. * ? ? ?
* ) , that at some stage begins to join the standard red - clump stars .
it could be useful to note , in this regard , that two metallicity ranges seem to favor the formation of ehb stars , through an optimum combination of msto stellar masses and subsequent mass - loss via stellar winds along the rgb . as displayed in fig .
11 , [ fe / h ] values of either @xmath6 dex or @xmath7 provide in principle the needed low mass at the onset of hb evolution in a standard reimers mass - loss framework with @xmath83 . to add even further to this challenging situation , however , one also has to report that high - resolution uv spectroscopy of a sample of six uv - strong ellipticals ( namely , ngc 1399 , 3115 , 3379 , 4472 , 4552 , and 4649 ) indicates for the claimed bhb component a moderately sub - solar ( @xmath84 ) metallicity @xcite . if confirmed , this result might lead to identify the _ metal - poor _ stellar component of giant ellipticals as the one responsible for the uv excess .
whatever the exact mechanism , this will cause hb morphology to readily assume a bimodal distribution , with a relative lack of warm stars at the intermedate @xmath85 range pertinent to the a spectral type ( see fig .
in this paper we have carried out a synoptic analysis of the different observational features that have to do with the uv luminosity excess in elliptical galaxies . as far as the canonical picture is assumed , with old stellar populations dominating early - type galaxy luminosity , the appearence of the `` uv upturn ''
should readily call for a profound change in the c - m diagram of galaxy stellar populations , not only involving the hot stellar component of the galaxy but also reverberating on red - giant evolution at the low - temperature regime . as we mainly deal with distant , unresolved stellar populations ,
our analysis has to rely on a combined approach , matching infrared and ultraviolet diagnostic tools in order to probe the main features of the stellar c - m diagram , starting with integrated galaxy photometry .
theory of surface - brightness fluctuations provides , in this sense , a natural and quite powerful way to go deep inside the problem and , as far as the infrared wavelength interval is considered , we have demonstrated theoretically that a straight and very clean relationship is in place between a macroscopic measure , such as the galaxy fluctuation magnitude , and the corresponding individual magnitude of the brightest stars in turn at the tip of the red - giant ( agb+rgb ) phases ( fig . 2 ) .
played in the @xmath28 band , this correlation leads , from a measurement of @xmath0 , to a value for @xmath34 : @xmath86 with a @xmath87 mag internal uncertainty . as we showed in sec . 2 ,
our ssp theoretical predictions find full support from the observations , and a direct check on the mc star clusters confirms the @xmath0 vs. @xmath34 relationship to be a much more general and deeply intrinsic property of stellar populations , virtually independent from any assumption about age , metallicity , imf , and mass loss parameters .
given its nature , this relationship can not , by itself , help disentangle the problem of age / metallicity degeneracy ; however , quite fruitfully , it provides us with a very direct probe of agb properties , in a number of relevant details that directly deal with the mass - loss impact and the mass of dying stars ( fig .
our effort toward exploring the infrared side of galaxy seds has a twofold aim since , as a consequence of the basic principle of energy conservation , any gram of stellar fuel spent to feed ultraviolet luminosity can not ( and will not ) be spent at longer wavelengths .
this has led to the key issue of this paper , summarized in fig .
6 , that the _ strengthening of the uv rising branch is always seen to correspond to a weakening in the agb luminosity extension _ , as traced by galaxy @xmath28 fluctuation magnitude .
this `` shortening '' in agb deployment is mainly recognized among giant ellipticals ( @xmath0 becomes fainter with increasing galaxy velocity dispersion , @xmath76 , see fig .
7 ) , and could mainly be ascribed to an age effect , as the agb tip naturally fades in luminosity with increasing age of the system ( fig .
3 ) , and high - mass galaxies are recognized to be older than systems of lower mass ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , the relationship in place likely calls for a more elaborated physical scenario , once the full range of observing evidences is added to our analysis . vs. lick mg@xmath67 index , for the galaxy sample in table [ tab_tot ] .
the lick index is assumed to trace galaxy metallicity according to the @xcite calibration , as reported on the top axis of the plot . the observed decrease in @xmath34 with increasing @xmath88
$ ] can be mostly explained if more metal - rich galaxies are also older , as in a standard monolythic scenario for galaxy formation . the dashed line on the plot marks the minimum luminosity required for stars to experience the thermal pulsing phase along their agb evolution , and thus end their evolution as pne .
] _ ( a ) _ besides being old , `` uv upturn '' galaxies are also metal rich ( i.e. , stronger in mg@xmath67 lick index ) . disregarding any change in mass - loss rate , stellar tracks predict slightly more massive stars to evolve off the ms at a fixed age , with increasing metallicity ( e.g. , *
* ; * ? ? ?
this leads to correspondingly more massive agb stars and a brighter agb luminosity .
facing the observed trend in galaxy distribution , as summarized in fig .
12 , metallicity effects evidently enter by mitigating the dimming action of age on @xmath34 with increasing galaxy mass . in any case
, the interplay between age and metal abundance actually makes the derived range for the reimers mass - loss parameter ( i.e. @xmath74 , as discussed in sec .
3 ) a safe upper limit . in fact , it suggests that metal abundance does _ not _ modulate by orders of magnitude mass - loss efficiency via stellar winds ..
conversely , also from data of single stars , @xcite argues that @xmath89 is metallicity - independent . from a theoretical point of view , @xcite models suggest that , if mass - loss is really proportional to metallicity , its effect to dim near - ir effective luminosities on average almost exactly offsets the brightening effect of metallicity itself . ]
_ ( b ) _ a match of galaxy @xmath0 data with the calibration of fig .
5 confirms that , all the way , brightest stars in ellipticals are genuine agb members , reaching the thermal - pulsing phase ( see also fig . 12 ) , and with the agb tip exceeding the rgb tip by some 0.5 - 1.5 mag .
in the temperature range of m giant stars , a major fraction of bolometric luminosity is emitted through the @xmath28 band , and bolometric correction is a nearly constant quantity that we can estimate from @xmath90 mag @xcite ; we can therefore straightforwardly translate the galaxy fluctuation magnitude into an estimate of the bolometric tip luminosity , @xmath91 , and therefrom of the corresponding stellar core mass . from our previous calibration ( eq . [ eq : calib ] ) , we can write then @xmath92\ ] ] ( where the sun has magnitude m@xmath93 ) . following @xcite , from the assumed core mass - luminosity relation for the solar metallicity range , this leads to @xmath94 .
\label{mctip}\ ] ] figure 13 reports the inferred core - mass distribution , at the pn onset , for our galaxy sample .
note that this is the _ maximum _ actual mass allowed to luminous stars in each galaxy environment , and demonstrates that the _ mass of dying stars tends to decrease with increasing uv - upturn strength _ , being in general @xmath95 m@xmath2 among giant ellipticals . for this mass range ,
pn lifetime is the largest possible , but the timescale for the nebula to be visible is critically constrained by the transition time ( @xmath96 ) needed by the post - agb stellar core to be hot enough to `` fire up '' the ejected envelope and become a hard uv emitter @xcite .
the evident drop of @xmath77 among strong uv - upturn galaxies ( fig .
8) might be a direct consequence , therefore , of an increasing blocking effect of @xmath96 along the inferred @xmath97 range ( i.e. , the stellar core takes longer to heat - up than the shell to evaporate ; * ? ? ?
* ) , to which one has to further add a size cut in the overall pn population , as a result of the ehb progenitors ( m@xmath98 m@xmath2 ) evolving as _ agb - manqu _ stars and therefore skipping the nebula event . _
( c ) _ we remarked , in sec .
3 , the importance of the integrated @xmath82 index as a fairly selective tracer of the warm ( t@xmath99 k ) stellar component in the galaxy stellar population .
a proper assessment of the photometric contribution from this range of temperatures is of paramount importance in the framework of early - type galaxy evolution , in order to single out any signature of recent ( i.e. , in the last few gyr or so ) star formation or , conversely , of intervening evolution of the hb morphology among old ssps , as in a more standard canonical scenario . as far as uv - upturn galaxies are concerned
, the study of @xmath82 distribution clearly points to _ a substantial lack of a - type stars _ in the galaxy mix ( see fig .
while , on one hand , this definitely secures the `` quiescent '' nature of these galaxies , it also poses , on the other hand , a stringent constraint on hb morphology in their old - age context .
in fact , a bimodal temperature distribution is required for the hb to assure , at a time , both an enhanced uv emission _ and _ a conveniently low @xmath82 feature .
thus , in these systems the expected prevailing bulk of red hb stars should be accompanied , at some point , by a residual population of blue ( metal - poor ? ) hb objects ( coincident with ehb stars in the current empirical classification scheme ) , peaked at about 20000 - 40000 k , in a proportion of , roughly , @xmath100 $ ] . on the other hand , to complete the picture , one can not neglect the masking effects of age distribution , facing a recognized evidence for low - mass ellipticals to display a more silent but also more continuous star formation along their entire galaxy life , that naturally feeds the a - star contribution , thanks to the bluer ms turn - off point ( msto ) exhibited by ssps in the @xmath101 gyr age range , and leads to a younger average " age , compared to high - mass systems .
this is what we observe , for instance , among the resolved stellar populations of the local group dwarf spheroidals @xcite ( see , in this regard , the illustrative location of m32 in fig . 9 , and also consider the discussion by * ? ?
( d ) _ along with our discussion of the @xmath0 vs. @xmath66 relation , we noticed , in fig .
6 , the presence of a few outliers about 0.7 mag brighter at infrared magnitudes , or alternatively @xmath102 mag `` bluer '' in the @xmath66 color , than the main galaxy population . in order to further investigate this issue
, we tracked the relevant objects also in other figures , whenever possible . if galaxy mass ( _ alias _
@xmath103 ) is considered as the leading physical parameter to compare outlier location with respect to the bulk of the galaxy distribution ( see , for instance fig .
7 ) , one must conclude that both ngc 4552 and ngc 1389 seem to have a brighter agb tip rather than a bluer @xmath66 color . with regard to ngc 1389 ,
@xcite find that the @xmath104 sbfs of ngc 1389 are also too bright compared to its ( @xmath105 ) color , a fact that would be consistent with either a higher than average metallicity given the age of its most recent burst of star formation or a longer lifetime of its tp - agb stars @xcite . in the case of ngc 4552 , however , @xcite contribute an interesting piece of information .
these authors measure near - ir sbfs for several galaxies in virgo , and their results are systematically fainter than those obtained by @xcite .
unfortunately , the two groups use slightly different filters ( jensen et al .
employ @xmath106 , vs. @xmath40 of pahre & mould ) , but the discrepancy is larger than can be ascribed to the effect of the filters . in particular , jensen et al .
find @xmath107 for ngc 4552 ; assuming @xmath108 , ngc 4552 would no longer be deviant in the @xmath104 vs. ( @xmath109 ) plane .
concerning ngc 1387 , the dearth of data for this galaxy in the literature ( see table 2 ) makes it hard to propose an origin for its departure from the @xmath104 vs. ( @xmath109 ) sequence .
vs. ( @xmath105 ) correlation determined by @xcite . ]
on the other hand , ngc 1387 and ngc 1389 are lenticular galaxies , like the merger remnant ngc 1316 , but so are ngc 3384 and ngc 4406 , both of which do not deviate from the correlation . at any rate ,
ngc 1387 and ngc 1389 constitute privileged candidates for any future `` in - depth '' investigation .
we would like to thank gustavo bruzual for providing us with his latest ssp models , in advance of publication , and livia origlia , for useful discussions .
the anonymous referee is also acknowledged for his / her competent suggestions , that greatly helped refine the main focus of the paper .
partial financial support is acknowledged from the italian miur , under grant inaf - prin05 1.06.08.03 , and mexican conacyt , under grant no .
48589-f , and dgapa , under grant in111007 .
bailyn , c. d. 1995 , , 33 , 133 blakeslee , j. p. , vazdekis , a. , & ajhar , e. a. 2001 , , 320 , 193 boothroyd , a. i. , & sackmann , i .-
1988 , , 328 , 641 bressan , a. , chiosi , c. , & fagotto , f. 1994 , , 94 , 63 bressan , a. , chiosi , c. , & tantalo , r. 1996 , , 311 , 425 brodie , j. p. , & huchra , j. p. 1990
, , 362 , 503 brown , t. m. 2004 , astrophys .
space sci .
, 291 , 215 brown , t. m. , ferguson , h. c. , davidsen , a. f. , dorman , b. 1997 , 482 , 685 brown , t. m. , ferguson , h. c. , stanford , s. a. , deharveng , j .- m .
1998 , , 504 , 113 brown , t. m. , bowers , c. w. , kimble , r. a. , sweigart , a. v. , & ferguson , h. c. 2000 , , 532 , 308 brown , t. m. , ferguson , h. c. , smith , e. , bowers , c. w. , kimble , r. a. , renzini , a. , & rich , r. m. 2003 , , 584 , l69 brown , d. , yi , s. , han , z. , & yoon , s .- j . 2006 , baltic astronomy , 15 , 13 bruzual , g. , 2007 , in proc .
of the iau symp .
241 `` stellar populations as building blocks of galaxies '' , eds .
a. vazdekis and r. peletier ( cambridge : cambridge univ .
press ) , p.125 ( see also astro - ph/0703052 ) bruzual , g. , & charlot , s. 2003 , , 344 , 1000 buonanno , r. , corsi , c. , bellazzini , m. , ferraro , f. r. , & pecci , f. f. 1997 , , 113 , 706 burstein , d. , bertola , f. , buson , l. m. , faber , s. m. , & lauer , t. r. 1988 , , 328 , 440 buson , l. m. , bertone , e. , buzzoni , a. , & carraro , g. 2006 , baltic astronomy , 15 , 49 busso , g. , moehler , s. , zoccali , m. , heber , u. , & yi , s. k. 2005 , , 633 , l29 buzzoni , a. 1989 , , 71 , 817 buzzoni , a. 1993 , , 275 , 433 buzzoni , a. 1995 , , 98 , 69 buzzoni , a. 1998 in t. zanzu , v. testa & m. bellazzini eds . , evolving evolution , oss . di cagliari , cagliari , p. 13
( see also astro - ph/9811382 ) buzzoni , a. 2008 , in d. valls - gabaud & m. chavez eds .
, `` resolved stellar populations '' , asp conf .
( asp : san francisco ) , in press ( see also astro - ph/0509602 ) buzzoni , a. , gariboldi , g. , & mantegazza , l. 1992 , , 103 , 1814 buzzoni , a. , mantegazza , l. , & gariboldi , g. 1994 , , 107 , 513 buzzoni , a. , arnaboldi , m. , & corradi , r. l. m. 2006 , , 368 , 877 cantiello , m. , raimondo , g. , brocato , e. , & capaccioli , m. 2003 , , 125 , 2783 castellani , v. , & renzini , a. 1968 , , 2 , 310 castellani m. , & tornamb a. 1991 , apj , 381 , 393 castellani , v. , giannone , p. , & renzini , a. 1970 , , 9 , 418 castellani m. , limongi m. , tornamb a. 1992 , , 389 , 227 castellani , v. , iannicola , g. , bono , g. , zoccali , m. , cassisi , s. , & buonanno , r. 2006 , , 446 , 569 catelan , m. , ferraro , f. r. , & rood , r. t. 2001 , , 560 , 970 cervio , m. , & luridiana , v. 2006 , , 451 , 475 cervio , m. , luridiana , v. , & castander , f. j. 2000 , , 360 , l5 cervio , m. , valls - gabaud , d. , luridiana , v. , & mas - hesse , j. m. 2002 , , 381 , 51 chabrier , g. 2003 , , 115 , 763 charbonnel , c. , meynet , g. , maeder , a. , & schaerer , d. 1996 , , 115 , 339 chiosi , c. , bertelli , g. , bressan , a. 1992 , , 30,235 code , a. d. , & welch , g. a. 1979 , , 228 , 95 cohen j. g. 1982 , , 258 , 143 colina , l. , bohlin , r. c. , & castelli , f. 1996 , instrument science report cal / scs-008 ( baltimore , stsci ) cox , j.p . & giuli , r.t . 1968 principles of stellar structure , gordon & breach sci .
ltd , new york .
dcruz , n. l. , dorman , b. , rood , r. t. , & oconnell , r. w. 1996 , , 466 , 359 dcruz , n. l. , morgan , s. m. , bhm - vitense , e. 2000 , , 120 , 990 demleitner , m. , accomazzi , a. , eichhorn , g. , grant , c. s. , kurtz , m. j. , & murray , s. s. 2001 , in astronomical data analysis software and systems x , asp conf .
238 , eds .
f. r. harnden , jr .
, francis a. primini , and harry e. payne ( san francisco : asp ) , p.321 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h.g .
, buta , r.j . , paturel , g. , fouque , p. 1991
, third reference catalog of bright galaxies .
springer verlag , heidelberg dorman , b. , rood , r.t . , oconnell , r.w .
1993 , , 419 , 596 dorman , b. , oconnell , r. w. , & rood , r. t. 1995 , , 442 , 105 demarque , p. , mengel , j. g. , & sweigart , a. v. 1972 , , 173 , l27 elson , r. a. w. , & fall , s. m. 1985 , , 299 , 211 elson , r. a. , & fall , s. m. 1988 , , 96 , 1383 ferguson , h.c . , &
davidsen , a.f .
1993 , , 408 , 92 frogel , j.a . ,
mould , j. , & blanco , v.m . 1990 , , 352 , 96 fusi pecci , & f. , renzini , a. 1976 , , 46 , 447 gonzlez , r. a. , liu , m. c. , & bruzual a. , g. 2004 , , 611 , 270 height0.4pt depth0.0pt width1.0 in .2005 , , 621 , 557 gonzlez - lpezlira , r. a. , albarrn , m. y. , mouhcine , m. , liu , m .
c. , bruzual - a . , g. , & de batz , b. 2005 , , 363 , 1279 greggio , l. , & renzini , a. 1990 , , 364 , 35 groenewegen , m. a. t. , smith , c. h. , wood , p. r. , omont , a. , & fujiyoshi , t. 1995 , , 449 , l119 hui x. , ford h.c . , ciardullo r. , jacoby g.h .
1993 , apj , 414 , 463 iben , i. j. , & rood , r. t. 1970 , , 161 , 587 iben , i. , jr . , & renzini , a. 1983 , , 21 , 271 jensen , j. b. , luppino , g. a. , & tonry , j. l. 1996 , , 468 , 519 jensen , j. b. , tonry , j. l. , & luppino , g. a. 1998 , , 505 , 111 jensen , j. b. , tonry , j. l. , barris , b. j. , thompson , r. i. , liu , m. c. , rieke , m. j. , ajhar , e. a. , & blakeslee , j. p. 2003
, , 583 , 712 johnson , h. l. 1966 , , 4 , 193 kaluzny , j. , & udalski , a. 1992 , acta astronomica , 42 , 29 kobayashi , c. , & arimoto , n. 1999 , , 527 , 573 kuntschner , h. , lucey , j. r. , smith , r. j. , hudson , m. j. , & davies , r. l. 2001 , , 323 , 615 larson , r.b .
1974 , , 166 , 585 lee , m. g. , freedman , w. l. , & madore , b. f. 1993 , , 417 , 553 lee , y .- w . , ree , c. h. , rich , r.m . , et al . 2005 , , 619 , l103 liu , m. c. , graham , j. r. , & charlot , s. 2002 , , 564 , 216 maraston , c. , greggio , l. , & thomas , d. 2001 , , 276 , 893 marigo , p. , girardi , l. , weiss , a. , groenewegen , m. a. t. , & chiosi , c. 2004 , , 423 , 995 marigo , p. , & girardi , l. 2007 , , 469 , 239 mateo , m. l. 1998 , , 36 , 435 mouhcine , m. , gonzlez , r. a. , & liu , m. c. 2005 , , 362 , 1208 oconnell , r. w. 1980 , , 236 , 430 oconnell , r. w. 1999 , , 37 , 603 paczyski , b. 1970 , acta astronomica , 20 , 47 pahre , m. a. , de carvalho , r. r. , & djorgovski , s. g. 1998 , , 116 , 1606 pahre , m. a. , & mould , j. r. 1994 , , 433 , 567 park , j .- h . , & lee , y .- w . 1997 , , 476 , 28 paturel , g. , petit , c. , prugniel , p. , theureau , g. , rousseau , j. , brouty , m. , dubois , p. , & cambrsy , l. 2003 , , 412 , 45 peimbert m. 1990 , rev .
, 20 , 119 persson , s. e. , murphy , d. c. , krzeminski , w. , roth , m. , & rieke , m. j. 1998 , , 116 , 2475 piersanti , l. , tornamb , a. , & castellani , v. 2004 , , 353 , 243 raimondo , g. , brocato , e. , cantiello , m. , & capaccioli , m. 2005 , , 130 , 2625 recio - blanco , a. , aparicio , a. , piotto , g. , de angeli , f. , & djorgovski , s. g. 2006 , , 452 , 875 ree , c. h. , lee , y .- w . , sohn , y .-
j . , yi , s. k. , yoon , s .- j . , &
the galex science team 2006 , baltic astronomy , 15 , 5 reimers , d. 1975 , mem . soc .
lige , 6th ser . ,
8 , 87 renzini , a. 1977 , in advanced stages in stellar evolution i. , eds .
iben jr . , a. renzini , d.n .
schramm ( heidelberg : springer verlag ) .
renzini , a. 2006 , , 44 , 141 renzini , a. , & buzzoni , a. 1986 , in spectral evolution of galaxies , eds . c. chiosi & a. renzini ( dordrecht : reidel ) , p. 195 renzini , a. , & fusi pecci , f. 1988 , , 1988 , 26 , 199 rich , r. m. , et al .
1997 , , 484 , l25 rose , j.a .
, & deng , s. 1999 , , 117 , 2213 schnberner , d. 1983 , , 272 , 708 schiavon , r.p . , caldwell , n. , & rose , j.a .
2004 , , 127,1513 searle , l. , wilkinson , a. , & bagnuolo , w. g. 1980 , , 239 , 803 seidel , e. , da costa , g. s. , & demarque , p. 1987
, , 313 , 192 skrutskie , m. f. et al .
1997 , assl vol . 210 : the impact of large scale near ir sky surveys , 25 stanghellini , l. , & renzini , a. 2000 , , 542 , 308 sweigart , a. v. 1997 , , 474 , l23 sweigart , a. v. , & gross , p. g. 1976 , , 32 , 367 sweigart , a. v. , brown , t. m. , lanz , t. , landsman , w. b. , hubeny , i. 2002 in omega centauri , a unique window into astrophysics .
asp conf .
, vol . 265 .
f. van leeuwen , j. d. hughes , & g. piotto ( san francisco : asp ) p.261 tantalo , r. , chiosi , c. , bressan , a. , & fagotto , f. 1996 , , 311 , 361 tonry , j. l. 1991 , , 373 , l1 tonry , j. , & schneider , d. p. 1988
, , 96 , 807 unglaub , k. , bues , i. 2001 , , 374 , 570 van den bergh , s. 1962 , , 67 , 486 van loon , j. t. 2000 , , 354 , 125 worthey , g. 1993 , , 409 , 530 worthey , g. 1994 , , 95 , 107 yi , s. k. , & yoon , s .- j .
2004 , , 291 , 205 yi s. , demarque p. , oemler a. j. 1998 , , 492 , 480 | relying on infrared surface brightness fluctuactions to trace agb properties in a sample of elliptical galaxies in the virgo and fornax clusters , we assess the puzzling origin of the `` uv - upturn '' phenomenon , recently traced down to the presence of a hot horizontal branch ( hb ) stellar component .
we find that the uv - upturn actually signals a profound change in the c - m diagram of stellar populations in elliptical galaxies , involving both the hot stellar component and red - giant evolution .
first , we encounter that the strengthening of the uv rising branch is always seen to correspond to a shortening in agb deployment ; this trend can be readily interpreted as an age effect , perhaps mildly modulated by metal abundance .
a comparison between galaxy @xmath0 magnitudes and population synthesis models confirms that , all the way , brightest stars in ellipticals are genuine agb members , reaching the thermal - pulsing phase , and with the agb tip exceeding the rgb tip by some 0.5 - 1.5 mag . the inferred core mass of these stars is found to be @xmath1 m@xmath2 among giant ellipticals .
coupled with the recognized severe deficiency of planetary nebulae in these galaxies , this result strongly calls for an even more critical blocking effect due to a lengthy transition time needed by the post - agb stellar core to become a hard uv emitter and eventually `` fire up '' the nebula . the combined study of galaxy @xmath3 color and integrated h@xmath4 index points , as an explanation for the uv - upturn phenomenon , to a composite hb with a bimodal temperature distribution , i.e. with both a red clump and an extremely blue component , in a relative proportion of roughly [ n(rhb):n(bhb ) ] @xmath5 [ 80:20 ] . as far as metallicity of the bhb stellar population
is concerned , we find that [ fe / h ] values of either @xmath6 dex or @xmath7 may provide the optimum ranges to feed the needed low - mass stars ( m@xmath8 m@xmath2 ) , that at some stage begin to join the standard red - clump stars . |
the o(3 ) @xmath1-model in two dimensions @xcite @xcite is the simplest asymptotically free field theory and , as is well - known , it has instanton solutions @xcite . a continuum bound on the action exists for configurations with non - zero topological charge .
we will show that , while in the standard lattice formulation of this model @xcite , there are topologically non - trivial configurations with a smaller action than the continuum bound , this is not the case when we use an improved action along the lines described above . for a similar discussion in the context of renormalization improved actions see @xcite .
we will not address in this paper the interesting question of whether short - range configurations satisfying the continuum bound in this model can still dominate in the continuum limit , as argued by lscher @xcite .
the @xmath0 @xmath1-model in the continuum is defined by the action , @xmath46 where @xmath47 is a 3-component real field satisfying the constraint @xmath48 ( which defines the surface of a sphere of unit radius ) .
the continuum topological charge in this model is given by , @xmath49 which is the number of times that space - time wraps around the @xmath47-sphere .
using the identity @xcite , @xmath50 it follows that , @xmath51)^2 , \end{aligned}\ ] ] and , since the second term is positive definite , @xmath52 continuum instantons saturate this bound @xcite . in a standard lattice treatment
the action is , @xmath53 where @xmath32 are the sites of a two dimensional lattice and @xmath54 .
the topological charge as defined by berg and lscher @xcite is given by , @xmath55 @xmath56 \ ; f[s , s+\hat{\mu}+\hat{\nu},s+\hat{\nu } ] ] ] \ } , \nonumber\\ f[s_1,s_2,s_3 ] \equiv \frac { 1 + \sum_{i < j }
\vec{n}(s_i)\cdot \vec{n}(s_j ) + i \vec{n}(s_1 ) \cdot [ \vec{n}(s_2 ) \times \vec{n}(s_3 ) ] } { [ 2 \ ;
\prod_{i < j } ( 1 + \vec{n}(s_i)\cdot \vec{n}(s_j))]^{1/2 } } \nonumber\\ \label{topo3}\end{aligned}\ ] ] it is not hard to understand this formula . consider the plaquette @xmath57 . the spin variables at the corners are four points on the sphere .
we can form two triangles with corners at these points and with sides along geodesics , @xmath58 and @xmath59 .
@xmath60 is simply the sum of the area of these two image triangles on the @xmath47-sphere .
if periodic boundary conditions are imposed , @xmath61 is necessarily an integer . with these definitions of the action ( [ o3stan ] ) and the topological charge ( [ topo3 ] )
, the analysis in @xcite@xcite showed that there are dislocations , i.e. configurations with non - zero topological charge that have a smaller action than the continuum bound ( [ o3bound ] ) .
we will show that this picture changes considerably when the action of an interpolation of @xmath47 is used .
we first interpolate the lattice spin variables to a finer lattice , by moving along geodesics on the sphere .
the interpolation will be done locally , i.e. the interpolated fields within a plaquette @xmath62 of the @xmath15 lattice will only depend on the four spins associated with this plaquette , i.e. @xmath63 . the points of the finer lattice contained in the plaquette at @xmath32 , are @xmath64 , with @xmath65 and multiples of @xmath44 .
the spin fields at these points will be denoted by @xmath66 .
we first define the interpolated spin fields at the corners in the obvious way , @xmath67 we now interpolate along the one dimensional boundaries of the plaquette , by moving along geodesics on the sphere . for @xmath68 , @xmath69}{\sin[\theta(0,0 ; 1,0 ) ] } \vec{n}(0 , 0 ) + \frac{\sin[\theta(0,0 ; 1,0)\ ; t_1]}{\sin[\theta(0,0 ; 1,0 ) ] } \vec{n}(1 , 0)\nonumber\end{aligned}\ ] ] @xmath70}{\sin[\theta(0,1;1,1 ) ] } \vec{n}(0 , 1 ) + \frac{\sin[\theta(0,1;1,1)\ ; t_1]}{\sin[\theta(0,1;1,1 ) ] } \vec{n}(1 , 1),\nonumber\\\end{aligned}\ ] ] where we have defined , @xmath71 = \arccos[\vec{n}(t_1 , t_2)\cdot \vec{n}(t'_1 , t'_2)].\end{aligned}\ ] ] the expression for spins on the boundaries along direction @xmath72 are analogous so we skip the formulae .
we now define the spins in the interior . in order to do this we first interpolate along the diagonal geodesic connecting @xmath73 and @xmath74 , @xmath75}{\sin[\theta(0,0 ; 1,1 ) ] } \vec{n}(0 , 0 ) + \frac{\sin[\theta(0,0 ; 1,1 ) t]}{\sin[\theta(0,0 ; 1,1 ) ] } \vec{n}(1 , 1).\end{aligned}\ ] ] for the interior points with @xmath76 , we obtain the spin by simply moving a fraction @xmath77 of the distance along the geodesic from @xmath78 to @xmath79 . similarly for points with @xmath80 ,
the spin variable is obtained by moving a distance @xmath77 along the geodesic linking @xmath81 and @xmath82 .
this corresponds to separately interpolating the interior points of the two image triangles @xmath83 and @xmath84 .
defining , @xmath85 the final expression is , @xmath86}{\sin[\alpha ] } \vec{n}(t_1 , 0 ) + \frac{\sin[\alpha\ ; t_2/t_1]}{\sin[\alpha]}\vec{n}(t_1 , t_1 ) \;\;\ ; t_1 \geq t_2 \nonumber\\ \vec{n}(t_1 , t_2 ) = \frac{\sin [ \beta\ ; ( 1- \tilde{t}_2/\tilde{t}_1 ) ] } { \sin[\beta ] } \vec{n}(t_1 , 1 ) + \frac{\sin[\beta\ ; \tilde{t}_2/\tilde{t}_1]}{\sin[\beta ] } \vec{n}(t_1 , t_1 ) \;\;\ ; t_2
\geq t_1 \label{interpol}\end{aligned}\ ] ] with @xmath87 .
the interpolation ( [ interpol ] ) is now a configuration on an @xmath29-lattice and we define the improved action to be , @xmath88 where @xmath31 is any point on the @xmath29-lattice and @xmath89 .
now , we can easily derive a bound on the action , noticing that by construction , @xmath90 where @xmath91 is the topological charge for the @xmath29-lattice configuration : it is given by ( [ topo3 ] ) , simply substituting @xmath47 by @xmath92 and @xmath32 by @xmath31 .
now we can expand @xmath91 in @xmath93 and find that , @xmath94 using ( [ trick ] ) one finds , @xmath95)^2 + o(f / b).\end{aligned}\ ] ] then , @xmath96 since it is not possible to prove that the @xmath43 terms are positive definite , we do not know how small an @xmath44 we need to be sufficiently close to the continuum bound . in order to address this question , we considered the discretization of a continuum instanton configuration with unit topological charge given by @xcite , @xmath97 @xmath98 where @xmath99 .
we take @xmath100 and @xmath101 to be real and define , @xmath102 the radius of the instanton is proportional to @xmath103 , while the center is located at @xmath104 , with @xmath105 .
generically there is always a critical , @xmath106 , below which @xmath107 .
we expect , and find numerically , that @xmath108 .
we consider the center to be situated at the center of the volume to reduce finite volume effects . obviously , the continuum instanton is not periodic .
we impose periodic boundary conditions by defining , @xmath109 where @xmath110 are coordinates on the infinite plane and @xmath111 are the coordinates on the lattice ( torus ) .
@xmath112 is then the number of lattice sites .
the connection between these two sets of coordinates is established through sterographic projection .
this deformation of the infinite volume instanton is small if the instanton size is much smaller than @xmath113 . for such configurations
therefore , the continuum action nearly saturates the bound . an alternative procedure would be to discretize the continuum instanton solutions of this model in a torus @xcite .
figure 1 sumarizes our results .
it represents the action of the discretized instanton configuration as the radius is varied .
the continuous line corresponds to the standard action ( [ o3stan ] ) , while the dashed lines correspond to the improved action for different values of the ratio @xmath44 .
it is clear that the standard action is problematic , since for @xmath114 the action is smaller than the continuum bound .
for the new action however the situation is different . not only is the continuum bound satisfied , but also as the instanton is shrunk to sizes of @xmath20 , a small barrier develops , separating the @xmath115 and @xmath116 sectors .
the existence of this barrier is easy to understand .
the new action is the action of the instanton which has been first discretized on the @xmath15-lattice and then interpolated . for a large instanton ( compared to @xmath15 )
, the interpolation should recover approximately the original configuration and so the action must be near that of a continuum instanton and show scale invariance . on the other hand , for an instanton of size @xmath117 a lot of information is lost in the discretization and the interpolation is not expected to give a configuration similar to the original continuum instanton . in general , the interpolation of small discretized instantons of @xmath20 will then be some other configuration that need not be even approximately an instanton , and consequently its action will be larger than the bound , since only instantons saturate the bound
. thus the action must increase as we decrease @xmath118 near @xmath106 , as is clearly seen in fig .
the other important point to notice is that the continuum bound ( [ o3bound ] ) is satisfied even for a ratio @xmath44 as large as @xmath119 .
this indicates that the extra effort required to use the improved action is clearly managable in this case .
we have presented a new action for non - abelian gauge theories which is better suited to studying topology on the lattice than the standard wilson action .
the idea is to use the wilson action of an interpolation to a finer lattice of the original lattice configuration . in this way , if the ratio of lattice spacings is small enough , the continuum bound on configurations with non - zero topological charge is satisfied . in non - abelian gauge theories
this is a sufficient condition for proper scaling of the topological susceptibility .
we also considered a new action for the @xmath0 @xmath1-model in two dimensions along the same lines .
a numerical study of discretized continuum instantons in this model indicates that already for a ratio of lattice spacings of @xmath119 , the continuum bound is satisfied .
although the results for the @xmath0 model are very promising , a separate numerical analysis is needed in the yang - mills case to determine the ratio there .
this research was supported by nsf - phys-92 - 18167 .
we thank u.j .
wiese for discussions .
is supported by the harvard society of fellows and also partially by the milton fund of harvard university .
g. thooft , phys . rev . * d14*(1976)3432 , phys
. rev . lett .
* 37 * ( 1976 ) 8 .
s. weinberg , phys .
rev . * d12 * ( 1975 ) 3583 .
e. witten , nucl .
phys . * b149*(1979 ) 285 .
g. veneziano , nucl . phys . * b159*(1979 ) 213 .
p. divecchia et al . , nucl . phys . *
b192 * ( 1981 ) 392 .
m. lscher , comm .
* 85 * ( 1982 ) 39 .
b. berg and m. lscher , nucl .
b190 * ( 1981 ) 412 .
m. lscher , nucl . phys .
* b200 * ( 1982 ) 61 .
m. gckeler et al . , phys . lett .
* 233b * ( 1989 ) 192 and references therein .
d.j.r . pugh and m. teper ,
. lett . * 224b*(1989 ) 159 and references therein .
m. kremer , et al .
phys.*b305*(1988 ) 109 .
p. hasenfratz and f. niedermayer , nucl
. phys . * b414*(1994 ) 785 .
t. degrand , a. hasenfratz , p. hasenfratz , f. niedermayer , nucl . phys . * b454*(1995 ) 587 and nucl . phys . * b454*(1995 ) 615 .
t. degrand , a. hasenfratz and d. zhu , hep - lat/9603015 .
m. gckeler , a. kronfeld , g. schierholz and u.j .
wiese , nucl .
phys . * b404*(1993 ) 839 .
p. hernndez and r. sundrum , hep - lat/9602017 .
a.a . belavin and a. m. polyakov , jetp lett * 22 * ( 1975 ) 245 .
h. eichenherr , nucl .
bf b146(1978 ) 215 ; e. cremmer and j. scherk , phys .
bf 74b ( 1978 ) 341 ; a. dadda , p. di vecchia and m. lscher , nucl .
bf b146(1978 ) 112 .
m. blatter , r. burkhalter , p. hasenfratz and f. niedermayer , hep - lat/9508028 .
see for example gauge fields and strings , a.m. polyakov , contemporary concepts in physics , volume 3 . | we propose a new lattice action for non - abelian gauge theories , which will reduce short - range lattice artifacts in the computation of the topological susceptibility .
the standard wilson action is replaced by the wilson action of a gauge covariant interpolation of the original fields to a finer lattice .
if the latter is fine enough , the action of all configurations with non - zero topological charge will satisfy the continuum bound . as a simpler example we consider the @xmath0 @xmath1-model in two dimensions , where a numerical analysis of discretized continuum instantons indicates that a finer lattice with half the lattice spacing of the original is enough to satisfy the continuum bound .
april , 1996 hutp-96/a015 + 0.2 in * a new lattice action for studying topological charge * .2 in * pilar hernndez * + and + * raman sundrum * 0.3 cm _ lyman laboratory of physics + harvard university + cambridge , ma 02138 , usa _ 0.7 cm field configurations with non - zero topological charge are expected to have a strong influence on the dynamics of asymptotically free theories . in qcd ,
such configurations are responsible for breaking axial symmetry and resolving the u(1 ) problem @xcite .
the study of these effects however requires non - perturbative techniques and one would expect that ultimately monte carlo methods on the lattice would be best suited to it .
the observable to consider is inspired by the classic large-@xmath2 analyses of witten and of veneziano @xcite , which showed that , @xmath3 where @xmath4 is the topological susceptibility . in the continuum it is given by , @xmath5 with @xmath6 being the topological charge density , @xmath7 .
\label{contq}\end{aligned}\ ] ] the topological charge , @xmath8 , is an integer if the field strength vanishes at infinity or if ( euclidean ) space - time is compact .
a continuum analysis also shows that the action of any configuration with non - zero topological charge must satisfy the following bound , @xmath9 a big effort has been devoted to the study of the topological susceptibility on the lattice .
there are several choices for the operator @xmath6 .
the naive discretization of ( [ contq ] ) does not yield integer values for @xmath10 and requires renormalization factors @xcite . on the other hand ,
the cleaner geometrical definition due to lscher @xcite gives an integer - valued topological charge and does not require renormalization .
here we will deal only with a geometrical definition very similar to lscher s .
the geometrical topological susceptibility is then obtained by computing , @xmath11 where @xmath12 is the volume of the lattice .
this definition is clearly equivalent to ( [ chi ] ) in finite volume .
from the continuum formula ( [ eta ] ) , it follows that the topological susceptibility in qcd should scale as @xmath13 in the continuum limit , namely @xmath14 where @xmath15 is the lattice spacing and @xmath16 is the leading term of the beta function . however , it was found , first in the @xmath0 @xmath1-model in two space - time dimensions @xcite@xcite and then in non - abelian gauge theories in four @xcite @xcite , that the standard actions in both cases give rise to short - range fluctuations with non - zero geometrical topological charge and smaller action than the continuum bound .
these fluctuations , often referred to as `` dislocations '' , overwhelm the contribution of the slowly varying fields , which would otherwise dominate in the continuum limit , and are expected to destroy the scaling of the topological susceptibility . on the other hand , if the bound is satisfied , the semiclassical continuum analysis of non - abelian gauge theories indicates that the susceptibility should indeed be ultraviolet finite , and therefore scale .
satisfying the continuum bound is thus a sufficient condition for scaling in non - abelian gauge theories @xcite@xcite@xcite@xcite .
@xmath1-model , the failure in finding scaling is not related to the existence of dislocations and is an essential problem of this model : the partition functional in the instanton sector shows a logarithmic ultraviolet divergence in the continuum analysis .
] this suggests that eq .
( [ contbound ] ) can be satisfied on the lattice by giving dislocations a larger action .
several proposals to solve this problem have been considered in the literature , like the cooling method @xcite and the use of improved actions @xcite @xcite . in this paper
we propose to use a new action which satisfies the continuum bound .
it is related in spirit to the actions proposed in @xcite , but may be simpler to implement .
the idea behind our new action is easy to understand once one realizes the mismatch between the geometrical definition of topological charge @xcite and the wilson action for gauge fields which allows dislocations to arise .
consider a continuum instanton @xmath17 which saturates the bound ( [ contbound ] ) , and discretize it on a lattice of spacing @xmath15 , @xmath18 where @xmath19 are the sites of the @xmath15 lattice .
the geometrical definition of topological charge assigns a non - zero value even to a lattice configuration ( [ dis ] ) obtained from very small instantons , of order of the lattice spacing @xmath15 . on the other hand ,
it is clear that the wilson action very poorly approximates the continuum action of the original instanton of @xmath20 , and in fact it is generically smaller and therefore does not satisfy the bound .
such rough configurations are dislocations and can destroy the scaling of the susceptibility .
an important observation is that the geometrical topological charge assigned to a lattice configuration can be understood as the naive topological charge of a _ continuum _ configuration obtained by smoothly interpolating the lattice configuration .
then it is clear that if , instead of using the standard wilson action of the original lattice configuration , we use the continuum action of the interpolated configuration , the continuum bound is necessarily satisfied , as first suggested in @xcite .
more concretely , in @xcite we described a procedure to obtain a continuum gauge field @xmath21 which interpolates any @xmath15-lattice configuration - hypercube and transversely continous across the @xmath15-boundaries , and when discretized according to ( [ dis ] ) gives back @xmath22 @xcite . ] .
the interpolation is local and gauge covariant , i.e. for a @xmath15-lattice gauge transformation @xmath23 , @xmath24 = a_\mu^\omega[u ] , \label{gi}\end{aligned}\ ] ] where @xmath25 is a gauge transformation in the continuum .
( other interpolations can also be used in this context @xcite . ) a geometrical topological charge of the @xmath15-lattice configuration is defined as the one associated to the interpolated field @xcite , @xmath26 .
\end{aligned}\ ] ] this definition has the same properties as lscher s original definition @xcite and requires roughly the same computational effort .
now , it is clear that replacing the standard wilson action by the continuum action of @xmath27 , i.e. @xmath28,\end{aligned}\ ] ] insures that the continuum bound is satisfied for the same reason as it is in the continuum .
notice that this is a perfectly gauge invariant action for the lattice field @xmath22 , by eq .
( [ gi ] ) .
however , for numerical purposes the continuum action is impractical @xcite .
the central observation of this paper is that it is not necessary to interpolate all the way to the continuum , but just to a finer lattice , with lattice spacing @xmath29 ( we will take @xmath30 to be integer ) . from now on ,
we refer to @xmath31 as the sites on the @xmath29-lattice and @xmath32 as the sites on the @xmath15-lattice ( @xmath33 , @xmath34 ) .
the interpolation procedure in @xcite gives a set of link variables @xmath35(x)$ ] , such that @xmath36 where @xmath37 is the continuum interpolation discussed above , at point @xmath31 . on the @xmath29 lattice
, we can simply choose the standard wilson action .
the partition functional will then have the form , @xmath38 } , \label{effact}\end{aligned}\ ] ] where , @xmath39 is the usual haar measure for non - abelian gauge fields on the @xmath15-lattice , and the wilson action in terms of the interpolated link variables @xmath40 $ ] is given by , @xmath41 = \frac{1}{g_0 ^ 2 } \sum_x \sum_{\mu\neq\nu } ( i - u_{\mu\nu}[x ] + h.c . ) , \nonumber\\ u_{\mu\nu}[x ] \equiv tr[u_\mu(x ) u_\nu(x+\hat{\mu } ) u_\mu^\dagger(x+\hat{\nu } ) u_\nu^\dagger(x ) ] .
\label{actna}\end{aligned}\ ] ] again , this action is gauge invariant , because the functional @xmath40 $ ] is gauge covariant @xcite . from eq .
( [ imp ] ) it then follows that , @xmath42 + o(f / b).\end{aligned}\ ] ] although the @xmath43 terms are not necessarily positive definite , clearly by taking @xmath44 small enough we can come arbitrarily close to satisfying the continuum bound , so that @xmath4 shows scaling .
determining how small this ratio must be in practice requires a numerical analysis , which is beyond the scope of this letter .
however , we have carried out a numerical analysis of this issue in a simplified model in two dimensions , which shares many features with four dimensional yang - mills theories and our results encourage us to believe that @xmath45 need not be very small in order to recover scaling of @xmath4 . |
all graphs in this paper are finite and simple .
a graph @xmath15 is a of a graph @xmath16 if @xmath15 can be obtained from a subgraph of @xmath16 by contracting edges .
we write @xmath17 if @xmath15 is a minor of @xmath16 . in those circumstances
we also say that @xmath16 has an @xmath15 .
our work is motivated by the following hadwiger s conjecture @xcite , which is perhaps the most famous conjecture in graph theory , as pointed out by paul seymour in his recent survey @xcite .
[ hc ] for every integer @xmath0 , every graph with no @xmath18 minor is @xmath1-colorable .
hadwiger s conjecture is trivially true for @xmath19 , and reasonably easy for @xmath20 , as shown by dirac @xcite .
however , for @xmath21 , hadwiger s conjecture implies the four color theorem .
wagner @xcite proved that the case @xmath22 of hadwiger s conjecture is , in fact , equivalent to the four color theorem , and the same was shown for @xmath23 by robertson , seymour and thomas @xcite .
conjecture remains open for @xmath24 . as pointed out by paul
seymour @xcite in his recent survey on hadwiger s conjecture , proving that graphs with no @xmath3 minor are @xmath4-colourable is thus the first case of hadwiger s conjecture that is still open .
it is not even known yet whether every graph with no @xmath3 minor is @xmath5-colorable .
kawarabayashi and toft @xcite proved that every graph with no @xmath3 or @xmath25 minor is @xmath4-colorable .
jakobsen @xcite proved that every graph with no @xmath26 minor is @xmath4-colorable and every graph with no @xmath27 minor is @xmath5-colorable , where for any integer @xmath28 , @xmath29 ( resp .
@xmath30 ) denotes the graph obtained from @xmath13 by removing one edge ( resp .
two edges ) . for more information on hadwiger s conjecture ,
the readers are referred to an earlier survey by toft @xcite and a very recent informative survey due to seymour @xcite .
albar and gonalves @xcite recently proved the following : [ k7k8 ] ( albar and gonalves @xcite ) every graph with no @xmath3 minor is @xmath12-colorable , and every graph with no @xmath31 minor is @xmath32-colorable the proof of theorem [ k7k8 ] is computer - assisted and not simple . in this paper
, we apply a kempe - chain argument ( see lemma [ wonderful ] below ) along with the fact that an induced path on three vertices is dominating in a graph with independence number two in order to give a much shorter and computer - free proof of theorem [ k7k8 ] .
in addition , we generalize it to the next step by proving the following .
[ main ] every graph with no @xmath6 minor is @xmath7-colorable , where @xmath33 .
we want to point out that our proof of theorem [ main ] does not rely on mader s deep result on the connectivity of contraction - critical graphs ( see theorem [ 7con ] below ) .
theorem [ main ] states that ( i ) every graph with no @xmath3 minor is @xmath12-colorable ; ( ii ) every graph with no @xmath31 minor is @xmath32-colorable ; and ( iii ) every graph with no @xmath34 minor is @xmath35-colorable .
we prove theorem [ main ] in section [ kt ] . applying the method we developed in the proof of theorem [ main ] and mader s deep result ( theorem [ 7con ] ) ,
we then prove two new results theorem [ maink8- ] and theorem [ maink8= ] .
[ maink8- ] every graph with no @xmath9 minor is @xmath10-colorable .
[ maink8= ] every graph with no @xmath11 minor is @xmath12-colorable .
our proofs of theorem [ maink8- ] and theorem [ maink8= ] are both short and computer - free and will be presented in section [ k8- ] and section [ k8= ] , respectively . to end this paper , we first propose a conjecture in section [ remarks ] .
we then apply lemma [ wonderful ] to prove that if conjecture [ conj3 ] ( see section [ remarks ] ) is true , then every graph with no @xmath13 minor is @xmath7-colorable for all @xmath14 .
our proof of the last result does not rely on the connectivity of contraction - critical graphs and the new idea we introduced yields a different / short proof of theorem [ main ] . to prove our results
, we need to investigate the basic properties of contraction - critical graphs . for a positive integer @xmath1 ,
a graph @xmath16 is if @xmath36 and any proper minor of @xmath16 is @xmath37-colorable .
lemma [ dirac ] below is a folklore result which is an extension of dirac s initial work @xcite on contraction - critical graphs . a proof of lemma [ dirac ]
can be easily obtained from the definition of @xmath38-contraction - critical graphs .
[ dirac](dirac @xcite ) every @xmath38-contraction - critical graph @xmath16 satisfies the following : for any @xmath39 , @xmath40 , where @xmath41 denotes the independence nunber of the subgraph of @xmath16 induced by @xmath42 ; no minimal separating set of @xmath16 is a clique .
lemma [ wonderful ] below on contraction - critical graphs turns out to be very powerful , as the existence of pairwise vertex - disjoint paths is guaranteed without using the connectivity of such graphs .
if two vertices @xmath43 in a graph @xmath16 are not adjacent , we say that @xmath44 is a of @xmath16 .
one possible application of lemma [ wonderful ] is depicted in figure [ wonderfullemma ] .
[ wonderful ] let @xmath16 be any @xmath38-contraction - critical graph .
let @xmath45 be a vertex of degree @xmath46 with @xmath47 and let @xmath48 with @xmath49 be any independent set , where @xmath50 and @xmath51 are integers . if @xmath52 is not a clique , then for any @xmath53 , where @xmath54 , @xmath55 , the vertices @xmath56 are all distinct , and for any @xmath57 , the set @xmath58 consists of @xmath59 missing edges of @xmath52 with @xmath60 as a common end , then for each @xmath57 there exist paths @xmath61 in @xmath16 such that each @xmath62 has ends @xmath63 and all its internal vertices in @xmath64 $ ] for all @xmath65 .
moreover , for any @xmath66 , the paths @xmath61 are vertex - disjoint from the paths @xmath67 .
to @xmath68 with @xmath69 and @xmath70.,width=220 ] let @xmath16 , @xmath68 , @xmath71 and @xmath72 be as given in the statement .
let @xmath15 be obtained from @xmath16 by contracting @xmath73 into a single vertex , say @xmath74 .
then @xmath15 is @xmath75-colorable .
let @xmath76 be a proper @xmath75-coloring of @xmath15 .
we may assume that @xmath77 .
then each of the colors @xmath78 must appear in @xmath52 , else we could assign @xmath68 the missing color and all vertices in @xmath71 the color @xmath79 to obtain a proper @xmath75-coloring of @xmath16 , a contradiction . since @xmath80
, we have @xmath81 for any two distinct vertices @xmath82 in @xmath52 .
we next claim that for each @xmath83 and each @xmath84 there must exist a path between @xmath60 and @xmath85 with its internal vertices in @xmath64 $ ] .
suppose not .
let @xmath83 and @xmath86 be such that there is no such path between @xmath60 and @xmath85 .
let @xmath87 be the subgraph of @xmath15 induced by the vertices colored @xmath88 or @xmath89 under the coloring @xmath90 . then @xmath91 .
notice that @xmath60 and @xmath85 must belong to different components of @xmath87 as there is no path between @xmath60 and @xmath85 with its internal vertices in @xmath64 $ ] . by switching the colors on the component of @xmath87 containing @xmath60
, we obtain a @xmath75-coloring of @xmath15 with the color @xmath88 missing on @xmath52 , a contradiction .
this proves that there must exist a path @xmath62 in @xmath87 with ends @xmath63 and all its internal vertices in @xmath92 $ ] for each @xmath83 and each @xmath84 .
clearly , for any @xmath66 , the paths @xmath61 are vertex - disjoint from the paths @xmath67 , because no two vertices of @xmath93 are colored the same under the coloring @xmath90 .
width6pt depth2pt + * remark . *
if @xmath94 in the statement of lemma [ wonderful ] , we simply write @xmath95 , and so @xmath72 is a matching of missing edges of @xmath52 . in this case , the paths @xmath96 are pairwise vertex - disjoint if @xmath97 .
if @xmath98 in the statement of lemma [ wonderful ] , we simply write @xmath99 . in this case , the paths @xmath100 have @xmath101 as a common end and are not necessarily pairwise internally vertex - disjoint if @xmath102 .
theorem [ 7con ] below is a deep result of mader @xcite and will be used in the proofs of theorem [ maink8- ] and theorem [ maink8= ] .
it seems very difficult to improve theorem [ 7con ] for small values of @xmath38 . for larger values of @xmath38
, some better results can be found .
kawarabayashi @xcite has shown that any minimal non - complete @xmath38-contraction - critical graph with no @xmath103 minor is @xmath104-connected , while kawarabayashi and yu @xcite have shown that any minimal such graph is @xmath105-connected .
chen , hu and song @xcite recently improved the bound further by showing that any minimal such graph is @xmath106-connected .
[ 7con ] ( mader @xcite ) for @xmath107 , every @xmath38-contraction - critical graph is @xmath5-connected .
we also need the following lemma in the proofs of theorem [ maink8- ] and theorem [ maink8= ] .
[ lem:7conn2k6 ] for any @xmath5-connected graph @xmath16 , if @xmath16 contains two different @xmath108-subgraphs , then @xmath109 .
let @xmath110 be two different @xmath108-subgraphs of @xmath16 with @xmath111 and @xmath112 .
let @xmath113 .
then @xmath114 .
we may assume that @xmath115 for all @xmath116 if @xmath117 .
assume that @xmath118 .
then @xmath119 has a @xmath120-subgraph of @xmath16 .
since @xmath16 is @xmath5-connected , it is easy to see that @xmath109 by contracting a component of @xmath121 into a single vertex .
so we may assume that @xmath122 .
then there exist @xmath123 pairwise disjoint paths @xmath124 between @xmath125 and @xmath126 in @xmath127 .
we may assume that @xmath128 has ends @xmath129 for all @xmath130 .
then @xmath131 is connected , so there must exist a path @xmath132 with one end , say @xmath133 , in @xmath134 , the other end , say @xmath135 , in @xmath136 , and no internal vertices in any of @xmath137 .
we may assume that @xmath133 lies on the path @xmath138 .
let @xmath139 be the subpath of @xmath140 with ends @xmath141 , and @xmath142 be the subpath of @xmath143 with ends @xmath144 .
now contracting @xmath139 onto @xmath145 , @xmath146 onto @xmath147 , @xmath142 and @xmath148 onto @xmath149 , @xmath150 onto @xmath151 , and each of @xmath152 to a single vertex , together with @xmath153 if @xmath117 , yields a @xmath9 minor in @xmath16 , as desired .
depth2pt + we need to introduce more notation . for a graph @xmath16
we use @xmath154 and @xmath155 to denote the number of vertices and minimum degree of @xmath16 , respectively . for a subset @xmath71 of @xmath156 , the subgraph induced by @xmath71
is denoted by @xmath157 $ ] and @xmath158 $ ] .
the degree and neighborhood of a vertex @xmath135 in @xmath16 are denoted by @xmath159 and @xmath42 , respectively . by abusing notation
we will also denote by @xmath42 the graph induced by the set @xmath42 .
we define @xmath160=n(v)\cup \{v\}$ ] , and similarly will use the same symbol for the graph induced by that set . for @xmath161 ,
if @xmath157 $ ] is connected , then we denote by @xmath162 the graph obtained from @xmath16 by contracting @xmath157 $ ] into a single vertex and deleting all resulting parallel edges and loops . for @xmath163
, we say that @xmath164 is to @xmath165 if each vertex in @xmath164 is adjacent to all vertices in @xmath165 , and @xmath164 is to @xmath165 if no vertex in @xmath164 is adjacent to any vertex in @xmath165 .
if @xmath166 , we simply say @xmath167 is complete to @xmath165 or @xmath167 is anti - complete to @xmath165 . the @xmath168 ( resp
@xmath169 ) of two vertex disjoint graphs @xmath16 and @xmath15 is the graph having vertex set @xmath170 and edge set @xmath171 ( resp .
@xmath172 ) .
results on the extremal function for @xmath6 minors will be needed to prove theorem [ main ] . before doing so , we need to define @xmath173-cockade . for
graphs @xmath110 and an integer @xmath38 , let us define an _
@xmath174-cockade _ recursively as follows .
any graph isomorphic to @xmath175 or @xmath176 is an @xmath174-cockade .
now let @xmath177 , @xmath178 be @xmath174-cockades and let @xmath16 be obtained from the disjoint union of @xmath177 and @xmath178 by identifying a clique of size @xmath38 in @xmath177 with a clique of the same size in @xmath178 . then the graph @xmath16 is also an @xmath174-cockade , and every @xmath174-cockade can be constructed in this way .
if @xmath179 , then @xmath16 is simply called an @xmath180-cockade .
the following theorem [ k7 ] was first shown by dirac @xcite for @xmath181 and by mader @xcite for @xmath182 .
[ k7](mader @xcite ) for every integer @xmath183 , a graph on @xmath184 vertices and at least @xmath185 edges has a @xmath13 minor . the edge bound in theorem [ k7 ]
is referred to as mader s bound for the extremal function for @xmath13 minors .
jrgensen @xcite and later the second author and thomas @xcite generalized theorem [ k7 ] to @xmath186 and @xmath187 , respectively , as follows .
[ k8](jrgensen @xcite ) every graph on @xmath188 vertices with at least @xmath189 edges either has a @xmath31 minor or is a @xmath190-cockade .
[ k9](song and thomas @xcite ) every graph on @xmath191 vertices with at least @xmath192 edges either has a @xmath193 minor , or is a @xmath194-cockade , or is isomorphic to @xmath195 .
it seems hard to generalize theorem [ k7 ] for all values of @xmath196 . in 2003 ,
seymour and thomas @xcite proposed the following conjecture . [
conj2](seymour and thomas @xcite ) for every @xmath197 there exists a constant @xmath198 such that every @xmath199-connected graph on @xmath200 vertices and at least @xmath185 edges has a @xmath13 minor . by theorem [ k9 ] ,
conjecture [ conj2 ] is true for @xmath201 .
we next prove the following lemma [ k_tk1 ] , which can be obtained from the ( computer - assisted ) proof of lemma 3.7 in @xcite . here
we give a computer - free proof of lemma [ k_tk1 ] so that the proof of theorem [ main ] is also computer - free .
[ k_tk1 ] for @xmath202 , let @xmath15 be a graph with @xmath203 vertices and @xmath204
. then @xmath205 .
suppose that @xmath15 has no @xmath206 minor .
then @xmath207 .
we claim that [ e : nok_t-3 ] ( ) @xmath208 .
suppose that @xmath209 .
let @xmath210 be isomorphic to @xmath211 . then @xmath212 .
if @xmath213 contains an induced @xmath214-path , say @xmath215 , with ends @xmath216 , where a @xmath214-path is a path with three vertices , then every vertex of @xmath217 is adjacent to either @xmath218 or @xmath219 because @xmath204 . by contracting the path @xmath215 into a single vertex
, we see that @xmath220>k_{t-2}$ ] and so @xmath205 , a contradiction .
thus @xmath213 does not contain an induced path on three vertices .
since @xmath204 , it follows that @xmath213 is a disjoint union of two cliques , say @xmath221 and @xmath222 . for @xmath223 , let @xmath224 .
since @xmath204 , @xmath225 is complete to @xmath226 for each @xmath227 .
thus @xmath228 is a clique for each @xmath229 and so either @xmath230 or @xmath231 is a clique of size at least @xmath232 , contrary to the fact that @xmath207 . height3pt width6pt
depth2pt let @xmath233 and let @xmath234 be a vertex with @xmath235 .
let @xmath236 $ ] .
since @xmath204 , @xmath237 is a clique of size @xmath238 . by , @xmath239 and
so @xmath240 . we next show that [ e:2cliques ] ( ) for any @xmath241 with @xmath242 , either @xmath243 $ ] contains two vertex - disjoint @xmath214-paths or @xmath244 $ ] is a disjoint union of two cliques .
suppose @xmath244 $ ] is not a disjoint union of two cliques .
then @xmath244 $ ] is connected because @xmath204 .
we next show that @xmath243 $ ] contains two vertex - disjoint induced @xmath214-paths . by , @xmath244
$ ] is not a clique and thus contains an induced @xmath214-path , say @xmath245 , with ends @xmath246 .
let @xmath247 , where @xmath248 .
clearly @xmath243 $ ] contains two vertex - disjoint induced @xmath214-paths if @xmath249 $ ] is not a clique , since @xmath250 is an edge for @xmath251 .
so we may assume that @xmath249=k_s$ ] .
first assume that @xmath167 is complete to @xmath252 .
by , @xmath253 is not complete to @xmath254 .
we may assume that @xmath255 .
clearly @xmath256 $ ] and @xmath257 $ ] are two vertex - disjoint @xmath214-paths if @xmath258 for some @xmath259 .
so we may assume that @xmath260 .
now either @xmath261 $ ] and @xmath262 $ ] ( if @xmath263 ) or @xmath264 $ ] and @xmath265 $ ] ( if @xmath266 ) are two vertex - disjoint @xmath214-paths .
next assume that @xmath167 is not complete to @xmath252 .
we may assume that @xmath267 .
then @xmath268 because @xmath204 . by symmetry
, we may assume that @xmath263 .
then @xmath269 . now either @xmath270 $ ] and @xmath271 $ ] ( if @xmath272 ) or @xmath273 $ ] and @xmath262 $ ] ( if @xmath274 ) are two vertex - disjoint @xmath214-paths , as desired . height3pt width6pt
depth2pt [ e : q = t-2 ] ( ) @xmath275 .
suppose @xmath276 . by theorem [ k7 ] , @xmath277 , which yields that @xmath278 and @xmath279 .
then @xmath15 is a graph on thirteen vertices .
clearly , @xmath280 . let @xmath281 be such that @xmath282 is maximum . since @xmath283
, we have @xmath284 .
if @xmath285 , then @xmath286=k_5 $ ] and @xmath287 . clearly
@xmath288 if @xmath289{\backslash}z$ ] has two vertex - disjoint induced @xmath214-paths . by , @xmath289{\backslash}z$ ]
is thus a disjoint union of two cliques , say @xmath290 . by
, we may assume that @xmath291 and @xmath292 . let @xmath293 .
by again , @xmath167 is not complete to @xmath294 and thus @xmath295 , contrary to the fact that @xmath296 .
thus @xmath297 .
let @xmath298 be the non - neighbor of @xmath219 .
by the choice of @xmath219 , every vertex in @xmath299 has at least one non - neighbor in @xmath237 and so @xmath300 . since @xmath301 , @xmath302 .
by , @xmath303 is not a clique and so @xmath304 is adjacent to at least one vertex , say @xmath74 , in @xmath303 , because @xmath204 . now
the edge @xmath305 is dominating @xmath237 , i.e. , every vertex in @xmath237 is adjacent to either @xmath219 or @xmath74 . notice that @xmath306 .
if @xmath289{\backslash}\{z , w\}$ ] contains two vertex - disjoint induced @xmath214-paths , say @xmath307 and @xmath308 , then @xmath288 by contracting the edge @xmath305 and the two @xmath214-paths @xmath307 and @xmath308 into three distinct vertices , respectively , a contradiction . thus @xmath289{\backslash}\{z , w\}$ ] does not contain two vertex - disjoint induced @xmath214-paths . by , @xmath309 is a disjoint union of two cliques , say @xmath310 and @xmath311 . since @xmath312
, we have @xmath313 . by , @xmath314
$ ] is not a clique .
let @xmath315 be such that either @xmath316 or @xmath317 .
since @xmath318 is adjacent to at most three vertices of @xmath237 , we see that @xmath319 , contrary to the fact that @xmath296 .
depth2pt by , @xmath275 .
if @xmath320 , then @xmath15 is a graph on nine vertices with @xmath321 .
thus there exists a vertex @xmath322 such that @xmath323 and so @xmath324 $ ] contains a @xmath325-subgraph because @xmath326 , contrary to .
hence @xmath327 .
now @xmath328 $ ] is a clique of size @xmath329 and @xmath330 .
clearly @xmath205 if @xmath331 contains two vertex - disjoint induced 3-paths , a contradiction .
thus by , @xmath299 is a disjoint union of two cliques , say @xmath221 and @xmath222 . for @xmath223 ,
let @xmath332 : v \text { is not adjacent to some vertex in } a_{3-i}\}$ ] .
since @xmath204 , @xmath225 is complete to @xmath226 for each @xmath227 .
thus @xmath333 is a clique for each @xmath229 and so at least one of them is of size at least @xmath334 , contrary to .
this completes the proof of lemma [ k_tk1 ] .
depth2pt + we are now ready to prove theorem [ main ] .
* proof of theorem [ main ] .
* suppose the assertion is false .
let @xmath16 be a graph with no @xmath6 minor such that @xmath16 is not @xmath7-colorable .
we may choose such a graph @xmath16 so that it is @xmath335-contraction - critical .
let @xmath336 be of minimum degree .
since @xmath195 and each @xmath190-cockade are @xmath337-colorable , and every @xmath338-cockade is @xmath4-colorable , it follows from theorem [ k7 ] , theorem [ k8 ] and theorem [ k9 ] that @xmath339 .
on the other hand , since @xmath16 is @xmath335-contraction - critical , by lemma [ dirac](i ) applied to @xmath340 , we see that @xmath341 .
clearly , @xmath342 , otherwise @xmath343 $ ] is a clique of size at least @xmath344 , contrary to the fact that @xmath16 has no @xmath6 minor .
thus @xmath342 and so @xmath345 .
then @xmath346 . by lemma [ dirac](i ) applied to @xmath340 again , we have @xmath347 . we next show that [ e : nok_t-1 ] ( ) @xmath16 has no @xmath348-subgraph .
suppose @xmath16 contains @xmath348 as a subgraph .
let @xmath349 be isomorphic to @xmath348 .
since @xmath350 , every vertex in @xmath15 is adjacent to at least one vertex in @xmath351 .
then @xmath351 is disconnected , since otherwise @xmath352 by contracting @xmath351 into a single vertex , a contradiction .
let @xmath177 be a component of @xmath351 .
then @xmath353 is a minimal separating set of @xmath16 . in particular
, @xmath354 is a clique , contrary to lemma [ dirac](ii ) .
depth2pt [ e : cnbr ] ( ) for any @xmath355 , @xmath356 .
suppose that there exists a vertex @xmath355 such that @xmath357 .
since @xmath347 , @xmath340 contains a clique of size @xmath358|\ge t-2 $ ] and so @xmath343 $ ] has a @xmath348-subgraph , contrary to .
depth2pt by lemma [ k_tk1 ] , @xmath359 .
let @xmath360 be such that @xmath361 .
clearly , @xmath218 is not adjacent to every vertex in @xmath362 , otherwise @xmath363>k_t$ ] , a contradiction .
let @xmath364 $ ] , where @xmath365|$ ] .
then @xmath218 is not adjacent to @xmath366 . by and ,
@xmath289\cap n(x)$ ] is not a clique .
let @xmath367 be a missing edge in @xmath299 .
by lemma [ wonderful ] applied to @xmath340 with @xmath368 , @xmath369 and @xmath370 , there exists a path @xmath128 with ends @xmath218 and @xmath371 and all its internal vertices in @xmath64 $ ] for each @xmath372 .
note that the paths @xmath373 have @xmath218 as a common end . by contracting all @xmath374 onto @xmath218
, we see that @xmath375 , a contradiction .
the primary purpose of this section is to prove theorem [ maink8- ] which states that every graph with no @xmath9 minor is @xmath10-colorable .
we need the following results .
theorem [ thm : k567-extremal ] was proved by dirac @xcite for the cases @xmath376 , and by jakobsen @xcite for the case @xmath377 .
[ thm : k567-extremal](dirac @xcite , jakobsen @xcite ) for @xmath378 , if @xmath16 is a graph with @xmath379 vertices and at least @xmath380 edges , then @xmath381 , or @xmath16 is a @xmath382-cockade when @xmath383 , or @xmath377 and @xmath16 is a @xmath384-cockade .
jakobsen @xcite also conjectured that theorem [ thm : k567-extremal ] extends to @xmath385 , which was confirmed by the second author @xcite : [ thm : k8-extremal](song @xcite ) if @xmath16 is a graph with @xmath188 vertices and at least @xmath386 edges , then @xmath109 or @xmath16 is a @xmath387-cockade .
the extremal function for @xmath388 minors remains open for @xmath389 .
the following lemma from @xcite will also be needed .
[ lem : delta5nbrhd](song @xcite ) let @xmath16 be a graph with @xmath390 and @xmath391 .
then either @xmath392 or @xmath16 is isomorphic to one of @xmath393 , @xmath394 , @xmath395 , @xmath396 , @xmath397 , or @xmath237 , where @xmath237 is the graph depicted in figure [ fig : j ] .
in particular , all of these graphs are edge maximal ( subject to not having a @xmath398-minor ) with maximum degree @xmath399 .
moreover , @xmath400 , @xmath401 , and @xmath402 .
.,width=150 ] notice that of the counterexamples listed in lemma [ lem : delta5nbrhd ] , only the graph @xmath237 has ten vertices , and none has exactly nine vertices .
we first prove the following lemma .
[ lem:10vertexalpha2 ] let @xmath16 be a graph with @xmath403 vertices and @xmath404 .
then either @xmath392 , or @xmath16 contains a @xmath405 subgraph , or @xmath16 is isomorphic to the graph @xmath237 depicted in figure [ fig : j ] . if @xmath406 , then by lemma [ lem : delta5nbrhd ] , either @xmath392 or @xmath16 is isomorphic to @xmath237 .
so we may assume that @xmath407 .
let @xmath408 be such that @xmath409 .
since @xmath404 , one can easily see that @xmath410 if @xmath411 .
hence we may further assume that @xmath412 .
then @xmath64 $ ] must be a @xmath413 as @xmath404 .
if @xmath343 $ ] also induces a @xmath413 , then @xmath16 contains a @xmath405 subgraph .
otherwise , some edge is missing from @xmath340 , say @xmath414 with @xmath415
. then since @xmath404 , each vertex in @xmath64 $ ] must be adjacent to either @xmath218 or @xmath219 .
thus @xmath416 , as desired .
this completes the proof of lemma [ lem:10vertexalpha2 ] .
depth2pt we are now ready to prove theorem [ maink8- ] .
let @xmath16 be a graph with no @xmath9 minor .
suppose for a contradiction that @xmath417 .
we may choose such a graph @xmath16 so that it is @xmath32-contraction - critical
. let @xmath336 be of minimum degree .
since @xmath16 is @xmath32-contraction - critical and has no @xmath9 minor , by lemma [ dirac](i ) applied to @xmath340 , we see that @xmath342 and @xmath418 .
on the other hand , since every @xmath419-cockade is @xmath5-colorable , by theorem [ thm : k8-extremal ] we see that @xmath420 .
thus @xmath421 . by lemma [ dirac](i ) applied to @xmath340 , we have [ e : alpha ] ( ) @xmath422 . we next show that [ e : noj ] ( ) @xmath340 is not isomorphic to the graph @xmath237 .
suppose that @xmath340 is isomorphic to the graph @xmath237 .
let the vertices of @xmath237 be labeled as depicted in figure [ fig : j ] . by lemma
[ wonderful ] applied to @xmath237 with @xmath423 and @xmath424 with @xmath425 , @xmath426 , there exist paths @xmath427 , @xmath428 , @xmath429 , @xmath430 , @xmath431 such that the paths @xmath427 , @xmath428 , @xmath429 , @xmath430 , @xmath431 have ends @xmath432 , @xmath433 , @xmath434 , @xmath435 , and @xmath436 , respectively , and all their internal vertices in @xmath437 $ ] .
moreover , the paths @xmath427 , @xmath428 , @xmath429 , @xmath430 are vertex - disjoint from the path @xmath431 . by contracting @xmath438 onto @xmath439 , @xmath440 onto @xmath441 , and @xmath442 $ ] into a single vertex , we see that @xmath443 , a contradiction .
depth2pt [ e : k5uk5 ] ( ) @xmath340 contains @xmath405 as a subgraph .
suppose that @xmath340 does not contain @xmath405 as a subgraph .
then by , and lemma [ lem:10vertexalpha2 ] , we see that @xmath444 .
let @xmath445 be a vertex such that @xmath446 . clearly , @xmath218 is not adjacent to every vertex in @xmath362 , otherwise @xmath363>k_8 ^ -$ ] , a contradiction .
let @xmath364 $ ] , where @xmath447|\ge1 $ ] .
then @xmath218 is not adjacent to @xmath366 .
clearly , @xmath448 is not a clique for all @xmath372 . by lemma [ wonderful ]
applied @xmath196 times to @xmath340 with @xmath449 , @xmath450 and @xmath98 ( where @xmath451 for @xmath452 ) , there exists a path @xmath128 between @xmath218 and @xmath371 with its internal vertices in @xmath64 $ ] for each @xmath372 .
note that the paths @xmath453 have @xmath218 as a common end . by contracting each @xmath374 onto @xmath218
, we see that @xmath454 , a contradiction .
depth2pt by , @xmath68 belongs to two different @xmath108-subgraphs of @xmath16 . by theorem [ 7con ]
, @xmath16 is @xmath5-connected . by lemma [ lem:7conn2k6 ] , @xmath109 .
this contradiction completes the proof of theorem [ maink8- ] .
we prove theorem [ maink8= ] in this section . the following result will be needed .
theorem [ thm : k8=extremal ] for the cases @xmath455 is due to dirac @xcite , and theorem [ thm : k8=extremal ] for the cases @xmath456 is due to jakobsen @xcite .
[ thm : k8=extremal](dirac @xcite , jakobsen @xcite ) for integer @xmath196 with @xmath457 , every graph with @xmath379 vertices and at least @xmath458 edges either contains a @xmath459-minor or is a @xmath460-cockade .
we are ready to prove theorem [ maink8= ] .
suppose the assertion is false .
let @xmath16 be a graph with no @xmath11 minor such that @xmath461 .
we may choose such a graph @xmath16 so that it is @xmath10-contraction - critical
. let @xmath336 be of minimum degree .
since @xmath16 is @xmath10-contraction - critical and has no @xmath11 minor , by lemma [ dirac](i ) applied to @xmath340 , we see that @xmath342 and @xmath462 . on the other hand , since each @xmath463-cockade is @xmath464-colorable , it follows from theorem [ thm : k8=extremal ] for @xmath186 that @xmath465 .
thus @xmath466 , and so @xmath467 .
it follows from theorem [ thm : k8=extremal ] for @xmath186 again that [ e : k8=:n9 ] ( ) @xmath16 contains at least 28 vertices of degree 9 .
since @xmath16 has no @xmath11 minor , by lemma [ dirac](i ) applied to @xmath340 , [ e : k8=:alpha ] ( ) @xmath422 .
we next show that [ e : k8=:deltanbr4 ] ( ) @xmath340 contains @xmath413 as a subgraph .
suppose that @xmath340 does not contain @xmath413 as a subgraph .
then @xmath468 and by , @xmath469 .
we claim that @xmath470 .
suppose that @xmath471 .
by lemma [ lem : delta5nbrhd ] applied to @xmath340 , we see that @xmath472 .
let @xmath445 be such that @xmath473 .
clearly @xmath218 has at least two non - neighbors in @xmath474 , otherwise @xmath343>k_8^=$ ] , a contradiction .
let @xmath475 be all non - neighbors of @xmath218 , where @xmath476|\ge2 $ ] .
since @xmath468 , @xmath477 must have a missing edge , say @xmath44 . by lemma [ wonderful ]
applied to @xmath340 with @xmath478 and @xmath479 , there exist @xmath480 paths @xmath481 such that each path @xmath128 has ends @xmath482 and all its internal vertices in @xmath437 $ ] . by contracting all the edges of each @xmath483 onto @xmath218 for all @xmath484 , we see that @xmath485 , a contradiction .
this proves that @xmath470 , as claimed .
let @xmath445 be such that @xmath218 has degree four in @xmath340 with @xmath486 maximum .
let @xmath487 be the set of all neighbors of @xmath218 in @xmath340 . since @xmath468 , @xmath289 \cap n(x)$ ] is not complete .
we may assume that @xmath488 .
by , @xmath489 $ ] induce a @xmath325 subgraph .
let @xmath490 $ ] .
we next show that ( @xmath491 ) each of @xmath492 has at most one neighbor in @xmath493 .
suppose , say @xmath494 , is adjacent to at least two vertices in @xmath493 .
then the subgraph induced on @xmath495 has a @xmath496 minor and thus @xmath343>k_8^=$ ] if @xmath497 is adjacent to all vertices in @xmath493 ( by contracting the path @xmath498 into a single vertex ) , a contradiction .
thus we may assume that @xmath497 is not adjacent to @xmath499 , where @xmath500 .
by lemma [ wonderful ] applied to @xmath340 with @xmath501 and @xmath502 , there exist @xmath227 paths @xmath503 such that for each @xmath504 , the path @xmath505 has ends @xmath506 and all its internal vertices in @xmath437 $ ] . by contracting all @xmath507 onto @xmath497 and the path @xmath508 into a single vertex , we see that @xmath509 , a contradiction .
this proves ( @xmath491 ) .
depth2pt we next claim that @xmath289\cap n(x)=k_5 ^ -$ ] .
suppose @xmath510 . by symmetry
, we may apply ( @xmath491 ) to the missing edge @xmath511 in @xmath340 , and so we see that each of @xmath512 has at most one neighbor in @xmath493 .
hence @xmath513 . on the other hand , since @xmath347
, each @xmath514 must be adjacent to at least one of the vertices in @xmath515 and @xmath516 , respectively , for all @xmath517 .
thus @xmath518 , a contradiction .
this proves that @xmath519 and thus @xmath340 does not have two independent missing edges .
next if @xmath520 , then @xmath521 and @xmath289\cap n(x)=k_5^=$ ] . since @xmath468
, we may assume that @xmath522
. then @xmath523 must be adjacent to both @xmath524 and @xmath497 by . by applying ( @xmath491 ) to the missing edges @xmath525 and @xmath526 , we see that @xmath527 is anti - complete to @xmath528 and @xmath494 has at most one neighbor in @xmath493 . by ( 2 ) , @xmath529 is complete to @xmath530 . since @xmath494 has at most one neighbor in @xmath493 , we may assume that @xmath531
. now @xmath532 has degree four in @xmath340 with @xmath533\cap n(x)=k_5 ^ -$ ] , contrary to the choice of @xmath218 .
n(x)=k_5 ^ -$ ] with @xmath534 the only missing edge , as claimed . since @xmath535 ,
each of @xmath529 and @xmath524 has at least one neighbor in @xmath493 . by
, each of @xmath536 is adjacent to at least one of @xmath537 , and so either @xmath529 or @xmath524 has at least two neighbors in @xmath493 . by symmetry , we may assume that @xmath529 has more neighbors in @xmath493 than @xmath524 .
on the other hand , each vertex in @xmath538 has at least one non - neighbor in @xmath493 as @xmath539 .
thus , @xmath529 has either two or three neighbors in @xmath493 .
we consider the following two cases
. has exactly two neighbors in @xmath493.,width=225 ] first , assume that @xmath529 has exactly two neighbors in @xmath493 , say @xmath540 .
then @xmath524 must have exactly two neighbors @xmath541 in @xmath493 . by ( @xmath491 ) , each of @xmath492 has at most one neighbor in @xmath493 .
we may assume that @xmath494 is not adjacent to @xmath541 , and @xmath542 .
by lemma [ wonderful ] applied twice to @xmath340 with @xmath501 and @xmath543 , there exist three vertex - disjoint paths @xmath544 and two vertex - disjoint paths @xmath545 such that the paths @xmath307 , @xmath308 , @xmath546 , @xmath547 , and @xmath548 have ends @xmath549 , @xmath550 , @xmath551 , @xmath552 and @xmath553 , respectively , and all their internal vertices in @xmath437 $ ] , as depicted in figure [ fig : case1 ] . notice that each @xmath128 is vertex - disjoint from @xmath554 for all @xmath555 and @xmath556 , @xmath307 and @xmath308 are vertex - disjoint but @xmath546 and @xmath308 are not necessarily vertex - disjoint .
if @xmath546 and @xmath308 have only @xmath557 in common , then contracting @xmath558 onto @xmath218 , @xmath559 onto @xmath497 , and @xmath560 onto @xmath494 , and @xmath561 into two distinct vertices yields a @xmath11 minor in @xmath16 , a contradiction .
thus @xmath546 and @xmath308 must have an internal vertex in common .
let @xmath74 be the first vertex on @xmath546 ( when @xmath546 is read from @xmath218 to @xmath557 ) that is also on @xmath308 .
then @xmath562 . let @xmath563 be the subpath of @xmath546 from @xmath218 to @xmath74 , @xmath564 be the subpath of @xmath308 from @xmath74 to @xmath557 .
notice that @xmath565 is vertex - disjoint from @xmath308 but not necessarily internally disjoint from @xmath307 .
now contracting @xmath566 , @xmath567 onto @xmath218 , @xmath568 onto @xmath557 , @xmath569 onto @xmath497 , @xmath560 onto @xmath494 , and @xmath561 into two distinct vertices yields another @xmath11 minor in @xmath16 , a contradiction .
has exactly three neighbors in @xmath493.,width=225 ] it remains to consider the case when @xmath529 has exactly three neighbors , say @xmath570 in @xmath493 . then @xmath524 is adjacent to @xmath532 . by ( @xmath491 )
, we may assume that @xmath523 is not adjacent to @xmath492 , and @xmath571 .
by lemma [ wonderful ] applied twice to @xmath340 with @xmath501 and @xmath572 , there exist vertex - disjoint paths @xmath573 and another path @xmath574 such that the paths @xmath575 have ends @xmath551 , @xmath576 , @xmath577 , and @xmath578 , respectively , and all their internal vertices in @xmath437 $ ] , as depicted in figure [ fig : case2 ] .
notice that @xmath307 is vertex - disjoint from @xmath554 for @xmath579 , but that @xmath574 is not necessarily internally vertex - disjoint from either @xmath547 or @xmath548 .
if @xmath574 and @xmath548 have only @xmath494 in common , then we obtain a @xmath11 minor by contracting @xmath307 and @xmath580 into two distinct vertices , @xmath581 , @xmath582 onto @xmath523 , and @xmath583 onto @xmath584 , a contradiction .
thus @xmath574 and @xmath548 must have an internal vertex in common .
let @xmath74 be the first vertex on @xmath574 ( when @xmath574 is read from @xmath523 to @xmath494 ) that is also on @xmath548 .
then @xmath585 . let @xmath586 be the subpath of @xmath574 from @xmath523 to @xmath74 , @xmath587 be the subpath of @xmath548 from @xmath74 to @xmath494 .
notice that @xmath588 is vertex - disjoint from @xmath548 but not necessarily internally disjoint from @xmath547 .
now we obtain another @xmath11 minor by contracting @xmath307 and @xmath580 into two distinct vertices , @xmath581 , @xmath589 onto @xmath523 , @xmath590 onto @xmath494 , and @xmath591 onto @xmath584 , a contradiction .
this completes the proof of .
depth2pt by , every vertex of degree 9 belongs to some @xmath108-subgraph of @xmath16 . by
, @xmath16 contains at least five different @xmath108-subgraphs . by theorem [ 7con ] ,
@xmath16 is @xmath5-connected and thus @xmath485 by lemma [ lem:7conn2k6 ] .
this contradiction completes the proof of theorem [ maink8= ] .
it seems very difficult to prove that every graph with no @xmath3 minor is @xmath5-colorable .
we establish in @xcite the properties of @xmath12-contraction - critical graphs with no @xmath3 minor to shed some light on this open problem . as pointed out by robin thomas ( personal communication with the second author )
, one might be able to settle this open problem by using the key ideas in @xcite .
however , it seems very hard to prove that every @xmath12-contraction - critical graph is @xmath12-connected , which means the truth of the seymour - thomas conjecture ( i.e. conjecture [ conj2 ] ) can not be applied easily to solve hadwiger s conjecture .
theorem [ k7 ] is such a nice result .
we believe that mader s bound for the extremal function for @xmath13 minors is true as follows : by theorem [ k9 ] , conjecture [ conj3 ] is true for @xmath201 .
as mentioned earlier , lemma [ wonderful ] turns out to be very powerful .
we believe that the application of lemma [ wonderful ] that we have developed in this paper is of independent interest . to end this section ,
we apply lemma [ wonderful ] along with a new idea ( namely , considering the chromatic number of @xmath340 ) to prove that the truth of conjecture [ conj3 ] implies that every graph with no @xmath13 minor is @xmath594-colorable for all @xmath14 .
since conjecture [ conj3 ] is true for @xmath201 , we see that theorem [ final ] implies theorem [ main ] .
suppose the assertion is false . by theorem [ k9 ]
, we have @xmath595 . among all minimum counterexamples ,
we choose @xmath16 so that @xmath16 has no @xmath13 minor and @xmath16 is @xmath596-contraction - critical .
let @xmath408 be such that @xmath597 . by the assumed truth of conjecture [ conj3 ] , @xmath598 .
on the other hand , by lemma [ dirac](i ) applied to @xmath340 , we see that @xmath342 and @xmath599 .
hence @xmath600 . by lemma
[ dirac](i ) again , we have suppose that @xmath603 .
let @xmath604 $ ] be isomorphic to @xmath605 .
since @xmath606 , every vertex in @xmath15 is adjacent to @xmath607 vertices in @xmath351 .
then @xmath351 is disconnected , otherwise @xmath352 by contracting @xmath351 into a single vertex , a contradiction .
let @xmath177 be a component of @xmath351 .
then @xmath353 is a minimal separating set of @xmath16 , in particular , @xmath354 is a clique , contrary to lemma [ dirac](ii ) .
this proves that @xmath601 . by
, we see that @xmath602.height3pt width6pt depth2pt suppose to the contrary that @xmath609 . by
, it is clear that @xmath610 .
thus @xmath611 for some @xmath612 .
let @xmath613 be the color classes of any proper @xmath1-coloring of @xmath340 .
we may assume that the color classes are ordered so that @xmath614 for @xmath615 and @xmath616 for @xmath617 . let @xmath618 .
notice that @xmath619 induces a clique in @xmath340 , and so @xmath620 by . by
, @xmath60 is adjacent to either @xmath621 or @xmath622 for each @xmath623 and each @xmath624 .
notice that if @xmath625 , then @xmath626 .
suppose that @xmath627 or that @xmath625 and @xmath628 have a common neighbor in @xmath629 , say @xmath630 .
by lemma [ wonderful ] applied to @xmath340 with @xmath631 and @xmath632 , there exist @xmath633 pairwise vertex - disjoint paths @xmath634 such that each @xmath505 has ends @xmath635 and all its internal vertices in @xmath636 $ ] . by contracting each @xmath505 to a single vertex for all @xmath637 , together with @xmath638 if @xmath639 ; and together with @xmath640 if @xmath625 where @xmath641 is a common neighbor of @xmath628 , we obtain a clique minor on @xmath642 vertices in the former case and @xmath643 vertices in the latter case , a contradiction .
thus @xmath625 and @xmath628 have no common neighbor in @xmath629 .
since each of @xmath628 is adjacent to either @xmath645 or @xmath646 , by symmetry , we may assume that @xmath645 is adjacent to @xmath101 and @xmath647 , but not to @xmath648 .
then @xmath646 is adjacent to @xmath648 .
we may assume that @xmath646 is not adjacent to @xmath101 .
for the worst case scenario , we may further assume that @xmath646 is not adjacent to @xmath647 .
by lemma [ wonderful ] applied twice to @xmath340 with @xmath649 and @xmath650 , there exist pairwise vertex - disjoint paths @xmath651 such that each @xmath505 has ends @xmath635 and all its internal vertices in @xmath636 $ ] ; and paths @xmath652 with ends @xmath653 , @xmath654 , @xmath655 and @xmath656 , respectively , and all their internal vertices in @xmath636 $ ] .
notice that each @xmath505 is vertex - disjoint from @xmath652 , and @xmath132 is vertex - disjoint from @xmath657 but not necessarily from @xmath658 .
let @xmath74 be the first vertex on @xmath132 ( when read from @xmath648 to @xmath645 ) that is also on @xmath658 .
note that @xmath74 could be @xmath645 .
let @xmath659 be the subpath of @xmath132 between @xmath74 and @xmath648 , as depicted in figure [ fig : finala1a2a3 ] . by contracting each @xmath505 to a single vertex for all @xmath660 , @xmath661 and @xmath662 onto @xmath646 , @xmath663 onto @xmath648 , and @xmath664 onto @xmath645 , together with the vertices @xmath665
, we obtain a @xmath13 minor , a contradiction .
depth2pt by and , we have @xmath667 .
suppose there exists a vertex @xmath360 such that @xmath218 has exactly @xmath668 neighbors in @xmath340 . then @xmath489 $ ] is a clique of @xmath340 with @xmath668 vertices .
furthermore , by , @xmath669 must have some missing edge , say @xmath44 .
then every vertex of @xmath489 $ ] is adjacent to at least one of @xmath133 and @xmath135 .
thus , by contracting the path @xmath670 to a single vertex , we can see that @xmath671 .
now we can assume without loss of generality that @xmath445 is such that @xmath672 .
clearly @xmath218 is not adjacent to every vertex in @xmath673 , or else @xmath674 > k_p$ ] , a contradiction .
let @xmath675 $ ] , where @xmath0 .
again , by , @xmath299 must have some missing edge , say @xmath44 . by lemma [ wonderful ] applied to @xmath340 with @xmath676 and @xmath677 , there exist paths @xmath678 such that each @xmath128 has ends @xmath482 and all internal vertices in @xmath636 $ ] . now by contracting each @xmath679 onto @xmath218 , we see that @xmath680 , a contradiction .
depth2pt + by and , @xmath340 does not contain @xmath681 as a subgraph and @xmath666 . by , @xmath682 .
let @xmath683 be the color classes of any proper @xmath684-coloring of @xmath340 .
we may assume that the color classes are ordered so that @xmath685 and @xmath616 for @xmath686 . by
, @xmath101 must be complete to some color class @xmath687 , say @xmath688 . by again
, @xmath647 and @xmath689 must have one common neighbor in some color class @xmath690 , say @xmath691 .
we may further assume that @xmath648 is adjacent to both @xmath692 .
by symmetry , we may assume that @xmath693 is adjacent to @xmath647 .
by lemma [ wonderful ] applied up to @xmath340 with @xmath694 and @xmath695 , there exist paths @xmath696 and pairwise vertex - disjoint paths @xmath697 such that @xmath696 have ends @xmath698 and @xmath699 , respectively ; each @xmath554 has ends @xmath635 and all such paths have their internal vertices in @xmath636 $ ] . by contracting @xmath700 onto @xmath693 , the edge @xmath701 onto @xmath648 , and each @xmath554 into a single vertex for @xmath702
, we see that @xmath680 , a contradiction .
b. albar , d. gonalves , on triangles in @xmath703-minor free graphs , arxiv:1304.5468 .
g. chen , z. hu , f. song , a new connectivity bound for linkages and its application to the hadwiger s conjecture , submitted .
g. a. dirac , a property of @xmath464-chromatic graphs and some remarks on critical graphs , j. london math .
27 ( 1952 ) 8592 .
g. a. dirac , homomorphism theorems for graphs , math .
( 1964 ) 6980 .
g. a. dirac , trennende knotenpunktmengen und reduzibilitt abstrakter graphen mit anwendung auf das vierfarbenproblem , j. reine agew .
204 ( 1960 ) 116131 .
h. hadwiger , ber eine klassifikation der strechenkomplexe , vierteljschr .
ges zrich 88 ( 1943 ) 133142 .
i. t. jakobsen , a homomorphism theorem with an application to the conjecture of hadwiger , studia sci .
( 1971 ) 151160 .
i. t. jakobsen , on certain homomorphism properties of graphs i , math .
31 ( 1972 ) 379404 .
i. t. jakobsen , on certain homomorphism properties of graphs ii , math .
52 ( 1983 ) 229261 .
l. k. jrgensen , contractions to @xmath31 , j. graph theory 18 ( 1994 ) 431448 .
k. kawarabayashi , b. toft , any @xmath5-chromatic graph has a @xmath3 or @xmath704 as a minor , combinatorica 25 ( 2005 ) 327353 .
k. kawarabayashi , on the connectivity of minimum and minimal counterexamples to hadwiger s conjecture , j. combin .
theory ser .
b 97 ( 2007 ) 144150 .
k. kawarabayashi , g. yu , connectivities for @xmath38-knitted graphs and for minimal counterexamples to hadwiger s conjecture , j. combin .
theory ser .
b 103 ( 2013 ) 320326
. w. mader , homomorphiestze fr graphen , math .
( 1968 ) 154168 .
ber trennende eckenmengen in homomorphiekritischen graphen .
( 1968 ) 243252 .
n. robertson , p. seymour and r. thomas , hadwiger s conjecture for @xmath108-free graphs , _
combinatorica * 13 * _ ( 1993 ) 279361 .
m. rolek , z - x .
song , properties of @xmath12-contraction - critical graphs with no @xmath3 minor , in preparation .
p. seymour , hadwiger s conjecture , manuscript .
song , the extremal function for @xmath9 minors , j. combin .
theory ser .
b 95 ( 2005 ) 300317 .
z - x . song , r. thomas , the extremal function for @xmath193 minors .
j. combin .
theory , ser .
b , 96 ( 2006 ) 240252 .
b. toft , _ a survey on hadwiger s conjecture _ , in : _ surveys in graph theory _ ( edited by g. chartrand and m. jacobson ) , congr .
115 ( 1996 ) , 249283 .
k. wagner , ber eine eigenschaft der ebenen komplexe , math .
( 1937 ) 570590 . | hadwiger s conjecture from 1943 states that for every integer @xmath0 , every graph either can be @xmath1-colored or has a subgraph that can be contracted to the complete graph on @xmath2 vertices .
as pointed out by paul seymour in his recent survey on hadwiger s conjecture , proving that graphs with no @xmath3 minor are @xmath4-colorable is the first case of hadwiger s conjecture that is still open .
it is not known yet whether graphs with no @xmath3 minor are @xmath5-colorable . using a kempe - chain argument along with the fact that an induced path on three vertices is dominating in a graph with independence number two
, we first give a very short and computer - free proof of a recent result of albar and gonalves and generalize it to the next step by showing that every graph with no @xmath6 minor is @xmath7-colorable , where @xmath8 .
we then prove that graphs with no @xmath9 minor are @xmath10-colorable and graphs with no @xmath11 minor are @xmath12-colorable .
finally we prove that if mader s bound for the extremal function for @xmath13 minors is true , then every graph with no @xmath13 minor is @xmath7-colorable for all @xmath14 .
this implies our first result .
we believe that the kempe - chain method we have developed in this paper is of independent interest . |
the standard architecture ( connection graph ) for a multiprocessor computer is the @xmath4-dimensional cube with @xmath5 or @xmath6 ( so the number of processors is 64 or 128 ) .
recently , the `` sierpinski gasket pyramid graph '' has been proposed by william , rajasingh , rajan & shantakumari @xcite as an alternative ( to the @xmath4-cube ) . after determining the elementary graph - theoretic properties of the sierpinski gasket pyramid graph , w - r - r&s proposed studying its `` message routing and broadcasting '' properties .
this is the first of several papers following up on that suggestion .
our goal with this paper has been to learn more about the structure of sierpinski graphs .
our goal for the next paper is to solve the edge - isoperimetric problem ( @xmath7 ) on the ( generalized & expanded ) sierpinski graph , @xmath0 .
@xmath0 has the same vertex set as @xmath1 , the hamming graph ( see below for definitions ) , but fewer edges , so it was natural to ask if @xmath0 could be embedded in @xmath1 .
if so , it would provide a context for @xmath0 which could lead to insight .
the structure of @xmath1 is relatively transparent , its @xmath7 has been solved ( see @xcite , p. 112 ) and it is closely related to the @xmath4-cube ( the sierpinski gasket pyramid graph s competition for multiprocessor architecture ) .
embedding @xmath0 into @xmath1 turned out to be an interesting and challenging problem in its own right and its solution has unexpected ramifications .
an _ ordinary _ _ graph , _
@xmath8 consists of a set @xmath9 , of _ vertices _ and a set @xmath10 , of pairs of vertices called _
edges_. @xmath11 , _ the complete graph on _
@xmath12 vertices has @xmath13 and @xmath14 .
the ( disjunctive ) product , @xmath15 is called the hamming graph .
two vertices ( @xmath4-tuples of vertices of @xmath16 ) have an edge between them if they differ in exactly one coordinate ( _ i.e. _ are at hamming distance @xmath17 ) .
@xmath0 , @xmath18 , @xmath19 was defined in 1944 by scorer , grundy and smith @xcite : @xmath20 . for @xmath21 , @xmath22 iff @xmath23
such that following 3 conditions hold : 1 .
@xmath24 for @xmath25 ; 2 .
@xmath26 ; and 3 .
@xmath27 and @xmath28 for @xmath29 .
the motivation for defining @xmath0 was that @xmath30 is the graph of the @xmath31-peg tower of hanoi puzzle with @xmath4 disks @xcite .
scorer , grundy and smith also pointed out that the graph defined by topologist wacsav sierpinski in 1915 , known as the sierpinski gasket graph , is a contraction of @xmath30 where every edge of @xmath30 not contained in a triangle ( @xmath32 ) is contracted to a vertex . the sierpinski gasket pyramid ( aka the sierpinski sponge ) , is a @xmath31-dimensional analog of the sierpinski gasket , and its graph is a similar contraction of @xmath33 .
jakovac @xcite generalized the construction to @xmath34 $ ] , the contraction of @xmath0 in which every edge of @xmath0 not contained in a @xmath32 is contracted to a vertex .
he worked out some properties of @xmath34 $ ] such as hamiltonicity and chromatic number ( @xmath35\right ) = m$ ] ) .
it seems apparent that the nomenclature of these structures is a muddle .
first came the sierpinski gasket , a topological curiosity , and then the family of graphs ( @xmath36 $ ] ) that comprise the boundary of the `` gasket '' at the @xmath37 stage of its construction .
the `` gasket '' , a 2-dimensional structure , was generalized to @xmath12 dimensions by replacing the equilateral triangle that sierpinski started with , by an @xmath12-simplex .
if @xmath38 $ ] is the sierpinski ( gasket ) graph , then @xmath34 $ ] should be called the `` generalized sierpinski graph '' .
scorer , grundy & smith noted that @xmath36 $ ] could be obtained from the tower of hanoi graph , @xmath39 , by contracting certain edges to a single vertex .
conversely , @xmath40 is derived from @xmath36 $ ] by expanding certain vertices into an edge ( and two vertices ) and @xmath0 is a generalization of that construction . for many purposes
@xmath0 is easier to deal with ( than @xmath34 $ ] ) because its vertices and edges are easily defined .
so we propose calling @xmath0 the `` generalized & expanded sierpinski graph '' or `` sierpinski graph '' for short ( when the meaning is clear from context ) .
all @xmath42 have @xmath43 `` interior '' neighbors .
these are the @xmath4-tuples that agree with @xmath44 in all coordinates except the @xmath37 ( the case @xmath45 in the definition of edges in @xmath0 ) .
if @xmath46 then @xmath44 has one other ( `` exterior '' ) neighbor : if @xmath47 , then @xmath48 , such that @xmath49 and by definition the exterior neighbor of @xmath44 is @xmath50 ( note that this relationship between @xmath51 and @xmath44 is symmetric ) .
thus @xmath52 , with @xmath53 , has degree @xmath43 and every other vertex has degree @xmath12 . summing the degrees of all vertices we get @xmath54 @xmath55 . since each edge is incident to two vertices , @xmath56 .
the vertices that agree in all except the last coordinate induce a complete subgraph , @xmath11 .
there are @xmath57 such @xmath11@xmath58 , nonoverlapping and containing all the vertices of @xmath0 .
this constitutes a @xmath59-decomposition of @xmath0 .
since any vertex is incident to at most one exterior edge , any triangle ( @xmath32 ) must contain at least two internal edges .
but then the third edge would also be internal to the same @xmath11 , so our @xmath11-decomposition is unique .
the vertices @xmath52 , for @xmath53 , are called _ corner _ vertices of @xmath0 .
@xmath60 is just a path of length @xmath61 .
its endpoints are the two corner vertices , @xmath62 & @xmath63 .
every other vertex is incident to two edges , one coming from ( the direction of ) @xmath62 and the other going toward @xmath63 . that it is a single path ( and not a path plus disjoint cycles )
is shown by the observation that the two vertices , @xmath64 , connected by the edge @xmath65 are binary representations of two consecutive integers .
one of them , say @xmath51 , has @xmath66 and the other @xmath67 .
the last @xmath68 components of @xmath51 contribute @xmath69 to its binary representation whereas that of @xmath44 contributes @xmath70 .
figure 1 shows diagrams of @xmath60 for @xmath71 with coordinates .
@xmath72{2.8919in}{\begin{center } \includegraphics [ trim=2.037533 in 5.776840 in 1.600797 in 3.493177 in , height=1.0533 in , width=2.8919 in ] % { figure1d.eps}% \\ figure 1-$s(n,2)$ for $ n=1,2,3 $ \end{center}}}% % endexpansion\ ] ] figure 2 shows diagrams of @xmath60 for @xmath71 with coordinates .
@xmath73{2.9914in}{\begin{center } \includegraphics [ trim=1.200598 in 2.406925 in 1.367135 in 1.205113 in , height=3.7239 in , width=2.9914 in ] % { figure2d.eps}% \\ figure 2-$s(n,3)$ for $ n=1,2,3 $ \end{center}}}% % endexpansion\ ] ] ( theorem 4.14 of @xcite ) the symmetry group of @xmath0 is @xmath74 , the symmetric group on @xmath12 generators .
let @xmath75 be any permutation of @xmath76 , and let it act on the vertices of @xmath0 in the obvious way ,
@xmath77 if @xmath22 then @xmath78 and@xmath79 so @xmath80 and @xmath81 is a symmetry of @xmath0 .
conversely , suppose @xmath82 is a symmetry of @xmath0 .
any symmetry , @xmath83 , of @xmath0 must take corners to corners so @xmath84 for some @xmath85 .
we then define a permutation on @xmath86 by @xmath87 .
we claim that when this action is extended as above , we get @xmath82 back again : by induction on @xmath88 ( in the definition of external edges ) . for @xmath89 . for @xmath90 ,
if @xmath91 then since the subgraph @xmath92 contains @xmath52 , its image under @xmath82 must contain @xmath93 and be @xmath94 . therefore @xmath95 .
@xmath96 must be a corner vertex of @xmath97 which means that @xmath98 for some @xmath99 since @xmath100 and @xmath101 are neighbors , @xmath102 and @xmath103 are also neighbors , so @xmath104 and @xmath105 . the pattern for higher @xmath88 is the same , so @xmath106 and the symmetry induced by @xmath107 is @xmath82 .
the vertex - sets of @xmath0 and @xmath1 are the same , @xmath2 .
however the edge - sets are different : @xmath108because @xmath109 also @xmath110 since the exterior edges of @xmath0 differ in two or more coordinates .
the question of whether @xmath0 can be embedded in @xmath1 is a special case of the subgraph isomorphism problem , well - known for it complexity ( it is np - complete @xcite ) .
however , we now show that such an embedding exists .
first we consider a couple of special cases .
we showed in example 1 that @xmath60 is a path from @xmath62 to @xmath63 .
so the question of embedding of @xmath60 into @xmath112 , the graph of the @xmath4-cube , is the graph - theoretic classic , `` does @xmath113 contain a hamiltonian path ( a path passing through each vertex exactly once ) .
the answer is yes ( there are many , the gray code being one ( see wikipedia on ' ' gray code " ) ) .
define @xmath115 as the graph with @xmath116 and @xmath117 where @xmath118 .
note that @xmath119 and @xmath120 . also note that @xmath115 is a subgraph of @xmath1 ( since @xmath121 and the pairs of vertices that comprise an edge of @xmath115 differ in exactly one coordinate , making it an edge of @xmath1 ) .
however , this does not prove that @xmath0 is embeddable as a subgraph of @xmath1 .
@xmath122 . in the next section
we will prove a more general theorem .
it would seem natural to extend lemma 1 to show that @xmath123 .
after many attempts to do so , we ran up against the following counterexample : consider the vertex @xmath124 in @xmath125 .
it shares an edge with the vertices @xmath126 @xmath127 @xmath128 ( @xmath129 ) as well as the vertices @xmath130 ( @xmath131 ) and @xmath132 ( @xmath90 ) .
so @xmath124 has degree @xmath133 in @xmath125 , but the maximum degree of any vertex in @xmath134 is @xmath31 .
the sierpinski graph , @xmath0 , has been defined analytically in section 1.2.2 .
however , @xmath0 may also be characterized recursively : @xmath135 and given @xmath136 for @xmath137 , we construct @xmath0 by making @xmath12 copies , @xmath138 , @xmath139 , and throwing in all edges of the form @xmath140 such that @xmath141 and @xmath142 . note that every copy of @xmath136 is connected to every other copy and
the corners of @xmath0 are the constant vertices @xmath143 @xmath139 .
the scorer - grundy - smith definition of the ( generalized and extended ) sierpinski graph may be regarded as a representation or coordinatization of the abstract graph
. our question of whether @xmath0 may be embedded in @xmath144is equivalent to asking whether there is another definition of that abstract graph ; a definition in which the set of vertices , @xmath145 , is the same , but the pairs of vertices , @xmath146 that constitute an edge differ in exactly one coordinate .
an obvious candidate was @xmath115 . that worked for @xmath114 ( lemma 1 ) but failed for @xmath147 .
there exists another coordinatization of @xmath0 ( call it @xmath148 ) that is a subgraph of @xmath1 .
i.e. the edges of @xmath149 consist of pairs of vertices that differ in exactly one coordinate .
by induction on @xmath4 : it is trivial for @xmath150 ( @xmath151 ) .
assume that the theorem is true for @xmath137 .
implicit in the statement of the theorem is the existence of a function , @xmath152 that takes each vertex of @xmath0 to its coordinates in @xmath149 .
we write this as @xmath153 .
we have represented @xmath0 recursively as @xmath154 ( the disjoint union of @xmath12 copies of @xmath136 ) , along with the external edges , @xmath155 by the inductive hypothesis we have @xmath156 @xmath157 taking @xmath158 to its coordinates in @xmath157 . for each @xmath159
we define@xmath160 by @xmath161 the @xmath162 are only the corner vertices of @xmath163 , but as we showed in theorem 1 , any permutation of the underlying set , @xmath86 ( and here we are applying the permutation @xmath164 ) , extends to a unique symmetry of @xmath136 .
that extension constitutes @xmath165 .
now we define @xmath166 the disjoint union of the component maps , @xmath167 , @xmath168 .
we claim then that @xmath169 @xmath170 @xmath149 is the isomorphism we have been seeking : edges internal to @xmath138 are preserved because @xmath171 @xmath172 is the composition of two isomorphisms @xmath173 and @xmath174 . by induction on @xmath4 ,
the vertices incident to an edge internal to @xmath175 @xmath157 differ in exactly one coordinate .
the external edges , @xmath176 are preserved because@xmath177 thus @xmath178 and @xmath179 differ in exactly one coordinate ( the first ) .
figure 3 shows @xmath180.@xmath181{2.9265in}{\begin{center } \includegraphics [ trim=1.197199 in 2.483964 in 1.504783 in 1.206214 in , height=3.6841 in , width=2.9265 in ] % { figure3d.eps}% \\ figure 3-$\widetilde{\widetilde{s}}(n,3)$ for $ n=1,2,3 $ \end{center}}}% % endexpansion\ ] ] note that in the case @xmath114 , @xmath182 is the identity on @xmath183 so @xmath184 and our proof includes lemma 1 .
geometric intuition had indicated that to embed @xmath0 into @xmath1 we had to give the copies of @xmath136 a twist , which led to the definition of @xmath185 .
the counterexample showed that just one twist was not adequate for @xmath186 , that we had to give the copies of @xmath185 another twist , leading to the definition of @xmath187 , _ etc . _
theorem 2 gives the solution of a wirelength problem . in laying out a graph , @xmath188 , on a graph , @xmath189
, edges of @xmath188 must be assigned to paths with distinct endpoints .
since the shortest such path is a single edge , an embedding of @xmath188 into @xmath189 achieves the minimum wirelength ( the sum of the lengths of all paths assigned to edges of @xmath188 ) and bandwidth ( the maximum length of any path assigned to an edge of @xmath188 ) of any layout .
thus @xmath190 and @xmath191 . in a sequel to this paper
we hope to present a solution of the bandwidth problem for laying out @xmath0 on a path , @xmath192 , of length @xmath193 .
the problems of laying out @xmath0 on a product of paths , @xmath194 such that @xmath195 so as to minimize wirelength or bandwidth appears very challenging and will probably require new methods . 1 .
the counterexample for @xmath147 & @xmath200 shows the subtlety of the problem , and 2 .
the proof for @xmath114 gave the solutions @xmath201 , @xmath202 of the recurrence inherent in the proof of theorem 2 .
this suggested extending the formulas to @xmath203 as a function of @xmath204 .
the resulting formulas are in the next section . [
[ inverting - varphileft - v_1v_2 ... v_nright - leftwidetildewidetildev_1widetildewidetildev_2 ... widetildewidetildev_nright ] ] inverting @xmath223 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ when @xmath111 these formulas reduce to @xmath229 and @xmath230 the embedding of @xmath60 , a path of length @xmath61 , into @xmath112 , the graph of the @xmath4-cube , given by theorem 2 , turns out to be exactly the gray code mentioned in section 3.0.1 .
the result is trivial for @xmath150 , @xmath231 being the identity .
if it is the gray code for @xmath137 , then by the recurrence at the beginning of section 4.4 , @xmath232 will be the same for the first half of @xmath233 as it was for @xmath234 since @xmath235 .
for the second half , we add @xmath236 to each component except the first , thereby complimenting it and reversing the order .
that is the ( recursive ) definition of the gray code .
more than fifty years ago , we noticed the similarity between the natural ( base @xmath237 ) code that assigns to each @xmath4-tuple , @xmath44 , of @xmath238s and @xmath17s the integer , @xmath239 , it represents in base @xmath237 , _ i.e. _ @xmath240 and the gray code ( see wikipedia )
. the gray code is defined recursively as a circular order on the vertices of the @xmath4-cube .
the natural binary code can also be described recursively .
but can the gray code be represented by a formula , similar to that for @xmath239 ?
it seemed that some of the coefficients in any such formula,@xmath241 would have to be negative .
but then the formula would take negative , as well as positive values , which presents a problem .
we tried various ways to get around this difficulty but without success .
however , the desired formula drops right out of our example 3 and section 4.5 . in example 3
we observed that @xmath60 is a path , with the vertices appearing in their natural ( base 2 ) order . at the end of section 4.5
we noted that @xmath242 carried @xmath60 to the gray code . with the inverse of @xmath243
we have our formula@xmath244 the essence of this relationship between @xmath245 and @xmath246 is in the wikipedia article `` gray code '' but presented in pseudocode rather than an algebraic formula . in reviewing the literature on sierpinski graphs , we were chagrined to realize that a principle case of our title s question already had an answer : the ( generalized & extended ) sierpinski graphs originated in work on the tower of hanoi puzzle @xcite . a position of the tower of hanoi puzzle with @xmath4 discs may be represented by @xmath247 , @xmath248 meaning that disc @xmath249 is on peg @xmath250 .
the position is uniquely determined by the @xmath4-tuple since the discs on a peg must be stacked smaller on larger ( the radius is decreasing in @xmath249 ) . when disc @xmath249 is moved from peg @xmath251 to @xmath252 , all discs on @xmath251 & @xmath252
must be larger than disc @xmath249 .
thus the smaller discs must all be stacked on the third peg , @xmath253,@xmath252 . when such a move is made , only one coordinate of the representing @xmath4-tuple , @xmath44 , changes ( @xmath254 goes from @xmath251 to @xmath252 ) .
so what we have done here is reverse engineer the work of scorer , grundy & smith in that case .
however , even in that case there are interesting differences .
let @xmath255 be the tower of hanoi graph with @xmath31 pegs and @xmath4 discs .
as described in the previous paragraph , vertices represent positions and edges represent the moving of a disc from one peg to another .
scorer , grundy & smith showed that @xmath255 and @xmath30 are different coordinatizations of the same abstract graph . therefore @xmath255 and @xmath256 are isomorphic and both are subgraphs of @xmath257 .
so , is @xmath258 @xmath256 ?
figure 4 shows @xmath255 , @xmath71 .
@xmath259{2.9974in}{\begin{center } \includegraphics [ trim=1.209095 in 2.146092 in 1.347592 in 0.949783 in , height=3.9825 in , width=2.9974 in ] % { figure4d.eps}% \\ figure 4-$t(n)$ for $ n=1,2,3 $ \end{center}}}% % endexpansion\ ] ] a glance at diagrams of @xmath256 & @xmath255 shows they are not equal . already for @xmath114
some coordinates are different .
is it possible to change our definition of @xmath260 a bit so that it takes @xmath30 to @xmath255 ?
@xmath261 is defined recursively by the following slight variant of @xmath262.@xmath263 the only difference is the factor of @xmath237 multiplying @xmath264 .
the diagram for @xmath265 differs from that for @xmath266 by an interchange @xmath267 in the second coordinate .
this is due to the fact that when the large disc is moved from peg @xmath238 , the small disc must be on peg @xmath17 or peg @xmath237 and then the large disc must go on the other peg ( @xmath237 or @xmath17 ) . since @xmath268 , @xmath269 and @xmath270 , multiplying @xmath271 by @xmath237 @xmath272 exactly represents this move .
again , the proofs of these theorems are essentially the same as for @xmath260 except we need the fact that @xmath283 for theorem 6 .
@xmath284 is actually the original scorer - grundy - smith transformation .
the tower of hanoi is a fundamental example of a recursive algorithm .
as such it is assigned as an exercise in programming courses .
the theorems above provide a `` hack '' ( illicit shortcut ) for that exercise : in @xmath40 the shortest path from @xmath62 to @xmath63 is @xmath285 , whose vertices appear in lexicographic order .
so @xmath286 gives the solution of the classical tower of hanoi puzzle , moving discs from @xmath62 to @xmath287 in @xmath288 steps .
more specifically , the @xmath289 coordinate of the @xmath290 position is@xmath291 where @xmath292 ( see , dad , algebra can solve the tower of hanoi puzzle , and evidently the chinese rings too ! ) .
clifford wolfe came up with a similar formula for @xmath293 ( http://www.clifford.at/hanoi/ ) by analysing the moves of individual discs:@xmath294 [ [ generalization - of - the - embeddings - varphileft - nright - tauleft - nright ] ] generalization of the embeddings @xmath295 , @xmath296 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ note that in each of these examples @xmath312 .
we may always assume , by way of normalizing to eliminate redundancy , that @xmath320 .
if @xmath321 , the symmetry induced by @xmath322 will put it in that form .
thus for each @xmath307 we can take @xmath323 to be @xmath324 , where gcd@xmath325 .
there are @xmath326 ways ( @xmath327 is the euler totient function ) to choose @xmath328 , so the number of nonisomorphic embeddings @xmath329 we have constructed is @xmath330 . in the words of hardy & wright ,
the order of @xmath331 is `` always ` nearly @xmath4 ' '' ( see wikepedia , `` euler s totient function ( growth of the function ) '' ) so for any @xmath332 , for @xmath12 suficiently large there are at least @xmath333 such embeddings .
one of the striking features of @xmath0 @xmath127 @xmath255 is that corners ( vertices of degree @xmath43 ) are constant @xmath334 . in @xmath256
there is only one constant corner , @xmath62 , the two other corners alternating @xmath238 & @xmath17 .
also , our mapping @xmath261 actually preserves the coordinates of corners ( @xmath335 ) .
can @xmath336 be generalized to @xmath337 , a subgraph of @xmath1 , isomorphic to @xmath0 and with constant corners ?
the recursive definition of @xmath338 that characterizes @xmath255 as its range , may be extended to @xmath339 for any odd @xmath12 by letting @xmath340 in the previous section .
then@xmath341 since multiplication by @xmath237 is invertible @xmath342 and @xmath343 ( note that @xmath344 , so it made no difference that we used @xmath237 instead of @xmath345 in section 4.8 , since @xmath200 there)@xmath311 by induction , the @xmath337 so defined has constant corners since @xmath346 .
however , the definition does not work for even @xmath12 because @xmath237 does not have a multiplicative inverse @xmath347 .
@xmath348 trivially has constant corners , but we can prove that there is no subgraph of @xmath349 isomorphic to @xmath350 with constant corners : ( by contradiction ) assume that @xmath351 is such a subgraph with corners @xmath352 .
the @xmath353 that constitute the @xmath354-covering of @xmath351 must lie on parallel lines ( rows or columns ) of @xmath349 . and each of those @xmath353 contain exactly one corner .
if the @xmath353 are rows , then the exterior edges must lie in the columns ( or _ vice versa _ ) .
but with one of the vertices in each column being a corner ( and not incident to an exterior edge ) the @xmath31 remaining vertices can only accommodate @xmath17 edge .
four columns then yield at most @xmath133 exterior edges but @xmath355 has @xmath356 exterior edges , a contradiction .
the same argument shows that @xmath357 does not exist for any even @xmath358 .
the question of @xmath337 for even @xmath358 and @xmath359 remains open .
also open is the question as to whether , for odd @xmath12 , any other ( nonisomorphic ) embeddings with constant corners exist . how to generalize the tower of hanoi puzzle to @xmath12 pegs , @xmath360
obviously , if we add more pegs and retain the same rules , the puzzle just becomes easier to solve ( although the minimum number of moves , even for @xmath361 ( and arbitrary @xmath4 ) , has never been completely determined .
the frame - stewart algorithm , which with 4 pegs takes about @xmath362 moves , is conjectured to be optimal ( see wikipedia , `` tower of hanoi ( four pegs and beyond ) '' .
scorer , grundy & smith @xcite presented several variants .
the one we found most amusing is called `` traveling diplomats '' .
the additional rules they propose for moving discs are described in terms of arcane diplomatic protocols for moving english diplomats from praha to geneva via two airlines that circulate ( with flights in both directions ) through @xmath363 major capitols , berlin , praha , rome , geneva & london .
so @xmath364 and the number of diplomats , @xmath4 , is arbitrary
. however , diplomats are strictly ordered by rank and ( per scorer , grundy & smith ) a : : " each member must always travel three stages by air , for consultation at the two intermediate towns .
b : : no member may start from , visit , or end up in a town at which one of his subordinates is stationed . "
the routes of the two airlines are presented by a `` map '' but there is an equivalent ( and more helpful ) diagram in figure 5 .
@xmath365{2.2174in}{\begin{center } \includegraphics [ trim=2.207469 in 3.006730 in 1.908381 in 3.818943 in , height=2.1179 in , width=2.2174 in ] % { figure5d.eps}% \\ figure 5-map of airline routes \end{center}}}% % endexpansion\ ] ] with this map it is easy to see that 1 .
any move , from town x to town y , can be made in 3 stages on a unique airline .
2 . there is just one town unvisited at each move , so by rule @xmath366 , all subordinates of the transient diplomat must be at that ( unvisited)town . from this one may deduce that a move from @xmath367 to @xmath368 , @xmath142 can be made iff @xmath369 which means that the graph of the traveling diplomats puzzle is @xmath370 .
scorer , grundy & smith also noted that the rules as they state them could be extended to @xmath371 for any prime @xmath372 .
we have constructed such graphs , @xmath337 , for all @xmath4 if @xmath12 is odd , and shown they do not exist for @xmath114 if @xmath358 is even . in our opinion , any such @xmath337 defines a generalized towers of hanoi puzzle : a disc may be moved from peg @xmath249 iff all smaller discs are on peg @xmath374 and then it may be only be moved to peg @xmath375 .
it follows then that @xmath376 and @xmath377 , making the graph of the game @xmath337 .
the tower of hanoi puzzle , as proposed by eduard lucas in 1881 , started with all @xmath4 discs on one peg ( say @xmath17 ) and ended with all of them on another peg ( say @xmath238 ) . some puzzles , such as rubik s cube , are started from a random position .
if the tower of hanoi is started from a random position , it would appear to be a much more complex problem .
it would be simple to select a position at random by labeling the sides of a triangular prism @xmath386 and tossing it @xmath4 times to generate a random member of @xmath387 . in the example , starting at @xmath384 , the largest disc starts on peg @xmath17 and must be moved to peg @xmath238 . in order to do that the three smaller discs must be moved to peg @xmath237 .
after the largest disc is transferred from peg @xmath17 to peg @xmath238 , the three smaller discs may be transferred from peg @xmath237 to peg @xmath238 .
thus the problem is recursively solvable and the same holds in general .
however , deciding the optimal move from a given position , such as @xmath384 seems overwhelming .
case @xmath17 : : @xmath396 , so replace @xmath397 by @xmath238 to get @xmath398 . by the definition of @xmath0 , @xmath22 and @xmath399 .
thus the unique shortest path from @xmath44 to @xmath62 starts with @xmath146 and continues with the unique shortest path from @xmath51 to @xmath400 case @xmath237 : : @xmath401 such that @xmath402 and @xmath403 for @xmath404 .
let @xmath398 be such that@xmath405{ll}% v_{i } & \text{if } i < h\\ 0 & \text{if }
v_{h } & \text{if } i > h \end{array } \right .
\text{.}%\ ] ] again @xmath22 and @xmath406 so we are done .
thus the antipodes of @xmath62 @xmath407 are those @xmath44 such that @xmath408 .
there are @xmath409 of them , each at distance @xmath288 from @xmath62 .
since @xmath410 is isomorphic to @xmath30 , we can use the correspondence @xmath261 and its inverse to solve the tower of hanoi problem with random starting position . for @xmath383 , @xmath411 @xmath412@xmath413 in general
so the matrix for @xmath273 is @xmath414{llll}% 1 & 0 & 0 & 0\\ 2 & 2 & 0 & 0\\ 2 & 1 & 1 & 0\\ 2 & 1 & 2 & 2 \end{array } \text{.}%\ ] ] surprisingly , this matrix is self - inverse , so it is also the matrix for @xmath415 . therefore @xmath416{llll}% 1 & 0 & 0 & 0\\ 2 & 2 & 0 & 0\\ 2 & 1 & 1 & 0\\ 2 & 1 & 2 & 2 \end{array } \right ) \left
( \begin{array } [ c]{l}% 1\\ 0\\ 2\\ 0 \end{array } \right ) = \left ( \begin{array } [ c]{l}% 1\\ 2\\ 1\\ 0 \end{array } \right ) \text{.}%\ ] ] @xmath417 .
the vertex one step closer to @xmath418 in @xmath419 is @xmath420 and the one after that is @xmath421 .
continuing on in this way we generate the minimum length path from @xmath422 to @xmath418 in @xmath423 .
from there we apply @xmath273 to obtain the minimum length path from @xmath424 to @xmath418 in @xmath425 : @xmath426{|r||cr}\hline $ \ell$ & $ s\left ( 4,3\right ) $ & \multicolumn{1}{|r|}{$t\left ( 4,3\right ) $ } \\\hline $ 14 $ & \multicolumn{1}{||r}{$1210 $ } & $ 1020$\\\hline $ 13 $ & \multicolumn{1}{||r}{$1201 $ } & $ 1010$\\\hline $ 12 $ & \multicolumn{1}{||r}{$1200 $ } & $ 1011$\\\hline $ 11 $ & \multicolumn{1}{||r}{$1022 $ } & $ 1211$\\\hline $ 10 $ & \multicolumn{1}{||r}{$1020 $ } & $ 1210$\\\hline $ 9 $ & \multicolumn{1}{||r}{$1002 $ } & $ 1220$\\\hline $ 8 $ & \multicolumn{1}{||r}{$1000 $ } & $ 1222$\\\hline $ 7 $ & \multicolumn{1}{||r}{$0111 $ } & $ 0222$\\\hline $ 6 $ & \multicolumn{1}{||r}{$0110 $ } & $ 0220$\\\hline $ 5 $ & \multicolumn{1}{||r}{$0101 $ } & $ 0210$\\\hline $ 4 $ & \multicolumn{1}{||r}{$0100 $ } & $ 0211$\\\hline $ 3
$ & \multicolumn{1}{||r}{$0011 $ } & $ 0010$\\\hline $ 2 $ & \multicolumn{1}{||r}{$0010 $ } & $ 0012$\\\hline $ 1 $ & \multicolumn{1}{||r}{$0001 $ } & $ 0001$\\\hline $ 0 $ & \multicolumn{1}{||r}{$0000 $ } & $ 0000$\\\hline \end{tabular}\ ] ] we can also write a formula to solve the traveling diplomats puzzle of @xcite : suppose we take @xmath383 and number the capitols , praha @xmath427 , geneva @xmath428 , paris @xmath429 , rome @xmath430 , and london @xmath431 in the order of the boac circuit .
the challenge is to transport all four diplomats from praha to geneva , so we start at @xmath432and end at @xmath433 on @xmath434 in @xmath435 the minimum path is @xmath436 , @xmath437 and mapping that path to @xmath438 by @xmath273 , whose matrix of coefficients @xmath439 is @xmath440{rrrr}% 1 & 0 & 0 & 0\\ 3 & 3 & 0 & 0\\ 3 & 4 & 4 & 0\\ 3 & 4 & 2 & 2 \end{array } \right)\ ] ] we get@xmath426{|r||rr}\hline $ \ell$ & $ s(4,5)$ & \multicolumn{1}{|r|}{$t(4,5)$}\\\hline $ 0 $ & $ 0000 $ & $ 0000$\\\hline $ 1 $ & $ 0001 $ & $ 0002$\\\hline $ 2 $ & $ 0010 $ & $ 0042$\\\hline $ 3 $ & $ 0011 $ & $ 0044$\\\hline $ 4 $ & $ 0100 $ & $ 0344$\\\hline $ 5 $ & $ 0101 $ & $ 0341$\\\hline $ 6 $ & $ 0110 $ & $ 0331$\\\hline $ 7 $ & $ 0111 $ & $ 0333$\\\hline $ 8 $ & $ 1000 $ & $ 1333$\\\hline $ 9 $ & $ 1001 $ & $ 1330$\\\hline $ 10 $ & $ 1010 $ & $ 1320$\\\hline $ 11 $ & $ 1011 $ & $ 1322$\\\hline $ 12 $ & $ 1100 $ & $ 1122$\\\hline $ 13 $ & $ 1101 $ & $ 1124$\\\hline $ 14 $ & $ 1110 $ & $ 1114$\\\hline $ 15 $ & $ 1111 $ & $ 1111$\\\hline \end{tabular}\ ] ] the matrix of coefficients @xmath441 for @xmath442 @xmath443 is @xmath440{rrrr}% 1 & 0 & 0 & 0\\ 4 & 2 & 0 & 0\\ 4 & 3 & 4 & 0\\ 4 & 3 & 1 & 3 \end{array } \right)\ ] ] which might be used to recall a random distribution of the four diplomats among the 5 capitols to back to praha in the most efficient manner .
the nonzero entries of the matrices in these two examples have pseudorandom characteristics since they are sums of exponentials reduced @xmath347 ( see wikipedia ) .
it is amusing that they act to bring order out of the apparent chaos of moves to solve tower of hanoi problems .
the key ingredient of this magic is the scorer - grundy - smith transform , which takes the combinatorial structure of @xmath444 and maps it onto the geometric structure of @xmath445 .
[ [ is - taus43rightarrow - t43-self - inverse - by - coincidence - or - principle ] ] is @xmath446 self - inverse by coincidence or principle ? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in example 7 we noticed that @xmath447 .
is this just a coincidence or the result of some underlying principle which make it true for all @xmath448 ?
the answer is somewhere in between : if @xmath200 it holds for all @xmath4 .
this can be seen from theorems 8 & 9 , since @xmath449 .
if @xmath364 it does not hold for @xmath383 ( shown in section 4.14 ) and seems unlikely for any @xmath4 if @xmath360 .
1 . answering the question of the title in the affirmative , and 2 . making explicit what was implicit in scorer , grundy & smith s paper @xcite .
they described a `` trilinear mapping '' from @xmath255 to @xmath30 in synthetic and qualitative terms . utilizing the self - similarity of the sierpinski graph
, we characterized that mapping by a recurrence and solved the recurrence .
what was synthetic and qualitative has become analytic , computational and surprisingly efficient .
hinz , andreas m. ; klavar , sandi ; milutinovi , uro and petr , ciril ; . with a foreword by ian stewart .
birkhuser / springer basel ag , basel ( 2013 ) .
xvi+335 pp . isbn : 978 - 3 - 0348 - 0236 - 9 ; 978 - 3 - 0348 - 0237 - 6 william , albert ; rajasingh , indra ; rajan , bharati & shanthakumari , a. : topological properties of sierpinski gasket pyramid network , in , proceedings ( part iii ) of an international conference ( icieis 2011 ) at kuala lumpur , maylaysia , november 14 - 16 , 2011 , pp .
431 - 439 , springer - verlag . | the ( generalized & expanded ) sierpinsi graph , @xmath0 , and the hamming graph , @xmath1 , have the same set of vertices , @xmath2 .
the edges of both are ( unordered ) pairs of vertices .
each set of edges is defined by a different property so that neither contains the other .
we ask if there is a subgraph of @xmath1 isomorphic to @xmath0 and show that the answer is yes .
the recursively defined embedding map , @xmath3 , leads to a number of variations and ramifications . among them
is a simple algebraic formula for the solution of the tower of hanoi puzzle . |
galaxy clusters , which are the most massive collapsed structures in the universe , are made up of hundreds to thousands of galaxies , a massive reservoir of hot ( @xmath1310@xmath14k ) plasma , and dark matter .
the latter dominates the mass budget , contributing @xmath1090% of the the total mass for a massive galaxy cluster ( e.g. , * ? ? ?
of the remaining @xmath1010% in mass , the hot intracluster medium ( icm ) outweighs the stars by a factor of roughly ten ( e.g. , * ? ? ?
* ; * ? ? ?
visible in x - rays as diffuse emission on mpc scales , the hot icm retains the imprint of the cluster history in its thermodynamic and chemical abundance profiles and in its x - ray morphology , allowing major events in the history of a given cluster to be inferred billions of years later .
the chemical abundance of the hot icm contains the cumulative enrichment history from various processes since the big bang , including mass loss from evolved stars , heavy element enrichment from supernovae , and dilution of metals due to mixing with low - metallicity gas infalling along cosmic filaments . in practice , with ccd - resolution x - ray spectra from _ chandra _ and _ xmm - newton _ , constraints on the icm metal abundance come primarily from the equivalent width measurement of the fe k@xmath15 emission line at 6.7 kev . assuming that the ratio of iron to other elements is constant across the universe , we can infer a metal abundance by comparing the iron abundance in a given cluster to that of the sun . in massive , low-@xmath16 clusters ( @xmath17 ) , the average observed metallicity is roughly a third of the solar value ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , assuming solar abundances from @xcite . in `` cool core '' clusters those with central cooling times significantly shorter than the age of the universe ( e.g. , * ? ? ?
* ) the metallicity peaks in the center ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , sometimes reaching values @xmath18 ( e.g. , * ? ? ?
this may be owing to the fact that the central , most massive galaxy ( which can enrich the icm via stellar mass loss ) tends to be in the cluster center for cool core clusters , leading to centrally - peaked stellar - to - gas ratios for these systems which is not present in non - cool core clusters .
the metal enrichment history of the icm remains a mystery .
recent studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have attempted to quantify how the metallicity evolves , both in the core and outer regions , for galaxy clusters at @xmath19 . using data from the _ chandra x - ray observatory _ , @xcite reported evolution of the metal abundance for 115 galaxy clusters at @xmath20 .
the sample , defined as all known clusters with existing data in the _ chandra _ archive at the time , is fairly representative of the true cluster population at low redshift , but is biased towards extreme ( well - studied ) systems at high redshift .
further , it is known that some of the most relaxed systems ( which exhibit enhanced metallicity in their cores ) have been missed by x - ray surveys because their cool cores appear point - like at large distances ( e.g. , the phoenix cluster ; * ? ? ?
the sample used by @xcite had few ( 12 ) clusters at @xmath21 , where they observed the strongest evolution .
more recent studies , first by @xcite and then @xcite , reported results from the _ xmm - newton _
telescope , employing a sample of all clusters in the archive at @xmath22 .
this sample likely suffers from similar selection biases to the study by @xcite .
these later works find no measurable evolution in the global metallicity of clusters at @xmath23 , although @xcite find marginal ( @xmath24 ) evidence for evolution in the centers of cool core clusters .
none of these studies find a strong dependence between cluster mass ( or temperature ) and metallicity , with @xcite reporting @xmath25 . here , we present the first study of icm metallicity evolution in a sample of galaxy clusters selected via the sunyaev zeldovich effect @xcite .
this selection yields a mass - limited , redshift - independent sample of clusters that is relatively unbiased to the presence or lack of a cool core @xcite .
we focus here on x - ray follow - up of galaxy clusters at @xmath26 , providing the best constraints on the global metallicity evolution and the radius - dependent evolution over this redshift range .
the data used in this analysis , from the _ chandra _ , _ xmm - newton _ , and _ suzaku _ telescopes , are described in 2 . in
3 we present the constraints on metallicity evolution , and describe the dependence of this evolution on the presence or lack of a cool core . in
4 we speculate on the cause of metal enrichment , both in the cores and outskirts of galaxy clusters , and compare our results to previous works . finally , in 5 we will summarize the results and comment on the ability of future surveys to improve these constraints . throughout this work
we assume h@xmath27 = 70 km s@xmath28 mpc@xmath28 , @xmath29 = 0.3 , @xmath30 = 0.7 .
this work is based on a sample of galaxy clusters selected via the sunyaev zeldovich ( sz ) effect in the 2500 square degree south pole telescope sunyaev zeldovich ( spt - sz ) survey @xcite .
this survey , completed in 2011 , discovered 516 galaxy clusters in the southern sky , at @xmath31 with masses @xmath32 m@xmath5 . for a more detailed description of the selection and cluster properties ,
the reader is directed to @xcite . for a subsample of 153 clusters we have x - ray follow - up from the _ chandra _ , _ xmm - newton _ , and/or _ suzaku _ telescopes .
the mass - redshift distribution for the clusters with follow - up x - ray data is shown in figure [ fig : mz ] .
this figure emphasizes the uniform mass selection and broad redshift coverage of our sample .
below , we describe in detail the follow - up strategy and data analysis for each of these three x - ray observatories . ) versus redshift for 516 clusters discovered in the full spt - sz survey ( grey points ; * ? ? ?
clusters observed with _
chandra _ , _ xmm - newton _ , and _ suzaku _ are colored red , blue , and green , respectively .
this figure demonstrates the clean mass selection of this sample .
masses in this plot are estimated from the sz significance , following @xcite.,scaledwidth=49.0% ] from the larger sample of 516 confirmed clusters in the spt - sz survey , we have performed a nearly mass - limited x - ray follow - up of 96 clusters at @xmath33 and m@xmath34 m@xmath5 with the _ chandra x - ray observatory_. the majority of these clusters were observed as part of either an _ x - ray visionary program _ to obtain shallow x - ray imaging of the 80 most massive spt - selected clusters at @xmath35 ( pi : benson ) or a large program to observe 10 spt - selected clusters at @xmath36 ( pi : mcdonald ) .
the remaining systems were observed through various smaller go ( pis : mcdonald , mohr ) and gto ( pis : garmire , murray ) programs , or were available in the archive .
the details of these observations are summarized in @xcite . for each spt - selected cluster
observed with _
chandra _ , the total exposure time was estimated with a goal of obtaining 2000 x - ray counts , assuming relationships between the sz detection significance and cluster mass , and between cluster mass and x - ray luminosity @xcite .
owing to significant scatter in these relations , coupled with uncertainty in preliminary cluster redshifts , the observed x - ray counts vary by a factor of @xmath102.5 around this goal ( see fig . 2 in * ? ? ?
we have much deeper data ( @xmath3710,000 counts ) on a subset of 7 well - studied clusters in this sample ( e.g. , bullet cluster ( pi : markevitch ) , el gordo ( pi : hughes ) , phoenix ( pi : mcdonald ) ) , which yields well - measured metallicities in a subsample of systems .
we note that , since we consider both the intrinsic scatter and measurement uncertainty in our fitting procedure , the inclusion of high - s / n systems will not strongly bias our results .
we will return to this point in 5.2 . for each cluster we estimate r@xmath38 , the radius within which the average enclosed density is 500 times the critical density ( @xmath39 ) , based on the sz - derived mass presented in @xcite .
we note that sz - derived masses may be underestimated for low-@xmath16 ( @xmath40 ) clusters and are only approximate for low - significance ( @xmath41 ) clusters @xcite
. however , this already - small bias has a reduced ( @xmath42 ) effect on the inferred radius , which has an even smaller effect on the integrated metallicity , given the flatness of the metallicity profile at @xmath43 @xcite .
x - ray data from _ chandra _ are reduced following procedures outlined in @xcite , @xcite , and @xcite using ciao v4.7 and caldb v4.6.8 , along with the latest blank - sky background and response files .
all exposures are initially filtered for background flares , before applying the latest calibration corrections and determining the appropriate ( epoch - based ) blank - sky background .
given that the typical angular size of a galaxy cluster at @xmath3 is only @xmath102.5@xmath44 , we are also able to extract off - source background files at distances of @xmath453r@xmath38 from the cluster center for each observation .
blank - sky background spectra are rescaled based on the observed 9.512kev flux , and subtracted from both on- and off - source spectra . any residual background
( e.g. , from unresolved point sources or galactic emission ) is accounted for by simultaneous modeling of on- and off - source spectra ( see below ) .
point sources are identified using an automated routine following a wavelet decomposition technique @xcite , and then visually inspected .
the center of the cluster is chosen by iteratively measuring the centroid in a 250500 kpc annulus , following @xcite .
this choice of center is insensitive to structure in the core and is a more accurate proxy for the center of the underlying dark matter potential than the x - ray peak . for each cluster observation
, we extract spectra in three different annuli : i ) @xmath46 ( total ) ; ii ) @xmath47 ( core ) ; and iii ) @xmath48 (
core - excised ) .
the first of these annuli will be compared with measurements from _ xmm - newton _ and _ suzaku _ , which , for the high-@xmath16 systems ( @xmath8 ) , do not have sufficient angular resolution to excise the core or identify whether the cluster harbors a cool core or not .
the combination of full - aperture data from all three telescopes will allow us to tightly constrain the evolution of the total metal content in clusters out to @xmath49 .
the latter two annuli , which we can only measure using the high angular resolution _
data , will provide weaker constraints on how this evolution depends on the presence or lack of a cool core , and whether it is stronger in the core than in the outskirts .
spectra are individually fit in combination with their respective background spectra ( see 2.2.1 ) over the energy range 0.510.0 kev with xspec ( v12.9.0 ; * ? ? ?
* ) , using a combination of a single - temperature plasma ( apec ; * ? ? ?
* ) , a soft x - ray galactic background ( apec , @xmath50 kev , @xmath51 , @xmath52 ) , a hard x - ray cosmic background ( bremss , @xmath53 kev ) , and a galactic absorption model ( phabs ) .
metallicity measurements are based on the solar abundances of @xcite . for each observation
, the normalization of the hard x - ray background is tied between the on- and off - source spectra , while the soft x - ray background is tied between all spectra for a given cluster ( i.e. , multiple obsids ) .
this difference is to account for the fact that the hard x - ray background should be exposure time dependent ( i.e. , larger fraction of the cosmic x - ray background ( cxb ) resolved into point sources for long exposure times ) , while the soft , diffuse x - ray background is not .
an example result of this fitting procedure is shown in figure [ fig : spectrum ] .
the inferred icm metallicity is most sensitive to the fe k@xmath15 emission line at 6.7 kev .
since the fit to this line is also strongly dependent on the cluster redshift , we allow the redshift to float in the fitting procedure within the 2@xmath54 uncertainties quoted by @xcite . for photometric redshifts ,
this uncertainty is typically @xmath55 .
for spectroscopic redshifts , we use a fixed uncertainty of 3000 km s@xmath28 ( @xmath56 ) . in a similar fashion , we allow the galactic absorption column to float within @xmath1015% of the measured value from @xcite , which tends to improve the continuum fit .
goodness - of - fit is determined using a modified version of the @xmath57 parameter , following @xcite , which has been shown to yield unbiased parameter estimates for spectra containing as few as @xmath1050 total counts . for each cluster ,
we use the markov - chain monte - carlo ( mcmc ) solver in xspec to determine the full probability distribution for the icm metallicity . using the metropolis - hastings algorithm
, we run 5 chains for each cluster , each with a length of 25000 and discarding the first 5000 steps . integrating over all other free parameters ( @xmath58 , @xmath59 , @xmath16 , source normalization , soft background normalization , hard background normalization ) yields the posterior distribution for @xmath60 ( hereafter @xmath61 ) . in figure
[ fig : fig1 ] , we show the derived @xmath61 for each cluster , highlighting two non - cool core clusters ( hereafter ncc ; spt - clj0102 - 4915 , spt - clj0546 - 5345 ) and two cool - core clusters ( hereafter cc ; spt - clj0000 - 5748 , spt - clj2344 - 4243 ) , where we use the central entropy ( @xmath62 ) as the classification metric , following @xcite . for both types ( cc / ncc ) , we show a low - s /
n case and a high - s / n case , to demonstrate the difference in metallicity constraints from system to system . in 3 , we will describe how these individual cluster constraints are incorporated into a framework which allows us to constrain the overall metallicity evolution . , and
grey points show the off - source background , normalized to the same area .
red and blue curves show the best - fit models for the on- and off - source spectra , respectively .
the presence of a redshifted iron emission line at @xmath105 kev allows us to constrain the metallicity of this system these constraints from 10,000 mcmc chains are shown in the inset ( see 2.2.2 for more details).,scaledwidth=49.0% ] ) at @xmath1 for 96 spt - selected clusters that have been observed with _
these curves were derived using the monte carlo markov chain solver in xspec , as described in 2.2.2 .
we highlight four clusters in color , which demonstrate weak ( green , black ) and strong ( blue , red ) constraints for low - metallicity ( blue , green ) and high - metallicity ( black , red ) clusters.,scaledwidth=49.0% ] a significant number of spt - selected clusters have been observed by _ xmm - newton _ as part of several different programs including , but not limited to , those focused on weak lensing mass calibration ( pis : benson , suhada ) and the study of the highest redshift spt - selected systems ( pis : benson , andersson ) . in
all , as of the end of 2015 , 69 spt - selected clusters have been observed by _ xmm - newton _ , with no preference in selection for the dynamical / cooling state of the cluster the full details of this sample are provided in bulbul et al .
( in prep ) .
given that the fraction of cool cores in this randomly - selected subsample should be representative of the true population , we are justified in combining these data with the larger , more complete sample of 96 clusters observed with _
chandra_. event files were calibrated using the xmm - newton science analysis system ( sas ) version 14.0.0 , and the most recent calibration files as of october 2015 .
calibrated , clean event files were produced after filtering for high intensity particle - background flares .
additional details of data reduction and analysis are described in detail in @xcite .
the images are extracted in a 0.47.0 kev bandpass from all instruments and pointings .
the particle background and soft proton background subtracted images were used to detect the point source in the field of view . detected point sources were excluded from the further analysis .
the source and background spectra were extracted within the overdensity radius r@xmath38 , as derived from @xcite .
we model the background with a superposition of four main components : quiescent particle background , cosmic x - ray background emission ( including galactic halo , local hot bubble , and unresolved extragalactic sources ) , solar wind charge exchange , and residual contamination from soft protons .
we use the _ rosat _ all - sky survey background spectrum which is extracted from an annulus from 0.5@xmath63 to 1.5@xmath63 surrounding the cluster center to model the soft x - ray background .
the 0.310 kev energy interval is used for mos spectra , whereas the 0.4 -10.0 kev band was used for the pn fits .
the remaining cluster model parameters and fitting process was identical to that used in the _ chandra _ analysis .
four spt - selected clusters were observed with _ suzaku _ during cycle 6 ( pi : e. miller ) .
one of the primary goals of this program was to constrain the metallicity of galaxy clusters at high redshift , with the four clusters spanning @xmath64 ( see figure [ fig : mz ] ) .
these systems were found to have slightly higher than average metallicities , with values ranging from 0.260.45 z@xmath5 @xcite . however , with only four clusters
, we were unable to make any claim on metallicity evolution with certainty . thus , we have included these four high - s / n measurements into a larger sample , which provide marginal improvements on the overall metallicity evolution of spt - selected clusters .
event data for xis0 , xis1 , and xis3 were reprocessed using the most current _ suzaku _ calibration products as of september 2015 .
point sources were identified from shallow _
data for each cluster using the ciao tool wavdetect , and masked out in the _ suzaku _ data to a radius of 1@xmath44 .
spectra for cluster emission were extracted from a region 4@xmath44 in radius centered on the x - ray peak ( about twice the typical r@xmath38 for these four clusters ) to account for the large size ( 2@xmath44 half - power diameter ) of the _ suzaku _ psf .
this is sufficient to collect 99% of the counts from an on - axis point source .
x - ray background spectra were extracted from the remaining source - free regions of each detector , excluding the central 5@xmath44 , the calibration source regions , and bad detector areas .
response files were produced for the source and x - ray background with xisrmfgen and xissimarfgen , the latter using a background - subtracted _ chandra _ image of the cluster as an input source for ray - tracing to produce the cluster auxiliary response file ( arf ) .
the x - ray background arf was constructed using a uniform , 20@xmath44 radius source .
particle background spectra were produced for each source and x - ray background region with the ftool xisnxbgen , with a filter to exclude parts of the spacecraft orbit with low geomagnetic cut - off rigidity ( cor2 @xmath45 6 gv ) and thus reduce the background .
this filter was also applied to the source and x - ray background spectra .
spectral fitting was performed with xspec v12.9 , fitting all three xis simultaneously with all free paramaters tied between instruments .
the particle background spectra were subtracted from the source and x - ray background spectra during spectral fitting .
the x - ray background model consisted of a one solar - abundance apec component to model the total galactic foreground , with temperature and normalization free ; and one power law component with fixed @xmath65 and free normalization to model the unresolved cosmic x - ray background .
both components were absorbed by the same @xmath66 column as the cluster model .
the x - ray background parameters were tied between the cluster and background regions .
the remaining cluster model parameters and fitting process was identical to that used in the _ chandra _ analysis .
) as a function of redshift . for each redshift bin , we combine the probability functions , @xmath67 , for the nearest 21 clusters in redshift space , computing the combined probability distribution for the metallicity of clusters at that redshift . in black , we show the median value of that distribution , which does not appear to evolve with redshift . in dark and light grey
we show the @xmath68 ( 68% ) and @xmath69 ( 95% ) confidence regions , respectively .
the horizontal red line shows a non - evolving metallicity , while the blue lines show the best - fit model from the following section .
this figure demonstrates that both the scatter and median values are well - fit by a non - evolving model , justifying our use of a simple , linear evolution term in
4.,scaledwidth=49.0% ] the ultimate goal of this work is to quantify how the integrated icm metallicity has evolved in galaxy clusters over the past @xmath109 gyr .
as a first step , we show in figure [ fig : zz_allxray ] the evolution in the icm metallicity for subsamples of 21 clusters from @xmath70 to @xmath71 , based on the full sample of clusters from all three x - ray observatories . in each redshift bin , we combine the p(z ) distributions for the 21 nearest clusters in redshift space , showing the median value of this combined probability distribution , as well as the 1@xmath54 ( dark grey ) and 2@xmath54 ( light grey ) widths .
this figure provides three initial insights into the evolution of the global ( @xmath72 r@xmath38 ) metal content of the icm .
first , there is no sign of complex ( e.g. , non - linear ) evolution in the median metallicity .
given the quality of these data , any evolution that does exist would be as well fit by a linear term in redshift ( e.g. , @xmath73 ) as it would by something more complex ( e.g. , @xmath74 ) .
second , the scatter in metallicity at a given redshift appears to be roughly symmetric about the median , such that it is well - approximated by a gaussian distribution with @xmath75z@xmath5 .
there does appear to be some asymmetry to the distribution ( because @xmath76 is unphysical ) , but this asymmetry is not present at the 1@xmath54 level .
finally , there does not appear to be any significant evolution in the scatter about the median metallicity out to @xmath49 .
the combination of these three insights implies that a simple linear evolution ( in redshift ) with non - evolving gaussian scatter is an appropriate choice of model given our data quality and the properties of the clusters in this sample . in the following section
, we will employ this simple model to provide constraints on the evolution of the average icm metallicity as a function of both redshift and radius .
given the large uncertainties inherent in our metallicity measurements , and the justification provided in the previous section , we opt for a simple linear model to express the evolution of metallicity as a function of redshift : @xmath77 where @xmath78 is chosen to roughly minimize covariance between @xmath79 and @xmath80 for the subsample of clusters observed with _
chandra _ ( @xmath81 ) .
we note that , while the data presented in this paper are of sufficient quality to constrain the two parameters in this simple model , they are not of sufficient quality to determine if a more complex functional form is justified , especially at @xmath8 . assuming gaussian intrinsic scatter ( independent of metallicity and redshift ) , characterized by a width c , the probability that a cluster at redshift @xmath16 has metallicity @xmath60 is given by : @xmath82 combining these probabilities for the full sample yields the posterior probability for the set of model parameters : @xmath83 where @xmath84 refers to a given cluster , @xmath85 is the prior probability ( taken to be flat for all parameters ) , and @xmath86 is the probability distribution for the metallicity of a given cluster , as obtained from xspec ( e.g. , figure [ fig : fig1 ] ) .
we can then write the approximate likelihood as : @xmath87 where @xmath88 denotes the @xmath89 sample ( from @xmath90 total samples ) drawn from the posterior @xmath60 distribution of the @xmath91 cluster , as given by the mcmc solver in xspec .
this posterior probability distribution of the model parameters is then explored using a uniformly spaced fine grid in all three parameters .
we note that , by including intrinsic scatter in this fit , we reduce the effect that a few high s / n systems can have on the fit outcome .
this approach is similar to considering a modified @xmath57 of the form @xmath92 . applying this formalism to the data described in 2 yields posterior distributions for @xmath93 , which we now assign the more physically - appropriate labels of @xmath94\right>$ ] , @xmath95 , and @xmath96 , respectively . in figure
[ fig : mcmc_allxray ] , we show the constraints on each of these three parameters for the full set of 153 clusters with _
chandra _ , _ xmm - newton _ , and/or _
suzaku _ data .
here we are considering the evolution of the _ global _ metallicity ( @xmath1 ) .
for this sample , we find @xmath94\right > = 0.23 \pm 0.01z_{\odot}$ ] , @xmath2 , and @xmath97 this best - fitting model is overplotted on the data in figure [ fig : zz_allxray ] .
the 1@xmath54 contours for all three parameters overlap between subsamples , suggesting that there is no systematic difference in either selection or treatment of the data .
figure [ fig : mcmc_allxray ] indicates that there is very little ( if any ) evolution in the global icm metallicity over the past @xmath109 gyr .
these data indicate that , in a typical galaxy cluster with @xmath98 m@xmath5 , the emission - weighted metallicity has changed by less than 40% since @xmath7 ( @xmath4595% confidence ) .
such a lack of evolution is consistent with previous studies , with this work providing the strongest constraints to date .
one can ask whether the evolution is stronger in the cores of clusters , where baryonic processes are important , than in the cluster outskirts , or whether the dynamical state of the cluster ( e.g. , relaxed or unrelaxed ) has any bearing on the observed metallicity . to address both of these questions
, we require the high angular resolution provided by _
chandra _ and , thus , will limit the remaining analysis in this section to the 96 clusters with _
chandra _ observations . .
in the upper panel , we show the constraints on the scatter and normalization , while in the lower panel we compare the slope and normalization terms . in both panels ,
we show constraints from each of the three telescopes individually , along with the combined constraints in grey .
contours represent confidence levels of 68% , 95% , and 99.7% .
overplotted in black dashed lines is a grid of evolutionary scenarios , to help contextualize these results .
the tight constraints on the evolutionary term imply that , at most ( with @xmath4595% confidence ) , 40% of the intracluster metals ( including the core ) in present - day clusters were created in the past @xmath108 gyr .
this suggests that the bulk of the metals in the icm were most likely created very rapidly early on.,scaledwidth=49.0% ] the measured evolution of the _ spatially - resolved _ icm metallicity for clusters observed with _
chandra _ is shown in figure [ fig : mcmc_chandra ] and summarized in table [ table : mcmc ] .
as described in 2 , we consider the metallicity in the core volume ( @xmath99r@xmath38 ) , the core - excised volume ( 0.15r@xmath100 r@xmath38 ) , and the full volume ( @xmath101 r@xmath38 ) .
figure [ fig : mcmc_chandra ] shows that these data , despite being of relatively low quality on a per - cluster basis ( @xmath102000 x - ray counts per cluster ) , provide excellent constraints on the metallicity evolution .
we exclude _ xmm - newton _ and _ suzaku _ data from this part of the analysis , due to the fact that , at @xmath102 , @xmath4520% of the flux from the core will be scattered outside of 0.15r@xmath38 . given the broad redshift range covered , disentangling this psf - driven bias in the core - excised metallicity from an actual evolution would be challenging .
we find that the core - excised metallicity has a mean value of @xmath94\right > = 0.17 \pm 0.02 $ ] z@xmath5 and @xmath103 , corresponding to a mean value at redshift zero of @xmath104\right > = 0.19 \pm 0.04 $ ] z@xmath5 .
this is remarkably consistent with the measured value from deep observations of the perseus cluster of @xmath105 z@xmath5 @xcite , and other nearby clusters @xcite , assuming solar abundances from @xcite .
we find no evidence for evolution in the metallicity outside of the core , independent of whether the cluster is a cool core or non - cool core , suggesting that the bulk of the metals outside of the core were created at @xmath8 .
again , this is consistent with earlier work by @xcite , who showed that the lack of azimuthal variations in the metallicity of the perseus cluster implies that metal enrichment happened very early in the cluster lifetime .
we note that there is very little intrinsic scatter in the core - excised metallicity ( @xmath106 z@xmath5 ) , and that the mean value , redshift slope , and scatter of the metallicity outside of the core is independent of the cooling state ( entropy ) of the core . in the cores of galaxy clusters ( @xmath107r@xmath38 ) ,
we find , for the full population , significantly higher metallicity . the mean value of @xmath94\right > = 0.32 \pm 0.03 $ ] z@xmath5 is offset from the core - excised value by 5@xmath54 .
further , if we divide the sample roughly in half into cc ( k@xmath108 kev @xmath109 ) and ncc ( k@xmath110 kev @xmath109 ) subsamples , the mean metallicity for each subsample is @xmath94\right>_{cc } = 0.39 \pm 0.03 $ ] z@xmath5 and @xmath94\right>_{ncc } = 0.24 \pm 0.05 $ ] z@xmath5 , respectively .
that is , there is a @xmath104@xmath54 offset between the core metallicity in cc and ncc clusters . despite this offset
, we measure no difference in the scatter of core metallicity between cc and ncc systems .
this dichotomy between the core metallicity in cc and ncc clusters is already well known and consistent with previous works ( e.g. , * ? ? ?
* ; * ? ? ?
the only significant metallicity evolution is measured in the cores of cc clusters , for which we measure @xmath111 z@xmath5 . while only significant at the 2@xmath54 level , it is worth noting that this evolution corresponds to , on average , @xmath1040% of the metals in cool cores arriving between @xmath7 and @xmath52 .
this is significantly shallower than the 77@xmath112% evolution measured by @xcite a point we will return to in the discussion .
this analysis has provided a clear picture of the metallicity evolution in galaxy clusters at @xmath113 .
we find no statistically significant evolution in outskirts of clusters , independent of the dynamical state of the core , and only marginal evidence for metallicity evolution in the inner regions of cool core clusters .
we confirm substantial differences between cc and ncc cores in terms of the average metallicity , with cc clusters having significantly higher average metallicity . outside of the core , cc and
ncc clusters are indistinguishable based on their metallicity .
most importantly , we find that @xmath4560% of the metals in the icm were already in place at @xmath7 ( with @xmath4595% confidence ) . in the following section
, we will place these results in the context of previous works and discuss their implications with regards to various enrichment scenarios .
l c c c c[h ! ] + + 0.0 0.15 r@xmath38 & & @xmath114 & @xmath115 & @xmath116 + 0.0 0.15 r@xmath38 & cc & @xmath117 & @xmath118 & @xmath119 + 0.0 0.15 r@xmath38 & ncc & @xmath120 & @xmath121 & @xmath122 + + 0.15 1.0 r@xmath38 & & @xmath123 & @xmath124 & @xmath125 + 0.15 1.0 r@xmath38 & cc & @xmath126 & @xmath127 & @xmath128 + 0.15 1.0 r@xmath38 & ncc & @xmath129 & @xmath130 & @xmath131 + + 0.0 1.0 r@xmath38 & & @xmath132 & @xmath133 & @xmath134 + 0.0 1.0 r@xmath38 & cc & @xmath135 & @xmath136 & @xmath137 + 0.0 1.0 r@xmath38 & ncc & @xmath138 & @xmath139 & @xmath140 + + + 0.0 1.0 r@xmath38 & & @xmath141 & @xmath142 & @xmath143 + [ table : mcmc ]
there are a number of previous studies which have attempted to constrain the metallicity evolution of the icm , both directly and indirectly ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . conveniently , @xcite have compiled data from several of these earlier works , and expressed the metallicity evolution in a consistent way . to allow a direct comparison to these works
, we will adopt the formalism of @xcite for the remainder of this section .
these authors describe the evolution of the icm metallicity as @xmath144 , with a normalization constant at @xmath3 , leading to the expression : @xmath145 we re - run our analysis using the same model , for consistency .
we also consider a new aperture , @xmath146 , which is similar to those used in previous works ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in figure
[ fig : mcmc_baldi ] we compare the results of this re - analysis to previous works @xcite . for clarity , we provide in table [ table : baldi ] two easily - interpreted quantities for each fit : the average metallicity predicted at @xmath52 and the predicted change in metallicity from @xmath7 to @xmath52 .
we derive these two estimates , and their uncertainties , based on the model fits provided in the literature , and compare to this work . in general , the work presented here agrees well with results from the literature , with this work providing the tightest constraints to date on the metallicity evolution .
much of the discrepancy between works can be attributed to different apertures ( see e.g. , column 2 of table [ table : baldi ] ) and different selection methods / criteria .
c c c c[htb ] + + * this work * & @xmath46 & @xmath147 & @xmath148 + b12 & @xmath149 & @xmath150 & @xmath151 + a09 & @xmath152 & @xmath153 & @xmath154 + m08 & @xmath46 & @xmath155 & @xmath156 + + + + + * this work * & @xmath48 & @xmath157 & @xmath158 + * this work * & @xmath146 & @xmath159 & @xmath160 + e15 & @xmath161 & @xmath162 & @xmath163 + e15@xmath164 & @xmath161 & @xmath165 & @xmath166 + b12 & @xmath167 & @xmath168 & @xmath169 + bls07 & @xmath170 & @xmath171 & @xmath172 [ table : baldi ] considering the integrated metallicity ( core included ) , the constraints on the evolutionary term , @xmath173 , are substantially improved by this work , compared to previous works .
we find @xmath174 , consistent with no evolution .
for comparison , @xcite find @xmath175 , which ( for a fixed normalization ) is consistent with anywhere from a 080% change in the emission - weighted metallicity since @xmath7 ( at 95% confidence ) , compared to 040% found in this study .
there is excellent agreement between our work and those of @xcite and @xcite , and some tension with the earlier results of @xcite . )
icm metallicity evolution from this work compared to previous works .
colored contours , which enclose 68% and 95% confidence regions , are taken from @xcite .
the grey contours show the significantly ( factor of 2 ) improved constraints on the evolutionary parameter @xmath173 from this work .
the lack of significant improvement in the normalization term stems from the fact that we are not actually measuring the metallicity more precisely than previous studies , nor do we have more clusters in our sample it is the redshift distribution of these clusters that leads to an improvement in the slope.,scaledwidth=49.0% ] the tightest constraints prior to this work came from @xcite , who utilized a sample of 83 galaxy clusters at @xmath176 , all observed with _
xmm - newton_. these clusters were drawn from the literature , so the selection effects are challenging to quantify precisely .
for their full sample , combining metallicities measured at all radii , @xcite found a metallicity normalization ( @xmath177 z@xmath5 ) that is significantly higher ( @xmath453@xmath54 ) than our measurement . we do not show a direct comparison to this measurement in figure [ fig : mcmc_baldi ] , since it is not an integrated measurement but rather a two - dimensional fit to both radius and redshift .
the high average metallicity found by this work for the full sample suggests that it may have a higher fraction of cool cores than is representative of the true population .
the strong evolution ( @xmath178 ) measured by @xcite may represent a redshift - dependent selection , with a higher - than - normal fraction of cool cores at low-@xmath16 and a lower - than - normal fraction high-@xmath16 , consistent with other x - ray surveys . if , instead , single apertures are considered ( e.g. , @xmath179r@xmath38 ) , our analysis and that of @xcite agree well , with measurements of @xmath180 z@xmath5 and @xmath181 z@xmath5 , respectively .
further , if only non - cool cores are considered , this agreement is even better , with @xcite finding @xmath182 z@xmath5 .
the larger disagreement on the _ core - excised _ metallicity when cool core clusters are included may be indicating an inability to fully excise the core for clusters at @xmath8 using data from _ xmm - newton_. for a typical cluster in our sample , at @xmath183 , 0.15r@xmath184 , which is similar in size to the on - axis fwhm of _ xmm - newton_. thus , we would expect the core - excised metallicity measurements of high-@xmath16 cool core clusters based on _ xmm - newton _ data to be biased slightly high .
in general , our results agree well with previous works and , in cases where there is disagreement , it is clear how both selection biases and differences in analyses could conspire to account for any discrepancies .
it is worth repeating that the results presented here also agree well with the indirect measurement of non - evolving icm metallicity from @xcite and @xcite .
these works found azimuthally - uniform metallicities in the outskirts of perseus and virgo , with @xmath185 z@xmath5 and @xmath126 z@xmath5 , respectively ( assuming the same solar abundances ) .
these azimuthally - uniform profiles were taken as evidence for a lack of metallicity evolution , since mixing times are long at large radii .
these works compare favorably with ours , both in terms of the average metallicity in the outskirts of clusters at @xmath52 ( @xmath157 z@xmath5 in this work ) , and in terms of the implied weak evolution of the icm metallicity .
the majority of the data used in this program comes from a series of large programs aimed at obtaining @xmath102000 x - ray counts per cluster . as such , the median number of counts in the 96 _ chandra _ spectra used in this work is @xmath101800 .
there are , however , 7 clusters in this sample with deep ( @xmath4510@xmath186 total counts ) x - ray observations .
these 7 systems account for 82% of the total signal from the _ chandra _ sample of 96 clusters . to test
how sensitive our analysis is to the contributions of a few systems , we determine the best - fitting parameters ( @xmath187 , @xmath95 , @xmath96 ) for the full 96-cluster _ chandra _ sample , as well as a subsample excluding these 7 high - s / n systems . for the full sample of 96
_ chandra_-observed clusters ( grey ) and a subsample excluding the 7 highest signal - to - noise systems .
the exclusion of these 7 systems , which comprise 82% of the total signal , does not have a strong effect on the outcome of this analysis , shifting the normalization and scatter by @xmath101@xmath54 , and leaving the slope relatively unchanged .
to first order , the effect of including a small number of high signal - to - noise measurements is to improve constraints on the measured scatter.,scaledwidth=49.0% ] in figure [ fig : mcmc_s2n ] we show constraints on the best - fitting metallicity evolution model for the full _ chandra _ sample in an aperture of @xmath1 . excluding the 7 highest s / n systems , which comprise 82% of the total x - ray signal , from this analysis has only a marginal effect .
we find shifts of @xmath188 , @xmath189 , and @xmath190 in the normalization , slope , and scatter of the metallicity evolution .
the highest signal - to - noise systems improve the constraints on the scatter by 27% , while providing relatively little improvement to the normalization and slope .
this test confirms that the primary effect of including a small number of precise measurements in this analysis is to improve constraints on the scatter , and that individual systems are not driving the measured evolution .
the results presented here strongly suggest that there has been relatively little change ( @xmath19140% ) in the icm metallicity since @xmath7 , consistent with observations of the outskirts of galaxy clusters , which show remarkable uniformity with azimuth @xcite .
these metals were likely formed in a mix of core - collapse supernovae ( ccsn ) and type ia supernovae the fact that the relative metal abundances are also uniform with azimuth imply that enrichment from both of these sources happened relatively early .
the relative metal abundances in the outskirts of clusters indicate that the enrichment was dominated by ccsn ( see review by * ? ? ?
* ) , with only a small fraction of the metals coming from type ia sne ( 1237% ; * ? ? ?
the data presented here are consistent with a picture in which the bulk of the metals present in the icm today were produced from ccsn during an early stage of rapid star formation , and removed from their host galaxies via starburst - driven winds ( see review by * ? ? ?
* ) . in the centers of cool core clusters , where we observe a peak in the metallicity
, it is thought that the metals are produced via type ia sne in the central bcg @xcite . in the centers of cool core clusters
, we estimate that emission - weighted metallicity has increased by @xmath192% in the intervening time since @xmath7 ( see figure [ fig : mcmc_chandra ] ) .
assuming our best estimate of the metal enrichment rate in the centers of cool core clusters clusters ( table [ table : mcmc ] ) , coupled with no evolution in the outer ( core excised ) metallicity , it would take @xmath1010 gyr to build the observed present - day central excess of metals in the centers of cool core clusters . thus ,
if our maximal likelihood estimate of the core and outer metallicity evolution is correct , we would expect nearly flat metallicity profiles in galaxy clusters at @xmath193 .
one possible avenue for the outskirts of galaxy clusters to become metal - enriched is for powerful radio jets from the central cluster galaxy to push metals out from the high - metallicity core to the low - metallicity outskirts .
such metal - enriching outflows have been predicted in simulations ( e.g. , * ? ? ?
* ) and observed in multiple systems ( e.g. , * ? ? ?
* ) . in the systems studied here , we find that the average metallicity outside of the core at @xmath3 is @xmath126 z@xmath5 and @xmath129 z@xmath5 for cool core and non - cool core clusters respectively ( see table [ table : mcmc ] ) .
if we assume that radio outflows are restricted to cool core clusters ( e.g. , * ? ? ?
* ) , the similarity in the core - excised metallicity suggests that only a small fraction of the total metals can be transported outside of the core , on average .
given the parameters in table [ table : mcmc ] , we can say ( with 95% confidence ) that @xmath19110% of the metallicity in an average cool core cluster is transferred outside of the core by agn feedback any higher fraction would be observable given our uncertainties .
this is consistent with the results presented by @xcite , who showed that fe - rich outflows extended to @xmath100.15r@xmath38 , but not beyond , and with simulations showing that agn feedback can not break self - similarity outside of cluster cores @xcite .
we present an analysis of the icm metallicity in 153 mass - selected galaxy clusters spanning @xmath0 .
this sample of clusters , observed with the _ chandra _ , _ xmm - newton _ , and _ suzaku _ x - ray satellites , provides the best constraints to date on the mean , scatter , and evolution of the icm metallicity .
the main results of this work are summarized as follows : * we find no evidence for evolution in the _ global _
( @xmath194 ) icm metallicity of clusters spanning @xmath0 .
we report that the emission - weighted metallicity has not changed by more than 40% since @xmath7 ( at 95% confidence ) .
* we find a @xmath453@xmath54 difference between the core ( @xmath195 ) metallicities of cool core and non - cool core clusters , consistent with earlier works .
* we find only a weak ( @xmath1033% ) radial increase in metallicity toward the centers of non - cool core clusters , compared to a @xmath45140% increase in the centers of cool core clusters .
* we find no evidence for metallicity evolution in the cores of non - cool core clusters ( @xmath103 ) . *
our best estimate ( @xmath111 ) suggests that the metallicity enhancement observed in the centers of low-@xmath16 cool cores may have been built slowly over the past @xmath1010 gyr . *
we find no evidence for evolution in the core - excised ( @xmath48 ) icm metallicity , and no difference between the core - excised metallicity between cool core and non - cool core clusters .
this implies that radio jets originating in the central cluster galaxy can not move a significant ( @xmath4510% ) fraction of the metals beyond @xmath196 in cool core clusters .
these data imply that the bulk of the metals in the icm were incorporated at early times ( @xmath197 ) , most likely during the peak of star formation at @xmath198 .
this work represents a total commitment of 9.1 ms from three of the most sensitive x - ray telescopes in orbit . as such ,
a significant improvement on this work is unlikely to come from deeper observations with current - generation telescopes , with the exception of potentially confirming a lack of highly - enriched cool cores at @xmath8 . a more precise accounting of the enrichment history of the icm awaits next - generation observatories , such as _ athena _ and _ x - ray surveyor _ , combined with samples of clusters at @xmath198 which should be available with the next generation of sz experiments ( e.g. , spt-3 g ; * ?
much of this work was enabled by generous gto contributions from steve murray , and was in progress at the time of his untimely death in 2015 .
he was a valued member of the center for astrophysics and a strong supporter of spt science - he will be greatly missed by all of us .
m. m. acknowledges support by nasa through contracts go4 - 15122a and go5 - 16141x ( chandra ) , and support by nasa through a hubble fellowship grant hst - hf51308.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , for nasa , under contract nas 5 - 26555 .
e. b. acknowledges support by nasa through contract nnx123ae77 g . the south pole telescope program is supported by the national science foundation through grants ant-0638937 and plr-1248097
this work was partially completed at fermilab , operated by fermi research alliance , llc under contract no .
de - ac02 - 07ch11359 with the united states department of energy .
partial support is also provided by the nsf physics frontier center grant phy-0114422 to the kavli institute of cosmological physics at the university of chicago , the kavli foundation , and the gordon and betty moore foundation .
support for x - ray analysis was provided by nasa through chandra award numbers 12800071 , 12800088 , and 13800883 issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa .
argonne national laboratory s work was supported under the u.s .
department of energy contract de - ac02 - 06ch11357 c. r. acknowledges support from the australian research council?s discovery projects scheme ( dp150103208 ) . , b. a. , ade , p. a. r. , ahmed , z. , et al .
2014 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 9153 , society of photo - optical instrumentation engineers ( spie ) conference series , 1 | we present the results of an x - ray spectral analysis of 153 galaxy clusters observed with the _ chandra _ , _ xmm - newton _ , and _ suzaku _ space telescopes .
these clusters , which span @xmath0 , were drawn from a larger , mass - selected sample of galaxy clusters discovered in the 2500 square degree south pole telescope sunyaev zeldovich ( spt - sz ) survey . with a total combined exposure time of 9.1 ms , these data yield the strongest constraints to date on the evolution of the metal content of the intracluster medium ( icm ) .
we find no evidence for strong evolution in the global ( @xmath1 ) icm metallicity ( @xmath2 ) , with a mean value at @xmath3 of @xmath4 z@xmath5 and a scatter of @xmath6 z@xmath5 .
these results imply that the emission - weighted metallicity has not changed by more than 40% since @xmath7 ( at 95% confidence ) , consistent with the picture of an early ( @xmath8 ) enrichment .
we find , in agreement with previous works , a significantly higher mean value for the metallicity in the centers of cool core clusters versus non - cool core clusters .
we find weak evidence for evolution in the central metallicity of cool core clusters ( @xmath9 ) , which is sufficient to account for this enhanced central metallicity over the past @xmath1010 gyr .
we find no evidence for metallicity evolution outside of the core ( @xmath11 ) , and no significant difference in the core - excised metallicity between cool core and non - cool core clusters .
this suggests that strong radio - mode agn feedback does not significantly alter the distribution of metals at @xmath12 .
given the limitations of current - generation x - ray telescopes in constraining the icm metallicity at @xmath8 , significant improvements on this work will likely require next - generation x - ray missions . |
as known , two nucleons can be ejected from the nucleus by two - body currents due to meson exchanges and delta - isobar excitation . but also , the real or virtual photon can hit , through a one - body current either nucleon of a correlated pair and both nucleons
are then ejected simultaneously from the nucleus .
the role and relevance of these two competing processes can be different in different reactions and kinematics .
it is thus possible to envisage situations where either process is dominant and various specific effects can be disentangled and separately investigated .
this gives ground for studies of short - range correlations ( src ) @xcite in a nucleus by means of the two - nucleon knockout processes .
various theoretical models for cross section calculations have been developed in recent years in order to explore the effects of ground - state @xmath5 correlations on @xmath6 @xcite and @xmath7 @xcite knockout reactions .
it appears from these studies that the most promising tool for investigating src in nuclei is represented by the @xmath8 reaction , where the effect of the two - body currents is less dominant as compared to the @xmath9 and @xmath7 processes .
measurements of the exclusive @xmath0o@xmath10c reaction performed at nikhef in amsterdam @xcite and mami in mainz @xcite have confirmed , in comparison with the theoretical results , the validity of the direct knockout mechanism for transitions to low - lying states of the residual nucleus and have given clear evidence of src for the transition to the ground state of @xmath11c .
one of the main ingredients in the transition matrix elements of exclusive two - nucleon knockout reactions is the two - nucleon overlap function ( tof ) .
the tof contains information on nuclear structure and correlations and allows one to write the cross section in terms of the two - hole spectral function @xcite .
the tof s and their properties are widely reviewed , e.g. , in @xcite .
in @xcite the tof s for the @xmath0o@xmath10c reaction are given by the product of a coupled and fully antisymmetrized pair function of the shell model and a jastrow - type correlation function which incorporates src .
a more sophisticated treatment is used in @xcite , where the tof s are obtained from an explicit calculation of the two - proton spectral function of @xmath0o @xcite , which includes , with some approximations but consistently , both src and long - range correlations ( lrc ) . a different method to calculate the tof s has been suggested in @xcite using the established general relationships connecting tof s with the ground state two - body density matrix ( tdm ) .
the procedure is based on the asymptotic properties of the tof s in coordinate space , when the distance between two of the particles and the center of mass of the remaining core becomes very large .
this procedure can be considered as an extension of the method suggested in @xcite , where the relationship between the one - body density matrix and the one - nucleon overlap function is established .
the latter has been applied @xcite to calculate the one - nucleon overlap functions , spectroscopic factors and to make consistent calculations of the cross sections of different one - nucleon removal reactions , such as @xmath12 , @xmath13 , and @xmath14 @xcite on @xmath0o @xcite and @xmath15ca @xcite , @xmath12 on @xmath16 mg , @xmath17si and @xmath18s @xcite , as well as @xmath19 on @xmath18s @xcite within various correlation methods .
the first aim of the present work is to apply the procedure suggested in @xcite to calculate tof s for @xmath0o using the tdm calculated in @xcite with the jastrow correlation method ( jcm ) , which incorporates the nucleon - nucleon src . as a second aim ,
the resulting two - proton overlap functions are used to calculate the cross section of the @xmath0o@xmath8 reaction for the transition to the @xmath20 ground and the @xmath21 excited states of @xmath11c .
the cross sections are calculated on the basis of the theoretical approach developed in @xcite .
in this section we present shortly the definitions and some properties of the tdm and related quantities in the overlap function representation . the method to extract the tof s from the tdm @xcite used in this work
is also given .
the tdm is defined in coordinate space as : @xmath22 where @xmath23 is the antisymmetric @xmath24-fermion ground state wave function normalized to unity and @xmath25 , @xmath26 are creation and annihilation operators at position @xmath27 .
the coordinate @xmath27 includes the spatial coordinate @xmath28 and spin and isospin variables .
the tdm @xmath29 is trace - normalized to the number of pairs of particles : @xmath30 of direct physical interest is the decomposition of the tdm in terms of the overlap functions between the @xmath24-particle ground state and the eigenstates of the @xmath31-particle systems , since tof s can be probed in exclusive knockout reactions . the tof s are defined as the overlap between the ground state of the target nucleus @xmath32 and a specific state @xmath33 of the residual nucleus ( @xmath34 ) @xcite : @xmath35 inserting a complete set of @xmath31 eigenstates @xmath36 into eq .
( [ tdm ] ) one gets @xmath37 the norm of the two - body overlap functions defines the spectroscopic factor @xmath38 a procedure for obtaining the tof s on the basis of the tdm suggested in @xcite uses the particular asymptotic properties of the tof s . in the case when two like nucleons ( neutrons or protons ) unbound to the rest of the system are simultaneously transferred , the following hyperspherical type of asymptotics is valid for the two - body overlap functions @xcite @xmath39 where @xmath40 and @xmath41 are the magnitudes of the relative and center - of - mass ( c.m . ) coordinates , @xmath42 and @xmath43 , respectively , @xmath44 is the nucleon mass and @xmath45 is the two - nucleon separation energy . for a target nucleus with @xmath46 and for large @xmath47 and @xmath48 a single term with @xmath49 ( corresponding to the smallest two - nucleon separation energy ) of the radial part of the tof @xmath50 can be expressed in terms of the tdm as @xmath51 where @xmath52 is constrained by the experimental values of the two - nucleon separation energy @xmath53 .
the relationship obtained in eq .
( [ p2of ] ) makes it possible to extract tof s with quantum numbers @xmath54 from a given tdm .
the coefficient @xmath55 and the constant @xmath56 can be determined from the asymptotics of @xmath57 .
the procedure described briefly in section 2 has been applied to calculate the two - proton overlap functions in the @xmath0o nucleus for the transition to the @xmath20 ground and the @xmath21 excited states of @xmath11c .
the tdm obtained in @xcite in the framework of the low - order approximation ( loa ) of the jastrow correlation method has been used @xcite .
the tof s for the @xmath58 and @xmath59 states are obtained in the jcm .
the result for @xmath58 state is presented in fig .
[ f:1s0 ] .
it is compared with the uncorrelated tof s obtained applying the same procedure to the uncorrelated tdm .
the notation for the partial waves in our case is @xmath60 .
two - proton overlap functions for the nucleus @xmath0o leading to the @xmath20 ground state of @xmath11c extracted from the jcm ( left ) and uncorrelated ( right ) two - body density matrices .
[ f:1s0],width=453 ] the spectroscopic factors corresponding to the @xmath58 and @xmath59 overlap functions are 0.958 and 0.957 , respectively . as a next step ,
we derive the total tof @xmath61 in terms of a sum over all possible partial components , i.e. @xmath62 we integrate the squared modulus of the total tof in eq .
( [ total ] ) over the angles and sum over the spin variables .
the result can be written in the form ( for the smallest value of @xmath63 ) : @xmath64 where the bar denotes the integration over the angles and summation over the spin variables , and @xmath65 is the radial part of the total tof obtained after the integration and summation . using the asymptotics of @xmath65 at @xmath66 , @xmath67 one can write : @xmath68 the results for the @xmath58 and @xmath59 partial components have a similar behaviour as previous ones , the main difference is that they are somewhat reduced in magnitude .
the spectroscopic factor corresponding to the total tof is equal to unity in the uncorrelated case and 0.965 in the jastrow case .
the jastrow tdm ( including only src ) is not `` rich '' enough to be able to explain realistically transitions to all the excited states of @xmath11c .
therefore , only the transition to the 1@xmath69 state is considered in the present paper as an example of the applicability of the method . in the case of the transition to the @xmath21 excited state of @xmath11c @xmath4-pairs in the states
@xmath70 give main contributions to the process .
the value of the spectroscopic factor e.g. for @xmath71 is 0.967 in the jastrow case and unity in the uncorrelated one .
the tof s obtained from the tdm within the jastrow correlation method have been used to calculate the cross section of the @xmath0o@xmath1c knockout reaction in one - photon exchange approximation @xcite . as an example
, the differential cross section calculated for the transition to the @xmath20 ground state of @xmath11c is shown in fig .
[ f : cs_sup ] for the kinematical setting considered in the experiment performed at mami @xcite .
o@xmath72 reaction as a function of the recoil momentum @xmath73 for the transition to the @xmath20 ground state of @xmath11c in the super - parallel kinematics with @xmath74 mev , @xmath75 mev and @xmath76 mev/@xmath77 .
the curves are obtained with different treatments of the tof : @xmath58 ( dashed line ) and @xmath59 ( dotted line ) as `` independent '' tof s in the jcm in the left panel and as partial components in the right panel .
[ f : cs_sup],width=453 ] the results are compared with the cross sections already shown in @xcite , where the tof is taken from a calculation of the two - proton spectral function ( sf ) @xcite , where a two - step procedure has been adopted to include both src and lrc . in the figure are also shown for a comparison the results obtained with a simpler approach , where the two - nucleon wave function is given by the product of the pair function of the shell model and of a jastrow type central and state independent correlation function ( sm+corr ) .
src are quite strong and even dominant for the @xmath58 state and much weaker for the @xmath59 state .
the role of the isobar current is strongly reduced for @xmath58 @xmath4 knockout , since there the magnetic dipole @xmath78 transition is suppressed @xcite . as a consequence , in the figures
the @xmath58 results are dominated by the one - body current and thus by src , while the @xmath79 current gives the main contribution to @xmath59 @xmath4 knockout .
it can be seen from fig .
[ f : cs_sup ] that the cross section calculated with the jastrow tof for the @xmath58 state is close to the sf and also to the sm+corr results at low values of @xmath73 , up to @xmath80 mev/@xmath77 . for @xmath81
mev/@xmath77 @xmath59 knockout becomes dominant with all the different treatments of the tof .
the results with the @xmath59 tof from the jastrow tdm is however much larger than the sf result and also larger than the sm+corr cross section .
the cross section calculated with the total tof , obtained from the combination of the @xmath58 and @xmath59 partial components , are shown in the right panel of figs .
[ f : cs_sup ] . in both kinematical settings the @xmath58 component dominates at low values of @xmath73 , while the @xmath59 component produces a strong enhancement at high momenta .
although obtained from a calculation of the tdm within the jcm where only src are included , the tof used in our calculations are able to reproduce the main qualitative features which were found in previous theoretical investigations .
this means that the procedure suggested in @xcite to calculate the tof s from the tdm can be applied and exploited in the study of two - nucleon knockout reactions .
the differential cross section calculated for the transition to the @xmath21 state of @xmath11c , at 11.3 mev excitation energy , is shown in fig .
[ f : cs_sup1 + ] in the same kinematical setting already considered for the @xmath20 state in fig . [
f : cs_sup ] . with respect to the other results
, the jastrow tof produces in the super - parallel kinematics a strong enhancement at high momenta , which makes the shape of the cross sections larger and flatter than that with the sf and sm+corr tof s .
o@xmath72 reaction as a function of the recoil momentum @xmath73 for the transition to the @xmath21 excited state of @xmath11c in the same kinematics as in fig .
[ f : cs_sup ] .
[ f : cs_sup1+],width=226 ]
the results of the present work can be summarized as follows : * the two - nucleon overlap functions ( and their norms , the spectroscopic factors ) corresponding to the knockout of two protons from the ground state of @xmath0o and the transition to the ground and @xmath21 excited states of @xmath11c are calculated using the recently established relationship @xcite between the tof s and the tdm . in the calculations the tdm obtained within the jcm @xcite is used .
though only src are accounted for in the jastrow tdm , the results can be considered as a first attempt to use an approach which fulfils the general necessity the tof s to be extracted from theoretically calculated tdm s corresponding to realistic wave functions of the nuclear states .
* the tof s extracted from the jastrow tdm are included in the theoretical approach of @xcite to calculate the cross section of the @xmath0o@xmath1c knockout reaction .
numerical results in different kinematics are compared with the cross sections calculated , within the same theoretical model for the reaction mechanism , with different treatments of the tof , in particular with the more refined approach of @xcite , where the tof s are obtained from a calculation of the two - proton spectral function of @xmath0o where both src and lrc are included .
the cross sections calculated in the present work , where the tof s are extracted from the jastrow tdm , confirm the dominant contribution of @xmath82 @xmath4 knockout at low values of recoil momentum , up to @xmath83 mev/@xmath77 .
the @xmath84 contribution is mainly responsible for the high - momentum part of the cross section at @xmath85 mev/@xmath77 . *
our method is applied in the present work only to the @xmath20 ground and the @xmath21 excited states of @xmath11c .
the main aim was to check the practical application of all steps of the method to a given state of the residual nucleus .
therefore , the results obtained for the @xmath0o@xmath1c reaction , which are able to reproduce the main qualitative features of the cross sections calculated with different treatments of the tof s , can serve as an indication of the reliability of the method , that can be applied to a wider range of situations and , as an alternative to an explicit calculation of the two - hole spectral function , to more refined approaches of the tdm @xcite .
one of the authors ( d. n. k. ) would like to thank the pavia section of the infn for the warm hospitality and for providing the necessary fellowship .
is grateful for support during his stay at the complutense university of madrid to the state secretariat of education and universities of spain ( n / ref .
sab2001 - 0030 ) .
three of the authors ( a.n.a .
, m.v.i . and d.n.k .
) are thankful to the bulgarian national science foundation for partial support under the contracts nos . @xmath86 - 809 and @xmath86 - 905 . | using the relationship between the two - particle overlap functions ( tof s ) and the two - body density matrix ( tdm ) , the tof s for @xmath0o@xmath1c reaction are calculated on the basis of tdm obtained with a jastrow - type approach .
the main contributions of the removal of @xmath2 and @xmath3 @xmath4-pairs from @xmath0o are taken into account in the calculations of the cross sections of the @xmath0o@xmath1c reaction using the jastrow tof s .
the contributions of the one - body and two - body delta currents are considered .
the results are compared with the calculations using tof s from other approaches . |
because of their extremely short kelvin - helmholtz timescales , massive protostars above 8 @xmath3 begin burning h while still accreting and while still buried deep in their natal cloud cores . when the massive star reaches the main - sequence and ionizes its surroundings , it forms an ultracompact h ii ( uchii ) region ( wood & churchwell 1989 ; churchwell 2002 ; hoare et al . 2007 ) , with typical sizes @xmath120.1 pc .
precursors of uchii regions are thought to be dense hot molecular cores ( e.g. , kurtz et al .
2000 ) , but early accretion phases are not as well understood as they are for low - mass stars . with spherical symmetry , radiation pressure on dust would lead to a limit at about 10 @xmath3 for the mass that could be accreted onto a star ( e.g. , wolfire & cassinelli 1987 ; kahn 1974 ; larson 1969 ) , or somewhat higher if momentum of infalling matter can overwhelm the dust sublimation radius . the way in which massive stars overcome this difficulty to reach @xmath840 - 150 @xmath3 is still a topic of current research .
one proposed solution is geometric : in scaling up the traditional view of low - mass star formation that involves accretion disks and collimated outflows ( keto 2003 ; mckee & tan 2003 ) , a massive protostar may circumvent the radiation pressure problem with non - spherical geometry . with an optically thick disk and an optically thin bipolar flow
, the disk can potentially shadow infalling material while radiative luminosity can , in principle , escape out the polar cavities without halting the accretion ( nakano 1989 ; jijina & adams 1996 ; yorke & sonnhalter 2002 ; krumholz et al .
recent numerical simulations by krumholz et al .
( 2009 ) argue that radiation pressure does not halt accretion , allowing the formation of very massive stars and multiple star systems from a common core .
on the other hand , massive stars tend to form in dense clustered environments where the effects of their neighbors may influence the star formation process . in clustered environments ,
coalescence of lower - mass protostars has been suggested as another way to get past the radiation - pressure limit ( e.g. , bonnell et al .
1998 ; bonnell & bate 2002 ; stahler et al .
2000 ; bally & zinnecker 2005 ) .
it is not yet clear from observations if the geometry of a disk plus a bipolar cavity is associated with the formation of the most massive stars up to @xmath8150 @xmath3 ( figer 2005 ; kroupa 2005 ) , while fragmentation in massive disks may also interfere with forming the most massive stars by accretion ( kratter & matzner 2006 ; krumholz et al .
2009 ) .
the massive star - forming complex w49a is an excellent laboratory in which to investigate the processes occurring during the enshrouded early lives of massive stars , despite its large distance of 11.4 kpc ( gwinn et al .
w49a contains one of the richest known clusters of uchii regions distributed across only a few pc , buried inside one of the most massive ( @xmath810@xmath13 @xmath3 ) giant molecular cloud cores in the galaxy , with a cumulative luminosity of well over 10@xmath14 @xmath15 ( see becklin et al . 1973 ; dreher et al . 1984 ;
welch et al . 1987 ; dickel & goss 1990 ; de pree et al .
1997 , 2000 ; smith et al .
its content of massive stars is similar to famous giant h ii regions like 30 doradus , the carina nebula , and ngc 3603 , but it is younger , with most of its o - type stars still embedded deep in the molecular cloud .
wood & churchwell ( 1989 ) noted several different morphological types of uchii regions : spherical / unresolved , cometary , shell , irregular or multiple - peaked , and bipolar ( added more recently in place of core / halo ) , all of which are seen in w49 ( de pree et al .
1997 , 2005 ) .
this compact cluster of 4050 uchii regions in w49 introduced the uchii region `` lifetime problem '' because the sound crossing time of this region exceeds the expected duration of the free expansion phase for any single uchii region ( welch et al .
( this synchronization may also imply that the burst of star formation was triggered by an external agent ; welch et al .
a statistical expression of the uchii region lifetime problem is that the fraction of uchii regions compared to exposed o - type stars in the galaxy suggests that the uchii region lifetime is a significant fraction of the o star lifetime , and is therefore considerbly longer than the free expansion time of the h ii region ( see wood & churchwell 1989 ) .
note , however , that bourke et al .
( 2005 ) suggest that some low - mass protostars may contaminate the wood & churchwell sample , so the implied lifetimes may be shorter ( see also hoare et al .
2007 ) .
about a dozen of these uchii regions are arranged in a remarkable 2 pc diameter ring at the center of w49 ( welch et al .
1987 ) , and the origin of this ring remains unknown .
the brightest member of the ring in the radio continuum is denoted source g , which breaks up into several components at sub - arcsecond resolution ( de pree et al .
1997 , 2000 ) .
source g is also the brightest source in the ring in the thermal - infrared ( ir ) ( smith et al .
2000 ) .
source g is of particular interest because it harbors the most luminous h@xmath0o maser outflow known in the milky way ( gwinn , moran , & reid 1992 ) .
the spatial confusion between radio continuum sources , h@xmath0o masers , and mid - ir emission has not been fully unraveled , however .
this is , in fact , one of the main results of the present study : we identify the driving source of the h@xmath0o maser outflow by its thermal - ir emission from dust , and it is apparently _ not _ an ionized uchii region . on larger scales ,
mid - ir emission is well - correlated with radio continuum emission from uchii regions in w49 ( smith et al .
2000 ) and in other uchii regions ( e.g. , [email protected] ; de buizer et al . 2002 ) , but this is generally not the case for h@xmath0o maser sources , which tend to have point - like ir emission but only weak or undetected radio emission ( tofani et al.1995 ) .
high angular resolution is key , due to w49 s large distance from earth of 11.4 kpc .
there have only been two previous mid - ir imaging studies of source g ( becklin et al .
1973 ; smith et al.2000 ) , and one low - resolution mid - ir spectrum has been obtained ( gillet et al .
none of these had sufficient angular resolution for meaningful comparison with the complex , multi - peaked radio continuum structure .
the mid - ir imaging with the 8 m gemini observatory helps remedy this , with angular resolution comparable to that achieved toward w49 with the vla at centimeter wavelengths .
we present our new mid - ir gemini images in 2 , including a discussion of how we spatially aligned the radio and ir data .
then in 3 we discuss results for sources in the w49 ring other than source g , and in 4 we discuss source g in more detail , comparing our new gemini images to the radio continuum and maser emission .
this allows us to construct a geometric working model for the complex source , and to model the ir emission in order to constrain the physical paramters of the embedded driving source . finally , in 5
we discuss the significance of these observations of source g and corresponding implications for the process of massive star formation and feedback .
we obtained thermal - ir images of w49a at wavelengths of 8.8 , 9.7 , 11.6 , and 18.5 @xmath17 m using michelle on the gemini north telescope .
table 1 lists observation dates and other details .
michelle is the facility mid - ir imager and spectrograph on gemini north , with a 320@xmath5240 pixel si : as ibc array , a pixel scale on the 8 m gemini north telescope of 0@xmath1099 , and a resulting field of view of 31@xmath17@xmath523@xmath18 .
the observations were taken with a @xmath815 north - south chop throw .
w49a is a complex , bright extended source , so some fainter emitting sources contaminated the reference sky positions , leading to negative image residuals in the chop - nod data .
these artifacts were corrected as well as possible using the same technique that we used previously on a large mosaic of the bright inner regions of the orion nebula ( see smith et al .
2004 , 2005 ) .
we subdivided the pixels without interpolation in each of the various images and then shifted and co - added them , to produce a final mosaic image at each of the four wavelengths we observed .
the measured fwhm in our final registered and co - added 812 @xmath17 m images was 0@xmath130@xmath135 , and about 50% larger at 18.5 @xmath17 m , roughly consistent with the expected diffraction limit .
figure [ fig : one ] shows the resulting co - added michelle images .
the observations were performed in service mode and targeted the water maser source w49a / g , but the @xmath830@xmath520field - of - view included a few other sources in the brightest central and western parts of the ring of uchii regions as well ( welch et al.1987 ) .
the observing conditions were not optimal ; while the image quality was good , the sky was non - photometric , and suitable standard star observations were not obtained .
the mid - ir images presented here could therefore not be absolutely flux calibrated and our discussion must be limited mainly to the observed morphology . our previous ground - based observations , however , have already measured the mid - ir fluxes of various sources in our field ( smith et al .
2000 ) , and we incorporate these and other available photometric data in our models described in 4.4 . in principle , the bandpasses of the 8.8 and 11.6 @xmath17 m filters allow for partial contribution of pah ( polycyclic aromatic hydrocarbon ) features . however , a low - resolution 813 @xmath17 m spectrum of source g published by gillet et al .
( 1975 ) shows only smooth continuum emission with deep 9.7 @xmath17 m silicate absorption , so we suspect that our images are not strongly influenced by extended pah emission .
they may , however , contain some extended silicate emission ( see below ) .
.gemini north / michelle observations of w49a [ cols="<,>,^,<",options="header " , ] note : @xmath18 is the inclination angle at which we view the system , and @xmath19 is the outflow opening angle
. figures [ fig : outflow]@xmath20 and [ fig : outflow]@xmath21 compare the 3.6 cm radio continuum morphology to that seen in our 11.6 and 18.5 @xmath17 m gemini images . clearly , the large - scale radio continuum is more elongated east / west , while the diffuse ir emission at 18.5 @xmath17 m seems to be elongated in the opposite direction .
the wings of the x - shaped ir nebulosity that emerge to the ne and se from g : irs1 seem to outline the brightest diffuse radio continuum emission in the eastern part of g ( sources g3 and g4 ) .
it seems likely that the mid - ir emission comes from warm dust at the limb - brightened boundary of an ionized cavity ; our model images discussed in the next section show that this interpretation is plausible .
the large cavity is likely to be a stellar wind - blown bubble ( e.g. , weaver et al . 1977 ) , since it is partly filled with x - ray emission .
the approximate center of the extended x - ray sources detected by tsujimoto et al .
( 2006 ) , which has a similar extent to the radio emission , is marked by the `` x '' in figures [ fig : outflow]@xmath20 and [ fig : outflow]@xmath21 . in that case , the hard x - rays are likely to be the hot stellar wind or jet decelerated in the reverse shock ( see fig .
[ fig : sketch ] , for example ) .
a similar geometry may apply to the western lobe of source g , where source b1 and other low - level diffuse radio emission may define a similar but smaller ionized cavity on the opposite side , which is bounded close to the source by the fainter ir extensions to the sw and nw from g : irs1 .
the western side of the putative bipolar outflow is more complicated , because there are other uchii regions ( sources b , d , and e ) projected along the same line of sight to the redshifted lobe , and because the diffuse radio emission from the redshifted lobe is fainter .
we are therefore less confident about the receding lobe s geometry , although its smaller size makes intuitive physical sense if its environment is denser .
a basic geometric model is sketched in figure [ fig : outflow]@xmath22 .
tilting the eastern ( left in fig . [ fig : outflow]@xmath22 ) cavity toward the observer is favored by several observational properties : 1 ) the eastern lobe is brighter , 2 ) the uchii regions to the west are all obscured in the mid - ir ( smith et al .
2000 ) , 3 ) the morphology of g : irs1 is consistent with the eastern polar axis being tilted toward us , and 4 ) this is the orientation of the @xmath825 km s@xmath10 co outflow observed on similar size scales by scoville et al .
( 1986 ) , where the blueshifted co emission is toward the east and the redshifted co emission is offset to the west .
the co outflow probably traces a dense sheath or cocoon surrounding the ionized cavity .
this orientation for the co outflow and ionized cavities is orthogonal to the water maser outflow ( the small white arrows in fig .
[ fig : outflow]@xmath22 ) .
one solution to this discrepancy , as noted above , may be that the water masers trace dense material in an equatorial torus or envelope , or may originate at the interface between the cavity and disk envelope close to the source ( mac low et al .
a similar orientation is seen in the omc-1 core ( e.g. , plambeck et al .
1982 ; greenhill et al . 1998 ) , and also for s140 irs1 ( hoare 2006 ) . from figure
[ fig : outflow]@xmath22 it is easy to picture how the linear water maser feature located 0@xmath11 west of g : irs1 could arise near the tangent point of the blueshifted outflow cavity . the overall outflow geometry for source g that is pictured in figure [ fig : outflow]@xmath22 is nearly identical to that envisioned earlier by dickel & goss ( 1990 ) , although their picture was based primarily on the different column densities observed toward various lines of sight .
in fact , dickel & goss argued that because of an abrupt change in column density , the edge of the blueshifted part of the outflow must reside between sources g and d , as we have pictured here and have associated with the linear water maser feature .
irs1 is not concident with the ionized uchii regions g2a , g2b , or g1 ( fig .
[ fig : sketch ] ) , showing that there are several centers of massive star formation activity even on size scales of @xmath80.1 pc .
thus , in priciple , we do not necessarily need the water maser outflow to be aligned with the larger cavity because it could simply be due to a different outflow with a different axis orientation .
for example , in omc-1 , the large - scale molecular outflow and cavity traced by mid - ir emission and the shocked h@xmath0 emission from the system of `` fingers '' defines one axis that runs se to nw ( allen & burton 1993 ; kaifu et al .
2000 ; kwan & scoville 1976 ; gezari et al . 1998 ; shuping et al . 2003 ; smith et al .
2005 ) , while source irc2 seems to drive a smaller collimated outflow nearly perpendicular to the large scale flow ( greenhill et al . 1998 , 2003
; bally et al . 2005 ) .
the omc-1 south core has several jets and molecular outflows in various directions that all originate in a @xmath80.01 pc region with multiple mid - ir sources ( smith et al . 2004 ;
zapata et al .
2004 ) . in order to verify that the morphology and orientation of the bipolar cavity are generally correct as sketched in figure [ fig : outflow]@xmath22 , and to constrain physical properties of the illuminating source , we conducted simulations of the dust emission to compare with the observed mid - ir spectral energy distribution ( sed ) and with the morphology observed in our new mid - ir images . as a starting point , we fit the sed of w49/g using the robitaille et al.(2006 , 2007 ) grid of young stellar object ( yso ) models and the online sed fitter . to construct the mid - ir seds
shown in figure [ fig : sed ] , we used the 813 @xmath17 m spectrum from gillet et al .
( 1975 ) , combined with ground - based photometry of source g from smith et al .
( 2000 ) at 12.3 , 12.8 , and 20.6 @xmath17 m , and photometry obtained at 3.6 , 4.5 , and 5.8 @xmath17 m with the infrared array camera ( irac ) on the _ spitzer space telescope _ as part of the glimpse survey ( benjamin et al .
( these additional data were necessary , as our gemini data did not have reliable photometric calibration stars , as noted earlier . )
the ground - based spectral data at 10 @xmath17 m had a large beam ( 22 ) compared to the photometric data ( @xmath82 at 3.6 , 4.5 , 5.8 and 4 at 12.3 , 12.8 , 20.6 ) , so we scaled the flux in the spectrum to match the imaging photometry .
it is therefore possible that the silicate absorption feature at 10 @xmath17 m is partially filled - in by silicate emission from the surrounding heated nebula .
our gemini images at 9.7 and 18.5 @xmath17 m in figure [ fig : one ] do show a halo that is more extended than at 11.6 @xmath17 m ( see also figs .
[ fig : three ] and [ fig : four ] ) , so extended silicate emission may contaminate the larger beam used by gillet et al .
( 1975 ) . with this possible contamination in mind ,
we fit the sed both with and without the 10 @xmath17 m spectral data .
three representative models are discussed below and summarized in table 2 .
the seds fit to both the broad - band data points and 10 @xmath17 m spectral data ( model a in fig .
[ fig : sed ] ) correspond to models that consist of a central source with mass 2535 @xmath3 , and luminosity 12@xmath510@xmath6 @xmath15 , a massive accreting envelope with an equivalent accretion rate @xmath2310@xmath2410@xmath9 @xmath3 yr@xmath10 , and disk mass 00.1 @xmath3 ( the mid - ir emission is dominated by envelope emission , so the sed fits are insensitive to inner disk mass ) .
note that the accretion rate is a model parameter prescribing the envelope density and mass , and is not a direct measure of the true accretion rate .
these well - fit sed models ( defined as having @xmath25 , where @xmath26 is the @xmath27 value of the best fit ) have viewing angles of 85 , and bipolar cavity opening angles of 11 - 18 .
fits that allow the silicate absorption feature to be deeper than indicated by the gillet et al.(1975 ) spectral data ( model b in fig .
[ fig : sed ] ) , on the other hand , give similar stellar properties ( a central stellar source of mass 2535 @xmath3 , and luminosity 23@xmath510@xmath6 @xmath15 ) , an envelope accretion rate @xmath28 5@xmath510@xmath24 @xmath3 yr@xmath10 , no disk , viewing angles of 3090 degrees , and very narrow bipolar cavities ( 23 ) .
this second group of model seds has deeper silicate absorption than the gillet et al .
( 1975 ) spectrum , consistent with the extended silicate emission mentioned above . however , the very narrow opening angle is in conflict with the observed mid - ir morphology .
our high - resolution mid - ir images in figure [ fig : one ] indicate a relatively large bipolar cavity opening angle ( @xmath830 ) , and the radio continuum , co , and water maser data argue for a viewing angle tilted from edge - on ( 4.3 ) .
therefore , we modified the model parameters accordingly in order to fit both the seds and structures in gemini images ( fig .
[ fig : one ] ) with a viewing angle of @xmath860 .
the significantly different result with model c highlights the value of simultaneously fitting both the sed and the spatially resolved morphology .
the resulting model ir spectrum is shown as model c ( black in fig .
[ fig : sed ] ) , and the model images at 11.6 and 18.5 @xmath17 m are compared to the observed gemini data in figure [ fig : model ] .
the qualitative agreement is quite good , especially in the cometary - shaped core .
note that this model did not account for the density gradient in the background cloud that was discussed in 4.1 and 4.3 , so the environment - dependent asymmetry in the large bipolar cavity ( see figure [ fig : outflow ] ) is not addressed by these models .
model c required a more massive star with a stellar mass of @xmath845 @xmath3 , luminosity @xmath83@xmath510@xmath6 @xmath15 , an equivalent envelope accretion rate @xmath29@xmath3010@xmath9 @xmath3 yr@xmath10 , and a large inner hole of 1700 au ( 0@xmath115 ; still unresolved in our images ) , which is larger than the expected dust sublimation radius .
the more massive star in model c has a hotter effective temperature and this fits the steeply rising 110 @xmath17 m spectrum better ( see whitney et al .
2004 ) , as does the large inner hole . also , the hotter and more massive star is in better agreement with the star that is needed to provide the ionizing photon flux of source g : de pree et al .
( 1997 ) estimated a source with and equivalent luminosity and ionizing flux of an o4 v type star to account for the observed radio continuum emission .
however , models of the infalling envelope include a parameter , @xmath31 , which is the centrifugal radius inside of which a disk would form .
the sed is well fit with @xmath325001000 au , consistent with possible disk formation .
the proper interpretation of the inner disk hole suggested by the model fits , within a radius of @xmath81700 au , is unclear because of the uncertainty in fitting a steep sed at short wavelengths in a source with high extinction .
we regard this inner hole as tentative and in need of confirmation , but we discuss it as a possibility in the following section .
the high angular resolution of our new mid - ir images obtained with michelle on gemini south reveal the extended ir morphology of source g , and show that the concentrated peak of source g : irs1 is coincident with a hot molecular core and the origin point of the powerful h@xmath0o maser in w49 .
this allows us to bring together several observational clues that shed light on the nature of the driving source of the outflow .
the key observational parameters to consider are : \1 .
the peak of the mid - ir 8 - 20 @xmath17 m emission is coincident with the origin point of the h@xmath0o maser outflow to within the registration uncertainty of our data . our positional uncertainty of 0@xmath107 is smaller than the size scale of the maser outflow and allows for meaningful comparison with high - resolution radio continuum data .
although this association with the water maser was suspected when the ir source was first discovered by becklin et al .
( 1973 ) , their measurements were made using a single element detector with a 7 diameter aperture , leaving much room for ambiguity given the complex structure of source g. our images show that several other faint sources would have contaminated that aperture as well .
dreher et al . ( 1984 ) compared the ir and radio continuum maps and located the ir source about 2 ( @xmath3320,000 au ) north of the position we adopt here , which is clearly in conflict with our new data and would have precluded it from having any role in the outflow .
the peak of the mid - ir 8 - 20 @xmath17 m emission is _ not _ coincident with any strong compact radio continuum source . although irs1 is surrounded by more diffuse radio continuum emission from source g on large scales ( several times larger than the whole maser outflow ) , the closest peak of compact radio continuum emission that signifies an uchii region is source g2a / b , which is offset more than 0@xmath135 or @xmath84,000 au to the north .
this is beyond our estimated positional uncertainty .
our finding that the compact source g : irs1 is coincident with the origin of the water masers and not radio source g2a / b supports the emerging view ( de buizer et al .
2005 ; tofani et al .
1995 ) that maser sources tend to be associated with mid - ir emission rather than radio continuum emission from uchii regions .
the peak of the mid - ir 8 - 20 @xmath17 m emission is coincident with a hot molecular core revealed by tracers such as ch@xmath34cn , seen in the high resolution mm - wavelength maps made by wilner et al .
( 2001 ) .
this hot core in source g also shows signs of accretion through inverse p cygni absorption profiles in cs ( williams et al .
this accretion signature in the hot core is consistent with the high effective envelope accretion rate in our model fits to the ir sed ( about 10@xmath9 @xmath3 yr@xmath10 ) .
the morphology of source g : irs1 suggests strongly that it is associated with not only the maser outflow , but a much larger bipolar outflow and cavity traced by co emission ( scoville et al .
1986 ) , radio continuum , and x - ray emission .
while the inner source itself shows no compact radio continum emission , diffuse radio continuum emission is present on large scales from the bipolar cavity , implying that some lyman continuum radiation escapes out the poles .
the large scale co outflow and the h@xmath0o maser outflow appear to be orthogonal .
diffuse x - ray emission partly fills the interior cavity of the eastern lobe ( tsujimoto et al .
2006 ) , perhaps marking a reverse shock in the polar outflow .
this combination of observed properties a strong mid - ir source associated with a hot molecular core and maser emission , but not strong radio continuum is quite similar to the hot core in orion that powers the extended bn / kl bipolar molecular outflow .
even the apparent geometry is similar , with the water maser outflow oriented orthogonal to the larger wide - angle bipolar molecular outflow in both cases .
thus , in many ways , source g in w49 appears to be a more luminous analog of the omc-1 outflow , and may provide critical insight to the formation of very massive stars . however , there is also an important difference between these outflows : unlike orion , at least one of w49a / g s outflow cavities is filled with ionized gas seen in the radio continuum ( see fig .
[ fig : outflow ] ) , and appears to have diffuse x - ray emission .
these ionized cavities span a region roughly 8 or almost 0.5 pc across , larger than the bn / kl outflow .
the ionizing flux to sustain source g that was deduced by de pree et al.(1997 ) requires the equivalent of at least six o6-type stars or a single o4 star .
thus , perhaps source g is simply in a somewhat more evolved wind - blown cavity stage than bn / kl , more akin to the bipolar h ii region s106 ( bally et al . 1998 ; smith et al .
interestingly , in the case of s 106 , the smaller of the two bipolar lobes is also more deeply embedded in the parent cloud .
normally , these observations of g : irs1 as a traditional hot molecular core that is still in a stage of active envelope accretion would seem to be at odds with its much larger cavity filled with radio continuum emission and x - ray emission , because one does not expect a hot core to produce significant ionizing flux . the high envelope accretion rate of 10@xmath9 @xmath3 yr@xmath10 that we infer from models of the ir emission would be sufficient to quench the ionizing flux if the source were spherical . one might naturally ask _ what ionizes the gas in the large outflow cavity ? _ we note two possible interpretations that have different implications
. given its bipolar geometry , g : irs1 itself may be the source of ionization for the outflow despite its high envelope accretion rate . as noted earlier , the equivalent ionizing flux of an o4 star needed to account for the radio continuum emission is in good agreement with the massive ( @xmath845 @xmath3 ) and luminous ( @xmath83@xmath510@xmath6 @xmath15 ) star that we infer from models of the ir emission . even though it is a strong mid - ir source inside a hot molecular core
, the embedded massive star may have become hot enough to generate a large uv luminosity that may be able to escape through lower density regions in the polar directions .
our model fits to the sed favor a relatively hot star . the required lyman continuum flux passing through a small polar `` nozzle '' would exceed the dust eddington limit ( radiation force on dust grains ) , but that is part of the basic notion behind using geometry to allow massive stars to form ( e.g. , tan & mckee 2004 ) .
it is unlikely that stellar radiation alone could have been the agent responsible for driving the powerful outflow observed in source g. from observations of the co outflow , scoville et al .
( 1986 ) derive an outflow mass of 138 @xmath3 and a momentum of 3500 @xmath3 km s@xmath10 .
located within @xmath354 of the central star , the outflow dynamical timescale for an average @xmath825 km s@xmath10 outflow speed is roughly 10@xmath11 yr . the momentum supplied by the stellar radiation field , @xmath36 , during this time falls short by a factor of @xmath850 if the stellar luminosity has been constant at the value of 3@xmath510@xmath6 @xmath15 that we infer from radiative transfer models .
similarly , the mass - loss rate required for the mechanical energy of a stellar wind to power the outflow over the same time period , assuming a typical o - star wind speed of 1500 km s@xmath10 , would need to be roughly 2@xmath510@xmath24 @xmath3 yr@xmath10 .
this is @xmath8100 times higher than typical main - sequence o star mass - loss rates in the relevant luminosity range ( see repolust et al.2004 ; smith 2006 ) .
however , such a high mass - loss rate is still only 20% of the effective envelope accretion rate that we infer from models .
therefore , the likely conclusion is that the large - scale co outflow in w49a / g was in fact driven by accretion , at least until very recently , and that the central star in source g is therefore an example of a massive star of @xmath845 @xmath3 that is forming by accretion from a disk .
our models imply that the inner disk has been cleared out to radii of 10002000 au , but this depends on the short - wavelength tail of the ir sed that is also severly affected by extinction , so we regard this result with some caution .
if true , this must have occured recently compared to the @xmath810@xmath11 yr age of the large co outflow .
the h@xmath0o maser outflow , on the other hand , is distributed over a smaller @xmath81 ( 0.06 pc ) region with speeds up to @xmath8100 km s@xmath10 , implying a dynamical age of only @xmath8500 yr .
this interpretation of w49/g directly supports the picture of single massive star formation by accretion advocated by yorke & sonnhalter ( 2002 ) , krumholz et al .
( 2005 ) , and others .
indeed , nearly all the physical parameters we infer for the central star and outflow of source g : irs1 ( the stellar mass and luminosity , the bipolar cavity opening angle , the envelope accretion rate , etc . )
are closely matched in the numerical simulation studied by krumholz et al .
source g : irs1 may be an excellent observational test case for developing models of massive star formation that include the disruptive effects of radiative feedback and stellar winds ( e.g. , krumholz et al . 2005 , 2009 ) . in this context ,
future observations of this source with facilities such as _ alma _ may be of great interest in order to confirm the tentative inner disk hole that we infer .
on the other hand , g : irs1 is in a crowded and observationally complex environment , leaving room for other possibilities .
a second conceivable interpretation is that the outflow cavity from source g is actually ionized by neighboring o stars that already formed within the same cloud core .
the best candidate for this type of symbiotic relationship is the nearby source g2a / b , located just a few thousand au to the north in projection ( see fig .
[ fig : sketch ] ) .
g2a / b appears to be a small bipolar uchii region where the putative polar axis is aligned with that of g : irs1 and its large - scale outflow .
this alignment makes it seem plausible that uv radiation generated by the o star in g2a / b could escape out the poles to ionize the environment of source g seen in diffuse radio continuum emission .
this interpretation highlights the degree to which dense clustered environments are important in massive star formation where feedback from neighboring massive stars may join forces in disrupting their common natal environment . in any case ,
g : irs1 may trace a brief but critical phase in massive star formation , akin to that of the bn / kl outflow in orion .
returning to the uchii region lifetime problem mentioned earlier , we note that among the 4050 uchii regions in w49a , g : irs1 is the only water maser outflow source and large - scale bipolar outflow .
( source a , undetected in the ir , appears to be a fledgeling bipolar cavity and circumstellar torus or disk ; de pree et al .
1997 , 2004 . )
if the uchii region lifetime really is of order 10@xmath6 yr , then it is likely that the maser outflow phase exemplified by source g is transient , lasting no longer than a few thousand years .
this is comparable to the dynamical timescale of the observed outflow in source g with a characteristic size scale of a few arcseconds and the fastest expansion speeds of @xmath8100 km s@xmath10 in the h@xmath0o maser outflow .
similarly , 5001000 yr is the dynamical timescale of the related outflow in the omc-1 core ( e.g. , odell et al .
2008 ) .
this returns our attention to the significance of the possible inner disk hole existing simultaneously with the bipolar outflow .
namely , the short timescale involved suggests that this inner disk is being cleared away at the same time that the h@xmath0o maser outflow is driven .
although direct radiative or stellar wind feedback from the central star has too little momentum to have powered the large scale co outflow over the past 10@xmath11 yr , it may drive the much younger maser outflow .
however , we caution that we only infer the inner disk hole from the absence of short - wavelength ir emission , so it would be interesting if future high - resolution techniques could directly resolve this putative inner disk hole , to catch g : irs1 in the act of destroying its own accretion disk and thereby fixing the central star s final mass .
we thank c. mckee and an anonymous referee for a helpful comments .
was partially supported by nasa through spitzer grants 1264318 and 30348 administered by jpl .
. appreciates continued support from the nsf .
this study has made use of data obtained with _
spitzer_/irac as part of the glimpse legacy survey . | we present high spatial resolution mid - infrared ( ir ) images of the ring of ultracompact h ii regions in w49a obtained at gemini north , allowing us to identify the driving source of its powerful h@xmath0o maser outflow .
these data also confirm our previous report that several radio sources in the ring are undetected in the mid - ir because they are embedded deep inside the cloud core .
we locate the source of the water maser outflow at the position of the compact mid - ir peak of source g ( source g : irs1 ) to within 0@xmath107 .
this ir source is not coincident with any identified compact radio continuum source , but is coincident with a hot molecular core , so we propose that g : irs1 is a hot core driving an outflow analogous to the wide - angle bipolar outflow in omc-1 .
g : irs1 is at the origin of a larger bipolar cavity and co outflow . the water maser outflow is orthogonal to the bipolar co cavity , so the masers probably reside near its waist in the thin cavity walls .
models of the ir emission require a massive protostar with @xmath245 @xmath3 , @xmath43@xmath510@xmath6 l@xmath7 , and an effective envelope accretion rate of @xmath810@xmath9 @xmath3 yr@xmath10 .
feedback from the central star could potentially drive the small - scale h@xmath0o maser outflow , but it has insufficient radiative momentum to have driven the large - scale bipolar co outflow , requiring that this massive star had an active accretion disk over the past 10@xmath11 yr . combined with the spatialy resolved morphology in ir images , g : irs1 in w49 provides compelling evidence for a massive protostar that formed by accreting from a disk , accompanied by a bipolar outflow .
[ firstpage ] h ii regions ism : individual ( w49a ) ism : jets and outflows stars : formation stars : pre main - sequence |
accurate astrometry is crucial in interpreting astronomical observations , allowing the association of sources observed in one region of the electromagnetic spectrum with their counterparts at other wavebands , and thus providing the full spectral energy distribution of an astrophysical object . for galactic objects such as x - ray binaries ,
high - precision astrometry can also provide estimates of the proper motion of the source ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and even the source distance via trigonometric parallax ( e.g. * ? ? ?
* ; * ? ? ?
at present , the highest precision can be achieved in the radio band using the technique of very long baseline interferometry ( vlbi ) . however , the high resolution of vlbi arrays that enables such high - precision astrometry also makes it impractical to use this technique to search for sources without a well - defined error circle from lower - resolution instruments , since the number of pixels to be searched can become unfeasibly large .
black hole x - ray binaries undergo occasional outbursts in which relativistically - moving jet knots are seen to move away from the central binary system .
accurate astrometry is also required to interpret observations of these jets , since knowledge of the position of the central binary allows us to compare the relative motion of approaching and receding jet components and hence to constrain the product @xmath2 , where @xmath3 is the jet speed as a fraction of the speed of light , and @xmath4 is the inclination angle of the jet axis to the line of sight @xcite .
knowledge of the doppler factor in turn helps constrain the energetics of the jets . as they evolve through their duty cycles , black hole x - ray binaries
pass through a range of canonical states , defined by their x - ray spectral and timing characteristics .
these x - ray states , representative of specific conditions within the accretion flow , are very well correlated with the behaviour of the associated outflow in the form of radio jets ( for a detailed discussion of the disc - jet coupling , see * ? ? ?
in the low / hard x - ray spectral state , observed at the beginning and end of an outburst , a steady , compact , flat - spectrum jet exists , which subsequently gives rise to bright , relativistically - moving , discrete ejecta as the x - ray spectrum softens at the peak of the outburst and the source moves through hard and soft intermediate states into a high / soft state . the compact radio jet
is then quenched , with no core radio emission being detected until the source moves back from the high / soft state through the intermediate states into the low / hard state once more .
the quenching of the core radio emission at the peak of the outburst implies that it is difficult to perform accurate astrometry at this time . to determine the position of the central binary system requires high - resolution vlbi observations during the low / hard state , to detect the compact jet originating from the core of the system
this requires the observations to be triggered sufficiently early in the outburst during the rise phase , or following the reverse transition back to the low / hard state at the end of the outburst , although the hysteresis effect @xcite coupled with the radio / x - ray correlation in the low / hard state @xcite implies that the radio emission is less bright in the latter case .
xte j1752 - 223 was discovered on 2009 october 23 by the _ rossi x - ray timing explorer _
( _ rxte _ ) during a routine scan of the galactic bulge region @xcite .
the spectrum and lack of pulsations in the x - ray band led @xcite to suggest that the source was a black hole candidate , a conclusion supported by @xcite and @xcite , who also claimed a black hole mass of 811@xmath5 and a source distance of @xmath6kpc from correlations between x - ray spectral and timing properties .
a 2-mjy radio counterpart was detected by @xcite during the hard state of the system , before its proximity to the sun precluded detailed study by pointed instruments on board x - ray satellites .
as the source emerged from this zone of avoidance , a state transition from the hard x - ray spectral state to the intermediate state was detected @xcite .
the outburst was intensively monitored in the x - ray band by the _ monitor of all - sky x - ray image ( maxi ) _
@xcite , _ rxte _
@xcite , and _ swift _ @xcite satellites , which observed behaviour fairly typical for a black hole x - ray binary system .
the initial radio detection was followed up by further monitoring with the australia telescope compact array ( atca ) , showing that the source had brightened by an order of magnitude by 2010 january 21 . a vlbi imaging campaign with the european vlbi network ( evn ) and very long baseline array ( vlba ) detected moving , decelerating and expanding jet components @xcite , suggesting an initially mildly relativistic jet . while the radio core of the system was not detected ,
its location was inferred to be between the two components detected in the vlba image of 2010 february 26 .
further target of opportunity vlba observations were made in 2010 april following the return of the source to the hard x - ray spectral state . the lack of an accurate core position rendered some of the interpretation of the vlbi images reliant upon various assumptions .
in this paper , we use optical astrometry in conjunction with vlbi radio imaging to determine the true core location , demonstrating how these two techniques can be highly complementary when used in parallel .
we describe our observations in section [ sec : observations ] , present our results in section [ sec : results ] , and re - interpret the existing vlbi observations in light of our newly - determined core position in section [ sec : discussion ] .
we observed the field containing xte j1752 - 223 using the inamori - magellan areal camera and spectrograph ( imacs ) instrument mounted on the 6.5-m baade magellan telescope at las campanas observatory .
we obtained four @xmath7-band and four @xmath8-band images on 2009 november 3 00:10:2600:19:28 utc with exposure times of 510s .
the observing conditions were good with a photometric sky and a seeing of 0.8 and 0.9 in the @xmath7-band and @xmath8-bands , respectively .
only a small section of one of the eight ccds of the imacs mosaic detector was read to sample a @xmath9 field of view at 0.11pixel@xmath10 .
we applied aperture photometry on each of the images using daophot within the image reduction and analysis facility ( iraf ) software package to compute the instrumental magnitudes of the detected stars .
flux calibration of the field was performed by observing two sloan digital sky survey ( sdss ) fields , and differential photometry was used to derive the source flux variability as a function of time .
the photometric results given here are with respect to the three field stars c1c3 shown in fig . [
fig : finder ] .
the average @xmath7-band and @xmath8-band magnitudes of xte j1752 - 223 were @xmath11 and @xmath12mag , respectively . at the end of its 20092010 outburst , xte
j1752 - 223 made a transition back to the hard state late in 2010 march @xcite .
as reported by @xcite , two vlba observations were made following this transition , on 2010 april 25 and 29 ( mjd 55311.46 and 55315.46 ) , under project code bb291 , at an observing frequency of 5ghz .
the data were taken in dual - polarization mode with a recording rate of 512mbps , corresponding to a total bandwidth of 64mhz per polarization .
xte j1752 - 223 was phase referenced to the nearby calibrator source j1755 - 2232 , from the third vlba calibrator survey ( vcs3 ; * ? ? ? * ) .
the position assumed for the calibrator source was 17@xmath1355@xmath1426285 , @xmath153210593 ( j2000 ) , although the current best position is 17@xmath1355@xmath14262845 , @xmath153210616 ( j2000 ) , with an uncertainty of 1.4mas .
since phase referenced positions are measured relative to the position of the calibrator source , the measured target position must therefore be corrected by the 24-mas difference between the assumed and true positions . in an attempt to perform accurate astrometry on the fading core of the system , one further epoch of vlba observations was taken on 2010 june 17 ( mjd 55364.31 ) , at an observing frequency of 8.4ghz , under project code bm346 .
the data were split into eight 8-mhz intermediate frequency pairs , corresponding to a total bandwidth of 64mhz in each of two independent polarizations .
we used the same phase reference source , j1755 - 2232 , but assumed the best known position from the vlba calibrator manual , as given above .
the phase referencing cycle time was 3min ( 2min on the target source , 1min on the calibrator ) , and we substituted every eighth scan on the target source for an observation of a nearby check source from the fifth vlba calibrator survey ( vcs-5 ; * ? ? ? * ) , j1751 - 1950 .
30min at the beginning and end of the observing run were used to observe bright calibrator sources at a wide range of elevations across the entire sky , in order to better calibrate unmodelled clock and tropospheric phase errors using the task delzn from the astronomical image processing system ( aips ) software package @xcite , thereby improving the accuracy of the phase transfer and hence the resulting astrometry .
data reduction was carried out according to standard procedures within aips .
we corrected for small changes to the earth orientation parameters used in the initial correlator model , and also for the ionospheric dispersive delay .
we used the measured system temperatures to calibrate the amplitude scale , and corrected the instrumental phase using the fringe finder source j1733 - 1304 from the international celestial reference frame ( icrf ; * ? ? ?
we performed fringe fitting on the phase reference source j1755 - 2232 , which was then subjected to iterative imaging and self - calibration .
the final image was used as a model for bandpass calibration before transferring bandpass , amplitude and phase solutions to xte j1752 - 223 .
the strong scattering along the line of sight towards the source led us to use only the shortest baselines ( up to 30m@xmath16 , where @xmath17 is the observing frequency ) in making the images , tapering the weights of the visibility data with a gaussian function of full width at half maximum 22.8m@xmath18 to further downweight the long baselines .
we performed astrometry on a 5-second @xmath7-band exposure .
we compared the positions of the stars against entries from the third u.s .
naval observatory ccd astrograph catalog ( ucac3 ; * ? ? ?
an astrometric solution was computed by fitting for the reference point position , the scale and the position angle , considering all the sources that are not saturated and appear stellar and unblended .
we obtain a solution with 0.031@xmath19 root - mean - square ( rms ) residuals from 26 stars ( ucac fit model magnitudes 1416.5 ) well distributed on the @xmath204@xmath21@xmath224@xmath21 image .
the positional accuracy of ucac3 is estimated to be @xmath23@xmath19 on stars of such magnitude @xcite .
in addition , the systematic uncertainty in tying the ucac3 stars to the international celestial reference system ( icrs ) is 5mas @xcite . for the accuracy on our stellar positions we adopt the linear sum of the residuals of the astrometry and the accuracy of the catalogue ( as the latter is probably a systematic error ) :
the resulting positional accuracy at 1@xmath24 is 0.046@xmath19 on both right ascension and declination .
thus our best optical position is @xmath25 this is 533.9mas to the south east of the position of component a ( corrected for the error in the assumed calibrator position ) as identified by @xcite in their vlba observations of 2010 february 11 . in the vlba observations of 2010
april 25 , a point - like radio source was detected with the vlba , at a level of @xmath26mjybeam@xmath10 , at a position consistent with the optical position of xte j1752 - 223 ( fig .
[ fig : image ] ) . after correcting for the error in the assumed calibrator position , the position of the radio source , relative to the vcs-3 calibrator j1755 - 2232 ( with an assumed position of ( j2000 ) 17@xmath1355@xmath14262845 , @xmath153210616 ) , was @xmath27 where the quoted uncertainties are only the statistical errors from the gaussian fitting algorithm ( jmfit within aips ) .
the systematic error due to the cumulative effect of the error in the assumed calibrator position and the 0.76 throw between calibrator and target @xcite is estimated to be 0.51mas .
this vlba position differs from the optically - derived core position by 30.4mas , along a position angle 107@xmath28 e of n. with the derived uncertainty of 46mas in both co - ordinates of the optical position , the radio source is well within the optical error circle .
given the positional agreement , and since xte j1752 - 223 was in the hard spectral state at the time of the radio observations , we conclude that this radio source is indeed a compact jet from the core of the system .
we note that the @xmath29-month time offset between the observations from which the optical and radio positions were determined will likely lead to a small positional shift between epochs owing to the parallax and proper motion signatures of the source , of order 1mas if the source is at a distance of a few kpc and participates in the galactic rotation . furthermore ,
since we see the radio emission from optical depth 1 at each frequency in a compact jet , the true core could be slightly offset along the jet axis from the position of the radio source .
times the rms noise of 0.066mjybeam@xmath10 , where @xmath30 the optically - derived core position is marked with a cross , and the @xmath31 optical error circle of radius @xmath32@xmath19 is shown in white .
co - ordinates have been corrected for the error in assumed calibrator position for the phase reference source .
the radio position lies well within the optical error circle , giving us confidence that we have correctly identified the radio core of the system . ] on 2010 april 29 , we detected a radio source at this same position , but it had faded to @xmath33mjybeam@xmath10 at 5ghz . on 2010
june 17 , the source was only marginally detected , at a level of @xmath34mjybeam@xmath10 at 8.4ghz .
the persistent , variable radio emission at this location also supports our conclusion that this radio source does indeed correspond to a compact jet from the core of the system . in none of these three hard state observations
was the radio core resolved .
our best constraint on the size scale of the compact jet comes from the first , brightest observation on 2010 april 25 , when the source was unresolved down to the beam size of @xmath35mas@xmath36 in position angle 5.5@xmath28 east of north .
this corresponds to a source size of @xmath37au , where @xmath38 is the source distance ( i.e. @xmath39au for the distance of 3.5kpc claimed by * ? ? ?
* ) . following the detection of the radio core , we re - reduced the vlba data of 2010 february 18 , 23 and 26 ( program code bb290 ) presented by @xcite , using a larger image size to search for receding ejecta to the south - east of the core .
no new components were detected , to @xmath40 upper limits of 0.62 , 0.74 , and 0.45mjybeam@xmath10 respectively .
using high - precision optical astrometry , we have been able to locate the core of the x - ray binary system xte j1752 - 223 . from vlbi observations made during the hard spectral state of the system ,
when the radio emission is dominated by a compact , unresolved core jet , we were able to further refine the core position . in light of this new information , we now reanalyze the vlbi data presented by @xcite . the vlba image of 2010 february 26 published by @xcite shows two components , labelled a and b. while component a was detected in all four vlbi images from 2010 february , component b is detected in only this one image , and was interpreted by @xcite as a receding component , leading the authors to infer that the true core of the system lay between components a and b in this image .
our new determination of the true core position implies that both these components are located to the north - west of the core ( table [ tab : components ] ) . from this
, we infer that components a and b must arise from separate ejection events .
while we are unable to reliably pinpoint the exact epoch of ejection of each of those events , we can use x - ray spectral and timing information together with constraints from the integrated radio light curves to obtain a rough estimate for component a. .[tab : components]angular separation from our newly - determined core position of the vlbi components detected by @xcite . [ cols="<,<,^,^,^ " , ] radio flares in x - ray binaries have been linked to the transition from a hard intermediate state ( hims ) to a soft intermediate state ( sims ) during a rapid phase of x - ray spectral softening at the peak of the outburst @xcite .
also associated with this transition are a sharp drop in the integrated rms variability of the x - ray emission and a reduction in the coherence of the associated quasi - periodic oscillations ( qpos ) , from high - coherence type c qpos associated with flat - topped noise in the power spectrum , to lower - coherence type a or type b qpos associated with weak red noise @xcite .
however , we note that @xcite found that while these changes in the variability properties were closely associated with radio ejection events , the association was not exact , such that one could precede the other by up to a few days .
on mjd55215.9 , xte j1752 - 223 was in a hims with a 2.2hz type c qpo @xcite .
the integrated rms variability then decreased from 25 to 18 per cent as the qpo frequency rose to 5.3hz by mjd55217.9 , and by mjd55218.8 , the observed qpos had changed from type c to type a / b .
this suggests that the transition from hims to sims occurred around mjd55218 .
supporting this inference is the bright ( 20mjy ) , flat - spectrum radio emission observed on mjd55217 between 1.2 and 19ghz @xcite .
this represents an increase of the radio brightness by an order of magnitude as compared to the initial 2-mjy radio detection in the rising hard state @xcite , suggesting the onset of a radio flare .
this increase in radio brightness corresponds to an increase of only a factor of @xmath41 in the 1550kev _ swift_/bat and 410kev _ maxi_/gsc x - ray count rates @xcite over the same period , which is not consistent with the radio / x - ray luminosity correlation @xcite found for the compact jets of many hard state black hole candidates ( although note an ever - increasing number of outliers ; e.g. * ? ? ?
while this would tend to support the interpretation of a radio flare on mjd55217 , the flat radio spectrum over more than a decade in frequency instead argues that this radio emission most probably still arises from a compact jet , rather than discrete , optically - thin transient ejecta . the anomalously bright radio emission could then correspond to the period of jet instability known to occur immediately preceding a large radio flare @xcite .
further radio information is available from the atca monitoring , which covered the entire outburst from the initial rising hard state through to the decay back to quiescence .
although a full analysis is beyond the scope of this paper and will be presented by brocksopp et al .
( in prep . ) , we summarize the relevant information here .
the integrated radio light curve shows at least two large flares , followed by a few smaller events .
the last flat - spectrum radio detection was made on mjd55217 , after which the 9-ghz radio flux density dropped to 3.3mjy on mjd55220 , before peaking at 9.9mjy on mjd55221 .
this suggests an initial ejection date between mjd55217 and 55220 .
the second flare was somewhat broader , with the rise phase beginning after mjd55226 and the flare peaking at 10.9mjy at 9ghz on mjd55242 .
the double - peaked light curve supports our conclusion from section [ sec : double ] that the outburst comprised at least two ejection events .
@xcite reported a sharp increase of the soft ( @xmath42kev ) x - ray flux and a decline of the hard ( @xmath43kev ) flux on mjd 55218 , with the emergence of a disk blackbody component in the x - ray spectrum .
this is consistent with the evidence from the x - ray timing and radio observations , suggesting mjd55218 as the likely date of the initial radio ejection event .
@xcite found that uniform deceleration of component a fitted their measurements better than a ballistic model with no deceleration .
however , their suggested deceleration parameters ( fig .
[ fig : angsep ] ) imply an ejection date of mjd55201.8 ( 2010 january 5 ) .
_ rxte _ was still sun - constrained on this date , so no pca observations are available to ascertain the x - ray state of the source , but the _ swift _ and _ maxi _ observations @xcite show that this was significantly prior to the beginning of the x - ray spectral softening , and we deem this unlikely as the true date of ejection . assuming an ejection date of mjd55218 , we are unable to fit the motion of the ejecta with a uniform deceleration model .
our best - fitting model has a reduced-@xmath44 value of 230.9 . either our assumed ejection date is wrong or the uniform deceleration model does not describe the data well .
a plausible alternative could be the scenario outlined by @xcite , whereby a shock wave propagates into the interstellar medium , sweeping up material as it moves , and decelerating such that the late - time behaviour approaches the sedov solution , @xmath45 .
allowing the zero point to float and fitting the measured positions of component a with a simplistic sedov model @xmath46 , where @xmath47 is the angular separation of the component from the core , we find @xmath48mas , @xmath49masd@xmath50 , and @xmath51 , with a reduced @xmath44 value of 0.2 .
although the zero time and position ( @xmath52 and @xmath53 , respectively ) do not correspond to our assumed ejection date and derived core position , when coupled with an initial coasting phase where the ejecta travel purely ballistically ( as proposed by * ? ? ?
* ) , this model appears to provide a plausible fit to the data ( fig .
[ fig : angsep ] ) .
the derived zero time , @xmath52 , is just consistent within error bars with an extrapolation of the expansion of component a @xcite to zero size , which occurs on mjd@xmath54 .
it also coincides with the rise phase of the second flare in the integrated radio light curves , suggesting that the breadth of the second flare could be due to the release of energy as component a begins to decelerate , possibly combined with the ejection of component b. however , if component b were ejected during this second radio flare , its non - detection in the three vlbi observations prior to mjd55253 is surprising .
one explanation could be that the intrinsic jet speed and inclination angle to the line of sight are both high enough for the emission to be doppler - deboosted until the jets have decelerated by sweeping up the surrounding gas .
alternatively , if the observed jet ejecta are shocks , the delay in the appearance of component b could arise from the time taken for the ejecta to either catch up with the slower - moving material ahead of them ( for internal shocks ) or to sweep up and interact with the surrounding gas ( for external shocks ) . with only the one vlbi detection of component
b and the lack of any signatures in the x - ray light curves that might correspond to a second ejection event , we can not further constrain the ejection date of component b , and do not discuss it further . in the absence of a precise ejection date for component a , constraints from the receding components , or from x - ray lightcurves of the ejecta as they decelerate , there are too few constraints to conduct a more meaningful fit to the full model of @xcite . however , the measured vlbi angular separations , the integrated radio light curves and the expansion of component a are all consistent with the deceleration , brightening and lateral expansion of that component close to mjd55232 , as derived from our model fitting .
thus , while we can not definitively verify the proposed scenario , it is certainly plausible
. should the model be applicable , the angular scale for deceleration ( @xmath55@xmath19 ) would be significantly smaller for xte j1752 - 223 than those derived by @xcite for xte j1550 - 564 ( 1217@xmath19 ) and h1743 - 322 ( 3@xmath19 ) . while the distance is not yet well - determined , if xte j1752 - 223 is indeed relatively nearby , as implied by the low hydrogen column towards the source @xcite and as derived via a more model - dependent method ( @xmath6kpc ; * ? ? ? * ) , then the discrepancy in the physical scale of the radius at which deceleration begins would be greater still .
( indicated by the grey vertical dotted line ) , followed by our fitted sedov phase ( section [ sec : deceleration ] ) .
for the assumed ejection date , this appears to provide a better match to the data than either the pure ballistic model or the pure deceleration model . ]
knowing the true core position , we can constrain the quenching factor of the compact jet in the soft state . during the three vlba observations of 2010 february @xcite , the core radio flux was in all cases
@xmath56 , corresponding to a flux density of @xmath57mjybeam@xmath10 . as compared to the 20mjy flux density measured on 2010 january 21 , this represents a core quenching factor of @xmath58
this is consistent with previous lower limits on the quenching factor of the compact core jets in the soft x - ray state ( e.g. * ? ? ?
the reinterpretation of component b of @xcite as an approaching component implies that their estimates of jet speed and inclination angle to the line of sight are no longer valid , since they relied on the ratios of sizes and flux densities of approaching and receding components .
since a re - examination of the vlba data from 2010 february did not show any receding components ( section [ sec : results ] ) , we can not make revised estimates of the physical parameters of the jets with any accuracy .
however , assuming symmetric approaching and receding ejecta , linear expansion of the components ( as fitted for component a of xte j1752 - 223 by * ? ? ?
* ) and a value for the index of the electron energy spectrum , @xmath59 , @xcite demonstrated that the ratio of the flux densities of approaching and receding components in a single image could be used to determine the product @xmath2 . using the measured flux density of 2.2mjy for component a on 2010 february 18 , and taking the @xmath40 upper limit of 0.62mjy for the flux density of the receding component , then assuming a canonical value for optically - thin ejecta of @xmath60
, we find a lower limit @xmath61 , such that @xmath62 and @xmath63 . as a caveat
, we note that if significant deceleration and consequent brightening of the ejecta has occurred prior to this image being taken , as is conceivable ( fig .
[ fig : angsep ] and section [ sec : deceleration ] ) , the flux densities of approaching and receding components will no longer be governed by symmetric ejection , adiabatic expansion and doppler boosting , so this analysis would not be valid .
should the upper limit on @xmath4 be valid however , it would argue that the time delay hypothesis proposed in section [ sec : deceleration ] is more probable than doppler deboosting as an explanation for the delayed appearance of component b. the time baseline between the three radio observations in 2010 april and june is not sufficiently large to measure a significant positional shift between epochs , particularly given the low significance of the latter two detections .
also , despite the 6-month time baseline between the observations from which the optical and radio positions were determined , the uncertainties are sufficiently large that we can not reliably determine the source proper motion between these epochs .
however , our high - precision measurement of the core position provides an initial data point for a measurement of the proper motion should the source undergo repeated outbursts in the future , or should it be sufficiently bright in quiescence to be detected with the new generation of sensitive radio instruments such as the expanded very large array ( evla ) or the high sensitivity array ( hsa ) following the completion of the ongoing bandwidth upgrade at the vlba .
long time baselines and high - precision astrometry are essential for the measurement of x - ray binary proper motions , and given the relative rarity of outbursts in the majority of sources and the relatively small proper motion signals ( typically a few milliarcseconds per year ) , it is important to take astrometric data at every possible opportunity . we encourage future astrometric observations of this source should it return to a bright hard state .
a measurement of the source proper motion would provide information about the formation mechanism of the compact object ( e.g. * ? ? ?
* ; * ? ? ?
we have demonstrated how accurate optical astrometry can be used in conjunction with high - resolution vlbi imaging in the hard state of an x - ray binary to locate the compact core of the system to sub - milliarcsecond accuracy .
our determination of the position of the core of xte j1752 - 223 mandated a re - interpretation of the published vlbi data from the 20092010 outburst of the source .
the two components detected in the vlba image of 2010 february 26 are both on the same side of the core , implying that there were at least two ejection events during the outburst , with the first likely occurring close to the transition from the hard intermediate state to the soft intermediate state on mjd55218 . with this extra constraint on the motion of component a , its angular separation from the core as a function of time can no longer be fit with a uniform deceleration model
. a plausible explanation could be ballistic motion out to some radius after which rapid deceleration occurred as the ejecta swept up the surrounding interstellar medium , and the jets transitioned to a sedov phase .
we constrain the quenching factor of the compact core radio jet in the soft state to be @xmath58 .
no receding ejecta are detected in any of the vlbi observations , and from the upper limit to their flux density , we constrain the product of jet speed and inclination angle , @xmath2 , to be @xmath64 .
pgj acknowledges support from the netherlands organisation for scientific research via a vidi grant .
the vlba is operated by the national radio astronomy observatory , a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
this paper includes data gathered with the 6.5-m magellan telescopes located at the las campanas observatory , chile .
this research has made use of nasa s astrophysics data system .
belloni t. , homan j. , casella p. , van der klis m. , nespoli e. , lewin w. h. g. , miller j. m. , mndez m. , 2005 , a&a , 440 , 207 bradshaw c. f. , fomalont e. b. , geldzahler b. j. , 1999 , apj , 512 , l121 brocksopp c. , corbel s. , tzioumis t. , fender r. , 2009 , the astronomer s telegram , 2278 brocksopp c. , corbel s. , tzioumis a. , fender r. , coriat m. , 2010 , the astronomer s telegram , 2400 curran p. a. , maccarone t. j. , casella p. , evans p. a. , landsman w. , krimm h. a. , brocksopp c. , still m. , 2011 , mnras , 410 , 541 dhawan v. , mirabel i. f. , rib m. , rodrigues i. , 2007 , apj , 668 , 430 fender r. et al . , 1999 ,
apj , 519 , l165 fender r. p. , belloni t. m. , gallo e. , 2004 , mnras , 355 , 1105 fender r. p. , homan j. , belloni t. m. , 2009 , mnras , 396 , 1370 gallo e. , fender r. p. , pooley g. g. , 2003 , mnras , 344 , 60 gallo e. , 2007 , in antonelli , l. a. et al .
, aip conf .
924 , the multicolored landscape of compact objects and their explosive origins .
phys . , new york , p.715 greisen e. w. 2003 , in information handling in astronomy : historical vistas , ed .
a. heck ( dordrecht : kluwer ) , 109 hao j. f. , zhang s. n. , 2009 , apj , 702 , 1648 homan j. , 2010 , the astronomer s telegram , 2387 kovalev y. y. , petrov l. , fomalont e. b. , gordon d. , 2007 , aj , 133 , 1236 ma c. et al . , 1998 , aj , 116 , 516 maccarone t. j. , coppi p. s. , 2003 , mnras , 338 , 189 markwardt c. b. et al .
, 2009a , the astronomer s telegram , 2258 markwardt c. b. , barthelmy s. d. , evans p. a. , swank j. h. , 2009b , the astronomer s telegram , 2261 miller - jones j. c. a. , blundell k. m. , duffy p. , 2004 ,
apj , 603 , l21 miller - jones j. c. a. , jonker p. g. , nelemans g. , portegies zwart s. , dhawan v. , brisken w. , gallo e. , rupen m. p. , 2009a ,
mnras , 394 , 1440 miller - jones j. c. a. , jonker p. g. , dhawan v. , brisken w. , rupen m. p. , nelemans g. , gallo e. , 2009b , apj , 706 , l230 mirabel i. f. , rodrigues i. , 2003 , science , 300 , 1119 mirabel i. f. , rodrguez l. f. , 1994 , nature , 371 , 46 mirabel i. f. , dhawan v. , mignani r. p. , rodrigues i. , guglielmetti f. , 2001 , nature , 413 , 139 mirabel i. f. , mignani r. , rodrigues i. , combi j. a. , rodrguez , l. f. , guglielmetti f. , 2002 , a&a , 395 , 595 muoz - darias t. , motta s. , pawar d. , belloni t. m. , campana s. , bhattacharya d. , 2010a , mnras , 404 , l94 muoz darias t. , motta s. , belloni t. , homan j. , 2010b , the astronomer s telegram , 2518 nakahira s. et al . , 2010 , pasj , 62 , l27 negoro h. et al . , 2010 , the astronomer s telegram , 2396 , 1 petrov l. , kovalev y. y. , fomalont e. , gordon d. , 2005 , aj , 129 , 1163 pradel n. , charlot p. , lestrade j .- f . , 2006 , a&a , 452 , 1099 shaposhnikov n. , 2010 , the astronomer s telegram , 2391 shaposhnikov n. , markwardt c. , swank j. , krimm h. , 2010 , apj , 723 , 1817 wang x. y. , dai z. g. , lu t. , 2003 , apj , 592 , 347 yang j. , brocksopp c. , corbel s. , paragi z. , tzioumis t. , fender r. p. , 2010 ,
mnras , 409 , l64 zacharias n. et al . , 2010 ,
aj , 139 , 2184 | using high - precision astrometric optical observations from the walter baade magellan telescope in conjunction with high - resolution very long baseline interferometric ( vlbi ) radio imaging with the very long baseline array ( vlba ) , we have located the core of the x - ray binary system xte j1752 - 223 .
compact radio emission from the core was detected following the state transition from the soft to the hard x - ray state .
its position to the south - east of all previously - detected jet components mandated a re - analysis of the existing vlbi data .
our analysis suggests that the outburst comprised at least two ejection events prior to 2010 february 26 .
no radio - emitting components were detected to the south - east of the core at any epoch , suggesting that the receding jets were doppler - deboosted below our sensitivity limit . from the ratio of the brightness of the detected components to the measured upper limits for the receding ejecta , we constrain the jet speed @xmath0 and the inclination angle to the line of sight @xmath1 . assuming that the initial ejection event occurred at the transition from the hard intermediate state to the soft intermediate state , an initial period of ballistic motion followed by a sedov phase ( i.e. self - similar adiabatic expansion ) appears to fit the motion of the ejecta better than a uniform deceleration model .
the accurate core location can provide a long time baseline for a future proper motion determination should the system show a second outburst , providing insights into the formation mechanism of the compact object .
[ firstpage ] x - rays : binaries radio continuum : stars stars : individual ( xte j1752 - 223 ) ism : jets and outflows |
thermodynamics of micro- and nano - scale systems exhibits distinctly different features from that of large systems due to influence of large thermal fluctuations @xcite .
typical energy changes are of the order of thermal energy per degree of freedom and consequently thermodynamics has to be modified at micro - scale .
these systems can be theoretically analyzed using stochastic thermodynamics .
exchange of energy between the particle and its surroundings becomes stochastic and yet one can clearly formulate the notion of work , heat and entropy production for a given microscopic trajectory of the particle @xcite .
recently obtained exact results ( fluctuation theorems@xcite ) put constraints on the distributions of the above mentioned stochastic quantities and are valid for systems driven far from equilibrium .
these theorems transform thermodynamic inequalities into equalities .
this area has become even more interesting with the development of experimental techniques .
using single - colloidal particle experiments several new key concepts have been verified .
information to energy conversion and validation of generalized jarzynski equality @xcite , landauer erasure principle @xcite , universal features in the energetics of symmetry breaking @xcite are to name a few .
micron sized heat engines have been experimentally realized by optically controlled motion of trapped colloidal particle @xcite .
there are several extensive studies on single bath nano - machines e.g. , information machines ( that can produce work using available information ) @xcite and molecular motors / thermal ratchets @xcite .
molecular motors are omnipresent in cellular as well as tissue level of many living organisms .
they are efficient enough to extract energy from a highly fluctuating environment and to convert it into mechanical work for cellular and / or intra - cellular logistics @xcite .
the conversion of energy into mechanical work repeatedly along a cycle working within multiple thermal bath is almost every where in our day - to - day life , spanning huge range of length and time scales .
for example , starting from high pressure steam locomotives , to a _ drinking bird _ @xcite and even biochemical reaction pathways for cellular respiration mechanism @xcite producing useful energy from nutrients - everywhere energy is transformed into mechanical work . similarly , by reversing the cycle
, we see that in various processes mechanical work is used to transfer heat from a cold source to a warm sink with an objective to cool down the cold source further ( refrigerator ) or to heat up the warm sink warmer ( heat pump ) .
though it is very important to study the work - energy ( or , vice versa ) conversion in all relevant scales , due to lack of experimental techniques for micro or nano world , it is relatively well explored in macro scale .
heat engines and refrigerators at nano - scale is a subject of current study @xcite .
detailed theoretical treatment of _ carnot - type _ micro heat engine , involving both quasistatic and non - quasistatic ( i.e. , finite cycle time ) features , have also been documented @xcite .
these features reveal the fundamental differences between micro and macro heat engines due to thermal fluctuations , reflected in the distributions of various thermodynamic quantities ( e.g. , work , heat exchange , efficiency etc . ) . unlike macro heat engines
, it has been shown that the system can work as a heat engine if the ratio of hot and cold bath temperatures is larger than a critical value @xcite . moreover , in non - quasistatic regime , the system works as a heat engine for cycle times larger than a critical value .
both the thresholds depend on the system parameters . fluctuations in thermodynamic quantities including efficiency of the system calculated over a large number of trajectories are significant not only in non - quasistatic regime but also in quasistatic regime , which is in clear contrast to the macro engines . several trajectories violate typical expectations from the second law of thermodynamics @xcite .
the non self - averaging nature of fluctuations in stochastic efficiency and other quantities requires detailed understanding of full probability distributions as opposed to the average behavior @xcite .
large deviation properties of such distributions are recently under theoretical investigations @xcite .
research on fluctuation relations for heat engines @xcite are being pursued .
it may also be noted that , for some of the heat engines studied so far , one may or may not recover carnot result in the quasistatic regime .
however , fluctuation theorems provide a bound on efficiency of an engine valid for any finite time cycle . recently
, novel theoretical approaches to capture the statistical properties of stochastic efficiency of micro heat engines and mesoscopic thermoelectric engines with broken time - reversal symmetry are being developed , particularly at long time limit @xcite , claiming universal properties of the large deviation function related to the statistics of stochastic efficiency . to our knowledge , so far , no such study exists for micro - refrigerators . in this article
we will focus on _ carnot - type _ single - particle refrigerator and its stochastic features . the refrigeration protocol used here is similar to the micro heat engine protocol used in @xcite but is running backward in time .
we believe that our model system is experimentally realizable using the technique already being used for micro heat engines .
we find new insights into far from equilibrium features of the concerned system .
for example , a major outcome of the present study is the variety of different modes of operation for such systems under the protocol . in fig .
[ modes ] we describe all the modes of operations which are thermodynamically possible for a system that works cyclically between a hot and a cold heat bath .
other four possibilities for heat exchanges and work are ruled out due to violation of first and second laws of thermodynamics . ) and a cold ( with temperature @xmath0 ) thermal baths : * a * : takes heat @xmath1 from hot bath and converts it partially into work @xmath2 and supplies the rest in the form of heat @xmath3 to the cold bath ( heat engine ) . *
b * : takes heat @xmath1 from the hot bath and with the help of work @xmath2 on it , supplies heat @xmath3 to the cold bath .
* c * : converts work @xmath2 on it , to heats @xmath1 and @xmath3 that enter into the hot and cold baths respectively.*b * and * c * are called heater of type - ii ( or , heater - ii ) and heater of type - i ( or , heater - i ) respectively .
* d * : takes heat @xmath3 from the cold bath and with the help of work @xmath2 on it , supplies heat @xmath1 to the hot bath ( refrigerator ) . when work is being done on ( by ) the system , it is positive ( negative ) .
when the system releases heat it is positive and negative otherwise.,width=359 ] our system , under the refrigeration protocol , exhibits all the modes depending on its cycle time and on the ratio of hot and cold bath temperatures , which can be controlled in experiments @xcite . throughout the paper ,
we consider the heat going from the system to the bath to be positive and that going from the bath to the system to be negative .
similarly , the work done on the system is taken as positive and that done by the system is considered negative .
we have also studied fluctuation in stochastic efficiency / coefficient of performance ( cop ) of engine / refrigerator and have shown that their fluctuations dominate the mean values .
probability distributions have been calculated both in the underdamped and overdamped regimes .
these distributions exhibit power law tails with non - universal exponents .
reliability of our system operating as a refrigerator / engine is also reported .
we first describe the model and the protocol in the next section . in section * 3 * we discuss the essentials of stochastic thermodynamics . in section * 4 * and * 5 * , we explain the results for quasistatic and non - quasistatic behaviour of micro - refrigerators in underdamped as well as in overdamped cases in detail .
finally we conclude by focusing on the major differences between micro and macro machines manifested by the model system .
we consider a brownian particle of mass _ m _ , position @xmath4 and velocity @xmath5 moving in a medium with friction coefficient @xmath6 . the particle is trapped in a harmonic potential whose stiffness constant @xmath7 is periodically changing with time period @xmath8 .
the underdamped equation of motion of the particle in contact with a thermal bath of temperature @xmath9 is given by @xcite @xmath10 where fluctuation dissipation relation between noise strength , temperature of the bath and friction coefficient is maintained . throughout the paper boltzmann constant @xmath11
is set to unity . in overdamped limit ,
the equation of motion reduces to @xmath12 the noises from the bath @xmath13 are gaussian distributed with zero mean and are delta correlated , i.e. , @xmath14 and @xmath15 .
the time dependent stiffness of the trap is given by , @xmath16 in the first step , the system undergoes an isothermal compression while in contact with the hot bath of temperature @xmath17 for @xmath18 . during this step ,
work is done on the system . at @xmath19 ,
the stiffness is instantaneously changed from @xmath20 to @xmath21 ( adiabatic expansion ) and simultaneously the system is connected to the cold bath at temperature @xmath0 , disconnecting it from hot bath . in the third step , isothermal expansion is carried out where the stiffness is changed from @xmath21 to @xmath22 during @xmath23 . during this step work
is extracted from the system .
finally the stiffness is changed by adiabatic compression from @xmath22 to @xmath21 . in this step
the system is detached from the cold bath and coupled with the hot bath again . a schematic diagram of the protocol is shown in fig . [ prot]a . and @xmath0 respectively .
black lines are the jumps of the protocol during the adiabatic steps . b. schematic representation for a cyclic process of stochastic refrigerator operating between two reservoirs kept at temperatures @xmath17 and @xmath0 .
the cycle consists of two isothermal steps and two adiabatic steps according to the time varying protocol @xmath7 .
the violet line denotes a one dimensional potential @xmath24 and the filled region denotes the corresponding steady state distribution.,width=396 ] one should note here that instantaneous jumps are not the only way to implement adiabaticity .
there are examples of smooth protocols where one changes the temperature of the bath and the stiffness of the harmonic trap smoothly over time such that at long time limit the phase space volume remains constant , providing no average heat dissipation and consequently establishing the adiabatic steps@xcite . in this paper
we restrict ourselves to this carnot - type refrigerator protocol .
this is the reverse of the protocol that has been studied in @xcite in context of single particle stochastic heat engine .
there are other important protocols , for example stirling - type protocol used in earlier studies @xcite , where the adiabatic steps are absent .
the comparative study between stochastic thermodynamics of micro machines driven by different protocols will be focused elsewhere .
our current work concerns stochastic thermodynamics of time periodic steady state ( tpss ) of the system .
it is well known fact @xcite that when a langevin system is subjected to a time periodic force , after initial transient ( duration of which depends on the details of system parameter ) , the system will settle to a tpss .
in tpss , the joint phase space distribution @xmath25 is time periodic , i.e. , @xmath26 .
before investigating further , we briefly recapitulate the principle of stochastic thermodynamics . using eq .
[ eom1 ] , change in internal energy , @xmath27 , for an underdamped langevin system can be written as @xmath28 where @xmath29 , @xmath30 and @xmath31 .
the integrals in the second step are performed according to stratonovich s rule . along a fluctuating trajectory of the particle , identifying the random variables @xmath32 , @xmath33 and @xmath34 respectively as total internal energy of the particle , thermodynamic work done on the particle and heat dissipated to the bath , one obtains first law of stochastic thermodynamics , valid for a single trajectory ( eq .
[ 1stlaw ] ) .
note that while @xmath33 and @xmath34 depend on the entire trajectory of the particle , @xmath35 depends only on its initial and final points .
similarly we can also write the first law in case of overdamped dynamics where @xmath36 . for the system concerned here , work done on the particle in isothermal steps for a typical cycle in tpss
is given by , @xmath37 the first integration in the r.h.s of above expression is defined along the path of isothermal compression and the second one is along isothermal expansion . by definition , in adiabatic steps @xmath38 and therefore work in adiabatic steps during this cycle in tpss is @xmath39 + \big[u\left({\tau}\right)-u\left({\tau}^{-}\right)\big ] \label{ad}\end{aligned}\ ] ] the first term of the above expression is the work done on the system along the path of adiabatic expansion while the second one is that along adiabatic compression . simulating the dynamics of the particle via eq .
[ eom1 ] , we numerically calculate @xmath40 and @xmath41 to obtain the total work for a cycle as , @xmath42 running the dynamics for a large number of cycles @xmath43 in tpss , we calculate work , averaged over all cycles , @xmath44 using eq.[1stlaw ] we calculate heat transfers in isothermal steps between the particle and the baths ( @xmath34 ) . heat transfer along isothermal compression is given by @xmath45 \label{heat1}\ ] ] and that along isothermal expansion is given by , @xmath46 .
\label{heat2}\ ] ] since the heat transfers in adiabatic steps are zero , the total heat transfer in a cycle is @xmath47 as before , running the cycle repeatedly for @xmath48 times , we define the average heats , @xmath49 these non - quasistatic results should asymptotically match with quasistatic results which can be recovered as @xmath50 with @xmath51 and @xmath52 . here
@xmath53s are the heat exchanges and @xmath54 is the total work in a cycle in quasistatic limit .
similarly , one can also calculate the change of total energy of the particle in a cycle , @xmath27 and its average over @xmath48 cycles followed by its quasistatic counterpart @xmath55 , taking @xmath56 limit .
therefore , for the system concerned here , all the relevant thermodynamic variables involved in the first law of stochastic energetics and their path - averages , both in non - quasistatic as well as quasistatic regime , can be calculated by simulating the underdamped and overdamped dynamics of the system .
quasistatic behavior of our system serves as an important benchmark for the simulation as it can also be obtained analytically . for any machine , quantification of its performance ( e.g. , efficiency )
is one of the important issues . in case of microscopic engine @xcite
, it has been shown that , in quasistatic regime average efficiency of the micro heat engine depends only on @xmath57 .
similarly , in case of micro refrigerator we will determine average cop in quasistatic regime .
we define cop for a cycle as @xmath58 for a trajectory of the system .
it fluctuates randomly from cycle to cycle . for large number of cycles ,
average stochastic cop is defined as , @xmath59 due to fluctuations of @xmath34 and @xmath33 in both non - quasistatic and quasistatic regime , @xmath60 , where @xmath61 is the conventional definition of average cop . for finite time cycle , using fluctuation theorems @xcite , it can be shown that @xmath62 .
note that this is valid for all cycle times .
however , no such bound exists for @xmath63 . being equipped with stochastic thermodynamics of our system , in the following sections , we explore quasistatic as well as non - quasistatic behavior of our system .
now we calculate thermodynamic quantities like , average work , heat exchanges and internal energy changes for different sub steps of a cycle in quasistatic limit . in this limit ,
the duration of the protocol is longer than all other time scales . during isothermal processes ,
though the protocol is being changed , the system instantaneously adjusts itself to the equilibrium corresponding to the value of the protocol at that instant .
hence the work done along any isothermal process is the free energy difference between the final and initial state . in the first step , i.e. , in isothermal compression , the work done on the system connected to the hot bath will be @xmath64 at @xmath65 , the system is in equilibrium with hot bath and the corresponding distribution is given by @xmath66.\ ] ] here , @xmath67 is the normalization constant .
the second step being instantaneous , there will be no heat dissipation and the work done on the particle is the instantaneous change of its internal energy , given by @xmath68 similar to the first step , the work done on the particle in the third step is , @xmath69 at @xmath70 , the system will be in equilibrium with the cold bath .
hence , the corresponding distribution will be @xmath71,\ ] ] with @xmath72 .
therefore , in the last step , i.e. , in adiabatic compression process the average work done on the particle is @xmath73 the average work done during the full cycle in the quasistatic process is @xmath74 as @xmath75 , we find that @xmath54 is always positive for any temperature difference .
this implies that , in the quasistatic limit , on an average , work is always done on the system . to obtain heat absorption from the cold bath ( @xmath3 )
first we have to calculate internal energy change along the third step .
the average internal energy at @xmath76 is @xmath77 .
since the system is in equilibrium with the cold bath at @xmath78 , the average internal energy will be @xmath0 .
this leads to the change in internal energy in the third step , @xmath79 .
using first law ( eq . [ 1stlaw ] ) we obtain average heat dissipated to the cold bath , @xmath80 similarly we can obtain the average heat , transferred to hot bath , @xmath81 though @xmath82 is always positive , @xmath1 and @xmath3 can take negative as well as positive values depending on the ratio between hot and cold bath temperatures . when @xmath83 , @xmath3 is negative and is positive elsewhere .
when @xmath84 , @xmath1 is positive and otherwise negative . the system will act as a refrigerator only when @xmath85 and @xmath86 i.e. , @xmath87 in the quasistatic limit .
for @xmath88 system will act as a stochastic heater i ( fig . [ modes ] ) where both @xmath1 and @xmath3 are positive .
when @xmath89 heat flows from hot bath to the system @xmath90 , but the system releases heat to the cold bath thereby acting as a stochastic heater of type ii .
it is evident that even after using a carnot - type refrigerating protocol one can obtain single particle refrigerator as well as heaters depending on @xmath17 and @xmath0 in quasistatic limit .
finally the cop of the system as a refrigerator in quasistatic limit is , @xmath91 it should be noted here that even in the quasistatic limit , cop in our system is much smaller than the carnot limit ( @xmath92 ) as shown in fig .
[ ud ] . , in red line ) and
the corresponding carnot bound ( @xmath93 , in green line ) as a function of @xmath57 .
blue filled circles denotes the values ( obtained from numerical simulations ) of @xmath94 at different values of @xmath57 for large cycle time.,width=226 ] we now turn our attention to the results obtained by simulation in non - quasistatic regime .
we evolve the system using discretised langevin dynamics with time step @xmath95 in the underdamped as well as overdamped limit [ eq . [ eom1 ] and eq .
[ eom2 ] ] .
the system is driven by time periodic protocol [ eq . [ protocol1 ] ] .
we follow heun s method @xcite .
we have set @xmath96 and @xmath97 .
all the physical quantities are in dimensionless form . throughout the paper
we have fixed @xmath98 and @xmath99 .
we make sure that , after the initial transient regime(@xmath100 cycle time ) , the system settles to a tpss i.e. , @xmath101 .
we use the same discretisation to numerically calculate the thermodynamic variables along a trajectory .
we consider at least @xmath102 cycles of operations and thermodynamical quantities are averaged over all these cycles . for each @xmath103 pair
, we calculate @xmath2 , @xmath104 , and @xmath105 with @xmath106 and a phase diagram [ fig .
[ phase1 ] ] for the operational modes of the system is obtained . from the phase diagram
it is evident that when @xmath107 , for any @xmath17 , @xmath108 , @xmath109 and @xmath110 . in this mode , on an average the system heats up the cold bath .
therefore it is a stochastic heater of type ii .
trajectories.,width=226 ] for large @xmath111 and high @xmath17 , @xmath112 and @xmath2 goes into the system whereas @xmath113 comes out of the system to heat up the cold bath , providing us again heater - ii . at large @xmath8 , but with intermediate @xmath17 , we obtain heater - i where @xmath112 flips its direction of flow to heat up the hot bath keeping everything else the same as before .
if we reduce @xmath17 further , we obtain stochastic refrigerator where , even @xmath113 flips its direction and flows from cold bath to the system while others remain same as heater - i .
interestingly for @xmath114 and @xmath115 , we observe that on an average , the system can do work , taking heat from the hot bath and releasing it partially to the cold bath .
it implies that even if we use a carnot - type refrigerating protocol , deep in non - quasistatic regime , for a particular range of @xmath8 and @xmath17 , our system can behave as a stochastic heat engine .
therefore , if we consider both non - quasistatic and quasistatic regimes , the system can act in all thermodynamically possible modes , depending on the values of @xmath116 .
it should be mentioned here that for large @xmath8 , phase boundaries are correctly predicted by the quasistatic limits calculated analytically .
the details of the phases and phase boundaries are depicted in fig .
[ phase1 ] .
note that , in the phase diagram crossing any phase boundary once , implies flipping the direction of any one of the @xmath112 , @xmath113 and @xmath2 .
one can further explore different modes of operations in fig .
[ w ] where we have plotted @xmath2 , @xmath117 and @xmath118 with respect to cycle time @xmath8 , for different @xmath17 . if temperature difference between two reservoirs is small , @xmath2 is always positive and @xmath117 is negative for small @xmath8 but positive for large @xmath8 .
the behavior of @xmath118 with @xmath8 is opposite to that of @xmath117 .
thus for small temperature differences , if we vary @xmath8 from small to large values , we obtain heater - ii first and then heater - i and finally a refrigerator .
for higher temperature differences , with increasing @xmath8 , @xmath117 becomes negative and @xmath118 becomes positive .
@xmath2 shows a remarkable non - monotonic ( convex ) behavior within a range of @xmath8 where it shows a dip that can even be below zero for higher temperature differences .
therefore , in this regime of temperatures , though we obtain heaters with very small and large @xmath8 , our system acts as a stochastic heat engine within a particular range of intermediate @xmath8 .
@xmath2 , @xmath117 and @xmath118 ultimately saturates with @xmath8 to values as found in quasistatic calculation .
we note from fig .
[ w]a that @xmath2 becomes independent of @xmath17 for a given @xmath0 at two different @xmath8 within which @xmath2 is non monotonic . versus @xmath8 for different @xmath17
, b : @xmath117 versus @xmath8 , c : @xmath118 versus @xmath8 , d : @xmath119 versus @xmath8 .
red : @xmath120 , green : @xmath121 : , blue : @xmath122 , cyan : @xmath123 ) .
two crossing points of all the curves in a are marked by broken black circles in a. at these points @xmath2 is independent of @xmath17.,width=377 ] in fig .
[ w]d , we have plotted average total entropy production ( @xmath124 ) per period .
it shows that initially @xmath119 increases with @xmath8 . above @xmath125 , @xmath119 tends to saturate to quasistatic value .
this sets a characteristic time scale in our system .
the characteristic time scale that determine the relaxation time of our system is @xmath126
. the entropy production increases monotonically with @xmath17 . in fig .
[ w]a , we see two special points where the value of @xmath2 for particular values of @xmath8 are same even though the hot bath temperature @xmath17 is changed .
we explore the probability distribution @xmath127 of work at those two values of @xmath8 for different @xmath17 as shown in fig .
[ wdist_un]a and [ wdist_un]b .
it is clear from the distributions that the fluctuations increases with increasing @xmath17 , keeping the mean unchanged .
@xmath127 is asymmetric with positive mean .
finite weight for @xmath128 arises from realisations which does nt act as refrigerator .
( red ) , @xmath121 ( green ) , @xmath122 ( blue ) , @xmath123 ( cyan ) .
the cycle time for a and b are @xmath129 ( first crossing point ) and @xmath130 ( second crossing point ) respectively .
the distributions are obtained by considering @xmath102 trajectories.,width=453,height=151 ] in the refrigerator regime of the phase diagram , most of the trajectories followed by the system maintains the refrigerating condition : @xmath131 , @xmath132 and @xmath133 .
however , there are considerable number of trajectories where the system does not follow this condition .
this number reduces with increasing @xmath8 but still remains finite even in quasistatic regime . due to fluctuations , the difference between @xmath134 and @xmath135 ( fig .
[ coeff ] a ) is prominent even at large @xmath8 .
the saturation values of @xmath134 and @xmath135 ( as given in table i ) are far below than the carnot limit ( @xmath136 in our case ) .
it should be noted that the saturation value of @xmath135 matches with our quasistatic result .
( black , broken ) and @xmath137 ( red , continuous ) with @xmath8 at @xmath17=0.12 .
the average quantities are obtained by considering @xmath102 trajectories.,width=226 ] @xmath138 , the distribution of stochastic cop , is shown in fig .
[ coeff - dist ] for two different cycle times , @xmath125 and @xmath139 ( both for @xmath120 ) .
@xmath140 can take values from @xmath141 to @xmath142 .
there are considerable number of realizations where @xmath140 can be negative or can occur beyond @xmath93 .
moreover , we have noticed ( see inset of fig . [ coeff - dist ] ) that the tail of @xmath138 decays as a power law ( @xmath143 ) for several decades .
the exponent @xmath144 depends on the system and protocol parameters ( e.g. , @xmath8 ) and are given in figure captions .
so far to our knowledge , no study exist for the probability distribution and its power law tails for stochastic cop .
it is not clear whether this power extends indefinitely for large values of @xmath140 .
however , given our numerical data we can calculate variance which is finite . for both distributions , variance @xmath145 ( see table [ table1 ] for more details ) .
thus @xmath63 ceases to be a good physical variable and one has to resort to the study of full probability distribution . from fig .
[ coeff - dist ] it is also apparent that , @xmath138 becomes sharper with increasing @xmath8 . from the phase diagram
we know that , with @xmath146 , our system is closer to the quasistatic limit . therefore , even in quasistatic regime , fluctuations in @xmath140 are significant . at @xmath147 ( red , continuous ) and @xmath148 ( black , broken ) at @xmath120 .
inset shows positive tails of the distributions plotted in log scale .
they behave as @xmath143 . for @xmath147 @xmath149 - 1.897 @xmath150 0.004 ( slope of the red , continuous line ) and for @xmath148 @xmath149 -1.875 @xmath150 0.004 ( slope of the black , broken line ) .
, width=226 ] in fig .
[ coeff_dist100_un ] distribution of cop is plotted at @xmath139 for different numbers of cycle(@xmath151 ) .
as @xmath151 increases , distribution becomes sharper .
the tail on the negative side disappears thereby indicating the fact that realisations leading to non - refrigerator mode reduces drastically .
hence our system works as a reliable refrigerator in large @xmath151 limit .
tail on the positive side also gets suppressed as @xmath151 increases .
the standard deviation decreases as @xmath151 increases . for @xmath152 , standard deviation becomes less than the mean value .
this implies that for large @xmath151 , @xmath153 becomes a well defined physical quantity .
average cop does not approach carnot limit for large @xmath151 . for different cycle numbers(@xmath151 ) at @xmath139 .
, width=226 ] in fig .
[ ref10]a and fig .
[ ref10]b we have plotted @xmath154 , the joint probability distributions of @xmath33 and @xmath155 over @xmath102 trajectories , at @xmath120 for two different cycle times @xmath125 and @xmath156 . for a particular cycle , @xmath140 can be positive when @xmath155 and @xmath33 has opposite signs , but only fourth quadrant of these joint probability distribution represent a refrigerator operation . in fig .
[ ref10]a , only @xmath157 trajectories follow the refrigerating condition whereas in fig .
[ ref10]b , it is enhanced to @xmath158 . therefore , the system will behave more reliably as a refrigerator if we run it close to quasistatic limit .
table i summarizes important results .
and @xmath155 at @xmath159 , b : distribution of @xmath33 and @xmath155 at @xmath160.,width=453 ] [ cols="^,^,^,^,^,^,^,^ " , ] we have plotted the distribution of stochastic efficiency in fig .
[ oeffdist ] at @xmath161 and @xmath123 , a typical point of the phase diagram where our system works as stochastic heat engine .
we found that only for @xmath162 cycles , the system works as a heat engine .
thus , even in overdamped regime , the engine operation is unreliable .
both @xmath138 and @xmath163 show power - law decays in their tails for several decades .
the exponent values are given in the figure captions .
the exponent of @xmath138 is different from that we have obtained in case of underdamped dynamics .
at @xmath161 at @xmath164 . here
@xmath165 and @xmath166 .
inset : the positive tail of the distribution behaves as a power law with exponent -2.155 @xmath150 0.014 ( slope of the straight line).,width=226 ]
we will conclude by focusing on the main results to point out the crucial differences between macro and micro refrigerators . in the macro - world
, idealized refrigerator is like an engine running backward in time , i.e. , using the work to extract heat from the cold bath and dumping the heat at hot bath .
we have shown that such a simple picture is not valid for single particle heat engine at nano - scale .
the engine running in a reverse order may produce refrigerator , stochastic heater of type i and ii as well as unreliable heat engine . for our model of study , fluctuations in stochastic cop and efficiency dominate their mean values even in the quasistatic case .
this implies that in such a situation mean is not a good physical variable and one must study the behaviour of full probability distributions which in all our cases contain power law behaviour in their tails with varied exponents .
cop and efficiency may exceed carnot bound . however , averaged cop and efficiency , as defined here , are bounded by the carnot value .
this can readily be shown by fluctuation theorem of heat engines . in every cycle , a macro thermal machine under refrigerating protocol acts as a refrigerator .
this is not true for micro refrigerators .
we know that the trajectory dependent work , heat and cop are randomly distributed variables over the cycles of a micro refrigerator
. the cycle time can be large or small , but there will always be a considerable number of cycles where the system will not run as a refrigerator due to fluctuations of these thermodynamic variables .
this makes the micro refrigerator unreliable in comparison to macro refrigerators .
this reliability , quantified as a fraction of cycles running as a refrigerator , increases as we increase the cycle time @xmath8 .
the obtained heat engine is most unreliable .
the phase diagram and thermodynamic quantities describing the micro refrigerators as well as the micro heat engines are crucially protocol dependent .
we have noticed in our other studies ( will be published elsewhere ) , that the micro refrigerators / heat engines , running under stirling protocol ( which is very similar to carnot protocol but devoid of the adiabatic jumps ) , though produce qualitatively similar phase diagrams as carnot - protocol , are less fluctuating in large cycle time limit .
moreover , only in the underdamped case in a small parameter space heat engine operation is possible .
the adiabatic jumps are key players behind the unreliability of micro heat engines in @xcite or refrigerators described in earlier sections . only in the overdamped quasistatic limit
we have shown that for our given specific protocol , refrigerator can work in a reversible mode of operation provided @xmath167 .
this clear from zero entropy production for this case ( e.g. , see fig .
[ ow ] d ) .
power output is found to be an useful optimization criterion for finite time heat engines . on the other hand , many optimization criteria were proposed in case of refrigerator . widely used target function was introduced in @xcite by giving equal footing for cop and cooling rate , known as @xmath168-criterion @xcite . in view of new findings in brownian refrigerators , it will be of immense interest to study these quantities in our system .
it may be noted that our studied engine or refrigerator do not reach carnot limit in the efficiency / cop in the quasistatic regime .
the engine is not a microscopic version of carnot engine .
there are other engines which reproduces carnot limit for efficiency in the quasistatic regime @xcite .
studies on differences in these two types of engines will be useful .
work along this direction is in progress .
authors a.s . , s.r . and
p.s.p thanks edgar roldan for several useful discussions .
a.m.j . thanks department of science and technology , india for financial support .
10 c. bustamante , j. liphardt and f. ritort , phys . today * 58 * ( 2005 ) 43 .
k. sekimoto , prog .
* 130 * ( 1998 ) 17 .
u. seifert , phys .
* 95 * ( 2005 ) 040602 .
d. dan and a. m. jayannavar , physica a. * 345 * ( 2005 ) 404 .
s. saikia , r. roy and a. m. jayannavar , phys .
a. * 369 * ( 2007 ) 367 .
p. jop , a. petrosyan and s. ciiberto , europhys . lett .
* 81 * ( 2008 ) 50005 .
a. saha , s. lahiri and a. m. jayannavar , phys .
e * 80 * ( 2009 ) 011117 .
k. sekimoto , stochastic energetics ( springer , 2010 ) .
s. saikia , a. m. jayannavar and m. c. mahato , phys .
e * 83 * ( 2011 ) 061121 .
f. ritort , poincare seminar * 2 * , ( 2003 ) 193 .
r. j. harris and g. m. schutz , j. stat .
( 2007 ) p07020 . c. jarzynski , annu .
rev . condens .
matter phys .
* 2 * ( 2010 ) 329 .
u. seifert , rep .
* 75 * ( 2012 ) 126001 .
s. lahiri and a. m. jayannavar , eur .
j. b * 87 * ( 2014 ) 141 .
s. lahiri and a. m. jayannavar , arxiv:1402.5588 s. toyabe , t. sagawa , m. ueda , e. muneyuki and m. sano , nature physics * 6 * ( 2010 ) 988 .
a. b@xmath169rut , a. arakelyan , a. petrosyan , s. ciliberto , r. dillenschneider and e. lutz , nature * 183 * ( 2012 ) 487 .
a. schadschneider , d. chowdhury and k. nishinari , stochastic transport in complex systems : from molecules to vehicles ( elsevier 2011 ) .
ralph lorenz , am .
* 74 * ( 2006 ) 677 .
m.yoshida , e. muneyuki and t. hisabori , nature reviews molecular cell biology * 2 * ( 2001 ) 669 - 77 marisa r. buchakjian1 and sally kornbluth , nature reviews molecular cell biology * 11 * ( 2010 ) 715 - 727 c. van den broeck and r. kawai , phys .
* 96 * ( 2006 ) 210601 . n. nakagawa and t. s. komatsu , europhys
* 75 * ( 2006 ) 22 .
r. marathe , a. m. jayannavar and a. dhar , phys . rev.e
* 75 * ( 2007 ) 030103 .
r. benjamin and r. kawai , phys . rev.e * 77 * ( 2008 ) 051132 .
t. schmiedl and u. seifert , europhys . lett .
* 81 * ( 2008 ) 2003 .
j. hoppenau , m. niemann , and a. engel , phys .
e * 87 * ( 2013 ) 062127 .
z. c. tu , phys .
e * 89 * ( 2014 ) 052148 .
v. holubec , j. stat .
( 2014 ) p05022 .
l. cerino , a. puglisi , and a. vulpiani , arxiv:1503.01434 g. verley , t. willaert , c. van den broeck and m. esposito , 2014 phys . rev .
e * 90 * 052145 .
jian - hua jiang , bijay kumar agarwalla , and dvira segal , 2015 phys .
. lett . * 115 * 04060 .
g. verley , t. willaert , c. van den broeck , and m. esposito , nat . commun . * 5 * ( 2014 ) 4721 .
t. r. gingrich , g. m. rotskoff , s. vaikuntanathan and p. l. geissler , 2014 new .
* 16 * 102003 . m. polettini , g. verley and m. esposito , 2015 phys . rev . lett . * 114 * 050601 .
k. proesmans and c. van den broeck , arxiv:1503.00497 g. verley , t. willaert , c. van den broeck , and m. esposito , phys .
e 90,052145 ( 2014 ) .
k. proesmans , b. cleuren , and c. v. d. broeck , arxiv preprint arxiv:1411.3531(2014 ) .
m. esposito , m. a. ochoa , and m. galperin , arxiv preprint arxiv:1501.03232 ( 2015 ) .
z. yan and j. chen , j. phys d appl .
* 23 * ( 1990 ) 136 .
s. velasco , j. m. m. roco , a. medina and a. calvo hern@xmath171nndez , phys .
* 78 * ( 1997 ) 3241 . c. de
tom@xmath171s , a. calvo hern@xmath171nndez and j.m.m .
roco , phys .
e * 85 * ( 2012 ) 010104 . c. de tom@xmath171s , j. m. m. roco , a. calvo hern@xmath171nndez , y. wang and z. c. tu , phys . rev .
e * 87 * ( 2013 ) 012105 .
y. yuan , r wang , j. he , y. ma and j. wang , phys.rev.e * 90 * ( 2014 ) 052151 . | we present a detailed study of a brownian particle driven by carnot - type refrigerating protocol operating between two thermal baths .
both the underdamped as well as the overdamped limits are investigated .
the particle is in a harmonic potential with time - periodic strength that drives the system cyclically between the baths .
each cycle consists of two isothermal steps at different temperatures and two adiabatic steps connecting them . besides working as a stochastic refrigerator ,
it is shown analytically that in the quasistatic regime the system can also act as stochastic heater , depending on the bath temperatures .
interestingly , in non - quasistatic regime , our system can even work as a stochastic heat engine for certain range of cycle time and bath temperatures .
we show that the operation of this engine is not reliable .
the fluctuations of stochastic efficiency / coefficient of performance ( cop ) dominate their mean values .
their distributions show power law tails , however the exponents are not universal .
our study reveals that microscopic machines are not the microscopic equivalent of the macroscopic machines that we come across in our daily life .
we find that there is no one to one correspondence between the performance of our system under engine protocol and its reverse .
stochastic particle dynamics , fluctuations , stochastic processes [ 2 ] [ 3]phys .
rev .
lett . *
# 1 * , # 2 ( # 3 ) [ 3]phys .
rev .
e * # 1 * , # 2 ( # 3 ) [ 3]phys .
rev . a * # 1 * , # 2 ( # 3 ) [ 3]j .
stat .
mech . *
# 1 * , # 2 ( # 3 ) [ 3]eur .
phys .
j. b * # 1 * , # 2 ( # 3 ) [ 3]rev . mod .
phys .
* # 1 * , # 2 ( # 3 ) [ 3]europhys .
lett . *
# 1 * , # 2 ( # 3 ) [ 3]j .
stat .
phys . *
# 1 * , # 2 ( # 3 ) [ 3]prog .
theor .
phys .
suppl . *
# 1 * , # 2 ( # 3 ) [ 3]physics today * # 1 * , # 2 ( # 3 ) [ 3]adv .
phys . *
# 1 * , # 2 ( # 3 ) [ 3]j .
. condens .
matter * # 1 * , # 2 ( # 3 ) [ 3]j .
phys . a * # 1 * , # 2 ( # 3 ) [ 3]new .
j phys . *
# 1 * , # 2 ( # 3 ) [ 3]applied phys .
lett . *
# 1 * , # 2 ( # 3 ) |
galaxy bias , the spatial clustering strength of galaxies relative to the matter field , encodes important information about the connection between the luminous baryons of galaxies and the dark matter ( dm ) halos that host them . the galaxy bias @xmath2 , defined as the ratio between the galaxy and matter correlation functions ( cfs ) @xmath3 , reflects both the clustering of galaxy - scale dm halos and how luminous galaxies populate halos of a given mass .
a self - consistent connection between the luminosity function ( lf ) , angular cf , and mass - to - light ratio of galaxies can be developed ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
cf estimates of bias , which require large samples of galaxies spanning a wide range of angular separations , are remarkably successful at low - redshift ( e.g. , * ? ? ?
* ; * ? ? ?
angular cf measurements on smaller samples have been performed for lyman break galaxies ( lbgs ) @xcite and lyman-@xmath4 emitters @xcite over a range of redshifts .
these studies have extended high - significance @xmath5 detections of bias out to @xmath6 . at the highest redshifts ( @xmath7 ) , limited samples have made significant detections of galaxy bias difficult .
@xcite used a @xmath8deg@xmath9 area observed by the ukirt infrared deep sky survey ultra deep survey and the subaru xmm - newton survey to measure the clustering of @xmath10 lbgs with @xmath11ab ( @xmath12 ) , but most of their systems lie at redshifts @xmath13 .
@xcite reported a measurement of bias for @xmath14-dropout galaxies in great observatories origins deep survey ( goods ) ( @xmath15 ) but no detection of clustering in the ultra deep field .
a study of @xmath16-dropout galaxies in the subaru deep field and goods - n by @xcite indicated that @xmath17 galaxies may be strongly clustered , but measuring bias from only 22 candidates is challenging .
given the difficulty of measuring clustering of high - redshift galaxies from the angular cf , an exploration of complementary approaches is warranted . as a salient example , @xcite used a counts - in - cells analysis @xcite to measure the clustering of redshift @xmath18 lbgs from statistical fluctuations in their number densities .
having obtained spectroscopic redshifts for @xmath19 lbgs in six fields , they split their samples into redshift slices and calculated the variance in galaxy counts in rectangular volumes .
the bias was then measured from the excess dispersion in the number counts beyond the poisson variance . in this _
letter _ , we present calculations that demonstrate the bias of high - redshift ( @xmath1 ) galaxies can also be measured from field - to - field dispersion in galaxy counts from photometric dropout samples .
specifically , we use monte carlo simulations to demonstrate that pure parallel programs with the hubble space telescope ( hst ) that obtain large numbers ( @xmath20 ) of statistically - independent samples of @xmath21 galaxies with the advanced camera for surveys ( acs ) or wide field camera 3 ( wfc3 ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) can provide a significant measurement of high - redshift galaxy bias . throughout
, we adopt the cosmological parameter values determined by the @xmath22-year _ wilkinson microwave anisotropy probe _ joint analysis @xcite .
consider a number @xmath23 of telescope pointings on the sky . at the location @xmath24 of the @xmath14-th field the expected number density of galaxies is @xmath25,\ ] ] where @xmath26 is the mean average galaxy number density , @xmath2 is the bias , and @xmath27 is the matter overdensity of the field .
the matter variance @xmath28 , averaged over many survey volumes @xmath29 , is @xmath30 where @xmath31 is the growth factor at redshift @xmath16 , @xmath32 is the matter power spectrum ( we use the @xcite transfer function ) , and @xmath33 is the fourier transform of the rectangular volume @xmath34 with angular area @xmath35 and depth @xmath36 at distance @xmath37 . the expected dispersion in galaxy counts owing to cosmic variance is simply @xmath38 ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. with a sample of @xmath23 fields , the distribution of galaxy counts @xmath39 , with mean @xmath40 per field , will have a dispersion @xmath41 that includes contributions from both cosmic and poisson variance .
the bias can be estimated as @xmath42 below , we demonstrate that this estimate can be used to measure the bias of high - redshift galaxies from multiple widely - separated hst pointings .
the accuracy of this method will depend on @xmath23 and the relative size of the cosmic and poisson variances , and we now explore these effects through monte carlo simulations .
to simulate high - redshift galaxy number counts , we require a model for the distribution of galaxy counts for each field in our sample .
below , we describe our model distribution and our monte carlo realizations of this distribution that act as our model observational samples .
the initial cosmological linear overdensity field is a gaussian random field with dispersion @xmath43 given by equation [ eqn : dm_variance ] . as the density field evolves owing to gravitation , the overdensity distribution changes to maintain the positivity condition on the density ( @xmath44 ) .
the shape of the quasi - linear overdensity distribution can be approximated by a lognormal @xcite .
the corresponding galaxy overdensity distribution will be lognormal , written as @xmath45,\ ] ] where @xmath46 and @xmath47 .
the expected cosmic variance is @xmath48 ( see below ) , so the quasi - linear distribution is appropriate . at each location @xmath24 , the number of galaxies observed will be a poisson - sampled random variate of equation [ eqn : galaxy_overdensity_pdf ] .
the probability of observing @xmath39 galaxies in fields of volume @xmath29 with cosmic variance @xmath49 , given the expected number @xmath50 , is @xmath51^{n}e^{-(1+{\delta_{\mathrm{g}}}){\bar{n}}}\ ] ] @xcite .
our simulations of the bias measurement consist of drawing discrete random samples from this distribution , as described below .
measuring the bias from the dispersion in galaxy counts requires an observational program suited to probing many high - redshift galaxy samples .
our presented method is motivated by hst programs ( programs 9488 , pi ratnatunga , 9575 and 9584 , pi sparks , 11700 , pi trenti , and 11702 , pi yan ) that use pure parallel observations with acs or wfc3 to probe @xmath52 galaxy populations .
while the total numbers of long - duration and multiple - passband observations in the hst archive are few ( @xmath53 ) , we expect that similar pure parallel observations will be obtained by hst in ongoing and future cycles .
we will therefore model future hst pure parallel programs that obtain acs @xmath14- and @xmath16-band data , which would allow for @xmath14-dropout selection at @xmath54 to @xmath55ab ( single orbit exposure , extrapolated from * ? ? ?
additionally , as an example comparison with wfc3 observations , we will model the yan wfc3 program that can select @xmath56 dropouts at @xmath57 , and reach @xmath58ab magnitude sensitivity ( single orbit exposure ) , but our method can be adapted for use with any wfc3 filter strategy .
we adopt these survey designs , summarized in table [ table : surveys ] , as our points of comparison for modeling observational measurements of high - redshift galaxy bias .
we wish to simulate the number of high - redshift galaxies these pure parallel surveys will discover in each field .
the mean abundance @xmath40 per field is determined by integrating a @xcite lf over the volume probed by each pointing . for modeling the acs observations of @xmath14-dropouts at @xmath59
, we adopt the measured @xmath54 lf parameters @xmath60 , @xmath61ab , and @xmath62 @xcite . with these values
, we expect @xmath63 per @xmath64 arcmin@xmath9 acs field in the magnitude range @xmath65 . to approximate the conversion between @xmath66 apparent and restframe @xmath67 used by @xcite , we use @xmath68 . for modeling the wfc3 observations the @xmath17 lf is uncertain at @xmath69 , but if we adopt the measured @xmath17 lf parameters @xmath60 , @xmath70ab , and @xmath71 @xcite , we expect @xmath72 per @xmath73 arcmin@xmath9
wfc3 field in the uv absolute magnitude range @xmath74 .
our results will be similar for other estimates of the @xmath17 lf ( e.g. , * ? ? ?
* ; * ? ? ?
galaxy counts for each field can be generated by discretely sampling the equation [ eqn : galaxy_count_pdf ] , with a variance @xmath75 and bias @xmath2 determined using the abundance - matching methodology presented in @xcite .
this method assumes the galaxies occupy dm halos with an abundance and bias appropriate for their high redshifts @xcite .
lcccccccccc @xmath76 & acs & 4 & @xmath55 & @xmath77 & -20.24 & -1.74 & 4.29 & 0.37 & 0.48 & 40 + @xmath78 & wfc3 & 4 & @xmath79 & @xmath77 & -19.91 & -1.77 & 1.04 & 0.50 & 0.98 & 130 the left panel of figure [ fig : bias_constraint ] shows @xmath80 realizations of the galaxy bias measurement for @xmath54 galaxies with @xmath65ab across @xmath81 ( light to dark blue ) widely - separated acs fields .
for this example magnitude range at redshifts @xmath82 , the mean number of galaxies per 11.3 arcmin@xmath9 is @xmath63 and @xmath83 .
the model bias calculated for this galaxy abundance is @xmath84 ( vertical dashed line ) .
the mean galaxy count per acs field @xmath40 and the sample variance @xmath85 are calculated from each realization .
the `` observed '' bias is then estimated using equation [ eqn : bias_measure ] , histogrammed , and normalized such that the area under each curve is unity .
for each @xmath23-sized sample the mean expected measured bias is accurate , with @xmath86 for @xmath87 decreasing to @xmath88 for @xmath89 . the expected uncertainty @xmath90 in the measured bias decreases from @xmath91 for @xmath87 to @xmath92 for @xmath89 , improving as @xmath93 .
when @xmath23 is small , the measured dispersion @xmath85 can be less than @xmath40 , providing @xmath94 in equation [ eqn : bias_measure ] .
such catastrophic failures happen infrequently for acs @xmath14-dropout samples ( @xmath95 for @xmath87 and @xmath96 for @xmath97 , shown as the @xmath98 bin ) .
given these results , we suggest that using the dispersion in galaxy counts determined from acs photometric samples in uncorrelated fields to measure bias will be a novel method for learning about the spatial clustering of @xmath54 galaxies .
the bias measurement significance increases from @xmath99 with @xmath100 to @xmath101 with @xmath89 fields for @xmath14-dropouts with magnitudes near @xmath102 , providing a method for measuring bias that increases in sensitivity with an increasing number of independent samples .
the right panel of figure [ fig : bias_constraint ] shows @xmath80 realizations of the bias measurement for @xmath17 galaxies with absolute magnitudes @xmath74ab across @xmath81 ( light to dark blue ) widely - separated wfc3 fields . for this magnitude range at redshifts @xmath103 ,
the expected mean number of galaxies per 4.7 arcmin@xmath9 is @xmath72 and @xmath104 .
the model bias calculated for this galaxy abundance is @xmath105 ( vertical dashed line ) .
the observed mean galaxy count per wfc3 field @xmath40 and the sample variance @xmath85 are calculated from each of the @xmath106 realizations .
the `` observed '' bias is then estimated using equation [ eqn : bias_measure ] , histogrammed , and normalized such that the area under each curve is unity .
these histograms demonstrate that characterizing the bias of rare @xmath17 galaxies will be challenging .
approximately @xmath107 pointings are needed to measure the bias of these galaxies with an appreciable signal - to - noise ( @xmath108 ) in an accurate way ( @xmath109 ) . with fewer fields ( @xmath110 ) ,
the expected uncertainty in the measured bias is large ( @xmath111 ) and overestimated ( @xmath112 ) .
when @xmath23 is small , the measured dispersion @xmath85 can be less than @xmath40 , providing @xmath94 in equation [ eqn : bias_measure ] . in these catastrophic failures ( @xmath113 for @xmath87 and @xmath114 for @xmath97 , shown as the @xmath98 bin ) , an unusually small poisson scatter limits sensitivity to the cosmic variance in number counts .
these failures decrease as the number of pointings increases ( to @xmath115 for @xmath89 ) .
the larger numbers of @xmath54 galaxies per acs field compared with the number of @xmath17 galaxies per wfc3 field ( see [ subsection : uncorrelated_acs_fields ] ) makes the @xmath54 acs experiment much easier .
while @xmath116 pointings are required to significantly ( @xmath108 ) measure the bias , such a determination would provide a crucial constraint on the nature of @xmath17 galaxies .
current hst programs may have too few fields to perform this experiment , but extensions or new pure parallel programs will likely provide enough pointings in the next few hst cycles to successfully measure the bias of @xmath17 from cosmic variance . pure parallel observations with acs and wfc3 provide a novel method to obtain statistically - independent samples of the high - redshift galaxy population .
however , the forthcoming hst multi - cycle treasury ( mct ) programs will probe greater numbers of contiguous pointings .
while the mct areas may be used to measure the angular correlation function of high - redshift galaxies , one may wonder whether contiguous areas may be used to measure bias from cosmic variance . here , we estimate the effects of correlations in the galaxy counts induced by large - scale density modes shared between nearby pointings .
we consider @xmath54 dropouts in acs pointings , as the improved statistics relative to @xmath17 dropouts more clearly illustrate the effects of field - to - field correlations .
the detailed correlation between nearby fields will depend on the geometry of the tiling and the separation between fields , and are difficult to calculate for general survey designs .
however , we estimate the field - to - field correlations in a contiguous tiling as follows ( for another estimate , see * ? ? ? * ) .
consider as before @xmath23 pointings , but arranged in a contiguous tiling .
the cosmic variance of an independent field with volume @xmath29 will depend on @xmath117 .
correspondingly , the galaxy counts in the larger volume @xmath118 will have a sample variance @xmath119 .
the number counts within the @xmath14-th field of the @xmath23 pointings within the contiguous survey will then be correlated with the other @xmath120 fields approximately at the level of @xmath121 this typically underestimates the correlation between very nearby or adjacent fields , but should approximate the average correlation of the most widely - spaced pointings within the contiguous area .
the matrix describing the correlation between fields @xmath14 and @xmath120 then has a simple form , with the diagonal elements @xmath122 and off - diagonal elements @xmath123 .
to model the counts in correlated fields , we use a standard cholesky decomposition technique to instill the distribution of galaxy overdensities with the correlation matrix @xmath124 , and then poisson - sample the correlated distribution to generate the counts in each field .
figure [ fig : bias_constraint_correlated ] shows the results of this procedure .
since the number counts in fields are correlated ( @xmath125 for @xmath87 and @xmath126 for @xmath89 ) , the dispersion @xmath85 in the acs galaxy counts becomes smaller than in the uncorrelated case .
the resulting estimates for the cosmic variance are correspondingly smaller , leading to a bias measurement that is underestimated ( @xmath127 for @xmath87 and @xmath128 for @xmath89 ) and is more susceptible to catastrophic ( measured @xmath129 ) failures ( @xmath130 for @xmath87 ) .
the simulations presented in [
subsection : uncorrelated_acs_fields ] and [ subsection : uncorrelated_wfc3_fields ] demonstrate that the bias of high - redshift galaxies can be measured directly from the variance in number counts between uncorrelated fields . using the connection between bias and dm halo mass calculated from cosmological simulations @xcite , the galaxy bias can be used to estimate the characteristic mass of dm halos with similar spatial clustering strength . for detailed discussions of the connection between halo mass and bias , we refer the reader to @xcite and @xcite .
figure [ fig : mass_constraint ] shows the characteristic halo mass estimated by converting the bias ( figure [ fig : bias_constraint ] ) using the @xcite bias - mass relation determined from cosmological simulations . for the case of @xmath65ab magnitude @xmath14-dropout galaxies at redshift @xmath54 with bias @xmath84 , the mass of halos with the same bias at @xmath54 is @xmath131 ( dashed line ) .
the bias can provide a precise estimate of a characteristic dm halo mass ( @xmath132 in @xmath133 for @xmath87 and @xmath115 in @xmath133 for @xmath89 ) that , while skewed , is accurate ( @xmath134/\log m = -0.03 $ ] for @xmath87 and @xmath135 for @xmath89 ) .
this measure of a characteristic halo mass only corresponds to the halo mass of @xmath14-dropout galaxies if each halo hosts one galaxy and mass strongly correlates with luminosity ( see * ? ? ?
* ) , but provides a convenient conceptual tool for discussing the approximate mass scale of high - redshift galaxies .
we have presented a simple counts - in - cells method ( e.g. , * ? ? ?
* ; * ? ? ?
* ) for measuring the bias of high - redshift galaxies from the field - to - field variation in number counts induced by cosmic variance .
the number of high - redshift galaxies of a given luminosity are measured in a large number independent , widely - separated fields .
the poisson contribution to the variance in the number counts across these fields is removed , leaving a cosmic variance contribution that depends on the bias and the mean matter overdensity fluctuations on the scale of the survey field volume .
we use monte carlo simulations of the bias measurement to estimate the effectiveness of the method for example galaxy populations , @xmath14-dropouts at @xmath54 in acs - sized fields and @xmath136-dropouts at @xmath17 in wfc3-sized fields .
our method can be adapted for other dropout filter strategies . at @xmath54 ( @xmath17 )
the method should provide a measurement of galaxy bias with @xmath137 for luminosities near @xmath138 when @xmath139 ( @xmath140 ) fields are used .
the uncertainty on the galaxy bias improves with an increasing number of independent fields as @xmath141 .
these requirements on the number of fields are concomitant with the expected number of pointings available from extensions to ongoing pure parallel hst programs @xcite , and the potential for using this method to measure high - redshift galaxy bias is therefore promising .
we also show that using correlated , nearby fields to perform the presented measurement typically leads to an underestimate of the bias owing to a correlation - induced decrease in the dispersion of galaxy number counts .
the measured bias can be associated with a characteristic mass of dm halos with similar clustering .
we show that the typical uncertainty in the bias measured using the presented method corresponds to an uncertainty in the inferred dm halo mass of @xmath142 a few percent in @xmath133 for @xmath14-dropouts . combining these estimates of halo mass with
the observed luminosities allow for an estimate of a mass - to - light ratio , which can constrain the connection between the galaxies observed lf and their clustering @xcite .
numerous effects can alter the correspondence between bias and halo mass , and we therefore view the presented method as primarily a measure of bias and interpret the halo mass estimates with caution .
our simulations assume that the high - redshift galaxy population is essentially volume limited and pure .
methods to handle incompleteness for lbg samples have already been engineered ( e.g. , * ? ? ?
* ; * ? ? ?
contamination at a fractional level of @xmath143 will lower the bias by an amount @xmath144 ( e.g. , * ? ? ? * ) . using simulations that include red @xmath145 galaxy interlopers with @xmath146 and a bias of @xmath147 @xcite , we find that the measured bias at @xmath54 is lowered by @xmath148 .
this systematic error is comparable to the statistical error on the bias when @xmath149 .
lastly , not every parallel field will be useful for measuring high - redshift galaxy counts owing , e.g. , to possible bright star or galactic reddening contamination , and the efficiency of the presented method may be correspondingly decreased .
i thank the anonymous referee for constructive suggestions , as well as richard ellis , chuck steidel , masami ouchi , and haojing yan for helpful discussions .
i am supported by a hubble fellowship grant , program number hst - hf-51262.01-a provided by nasa from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 . | as deeper observations discover increasingly distant galaxies , characterizing the properties of high - redshift galaxy populations will become increasingly challenging and paramount .
we present a method for measuring the clustering bias of high - redshift galaxies from the field - to - field scatter in their number densities induced by cosmic variance .
multiple widely - separated fields are observed to provide a large number of statistically - independent samples of the high - redshift galaxy population .
the expected poisson uncertainty is removed from the measured dispersion in the distribution of galaxy number counts observed across these many fields , leaving , on average , only the contribution to the scatter expected from cosmic variance . with knowledge of the @xmath0-cold dark matter power spectrum , the galaxy bias
is then calculated from the measured cosmic variance .
the results of cosmological n - body simulations can then be used to estimate the halo mass associated with the measured bias .
we use monte carlo simulations to demonstrate that hubble space telescope pure parallel programs will be able to determine galaxy bias at @xmath1 using this method , complementing future measurements from correlation functions . |
in the main text we discussed the generation of negatively oriented loops in @xmath120 through a cusp singularity . in this section
we discuss the second mechanism for generating negatively oriented area the formation of a self tangent . in this process
, one section of the curve intersects a different section - see the lower panel in fig.([fig : supinfo_kiss_1 ] ) .
.,title="fig : " ] + .,title="fig : " ] + .,title="fig : " ] let us write the necessary and sufficient conditions for @xmath100 to have a point at which it is tangent to itself for some @xmath121 ( e.g. point @xmath122 in the middle panel in fig.[fig : supinfo_kiss_1 ] ) : @xmath123 the first two equations assure that the curve intersects itself , and the last equation assures that at the intersection point the curve is tangent to itself . using the equation for @xmath81 given in the main text , the right hand side of the last equation can be expressed as @xmath124 however , at the tangent point @xmath125 .
we can therefore write @xmath126 moreover , as discussed in the main text , the range of values that @xmath85 can have is bounded by the extreme values of @xmath95 .
thus , given @xmath93 , it can be verified for a proposed protocol @xmath84 that the equation above does not hold for any @xmath127 and @xmath128 for which @xmath129 and any @xmath118 in the range @xmath130 note that if @xmath131 for some constants @xmath97 and @xmath98 , as in the example in the main text , and furthermore @xmath132 undergoes only a single oscillation in each period , then eq.([eq : edot_ratio ] ) never holds for any finite @xmath86 .
in this section we explain the transformation from @xmath133 to @xmath84 .
we start with the equation : @xmath134 for any re - parametrization @xmath69 , the equation for @xmath135 is : @xmath136 where @xmath137 and @xmath138 , as in the main text . choosing @xmath69 such that @xmath139 for some constant @xmath140 , would cancel the @xmath141 dependence in the equation for @xmath81 .
therefore , we choose @xmath142 this specific choice of @xmath140 makes @xmath73 , namely when @xmath32 varies from @xmath143 to @xmath144 both @xmath74 , @xmath145 and @xmath146 complete a period , and hence @xmath147 traces exactly the same curve as @xmath31 . the price for
this choice is rescaling of @xmath3 : the equation for @xmath81 becomes : @xmath148 where @xmath149
to show that in the brownian heat engine model discussed in the main text the phase difference between @xmath118 and @xmath132 is the only reason for the decrease in the power , let us first express both @xmath85 and @xmath150 as a fourier series : @xmath151 and @xmath152 . in the fourier basis , the power can be expressed as @xmath153 from this equation it is evident that the power depends on @xmath3 in three distinct ways : ( i ) the _ phase difference _ between @xmath82 and @xmath118 .
note that the ` drift ' of @xmath85 to the left , discussed in the main text , changes exactly this phase difference ; ( ii ) besides the phases @xmath154 , there is additional @xmath3 dependence in the term @xmath155 .
the last factor is ( iii ) the _ spectral overlap _ between @xmath132 and @xmath118 even if @xmath156 is very large for some given @xmath157 , it will not contribute to the power if @xmath158 , namely the spectral overlap between them is zero .
the above three factors are quite general and follows from the definition of the power .
let us next consider the specific model described in the main text .
first , we note that in this case the transformation between @xmath159 and @xmath118 is linear and time invariant , therefore a frequency shift is impossible and the spectral overlap does not changes with @xmath3 .
moreover , the amplitudes @xmath155 can be exactly expressed by taking the fourier series of eq.([eq : w_vs_r ] ) : @xmath160 note that all the amplitudes @xmath155 are monotonically decreasing as a function of @xmath3 .
thus , decreasing @xmath3 necessarily increases the amplitudes of all the fourier coefficients , hence the changes in amplitudes ( which shrinks the width in the @xmath19 direction ) can not explain the power decreasing .
the only reason that the power can decreases with decreasing @xmath3 is the phase difference between @xmath118 and @xmath132 , which deforms the shape of the curve in the @xmath28 $ ] state space .
in the main text we analyzed the brownian particle in a harmonic potential model for a heat engine . another simple model for heat engines which is commonly used
@xcite is the markovian two level system .
this engine is made of two states ( commonly referred to as `` levels '' ) in which the system can be at any moment .
the energy difference between these levels is denoted by @xmath161 and it is periodic in time . for simplicity
we assume that the lower level s energy is fixed in time at @xmath162 , and the upper level s energy is time dependent .
the probability to be in the @xmath63 s level , denoted by @xmath163 , varies according to the master equation , @xmath164 in the master equation , @xmath165 is the rate transition matrix , which can be parametrized by @xmath166 where @xmath167 can be an arbitrary function .
this parametrization is physically motivated by the generalized arrhenius form .
for a two level system , conservation of probability implies @xmath168 , hence we denote the probability to be in the upper state simply as @xmath169 .
the probability to be in the lower level is hence @xmath170 .
substituting eq.([eq : arrhenius ] ) in the master equation gives @xmath171 the infinitesimal heat flow from the bath to the system is given by @xmath172 and the work and efficiency can be interpreted , as in the main text , as areas in the @xmath173 $ ] state space .
following the same steps explained in the main text , we first change to a dimensionless time units , @xmath174 , to get @xmath175 in this case , as in the model described in the main text , we can not vary any of the two control parameters @xmath176 or @xmath116 without affecting the periodic solution of @xmath177 .
we therefore change the time parametrization to @xmath178 the equation for @xmath179 is given by @xmath180 = \hat \tau \left(\bar p(s)-\bar \pi(s)\right)\end{aligned}\ ] ] where we defined @xmath181 the above equation is exactly analogous to eq.(6 ) in the main text , where @xmath173 $ ] is the state space and @xmath182 $ ] is the new control space .
together with the interpretation of the work and efficiency as areas , the rest of the results can be applied equally well to this model .
to prove that no protocol in the brownian particle in harmonics potential heat engine model can attain exactly the carnot efficiency with non - zero power , namely at finite time , we first show that if a protocol achieve exactly the carnot efficiency , then in a complete cycle heat is exchanged with only two baths : the coldest and hottest one . in other words , while it might be possible that in a cycle with the carnot efficiency heat is exchanged with many baths at various temperatures , the net heat exchanged in a complete cycle is zero for all the baths , except the one at the hottest and coldest temperatures . to show this , consider a heat engine which interacts with three heat baths , at temperatures @xmath183 .
let us denote the total heat exchange with these baths by @xmath184 , @xmath185 ( which might be either positive or negative ) , and @xmath186 respectively .
the extracted work is given by @xmath187 .
first , consider the case in which @xmath188 .
the efficiency in this case is given by @xmath189 but the total entropy production in a cycle is non - negative .
as there is no net entropy production in the system itself after completing a cycle when it is in the periodic state , the total entropy production is given by @xmath190 , thus @xmath191 since @xmath192 and both @xmath193 and @xmath194 are positive , the efficiency is necessarily smaller then the carnot efficiency . to show that no piecewise continuous protocol can attain exactly the carnot efficiency in this model
, we will next show by contradiction that if such a protocol exist , then there exist another protocol , with an even larger efficiency .
this would clearly violate the second law ( which is built into the model ) , since the efficiency of this other protocol is larger then the carnot efficiency .
assume that there exist a protocol , @xmath114 that attains exactly the carnot efficiency with non - zero power at finite time @xmath3 .
then , by the argument given above , there is only total heat exchange with two baths , hence @xmath200 where @xmath186 and @xmath184 for it to be an engine .
moreover , it must interact for some finite time with the coldest bath , otherwise it will not exchange heat with the coldest bath or with any bath other then the hottest bath , so by the second law its power must be zero .
let us denote one of the intervals in which the engine is in contact with the coldest bath by @xmath201 .
in what follows , we shows that it is possible to deform only the part of the protocol in which it touches the cold bath , namely in the interval @xmath201 , and increase @xmath202 ( and hence the efficiency ) without changing the temperature or the rest of the protocol . to this end
we use @xmath203 where for simplicity we use the notation @xmath204 .
the protocol deformation , denoted by @xmath205 , changes also @xmath19 by @xmath206 , but must not change @xmath207 , to avoid changing the carnot efficiency .
therefore , the deformation must be constrained by : @xmath208 additional restriction on the deformation comes from the fact that @xmath19 is a solution of an o.d.e .
, thus it must be continues ( but not necessarily differentiable ) , namely @xmath209 . in principle
, the deformation is defined by @xmath210 , and the deformation @xmath206 can be calculated from the equation above
. however , the constraint @xmath209 makes it simpler to define the deformation in terms of @xmath206 , and solve for @xmath205 .
let us choose the variation fo the form @xmath211 for some function @xmath212 ( and infinitesimal @xmath213 ) .
then from the above equation : @xmath214 but on the other hand @xmath215 equating these two equation and solving for @xmath141 , one finds that @xmath216 next , let us consider the change in the heat exchange with the cold bath : @xmath217 hence @xmath218 using the results above we can write : @xmath219 can we choose a deformation @xmath212 that will make @xmath220 , hence increasing the efficiency ? to do this we need to choose @xmath221 such that the above integral is positive , and such that @xmath222 and @xmath223 .
this can be done as follows : we choose @xmath221 to be zero everywhere , except for a very small section , of length @xmath224 , somewhere at a location between @xmath225 and @xmath128 such that in this interval @xmath65 and @xmath226 are both continuous .
@xmath224 is chosen to be short enough such that in this interval both @xmath227 and @xmath228 are constant to a good approximation , and such that @xmath229 : this is always possible since @xmath230 by definition ( it is a variance ) , and there must be a place where @xmath231 since in a heat engine @xmath232 , and @xmath233 .
let us denote the starting point of the interval by @xmath234 . in this interval
we choose @xmath235 note that @xmath236 in both the end points of the interval , and its derivatives at the end points are zero , so @xmath212 is both continues and differentiable .
its integral is given by @xmath237 moreover , @xmath238 since the prefactors of both @xmath212 and @xmath239 in the interval ( in which @xmath212 and @xmath239 are different then zero and do not change much ) , we can take them out of the integral , @xmath240 note that there are two scenarios in which the carnot efficiency can nevertheless be attained : ( i ) there is no section @xmath201 in which the system is in contact with the coldest bath .
as explained , the power in this case is zero .
( ii ) in the limit @xmath45 , eq.([eq : delta_lambda ] ) gives finite @xmath241 only if @xmath242 for all @xmath32 , so our choice of @xmath212 is inconsistent , and the argument for no protocol with carnot efficiency is invalid .
this should be expected there are definitely protocols that saturate the carnot bound in the @xmath45 limit . | thermodynamics places a limit on the efficiency of heat engines , but not on their output power or on how the power and efficiency change with the engine s cycle time . in this letter
, we develop a geometrical description of the power and efficiency as a function of the cycle time , applicable to an important class of heat engine models .
this geometrical description is used to design engine protocols that attain both the maximal power and maximal efficiency at the fast driving limit .
furthermore , using this method we also prove that no protocol can exactly attain the carnot efficiency at non - zero power . [ [ introduction ] ] introduction + + + + + + + + + + + + heat engines - machines that exploit temperature differences to extract useful work , are modeled as operating in either a non - equilibrium steady - state , e.g. thermoelectric @xcite or chemical potential @xcite driven engine , or as a cyclic engine , where external parameters and temperature are varied periodically in time , e.g. the carnot , otto , stirling and the diesel cycles @xcite
. both types of engines are characterized by two main figures of merit : efficiency and power . in steady - state heat engines , currents generated by the temperature difference flow against some affinities ( thermodynamical forces ) , e.g.
electrical @xcite or chemical potentials @xcite , generating useful work . from a practical standpoint ,
the interest in these engines is limited , since the majority of heat engines are better modeled as cyclic engines .
one of the main motivations to study steady - state heat engines , however , is the hope that they share universal characteristics with cyclic engines , which are generically more difficult to analyze .
an example for such a characteristic behavior is the relationship between power and efficiency : in all steady state heat engines , the affinity at maximal power does not equal to the affinity at maximal efficiency , unless one of the heat baths is at infinite or zero temperature .
heat engines that attain their maximal power and maximal efficiency ( which is either the carnot efficiency or a lower value ) at different working conditions are here defined as _ heat engines with a power - efficiency trade - off_. the power - efficiency trade - off is the subject of many recent studies @xcite .
less is known about the efficiency and power of cyclic heat engines , but a lot of research effort has been devoted to understanding them in recent years @xcite .
the operation of a cyclic engine is characterized by a protocol that describes the time dependence of key variables along the cycle e.g. piston position and temperature .
the set of feasible protocols however , is strongly bounded by a set of engine specific and hence non - generic constraints .
maximizing power or efficiency is , therefore , a nontrivial constrained optimization problem .
nevertheless , there is a natural optimization problem in these engines which is both simpler and of practical importance : optimization with respect to overall cycle time . in most cyclic engines ,
the protocol is determined up to a rescaling of the cycle time . in other words ,
the overall rate at which the engine operates can be varied .
it is thus natural to consider the efficiency and power of a heat engine with a fixed protocol as a function of the cycle time .
steady - state heat engines have a characteristic trade - off between power and efficiency as a function of the affinity .
do cyclic heat engines have a corresponding power - efficeincy trade - off as a function of their cycle time ?
analytical @xcite , numerical @xcite and experimental @xcite results for certain driving protocols seem to suggest that this might be the case : increasing the cycle time increases the efficiency , with the maximal efficiency ( which is possibly lower then the carnot limit as in the diesel , miller and sargent cycles ) only attained in the quasi - static limit namely at infinitely long cycle time , where the power vanishes . on the other hand , driving the engine faster increases the power at the expense of efficiency , until eventually , at fast enough driving , the dissipation rate becomes significant and causes a decrease in power . yet , as we demonstrate , this behavior is not universal , and there is no inherent trade - off between power and efficiency as a function of the cycle time , although such a trade - off always exists in cycles that exactly attain the carnot bound .
here we analyze a class of cyclic heat engine models , referred to as _ geometric heat engines _ , which includes the paradigmatic examples of a brownian particle in a parabolic potential @xcite and the two - state markovian engines @xcite , but is not limited to these models . in this class ,
the work and heat can be interpreted as areas in state space ( the space of all the possible states of the engine ) defined by the periodic trajectory of the engine s state .
this interpretation has two consequences : ( i ) work and heat can be formulated as parametrization independent quantities , and ( ii ) power can vanish due to the formation of singularities in the engine s trajectory in state space .
we then introduce a time re - parametrization that effectively decouples the state space variables .
this decoupling , together with an understanding of the aforementioned singularities , enables the design of novel engine driving protocols that avoid power losses at short cycle times . as an example
, we construct a protocol whose power and efficiency are both maximized at the infinitely fast driving limit .
this proves that cyclic heat engines do not have an inherent trade - off between power and efficiency as a function of their cycle time . achieving the maximum efficiency at finite power , however , comes with a price : we prove that in this class of models the carnot limit can not be attained at non - zero power .
therefore , to avoid the trade - off , the efficiency must be lower then the carnot limit .
similar relations between the carnot limit and power were discussed in @xcite in the linear response regime , and very recently for arbitrary markovian dynamics with local detailed balance in @xcite .
[ [ model - description ] ] model description : + + + + + + + + + + + + + + + + + + for simplicity , we focus here on a specific model , and subsequently show that our results are valid for a larger class of models .
this model consists of an overdamped brownian particle confined to one spatial dimension , in a time dependent harmonic potential @xmath0 , coupled to a heat bath with a time dependent inverse temperature @xmath1 .
this model was suggested in @xcite and experimentally realized in @xcite .
both @xmath2 and @xmath1 are periodic with a cycle time @xmath3 .
and @xmath4 are twice differentiable . ] the engine s protocol , which is a time - parametrized closed curve in the _ control space _ - the space of external control parameters , is denoted by @xmath5 where the subscript @xmath6 indicates that the protocol is parametrized by the time and the superscripts @xmath7 indicate that this protocol is defined in the @xmath8 $ ] control space . the probability density to find the particle at point @xmath9 at time @xmath6 , denoted by @xmath10 , evolves according to the fokker - planck equation , @xmath11p(x , t),\ ] ] where for simplicity we assumed the mobility to be one half . under this evolution ,
an initial gaussian distribution centered at the origin remains a centered gaussian with time dependent width and height @xcite , which are related through probability conservation .
this gaussian distribution can be parametrized by a single parameter
the variance , @xmath12 , which evolves in time according to @xcite @xmath13 after sufficiently long time , the system reaches a periodic state , @xmath14 .
we assume that the engine already relaxed to this periodic state . to describe the engine s properties when the same protocol is performed at different cycle times , it is useful to consider the protocol in terms of a dimensionless time parameter , @xmath15 in the interval @xmath16 , rather than @xmath17 . defining the protocol as @xmath18 , allows us to treat the cycle time @xmath3 as a parameter , independent of other characteristics of the protocol . the time evolution of the width @xmath19 , governed by eq.([eq : w_dot_t ] )
, can be written in terms of @xmath20 as : @xmath21 note that the solution @xmath22 of the above equation is explicitly @xmath3 dependent .
the infinitesimal heat exchange between the system and the bath is given by @xmath23 .
with this convention , @xmath24 implies that heat flows into the system @xcite . by conservation of energy , the total work extracted in a cycle and the corresponding power are given by @xmath25 the equation for the work @xmath26 has a geometrical interpretation : @xmath26 is half the oriented area bounded by @xmath27 , namely by the curve in the @xmath28 $ ] state space .
an important consequence of the geometrical interpretation is that the work is parametrization independent . by this
we mean that if some other driving protocol , @xmath29 , operating at some cycle time @xmath30 , happens to trace the same curve in the @xmath28 $ ] space as @xmath31 but with a different @xmath32 parametrization , then the extracted work is equal in the two protocols , even though @xmath33 . to define the efficiency , we need to quantify the cost of any protocol in terms of the heat extracted from the heat baths during a cycle
this can be done by accounting for only the sections of the cycle during which heat flows from the bath into the system : @xmath34ds , \end{aligned}\ ] ] where @xmath35 $ ] is the heaviside step function .
this integral can be interpreted geometrically as half the area under the sections of @xmath31 in which @xmath19 decreases.$ ] is not geometrical , since it is not parametrization independent . however , if we consider only re - parameterizations @xmath36 in which @xmath37 is a monotonically increasing function , then the geometric interpretation can be applied . ] with these definitions of work @xmath26 and heat @xmath38 , which are consistent with the laws of thermodynamics @xcite , the efficiency is given by @xmath39 , and it can be interpreted as the ratio between the two corresponding areas . the interpretation of the work and heat as areas is done in the @xmath28 $ ] state space , which describes the instantaneous state of the engine , whereas the engine s driving protocol is defined in the @xmath8 $ ] control space , which is the space of the external control parameters .
this makes it difficult to directly relate the protocol to the areas associated with work and heat .
however , the protocol can be defined equally well in terms of @xmath40 instead of @xmath4 , namely as @xmath41 .
note that @xmath42 is the width of the boltzmann distribution for the potential @xmath43 .
the main advantage of defining the protocol as @xmath44 is that in the quasi - static limit , @xmath45 , eq.([eq : w_dot ] ) implies that @xmath46 and @xmath47 , therefore we can unify the control space and state space . with decreasing @xmath3 , the contour that defines the protocol , @xmath44 , continuously deforms into the contour @xmath31 , which has the geometrical interpretation for work and heat .
therefore , we study this deformation next . [ [ finite - cycle - time ] ] finite cycle time : + + + + + + + + + + + + + + + + + + consider first a simple driving protocol , in which @xmath48 traces a circle at a uniform rate in control space as shown in the upper inset of fig([fig : circular_protocol ] ) . the upper panel shows the efficiency and power as a function of the cycle time @xmath3 . as can be seen , this protocol has the typical behavior described above : the efficiency is a monotonically increasing function of @xmath3 , attaining its maximum as @xmath45 . the power is zero where the efficiency is maximal , and attains its maximum at about @xmath49 .
below about @xmath50 , both the power and efficiency become negative , since the work changes its sign . for
the work to become negative , the area bounded by @xmath31 must changes its orientation .
how can this comes about ? for a circular protocol in the @xmath8 $ ] control space ( inset ) .
this protocol attains its maximal power ( @xmath51 ) at the @xmath45 limit .
the carnot limit for this protocol is @xmath52 . at high driving rates ,
the power and efficiency change their sign . *
lower panel : * the @xmath28 $ ] curves of the above protocol for few values of the cycle time @xmath3 . at
about @xmath53
the curve develops a cusp , which for even smaller @xmath3 evolves into a loop with a negative orientation .
the inset shows a blow - up of the cusp and a negatively oriented loop.,title="fig : " ] for a circular protocol in the @xmath8 $ ] control space ( inset ) .
this protocol attains its maximal power ( @xmath51 ) at the @xmath45 limit .
the carnot limit for this protocol is @xmath52 . at high driving rates ,
the power and efficiency change their sign . *
lower panel : * the @xmath28 $ ] curves of the above protocol for few values of the cycle time @xmath3 . at
about @xmath53
the curve develops a cusp , which for even smaller @xmath3 evolves into a loop with a negative orientation .
the inset shows a blow - up of the cusp and a negatively oriented loop.,title="fig : " ] there are only two generic ways in which the area orientation of a closed curve in 2d can change through continuous deformations : a cusp singularity or a self - tangent singularity @xcite .
we next analyze the formation of the cusp singularity , whereas the formation of the self - tangent singularity is analyzed in the si .
a cusp in @xmath31 emerges when both @xmath54 and @xmath55 have an extreme point at the same value of @xmath32 @xcite . in other words , if varying @xmath3 causes the value of @xmath56 at which @xmath57 to pass through @xmath58 for which @xmath59 , then a cusp is formed when @xmath60 , and developed into a loop with an inverted orientation .
this loop decreases the power and efficiency , enabling the work to vanish at some non - zero @xmath3 . in the specific example described in fig.([fig : circular_protocol ] ) , a cusp singularity is generated at @xmath61 . below this @xmath3
, the cusp evolves into a negatively oriented loop , reducing the power and efficiency . realizing that the deterioration of the power can be related to the formation of negatively oriented areas immediately raises the question : can a protocol , in which negatively oriented areas never occur , be designed ? to avoid the singularities , the locations @xmath62 of the extreme points of @xmath54 , where the index @xmath63 label the different extreme points , must be considered .
however , even if we know @xmath64 , manipulating the protocol to avoid the singularities is challenging , since varying either @xmath65 or @xmath66 varies @xmath67 as well . to simplify the analysis , we take advantage of the fact that areas are parametrization invariant . instead of looking on the actual parameterization , we consider the following re - parametrized quantities @xmath68 where the re - parametrization @xmath69 is given by @xmath70 note that @xmath71 , i.e. @xmath69 is monotonically increasing with @xmath32 , and moreover @xmath72 and @xmath73 . also note that the transformation is given in terms of @xmath74 rather then @xmath65 .
this turns out to be useful in what follows . although @xmath75 and @xmath76 , their corresponding curves , @xmath31 and @xmath77 , trace _ the same contour _ in the engine s state space , @xmath28 $ ] , hence they have the same work and efficiency . in this re - parameterization ,
@xmath78 where @xmath79 .
although @xmath80 , the two curves are different parameterizations of the same contour in the @xmath28 $ ] state space , and hence they enclose the same area and represent heat engines with the same values of heat and work . the main advantage of this re - parametrization is that the equation for @xmath81 ( eq .
[ eq : w_vs_r ] ) is independent of the potential width @xmath82 .
therefore , using the @xmath83 parametrization to define the protocol , @xmath84 can be manipulated without affecting @xmath85 , but at the price of rescaling @xmath3 .
[ [ the - tau - dependence - of - bar - wss - extreme - points . ] ] the @xmath3 dependence of @xmath85 s extreme points .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to avoid cusp generation , we next explain how the extreme points of @xmath85 change as a function of @xmath86 .
let us denote by @xmath56 any @xmath32 at which @xmath85 has an extreme point , namely @xmath87 . by taking the derivative of eq.([eq : w_vs_r ] ) and using @xmath88 it follows that @xmath89 these two equations can be interpreted as follows : in the quasi - static limit , @xmath90 , @xmath91 everywhere , including the extreme points .
decreasing @xmath86 , the maximal points of @xmath85 , for which @xmath92 , `` slide '' along the @xmath93 curve down and to the right ( where the slope of @xmath93 is negative ) , and similarly , the minimal points of @xmath85 ( where @xmath94 ) slides up and to the right along the positive slope of @xmath95 .
physically , this means that decreasing @xmath86 results in a flattening of @xmath85 , with an increasing phase - lag between @xmath85 and @xmath95 .
example for this behavior can be seen in the lower panel of fig.([fig : resub_fig_2_eff_power_graph ] ) . at this point
one might worry about a different effect that have the potential to decrease the power with decreasing @xmath86 : the range covered by @xmath85 , namely the range between its maximum and minimum , decreases with decreasing @xmath86 , eventually shrinking the area .
this might cause the power to decrease , even without the phase - lag , which is manifested e.g. in the generation of negatively oriented loops . in the si
we show that for the specific model considered here , this is not the case .
[ [ designing - a - protocol - without - a - power - efficiency - trade - off . ] ] designing a protocol without a power - efficiency trade - off .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to avoid the power deterioration , we should design a protocol that does not form singularities .
this can always be done by choosing @xmath96 for some constants @xmath97 and @xmath98 , and @xmath84 that has a single minimum and maximum .
in such a protocol , the extreme points of @xmath85 coincide with those of @xmath84 only in the @xmath99 limit . as discussed in the si
, such a protocol is also assured not to have a self tangent singularity for any finite @xmath86 . in the @xmath99 limit
, @xmath100 bounds no area and the work and efficiency are zero .
when @xmath86 decreases , the phase lag between @xmath84 and @xmath85 ` inflates ' the area . for demonstration
, we show in fig.([fig : resub_fig_2_eff_power_graph ] ) the example @xmath101 , for which @xmath85 can be solved analytically .
as shown in the figure , when @xmath86 decreases , @xmath85 shifts to the right and its amplitude decreases .
this protocol indeed does not have any trade - off between power and efficiency , and in fact both the maximal power and maximal efficiency are attained asymptotically at the fast driving limit . .
the protocol is given by @xmath102 .
the inset shows the protocol in the @xmath8 $ ] control space .
the protocol traverses the line back and forth , without covering any area . *
lower panel : * @xmath85 for various values of @xmath3 .
when @xmath103 , @xmath91 .
as can be seen , the maximum of @xmath85 `` slides '' down the negative slope when @xmath3 decreases , and the minimum if @xmath104 `` slides up '' on the positive slope .
overall , @xmath85 shifts to the right and its amplitude decreases with decreasing @xmath3.,title="fig : " ] .
the protocol is given by @xmath102 .
the inset shows the protocol in the @xmath8 $ ] control space .
the protocol traverses the line back and forth , without covering any area . *
lower panel : * @xmath85 for various values of @xmath3 .
when @xmath103 , @xmath91 .
as can be seen , the maximum of @xmath85 `` slides '' down the negative slope when @xmath3 decreases , and the minimum if @xmath104 `` slides up '' on the positive slope .
overall , @xmath85 shifts to the right and its amplitude decreases with decreasing @xmath3.,title="fig : " ] to implement this protocol in an experimental realization as in @xcite , it is not useful to specify it as @xmath84 and @xmath93 , since in the laboratory the protocol @xmath55 and @xmath105 should be performed . however , as @xmath69 is given in terms of @xmath106 ( eq . [ eq : lambda_s ] ) , it can be calculated from the designed protocol , inverted into @xmath107 such that @xmath108 , and then @xmath109 and @xmath110 .
the resulting protocol in the @xmath8 $ ] space is shown in the inset of fig.([fig : resub_fig_2_eff_power_graph ] ) .
note that the maximal efficiency in this protocol is only @xmath111 ( attained in the limit @xmath112 ) , compared to the carnot efficiency calculated from the minimal and maximal temperatures which is @xmath113 .
this is expected , since in this protocol the engine exchanges heat with many heat baths at intermediate temperatures , hence it can not attain the carnot efficiency computed with the extreme temperatures .
however , _ this protocol does not suffer from a tradeoff between power and efficiency .
_ with the tools developed above , it is natural to ask : is it possible to design a protocol that achieves the carnot limit but has non - zero power , namely operate at a finite cycle time @xmath3 ? as we show in the si , the answer is no .
this is proven by showing that for any piecewise continuous protocol @xmath114 attaining the carnot efficiency at some finite cycle time @xmath3 , it is possible to slightly deform the protocol into a new protocol @xmath115 which has the same @xmath116 ( hence the same carnot bound ) but different @xmath65 , such that the efficiency of the new protocol @xmath115 is strictly larger then the efficiency of @xmath114 , hence larger then the carnot limit .
this would violate the second law .
[ [ applicability - to - other - models . ] ] applicability to other models .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the above analysis was made possible due to three properties of the model : ( i ) the interpretation of the work and incoming heat as areas ; ( ii ) the 2d nature of the state space , which enabled simple characterization of all the possible ways in which the area can change its orientation , and ( iii ) the ability to separately control the work parameter @xmath84 without influencing the other state parameter , @xmath85 .
this required changing the time parametrization from @xmath32 to @xmath69 .
what class of models share these properties ?
the first property is common in a wide range of models , including any markovian finite state system and any brownian engine in which both the time dependent potential @xmath117 and the periodic solution for the probability density @xmath10 can be described by a finite number of parameters at each given time @xmath6 .
the second property is not a true limitation , however , at higher dimensions the analysis of the possible ways in which the area changes its sign is significantly complicated .
we do not know what is the exact class of models in which the last condition is possible , however , as shown in the si , in addition to the example given above , this class contains also any 2-levels markovian model . [ [ acknowledgments ] ] acknowledgments + + + + + + + + + + + + + + + we would like to thank c. jarzynski for useful discussions and y. sagi , s. rahav , s. kotler , o. hirshberg , v. alexandrov , j. hopfield for discussion and comments .
o.r . acknowledges the financial support from the james s. mcdonnell foundation .
r.p . would like to thank s. christen and a. levine for their help and support .
financial support from fullbright foundation , eric and wendy schmidt fund and the janssen fellowship are greatfully acknowledged .
y.s . acknowledge financial support from the u.s .
army research office under contract number w911nf-13 - 1 - 0390 .
26 natexlab#1#1 bibnamefont # 1#1 bibfnamefont # 1#1 citenamefont # 1#1 url # 1`#1 ` urlprefix [ 2]#2 [ 2][]#2 , * * , ( ) .
, , , * * , ( ) . , , , * * , ( ) . ,
_ _ ( , ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( )
. , * * , ( ) . , , , , * * , ( ) . , ,
, * * , ( ) . ,
* * , ( ) .
, , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , ,
* * , ( ) . ,
* * , ( ) . , , , , * * , ( ) . , . , * * , ( ) . , _ _ ( , ) .
supplementary information + this supplementary information has 5 sections : in the first section , we consider the self - tangent singularity ; in the second section , we discuss the transformation @xmath69 ; in the third section we show that any decrease in the power at low @xmath3 comes only from the phase between @xmath118 and @xmath119 and not from the amplitude of @xmath85 ; in the fourth section , we show additional important example for geometrical heat engine ; in the last section , we show that no protocol in these engines can attain the carnot efficiency with non - zero power . |
recently a new global next - to - leading order ( nlo ) analysis @xcite of spin - dependent parton distribution functions ( pdfs ) was performed by the jam ( jefferson lab angular momentum ) collaboration @xcite , in which particular attention was paid to the valence quark - dominated region at high parton momentum fractions @xmath0 and low four - momentum transfers @xmath2 .
this region requires careful treatment of the potentially important @xmath3 power corrections associated with target mass and higher twist contributions to the inelastic cross sections , as well as nuclear smearing effects when using deuterium and @xmath4he data .
the analysis @xcite indeed found significant effects on the polarized leading twist pdfs when twist-3 and twist-4 power corrections in both the spin - dependent @xmath5 and @xmath6 structure functions were taken into account .
in particular , the @xmath7 distribution ( defined as @xmath8 ) was found to have a significantly larger magnitude at @xmath9 than in previous global analyses , driven partly by a large and positive twist-4 correction to the neutron @xmath5 structure function .
analyses such as those in ref .
@xcite that systematically incorporate subleading effects in an effort to accommodate data over a broad range of kinematics can therefore provide a more solid basis for extracting reliable information on pdfs and their uncertainties , especially in regions such as at large @xmath0 where data are relatively scarce @xcite .
in fact , the absence of high - precision polarization data at high @xmath0 , particularly for the neutron ( or @xmath4he ) , has meant that spin - dependent pdfs are essentially unconstrained for @xmath10 .
this is rather unfortunate , given that polarized pdfs , and ratios of polarized to unpolarized pdfs , are quite sensitive to the details of nonperturbative quark - gluon dynamics in the nucleon at high @xmath0 , with theoretical predictions differing in some cases even in sign @xcite .
in the simplest quark models , for example , spin - flavor symmetry implies constant ratios of pdfs , @xmath11 , @xmath12 and @xmath13 .
symmetry breaking effects , which typically generate a larger energy for axial vector spectator diquark configurations compared to scalar diquarks , generally raise the @xmath14 ratio to unity in the @xmath1 limit , while keeping @xmath15 unchanged from the su(6 ) value @xcite .
calculations of one - gluon exchange in perturbative qcd ( pqcd ) , on the other hand , predict that @xcite @xmath16 where @xmath17 is the spin - averaged distribution , for all quark flavors @xmath18 .
similar expectations arise also from arguments based on local quark - hadron duality @xcite . while most global pdf analyses do not include conditions such as in eq .
( [ eq : x1limit ] ) in order to avoid introducing theoretical bias into the pdf extraction , brodsky , burkardt and schmidt @xcite proposed a simple parametrization of pdfs in which the large-@xmath0 constraints of eq .
( [ eq : x1limit ] ) were built in .
only a limited set of data was analysed in ref .
@xcite , however , although a subsequent global analysis utilizing the pqcd expectations was performed by leader , sidorov and stamenov ( lss ) @xcite .
this found that a reasonably good fit to the available data was indeed possible , with the feature of a steep rise in the @xmath15 ratio at intermediate values of @xmath0 .
later high - precision data from the e99 - 117 experimental at jefferson lab @xcite observed the first evidence of a turn over in the @xmath19 neutron polarization asymmetry from negative to positive values , although at somewhat larger values of @xmath0 ( @xmath20 ) than those in the fit of ref .
@xcite . in a more recent analysis , avakian _
@xcite showed that inclusion of @xmath21 components in the lowest three - quark fock state of the nucleon , in addition to the usual @xmath22 configurations , can generate additional terms that behave as @xmath23 at large @xmath0 , which can play an important role numerically .
generalizing the pqcd - inspired parametrization from ref .
@xcite to include the subleading @xmath24 terms , avakian _ et al .
_ showed that the large-@xmath0 asymmetry data could be well fitted while preserving the asymptotic constraints of eq .
( [ eq : x1limit ] ) .
in particular , the new @xmath15 ratio was found to remain negative until @xmath25 , as suggested by the e99 - 117 data @xcite , before rising towards unity at higher @xmath0 values . while the analysis of ref .
@xcite showed the potential of high-@xmath0 data to reveal information about the orbital angular momentum ( oam ) of quarks in the nucleon , it was not based on a comprehensive global analysis of all available data .
the goal of the present work is to examine the behavior of spin - dependent pdfs in the @xmath26 region in the context of a global qcd analysis , including the effects of the @xmath1 constraints in eq .
( [ eq : x1limit ] ) and of the @xmath24 terms inspired by pqcd .
we begin our discussion in sec .
[ sec : fits ] by summarizing the recent global analysis @xcite from the jam collobaration , which we use as the baseline fit for our large-@xmath0 studies . to explore the dependence on the choice of parametrization and
allow for more direct connection with quark orbital angular momentum , we also consider a simplified functional form which uses a smaller number of parameters .
the effects on the fits of additional terms in the pdf parametrizations induced by nonzero orbital angular momentum are investigated , together with the impact on the @xmath27 and @xmath28 pdfs from imposing constraints for the @xmath1 behavior from perturbative qcd . in sec .
[ sec:12gev ] we repeat the global analysis using in addition pseudodata generated at the kinematics of future data from several experiments planned at the 12 gev energy upgraded jefferson lab , and quantify the resulting reduction in the pdf errors at high @xmath0 . finally in sec .
[ sec : conc ] we draw some conclusions of the present analysis and outline steps for future work .
for our exploration of the large-@xmath0 region we use as a baseline the pdfs from the jam analysis in ref .
the jam pdfs were obtained from a global nlo fit to all available data on inclusive polarized deep - inelastic scattering asymmetries for and @xmath29 gev .
inclusion of low-@xmath2 and low-@xmath30 data necessitated a careful treatment of the subleading @xmath3 contributions , to both the @xmath5 and @xmath6 structure functions , from target mass and higher twist corrections , as well as nuclear smearing effects for deuterium and @xmath4he data . by fitting directly the longitudinal and transverse polarization asymmetries , where available , one avoids introducing biases that would otherwise arise in fits to the spin - dependent structure functions , which are often extracted from the experimental asymmetries under different assumptions for the spin - averaged structure functions .
a standard parametrization was used for the polarized quark , antiquark and gluon distributions in terms of four parameters plus an overall normalization , @xmath31 where @xmath32 , at the input scale @xmath33 . at large values of @xmath0 the antiquark and gluon pdfs
play a negligible role , so that the inclusive dis data alone are sufficient to determine the @xmath34 and @xmath7 distributions individually .
a total of over 1,000 data points were used in the analysis , giving an overall @xmath35 per data point of 0.98 ( see table [ tab : chi2 ] ) .
the resulting @xmath34 and @xmath7 distributions are illustrated in fig .
[ fig : qplus ] , together with the ratios to the spin - averaged pdfs , @xmath36 and @xmath37 , at @xmath38 gev@xmath39 .
.@xmath35 values per number of data points @xmath40 for the various fits discussed in this analysis , including the jam , simp and oam fits , with or without the @xmath1 constraint in eq .
( [ eq : x1limit ] ) , and including leading twist ( lt ) contributions only . for the jam fit we also considered the cases where the @xmath7 pdf was forced to cross zero at @xmath41 and @xmath42 . + [ cols="<,^,^,^,^,^,^",options="header " , ] [ tab : chi2 ] and @xmath7 * ( left ) * and their ratios @xmath36 and @xmath37 to the spin - averaged distributions * ( right ) * at a scale of @xmath43 gev@xmath39 . the distributions from the jam analysis @xcite ( black solid curves )
are compared with those using a more basic parametrization ( orange dotted curves ) , the `` oam '' parametrization inspired by nonzero orbital angular momentum considerations ( blue short - dashed curve ) , and the `` jam+ '' ( red long - dashed curves ) and `` oam+ '' ( green dot - dashed curves ) fits which include the @xmath1 constraint from eq .
( [ eq : x1limit]).,title="fig:",width=302 ] and @xmath7 * ( left ) * and their ratios @xmath36 and @xmath37 to the spin - averaged distributions * ( right ) * at a scale of @xmath43 gev@xmath39 . the distributions from the jam analysis @xcite ( black solid curves ) are compared with those using a more basic parametrization ( orange dotted curves ) , the `` oam '' parametrization inspired by nonzero orbital angular momentum considerations ( blue short - dashed curve ) , and the `` jam+ '' ( red long - dashed curves ) and `` oam+ '' ( green dot - dashed curves ) fits which include the @xmath1 constraint from eq .
( [ eq : x1limit]).,title="fig:",width=302 ] unlike the @xmath2 dependence of pdfs , which is determined by the @xmath2 evolution equations to a given order in the strong coupling @xmath44 @xcite , the @xmath0 dependence of pdfs is generally not accessible directly from pqcd calculations .
an exception is the kinematic region at large @xmath0 , where hard gluon exchange between the quarks in the leading three - quark fock state component of the nucleon can be used to determine the dominant contributions to the @xmath0 dependence of the pdfs in the @xmath1 limit @xcite .
typically one finds that the quark pdfs in the nucleon behave as @xmath45 , where @xmath46 is the minimum number of partons that are spectators to the hard collision @xcite , so that for @xmath47 the leading exponent is equal to 3 .
more generally , the exponent on @xmath48 also depends logarithmically on @xmath2 @xcite , although the scale from which this should evolve is _ a priori _ unknown .
nevertheless , global pdf fits do find parameters @xmath49 in eq .
( [ eq : jam ] ) that are close to the pqcd ( or quark `` counting rule '' ) predictions ; for the jam fit , for instance , one has @xmath50 and @xmath51 for @xmath34 and @xmath7 , respectively , at the input scale @xmath33 .
of course , the additional polynomial terms in ( [ eq : jam ] ) with coefficients @xmath52 and @xmath53 obscure the direct connection between the @xmath0 dependence of the fitted distributions and the predicted pqcd behavior . to make the connection more explicit , we consider a fit based on a simplified functional form , with parameters @xmath52 and @xmath53 in eq .
( [ eq : jam ] ) set to zero . the resulting fit , labeled `` simp '' in fig . [ fig : qplus ] , gives similar @xmath34 and @xmath7 distributions to those from the full jam analysis , albeit with a slightly larger overall @xmath35 value .
the leading @xmath48 exponents in this case are reduced slightly to @xmath54 and @xmath55 . as well as specifying the leading @xmath1 behavior of the pdfs
, the pqcd counting rules also predict that the dominant contribution to the cross section , in a reference frame where the nucleon is moving fast along the @xmath56-axis , is from scattering off quarks with the _ same _ helicity as that of the nucleon .
this implies that asymptotically the helicity - aligned distributions dominate both the unpolarized and polarized pdfs , so that the ratio @xmath57 as @xmath1 for all quark flavors @xmath18 , as in eq .
( [ eq : x1limit ] ) . in this scenario
the @xmath58 polarization asymmetries in dis are therefore expected to approach unity for both the proton and neutron .
unfortunately , current data can not definitively confirm the pqcd expectations . while the proton @xmath59 asymmetries , which have been measured to @xmath60 in the dis region , are consistent with an approach towards unity in the @xmath1 limit , the neutron ( or @xmath4he ) data extend only to @xmath61 and are generally consistent with a zero or negative asymmetry .
the dearth of high-@xmath0 data means that the spin - dependent pdfs , and particularly the @xmath7 distributions , are essentially unconstrained above this region . consequently the spin - dependent pdfs obtained from global analyses often violate the positivity condition @xmath62 at large @xmath0 (
although strictly speaking these need not be satisfied beyond leading order ) .
this can be seen in fig .
[ fig : qplus ] for both the jam and simp ( and other ) fits , and in general will be a feature of any global fit which does not _ a priori _ impose the positivity constraint ( or else fit the individual helicity - aligned and helicity - antialigned distributions separately ) .
note that in the present work the unpolarized fits are based on the recent nlo global analysis from ref .
ideally one should perform a global fit of polarized and unpolarized data simultaneously and extract the helicity - aligned and antialigned distributions
. this will be particularly important in future analyses that include semi - inclusive data sets , where transverse momentum distributions , which are expected to be different for different helicity states @xcite , will play a more important role .
gev@xmath39 with proton * ( left ) * and @xmath4he * ( right ) * @xmath58 polarization asymmetry data for similar @xmath2 values .
the proton data are from the slac e143 @xcite and hermes @xcite experiments , while the @xmath4he data are from the slac e142 @xcite , hermes @xcite and jefferson lab e99 - 117 @xcite experiments.,title="fig:",width=302 ] gev@xmath39 with proton * ( left ) * and @xmath4he * ( right ) * @xmath58 polarization asymmetry data for similar @xmath2 values .
the proton data are from the slac e143 @xcite and hermes @xcite experiments , while the @xmath4he data are from the slac e142 @xcite , hermes @xcite and jefferson lab e99 - 117 @xcite experiments.,title="fig:",width=302 ] to explore the effect of the @xmath1 constraints ( [ eq : x1limit ] ) on pdfs in the context of the jam global analysis , we consider a modified fit in which the @xmath36 and @xmath37 ratios are both forced to unity at @xmath63 .
the resulting fit , denoted by `` jam+ '' in fig .
[ fig : qplus ] and table [ tab : chi2 ] , shows that one can indeed obtain a reasonable description of data , consistent with the pqcd @xmath1 limit , with an overall @xmath64 that is only slightly larger than for the unconstrained fit .
this is confirmed also in fig .
[ fig : data ] , where the jam and jam+ fits are compared with data on the @xmath58 asymmetries for the proton and @xmath4he from slac , hermes , and jefferson lab .
the increase in @xmath35 is associated with the reduced magnitude of @xmath7 in the intermediate-@xmath0 region , @xmath65 , which in order to maintain the normalizations required by the triplet and octet axial vector charges @xcite , becomes slightly more negative ( with larger magnitude ) at smaller @xmath0 , @xmath66 , where considerably more data exist .
the @xmath34 distribution , on the other hand , which was already large and positive in the jam fit , undergoes relatively little change with the @xmath63 constraint .
note that the constraint ( [ eq : x1limit ] ) can not be accommodated by the simp parametrization , as without nonzero @xmath52 or @xmath53 terms in eq .
( [ eq : jam ] ) the @xmath7 distribution can not change sign at any @xmath0 .
interestingly , the turn - over in @xmath7 from negative to positive values occurs at relatively large values of @xmath0 , @xmath67 , which would be challenging to observe experimentally .
this is significantly higher than the turn - over found in the earlier lss analysis @xcite at @xmath68 , which was subsequently found to be in conflict with the neutron asymmetry data from the e99 - 117 experiment at jefferson lab @xcite .
indeed , the existing data tend to disfavor fits with positive @xmath69 quark polarization over the measured @xmath0 range .
we studied this by forcing a zero crossing in @xmath7 at @xmath70 , with the distribution becoming positive for @xmath71 . for the jam fit with @xmath72
the @xmath73 increased slightly compared to the jam+ fit , but the increase was significantly greater , to @xmath74 , when the crossing was set at a lower value , @xmath75 .
of course , the behavior of leading twist pdfs at large @xmath0 is also influenced to some extent by the effect of higher twist corrections , which become more important as @xmath1 .
using either the jam or simp parametrizations _ without _ including the @xmath3 corrections generally results in a significantly worse fit , with @xmath73 values increasing form @xmath76 to @xmath77 in table [ tab : chi2 ] , regardless of whether the constraint ( [ eq : x1limit ] ) is imposed or not .
this supports the findings of ref .
@xcite that the @xmath2 dependence of the data over the range considered here can not be accommodated by the parametric form in eq .
( [ eq : jam ] ) with @xmath2 corrections from @xmath2 evolution only . to address the problem of the rapid rise of @xmath7 at too low values of @xmath0 , avakian _
_ @xcite generalized the pqcd calculations for the @xmath1 behavior of pdfs by considering components of the lowest three - quark fock - state wave function with nonzero orbital angular momentum , @xmath21 , in addition to the usual @xmath22 configurations .
the @xmath21 contributions were found to generate additional terms that behave as @xmath78 at large @xmath0 .
although formally subleading in the @xmath1 limit compared with the dominant @xmath79 contributions expected from the @xmath22 component , numerically the @xmath80 terms can play an important role .
in particular , avakian _
found that by using the pqcd - inspired parametrization from ref .
@xcite supplemented by the subleading @xmath24 terms , the large-@xmath0 asymmetry data could be well fitted while preserving the asymptotic constraints of eq .
( [ eq : x1limit ] ) .
furthermore , the resulting @xmath37 ratio remained negative until @xmath25 , as suggested by the e99 - 117 data @xcite , before rising towards unity at higher @xmath0 values . to explore the importance of the additional @xmath24 terms in the context of a global qcd analysis of all data , at small and high @xmath0
, we use as a basis the simplified parametrization with @xmath81 in eq .
( [ eq : jam ] ) , together with the @xmath80 term inspired by the oam arguments , @xmath82 with arbitrary relative normalization @xmath83 .
it is reasonable as a first approximation to assume that the @xmath84 behavior of the oam - inspired term is the same as the standard term , @xmath85 . to reduce the number of parameters that can be reliably determined from the existing data , we also fix @xmath86 in accordance with the pqcd derivation @xcite , even though the corresponding power of @xmath48 in the first term of eq .
( [ eq : oam ] ) remains a free parameter .
the resulting fit , denoted by `` oam '' in fig .
[ fig : qplus ] , is of comparable quality to the jam fit ( @xmath87 ) , with similar @xmath34 and @xmath7 distributions at moderate @xmath88 , but differing at higher @xmath0 values , where there are no constraints from data .
if one includes in addition the @xmath63 constraint from eq .
( [ eq : x1limit ] ) , the effect on the new constrained fit , labelled `` oam+ '' in fig .
[ fig : qplus ] , is again similar to that on the jam+ fit .
namely , the @xmath7 pdf is forced to become positive at @xmath89 , and the reduced magntitude forces the distribution at smaller @xmath0 values to become more negative in order to preserve the sum rules .
the @xmath34 distribution remains relatively unchanged , and the overall @xmath90 is comparable to that for jam+ .
the oam and oam+ fits with lt contributions only are once again considerably worse than the full fits including higher twist effects , indicating that the need for subleading corrections is independent of the parametric form chosen for the lt component .
the results of the above fits suggest that with the additional flexibility afforded by the @xmath24 terms in eq .
( [ eq : oam ] ) , the current data certainly _ can _ be accommodated with the oam - inspired parametrization . on the other hand
, the jam and jam+ fits based on the standard parametrization in eq .
( [ eq : jam ] ) give perfectly good descriptions of the available data over the entire range of kinematics , and _ do not need _ the introduction of the additional @xmath80 terms .
the constraint from eq .
( [ eq : x1limit ] ) , when imposed on the standard pdfs , can be satisfied without substantially modifying the distributions in the regions constrained by data .
one should also caution , however , that the @xmath24 term in the oam - inspired parametrization ( [ eq : oam ] ) can not at present be directly related to the component of the nucleon s spin carried by the quark orbital angular momentum @xcite .
its appearance in the present analysis serves more to illustrate the possible role played by oam in understanding pdfs at large @xmath0 , and to explore the systematic uncertainties that may arise from different assumptions about the functional forms used for the pdf parametrizations .
fits including only terms with @xmath22 and @xmath21 @xcite , which can be interpreted in terms of relative contributions from different orbital states , will be reported elsewhere @xcite .
a scenario in which one finds qualitatively different fits with comparable @xmath35 values , or fits which differ by amounts that are larger than the uncertainties from the propagation of experimental errors , indicates a lack of information at large @xmath0 , and an underestimate of the systematic errors in this region . in the absence of clearer theoretical constraints at @xmath91
, the problem can be best addressed of course by the availability of new data at higher @xmath0 values than are currently available , which we discuss in the next section .
constraining the behavior of the polarization asymmetries @xmath58 , and consequently of the spin - dependent pdfs , in the limit as @xmath1 is one of the featured goals of the experimental physics program planned for the 12 gev energy upgraded cebaf accelerator at jefferson lab .
data from several experiments are expected to be collected for values of @xmath0 as high as @xmath92 for dis kinematics @xcite , and even higher @xmath0 in the nucleon resonance region .
this should significantly reduce the pdf uncertainties for @xmath93 , especially for the @xmath7 distribution , which will be more strongly constrained by new data on the @xmath4he asymmetry . *
( left ) * and @xmath7 * ( right ) * pdfs for the jam fit at @xmath38 gev@xmath39 ( gray band ) and for jam including pseudodata expected from planned jefferson lab 12 gev experiments @xcite ( red hashed area).,title="fig:",width=302 ] * ( left ) * and @xmath7 * ( right ) * pdfs for the jam fit at @xmath38 gev@xmath39 ( gray band ) and for jam including pseudodata expected from planned jefferson lab 12 gev experiments @xcite ( red hashed area).,title="fig:",width=302 ] to estimate the possible impact of the new jefferson lab data we use the projected statistical and systematic uncertainties for the proposed experiments at the @xmath0 and @xmath2 values where the asymmetries will be measured @xcite .
the pseudodata are generated by randomly distributing the central values of the points around the jam fit in fig .
[ fig : data ] for hydrogen , deuterium and @xmath4he targets ( distributing them around any of the other fits considered in this analysis would be equally suitable ) .
the reduction in the pdf uncertainties , illustrated in fig .
[ fig:12gev ] , is significant , with the relative error on @xmath34 and @xmath7 decreasing by @xmath94 for @xmath95 at the input scale @xmath38 gev@xmath39 .
reductions in the spin - dependent pdf errors such as these , combined with similarly dramatic reductions expected for the uncertainty on the unpolarized @xmath69 quark distribution ( or the @xmath96 ratio ) @xcite , should at the very least allow one to discriminate between a @xmath15 ratio that remains negative , as in simple quark models , and one that approaches @xmath97 in the @xmath1 limit , as predicted by pqcd arguments . beyond this
there may be additional constraints on the @xmath1 behavior of spin - dependent pdfs from an electron - ion collider @xcite , particularly if the spectator tagging technique @xcite in semi - inclusive dis from the deuteron or @xmath4he can be extended to polarized beams and targets .
the aim of this analysis was to investigate whether existing data from polarized lepton nucleon dis is able to provide any constraints on the @xmath1 behavior of spin - dependent pdfs in the context of a global qcd analysis .
using the recent jam fit as a baseline , we showed that demanding the polarized to unpolarized pdf ratios @xmath98 to approach unity at @xmath63 results in equally good fits to the available data , even though the resulting changes to the @xmath7 pdf are significant in the intermediate-@xmath0 region . with dramatically different behaviors for the @xmath37 ratio allowed for @xmath65 ,
this highlights the critical need for precise data sensitive to the @xmath69 quark polarization at large @xmath0 values .
we have investigated the recent suggestion that inclusion of fock states in the nucleon wave function with nonzero orbital angular momentum gives rise to additional contributions to pdfs proportional to @xmath23 @xcite which could play an important role numerically . employing an extension of the typical functional form used in standard pdf analyses which allows for the @xmath80 dependence , we find that the generalized parametrization is also able to provide a good description of the existing dis data , with or without the @xmath63 constraint . while there has been a first indication of a rise above unity of the neutron ( @xmath4he ) polarization asymmetry for @xmath99 @xcite , the data still generally prefer a negative @xmath7 distribution at large @xmath0 even with the @xmath63 limit built in , although the cross over to positive values depends on the parametrization chosen ( at @xmath100 for the jam+ and @xmath101 for oam+ ) .
further progress on this problem is expected soon with new data expected from several experiments at the 12 gev energy upgraded jefferson lab , which aim to measure polarization asymmetries of protons , deuterons and @xmath4he up to @xmath102 in dis kinematics @xcite . using the projected statistical and systematic errors from these experiments , we explored the possible impact on the pdfs and their uncertainties in this region .
we find reductions in both the @xmath34 and @xmath7 pdfs of up to @xmath103 for @xmath104 in the jam fit , with significant reductions also at smaller @xmath0 values .
this should considerably narrow the range of possible asymptotic @xmath1 behaviors of the @xmath98 ratios , and for the first time provide critical tests of the various theoretical scenarios that have been proposed to describe pdfs in the large-@xmath0 region @xcite .
we thank a. accardi for helpful discussions , and j .-
chen , s. kuhn , g. schnell , b. wojtsekhowski and x. zheng for communications about existing and future polarization asymmetry experiments .
this work was supported by the doe contract no .
de - ac05 - 06or23177 , under which jefferson science associates , llc operates jefferson lab .
jefferson lab experiments pr12 - 06 - 109 , s. kuhn _
et al . _ ; pr12 - 06 - 110 , j .-
chen _ et al . _ ; pr12 - 06 - 122 , b. wojtsekhowski _ et al .
_ , spokespersons .
jefferson lab experiments c12 - 10 - 103 [ marathon ] , g. g. petratos _ et al . _ ; e12 - 10 - 102 [ bonus12 ] , s. bltmann _ et al . _ ; e12 - 10 - 007 [ solid ] , p. souder , spokespersons . | we investigate the behavior of spin - dependent parton distribution functions ( pdfs ) at large parton momentum fractions @xmath0 in the context of global qcd analysis .
we explore the constraints from existing deep - inelastic scattering data , and from theoretical expectations for the leading @xmath1 behavior based on hard gluon exchange in perturbative qcd .
systematic uncertainties from the dependence of the pdfs on the choice of parametrization are studied by considering functional forms motivated by orbital angular momentum arguments .
finally , we quantify the reduction in the pdf uncertainties that may be expected from future high-@xmath0 data from jefferson lab at 12 gev . |
in order to construct an inflation model in supergravity ( with slow roll conditions ) successfully , one must at least explain observations appropriately @xcite .
if this is the case , the model may have possibilities to predict objects of supergravity properly , such as gravitino mass , gaugino mass and the masses of superpartners of scalar fields . if the predicted values are experimentally testified , then the supergravity model has some plausibility as a true theory . in this paper
we will propose a new modular invariant model , which satisfies the above requirement .
it is convenient to introduce the dilaton superfield @xmath0 , a chiral superfield @xmath4 ( gaugino condensate superfield ) , and the modular superfield @xmath5 . here , the other matter fields are set to zero for simplicity .
moreover , we identify the inflaton field with the dilaton . because the inflation concerns planck scale physics , the dilaton seems the most appropriate candidate for the inflaton field .
( though we have exclusively restricted our attention to a model similar to ref.@xcite , other models derived from another type of compactification seem very interesting . among them
, the kklt model @xcite attracts our interest , where the moduli of the superfield @xmath5 play essential roles . )
seven cases of parameter choices will be discussed here , among which some examples show the possibility of observation of gauginos by lhc experiments , which will give some hints of identity of dark matters .
+ we first review the inflationary parameters obtained in the slow roll approximation ( sra ) @xcite .
the slow - roll parameters ( in planck units @xmath6 ) are defined by : @xmath7 the slow - roll condition is well satisfied , and the @xmath8-problem can just be resolved as shown below .
using sra , number of @xmath9-folds @xmath10 at which a comoving scale @xmath11 crosses the hubble scale @xmath12 during inflation is given by : @xmath13 where the integral extends from @xmath14 to @xmath15 , where @xmath14 is the value of @xmath0 at the end of inflation and @xmath15 is that at the beginning of inflation .
next , a scalar spectral index @xmath16 for the scale dependence of the spectrum of density perturbation and its tilt @xmath17 are defined by : @xmath18 these are approximated in the slow - roll paradigm as : @xmath19 where @xmath20 is an extra slow - roll parameter that includes the third derivative of the potential as follows : @xmath21 substituting @xmath15 into these expressions , we get finally the following estimate for the spectrum of the density perturbation @xmath22 caused by slow - rolling dilaton : @xmath23 the spectrum of the density perturbation of tensor type @xmath24 is given by @xmath25 the spectral index @xmath26 is then given as @xmath27 the tilt is given by @xmath28 moreover , the ratio between scalar power spectrum and tensor one , which is denoted by @xmath29 , is also very important : @xmath30 the upper limit @xmath31 is given by seven years observation by wmap .
we hope that the planck satellite might give the direct signal of primordial tensor perturbation soon . or ,
if the observation of primordial non - gaussianity indirectly proves the contribution of tensor mode from nonlinear terms of the perturbation spectrum , the value of the ratio will be improved .
the current @xmath32 limit from the 7-year data of wmap is the primordial non - gaussianity parameter @xmath33 , and the planck satellite is expected to reduce the uncertainty by a factor of four in a few years from now@xcite .
we will show the prediction of the ratio corresponding to each parameter choice in tables [ tab : infparam1 ] and [ tab : infparam2 ] .
seven cases of parameter choices will be discussed in this paper , which fit very well with the 7-year wmap observations .
it is the end of inflation , when the slow - roll parameter @xmath34 or @xmath35 reaches the value 1 . after passing through the minimum of the potential
, reheating will begin .
+ in section 2 , a modified string - inspired modular invariant supergravity is newly proposed . in section 3 , the inflationary cosmology and inflationary trajectory of this model
are presented , in which the numerical values are explained by using the case 1 of parameters . in section 4 , super higgs mechanism and gravitino production from heavy scalar field
are discussed . in section 5 , reheating temperature is estimated by calculating the decay rate of inflaton into gauginos in minimal supersymmetric standard model ( mssm ) . in section 6 , numerical predictions by tuning parameters are explained and shown in tables [ tab : infparam1 ] and [ tab : infparam2 ] .
finally , a short summary is given in section 7 .
@xmath36 , @xmath37 supergravity from @xmath38 heterotic string by dimensional reduction has no - scale structure with @xmath39 gauge group @xcite .
the modular invariant supergravity model was proposed , where the khler potential and the superpotential are given as : @xmath40.\end{aligned}\ ] ] here @xmath8 is dedekind s @xmath8 function , @xmath41 is a free parameter of the theory , @xmath0 is the dilaton , @xmath5 is the moduli and @xmath4 is a complex scalar superfield defined by the gaugino condensation @xmath42 of the @xmath43 hidden sector@xcite .
the renormalization group parameter @xmath44 can correspond to the @xmath43 hidden sector gauge group .
however , if we derive the scalar potential from these formulae in the einstein frame , it seems impossible to solve difficulties such as the @xmath8-problem and/or the negative vacuum energy .
we would like to propose a new modular invariant @xmath36 supergravity , which solve such problems .
a modified string - inspired modular invariant supergravity is proposed here to apply it to inflationary cosmology .
+ in the original model , the massless goldstino was given by the the dilatino @xmath45 as shown in our former study , because @xmath46 , where @xmath47 is defined@xcite by @xmath48 and @xmath49 as @xmath50m^j-\gamma_k^{ij}m^k,\qquad \gamma_i^{jk}=g_{\ \ \ i}^{-1l}\partial^jg_l^k.\end{aligned}\ ] ] . in order to extend the original model to realize the slow roll inflation with @xmath46 ,
the simplest choice is to add a term linear term in @xmath0 , like @xmath51 : @xmath52,\end{aligned}\ ] ] where @xmath53 and @xmath54 are new parameters that should be determined from observations .
then the scalar potential @xmath55 ) $ ] is obtained as : + @xmath56 \right|^2 \right .
\nonumber \\[5pt ] & & + \frac{1}{t+t^*-|y|^2 } \left| \strut \alpha + \beta s + 3 b y^3 \ln \left [ c\>e^{s/3b}\>y\eta^2(t ) \right ] - ( s+s^ * ) ( y^3 + \beta )
\right|^2 \nonumber \\[5pt ] & & + 6 b^2 |y|^6 \left\ { \left ( 1- \frac{\alpha + \beta s^*}{b{y^*}^3 } \right ) \frac { \eta ' ( t ) } { \eta ( t ) } + \left ( 1- \frac{\alpha + \beta s}{by^3 } \right ) \frac { \eta ' ( t^ * ) } { \eta ( t^ * ) } \right .
\nonumber \\[5pt ] & & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \left .
\left . + 2 ( t+t^ * ) \left| \frac { \eta ' ( t ) } { \eta ( t ) } \right|^2 \right\ } \right],\end{aligned}\ ] ] which is explicitly modular invariant in @xmath5 . instead of imposing @xmath57
, we will assume @xmath58 , which is a rather good approximation , i.e. , @xmath59=0,\end{aligned}\ ] ] then a relation between @xmath0 and @xmath4 is obtained as follows : @xmath60 in order to check the validity of this approximation , numerical estimations of @xmath61 are shown at tables [ tab : infparam1 ] and [ tab : infparam2 ] .
because inflation is concerned with planck scale physics , the dilaton can be one of the strong candidates to be identified with the inflaton@xcite .
we will show here only the results from case 1 among the seven parameter choices for @xmath53 , @xmath54 , for which the potential @xmath62 at @xmath63 has a stable minimum .
results for the other cases are summarized at tables [ tab : infparam1 ] and [ tab : infparam2 ] .
+ first we show the result with the parameter choice , @xmath64 , @xmath65 and @xmath66 ( case 1 ) .
hereafter we fix @xmath63 ( @xmath67 , @xmath68 , @xmath69 ) and @xmath44 corresponding to the @xmath43 gauge group .
the minimum of the potential is given by @xmath70 the parameters of inflation are predicted as follows : @xmath71 the gravitino mass @xmath72 and and susy breaking scale @xmath73 are predicted as : @xmath74 we show the potential @xmath75 minimized with respect to @xmath4 in fig .
1 , the evolution of the slow - roll parameters in fig . 2 , and the stability of the potential minimum at @xmath63 in fig .
3 . + this case seems to explain the wmap observations well .
the slow - roll parameters ( in planck units @xmath6 ) are defined by @xmath76 the slow - roll condition is well satisfied , and the @xmath8-problem can be resolved . using the sra
, the number of @xmath9-folds at which a comoving scale @xmath11 crosses the hubble scale @xmath12 during inflation is given by : @xmath10 is also calculated by : @xmath77 by integrating from @xmath14 to @xmath15 , fixing the parameters @xmath78 as well as @xmath79 .
that is , our potential has the ability to produce the cosmologically plausible number of @xmath9-folds .
the scalar spectral index @xmath80 for the scale dependence of the spectrum of density perturbation , and its tilt @xmath81are estimated by substituting @xmath15 into these formula : @xmath82 & & \alpha_{s*}=\frac{dn_s}{d\ln k}\sim -4.314 \times 10^{-4}. \end{aligned}\ ] ] finally , estimating the spectrum of the density perturbation caused by the slow - rolling dilaton , we obtain : @xmath83 the tensor power spectrum @xmath84 and its ratio @xmath29 to @xmath22 are estimated as @xmath85 & & r=\left ( \frac{\mathcal{p_{t^*}}}{\mathcal{p_{r^ * } } } \right ) = 6.755 \times 10^{-2}.\end{aligned}\ ] ] . ]
@xmath86 + . ] @xmath86 + . ] @xmath86 + for the potential at the minimum for the case 1 , moreover , the energy scale is @xmath87 , which is non - negative .
the results for the remaining six cases are almost identical to case 1 , except the supergravity proper quantities , which are shown in tables [ tab : infparam1 ] and [ tab : infparam2 ] .
it is the end of inflation , when the slow - roll parameter @xmath34 or @xmath35 reaches the value 1 . after passing through the minimum of the potential
, reheating will begin .
let us consider the super higgs mechanism in our model .
the inflatino field @xmath45 with mass @xmath88 , which is the susy partner of the inflaton ( dilaton ) field @xmath0 , can play the role of the higgsino field .
because the metric elements satisfy @xmath89 in the khler metric @xmath90 , @xmath0 does not mix with @xmath91 .
then the terms that cause the super higgs mechanism ( shm ) are selected as @xmath92 first , we calculate the first , the fourth and the sixth terms as follows : @xmath93 now the last term of eq.([shm ] ) implies the mass of @xmath45 , which is proved to be exactly zero in our model .
the first term can be identified with the mass term of the massive gravitino field , whose mass is given by @xmath94 .
the second , the third and the fifth terms are the hermite conjugates of eq.([shm ] ) , and similar expressions can be derived .
this is the scenario of the super higgs mechanism .
the predicted values of the gravitino mass @xmath95 and the scale of susy breaking@xcite @xmath96 in our model are summarized in table [ tab : infparam1 ] .
now we will investigate the gravitino production from heavy scalar bosons after inflation .
the interaction terms between the scalar fields @xmath97 and the gravitino @xmath98 in the total lagrangian density of supergravity are selected as follows @xcite : @xmath99 these interaction terms are expanded in terms of the shifts @xmath100 from the stable values for each of the @xmath97 s , i.e. , @xmath101 . in our model
there are three scalar fields @xmath102 corresponding to the @xmath97 s .
@xmath102 are canonically normalized by @xmath103 , where @xmath104 are the coefficients of canonical normalization . since @xmath105 s
are also affected by the normalization , the normalized @xmath106 s are replaced with @xmath107s , and the @xmath104 s are replaced by the @xmath108 s at the stable points @xmath109 . by using these formula and the general relation for the gravitino mass @xmath110
, the interaction terms are obtained as : @xmath111\psi_\nu.\end{aligned}\ ] ] the helicity @xmath112 part of the massive gravitino is defined by the tensor product of a vector and a spinor as @xcite @xmath113 where the coefficient of each term is a clebsch - gordan coefficient , @xmath114 is the wave function of vector field with helicity @xmath115 , and @xmath116 is the spinor wave function with helicity @xmath117 . the decay rate @xmath118 is obtained as @xcite : @xmath119 after the scalars @xmath102 are canonically normalized and the masses diagonalized , the mass eigenstates are denoted by @xmath120 , and the masses are calculated as @xmath121 m_{t '' } & = & 2.265 \times 10^{12 } { \rm gev},\end{aligned}\ ] ] for the case 1 .
therefore , we obtain for the decay process @xmath122 , by using the above formula : @xmath123 \tau ( y '' \rightarrow \psi_{3/2}+\psi_{3/2 } ) & = & 1.376 \times 10^{-29 } \,\ , { \rm sec}.\end{aligned}\ ] ] this process occurs almost instantly . for the other cases ,
see tables [ tab : infparam1 ] and [ tab : infparam2 ] .
as an example , the decay rate of @xmath124 into gauginos is estimated in our model . by using the term @xmath125 , the interaction between @xmath0 and the gauginos @xmath126 s
are given as @xmath127-\frac{1}{2}f^i_{ab}(\phi)\tilde{\mathcal{d}}_\mu\left[\lambda^a\sigma^{\mu}\bar{\lambda}^b\right ] \nonumber \\ & & \qquad -\frac{1}{4}\frac{\partial f_{ab}(\phi)}{\partial \phi}e^{k/2}g_{\phi\phi^*}d_{\phi^*}w^*\lambda^a\lambda^b + \frac{1}{4}\left(\frac{\partial f_{ab}(\phi)}{\partial \phi}\right)^*e^{k/2}g_{\phi\phi^*}d_{\phi}w\bar{\lambda}^a\bar{\lambda}^b .
\label{gaugino_decay}\end{aligned}\ ] ] by looking at the first term of ( [ gaugino_decay ] ) , the @xmath126 s are also canonically normalized as @xmath128 .
the interactions come from the third and fourth terms .
the terms include @xmath129 , which implies the auxiliary field of @xmath130 in global susy theory and it is replace by @xmath131 .
+ by expanding @xmath132 around the stable point , the interaction terms are given as @xmath133\lambda^a\lambda^b + { \rm h.c . } .\end{aligned}\ ] ] when @xmath134 , @xmath73 implies the ssb scale of the model and is estimated as @xmath135 , since @xmath136 and @xmath137 take values close to the planck scale .
therefore , as the first term is far smaller than the second and actually negligible , @xmath138 remains .
the derivative term of @xmath139 can be replaced by @xmath140 . then the decay rate @xmath141 can be estimated as : @xmath142 the reheating temperature @xmath143 is derived from the boltzmann equation by using the decay rate , and is given by @xmath144 where @xmath145 is the number of effective degrees of freedom of mssm , i.e. @xmath146 and numerically given above by inserting the decay rate from the canonically normalized inflaton field @xmath124 .
+ by using the relation @xmath147 , which holds for the mass of susy particles ( sp)@xcite , the gaugino masses are estimated as + @xmath148 then the decay rate of @xmath149 and the reheating temperature are estimated as @xmath150 because the primordial gravitinos decay very rapidly and the reheating temperature is lower than the gravitino mass , gravitino reproduction will not occur after reheating .
the the effect on the standard big bang nucleosynthesis ( bbn ) scenario @xcite ) may be negligible in our model ( see also for the bound of reheating temperature@xcite ) .
all these numerical values are shown in tables [ tab : infparam1 ] and [ tab : infparam2 ] , and will be explained in next section .
the numerical predictions of inflationary variables coming from the scalar potential in sra are shown in tables [ tab : infparam1 ] and [ tab : infparam2 ] .
those values are extremely well fitted to the wmap observations for all the cases 1 to 7 . here
it will be instructive to revisit our scenario , which should be investigated by experiments .
the renormalization group @xmath152 function is assumed to be fixed , corresponding to the @xmath43 hidden sector gauge group , i.e. , @xmath153 because the @xmath54 function is defined as @xmath154 , therefore , @xmath155 .
next we also fix the modular field at @xmath63 , where the potential is required at least to be stable . varying the parameters @xmath151 and @xmath41
, we can see that the slow roll inflation parameters are almost completely compatible with 7-year wmap observations in the seven cases we have chosen . in these numerical estimates of inflation , the parameter @xmath54
is intimately related to the value of the power spectrum @xmath156 .
we find @xmath157 which realizes the slow roll inflation to explain the wmap observations almost exactly .
the other two parameters @xmath53 and @xmath41 are essentially playing the roles to determine the prediction of the susy part of the theory , such as gravitino and gaugino masses . the parameter @xmath53 is fixed to @xmath158 from case 2 to 7 . the inflationary variables @xmath159 , @xmath160 , @xmath161 , and @xmath156 are then calculated .
all the values fit extremely well to the wmap observations .
the values of @xmath162 and @xmath29 are new predictions of our present model , in which the @xmath29 s appear to be about @xmath163 in all cases .
we expect that the planck satellite will prove these predictions soon .
we estimated the values of variables of supergravity , i.e. , the gravitino mass @xmath164 , the susy breaking scale @xmath165 , the order of the gaugino masses @xmath166 , and the reheating temperature @xmath167 , as well as the decay rates of gravitino and heavy scalar @xmath4 .
the dependences on the parameter @xmath41 of the susy sector can be seen in the tables , among which some cases show the possibility of gauginos to be observed by lhc or other experiments aiming to search dark matter . the validity of our model may be clearly checked by the experiments .
the parameter dependence of our model will be explained more thoroughly in a separate paper .
+ we summarize our results for the seven cases in tables [ tab : infparam1 ] and [ tab : infparam2 ] .
.predictions of inflation parameters and susy particles ( @xmath64 , @xmath168 ) [ cols="^,^,^,^,^,^",options="header " , ] @xmath86 +
by modifying the original string - inspired modular invariant supergravity , which are known to explain wmap observations appropriately , a mechanism of ssb and gravitino production just after the end of inflation as been investigated .
the model we used cleared the @xmath8-problem and the negative energy problem of the potential at the stable point , and appeared to reproduce successfully the observations on the inflation era . the assumption to set @xmath58 instead of imposing @xmath57 in order to determine the minimum of the scalar potential , which seems to be a rather good approximation , might give rise to some problems .
however , we were able to check that @xmath169 itself has a very small susy breaking contribution and therefore is no flat direction .
seven cases of parameter choices have been discussed here , among which some examples ( cases 2 to 5 ) show the possibility of gauginos to be observed by lhc experiments , which will give some hints on the identity of dark matters .
we would like to emphasize that the ratio @xmath170 between the scalar power spectrum and the tensor one is predicted to be almost the same for all seven cases , and this value seems to be in the range possibly observed by the planck satellite soon .
+ because the supergravity seems the most plausible framework to explain new physics , including undetected objects , the supergravity model of inflation which here described will open the hope to construct a realistic theory of particles and cosmology in this framework .
m. j. hayashi , t. watanabe , i. aizawa and k. aketo , mod .
lett . a * 18 * ( 2003 ) , 2785 , hep - ph/0303029 . + m. j. hayashi and t. watanabe , _ proceedings of ichep 2004 , beijing _ ( world scientific , 2005 , pp.423 - 426 ) , hep - ph/0409084
. m. j. hayashi , s. hirai , t. takami , y. okame , k. takagi and t. watanabe , int .
j. mod .
a * 22 * , ( 2007 ) 2223 ; _ frontiers of fundamental and computational physics _
( american institute of physics , 2008 , pp.74 - 79 ) .
j. m. bardeen , phys .
d * * ( ) , .
+ h. kodama and m. sasaki , prog .
+ m. sasaki , prog .
phys . * * ( ) , .
+ v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .
rep . * * ( ) , . | a new modified string - inspired modular invariant supergravity model is proposed and is applied to realize the slow roll inflation in einstein frame , so that the model explains wmap observations very well .
gravitino mass and their production rate from scalar fields are estimated at certain values of parameters in the model .
seven cases of parameter choices are discussed here , among which some examples show the possibility of observation of gauginos by lhc experiments , which will give some hints of identity of dark matters .
the reheating temperature , which is estimated by the stability condition of boltzmann equation by using the decay rates of the dilaton @xmath0 into gauginos , is lower than the mass of gravitino .
therefore no thermal reproduction of gravitinos happens .
the ratio between the scalar and tensor power spectrum is predicted to be almost the same for the seven cases under study , and its value @xmath1 seems in the range possibly observed by the planck satellite soon .
the plausible supergravity model of inflation , which will be described here , will open the hope to construct a realistic theory of particles and cosmology in this framework , including yet undetected objects . * dilatonic inflation , gravitino and reheating + in modified modular invariant supergravity * + yuta koshimizu@xmath2 , toyokazu fukuoka@xmath2 , kenji takagi@xmath3 , hikoya kasari@xmath2 + and mitsuo j. hayashi@xmath2 + @xmath2_department of physics , tokai university , hiratsuka , 259 - 1292 , japan _ + @xmath3_company vsn , minato , shibaura , 108 - 0023 , japan _ + e - mail : [email protected] |
constitutive equations in polymer melts or concentrated solutions are mathematical relationships between the stresses and the external flow conditions .
they represent the inherent properties of the polymer system , and should be derived from the knowledge of polymer chain structures , configurations , interactions and polymer dynamics . in practice , a constitutive model is needed to idealize these microscopic characteristics of polymer chains , due to the complex interactions of this many - chain system ( larson , 1988 ) . as polymer melts own some characteristics of universality ,
that is to say , some physical properties of polymer melts are independent on the microscopic chemical structures , we can coarse grain a real polymer chain as a smooth thread connected by segments ( watanabe , 1999 ) . this coarse - grained chain , just like the real chain , has an enormous number of configurations and can be usefully described statistically as gaussian chain or non - gaussian chain .
in fact , the dynamics in our study is the segmental dynamics , which is based on a spatial scale of a chain segment and those chains are only distinguished by their lengths and abilities of extension .
the interactions between chains are intricate .
however , the attractive interactions between segments are tended to screen out by the excluded volume effect in polymer melts or concentrated solutions ( de gennes , 1979 ) , leaving the topological constraints to play a dominant role in such a system at equilibrium .
a basic understanding of chain dynamics is the key - point to develop constitutive equations for melts or concentrated solutions of flexible polymers .
the proposition of the idea of ` reptative ' motion by de gennes ( 1971 ) is a milestone in the development of polymer dynamics of the concentrated system .
in fact , in a flow the long linear polymer chains become oriented and stretched , which generates internal stress as the response to external disturbance .
however , the orientation and stretch of chains will relax in the course of time due to the motion of non - crosslinked polymer chains .
the major mechanisms of relaxation concerned are : reptation ( doi and edwards , 1978a , b , c , 1979 ) , convection of segments along primitive chains ( doi , 1980 ; marrucci , 1986 ) , fluctuations of the contour length ( doi and edwards , 1986 ; pearson and helfand , 1984 ; mead et al . , 1998 ) , reptative constraint release ( marrucci , 1985 ; des cloizeaux , 1988 , 1990 ; leygue et al . , 2005 , 2006a , b ) , and convective constraint release ( ccr ) ( marrucci , 1996 ; ianniruberto and marrucci , 1996 ; milner et al . , 2001 ; graham et al . , 2003 ) .
the first three kinds of relaxation are caused by the motion of the primitive chains , which can happen only at the ends ; while the other two are induced by the motion of surrounding chains , which can occur not only at their ends but also in some other place along the primitive chains .
the optical - stress law guarantees a linear relationship between the stress and the orientational anisotropy of segments for a highly crosslinked rubber system ( treloar , 1975 ) .
the subsequent studies also show its validity for non - crosslinked polymer system ( janeschitz - kriegl , 1975 ) .
the segments in the crosslinked rubber refer to chain portions between two permanently tethered crosslink points .
the orientations of these segments are uniform .
the segments in the non - crosslinked polymer system refer to entanglement segments .
the orientations of these segments are non - uniform , as the segments can move freely . under fast flow
the polymer chains not only become oriented but also significantly stretched .
the contribution from segmental stretch to stress is proportional to the square of stretch ratio for gaussian chains ( doi and edwards , 1986 ) , while this relationship is much more complex for non - gaussian chains . in the original doi and edwards ( de ) model , due to the faster relaxation of segmental stretch
, the chain is assumed to remain its equilibrium length all the time .
the stress contribution comes only from segmental orientation .
de theory predicts a maximum in shear stress @xmath0 with respect to the shear rate , and followed by a sharp decrease asymptotically as @xmath1 , which has not been observed by experiments .
the neglect of segmental stretch results in a monotonic increase in the first normal stress difference in the start - up of a simple shear flow , however , an overshoot appears in experiment . in order to improve this deficiency of the original de model , several `
chain stretch ' models are proposed . in the uniform stretch model , it is assumed a uniform segmental stretch along the chain , such as larson s ` partially convected strand ' model ( 1984 ) and also the one proposed by pearson et al . , ( 1989 ) .
the non - uniform segmental stretch model is proposed by marrucci and grizzuti ( 1988 ) .
the segmental stretch models are called as demg model ( mead and leal , 1995a , b ) .
the extension of chain segments creates some segment ( or saying pseudo _ defect _
( de gennes , 1971 ) ) sources along the chain , which is equivalent to reduce the diffusion coefficient .
the inclusion of chain stretch also changes the feature in the start - up of shearing .
a serious deficiency in all these models is that in the terminal region excessive shear thinning is inevitable due to the lacking of an extra relaxation mechanism to increase the rate of creating new segments , which is equal to the rate of annihilating old segments , or equivalently saying to avoid high orientation of the chain to the direction of flow .
the mead - larson - doi ( mld ) model incorporating ccr to the demg model qualitatively improves the description of the phenomenon of excessive shear thinning and other rheological properties in steady state and transient shearing flow ( mead et al . , 1998 ) .
recently pattamaprom and larson ( 2001 ) proposed a toy mld model by making an extension of the mld model in a simple way and adding reptative constraint release to mld equations .
in all these models , the rate of ccr is evaluated at the tube end , which is a uniform rate of ccr .
ccr represents the ability of the flow to convect or release entanglements , and is determined by segmental orientation and stretch . as the segmental orientation and stretch in the non - crosslinked polymer melts are non - uniform , the ccr is non - uniform .
the rate of ccr is a function of both the contour variable and time .
we assume that when the entanglement is convected away from the primitive chain segment to some other place , the previously formed environment has changed and this entanglement then becomes released .
based on this idea , we develop a new contour - variable model including non - uniform segmental stretch , non - uniform rate of ccr and non - gaussian chain to describe the rheological behavior of linear chain entangled system in simple shearing and transient flows . however , we dropped the reptative constraint release term in the present work as we focus on the high - shear - rate region , where the segmental orientation and stretch are obvious . in the low - shear - rate region ,
the reptative constraint release occurs on a time scale of the reptation time of the whole chain and one event only relaxes a small part of the chain ( graham et al . ,
2003 ) ; while in the high - shear - rate region , the ccr dominates .
thus in both of these two cases it is valid to drop the reptative constraint release term . while in some middle region
, some quantitative deviations occur if we drop the reptative constraint release term .
we also neglected the contribution from the fluctuations of the contour variable , as the problem is considered in the mean - field level .
this paper is organized as follows . in section [ sec .
2 ] we develop a contour - variable dependent rate of ccr to describe the flow induced constraint release and construct the new equations which the segmental stretch and the tube survival probability satisfy , respectively .
we promote our contour - variable model of constitutive equations in the end of this section . in section [ sec .
3 ] we give the results and discussion , in which we focus on the simple steady shear flow and the start - up of steady shear flow . in section [ sec .
4 ] , we give the conclusions . the process and methods used for the calculation of constitutive equations are shown in the appendix .
ccr represents the entanglements relaxation mechanics caused by the convective motion of surrounding chains .
it is not important when the concentrated system is at rest or undergoes a slow flow .
however , when the velocity of the flow is comparable to the inverse of reptation time of the primitive chain , ccr will play a key role in determining the dynamics of the system .
taking into account the ccr mechanics , ianniruberto and marrucci ( 1996 ) proposed a contour location independent rate of ccr . in this model , the rate of ccr is only caused by the retraction of the primitive chain ends . so
the rate of ccr is proportional to the velocity gradient of the movement of the primitive chain ends .
however , not all convections or deformations will release constraint . in order to exclude the case of affine deformations , in which no constraint release occurs , mead et al .
gives the rate of ccr as the difference between the rate of convection of the entanglements and the rate of primitive chain retraction ( mead et al . , 1998 ; viovy et al . , 1983 ) .
they only evaluate the rate of ccr at the chain ends , since they think the topological constraint can be released only by the motion of chain ends ( ianniruberto and marrucci , 1996 ; marrucci et al . , 2001 ) .
( we call it the end - relaxation mechanism in the follows . ) in the present work , we focus on the non - uniform behavior of ccr .
let us consider a primitive chain segment confined by a tube segment which was created at past time @xmath2 , as called the @xmath2-segment .
the deformation of this chain segment will be released when the entanglements forming the tube segments are convected away by the flow .
it is an irreversible process .
this @xmath2-segment will never be created again .
when this happens , some hidden entanglements become active ( ianniruberto and marrucci , 2000 ) , which instantaneously makes the released part ( e.g. the @xmath2-segment ) become a part of a new tube .
thus we can say that the rate of creating a new segment is the same as that of annihilating an old segment .
the contribution to the stress for a specified tube only comes from the part that remain unreleased .
the released part of the original tube , which becomes a part of a new tube , will still have contributions to the stress .
however , the stress should be calculated from the new tube .
define @xmath3 as the equilibrium contour variable , which runs from @xmath4 to @xmath5 , where @xmath6 is the equilibrium length of the primitive chain .
when the chain is stretched , its primitive length @xmath7 can be larger than @xmath6 .
we then define a local stretch function @xmath8 which takes the value from @xmath9 to @xmath10 , to describe the segmental stretch .
we can define @xmath11 as a local strain .
it describes the local extension .
the rate of ccr @xmath12 can be written as the following form @xmath13 where is the velocity gradient tensor of the flow , which , we suppose , is homogeneous , although in general it is not ; @xmath14 is the orientational tensor , @xmath15 . apparently , it is a local approximation form . the first term in eq .
( 1 ) describes the rate of entanglements convection at contour variable @xmath3 and @xmath16 .
the second term is used to exclude the case of affine deformation , in which there are no environment changes with respect to the primitive chain and no ccr happens . as @xmath17 , @xmath18 , from eq .
( 1 ) we then find no ccr happens at the chain ends of the primitive chain .
in fact the ends of the tube change instantaneously with the ends of the primitive chain during movement , there is no relative motion between the ends of the tube and the ends of the primitive chain . the primitive chain and the tube
are intrinsically synonymous , we treat them as different things only for convenience in describing the motion of the chain considered and the motion of its surrounding chains .
the difficulty of the flow - reversal problem has be pointed out ( wapperom and keunings , 2000 ; ianniruberto and marrucci , 2001 ) .
if the present model reduces to the uniform and single - relaxation - time version , this problem can be amended using the absolute value of @xmath19 in eq .
( 1 ) . in the present non - uniform case
, we suppose that the above expression can amend this difficulty . in the following
we will not write the absolute value sign in the rate of ccr by considering the flow only in one direction .
the dynamics of segmental stretch is described by the equation of motion , in which we neglect the acceleration term .
it is derived from the force balance between the segmental extension force caused by the flow and the entropic elastic force .
if the chain is considered to be non - gaussian , this equation can be written as ( mead and leal , 1995a , b ) @xmath20 \left(\frac{\partial^2 s}{\partial s_{0}^{2}}\right).\ ] ] here @xmath21 is the relative pre - averaged tangential velocity with respect to the center of chain @xmath22 the parameter @xmath23 is the number of entanglements in a primitive chain , where @xmath24 is the molecular weight between two entanglements .
@xmath25 ( doi and edwards , 1986 ) , where @xmath26 is the reptation time and @xmath27 is the rouse relaxation time .
@xmath28 defines the diffusion coefficient along the primitive path , and is related to the reptation time by @xmath29 .
@xmath30 is a coefficient related to non - gaussian behavior , which is given by @xmath31 where @xmath32 , and @xmath33 is the maximum length the primitive chain can be stretched to .
@xmath30 ensures that the magnitude of a fictitious force acting along the primitive chain is equal to @xmath34 when there is no segmental extension ( doi and edwards , 1986 ; marrucci and grizzuti , 1988 ) . here
, @xmath35 is the step length of the primitive chain and is a constant in the present model . a good discussion for its magnitude is given by milner ( 2005 ) .
the inverse langevin function @xmath36 with the fractional extension @xmath37 is used to describe the tension of a non - gaussian finitely extensible polymer segment ( treloar , 1975 ) , which satisfies the following equation @xmath38 expanding @xmath36 in a taylor series and using pad approximation ( cohen , 1991 ) , we can obtain @xmath39 we differentiate both sides of eq .
( 2 ) with respect to @xmath40 and obtain @xmath41\left ( \frac{\partial q}{\partial s_0 } \right ) ^2.\ ] ] insert eq .
( 6 ) into the above equation , we obtain @xmath42 taking into account the effect of ccr and following the argument by mead , larson and doi ( 1998 ) , finally we obtain @xmath43 the parameter @xmath44 comes from the fact that the magnitude of the rate of entanglements reduction due to ccr is two times as large as the rate of segmental retraction . the boundary condition and the initial condition are @xmath45 and @xmath46 respectively .
the boundary condition is specified by the untethered condition of the chain ends which can never be stretched .
the process of the segmental orientation can be described by the tube survival probability @xmath47 , which means that a tube segment created at past time _
_ t__@xmath48 still can be found at location @xmath49 and time @xmath16 . in the original de model
, @xmath47 satisfies a diffusion equation .
when the segmental stretch is taken into account , such as the demg model , a convective term has to be added to the de model .
this term accounts for the flow induced convection of the _ defects _ to the ends of the primitive chain and the elongation of the tube segments .
however , the orientation relaxation induced by constraint release has not been considered . without
that it makes the prediction of the rheological properties by these models qualitatively different from the experimental results . based on our understanding for the effect of constraint release , we modify the ccr term in the equation of the probability of tube survival in mld model and drop the term which comes from the reptative constraint release . including the contribution from chain stretch , the equation for the probability of the tube survival is given by @xmath50 where @xmath51 is the tube strain function . the boundary condition and the initial condition are @xmath52 and @xmath53respectively
it is convenient to transform the independent variables @xmath49 and @xmath16 to other independent variables @xmath40 and @xmath16 ( pearson et al . , 1991 ) .
@xmath54 where @xmath55 is the function of local strain .
when we consider the ccr mechanism , a new term must be added to eq . ( 11 ) , which is @xmath56 here , @xmath57 is the orientational tensor .
then we obtain @xmath58 where @xmath59 is the switch function ( mead et al . , 1998 ) .
when @xmath60 is large , the contribution from the last term in the right hand side of eq .
( 12 ) is much smaller .
however , when @xmath60 approaches to unity , this term plays an important role .
this is arisen from the large difference in the time scale between the reptation time and the rouse retraction time .
it means that the relaxation of the tube orientation caused by constraint release of surrounding chains should start to happen just when the primitive chain nearly finish its retraction .
the boundary condition is @xmath61 and the initial condition is @xmath62 note that @xmath63 is the probability function , which has a range from 0 to 1 .
the stress is the sum of tensile force of the primitive chain projecting to some direction @xmath64 by averaging over the conformations of primitive chains .
those primitive chains penetrate a unit area plane with its normal direction @xmath64 .
the tensile force acting on segment @xmath40 , along the non - gaussian primitive chain at time @xmath16 can be expressed by the inverse langevin function @xmath65 when the pad approximation is used , we have @xmath66 where @xmath30 has the same meaning as in eq .
( 4 ) , and is used to conform that when @xmath67 , @xmath68 is equal to the fictitious force acting on every segment ( marrucci and grizzuti , 1988 ) .
the stress can be written as @xmath69 where @xmath70 is the number of polymers in a unit volume and _ n _ being the degree of polymerization .
we decouple the average segmental tension from the average segmental orientation in the calculation of the first line to the second line .
@xmath71 is the plateau modulus , @xmath72 .
@xmath73 can be expressed by @xmath74 here , @xmath75 is the de strain tensor without the independent alignment approximation given by @xmath76 where @xmath77 is the deformation tensor ( doi and edwards , 1986 ) .
the results reported in this paper will focus on steady and start - up of simple shear flow of monodisperse concentrated polymeric systems .
we scale the equations for simplicity .
the time variable is scaled by the reptation time , @xmath78 , and the spatial variable is scaled by the length of primitive chain at equilibrium , @xmath79 . for convenience , we use the same symbol to denote the scaled quantities .
( 9 ) , ( 12 ) and ( 15 ) become @xmath80 @xmath81 @xmath82 respectively . here , @xmath21 is given by eq .
( 3 ) , the relationship of which before scaling and after scaling is @xmath83 .
the stress tensor has been scaled as @xmath84 .
@xmath73 is given by eq .
( 16 ) . for the case of shear flow , @xmath85 .
the shear viscosity is defined by @xmath86 , where @xmath0 is the shear stress and the primary normal stress coefficient is given by @xmath87 , where @xmath88 is the first normal stress difference .
the boundary conditions are @xmath89 @xmath90 and the initial conditions are @xmath46 @xmath62 these equations are nonlinear and have to be solved numerically . combining the finite differencing and the newton - raphson method ,
we can obtain the solutions of these equations .
the details are shown in the appendix .
the constitutive parameters in these equations include @xmath91 and @xmath92 . here ,
@xmath91 is the number of entanglements per chain and can be specified by fitting the steady state predictions of the model and the experimental data .
a larger @xmath91 corresponds to a shorter rouse relaxation time .
@xmath92 is the ratio between the equilibrium and the maximum length of the primitive chain , reflecting the extensibility of the chain . in this section ,
we show the dynamic properties of the chain segments and rheological behavior under steady shear flow .
1 gives the non - uniform segmental stretch of the primitive chain with different shear rates .
the parameters are chosen as @xmath93 and @xmath94 which denote the gaussian chain . due to
the untethered fact of the chain ends , the segments at chain ends remain unstretched .
there exists a maximum value of stretch at the center .
the extension of the segments equivalently reduces the diffusion coefficient by @xmath95 in eq . ( 19 ) .
so the equivalent reduction of diffusion is much more at the center .
in addition to a small value of the rate of segmental convection around the center , we find that the rate of tube relaxation , or the rate of tube creation , is relatively small at the center .
the cancellation of gradually increasing rate of convection and decreasing rate of effective diffusion from the center to ends induces a flat region of the rate of the tube relaxation around the center . according to the calculation from eq .
( 16 ) under steady state shear , we find that around the center of the tube the value of @xmath96 is relatively small . at the ends , where the segments of tube disappear at their creating time , @xmath97 .
therefore , there must be a maximum value of @xmath96 in some location between the center and the ends .
this can be seen from fig . 2 when the shear rate is high enough .
it is very interesting that when the shear rate is high enough , the maximum contribution to stress from segmental orientation comes neither from the ends of the chain nor from the center , but from some other location in between .
moreover , this location will move towards the ends when the shear rate grows higher . from the calculation we also find that the segmental orientation is insensitive to the entanglement number of a primitive chain , denoted as @xmath91 which determines the segmental extension in steady state .
3 shows the differences of @xmath96 between the present model and demg model for @xmath93 and @xmath98 , respectively .
the result shows that including of the ccr term will obviously enhance the value of @xmath96 .
therefore , the magnitude of the stress in fast flow is seriously underestimated in the demg model .
the comparisons of the shear stress and the steady state shear viscosity with different values of @xmath91 are given in fig .
4 . in the linear region , where the steady shear rate is smaller than the inverse of reptation time ,
both the shear stresses and steady state shear viscosity are independent of @xmath91 , because in these cases the shear rates are not high enough to stretch the chain segments . when the shear rate is high enough , the segments will be stretched to many times of their equilibrium length .
the larger extension of the segment is , the more it contributes to the shear stress . on the other hand , the rouse relaxation time @xmath27 determines the magnitude of segmental extension .
if @xmath27 is larger , which corresponds to a smaller @xmath91 , the shear stress will be larger .
4 also shows the shear viscosity of the steady state which is defined as @xmath99 .
when the shear rate is smaller than the inverse of reptation time , it behaves as newtonian fluid . while in the high - shear - rate region it markedly depends on the shear rate .
different segmental stretch results in the differences of the first normal stress difference in the high - shear - rate region , which can be seen in fig .
the same characteristics for the primary normal stress coefficients are also shown in fig .
up to now the chains are approximated as gaussian chains ( @xmath94 ) , which is valid if the stains are not too large . however , under fast flow the chains are largely stretched and approach to their limiting extensible values , the gaussian treatment is no longer valid . in the following
, we will use the non - gaussian treatment of the single chain taking into account the finite extensibility of the chain ( @xmath100 ) .
6 shows the comparisons of the shear stress and the first normal stress difference with respect to the shear rate with different extensibilities for @xmath101 and @xmath102 , respectively .
the parameter @xmath91 is chosen as @xmath93 .
the calculation also indicates that for a specified value of the shear rate segmental orientation is not sensitive to segmental extensibility . from fig .
6 it can be found that the curves of shear stress and first normal stress difference are independent of segmental extensibility in the small shear rate region where the stress contribution from segmental stretch can be ignored . in the high - shear - rate region ,
the differences of shear stress and the differences of the first normal stress caused by the differences of segmental extensibility are mainly contributed from segmental extension .
the segmental orientation is only determined by the value of the shear rate , and is independent of the segmental stretch , because they are considered as independent processes .
6 also shows the comparisons between the predictions of the present model and the experimental data ( pattamaprom and larson , 2001 ) of the stresses and the first normal stress differences under steady shear for different parameters of @xmath92 ; fig .
7 shows the comparisons with experimental data of the stresses and the first normal stress differences under steady shear for different parameters of @xmath91 . by fitting the experimental data , we find that incorporating the non - uniform features to the constitutive models does not affect rheological properties .
moreover , the non - uniform model gives us the detailed descriptions of the chain dynamics . in the present model
there are two modifiable parameters of @xmath91 and @xmath92 . by fitting the experimental data to the predictions of the present model ,
we can obtain the values of @xmath91 and @xmath92 , respectively .
the comparison between them will show you the relative abilities of the retraction and the extension of the chains .
this means that , under the same magnitude of shear rate , different abilities of retraction or extension will result in different magnitudes of stresses , although the molecules composing the materials may have the same volume fractions and molecular weights . from these two figures , quantitative deviation from the experimental data exists , especially the first normal stress difference in the high - shear - rate region .
these deviations may attribute to the nonlocal " interactions between chains . by comparing with the uniform models by pattamaprom and larson ( 2001 )
, we also find that the effect of non - uniform ccr on the stress and the normal stress is minor .
the details of segmental orientation and stretch are not important in determining the rheological properties in steady simple shear . the transient behavior of chain stretch and orientation , as well as the rheological properties , are discussed in this section .
the steady shear is imposed to the system at @xmath103 . in fig .
8 the segmental stretch varying with time at different shear rates are shown , and the constitutive parameters are chosen as @xmath104 .
the contour variable is focused at @xmath105 .
before this segment reaches its steady value of extension , it passes a maximum value which is dependent on the shear rate , since other parameters are fixed .
this maximum value will appear earlier when the shear rate increases .
9 gives the behavior of segmental orientation @xmath106 with the constitutive parameters chosen as the same in fig .
we focus on the location @xmath105 .
if the shear rate is of the order of @xmath107 or larger , the value of @xmath106 will also pass a maximum value before reaching its steady value .
the appearance of the maximum value is not due to the fact of segmental stretch , but the existence of a maximum value in the strain @xmath108 with respect to the time .
10 and fig .
11 show the transient behavior of the shear stress and the first normal stress difference , respectively , where overshoots are predicted in the present model . for the shear stress ,
when the shear rate is higher than the inverse of reptation time , the overshoot appears .
the overshoot of the first normal stress difference appears at a much higher shear rate in the present model than that in other models .
the reason is that the appearance of maximum in the first normal stress difference is determined by the evolvement of segmental stretch which should also have a maximum in the course of time .
however , the rate of ccr we obtained contains a local strain @xmath109 , which largely decreases the maximum value of segmental stretch and subsequently the stress .
thus the magnitude of the start - up shear rate , which can produce a overshoot , increases .
12 and fig .
13 show the comparisons of the shear viscosities and the first normal stress differences with the experimental results , respectively .
we choose the shear viscosities instead of the shear stresses for convenience to compare with the experimental data ( pattamaprom and larson , 2001 ) , in which the material is sample one . the parameters in these two figures are both chosen as @xmath93 and @xmath94 , @xmath110 , @xmath102 , respectively .
the results indicate that only when the shear rate is high , i.e. @xmath111 , the differences caused by the different values of @xmath92 are predicted by the present model
. if the chains are less extensible , which corresponds to a large value of @xmath92 , the segmental extension will be smaller under the same shear rate at the same time .
this leads to smaller shear viscosities as well as the first normal stress differences .
in fact , the segmental orientation is mainly determined by the magnitude of the start - up shear rate . when the shear rate is fixed the contribution from orientation to shear stress
is specified .
thus , the magnitude of segmental stretch plays a critical role in determining the final value of the shear stress , and subsequently the shear viscosity .
this is kept when the first normal stress difference is considered . by fitting the experimental data we find that the predictions of the present model have the same tendency with the experimental curves . on the other hand ,
there are large inconsistencies when the quantitative values are concerned , especially the magnitude of overshoots . both the shear stress and the normal stress overshoots in the present model are very weak comparing to the experimental data
it is obvious that the chain stretch is not enough , which may arise from the inclusion of local approximation in ccr in the present model .
in this paper , we develop a contour - variable model to describe the non - uniform behavior of segments under steady shear and start - up of steady shear , and the subsequent contributions to the shear stress and the first normal stress difference .
the present model is a modification of the mld model with a modified ccr term and non - gaussian chain treatment of the single chain .
we drop the contour length fluctuation contribution and neglect the reptative constraint release term , since both of them contribute less than the shear rate dependent ccr term in the high - shear - rate region .
the numerical results predict strong dependencies of the segmental stretch and orientation on the contour variable .
the maximum extension happens in the center of the chain due to the demand of symmetry .
the fact that segmental extension passes a maximum during the start - up of steady shear flow is the key reason why the first normal stress difference has a maximum .
the segmental orientation also shows a maximum in the center when shear rate is small , and this is mainly attributed to the reptation .
however , when the shear rate is high enough the segmental convection begins to work .
a maximum of segmental orientation appears in some locations between the center and the end , where the stress contribution from orientation is the largest .
the inclusion of ccr entirely promotes the magnitude of orientation of each segment rather than changes the distribution of the value of segmental orientation . the non - uniform segmental orientation and stretch subsequently contribute to rheological properties , which qualitatively agree with experimental data , although quantitative deviation still exists . when the shear rate is high , i.e. @xmath112 , the segmental strain is larger than 1 .
the difference of the segmental extensibilities is the main reason for the difference of rheological behavior .
the same properties hold in the start - up of steady shear .
this validates our pre - approximation of the independence of segmental stretch and orientation .
a few approximations have been used in the present model .
one is the local approximation of ccr .
ccr happens once a tube segment is convected away from its original location , since the relative position to the primitive - chain segment confined by this tube segment changes .
the convection of a segment may not be only determined by its local conditions , but all other segments elsewhere .
in fact , the present model only predicts the overshoots in shear viscosity and first normal stress difference under the start - up of steady shear , but does not predict undershoots which have been observed in experiments ( huppler et al .
, 1967 ; mhetar and archer , 2000 ) .
it is well known that the appearance of overshoots is due to segmental stretch .
how about the undershoots ?
pattamaprom and larson ( 2001 ) predicted that the undershoots would appear if the dependence of the orientation and stretch on the contour variable was taken into account .
the present model indicates that the inclusion of `` local '' non - uniform rate of ccr to the constitutive equations does not predict undershoots .
undershoots may appear if we include the nonlocal interaction into the model .
that is the coupling between chains , e.g. the hydrodynamic interactions , which are neglected in the present model .
another is that we suppose the shear flow is homogeneous , which in general is not .
at last we can say that in polymer melts under shear non - uniform segmental orientation and stretch are hard to be observed directly in experiments , but their effect on rheological behavior can be detected , though the influences of the microscopic non - uniform behavior on them is not remarkable .
however , the processes of crystallization and the dynamics of phase separation ( maurits and fraaije , 1997 ) should be significantly affected by these contour dependent quantities , as the segmental dynamics plays a crucial rule in these processes .
in fact the kinetic coefficients are related to the non - uniform features ( kawasaki and sekimoto , 1988 ; kawakatsu , 1997 ) .
it is of great significance to investigate the segmental dynamics and their affiliated properties theoretically by microscopic or mesoscopic models .
we thank dr .
bing miao for the helpful discussion and suggestions .
this work is supported by national natural science foundation of china ( nsfc ) 20490222 , 20874111 , 50821062 and the grant from chinese academy of sciences kjcx2-yw-206 . in this appendix
we describe the process and methods for the calculation of constitutive equations .
the de strain tensor without the independent alignment approximation , i.e. eq . ( 17 ) , is integrated numerically .
the unit vector of orientation is chosen as @xmath113 @xmath114 where @xmath115 is the shear deformation , @xmath116 denotes the average over the isotropic state , i.e. , @xmath117 then we have @xmath118^{1/2}.\end{aligned}\ ] ] here , the variable transformations , i.e. @xmath119 , @xmath120 , are introduced .
then we have @xmath121^{1/2}}.\ ] ] with the same procedure , we obtain @xmath122^{1/2}}.\end{aligned}\ ] ] the two - dimensional integrals of eqs .
( a2)-(a4 ) are integrated out using the extended _
simpson s rule _ in each dimension , respectively .
the constitutive equations under start - up of steady shear are solved numerically .
we discretize eq .
( 18 ) using the crank - nicolson scheme , and define @xmath123 as @xmath124where _
i _ and _ j _ denote the discretized contour variable grid points and time variable grid points , respectively . here ,
@xmath125 is given by @xmath126 ^ 2}\frac{q(i , j)}{1+q(i , j ) } \left[q(i+1,j)-2q(i , j)+q(i-1,j)\right ] \nonumber \\ & & + \frac{3z\beta\alpha}{\pi^2 } \frac{\delta t}{(\delta s)^2 } \frac{4\alpha^3q^3(i , j)+12\alpha q}{\left[1-\alpha^2 q^2(i , j)\right]^3}\frac{q(i , j)}{1+q(i , j ) } \frac{\left[q(i+1,j)-q(i-1,j)\right]^2}{4 } \nonumber \\ & & + \frac{1}{2}\delta t\dot \gamma s_{xy}(i , j ) \frac{3q^2(i , j)-q^3(i , j)}{1+q(i , j)}\end{aligned}\ ] ] where @xmath127 and @xmath128 are the contour and time grid steps , respectively .
the contour grid step is a constant in our calculation , while there are two kinds of time step if the shear rate is higher than @xmath129 .
the grids density around the maximum value of @xmath130 is larger than that the place far from it , so that the effects arising from the sharp changes of the value of @xmath130 around the maximum value can be observed .
there are @xmath131 coupled equations if we set @xmath132 for each time grid , and @xmath133 is the grid number of of the contour coordinate .
the newton - raphson method is used to find their roots at each time grid .
the number of time step is @xmath134 . in this paper , we choose @xmath135 and @xmath128 has a range from @xmath136 to @xmath137 depending on the shear rate . for a larger shear rate , we choose a smaller @xmath138 .
@xmath134 is determined by the time grid steps .
( 19 ) can be written as @xmath139 where @xmath140 @xmath141 @xmath142 eq .
( a7 ) can be discretized as @xmath143 \nonumber \\ & & -a_1(i , j+1)\frac{1}{2 } \left[\left(\frac{\partial g}{\partial s_0}\right)_{(i , j ) } + \left(\frac{\partial g}{\partial s_0}\right)_{(i , j+1 ) } \right ] \nonumber \\ & & + a_3(i , j+1)\frac{1}{2 } \left(g(i , j)+g(i , j+1)\right),\end{aligned}\ ] ] where @xmath144 @xmath145 the complete procedure is as follows . for every time step ,
we first assume that @xmath146 .
we solve the nonlinear equations of @xmath132 ( @xmath123 is given by eq .
( a5 ) ) , then we obtain @xmath147 .
insert these into eq .
( a8 ) we obtain @xmath148 . from eq .
( 16 ) we obtain the new values of @xmath149 . by inserting these values back to eq .
( a5 ) and let @xmath132 , we obtain the new values of @xmath147 , then new @xmath148 , and subsequently @xmath149 .
repeat the above steps until the expected convergent value ( @xmath150 ) is reached .
the calculation of constitutive equations under steady shear is much easier than that under the start - up of steady shear , as in the former case the orientational tensor @xmath151 and strain @xmath60 are time independent , and the calculation procedure are the same as in the case of start - up of steady shear at a certain time grid . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ cohen a ( 1991 ) a pad approximant to the inverse langevin function .
rheol acta 30 : 270 - 273 graham rs , likhtman ae , mcleish tcb , milner st ( 2003 ) microscopic theory of linear , entangled polymer chains under rapid deformation including chain strecth and convective constraint release .
j rheol 47 : 1171 - 1200 watanabe h ( 1999 ) viscosity and dynamics of entangled polymers .
prog polym sci 24 : 1253 - 1403 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig . 6 * comparison of the stresses and the first normal stress differences under steady shear between the present model ( lines ) and the experimental data ( closed symbols ) .
the parameters are chosen as @xmath93 and @xmath94 , @xmath153 , @xmath154 , @xmath102 , respectively . *
* comparison of the stresses and the first normal stress differences under steady shear between the present model ( lines ) and the experimental data ( closed symbols ) .
the parameters are chosen as @xmath94 and @xmath155 , @xmath156 , @xmath157 , @xmath158 , respectively . | constitutive equation , ccr , contour variable , polymer dynamics |
the breaking of the colour - electric string between two static sources is a prime example of a strong decay in qcd @xcite .
recently , we reported on an investigation of this two state system @xcite , with a wave function @xmath1 created by a @xmath2 operator and a wave function @xmath3 created by a four - quark @xmath4 operator , where @xmath5 .
@xmath6 denotes a static source and @xmath7 is a light quark .
we determined the energy levels @xmath8 and @xmath9 of the two physical eigenstates @xmath10 and @xmath11 which we decomposed into the components , @xmath12 we characterise string breaking by the distance scale @xmath13 at which @xmath14 is minimized and by the energy gap @xmath15 .
while these energy levels and @xmath13 are first principles qcd predictions , the mixing angle @xmath16 is ( slightly ) model dependent : within each ( fock ) sector there are further radial and gluonic excitations and we truncated the basis after the four quark operator . in order to obtain dynamical information on the string breaking mechanism , we are studying the spatial energy and action density distributions within the two state system .
in doing so one can address questions about the localisation of the light @xmath17 pair that is created when @xmath18 is increased beyond @xmath13 .
the energy density will decrease fastest in those places where @xmath17 creation is most likely .
perturbation theory suggests that light pair creation close to one of the static sources is favoured by the coulomb energy gain while aesthetic arguments might suggest a symmetric situation with @xmath17 dominantly being created near the centre .
first results on this investigation are presented here .
we use @xmath0 wilson fermions at a quark mass slightly smaller than the physical strange quark and a lattice spacing @xmath19 fm and find @xcite , @xmath20 and @xmath21 mev , where the errors do not include the phenomenological uncertainty of assigning a physical scale to @xmath22 fm . using data on the @xmath2 potential and the static - light meson mass @xmath23 , obtained at different quark masses ,
we determine the real world estimate , @xmath24 fm , where the errors reflect all systematics .
an extrapolation of @xmath25 however is impossible , without additional simulations at lighter quark masses .
these results became possible by combining a variety of improvement techniques : the necessary all - to - all light quark propagators were calculated from the lowest eigenmodes of the wilson - dirac operator , multiplied by @xmath26 , after a variance reduced stochastic estimator correction step .
the signal was improved by employing a fat link static action .
many off - axis distances were implemented to allow for a fine spatial resolution of the string breaking region .
for details see ref .
@xcite .
+ great care was also spent on optimizing the overlap with the ground states , within the @xmath1 and @xmath3 sectors , using combinations of ape @xcite and wuppertal @xcite smearing .
our ape smearing consists of the replacement , @xmath27 @xmath28 denotes a projection operator , back onto the @xmath29 group , and @xmath30 @xmath31 , is the spatial staple - sum , surrounding @xmath32 .
we choose @xmath33 and define our ape smeared links , @xmath34 . for the projection we use , @xmath35 the spatial transporters within our @xmath2 states are products of ape smeared links , taken along the shortest lattice distance between the two sources .
the ape smeared links are also employed for the parallel transport within the wuppertal smearing of light quark sources , used to improve the static - light meson operators : @xmath36 we set @xmath37 and take the linear combination @xmath38 as our smearing function .
best results are obtained by using smeared - local quark propagators .
the quality of the overlap with the static - light mesonic ground state is visualized in the effective mass plot figure [ fig : effmass ] .
note that we display the effective masses up to physical distances @xmath39 fm .
+ we were able to achieve values of @xmath40 and @xmath41 at @xmath42 for the overlaps of our test wave functions @xmath43 with the respective states on the right hand sides of eqs .
( [ eq:1 ] ) and ( [ eq:2 ] ) .
we display these results in figure [ fig : overlap ] .
the almost optimal value @xmath44 was essential to allow @xmath45 , @xmath46 and @xmath16 to be fitted from correlation matrix data , @xmath47{w11.ps}\end{minipage } } } & \sqrt{n_f}{\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w12.ps}\end{minipage}}}\\&\\ \sqrt{n_f}{\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w21.ps}\end{minipage}}}&-n_f{\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w22c.ps}\end{minipage}}}+{\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w22d.ps}\end{minipage}}}\end{array}\right),\nonumber\end{aligned}\ ] ] obtained at moderate euclidean times : @xmath48 for @xmath49 and @xmath50 for the remaining 2 matrix elements @xmath51 and @xmath52 at @xmath42 .
we follow ref .
@xcite and define action and energy density distributions , @xmath53,\\ \epsilon_n({\mathbf x})=\frac12\left[{\mathcal e}_n({\mathbf x})-{\mathcal b}_n({\mathbf x})\right ] , \label{eq : eps}\end{aligned}\ ] ] where @xmath54 @xmath55 are our approximations to the creation operators of the states @xmath56 , obtained from the diagonalisation of @xmath57 .
we have suppressed the distance @xmath18 from the above formulae and @xmath58 denotes the ground state ( dominantly @xmath2 at @xmath59 ) and @xmath60 the excitation ( dominantly @xmath4 at @xmath59 ) .
electric and magnetic fields are calculated from the plaquette , @xmath61,\ ] ] @xmath62,\end{aligned}\ ] ] where @xmath63 and , @xmath64 we implement two different operators with the same continuum limits : in one case we identify @xmath65 with the link @xmath66 connecting @xmath67 with @xmath68 . additionally , we used smeared operators , @xmath69 where @xmath70 and @xmath71 is the sum over all six staples enclosing @xmath66 , in the three forward and in the three backward directions .
the @xmath72-value was tuned to maximize the average plaquette , calculated from the smeared links . for the un - smeared plaquette @xmath73 in eq .
( [ eq : nn ] ) while for smeared plaquettes
@xmath74 is adjusted such that the vacuum expectation value of the average plaquette remains unchanged .
+ + the plaquette smearing enhances the signal / noise ratio . due to this smearing and the fat link static action used , the peaks of the distributions around the source positions ( that will diverge in the continuum limit ) are less singular than in previous studies of @xmath75 gauge theory at similar lattice spacings @xcite . in the continuum limit the results from smeared and un - smeared plaquette probes will coincide , away from these self energy peaks .
the draw back of plaquette smearing is that exact reflection positivity is violated .
however , our wave functions are sufficiently optimized to compensate for this .
we insert the @xmath76 and @xmath77 operators at position @xmath78 into the correlation matrix @xmath57 , eq . ( [ eq : sb ] ) . for even@xmath79odd @xmath80-values
we average @xmath81 over the two adjacent time slices , respectively . using the fitted ground state overlap ratio @xmath82 and the mixing angle @xmath16 as inputs ,
we calculate the action and energy density distributions eqs .
( [ eq : sig ] ) and ( [ eq : eps ] ) in the limit of large @xmath83 via eq .
( [ eq : cal ] ) from the measured matrix elements .
the distributions agree within errors within the time range @xmath84 .
the results presented here are based on our @xmath85 analysis .
to set the stage , we display the main results of ref .
@xcite in figures [ fig : pot ] and [ fig : angle ] . in the first figure we also speculate about the scenario in the real world with possible decays into @xmath4 as well as into @xmath86 . for our parameter settings and @xmath0
string breaking occurs at a distance @xmath87 fm . in figure
[ fig : angle ] we show the mixing angle as a function of the distance .
the @xmath4 content of the ground state is given by @xmath88 .
within our statistical errors @xmath16 reaches @xmath89 at @xmath90 .
remarkably , there is a significant four quark component in the ground state at @xmath59 while for @xmath91 the limit @xmath92 is rapidly approached .
+ our basis vectors @xmath1 and @xmath3 are no eigenstates of the system .
the transition rate , @xmath93 , can be related to energy gap and mixing angle : @xmath94 .
this means that @xmath95 . with a transition rate of only about 25 mev in the string breaking region and even smaller @xmath96-values at @xmath91 , a detection of the ground state contribution to
the standard wilson loop at @xmath91 , where @xmath10 contains little @xmath1 admixture , is hopeless : one would have to resolve the correlation function at times of order @xmath97 fm ! for a complementary view on the problem , in the context of the breaking of higher representation strings in pure gauge theories , see e.g. ref . @xcite .
+ + + + diagrammatically , @xmath96 can be represented as , @xmath98{w12.ps}\end{minipage}}}}{\left({\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w11.ps}\end{minipage}}}\times n_f^2{\mbox { \begin{minipage}{16pt}\includegraphics[width=16pt]{w22d.ps}\end{minipage}}}\right)^{1/2 } } \propto \sqrt{\frac{n_f}{n_c}},\ ] ] where @xmath99 is the number of colours : the large @xmath100 expectation for the minimal energy gap is , @xmath101 .
obviously , precision studies at different @xmath99 and @xmath102 are needed to establish the validity range of this prediction .
the decay of the adjoint potential into two gluelumps , where @xmath103 , also fits nicely into this context . the quality of our density distribution data is depicted in figure [ fig : transverse ] for the ground state at a separation slightly smaller than @xmath13 , as a function of the transverse distance @xmath67 from the @xmath2 axis . due to cancellations between the magnetic and electric components
the energy density is much smaller than the action density : for the comparison we have multiplied the energy density data by the arbitrary factor of 29 .
note that the ratio @xmath104 will diverge like @xmath105 in the continuum limit .
the differences between the shapes of the energy and action density distributions are not statistically significant .
below we only visualise the more precise action density results .
we employ several off - axis separations . assuming rotational symmetry about the interquark axis , each point is labelled by two coordinates .
@xmath67 denotes the distance from the @xmath2 axis and @xmath106 denotes the longitudinal distance from the centre point .
we define an interpolating rectangular grid with perpendicular lattice spacing @xmath107 and the longitudinal spacing slightly scaled , such that the static sources always lie on integer grid coordinates .
we then assign a quadratically interpolated value to each grid point @xmath108 , obtained from points in the neighbourhood , @xmath109 . on the axis the data points are more sparse and
we relax the condition to @xmath110 while for the singular peaks we maintain the un - interpolated values . in figures [ fig : frame ] and [ fig :
frame2 ] we display ground state action density distributions for different distances around @xmath13 .
an mpeg animation has been published in ref .
the colour encodes the relative statistical errors and the lattice mesh represents our spatial resolution @xmath107 . as already evident from the mixing angle @xmath111 of figure [ fig : angle ] above , string breaking takes place within a small region around @xmath13 .
all distributions are very similar to linear superpositions of string and broken - string states , with no non - trivial spatial dependence : string breaking appears to resemble an instantaneous process , without evidence of localisation of the @xmath17 pair creation .
we wish to relate the static limit results to strong decay rates of quarkonia . in the non - relativistic limit of heavy quarks , potential `` models '' provide us with the natural framework for such studies .
in fact at short distances , @xmath112 , potential `` models '' can be derived as an effective field theory , potential nrqcd , from qcd @xcite .
one can in principle add a @xmath4 sector , as well as transition terms between the two sectors , to the @xmath2 pnrqcd lagrangian .
strong decays would then be a straightforward non - perturbative generalisation of the standard multipole treatment of radiative transitions in qed .
unfortunately , transitions such as @xmath113 can hardly be classed as `` short distance '' physics .
so , some modelling is required instead .
the natural starting point again is a two channel potential model which might still have some validity beyond the short distance regime : @xmath114 with @xmath115 @xmath116 denotes the heavy quark mass and @xmath23 is the mass of a @xmath117 meson .
the wavefunction has two components , @xmath118 and the potential is given by , @xmath119 where we have normalized the zero point energy to @xmath120 and @xmath121 rotates our fock basis @xmath122 into the eigenbasis @xmath123 .
we neglect @xmath117-@xmath124 interactions and set @xmath125 .
we further adjust the difference @xmath126 to the experimental value which affects our phase space factor .
we then follow ref .
@xcite and calculate the decay rate by multiplying phase space with the overlap integral between the @xmath127 wave function , the interaction term @xmath96 and the @xmath4 continuum . in doing so , we assume that the interaction does not contain any spatial distribution but only depends on the distance @xmath18 .
this instantaneous picture is supported by our action density measurements above .
+ in figure [ fig : wave ] we display our radial @xmath127 wave function @xmath128 .
the decay rate depends rather sensitively on the positions of the nodes .
we obtain a preliminary result @xmath129 \approx 5 $ ] mev , which is about half the experimental value .
this appears very reasonable , given the crudeness of the model and the fact that the gap @xmath25 will increase with lighter , more realistic sea quark masses .
we are studying the situation and systematics in more detail .
we were able to resolve the string breaking problem in @xmath0 qcd , at one value of the lattice spacing @xmath130 gev and of the sea quark mass , @xmath131 .
it was also possible to study the dynamics of string breaking in detail and to resolve spatial colour field distributions .
the breaking of the string appears to be an instantaneous process , with de - localized light quark pair creation .
while a direct lattice study of strong decay rates such as @xmath132 or @xmath133 is at present virtually impossible , investigations in the static limit can help constraining models .
studying the energy between pairs of static - light mesons can also be viewed as a milestone with respect to future calculations of @xmath134 forces , which are related to nucleon - nucleon interactions @xcite .
the computations have been performed on the ibm regatta p690 + ( jump ) of zam at fz - jlich and on the alice cluster computer of wuppertal university .
this work is supported by the ec hadron physics i3 contract rii3-ct-2004 - 506078 , by the deutsche forschungsgemeinschaft and by pparc .
99 c. michael , pos lat2005 , 008 ( 2005 ) [ arxiv : hep - lat/0509023 ] .
bali _ et al .
_ , arxiv : hep - lat/0409137 .
bali _ et al . _ [ sesam ] , phys . rev .
d 71 , 114513 ( 2005 ) [ arxiv : hep - lat/0505012 ] .
z. prkacin _ et al .
_ pos lat2005 , 308 ( 2005 ) [ arxiv : hep - lat/0510051 ] .
m. albanese _ et al . _
[ ape ] , phys .
lett . 192b , 163 ( 1987 ) . m. teper , phys . lett . 183b , 345 ( 1987 ) .
s. gsken _ et al .
_ , phys .
b 227 , 266 ( 1989 ) .
bali , k. schilling and c. schlichter , phys .
d 51 , 5165 ( 1995 ) .
f. gliozzi and a. rago , nucl .
b 714 , 91 ( 2005 ) [ arxiv : hep - lat/0411004 ] .
n. brambilla , a. pineda , j. soto and a. vairo , arxiv : hep - ph/0410047 .
i.t . drummond and r.r .
horgan , phys .
b 447 , 298 ( 1999 ) [ arxiv : hep - lat/9811016 ] .
d. arndt , s. r. beane and m. j. savage , nucl .
phys . a 726 , 339 ( 2003 ) [ arxiv : nucl - th/0304004 ] . | we numerically investigate the transition of the static quark - antiquark string into a static - light meson - antimeson system . improving noise reduction techniques , we are able to resolve the signature of string breaking dynamics for @xmath0 lattice qcd at zero temperature .
we discuss the lattice techniques used and present results on energy levels and mixing angle of the static two - state system .
we visualize the action density distribution in the region of string breaking as a function of the static colour source - antisource separation .
the results can be related to properties of quarkonium systems . |
globular clusters ( gcs ) are thought to be formed in bursts of star formation within molecular protocluster clouds , producing gravitationally bound clusters of 10@xmath210@xmath3 stars which represent the best known approximations of a `` simple stellar population '' ( hereafter ssp ) , i.e. , a coeval population of stars with a single metallicity ( e.g. , * ? ? ?
* and references therein ) .
extensive photometric observations of extragalactic gc systems have now yielded a substantial database of gc colors and magnitudes to investigate overall trends among subpopulations of gcs and between properties of gc subpopulations and those of their host galaxies .
the advent of the _ advanced camera for surveys ( acs ) _ aboard the _ hubble space telescope ( hst ) _ was pivotal in this area due to its unique high sensitivity and spatial resolution over a field of view of @xmath412 arcmin@xmath5 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
one main result of these studies was that gcs in luminous galaxies typically follow a bimodal color distribution .
multi - object spectroscopy studies showed that both `` blue '' and `` red '' gc subpopulations are nearly universally old , with ages @xmath68 gyr ( e.g. , * ? ? ?
* ; * ? ? ?
this implies that the color bimodality is mainly due to differences in metallicity ( see also * ? ? ?
typical metallicity values for metal - poor ( blue ) gcs in giant elliptical galaxies are [ z / h ] = @xmath7 , while metal - rich ( red ) gcs average [ z / h ] = @xmath8 @xcite .
a more recently discovered feature among gc systems was a color - magnitude relation among gcs in the metal - poor subpopulation , in the sense that their colors become progressively redder at higher luminosities @xcite .
this relation , colloquially called the `` blue tilt '' , is typically interpreted as a mass - metallicity relation ( hereafter mmr ) , although the various analyses of the effect in individual galaxies have not yet converged to an agreement on the significance of the tilt and the mass range in which the correlation is significant ( see , e.g. , @xcite and @xcite versus @xcite , @xcite , and @xcite ) .
recently , @xcite performed a detailed investigation of the color - magnitude relation in the largest sample of gcs to date by far ( @xmath916,900 gcs , see section [ s : sample ] ) , using @xmath10 and @xmath11 filters .
three main conclusions of @xcite regarding the blue tilt were that _
( i ) _ the blue tilt is firmly present for gc masses @xmath12 while it becomes more pronounced at @xmath13 ; _ ( ii ) _ the tilt is more pronounced for gcs associated with high - mass galaxies ( @xmath14 ) than for gcs in lower - mass galaxies ; and _ ( iii ) _ the tilt is more pronounced for gcs at smaller galactocentric distances .
notably , @xcite also found evidence for the presence of a `` red tilt '' , i.e. , a color - magnitude relation among gcs in the metal - rich subpopulation .
this red tilt is seen among gcs in the full magnitude range sampled by @xcite , albeit only within their `` high - mass '' subset of galaxies .
the vast majority of studies of the mmr among gcs explain it as being due to gc self - enrichment by ejecta of supernovae of type ii ( snii ; e.g. , * ? ? ?
* ; * ? ? ?
. however , several features of the blue tilt mentioned above seem to indicate that gc self - enrichment alone can not fully explain the observed mmr .
for example , most theoretical work on gc self - enrichment in heavy ( fe - peak ) elements indicates that it is only effective at masses @xmath15 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) whereas the blue tilt in elliptical galaxies persists down to gc masses that are an order of magnitude loweral ) are commonly found within galactic gcs , and there are indications that the amplitude of such variations scale with gc mass ( e.g. , * ? ? ? * ; * ? ? ?
. however , such variations only cause a small blueing of integrated @xmath16 colors ( @xmath17 * ? ? ?
* ) , which can not cause the blue or red tilts . ] .
furthermore , one would not expect to see significant environmental dependences of the slope of the mmr ( such as the tilt being more pronounced for higher - mass galaxies or for smaller galactocentric distances ) if gc self - enrichment were the only relevant cause .
instead , the presence of such environmental variations seem to indicate an effect ( or effects ) whose strength is driven by the local tidal field , such as star cluster disruption .
recent investigations of the distribution of fe - peak element abundances within galactic gcs are starting to yield clues in this regard .
statistically significant internal variations in [ fe / h ] have so far been detected within a few ( metal - poor ) gcs : @xmath18cen ( e.g. , * ? ? ?
* ) , m54 @xcite , and m22 @xcite .
these gcs have current masses of @xmath19 , @xmath20 , and @xmath21 respectively @xcite .
note that the mass of m22 is comparable to the onset of the blue tilt found by @xcite , lending credence to the potential relevance of gc mass loss in this context .
meanwhile , recent studies of the central regions of nearby massive elliptical galaxies revealed very strong gravity - sensitive features in their near - ir spectra @xcite , indicating a stellar initial mass function ( imf ) that was steeper at subsolar masses ( @xmath22 in @xmath1 ) than the canonical @xcite or @xcite imfs , which have @xmath23 in the same mass range .
this was subsequently found to be consistent with studies of stellar @xmath24 ratios of massive early - type galaxies derived from gravitational lensing ( e.g. , * ? ? ?
* ) and stellar dynamics @xcite .
in the same vein , @xcite reported a systematic offset between optical colors of giant elliptical galaxies and the mean colors of their massive metal - rich gcs , in the sense that those gcs were bluer than their parent galaxies by 0.120.20 mag at a given galactocentric distance ( a 722 @xmath25 effect ) .
however , measurements of lick - system spectral indices that are sensitive to age and metallicity ( but not to imf variations ) for such galaxies and their metal - rich gcs _ were _ found to be consistent with one another at a given galactocentric distance , within small uncertainties .
* hereafter paper i ) showed that this paradox can be explained if the stellar imf was steep at subsolar masses : @xmath2612 gyr of dynamical evolution would then cause the average massive gc to become bluer by an amount consistent with the observed color difference between the metal - rich gcs and their parent galaxies .
to investigate these recent indications of the relevance of the imf shape and dynamical evolution with regards to the colors of gcs and features of the observed mmr among gcs in nearby galaxies , we conduct a study of the effects of mass loss of gcs born with different imf shapes to the mmr among gcs in giant elliptical galaxies and compare those effects to the observed mmr and the effect of self - enrichment .
in comparing results of our calculations with observed features of the gc mmr in giant elliptical galaxies , we adopt the parametrizations of the observed color - magnitude relation and the mmr among metal - poor and metal - rich gcs by ( * ? ? ?
* see their table 1 ) .
this choice was made because their mmr study gathered the largest photometric sample of gcs to date by far ( @xmath916,900 gcs ) , namely by combining all gcs detected in the acs fornax and virgo surveys of early - type galaxies .
these _ hst _ surveys were carried out using a homogeneous measurement technique in which photometric aperture - size corrections were determined from convolutions of the best - fit gc profiles with local point - spread functions on the image ( see @xcite for details ) .
briefly , gc masses and metallicities in @xcite were derived from hst / acs photometry in the f475w ( hereafter @xmath10 ) and f850lp ( @xmath11 ) filters , corrected for galactic foreground extinction .
metallicities ( in terms of [ fe / h ] on the @xcite scale ) were derived using the broken linear transformation of @xcite , while gc masses were derived from @xmath11-band absolute magnitudes , using @xmath27 , which is correct to within 5% in the range @xmath28 } \la + 0.2 $ ] for an ssp @xcite .
we refer to @xcite for further details regarding the determination of color - magnitude relations and mmrs among their various subsamples of gcs .
it is well known that mass loss of star clusters by dynamical evolution affects the shape of their stellar mf , because the escape probability of stars from their parent cluster increases with decreasing stellar mass ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this effect is widely used to explain observed stellar mfs in ancient galactic gcs , some of which are flatter than canonical imfs ( e.g. , * ? ? ?
* ; * ? ? ?
such flat mfs in gcs can explain their low observed mass - to - light ratios when compared with predictions of ssp models that use canonical imfs ( e.g. , * ? ? ?
* ) . to evaluate the influence of the shape of the imf on the present - day color - magnitude relation among gcs
, we follow paper i and use the gc evolution model of ( * ? ? ?
* hereafter k09 ) which incorporates the effects of stellar evolution , dynamical evolution in a tidal field , mass segregation , and stellar remnant retention . for the purposes of the current study , we select k09 models that _ ( i ) _ feature `` default '' kick velocities of stellar remnants ( white dwarfs , neutron stars , and black holes ) , and _
( ii ) _ produce gcs with masses @xmath29 at an age of 12 gyr , i.e. , gcs such as those selected by @xcite to derive their mmrs .
we further consider @xcite profiles with @xmath30 values of 5 and 7 and environmentally dependent cluster dissolution time scales @xmath31 of 0.3 , 0.6 , 1.0 , and 3.0 myr is defined by @xmath32 where @xmath33 is the cluster disruption time and @xmath34 sets the mass dependence of cluster disruption .
see @xcite and k09 for details . ] . for reference ,
@xmath35 myr yields a good fit to the globular cluster mass function of all surviving galactic gcs @xcite .
however , rather than using a single @xmath31 value for a given galaxy , we consider that gc systems in galaxies are expected to exhibit a _ range _ of @xmath31 values .
for instance , the rate of mass loss from gcs by two - body relaxation , @xmath36 , effectively scales with the mean gc half - mass density @xmath37 as @xmath38 for tidally limited gcs ( e.g. , @xcite ) . among gcs
for which two - body relaxation has been the dominant mass loss mechanism , low-@xmath37 gcs thus feature larger @xmath31 values than high-@xmath37 gcs at a given @xmath39 .
similarly , gcs on eccentric orbits with large perigalactic distances @xmath40 should have larger @xmath31 values than otherwise ( initially ) similar gcs with smaller values of @xmath41 . finally , disruption rates of gcs in dense environments such as central regions of gas - rich galaxy mergers at high redshift were likely higher than in current quiescent galaxies @xcite due to the enhanced strength and rate of tidal perturbations . in paper
i , we explored the influence of the imf on the present - day mf at subsolar masses for the star cluster families mentioned above . for an ensemble of such gcs
, our calculations showed that the overall average _ present - day _ mf slope in the stellar mass range 0.30.8 @xmath42 ( hereafter designated @xmath43 ) is consistent with the kroupa mf for any imf slope @xmath44 ( see figure 7 of paper i ) .
however , the _ mass dependence _ of @xmath43 ( averaged over all k09 models that produce gcs of a given mass ) , is found to be significantly stronger for steeper imfs .
this is illustrated in figures [ f : mf_and_m_lz_vs_logm]a[f : mf_and_m_lz_vs_logm]c , which show @xmath43 versus @xmath39 for all k09 models that produce gcs with @xmath45 at an age of 12 gyr , as well as the average model , for three imfs : _
( i ) _ a `` bottom - heavy '' imf with @xmath46 for @xmath47 and @xmath48 for @xmath49 ( this imf is hereafter simply referred to as @xmath46 ) , _ ( ii ) _ the @xcite imf ( i.e. , @xmath48 over the full stellar mass range ) , and _
( iii ) _ the @xcite imf .
as described in detail in paper i , the stronger cluster mass dependence of @xmath43 among ancient gcs born with a steeper imf is mainly caused by the weaker effect of retained massive stellar remnants on the escape rate of massive stars in gcs with steeper imfs relative to that in gcs with flatter imfs .
the effect of 12 gyr of dynamical evolution on the @xmath50 ratio as a function of @xmath39 is similar to that on @xmath43 .
this is shown in figures [ f : mf_and_m_lz_vs_logm]d[f : mf_and_m_lz_vs_logm]f .
as expected , steeper imfs feature higher @xmath50 ratios for the most massive gcs .
however , steeper imfs also feature a stronger cluster mass dependence of @xmath50 , and a larger spread of @xmath50 ratios among different gc families at a given current mass .
this result renders it difficult to constrain the imf slope for individual ancient gcs through direct measurements of @xmath51 or @xmath52 , especially for gcs with @xmath53 .
however , the large spread of observed mf slopes and @xmath52 ratios among galactic gcs @xcite generally appears more consistent with a bottom - heavy imf than with a kroupa or chabrier one . in the remainder of this paper , the cluster mass @xmath50 relations shown by the thick solid lines in figs .
[ f : mf_and_m_lz_vs_logm]d[f : mf_and_m_lz_vs_logm]f will be taken into account when comparing model predictions to the observed mmr from @xcite , since the latter study used @xmath54 independent of cluster mass . the impact of @xmath43 on the integrated @xmath16 color of ancient ( 12 gyr old ) gcs is shown in figure [ f : color_vs_alpha ] for [ z / h ] = @xmath561.3 and [ z / h ] = @xmath560.2 , i.e. , the two typical metallicity values for the metal - poor and metal - rich gc subpopulations in giant elliptical galaxies , respectively .
these curves were determined from padova isochrones @xcite .
after rebinning the isochrone tables to a uniform bin size in stellar mass @xmath57 using linear interpolation , weighted luminosities of the population were derived for the relevant passbands by normalizing individual stellar luminosities @xmath58 as follows : @xmath59 where @xmath60 is the maximum stellar mass reached in the isochrone table , @xmath61 , and @xmath62 is the mf slope in the mass range @xmath63 .
note that the effect of the imf shape on the @xmath16 color is larger for metal - poor populations than for metal - rich ones .
this difference can be understood by considering the shape of the isochrones in color - magnitude diagrams ( cmds ) at different metallicities .
this is illustrated in figure [ f : isocomparison ] which shows @xcite isochrones for [ z / h ] = @xmath561.3 and [ z / h ] = @xmath560.2 ( at an age of 12 gyr ) in a ( @xmath64)/2 vs. @xmath16 cmd)/2 was chosen as the ordinate for the cmd in order to show the luminosity exactly in between the @xmath10 and @xmath11 passbands . ] .
note that metal - rich populations feature a strong curvature of the red giant branch ( rgb ) in these passbands , rendering @xmath16 colors of upper rgb stars that are significantly redder than those of lower main sequence ( ms ) stars .
conversely , the curvature of the rgb in the cmd for metal - poor populations is much weaker , and upper rgb stars are _ bluer _ than lower ms stars for metal - poor populations .
as the overall range of ( @xmath64)/2 magnitudes spanned by the isochrones is similar for metal - poor and metal - rich populations , the impact of the mf slope at subsolar masses on the integrated colors is larger for metal - poor populations than for metal - rich ones .
color vs. @xmath43 using @xcite isochrones for an age of 12 gyr and the two [ z / h ] values @xmath561.3 ( solid line ) and @xmath560.2 ( dashed line ) .
for convenience , dotted lines indicate values for @xmath16 and @xmath62 for the kroupa imf ( k01 ) , the salpeter imf ( s55 ) and an imf with @xmath65 .
[ f : color_vs_alpha ] , width=215 ] = @xmath561.3 ( black lines ) and @xmath560.2 ( red lines ) in a @xmath66 vs. @xmath55 color - magnitude diagram .
the agb phase is indicated with dashed lines . _
right panel _
: illustration of the relative weights of stellar luminosities implied by two mass functions : the kroupa imf ( solid line ) and @xmath46 ( dashed line ) .
see discussion in section [ s : gzevol ] .
[ f : isocomparison ] , width=313 ] to transform the gc mass and imf - dependent effect of gc disruption on @xmath43 at an age of 12 gyr seen in figs .
[ f : mf_and_m_lz_vs_logm]a[f : mf_and_m_lz_vs_logm]c to observable properties , we apply the @xmath43 vs. @xmath16 relations shown in figure [ f : color_vs_alpha ] to the k09 models for the three imfs shown in figure [ f : mf_and_m_lz_vs_logm ] . figure [ f : mcurr_vs_color ] shows the resulting color - mass relations among ancient gcs for the three imfs .
for comparison , we overplot the observed color - mass relations among metal - poor and metal - rich gc subpopulations in giant elliptical galaxies taken from ( * ? ? ? *
i.e. , their
fcs+vcs high - mass galaxy sample ) , after taking into account the dependence of @xmath50 on @xmath39 ( and its dependence on the imf ) shown in figs .
[ f : mf_and_m_lz_vs_logm]d[f : mf_and_m_lz_vs_logm]f .
figures [ f : mcurr_vs_color]b and [ f : mcurr_vs_color]c show that the observed color - mass relations among gcs in giant ellipticals are significantly more pronounced than those predicted by the k09 gc disruption models that use the kroupa imf ( and , to a lesser extent , those using the salpeter imf as well ) .
this is especially true for the case of the metal - poor gcs ( i.e. , the ` blue tilt ' ) .
this is consistent with the findings of @xcite who used the older gc disruption models of @xcite ( which use the kroupa imf ) to evaluate the impact of dynamical evolution to the color - mass relations .
hence , @xcite concluded that dynamical evolution of star clusters can not be responsible for the shape of the blue tilt seen in giant elliptical galaxies .
however , the color - mass relations predicted by the k09 models for @xmath46 are more similar to the observed relations ( see fig . [
f : mcurr_vs_color]a ) .
specifically , the `` average k09 model '' for @xmath46 reproduces the ` red tilt ' observed among metal - rich gcs in giant ellipticals almost perfectly , while the same model reproduces the ` blue tilt ' quite well for gc masses up to log @xmath67 , beyond which the observed blue tilt becomes more pronounced than the k09 model prediction .
as shown in the previous section , dynamical evolution of gcs can explain the shape of the color - mass relation among gcs in giant elliptical galaxies up to log @xmath67 if the imf was steep at subsolar masses ( @xmath68 ) , i.e. , significantly steeper than the kroupa imf for which @xmath69 at those masses .
however , the fact that the blue tilt becomes more pronounced for log @xmath70 does not seem to be compatible with dynamical evolution and requires a different explanation . in this context , previous studies already tested the potential efficacy of stochastic fluctuations within the stellar cmds , contamination of the gc samples by tidally stripped dwarf galaxy nuclei , accretion of gcs found in present - day low - mass galaxies , and field star capture by gcs @xcite .
none of these effects were found to be likely contributors to the shape of the blue tilt in a significant manner . on the other hand , one effect known to have the potential to cause
a significant mass - metallicity relation among massive gcs is that of gc self - enrichment by means of retention and reprocessing of snii ejecta ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the remainder of this section , we re - examine the gc self - enrichment scenario of ( * ? ? ?
* hereafter bh09 ) and its expected effect on the mass vs. @xmath16 relation among gcs by taking into account the effects of gc mass loss by both stellar evolution and dynamical evolution for different imf shapes . the latter effects can be divided up as follows : _
( i ) _ the effect of gc self - enrichment on the @xmath16 colors of gcs .
this effect becomes increasingly important for gc masses above a certain minimum mass ; _ ( ii ) _ the influence of gc mass loss on the present - day masses ( and hence magnitudes ) at which a given effect of gc self - enrichment is reached ( including the dependence thereof on the adopted imf ) ; and _ ( iii ) _ the influence of dynamical evolution on the @xmath16colors of gcs ( and its dependence on the adopted imf ) .
the latter effect was shown and discussed above in section [ s : gzevol ] .
we refer to bh09 for a full description of their gc self - enrichment model , including its assumptions , caveats and comparisons with other such models .
briefly , the bh09 model is built upon the assumptions that the energy of the snii ejecta is fully converted to kinetic energy , and that self - enrichment occurs when this kinetic energy is equal to or less than the binding energy of the primordial gas cloud that formed the gc .
the metallicity of the protocluster cloud is parameterized as follows ( see equation ( 7 ) of bh09 ) : @xmath71 where @xmath72 represents the metallicity of the gas cloud prior to self - enrichment , the factor 10@xmath73 represents the sn yield used by bh09 , @xmath74 represents the star formation efficiency ( for which we adopt bh09 s value of 0.3 ) , and @xmath75 denotes the fraction of metals retained within the cloud and incorporated during the formation of the lower - mass stars . in deriving @xmath75 ,
bh09 define their `` reference model '' as one in which the radial mass density profile of the protocluster cloud falls off as an isothermal sphere ( i.e. , @xmath76 ) . for this case
they find @xmath77 ( see equation 28 of bh09 ) where @xmath78 is the energy released per snii ( taken to be 10@xmath79 erg ) , @xmath80 is the truncation radius of the protocluster cloud ( taken as 1 pc in bh09 s reference model ) , and @xmath81 is the protocluster cloud mass .
to estimate protocluster cloud masses from observed present - day gc masses , we use @xmath82 where @xmath83 is the initial gc mass and @xmath84 is the cumulative mass loss of the gc at an age of 12 gyr , consisting of mass loss by stellar evolution ( @xmath85 ) and by dynamical evolution ( @xmath86 ) .
we determine values for @xmath87 by using the `` average '' models of k09 . ] which were shown by open circles in fig .
[ f : mcurr_vs_color ] .
figure [ f : mcurr_minit ] depicts @xmath88 versus @xmath39 for the three imfs considered in this paper .
note the significant dependence of the cumulative gc mass loss on the imf shape , especially for the most massive gcs .
this effect is mainly caused by @xmath85 being significantly larger for imfs with shallower slopes .
metallicities predicted by the bh09 reference model through equation ( [ eq : f_z ] ) are converted to observed @xmath16 colors by means of the relation of @xcite , adopting @xmath89 = 0.016 @xcite and @xmath90 = [ fe / h ] + 0.25 ( @xmath91 0.10 ) as found for galactic gcs as well as gcs in giant early - type galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . to compare model predictions with the observed gc color - mass relations , we assume pre - enrichment levels of [ z@xmath92/h ] = @xmath561.45 and [ z@xmath92/h ] = @xmath560.25 for the metal - poor and metal - rich gc subpopulations , respectively
these values yield @xmath16 colors that agree with those of the observed red and blue tilts , respectively , at the low - mass limit of the gc sample . in the top panels of figure [ f : massloss_all ] ( i.e. , figs . [
f : massloss_all]a[f : massloss_all]c ) we plot the color - mass relations that would be expected if gcs are self - enriched according to the reference bh09 model , after which the gcs undergo 12 gyr of mass loss according to the `` average '' k09 models for the three different imfs . these diagrams are based on fig .
[ f : mcurr_vs_color ] after removing the symbols associated with the k09 model predictions .
the dotted lines indicate the color - mass relation due purely to self - enrichment , while the dashed lines do so for the case of pre - enrichment _ plus _ self - enrichment .
each set of lines is drawn twice ( using different pen colors ) to indicate the predictions with and without the effect of gc mass loss .
concentrating on the results for the metal - poor gc subpopulation , we see that the effect of gc mass loss on the bh09 model predictions is significant .
specifically , the color - mass relation predicted by the standard bh09 model is consistent ( to within the uncertainties ) with the observed color - mass relation from @xcite for kroupa and salpeter imfs . for the @xmath93 imf ,
the predicted colors are somewhat bluer than the observed color - mass relation in the mass range @xmath94 .
conversely , looking at the predictions for the metal - rich gc subpopulation , we see that the effect of self - enrichment is expected to become observable only for current gc masses @xmath95 , whereas the red tilt is observed down to significantly lower gc masses .
the combined effects of gc self - enrichment and the evolution of the stellar mass function in gcs on their @xmath16 colors are illustrated in the bottom panels of figure [ f : massloss_all ] ( i.e. , figs . [
f : massloss_all]d[f : massloss_all]f ) .
this is done by first converting the gc mass - dependent @xmath16 colors due to the effect of dynamical evolution of gcs on @xmath96 ( shown in figs .
[ f : mf_and_m_lz_vs_logm]a c and [ f : color_vs_alpha ] ) to metallicity space using the transformation of @xcite . to this
we add the metallicity due to gc self - enrichment ( i.e. , the equivalent of the blue dotted lines in figs .
[ f : massloss_all]a c ) .
the total metallicities are then converted back to @xmath16 using the @xcite transformation as before .
[ f : massloss_all]d[f : massloss_all]f reveal an interesting result for the metal - poor gc subpopulation : when adding the simulated effects of gc self - enrichment ( including 12 gyr of gc mass loss ) and the evolution of @xmath96 in gcs on their @xmath16 colors , the resulting color - mass relation becomes consistent with the observed blue tilt among metal - poor gcs in giant ellipticals _ for all imf shapes considered here_. in fact , differences between results that use different imf shapes are smaller than the uncertainties in this respect .
this result is due to the fact that while the effect of the mass - dependent evolution of @xmath96 ( due to dynamical evolution of gcs ) on the @xmath16 colors is significantly stronger for steep imfs ( e.g. , @xmath973.0 ) than for salpeter or kroupa imfs , the opposite is true for the effect of gc mass loss due to stellar evolution ( and hence the impact of gc self - enrichment for gcs with @xmath98 ) , and the two effects cancel out each other to within the uncertainties .
the situation is different for the metal - rich gc subpopulation .
the effect of gc self - enrichment on the @xmath16 colors of metal - rich gcs is much weaker than for metal - poor gcs : our calculations indicate that only the most massive metal - rich gcs ( with @xmath99 ) will have had their @xmath16 colors slightly affected by self - enrichment ( by @xmath100 relative to less massive gcs ) .
conversely , the effect of the mass - dependent evolution of @xmath96 on the @xmath16 color causes an approximately linear color - mass relation over the full range of gc masses considered , and the tilt of this relation is only significant for bottom - heavy imfs ( cf .
section [ s : gzevol ] ) .
the presence of a color - mass relation among metal - rich gcs in giant ellipticals down to masses @xmath101 , as seen in the large sample of @xcite , therefore seems to indicate that the metal - rich gcs in giant ellipticals were formed with a bottom - heavy imf ( @xmath102 ) .
this result reinforces the conclusions of paper i as well as those of various recent galaxy studies based on gravity - sensitive spectral features in the red part of the optical spectrum ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and dynamical @xmath52 values that exceed those predicted by stellar population models based on a kroupa imf ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . as to the calculations of the effects of gc self - enrichment done in section [ s : selfenr ] , the main differences between our results and those of both bh09 itself and @xcite are related to the treatment of gc mass loss .
specifically , bh09 did not take gc mass loss into account when comparing their simulations with observed gc color - magnitude relations in giant elliptical galaxies .
@xcite did implement an aspect of gc mass loss , using the gc dissolution scenario of @xcite to estimate gc mass loss .
however , the latter study only considered gc mass loss due to two - body relaxation and hence did not incorporate the important effect of mass loss due to stellar evolution ( @xmath940% of the initial mass for a kroupa or chabrier imf at an age of 12 gyr : see , e.g. , @xcite or @xcite ; see also fig .
[ f : mcurr_minit ] ) .
the impact of a full treatment of gc mass loss ( and its dimming effect ) on self - enrichment estimates is significant .
one illustration of this is that @xcite noted two main apparent problems with their implementation of the reference model of bh09 in reproducing the blue tilt seen among gcs in giant ellipticals : _
( i ) _ the model predicts self - enrichment to become important only at masses @xmath103 whereas the blue tilt starts at a mass that is an order of magnitude lower , and _
( ii ) _ the mass - metallicity relation predicted by the model is significantly more pronounced in the self - enrichment regime than the observed blue tilt .
consequently , @xcite looked into changes to the assumptions of the radial density profile and star formation efficiency in the reference bh09 model to improve the agreement of the model predictions with the data .
in particular , mieske et al .
needed to make the radial density profile of the protocluster cloud significantly steeper than an isothermal profile to make the self - enrichment model fit the observed mmr , which seems physically unlikely . however , as shown in fig .
[ f : massloss_all ] , the implementation of a full treatment of gc mass loss ( for which the impact to the mass - metallicity relation depends on gc mass as well as the imf ) renders a situation where the reference bh09 model reproduces the blue tilt exhibited by the massive gcs very well . in this section
we assess to what extent our results shown in fig .
[ f : massloss_all ] might depend on the assumptions and ingredients of the color - metallicity relation and the gc dynamical evolution model used to derive them . to translate metallicity variations into @xmath16 color changes in section [ s : selfenr ] , we followed @xcite and used the broken linear function of @xcite which was derived from integrated - light @xmath10 and @xmath11-band photometry of 95 gcs in the milky way , m49 , and m87 that have spectroscopic metallicities . however , the scatter in the peng et al .
relation is significant , especially at intermediate metallicities ( @xmath104 } \la -0.5 $ ] ) and supersolar ones , and uncertainties of several individual data points are substantial . to assess the uncertainty associated with the choice of a particular [ z / h](@xmath16 ) relation , we apply two other such relations from the recent literature for comparison purposes : _
( i ) _ the quartic polynomial function of @xcite , which was derived from the same data as the @xcite function , and _
( ii ) _ the broken linear function of @xcite , which was derived from an independent empirical database of colors and spectroscopic metallicities of 903 extragalactic gcs .
as might be expected , the three [ z / h](@xmath16 ) relations imply different amounts of gc pre - enrichment needed to let the mean color of the modeled blue and red tilts agree with the observed ones .
specifically , the @xcite relation implies [ z@xmath92/h ] = @xmath561.37 and [ z@xmath92/h ] = -0.22 for the blue and red gcs , respectively , while the [ z@xmath92/h ] values are @xmath561.20 and @xmath560.18 when using the @xcite relation .
the sensitivity of our results to the choice of a given different color - metallicity relation is illustrated in figure [ f : cmcompare1 ] , which consists of three copies of figs .
[ f : massloss_all]c and [ f : massloss_all]f , one for each color - metallicity relation .
while the resulting shapes of the blue and red tilts depend somewhat on the color - metallicity relation in a quantitative sense , they are all consistent with one another to within the 1-@xmath25 uncertainties . moreover , our main results do not change : the addition of the effect of 12 gyr of gc mass loss to that of self - enrichment yields a good fit to the observed blue tilt , while the shape of the observed red tilt indicates a bottom - heavy imf at sub - solar masses for the metal - rich population . to estimate the variation in the effects of 12 gyr of gc mass loss among different gc dynamical evolution models
, we compare the k09 model with that of ( * ? ? ?
* hereafter fz01 ) .
the main difference between these two models is that fz01 employs a constant , gc mass - independent value for the ( present - time ) @xmath52 to compare model masses with observed luminosities , while k09 uses gc mass - dependent mass loss rates that result in gc mass - dependent @xmath52 rates ( see fig .
[ f : mf_and_m_lz_vs_logm ] ) .
the two models also use different parametrizations of gc mass loss by two - body relaxation , and k09 incorporates a prescription of the effect of stellar remnant retention which fz01 did not .
however , in spite of these different model ingredients , the only significant differences between the gc mass functions ( gcmfs ) for the galactic gc system at an age of 12 gyr as inferred by the two models show up at _ low _ gc masses ( @xmath105 , see * ? ? ?
for the massive gcs considered in this paper ( @xmath106 ) , the difference between the gcmfs inferred by the two models are only of order 10% for any given mass bin , which is within the 1@xmath25 poisson error bars ( see figure [ f : mfcompare ] ) . since these differences are smaller than the differences between k09 models with different values of dissolution timescales @xmath31 employed in this paper ( illustrated by the blue error bars in figs .
[ f : massloss_all]a[f : massloss_all]c ) , we believe our results on the amounts of gc mass loss , and their uncertainties , are robust . , assuming @xmath107 ( open squares and 1-@xmath25 error bars ) .
the solid line represents the gcmf calculated from the gc luminosity function derived by @xcite using a model involving mass - dependent @xmath52 ratios and dissolution timescales @xmath31 determined for each individual gc ( see their fig .
the dashed line represents a fit to the galactic gcmf using a model assuming a mass - independent @xmath52 ratio ( as in * ? ? ?
data are from @xcite .
note the similarity of the two curves in the mass range of gcs considered in this paper .
[ f : mfcompare ] , width=253 ] as mentioned in the introduction , the color - mass relations among gcs show environmental dependencies @xcite .
firstly , gcs associated with high - mass galaxies exhibit more pronounced blue _ and _ red tilts than do gcs in lower - mass galaxies .
secondly , the blue tilt is found to be more pronounced among gcs at smaller galactocentric distances than among those further out .
one might expect such environmental dependences to exist among ancient gc systems due to dynamical evolution ( i.e. , tidal shocks and two - body relaxation ) , as tidal fields are stronger in more massive galaxies and at smaller galactocentric distances .
a recent illustration of this effect was shown by @xcite who performed n - body simulations of gcs in a realistic milky way - like potential and found a strong trend of increasing mass loss with decreasing galactocentric distance for a given type of gc ( initial mass , half - mass radius , concentration ; see also * ? ? ?
* ) . to simulate the effect of environmentally varying gc dissolution timescales on the color - mass relations among gcs for a given imf ,
we select gcs whose cumulative mass loss over 12 gyr of stellar and dynamical evolution are at the 20% and 80% percentiles of the range found among the gc families considered in this paper ( see figure [ f : mcurr_minit]c ) , and compare the resulting color - mass relations with that derived for the `` average '' k09 model . this comparison is depicted in figure [ f : cmcompare2 ] for the case of the kroupa imf and the @xcite color - metallicity relation .
it is obvious that the shape of the blue tilt is indeed sensitive to the amount of cumulative gc mass loss ( and its impact on self - enrichment in the observed color - mass frame ) : the blue tilt only shows up for @xmath108 for the `` low '' mass loss case while it is present and pronounced all the way down to the minimum mass of the gc sample for the `` high '' mass loss case .
note that this behavior is entirely consistent with the observed environmental dependencies of the blue tilt found by @xcite .
note also that in contrast with the blue tilt , the dependence of the shape of the _ red _ tilt on the cumulative amount of gc mass loss is negligible in terms of the impact of gc mass loss on self - enrichment in the observed color - mass frame .
this suggests that the presence of the red tilt in massive ellipticals and its absence in low - mass ellipticals is _ not _ mainly due to differences in cumulative gc mass loss .
instead , we suggest that it indicates that metal - rich gcs in massive ellipticals were formed with a bottom - heavy imf , as already mentioned in section [ s : combined ] . several studies of the blue tilt in giant elliptical galaxies have pointed out that the color - magnitude relations among metal - poor gcs can vary significantly from galaxy to galaxy , even among massive galaxies with similar luminosities . a well - known case in point
is m49 , the brightest giant elliptical galaxy in the virgo cluster , which does not show an obvious blue tilt whereas ngc 1399 and m87 , two other giant elliptical galaxies with similar luminosities , do @xcite . while such differences might seem difficult to reconcile within a scenario in which the mass - metallicity relation among massive gcs is primarily due to self - enrichment
, one should consider the following .
first and foremost , we emphasize that the main purpose of this study has been to compare calculations of the impact of gc mass loss and self - enrichment to the _ average _ color - mass relations among gcs in a large sample of giant elliptical galaxies .
several factors render averaging over many galaxies a necessity in this context .
as shown and discussed above , @xmath2612 gyr of dynamical evolution can cause several differences between different families of gcs ( especially for the case of bottom - heavy imfs ) .
one example of such differences is the relative amount of mass lost ( see section [ s : t_0 ] and figs .
[ f : mcurr_minit ] and [ f : cmcompare2 ] ) .
this results in a spread of current luminosities among gcs that started out with a given initial mass , and hence with a similar amount of self - enrichment .
another property that is quite sensitive to the initial gc conditions is the change of the slope of stellar mass function at subsolar masses and its associated change of integrated @xmath16 color after 12 gyr ( see figs .
[ f : mf_and_m_lz_vs_logm]a c and [ f : color_vs_alpha ] as well as paper i ) .
the 1-@xmath25 uncertainties associated with these two effects in the color - mass plane are indicated in figs .
[ f : massloss_all]d f .
last but not least , there is a significant scatter among colors of metal - poor gcs in individual giant ellipticals ( @xmath109 } \simeq 0.3 $ ] ; e.g. , @xcite ) , which translates to @xmath110 .
this is most likely mainly due to a spread in pre - enrichment levels ( e.g. , bh09 ) . for gcs near the massive end of the gc mass function ( where gcs are scarce and hence stochastic variations between gcs are significant )
, the impact of this scatter on the resulting color - mass relation can be significant , especially for galaxies with relatively small specific gc frequencies .
note in this respect that m49 hosts significantly fewer gcs than ngc 1399 and m87 do , see figure 3 of @xcite .
that said , important steps forward in the context of understanding the nature of the color - mass relations of gcs as well as their variations between galaxies can already be made with currently available instrumentation .
one such step will be to obtain accurate photometry and size measurements of gcs _ outside the central few kpc _ sampled by the currently available hst observations of nearby giant ellipticals ( typically only a central pointing ) .
this will not only yield color - magnitude data for gcs in weaker tidal fields which will allow one to evaluate differences in sizes and hence in dynamical evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the lower background levels of the underlying diffuse galaxy light in the outer regions will also allow high - s /
n measurements of spectroscopic metallicities of gcs in the upper @xmath263 mag of the gc luminosity function with 8 - 10 m class telescopes ( e.g. , * ? ? ?
* ; * ? ? ?
the latter will be important to decipher the true color - metallicity relation ( rather than one derived from colors and [ fe / h ] values of galactic gcs ; for example , the few galactic gcs that are moderately metal - rich ( @xmath111 \ga -0.5 $ ] ) have significant foreground extinction ) and its impact on the nature of the blue tilt .
driven by several lines of recent evidence for a `` bottom - heavy '' stellar imf in massive elliptical galaxies , we have investigated the influence of the shape of the imf on the color - magnitude relations for the metal - poor and metal - rich subpopulations of gcs in such galaxies . to this end
we used mass loss calculations for gcs for three imf shapes and a large range of initial gc masses , in conjunction with calculations of integrated - light colors from isochrones .
these calculations were used for two main purposes : _ ( i ) _ to evaluate the effect of evolution of the stellar mass function at subsolar masses on the color - magnitude relations , and _
( ii ) _ to evaluate the impact of gc mass loss on the ( * ? ? ? * bh09 ) model of gc self - enrichment when comparing model predictions to observed color - magnitude diagrams .
our results are compared to observed color - magnitude relations among a very large set of gcs in giant elliptical galaxies from @xcite .
our main conclusions are as follows . * as to the observed color - magnitude relation among the metal - rich gcs in giant elliptical galaxies ( often referred to as the `` red tilt '' )
, we found that the effect of dynamical evolution of gcs on the stellar mass functions over a hubble time can explain the observed slope of the red tilt along its full extent _ if ( and only if ) the imf was bottom - heavy _
( e.g. , @xmath0 in @xmath1 at subsolar masses ) . + [ -3.3ex ] * conversely , we find that gc self - enrichment is only able to cause a small reddening for the most massive metal - rich gcs ( i.e. , current masses @xmath112 ) .
hence we argue that the observed color - magnitude relation among metal - rich gcs in massive elliptical galaxies constitutes evidence for a bottom - heavy imf in such galaxies .
+ [ -3.3ex ] * as to the `` blue tilt '' among metal - poor gcs , we find that the influence of dynamical evolution of gcs on the stellar mass functions over a hubble time can reproduce the observed tilt only up to a current gc mass @xmath113 , and only if the imf was bottom - heavy . beyond this mass ,
the observed blue tilt is more pronounced than what can be explained by the evolution of stellar mass functions in gcs , so that additional effects ( such as self - enrichment ) are still required to reproduce the full extent of the observed blue tilt .
+ [ -3.3ex ] * when considering the dimming effect of 12 gyr of mass loss in color - magnitude diagrams of gcs , we find that the `` reference '' gc self - enrichment model of bh09 reproduces the shape of the observed blue tilt very well in the cases of kroupa or salpeter imfs .
this result differs from those of previous studies , which did not take gc mass loss by stellar evolution into account .
+ [ -3.3ex ] * when adding the simulated effects of gc self - enrichment and the evolution of stellar mass functions in metal - poor gcs on their @xmath16 colors , the resulting color - mass relation becomes consistent with the observed blue tilt _ for all imf shapes considered here_. this result is due to the fact that while the effect of the gc mass - dependent evolution of stellar mass functions to the @xmath16 colors is significantly stronger for steep imfs ( e.g. , @xmath1143.0 ) than for salpeter or kroupa imfs , the opposite is true for the effect of gc mass loss due to stellar evolution , and the two effects cancel out each other to within the uncertainties .
+ [ -3.3ex ] * we looked into the sensitivity of our results to the choice of the color - metallicity relation . while details such as the mean pre - enrichment metallicity required to make our modeled color - mass relations fit the observed blue and red tilts do differ slightly among color - metallicity relations from the recent literature , the _ shapes _ of the modeled color - mass relations
are found to be consistent from one color - metallicity relation to another .
similarly , the sensitivity of our results to the choice of gc dynamical evolution model was found to be insignificant in the relevant range of gc masses .
+ [ -3.3ex ] * differences in cumulative gc mass loss expected for different galactic environments ( e.g. , inner versus outer regions of massive galaxies ) are found to affect the shape of the blue tilt in a way that is qualitatively consistent with the observed variations . however , the impact of varying gc mass loss on the shape of the _ red _ tilt is found to be negligible .
we argue that further insights into the color - magnitude relations of gcs as well as their variations between galaxies can be obtained by means of high - resolution imaging as well as high - s / n spectroscopy of gcs outside the central few kpc sampled by the currently available hst observations of nearby giant ellipticals . with such information
, one will be able to constrain the true color - metallicity relation as well as environmental influences on the properties of the blue tilt .
we acknowledge useful discussions with jeremy bailin , charlie conroy , mike fall , mark krumholz , and enrico vesperini .
we highly appreciated the referee s insightful and constructive comments which improved the paper .
we made heavy use of the sao / nasa astrophysics data system while writing this paper .
pg was partially supported during this project by nasa through grant hst - go-11691 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas526555 .
jmdk acknowledges the hospitality of the aspen center for physics , which is supported by the national science foundation through grant no .
phy-1066293 .
bailin , j. , & harris , w. e. 2009 , , 695 , 1082 ( bh09 ) baumgardt , h. , & makino , j. 2003 , , 340 , 227 bellazzini , m. , ibata , r. a. , chapman , s. c. , et al .
2008 , , 136 , 1147 blakeslee , j. p. , cantiello , m. , & peng , e. w. 2010 , , 710 , 51 brodie , j. p. , & strader , j. 2006 , , 44 , 193 brodie , j. p. , usher , c. , conroy , c. , et al .
2012 , , 759 , l33 bruzual , g. a. & charlot , s. 2003 , , 344 , 1000 cappellari , m. , mcdermid , r. m. , alatalo , k. , et al .
2012 , , 484 , 485 carretta , e. , bragaglia , a. , gratton , r. g. , et al . 2010a , , 516 , a55 carretta , e. , bragaglia , a. , gratton , r. g. , et al .
2010b , , 714 , l7 chabrier , g. 2003 , , 115 , 763 cenarro , a. j. , beasley , m. a. , strader , j. , brodie , j. p. , & forbes , d. a. 2007 , , 134 , 391 cohen , j. g. , blakeslee , j. p. , & ct , p. 2003 , apj , 592 , 866 conroy , c. , & van dokkum , p. g. 2012 , , 760 , 71 ct , p. , blakeslee , j. p. , ferrarese , l. , et al .
2004 , , 153 , 223 de marchi , g. , paresce , f. , & pulone , l. 2007 , , 656 , l65 fall , s. m. , & zhang , q. 2001 , apj , 561 , 751 ( fz01 ) fall , s. m. , chandar , r. , & whitmore , b. c. 2009 , , 704 , 453 fellhauer , m. , kroupa , p. , & evans , n. w. 2006 , , 372 , 338 forbes , d. a. , spitler , l. r. , harris , w. e. , et al .
2010 , , 403 , 429 geisler , d. , lee , m. g. , & kim , e. 1996 , , 111 , 1529 georgiev , i. y. , hilker , m. , puzia , t. h. , goudfrooij , p. , & baumgardt , h. 2009 , , 396 , 1075 gieles , m. , heggie , d. c. , & zhao , h. 2011 , , 413 , 2509 goudfrooij , p. 2012 , apj , 750 , 140 goudfrooij , p. , gilmore , d. , whitmore , b. c. , & schweizer , f. 2004 , , 613 , l121 goudfrooij , p. , & kruijssen , j. m. d. 2013 , apj , 762 , 107 ( paper i ) goudfrooij , p. , schweizer , f. , gilmore , d. , & whitmore , b. c. 2007 , , 133 , 2737 gratton , r. g. , carretta , e. , & bragaglia , a. 2012 , , 20 , 50 harris , w. e. 1996 , , 112 , 1487 harris , w. e. , whitmore , b. c. , karakla , d. , et al .
2006 , , 636 , 90 harris , w. e. , kavelaars , j. j. , hanes , d. a. , pritchet , c. j. , & baum , w. a. 2009 , , 137 , 3314 harris , w. e. , spitler , l. r. , forbes , d. a. , & bailin , j. 2010 , , 401 , 1965 hnon , m. 1969 , , 2 , 151 johnson , c. i. , & pilachowski , c. a. 2010 , , 722 , 1373 jordn , a. , blakeslee , j. p. , ct , p. , et al . 2007a , , 169 , 213 jordn , a. , mclaughlin , d. e. , ct , p. , et al . 2007b , , 171 , 101 jordn , a. , peng , e. w. , blakeslee , j. p. , et al .
2009 , , 180 , 54 king , i. r. 1966 , , 71 , 64 kirby , e. n. , guhathakurta , p. , & sneden , c. 2008 , , 682 , 1217 kroupa , p. 2001
, , 322 , 231 kruijssen , j. m. d. 2009 , , 507 , 1409 ( k09 ) kruijssen , j. m. d. & lamers , h. j. g. l. m. 2008 , , 490 , 151 kruijssen , j. m. d. , & portegies zwart , s. f. 2009 , , 698 , l158 kruijssen , j. m. d. , pelupessy , f. i. , lamers , h. j. g. l. m. , et al .
2012 , , 421 , 1927 kundu , a. 2008 , , 136 , 1013 lamers , h. j. g. l. m. , gieles , m. , bastian , n. , et al .
2005 , , 441 , 117 lsker , r. , van den bosch , r. c. e. , van den ven , g. , et al .
2013 , , 434 , l31 madrid , j. p. , hurley , j. r. , & sippel , a. c. 2012 , , 756 , 167 maraston , c. 2005 , , 362 , 799 marino , a. f. , sneden , c. , kraft , r. p. , et al .
2011 , , 532 , a8 marigo , p. , girardi , l. , bressan , a. , et al .
2008 , , 482 , 883 mclaughlin , d. e. , & fall , s. m. 2008 , , 679 , 1272 mclaughlin , d. e. , & van der marel , r. p. 2005
, , 161 , 304 mieske , s. & baumgardt , h. 2007 , , 475 , 851 mieske , s. , jordn , a. , ct , p. , et al .
2006 , , 653 , 193 mieske , s. , jordn , a. , ct , p. , et al .
2010 , , 710 , 1672 morgan , s. , & lake , g. 1989 , , 339 , 171 norris , j. e. , & da costa , g. s. 1995 , , 447 , 680 paust , n. e. q. , reid , i. n. , piotto , g. , et al .
2010 , , 139 , 476 peng , e. w. , jordn , a. , ct , p. , et al .
2006 , , 639 , 95 peng , e. w. , jordn , a. , blakeslee , j. p. , et al .
2009 , , 703 , 42 peng , e. w. , ferguson , h. c. , goudfrooij , p. , et al .
2011 , , 730 , 23 puzia , t. h. , kissler - patig , m. , thomas , d. , et al .
2005 , , 439 , 997 puzia , t. h. , kissler - patig , m. , & goudfrooij , p. 2006
, , 648 , 383 recchi , s. , & danziger , i. j. 2005 , , 436 , 145 salpeter , e. e. 1955 , , 121 , 161 sarajedini , a. , & layden , a. c. 1995 , , 109 , 1086 shetrone , m. d. , ct , p. , & sargent , w. l. w. 2001 , , 548 , 592 smith , r. j. , lucey , j. r. , & carter , d. 2012 , , 421 , 2982 sonnenfeld , a. , treu , t. , gavazzi , r. , et al .
2012 , , 752 , 163 spitler , l. r. , larsen , s. s. , strader , j. , et al .
2006 , , 132 , 1593 strader , j. , & smith , g. h. 2008 , , 136 , 1828 strader , j. , brodie , j. p. , spitler , l. r. , & beasley , m. a. 2006 , , 132 , 2333 tortora , c. , romanowsky , a. j. , & napolitano , n. r. 2013 , , 765 , 8 usher , c. , forbes , d. a. , brodie , j. p. , et al .
2012 , , 426 , 1475 vandenberg , d. a. , gustaffson , b. , edvardsson , b. , eriksson , k. , & ferguson , j. 2007 , , 666 , l105 van dokkum , p. g. , & conroy , c. 2010 , , 468 , 940 vesperini , e. , & heggie , d. c. 1997 , , 289 , 898 waters , c. z. , zepf , s. e. , lauer , t. r. , & baltz , e. a. 2009 , , 693 , 463 zaritsky , d. , colucci , j. e. , pessev , p. m. , bernstein , r. a. , & chandar , r. 2012 , , 761 , 93 zinn , r. j. , & west , m. j. 1984 , , 55 , 45 | several recent studies have provided evidence for a `` bottom - heavy '' stellar initial mass function ( imf ) in massive elliptical galaxies . here
we investigate the influence of the imf shape on the recently discovered color - magnitude relation ( cmr ) among globular clusters ( gcs ) in such galaxies .
to this end we use calculations of gc mass loss due to stellar and dynamical evolution to evaluate _
( i ) _ the shapes of stellar mass functions in gcs after 12 gyr of evolution as a function of current gc mass along with their effects on integrated - light colors and mass - to - light ratios , and _
( ii ) _ their impact on the effects of gc self - enrichment using the 2009 `` reference '' model of bailin & harris . as to the class of metal - poor gcs
, we find the observed shape of the cmr ( often referred to as the `` blue tilt '' ) to be very well reproduced by bailin & harris reference self - enrichment model once 12 gyr of gc mass loss is taken into account .
the influence of the imf on this result is found to be insignificant .
however , we find that the observed cmr among the class of _ metal - rich _ gcs ( the `` red tilt '' ) can only be adequately reproduced if the imf was bottom - heavy ( @xmath0 in @xmath1 ) , which causes the stellar mass function at subsolar masses to depend relatively strongly on gc mass .
this constitutes additional evidence that the metal - rich stellar populations in giant elliptical galaxies were formed with a bottom - heavy imf . |
there are a lot of answers to a question `` why does not a knot come loose ? '' .
the next example gives one of the answers . in figure [ trefoil ] ,
the trefoil knot diagram is partitioned by 3 vertices into 3 edges @xmath0 and @xmath1 .
the edge @xmath2 is over @xmath3 , @xmath3 is over @xmath1 and @xmath1 is over @xmath2 .
this relation is known to be a three - way deadlock .
it seems that this is one of the factors that why a knot can not come loose .
throughout this paper we work in the piecewise linear category .
we shall study knots and links in the three - dimensional euclidean space @xmath4 . for the standard definitions and results of knots and links
, we refer to @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite .
let @xmath5 be a knot or link and @xmath6 be a diagram of @xmath5 .
we partition @xmath6 by @xmath7 vertices into @xmath7 edges @xmath8 .
we assume that each component of @xmath5 has at least one vertex .
we call such a diagram @xmath6 an _ @xmath7-partitioned diagram_. an @xmath7-partitioned diagram @xmath6 is an
_ @xmath7-cycle presentation _ of @xmath5 if 1 .
each edge of @xmath6 has no self - crossing . 2 .
for any pair of two edges @xmath9 and @xmath10 , exactly one of the following holds . 1 .
@xmath9 is over @xmath10 at every crossing of @xmath9 and @xmath10 .
@xmath9 is under @xmath10 at every crossing of @xmath9 and @xmath10 .
3 . there is no crossing of @xmath9 and @xmath10 .
we define an _ edge number _ @xmath11 of @xmath5 as the minimum number of @xmath7 where @xmath7 is taken over all @xmath7-cycle presentation of @xmath5 .
we can make a directed graph @xmath12 from an @xmath7-cycle presentation @xmath6 as follows . 1 . for each edge @xmath9 , we assign a vertex @xmath13 to @xmath12 .
2 . for two edges @xmath9 and @xmath10 which have at least one
crossing , if @xmath9 is over ( resp . under )
@xmath10 , then we assign an oriented edge from @xmath9 to @xmath10 ( resp . from @xmath10 to @xmath9 ) to @xmath12 .
figure [ digraph ] shows a digraph @xmath12 obtained from a 3-edge presentation @xmath6 in figure [ trefoil ] . ]
if we exclude the condition ( 2 ) in the definition of an @xmath7-cycle presentation , then any knot has a 2-cycle presentation .
figure [ trefoil2 ] will help you to show this .
[ non - trivial ] a knot @xmath5 is non - trivial if and only if @xmath14 . [ 3-cycle ] let @xmath5 be a knot with @xmath15 and @xmath6 a 3-cycle presentation of @xmath5 . then @xmath12 is an oriented 3-cycle .
[ main ] for any minimal @xmath7-cycle presentation @xmath6 , 1 .
@xmath12 is connected .
@xmath12 is not a path . for a non - trivial knot @xmath5 and any @xmath7-cycle presentation @xmath6 of @xmath5
, @xmath12 contains at least one oriented cycle .
[ exist ] for any @xmath16 @xmath17 , there exists a knot such that @xmath18 and @xmath15 .
[ bridge ] @xmath19 .
@xmath20 .
find a knot @xmath5 with @xmath21 .
we expect that @xmath22 is a candidate .
( of proposition [ non - trivial ] ) ( @xmath23 ) suppose that @xmath24 and let @xmath6 be a 1-cycle presentation of @xmath5 .
then by the definition ( 1 ) of an @xmath7-cycle presentation , @xmath6 has no crossing .
hence @xmath5 is trivial .
next suppose that @xmath25 and let @xmath6 be a 2-cycle presentation of @xmath5 .
@xmath6 has two edges @xmath2 and @xmath3 and without loss of generality we may assume that @xmath2 is over @xmath3 at every crossings of @xmath2 and @xmath3 . then by the definition ( 1 ) of an @xmath7-cycle presentation , @xmath6 is a 1-bridge presentation with an over - bridge @xmath2 and an under - bridge @xmath3 .
hence @xmath5 is trivial .
( of proposition [ 3-cycle ] ) let @xmath6 be a 3-cycle presentation of @xmath5 with three edges @xmath0 and @xmath1 .
suppose that @xmath12 does not form an oriented 3-cycle .
without loss of generality , we may assume that @xmath2 is over @xmath3 at every crossings of @xmath2 and @xmath3 , @xmath3 is over @xmath1 at every crossings of @xmath3 and @xmath1 , and @xmath2 is over @xmath1 at every crossings of @xmath2 and @xmath1 .
then @xmath6 is a descending diagram by specifying an orientation in the order @xmath27 .
therefore @xmath5 is trivial and this contradicts @xmath15 and proposition [ non - trivial ] .
hence @xmath12 forms an oriented 3-cycle . 1 .
for any successive vertices @xmath29 of @xmath12 , both of them can not be a source nor sink in @xmath30 .
2 . for any successive non - adjacent vertices @xmath29 of @xmath12 ,
there exists a vertex @xmath31 in @xmath32 such that @xmath33 is an oriented path , \(1 ) suppose without loss of generality that there exist successive vertices @xmath29 of @xmath12 such that both of them are source in @xmath30 . since the subarc @xmath36 is over all other edges @xmath37 , there is an isotopy of @xmath6 such that @xmath36 has no self - crossing
. then we can regard the subarc @xmath36 as a single edge @xmath38 , and the new edge @xmath38 satisfies the definition ( 1 ) and ( 2 ) of an @xmath7-cycle presentation .
see figure [ reduce1 ] .
hence we obtain an @xmath39-cycle presentation of @xmath5 and this contradicts @xmath28 .
\(2 ) suppose without loss of generality that there exist successive non - adjacent vertices @xmath40 such that for any vertex @xmath31 in @xmath32 , the edge @xmath41 ( resp .
@xmath42 ) has an orientation from @xmath13 to @xmath31 ( resp . from @xmath43 to @xmath31 ) .
we regard the subarc @xmath36 as a single edge @xmath38
. then @xmath38 satisfies the definition ( 1 ) and ( 2 ) of an @xmath7-cycle presentation .
see figure [ reduce2 ] .
hence we obtain an @xmath39-cycle presentation of @xmath5 and this contradicts @xmath28 .
_ `` proof '' of conjecture [ bridge ] .
_ let @xmath5 be a knot with @xmath28 and @xmath6 an @xmath7-cycle presentation of @xmath5 on the 2-sphere @xmath47 dividing the 3-sphere @xmath48 into two 3-balls @xmath49 and @xmath50 .
we push all subarcs of @xmath6 , which are regular neighbourhoods of vertices , into @xmath50 , and pull the rest of @xmath6 into @xmath49
. then @xmath5 intersects @xmath47 in @xmath51-points , and @xmath52 consists of properly embedded @xmath7-arcs in @xmath53 .
conjecture [ bridge ] will be proved if we can show that the @xmath7-string tangle @xmath54 is trivial . | we introduce a new numerical invariant of knots and links made from the partitioned diagrams .
it measures the complexity of knots and links . |
quantum chromodynamics ( qcd ) offers the description of hadrons in terms of quarks and gluons .
there are two basic ingredients of that picture : bound state dynamics of hadrons and fundamental interactions of quarks and gluons . while the former is still rather elusive to existing theoretical tools , the latter is rather well understood . the description of the hadronic processes at large momentum transfer is realized by making use of the factorization of high and low energy ( short and long distance ) dynamics .
the existence of asymptotic freedom makes then the high energy part tractable to the perturbative calculation , i.e. , the perturbative qcd ( pqcd ) .
exclusive processes are defined as the scattering reactions in which the kinematics of all initial and final state particles are specified , like , for example , the processes defining the hadron form factors ( @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 ) , the two - photon annihilation processes ( @xmath5 , @xmath6 , @xmath4 ) , the hadron scatterings ( @xmath7 , @xmath8 , @xmath4 ) , the decays of heavy hadrons ( @xmath9 , @xmath10 , @xmath4 ) etc .
the hard exclusive reactions , i.e. , the exclusive reactions at large momentum transfer ( or wide - angle ) , can be described by the so - called hard - scattering picture @xcite .
the basis of this picture is the factorization of short and long distance dynamics , i.e , the factorization of the hard - scattering amplitude into the elementary hard - scattering amplitude and hadron distribution amplitudes one for each hadron involved in the process .
usually the following standard approximations are made .
hadron is replaced by the valent fock state , collinear approximation , in which hadron constituents are constrained to be collinear , is adopted , and the masses are neglected . for example , in the case of the pion this leads to replacing the pion by @xmath11 ( correct flavour structure has to be taken into account ) , adopting @xmath12 , @xmath13 , where @xmath14 , @xmath15 and @xmath16 are pion , quark and antiquark momenta , respectively , while @xmath17 is the longitudinal momentum fraction , and taking @xmath18 .
generally , the hard - scattering amplitude is then represented by the following convolution formula : @xmath19 \quad { t_h}(x_{j } , \ , { q^2},\ , { \mu_{f}^{2 } } ) \quad { \prod_{{h}_{i } } } \ , \phi_{{h}_{i}}(x_{j } , \ , { \mu_{f}^{2 } } ) \ , , \ ] ] @xmath20 = \prod_{j=1}^{{n}_{h_i } } dx_{j } \ , \delta(1-\sum_{k=1}^{{n}_{h_i } } x_{k } ) \ , , \label{eq : conv}\ ] ] where @xmath21 is the process - dependent elementary hard - scattering amplitude , @xmath22 is the process - independent distribution amplitude ( da ) of the hadron @xmath23 , @xmath24 denotes the large momentum transfer while @xmath25 is the factorization scale at which the separation between short and long distance dynamics takes place . within this framework leading - order ( lo )
predictions have been obtained for many exclusive processes .
it is well known , however , that , unlike in qed , the lo predictions in pqcd do not have much predictive power , and that higher - order corrections are essential for many reasons . in general , they have a stabilizing effect reducing the dependence of the predictions on the schemes and scales .
therefore , to achieve a complete confrontation between theoretical predictions and experimental data , it is very important to know the size of radiative corrections to the lo predictions .
the list of exclusive processes at large momentum transfer analyzed at next - to - leading order ( nlo ) is very short and includes only three processes : the meson electromagnetic form factor @xcite , the meson transition form factor @xcite , and the process @xmath26
( @xmath27 , @xmath28 ) @xcite .
we note here that the meson transition form factor belongs to the same class of processes as the deeply virtual compton scattering ( dvcs ) @xcite ( @xmath29 ) , which recently has been extensively studied in the context of general parton distributions ( gpds ) @xcite . regarding the elementary hard - scattering amplitude , these two processes , or correspondingly subprocesses @xmath30 and @xmath31 ,
differ only in kinematic region and are related by crossing .
still , the nlo correction to dvcs has been calculated independently @xcite .
analogous connection exists between the pion electromagnetic form factor and the deeply virtual electroproduction of mesons ( dvem ) @xcite ( @xmath32 , the momentum transfer @xmath33 between the initial and the final proton is negligible , while the virtuality of the photon is large ) . in the context of subprocesses
, there is a connection between @xmath34 and @xmath35 . in @xcite
the use has been made of the nlo results for the pion electromagnetic form factor to obtain the nlo prediction for the specific case of dvem ( electroproduction of the pseudoscalar flavour non - singlet mesons ) . in this work
we mostly discuss the meson form factor calculations . at the next - to - next - to - leading order ( nnlo ) only the @xmath36-proportional terms for the deeply virtual compton scattering and pion transition form factor have been explicitly calculated @xcite .
the use of conformal constraints made it possible to circumvent the explicit calculation and to obtain the full nnlo results for the pion transition form factor @xcite . finally , let us note that apart from the deeply virtual region , also the wide angle region has been investigated in the literature in the context of the compton scattering ( wacs ) , as well as , electroproduction of mesons ( waem ) @xcite .
the nlo corrections were calculated only for the wacs @xcite . in this paper the introduction to hard - scattering picture for exclusive processes is given in sec .
[ sec : hsp ] .
the characteristic properties of the pqcd predictions regarding the importance of higher - order corrections and the renormalization scale ambiguities are explained in sec .
[ sec : pqcdprediction ] .
section [ sec : higherorder ] is devoted to the short review of exclusive processes calculated to the next - to - leading order ( nlo ) in the strong coupling constant @xmath37 : meson electromagnetic form factor ( @xmath38 ) , photon - to - meson transition form factor ( @xmath39 ) , meson pair production : @xmath40 . in sec .
[ sec : nnlo ] the next - to - next - to - leading order ( nnlo ) prediction for the photon - to - pion transition form factor obtained using conformal symmetry constraints is explained . finally , in sec .
[ sec : concl ] the summary and conclusions are given .
let us explain the basic ingredients of the standard hard - scattering picture by taking as an example the simplest exclusive quantity , i.e. , the photon - to - pion transition form factor @xmath41 appearing in the amplitude of the process @xmath42 .
at least one photon virtuality has to be large and we take here the simple case : @xmath43 and @xmath44 .
the full amplitude is of the form @xmath45 and the transition form factor can be represented by a convolution @xmath46 here @xmath47 and @xmath25 is a factorization scale .
the elementary hard - scattering amplitude @xmath21 obtained from @xmath48 is calculated using the pqcd . by definition , @xmath21 is free of collinear singularities and has a well defined expansion in @xmath49 , with @xmath50 being the renormalization ( or coupling constant ) scale of the hard - scattering amplitude .
thus , one can write @xmath51 the diagrams contributing to lo and representative diagrams contributing to nlo order are displayed in figs .
[ f : tlo ] and [ f : tnlo ] , respectively .
amplitude.,width=340 ] .,width=302 ] when evaluating the nlo amplitude one encounters the uv and collinear singularities .
the former are removed by coupling constant ( @xmath37 ) renormalization introducing the scale @xmath50 , while the latter are factorized into the da at the scale @xmath25 .
the pion distribution amplitude is defined in terms of the matrix elements of composite operators : @xmath52 . while the da form is taken as an ( nonperturbative ) input at some lower scale @xmath53 ( @xmath54 ) , its evolution to the factorization scale @xmath25 ( @xmath55 ) is governed by pqcd .
the da can hence be written in a form @xmath56 where @xmath57 denotes the evolution part of the da . in latter
the resummation of @xmath58 terms is usually included , and @xmath57 is obtained by solving the evolution equation @xmath59 where @xmath60 represents the perturbatively calculable evolution kernel .
one often introduces the distribution amplitude @xmath61 normalized to unity @xmath62 , and related to @xmath55 by @xmath63 where @xmath64 gev is the pion decay constant and @xmath65 is the number of colours .
the solutions of the evolution equation ( [ eq : eveq ] ) combined with the nonperturbative input can then be written in a form of an expansion over gegenbauer polynomials @xmath66 which represent the eigenfunctions of the lo evolution equation : @xmath67 \ , .\ ] ] here @xmath68 denotes the sum over even indices . the nonperturbative input @xmath69 as well as
the evolution is now contained in @xmath70 coefficients .
they have a well defined expansion in @xmath37 : @xmath71 where @xmath72 represent the lo and nlo @xcite parts whose exact form in @xmath73 factorization scheme is given in , for example , @xcite .
as the da input one often takes the asymptotic function @xmath74 being the solution of the da evolution equation for @xmath75 and the simplest possibility .
we list here two more choices from the literature @xmath76 \phi_{bms}\mbox{\cite{bakulevms01 } } : & b_2(1\ , \mbox{gev}^2)=0.188 \qquad b_4(1\ , \mbox{gev}^2)=-0.13 \ , .
\end{array}\ ] ] the cz distribution amplitude is nowadays mostly ruled out ( see @xcite and references therein ) , and it is believed that even at lower energies the pion da is close to asymptotic form but probably end - point suppressed like the bms da .
in this section we would like to discuss some properties inherent to all pqcd predictions .
let us first briefly discuss the expansion parameter , i.e. , the qcd coupling constant .
the qcd @xmath77 function given by @xmath78 is negative ( @xmath79 , @xmath80 is the number of active flavours ) and theory is asymptotically free . the usual one - loop solution of the renormalization group equation ( [ eq : beta ] )
is given by @xmath81 and , obviously , @xmath82 , while , for example , @xmath83 , @xmath84 .
obviously , low - energy behaviour of such @xmath37 represents a problem due to the existence of the landau pole : @xmath85 . in the literature one can encounter several prescriptions to improve @xmath37 in low - energy region . for example , `` frozen '' coupling constant @xmath86 or analytical coupling @xcite @xmath87 \ , .\ ] ] but even these improved forms of @xmath37 give rather large values at lower energies
thus , lo qcd predictions do not have much predictive power and higher - order corrections are important .
generally , the pqcd amplitude can be written in a form @xmath88 where @xmath24 is some large momentum and as usual @xmath50 represents the renormalization scale .
the truncation of the perturbative series to finite order introduces the residual dependence of the results on the renormalization scale @xmath89 and scheme ( to the order we are calculating these dependences can be represented by one parameter , say , the scale ) .
inclusion of higher order corrections decreases this dependence .
nevertheless , we are still left with intrinsic theoretical uncertainty of the perturbative results .
one can try to estimate this uncertainty ( see , for example , @xcite ) or one can try to find the `` optimal '' renormalization scale @xmath89 ( and scheme ) on the basis of some physical arguments . in the latter case , one can assess the size of the higher order corrections and of the expansion parameter .
these values can then serve as a sensible criteria for the convergence of the expansion .
the simplest and widely used choice for @xmath89 is @xmath90 , and the justification is mainly pragmatic .
however , physical arguments suggest that the more appropriate scale @xmath89 is lower .
namely , since each external momentum entering an exclusive reaction is partitioned among many propagators of the underlying hard - scattering amplitude , the physical scales that control these processes are inevitably much softer than the overall momentum transfer . there are number of suggestions in the literature . according to fastest apparent convergence ( fac ) procedure @xcite ,
the scale @xmath89 is determined by the requirement that the nlo coefficient in the perturbative expansion of the physical quantity in question vanishes , i.e. , one demands @xmath91 . on the other hand , following the principle of minimum sensitivity ( pms ) @xcite one mimics the independence of the all order expansion on the scale @xmath89 , and one chooses the renormalization scale @xmath89 at the stationary point of the truncated perturbative series : @xmath92 . in the brodsky - lepage - mackenzie ( blm )
procedure @xcite , all vacuum - polarization effects from the qcd @xmath77-function are resummed into the running coupling constant .
according to blm procedure , the renormalization scale best suited to a particular process in a given order can , in practice , be determined by setting the scale demanding that @xmath77-proportional terms should vanish : @xmath93 and @xmath94 as it is known , the relations between physical observables must be independent of renormalization scale and scheme conventions to any fixed order of perturbation theory . in ref .
@xcite was argued that applying the blm scale - fixing to perturbative predictions of two observables in , for example , @xmath73 scheme and then algebraically eliminating @xmath95 one can relate any perturbatively calculable observables without scale and scheme ambiguity , where the choice of blm scale ensures that the resulting `` commensurate scale relation '' ( csr ) is independent of the choice of the intermediate renormalization scheme . following this approach , in paper by brodsky _
et al._@xcite the several exclusive hadronic amplitudes were analyzed in @xmath96 scheme , in which the effective coupling @xmath97 is defined from the heavy - quark potential @xmath98 . the @xmath96 scheme is a natural , physically based scheme , which by definition automatically incorporates vacuum polarization effects .
the @xmath99 scale which then appears in the @xmath96 coupling reflects the mean virtuality of the exchanged gluons .
furthermore , since @xmath96 is an effective running coupling defined from the physical observable it must be finite at low momenta , and the appropriate parameterization of the low - energy region should in principle be included . the scale - fixed relation between the @xmath95 and @xmath96 couplings
is given by @xcite @xmath100 where @xmath101 is defined from the heavy - quark potential @xmath102 and @xmath103
as mentioned in sec . [ sec : intro ] , only the small number of exclusive processes have been analyzed in higher orders .
when higher order calculations are explicitly performed , usually the dimensional regularization together with the @xmath73 renormalization scheme is applied . the photon - to-@xmath104 transition form factor appearing in the amplitude
@xmath107 takes the form of an expansion @xmath108 + \cdots \ , , \end{aligned}\ ] ] where only the parts that can be found in the literature as explicitly calculated from the contributing feynman diagrams are written .
there are 2 lo diagrams contributing to the subprocess amplitude @xmath109 and displayed in fig .
[ f : tlo ] . furthermore , there are 12 one - loop diagrams contributing at nlo order @xcite .
the representative diagrams are given in fig .
[ f : tnlo ] . in the case of the photon - to-@xmath105 ( @xmath106 )
transition form factor the two - gluon states also contribute ( @xmath110 ) giving rise to 6 more diagrams at nlo @xcite . in @xcite
the @xmath36-proportional nnlo terms were determined from the 12 two - loop feynman diagrams obtained from the one - loop diagrams by adding the gluon vacuum polarization bubble .
the numerical predictions for @xmath111 are displayed in fig .
[ f : num1 ] . obviously , the results obtained using the cz da overshoot the experimental data .
the blm scale for the asymptotic da amounts to @xmath112 , while @xmath113 . in the @xmath96 scheme for the coupling constant scale
one obtains @xmath114 .
the spacelike pion electromagnetic form factor @xmath115 appearing in the amplitude @xmath116 , takes the form of an expansion @xmath117 there are 4 diagrams that contribute to the amplitude @xmath118 at lo ( see fig .
[ f : lo ] ) , and 62 one - loop diagrams at nlo @xcite ( see fig .
[ f : nlo ] ) .
.,width=264 ] amplitude.,width=302 ] numerical predictions for @xmath115 are displayed in fig .
[ f : num2 ] . .
the shaded area denotes the range of the total nlo prediction and offers the way to asses the theoretical uncertainty.,width=340 ] we comment the asymptotic da results . for @xmath119 nlo corrections are rather large : the ratio ( nlo correction / lo prediction ) is @xmath120 until @xmath121 gev@xmath122 is reached ! on the other hand ,
the blm scale @xmath123 is very small and hence @xmath37 is large
. the @xmath96 scheme offers the possible way out . in this scheme the scale amounts to @xmath124 ( @xmath125 and nlo corrections @xmath126 for @xmath127 gev@xmath122 ) @xcite .
finally , the amplitude of the process @xmath128 takes the form @xmath129 there are 20 diagrams that contribute at lo order @xmath130 . at nlo order 454
one - loop diagrams contribute to the nlo prediction .
the existing result from the literature @xcite covers only the special case of the equal momenta da , i.e. , @xmath131 .
the numerical result is thus not particularly realistic .
new ( general ) nlo calculation is in preparation and for that purpose convenient general analytical method for evaluation of one - loop feynman integrals has been developed @xcite .
recently , the conformal symmetry constraints were used to obtain the nnlo prediction for the photon - to - pion transition form factor @xcite .
the crucial ingredients of this approach lie in the fact that the massless pqcd is invariant under conformal transformations provided that the @xmath77 function vanishes , and that @xmath132 belongs to a class of two - photon processes calculable by means of the operator product expansion ( ope ) .
one can then make use of the predictive power of the conformal ope ( cope ) , the dis results for the nonsinglet coefficient function of the polarized structure function @xmath133 known to nnlo order @xcite and the explicitly calculated @xmath77-proportional nnlo terms @xcite .
let us first introduce the basic ingredients of the formalism .
for the general case of the pion transition form factor @xmath134 one expresses the results in terms of @xmath135 it is convenient to turn the convolution formula ( [ eq : fpigammaconv ] ) into the sum over conformal moments @xmath136 here @xmath137 is the @xmath138th conformal moment of the elementary hard - scattering amplitude , while @xmath139 where @xmath140 represents composite conformal operator , and @xmath141 . for the thorough review of the conformal transformations and their applications we refer to @xcite . on the quantum level conformal symmetry
is broken owing to the regularization and renormalization of uv divergences : coupling constant renormalization resulting in @xmath77-proportional terms and the renormalization of composite operators .
the latter represents the origin of non - diagonal nlo anomalous dimensions in @xmath73 scheme and can be removed by finite renormalization of the hard - scattering and distribution amplitude , i.e. , by the specific choice of the factorization scheme .
first , we pose the question weather we can find a factorization scheme in which conformal symmetry holds true up to @xmath77-proportional terms ?
renormalization group equation for the operators @xmath142 , equivalent to the da evolution equation ( [ eq : eveq ] ) , is given by @xmath143 where the anomalous dimension matrix , corresponding to the evolution kernel ( [ eq : kernel ] ) , is given by @xmath144 with @xmath145 .
since the conformal symmetry holds at lo , the anomalous dimensions are diagonal at lo .
similarly non - diagonal terms present in the @xmath73 scheme beyond lo , originate in breaking of conformal symmetry due to the renormalization of the composite operators .
in contrast , the conformal subtraction ( cs ) scheme defined by @xmath146 and @xmath147 preserves the conformal symmetry up to @xmath77-proportional terms .
second , we ask weather and how can we use the predictive power of conformal symmetry ? the process of interest belongs to quite a large class of two - photon processes calculable by means of ope @xcite .
dvcs , deeply inelastic lepton
hadron scattering ( dis ) and production of various hadronic final states by photon
photon fusion belong to this class of processes .
such processes can be described by a general scattering amplitude given by the time - ordered product of two - electromagnetic currents sandwiched between the hadronic states . for a specific process
, the generalized bjorken kinematics at the light - cone can be reduced to the corresponding kinematics , while the particular hadron content of the process reflects itself in the non - perturbative part of the amplitude .
hence , the generalized hard - scattering amplitude enables us to relate predictions of different two - photon processes on partonic level .
conformal ope ( cope ) for two - photon processes works under the assumption that conformal symmetry holds ( cs scheme and @xmath77=0 ) , and the wilson coefficients are then , up to normalization , fixed by the ones appearing in dis structure function @xmath133 ( calculated to nnlo order ) .
conformal symmetry breaking terms proportional to @xmath77 function alter cope result .
one can make use of @xmath77-proportional nnlo terms explicitly calculated in @xmath73 scheme @xcite .
we note here that there exists freedom in defining @xmath77-proportional terms in cs scheme and hence we speak of cs scheme , @xmath148 scheme , @xmath149 ( for detailed explanation see @xcite ) .
finally we list the numerical results for the special case @xmath150 and @xmath151 , i.e. , asymptotic da . for @xmath152 @xmath153 \quad \mbox{in}\quad \left\ { { \mbox{cs } \atop \overline{\mbox{cs } } } \right.\ ; \mbox{-scheme}\ , , \end{aligned}\ ] ] while for the blm prescription @xmath154 @xmath155 \,.\ ] ] here @xmath156 one notices that , similarly to the pion electromagnetic form factor results presented in the preceding section , for @xmath50 equal to the characteristic scale of the process the qcd corrections are large , while for the blm scale these corrections are smaller but the scale itself is also rather small leading to large expansion parameter @xmath37 .
the @xmath96 scheme could as in the case of the pion electromagnetic form factor offer the way out and physically better motivated description of the transition form factor .
we mention , that , as already noticed in @xcite and shown in @xcite , the significance of higher - conformal moments decreases with @xmath157 and that with decreasing @xmath157 the difference between various schemes also decreases .
hence , small @xmath157 region is suitable for a novel test of pqcd .
although the higher - order qcd corrections are important , only few exclusive processes have been explicitly calculated to nlo order .
the inclusion of higher - order corrections stabilizes the dependence on renormalization scale .
still , the usual choice @xmath158 $ ] leads to large corrections .
other choices of scales ( blm , @xmath96 scheme ) are preferable and more physical .
more effort in calculating higher - order corrections are needed and some tools applicable to the kinematic region of interest are underway .
furthermore , for some processes ( example : nnlo calculation of photon - to - pion transition form factor ) , one can make use of the predictive power of conformal symmetry to avoid cumbersome higher - order calculations .
i would like to take the opportunity to thank p. kroll , b. meli , d. mller and b. nii for fruitful collaborations in challenging higher order calculations , my collaborators a. p. bakulev , n. g. stefanis and w. schroers for interesting extensions of these investigations , as well as , h. w. huang , r. jakob , m. schrmann and w. schweiger for collaborating on stimulating and demanding lo calculations .
this work was supported by the ministry of science and technology of the republic of croatia under contract no . 0098002 .
g. p. lepage and s.
j. brodsky : phys .
b * 87 * , 359 ( 1979 ) ; g. p. lepage and s. j. brodsky : phys .
lett . * 43 * , 545 ( 1979 ) [ erratum - ibid . * 43 * , 1625 ( 1979 ) ] ; a. v. efremov and a. v. radyushkin : theor .
phys . * 42 * , 97 ( 1980 ) [ teor . mat .
fiz . * 42 * , 147 ( 1980 ) ] ; a. v. efremov and a. v. radyushkin : phys .
b * 94 * , 245 ( 1980 ) ; a. duncan and a. h. mueller : phys .
b * 90 * , 159 ( 1980 ) ; a. duncan and a. h. mueller : phys .
d * 21 * , 1636 ( 1980 ) g. p. lepage and s. j. brodsky : phys .
d * 22 * , 2157 ( 1980 ) r. d. field , r. gupta , s. otto and l. chang : nucl
. phys . *
b186 * , 429 ( 1981 ) f. m. dittes and a. v. radyushkin : sov .
j. nucl .
phys . * 34 * , 293 ( 1981 ) [ yad .
fiz . * 34 * , 529 ( 1981 ) ] m. h. sarmadi : ph .
d. thesis , university of pittsburgh , 1982 , umi 83 - 18195 r. s. khalmuradov and a. v. radyushkin : sov .
j. nucl .
phys . * 42 * , 289 ( 1985 ) [ yad .
fiz . * 42 * , 458 ( 1985 ) ] e. p. kadantseva , s. v. mikhailov and a. v. radyushkin : yad .
* 44 * , 507 ( 1986 ) [ sov .
j. nucl .
phys . * 44 * , 326 ( 1986 ) ] e. braaten and s. tse : phys .
d * 35 * , 2255 ( 1987 ) b. meli , b. nii and k. passek : phys .
d * 60 * , 074004 ( 1999 ) [ hep - ph/9802204 ] f. del aguila and m. k. chase : nucl .
* b193 * , 517 ( 1981 ) e. braaten : phys .
d * 28 * , 524 ( 1983 ) p. kroll and k. passek - kumeriki : phys . rev .
d * 67 * ( 2003 ) 054017 [ hep - ph/0210045 ] ; b. nii : phys .
d * 35 * , 80 ( 1987 ) i. v. musatov and a. v. radyushkin : phys .
d * 56 * , 2713 ( 1997 ) [ hep - ph/9702443 ] n. g. stefanis , w. schroers and h. c. kim : eur .
j. c * 18 * , 137 ( 2000 ) [ hep - ph/0005218 ] d. mller , d. robaschik , b. geyer , f .- m .
dittes , and j. hoeji : fortschr .
42 , 101 ( 1994 ) .
x. d. ji : phys .
d * 55 * , 7114 ( 1997 ) [ hep - ph/9609381 ] ; a. v. radyushkin : phys .
d * 56 * , 5524 ( 1997 ) [ hep - ph/9704207 ] .
l. mankiewicz , g. piller , e. stein , m. vanttinen and t. weigl : phys .
b * 425 * , 186 ( 1998 ) [ hep - ph/9712251 ] x. d. ji and j. osborne : phys .
d * 58 * , 094018 ( 1998 ) [ hep - ph/9801260 ] a. v. belitsky and d. mller : phys .
b * 417 * , 129 ( 1998 ) [ hep - ph/9709379 ] a. v. radyushkin : phys . lett .
b * 385 * , 333 ( 1996 ) [ hep - ph/9605431 ] ; j. c. collins , l. frankfurt and m. strikman : phys . rev .
d * 56 * , 2982 ( 1997 ) [ hep - ph/9611433 ] a. v. belitsky and d. mller : phys .
b * 513 * , 349 ( 2001 ) [ hep - ph/0105046 ] a.v .
belitsky and a. schfer : nucl . phys . *
b527 * , 235 ( 1998 ) [ hep - ph/9801252 ] b. meli , b. nii and k. passek : phys . rev .
d * 65 * , 053020 ( 2002 ) [ hep - ph/0107295 ] b. meli , d. mller and k. passek - kumeriki : phys .
d * 68 * , 014013 ( 2003 ) [ hep - ph/0212346 ] h. w. huang and p. kroll , eur .
j. * c17 * , 423 ( 2000 ) , [ hep - ph/0005318 ] .
h. w. huang , p. kroll and t. morii : eur .
j. c * 23 * , 301 ( 2002 ) [ erratum - ibid .
c * 31 * , 279 ( 2003 ) ] [ hep - ph/0110208 ] .
h. w. huang and t. morii : phys .
d * 68 * , 014016 ( 2003 ) [ hep - ph/0305132 ] .
d. mller : phys .
d * 49 * , 2525 ( 1994 ) ; phys .
d * 51 * , 3855 ( 1995 ) [ hep - ph/9411338 ] v. l. chernyak and a. r. zhitnitsky : phys .
rept . * 112 * , 173 ( 1984 ) a. p. bakulev , s. v. mikhailov and n. g. stefanis : phys .
b508 , 279 ( 2001 ) [ hep - ph/0103119 ] b. meli , b. nii and k. passek : fizikab * 8 * , 327 ( 1999 ) ; d. v. shirkov and i. l. solovtsov : phys . rev
. lett . * 79 * , 1209 ( 1997 ) [ hep - ph/9704333 ] g. grunberg : phys .
b * 95 * , 70 ( 1980 ) ; phys .
d * 29 * , 2315 ( 1984 ) p. m. stevenson : phys .
b * 100 * , 61 ( 1981 ) ; phys .
d * 23 * , 2916 ( 1981 ) ; nucl .
b * 203 * , 472 ( 1982 ) ; nucl .
b * 231 * , 65 ( 1984 ) s. j. brodsky , g. p. lepage and p. b. mackenzie : phys .
d * 28 * , 228 ( 1983 ) s. j. brodsky and h. j. lu , phys .
d * 51 * , 3652 ( 1995 ) [ hep - ph/9405218 ] s. j. brodsky , c. ji , a. pang and d. g. robertson , phys .
d * 57 * , 245 ( 1998 ) [ hep - ph/9705221 ] a. p. bakulev , a. v. radyushkin and n. g. stefanis : phys .
d * 62 * , 113001 ( 2000 ) [ hep - ph/0005085 ] b. meli , b. nii and k. passek : in proc . of joint int / jlab workshop ,
newport news 1999 , 279 - 286 , hep - ph/9908510 g. duplancic and b. nizic : eur .
j. c * 20 * , 357 ( 2001 ) [ hep - ph/0006249 ] ; eur .
j. c * 24 * , 385 ( 2002 ) [ hep - ph/0201306 ] ; hep - ph/0303184 .
e. b. zijlstra and w. l. van neerven : nucl . phys . * b417*,61 ( 1994 ) [ erratum - ibid .
b426:245 , ( 1994 ) ] v. m. braun , g. p. korchemsky and d. muller : prog .
phys . * 51 * , 311 ( 2003 ) [ hep - ph/0306057 ] m. diehl , p. kroll , and c. vogt : eur .
j. c22,439 ( 2001 ) . | the short review of the higher order corrections to the hard exclusive processes is given .
different approaches are discussed and the importance of higher - order calculations is stressed . |
recent years have seen improvements in the calculational techniques for one - loop amplitudes @xcite . besides calculations of new amplitudes for phenomenological @xcite purposes , a number of infinite sequences of amplitudes have been obtained @xcite . at two- or higher - loops
there has not been analogous progress as yet , although there is a need for such calculations in the analysis of experiments .
for example , experiments at lep are currently sensitive to next - to - next - to leading order corrections to @xmath1 jets @xcite ; this requires two - loop amplitudes as an input .
before this and other two - loop calculations can be undertaken , a number of technical issues must be addressed .
it would be useful to have an efficient technique for obtaining compact analytic expressions for two - loop amplitudes .
some first steps in developing such a technique have recently been reported using ideas motivated by string theory @xcite .
in particular , reuter , schmidt and schubert @xcite have performed an elegant evaluation of two - loop quantities using a world - line approach .
kim and nair @xcite have also set up a multi - loop formulation of the recursive approach @xcite .
another technical hurdle is that many of the required integrals are uncalculated @xcite .
the construction of a numerical program to extract jet cross - sections would also require significant work ; in particular , one would need to extend the known techniques for handling the infrared divergent corners of phase space @xcite . in this letter
we explore the use of cutting rules @xcite to obtain compact analytical expressions for two - loop amplitudes .
in particular , we compute the @xmath2 supersymmetric four - gluon amplitudes in terms of a basic set of scalar integral functions . amplitudes in @xmath3 super - yang - mills are particularly simple and are therefore a good starting point . at tree - level @xcite and one - loop , infinite sequences of gauge theory amplitudes have been explicitly constructed .
can one find such sequences for higher - loop amplitudes ?
in this letter we also present a pattern for the leading - color four - gluon amplitudes to all orders of perturbation theory consistent with two - particle cuts .
we do not , however , have a proof that the pattern is the complete answer and it is possible that a more complete analysis including three- or higher - particle cuts might reveal additional terms . in the past , cutting rules @xcite have been widely used in field theory and provide powerful constraints on the form of amplitudes . in order to construct complete amplitudes one must have control of the ambiguities and subtractions . at one - loop
it has been shown that for amplitudes satisfying a power counting criterion , such as supersymmetric ones , one can obtain the complete answer from four - dimensional cuts and knowledge of the set of integral functions that can appear in the result @xcite . for amplitudes
not satisfying the power counting criterion one may unambiguously obtain the amplitudes by computing the cuts to all orders in the dimensional regularization parameter @xcite , since all terms develop cuts . at two loops
much less is known about the integral functions . nevertheless , for the supersymmetric two - loop amplitudes considered in this letter one can perform a complete reconstruction .
at tree and one - loop levels it has proven useful to separate the color factors from the kinematics @xcite .
( the reader may consult various review articles @xcite for notation and further details . )
the same type of color decomposition is also useful at the two - loop level .
the color decomposed expression for a four - point amplitude with all particles in the adjoint representation is @xmath4 \
, \bigl ( a^{\lc}_{4;1,1}(\si(1 ) , \si(2 ) , \si(3 ) , \si(4 ) ) \cr & \hskip 4 cm + { 1\over n_c^2 } a^{\sc}_{4;1,1}(\si(1 ) , \si(2 ) , \si(3 ) , \si(4 ) ) \bigr ) \cr & \hskip 1 cm + g^6 \hskip -.2 cm \sum_{\si \in { s_4/z_2 ^ 3 } } n_c \tr[t^{a_{\si(1)}}t^{a_{\si(2 ) } } ] \tr[t^{a_{\si(3)}}t^{a_{\si(4 ) } } ] \ , a_{4;1,3}(\si(1 ) , \si(2 ) ; \si(3 ) , \si(4))\ , , \cr } \equn\label{twoloopcolor}\ ] ] where @xmath5 and @xmath6 are ` partial amplitudes ' .
our notation for @xmath7 is that @xmath8 is the number of external legs , and @xmath9 and @xmath10 label the position at which the color traces are split ; for @xmath11 there can be up to three color traces so for a generally consistent notation two indices are required .
we have explicitly broken up the single color trace partial amplitude into leading color @xmath12 and subleading color @xmath13 pieces .
( by subleading color we mean expressions suppressed in powers of @xmath14 . )
for each external leg we have abbreviated the dependence on the outgoing external momenta , @xmath15 , and polarizations , @xmath16 , by the label @xmath17 .
the notation ` @xmath18 ' denotes the set of all permutations of four objects @xmath19 , omitting the cyclic transformations .
the notation ` @xmath20 ' refers again to the set of permutations of four objects omitting those permutations which exchange labels within a single trace or exchange the two traces .
that is , @xmath21 .
this decomposition has a straightforward generalization to an arbitrary number of external legs .
a similar decomposition exists when some of the particles are in the fundamental representation .
the partial amplitudes are independently gauge invariant and may therefore be calculated separately .
the high degree of supersymmetry present in @xmath2 amplitudes considerably simplifies their analytic structure .
supersymmetric amplitudes with external legs that have either all helicities identical or all but one identical vanish by a supersymmetry identity @xcite .
the non - vanishing maximally helicity violating ( mhv ) amplitudes , where two of the external helicities are of one type and the rest of the other type , are especially simple . at four and five
points all the non - vanishing amplitudes are mhv . for these reasons
four - point @xmath2 supersymmetric amplitudes are a natural starting point for evaluating higher - loop amplitudes . at one - loop @xmath0 amplitudes may be considered as pieces of qcd amplitudes ; similarly at two and higher loops we may expect a study of @xmath0 amplitudes to be of some use for obtaining qcd amplitudes .
the tree - level four - gluon partial amplitudes are @xmath22 these are identical to the ones of standard qcd , since intermediate scalars or fermions do not propagate at tree level .
the @xmath23 superscripts on the leg labels represent the helicities of the external gluons .
we have chosen to represent the amplitudes in terms of the spinor helicity formalism @xcite , where gluon polarizations are replaced by spinor inner - products .
( the reader may consult review articles for details @xcite . )
we use the compact notation @xmath24 \ , , $ ] where @xmath25 are massless weyl spinors labeled by the sign of their helicities and with normalization @xmath26 . at one - loop
the leading color @xmath2 partial amplitudes are rather simple and given by @xmath27 where @xmath28 is the one - loop scalar integral .
the mandelstam variables are defined as @xmath29 and @xmath30 .
the massless @xmath31 super - multiplet consists of one gluon , four weyl fermions and six real scalars , whose contributions are summed over to obtain the result ( [ oneloopnfour ] ) . the integral ( [ oneloopbox ] )
has a simple representation in terms of logarithms .
( see for example ref .
@xcite . ) the remaining subleading color partial amplitudes may easily be computed in terms of sums of permutations of the expression ( [ oneloopnfour ] ) .
the one - loop @xmath32 four - gluon amplitude , in @xmath33 dimensions , was first obtained by green , schwarz and brink from the low energy limit of superstring theory @xcite .
amplitudes , of course , depend on the particular form of dimensional regularization .
the amplitudes ( [ oneloopnfour ] ) are quoted in the dimensional reduction scheme @xcite or equivalently the four - dimensional helicity scheme @xcite , which preserve supersymmetry .
the above expression turns out to be valid to all orders of of the dimensional regularization parameter @xmath34 @xcite .
furthermore , when @xmath35 is expressed in terms of formal polarization vectors and spinors instead of helicities , it is valid in any dimension .
this will be useful later when we construct two - loop amplitudes . when using cuts to obtain the @xmath0 gluon amplitudes we also need tree and one - loop amplitudes with either two external scalars or fermions , since these particles can also cross the cuts .
the maximally helicity violating amplitudes with two external fermions or scalars may be obtained directly from the gluon amplitudes using the supersymmetry identity @xcite @xmath36 where leg @xmath9 is the only negative helicity gluon on the left - hand - side , @xmath37 is either a scalar or fermion and @xmath38 is the absolute value of its helicity ( @xmath39 or @xmath40 ) .
the legs without subscripts are taken to be gluons . since this identity follows from general supersymmetry relations , it is valid to all orders in the perturbative expansion .
the mhv @xmath0 gluon amplitudes also satisfy the identity @xcite @xmath41 where @xmath17 and @xmath9 are the only negative helicity legs on the left - hand - side and @xmath42 and @xmath43 are the only negative helicities on the right - hand - side .
furthermore , it implies that different cuts are related by simple relabelings , up to an overall prefactor of @xmath44 , where @xmath17 and @xmath9 label the negative helicities @xcite . in four dimensions the @xmath8-loop
@xmath0 super - yang - mills amplitudes are ultraviolet finite @xcite .
an important ingredient in mandelstam s demonstration of finiteness is that for each external leg a power of external momentum may be extracted from each superspace feynman diagram . more generally , in @xmath33 dimensions with @xmath45 supersymmetries ( where @xmath45 is counted in four dimensions ) , power counting constraints on the form of the @xmath46 loop effective action lead to the ultraviolet finiteness condition @xcite , @xmath47 in particular , at two - loops , @xmath0 amplitudes are ultraviolet finite for @xmath48 .
-.7 cm 1.5 cm -.75 cm
first we briefly review the cut construction method @xcite , extending it to the case of two loops . following the discussion in ref .
@xcite , a convenient way to obtain amplitudes is by considering cuts of unrestricted loop momentum integrals . in this way
, one may simultaneously construct the imaginary and associated real parts of the cuts .
consider for example , the cuts of the two - loop amplitude @xmath49 . at two loops
one must consider both two- and three - particle cuts . in each channel
there may be multiple contributing cuts .
for example , in the @xmath50 channel there are two two - particle cuts , as depicted in .
the first of these has the explicit representation @xmath51 where @xmath52 and @xmath53 are the momenta of the cut legs and the sum runs over all particle types ( including helicity ) @xmath54 and @xmath55 which may propagate across the two cut lines .
we may use the on - shell conditions @xmath56 and @xmath57 in the integrand ( but not on the cut propagators ) since this equation is valid only for those terms which have explicit @xmath52 and @xmath53 propagators .
it is convenient to apply the color decomposition ( [ twoloopcolor ] ) before computing the various cut contributions .
the three - particle @xmath50-channel cut is depicted in .
the @xmath58-channel cuts , of course , have a similar structure to the @xmath50-channel cuts . by combining all cuts into a single function ,
one obtains the full amplitude . when combining the cuts ,
care must be exercised not to over - count a particular term . -.7 cm -.75 cm as discussed in refs .
@xcite , by computing cuts to all orders in the dimensional regularization parameter , one may perform a complete reconstruction of a massless loop amplitude .
this follows from dimensional analysis since every term in an amplitude must have a prefactor of powers of @xmath59 , which necessarily have cuts . for reasons of technical simplicity
, it is convenient to use helicity amplitudes which implicitly replace the @xmath60-dimensional momenta of the cut lines by four - dimensional momenta .
however , to do so we must ensure that errors , especially through @xmath61 , are not introduced . in section [ validitysection ] , we shall argue that at least for the @xmath0 four - point amplitudes there are no such errors .
we also find it convenient to perform the cut construction in components instead of using superfields .
the potential advantage of a superfield formalism would be that one would simultaneously include contributions from all particles in a supersymmetry multiplet .
however , for maximally helicity violating amplitudes , the supersymmetry identities are sufficiently powerful that once the contribution from one component is known the others immediately follow .
a component formulation is also more natural for extensions to qcd .
we now proceed with the explicit construction of the @xmath0 four gluon partial amplitudes .
first we consider the leading color amplitude @xmath62 . in the cut construction
we use the amplitudes ( [ treeampls ] ) and ( [ oneloopnfour ] ) as inputs . for the @xmath50-channel cuts depicted in , with the helicity configuration @xmath63
, only the internal gluon loop contributes in the configuration where the cut gluons on each side of the cut have helicity opposite to that of the external legs . for fermion , scalars or gluons with a different helicity configuration , propagating across the cut , the amplitudes on both sides of the cut vanish by a supersymmetry identity .
this then gives for , @xmath64\right|_{\ell_1 ^ 2 = \ell_2 ^ 2 = 0 } \hskip -.6 cm \
, , \cr } \equn\label{scutsusyb}\ ] ] where @xmath65 and @xmath66 and we have used the amplitudes in eqs .
( [ treeampls ] ) and ( [ oneloopnfour ] ) .
we have also used , for example , @xmath67 to rationalize the denominators .
the numerator of the integrand is @xmath68 \cr & = -st \ , ( p - k_1)^2 ( p + k_4)^2 \ , , \cr } \equn\ ] ] where @xmath69 = { 1\over 2 } \tr[(1+\gamma_5 ) \cdots]$ ] .
the @xmath70 term in the trace does not contribute because a four - point amplitude has only three independent momenta to contract into the totally anti - symmetric levi - civita tensor .
the above sewing algebra is particularly simple because it is identical to that of the one - loop case , discussed in ref .
@xcite .
thus , after canceling numerator factors against propagators and identifying the remaining integral as a two - loop scalar integral we obtain @xmath71 where the two - loop planar scalar double - box integral is @xmath72 the evaluation of the two - particle @xmath50-channel cut in b is similar and gives the same result .
the evaluation of the @xmath58-channel two - particle cuts are again similar , but a bit more involved since all particles in the super - multiplet contribute . however , after summing over the contribution of all particles , with the help of the supersymmetry identities and spinor identities , the integral appearing in the @xmath58-channel cut coincides with the one appearing in the @xmath50-channel cut in , but with @xmath50 and @xmath58 interchanged . -.7 cm
-.75 cm the three - particle cuts reproduce the results obtained from the two particle cuts without adding any new integral functions . the result of evaluating the three - particle @xmath50-channel cut in is depicted in .
the scalar box integral @xmath73 appears twice because it has two distinct three - particle cuts .
this calculation is more complicated than the two - particle cut calculation since one must sum over a total of sixteen intermediate particle and helicity configurations .
nevertheless the @xmath0 supersymmetry ensures that the various terms combine neatly .
combining all cuts into a single function that has the correct cuts in all channels yields the final result for the leading color @xmath0 four - gluon partial amplitude , @xmath74 this is depicted in .
we comment that although substantial progress has been made in the calculation of such integrals @xcite , the scalar integral appearing in this amplitude has not yet been evaluated in terms of known functions .
this amplitude exhibits the cyclic symmetry for the leading color partial amplitudes ( up to an overall @xmath75 ) implied by the supersymmetry identity ( [ cyclicidentity ] ) , as expected .
we may also use the supersymmetry identity ( [ cyclicidentity ] ) to immediately obtain the leading color partial amplitude with non - adjacent negative helicities . -.7 cm
-.75 cm the computation of the subleading color pieces is similar to the leading color computation , so we only quote the results .
there are two types of subleading color contributions .
firstly , there are contributions arising from planar diagrams .
these can be determined by evaluating the color algebra for the feynman diagrams and are given by nothing more than sums of permutations of arguments of the same functions that appear at leading color .
secondly , there is a new contribution characterized by diagrams with non - planar topology .
the cut construction of the non - planar contributions is quite similar to the planar one , with the main difference being the appearance of non - planar scalar integrals in the final results .
the results for the subleading in @xmath14 partial amplitudes are @xmath76 where @xmath77 and @xmath78 .
the planar scalar integral is defined in , while the non - planar one , depicted in , is @xmath79 -.7 cm -.75
cm a simple consistency check on the amplitudes ( [ leadingcolorresult ] ) and ( [ subleadingcolorresult ] ) is that they satisfy the expected finiteness condition ( [ finiteness ] ) .
we now discuss potential errors arising from our use of four - dimensional helicity amplitudes in the cuts .
such errors would not occur if the sewing were done using formal spinors and polarization vectors . when using the four - dimensional spinor helicity formalism potential errors will be of the form , @xmath80 where we have explicitly separated the loop momenta into four- and @xmath81-dimensional parts .
( a discussion of the @xmath81-dimensional parts of loop momenta can be found , for example , in refs .
terms in the integrand proportional to @xmath81-dimensional momenta @xmath82 or @xmath83 , may in principle have been dropped in the above cut construction .
note that terms containing such factors are necessarily suppressed by a power of @xmath34 .
however , this suppression may be cancelled if the integrals contain poles in @xmath34 .
the @xmath84 amplitudes are , however , special .
we may use the fact that the algebra associated with the two - particle cuts is identical at one and two loops . as discussed in section [ twoloopcutsection ] ,
the one - loop amplitudes ( [ oneloopnfour ] ) are valid in any dimension when expressed in terms of formal polarizations and spinors .
since the one - loop amplitudes are proportional to @xmath35 , the two - loop sewing procedure for the two - particle cuts using formal spinors and polarizations reproduces the one - loop result , except for the prefactors and the two - loop scalar integral function .
this implies that all feynman diagrams with two - particle cuts have been accounted for exactly .
this , of course , does not rule out errors of the type in from diagrams with no two - particle cuts .
two examples of such diagrams are given in .
here we may appeal to the superspace power counting rule that requires the extraction of a power of external momentum for each external leg of a superspace feynman diagram @xcite . after extracting four powers of external momenta ,
four - point diagrams with no two - particle cuts can not contain any powers of @xmath85 in the numerators of their integrands since there are no remaining powers of loop momenta in the non - vanishing diagrams .
this means that there are no potential errors of the type in arising from three - particle cuts . a more complete discussion of the @xmath85 terms , especially for theories with less supersymmetries ,
will be presented in the future . -.7
cm -.75 cm
following the same cut construction procedure used for the two - loop amplitudes , we have found a pattern for the @xmath8-loop @xmath84 four - gluon leading color partial amplitudes .
the three - loop leading color partial amplitude is given in .
note that there are one - loop pentagon sub - diagrams .
this complicates the analysis of the three - particle cuts since one - loop pentagons can be reduced to sums over box integrals @xcite . in some cuts
it is the box integrals that appear and in some it is the pentagon ; this must be disentangled in order to identify the form appearing in . -.7 cm -.75 cm observing the results for the leading color one- , two- and three - loop @xmath0 amplitudes one can recognize a pattern which can be used to construct the @xmath86-loop amplitude from the @xmath8-loop result .
the pattern is that one takes each @xmath8-loop graph in the @xmath8-loop amplitude and generates all the possible @xmath86-loop graphs by inserting a new leg between each possible pair of internal legs .
diagrams where triangle or bubble subgraphs are created should not be included .
the new loop momentum including an additional factor of @xmath87 in the numerator is integrated over , where @xmath52 and @xmath53 are the momenta flowing through each of the legs to which the new line is joined .
( this is depicted in ) .
momentum conservation ensures that it does not matter on which side of the new line the momentum pair @xmath52 and @xmath53 are taken .
note that no four - point vertices are created by this procedure .
each distinct @xmath86-loop graph should be counted once , even though they can be generated in multiple ways .
the @xmath86-loop amplitude is then the sum of all distinct @xmath86-loop graphs . -.7 cm
-.75 cm we have verified that this pattern is consistent with two - particle cuts to all loop orders and with four - dimensional three particle cuts , up to five loops .
the two - particle cuts are not difficult to check because the same algebra appears at any loop order .
it is , however , possible that there are further terms in the amplitudes . in order to show that nothing has been missed
, one would need to demonstrate that higher order in @xmath34 terms in lower - loop amplitudes do not cause errors through @xmath61 .
one would also need to verify the consistency of higher - particle cuts .
presumably , a similar pattern can be found for subleading in color contributions .
in this letter we have made an initial step in applying the one - loop cut construction technique @xcite reviewed in ref .
@xcite to two - loop amplitudes . as an explicit example we computed the two - loop @xmath0 supersymmetric four - gluon amplitudes in terms of a basic set of scalar integrals .
( the amplitudes with two fermions or scalars and two gluons trivially follow from a supersymmetry identity @xcite . )
we also presented a pattern for the leading - color @xmath84 four - gluon amplitudes to all loop orders consistent with two - particle cuts .
for the two - loop @xmath84 amplitudes presented in this letter , we showed that the cut construction yields the complete amplitude without any rational function or subtraction ambiguities .
this argument was based on the fact that the one - loop four - point amplitudes are known in any dimension and that the amplitudes satisfy a power counting criterion related to the finiteness of the theory .
a next step , along the lines of this letter , would be to obtain amplitudes in @xmath88 supersymmetric theories as a precursor to computing qcd amplitudes .
however , before two - loop amplitudes can be used in phenomenological studies , one would need a numerically stable method for evaluating the basic integral functions that occur @xcite .
we thank l. dixon and d.a .
kosower for comments which greatly improved this letter .
we also thank d. dunbar , a. grant and a.g .
morgan for a number of helpful comments .
this work was supported by the doe under contract de - fg03 - 91er40662 and by the alfred p. sloan foundation under grant br-3222 . z. bern , l. dixon and d.a .
kosower , phys .
lett.70:2677 ( 1993 ) , preprint hep - ph/9302280 ; + z. kunszt , a. signer and z. trcsnyi , phys .
b336:529 ( 1994 ) , hep - ph/9405386 ; + z. bern , l. dixon and d.a .
kosower , nucl .
phys.b437:259 ( 1995 ) , hep - ph/9409393 ; + z. trocsanyi , phys .
lett . 77:2182 ( 1996 ) , hep - ph/9610499 ; + w.b .
kilgore and w.t .
giele , preprint hep - ph/9610433 + a. signer and l. dixon , phys .
lett . 78:811 ( 1997 ) , hep - ph/9609460 ; + e.w.n . glover and d.j .
miller , preprint hep - ph/9609474 ; + z. bern , lance dixon , d.a .
kosower , and s. weinzierl , preprint hep - ph/9610370 , to appear in nucl .
b. z. bern , l. dixon and d.a .
kosower , _ proceedings of strings 1993 _ , eds . m.b .
halpern , a. sevrin and g. rivlis ( world scientific , singapore , 1994 ) , hep - th/9311026 ; + z. bern , g. chalmers , l. dixon and d.a .
kosower , phys .
lett . 72:2134 ( 1994 ) , hep - ph/9312333 .
k. roland , phys .
289b:148 ( 1992 ) ; + g. cristofano , r. marotta and k. roland , nucl .
b392:345 ( 1993 ) ; + a. pasquinucci and k. roland , nucl .
b457:27 ( 1995 ) , hep - th/9508135 ; + p. di vecchia , a. lerda , l. magnea and r. marotta , phys .
b351:445 ( 1995 ) , hep - th/9502156 ; + p. di vecchia , a. lerda , l. magnea , r. marotta , and r. russo , nucl . phys.b469:235 ( 1996 ) ; + hep - th/9601143 ; + m.g .
schmidt and c. schubert , phys .
b331:69 ( 1994 ) , hep - th/9403158 ; nucl .
suppl . 39bc:306 ( 1995 ) , hep - ph/9408394 .
usyukina and a.i .
davydychev , phys .
lett . 298b:363 ( 1993 ) ; 305b:136 ( 1993 ) ; 332b:159 ( 1994 ) , hep - ph/9402223 ; 348b:503 ( 1995),hep - ph/9412356 ; + d. kreimer , phys .
b347:107 ( 1995 ) , hep - ph/9407234 ; mod .
a9:1105 ( 1994 ) , hep - ph/9407234 ; + a.i .
davydychev and j.b .
tausk , preprint hep - ph/9504431 . s.d .
ellis , z. kunszt and d.e .
soper , phys .
d40:2188 ( 1989 ) ; phys . rev .
lett . 64:2121 ( 1990 ) ; phys .
69:1496 ( 1992 ) ; + w.t .
giele and e.w.n .
glover , phys .
d46:1980 ( 1992 ) ; + w.t .
giele , e.w.n . glover and d.a .
kosower , nucl .
b403:633 ( 1993 ) , hep - ph/9302225 ; + s. frixione , z. kunszt and a. signer , nucl .
b467:399 ( 1996 ) , hep - ph/9512328 ; + s. catani and m.h .
seymour , preprint hep - ph/9605323 .
j.e . paton and h .-
chan , nucl .
b10:519 ( 1969 ) ; + f.a .
berends and w.t .
giele , nucl .
b294:700 ( 1987 ) ; + d.a .
kosower , b .- h .
lee and v.p .
nair , phys .
201b:85 ( 1988 ) ; + m. mangano , s. parke and z. xu , nucl .
b298:653 ( 1988 ) ; + z. bern and d.a .
kosower , nucl . phys . b362:389 ( 1991 ) . m. mangano and s.j .
parke , phys . rep .
200:301 ( 1991 ) ; + l. dixon , preprint hep - ph/9601359 , in _ proceedings of theoretical advanced study institute in elementary particle physics ( tasi 95 ) _ , ed .
soper .
berends , r. kleiss , p. de causmaecker , r. gastmans and t.t .
wu , phys .
lett . 103b:124 ( 1981 ) ; + p. de causmaecker , r. gastmans , w. troost and t.t .
wu , nucl .
b206:53 ( 1982 ) ; + r. kleiss and w.j .
stirling , nucl .
phys . b262:235 ( 1985 ) ; + j.f .
gunion and z. kunszt , phys .
lett . 161b:333 ( 1985 ) ; + z. xu , d .- h .
zhang and l. chang , nucl .
b291:392 ( 1987 ) .
w. van neerven and j.a.m .
vermaseren , phys .
lett . 137b:241 ( 1984 ) ; + z. bern , l. dixon and d.a .
kosower , phys .
302b:299 ( 1993 ) , hep - ph/9212308 ; erratum _ ibid . _ 318b:649 ( 1993 )
; + z. bern , l. dixon and d.a .
kosower , nucl .
b412:751 ( 1994 ) , hep - ph/9306240 . | using cutting techniques we obtain the two - loop @xmath0 super - yang - mills helicity amplitudes for four - gluon scattering in terms of scalar integral functions .
the @xmath0 amplitudes are considerably simpler than corresponding qcd amplitudes and therefore provide a testing ground for exploring two - loop amplitudes .
the amplitudes are constructed directly in terms of gauge invariant quantities and therefore remain relatively compact throughout the calculation .
we also present a conjecture for the leading color four - gluon amplitudes to all orders in the perturbative expansion .
# 1_#1 _ = 0.1 cm = -2.5 cm = -2.0 cm = 23.2truecm = 16.7truecm 25 pt # 1fig .
[ # 1 ] # 1eq .
( [ # 1 ] ) plus 1000pt minus 1000pt # 1 # 1= to # 1 [ cols="^",options="header " , ] # 1eq . (
[ # 1 ] ) # 1eq . ( [ # 1 ] ) # 1eqs .
# 1 # 1eqs .
# 1 # 1fig .
[ # 1 ] # 1figs .
# 1 # 1section [ # 1 ] # 1section [ # 1 ] # 1appendix [ # 1 ] # 1appendix [ # 1 ] # 1 = = = by - by - by by 2 -to # 1.#2#1#2 # 1.#2 # 1.#2(#1#2 ) # 1.#2.#3 |#2 | # 1.#2.#3 ^-|#2 |^+ # 1.#2.#3 ^+|#2 |^+ # 1.#2.#3 ^-|#2 |^- # 1.#2.#3 ^+|#2 |^- # 1.#2.#3 ^-|#2 |^+
@=11 p # 1#2#3#4(#1#2#3#4 ) hep - ph/9702424 ucla/97/tep/4 + february , 1997 + 2.cm * two - loop four - gluon amplitudes in n=4 super - yang - mills * 2.cm z. bern , j.s .
rozowsky and b. yan 0.5 cm _ department of physics + university of california at los angeles + los angeles , ca 90095 - 1547 _ |
it is known that inclusive heavy quark production is a calculable process in perturbative qcd , since the heavy quark mass acts as a cut - off for the final state collinear singularities .
thus , the process e^+
e^- z/ q+ x , where @xmath2 is the heavy quark and @xmath3 is anything else , is calculable . its cross section can be expressed as a power expansion in the strong coupling constant [ eq : sigma ] ( x , e , m ) = _ n=0^ a^(n)(x , e , m , ) ^n ( ) , where @xmath4 is the centre - of - mass energy , @xmath5 is the mass of the heavy quark , @xmath6 is the renormalization scale , and ( ) = . as usual
we define x = , where @xmath7 and @xmath8 are the four - momenta of the intermediate virtual boson and of the final heavy quark @xmath2 . the cross section ( [ eq : sigma ] ) , normalized to the total cross section , is sometimes referred to as the heavy quark fragmentation function in @xmath1 annihilation . when @xmath9 is not too large , the truncation of eq .
( [ eq : sigma ] ) at some fixed order in the coupling can be used to compute the cross section . on the other hand , if @xmath10 , the @xmath11 order coefficient of the expansion will in general contain up to @xmath12 powers of @xmath13 , thereby spoiling the convergence of the expansion .
these large logarithms can be resummed , according to the method described in ref .
first of all , since @xmath14 is large , one is entitled to neglect the terms that are suppressed by powers of @xmath15 .
one then observes that , in this limit , the inclusive heavy quark cross section must satisfy the factorization theorem formula [ eq : factorization ] ( x , e , m ) = _ i _ x^1 @xmath16 _ i@xmath17 , where @xmath18 are the @xmath19-subtracted partonic cross sections for producing the parton @xmath20 , and @xmath21 are the @xmath19 fragmentation functions for the parton @xmath20 into the heavy quark @xmath2 .
in order for eq .
( [ eq : factorization ] ) to hold , it is essential that one uses a renormalization scheme where the heavy flavour is treated as a light one , like the pure @xmath19 scheme .
thus @xmath18 has a perturbative expansion in terms of @xmath22 with @xmath23 flavours , where @xmath23 includes the heavy one .
the scale @xmath6 is the factorization and renormalization scale .
it should be chosen of the order of @xmath4 , in order to avoid the appearance of large logarithms of @xmath24 in the partonic cross section .
the @xmath19 fragmentation functions @xmath25 obey the altarelli - parisi evolution equations : [ eq : ap ] ( x,,m ) = _
j^1_x p_ij@xmath26 _ j(z,,m ) . the altarelli - parisi splitting functions @xmath27 have the perturbative expansion [ eq : apfunctions ] p_ij(x , ( ) ) = ( ) p^(0)_ij(x ) + ^2 ( ) p^(1)_ij(x ) + o(^3 ) , where @xmath28 are given in ref .
@xcite and @xmath29 have been computed in refs .
the only missing ingredients for the calculation of the inclusive cross section are the initial conditions for the @xmath19 fragmentation functions .
these were obtained at the nlo level in ref .
@xcite by matching the @xmath30 direct calculation of the process ( i.e. formula ( [ eq : sigma ] ) ) with the expansion of formula ( [ eq : factorization ] ) at order @xmath31 .
they have the form _
q(x,_0,m)&=&(1-x)+(_0 ) d^(1)_q(x,_0,m)+o(^2 ) + [ eq : ini ] _
g(x,_0,m)&=&(_0 ) d^(1)_g(x,_0,m)+o(^2 ) , all the other components being of order @xmath32 . thus , in order to compute the nlo resummed expansion , one takes the initial conditions eqs .
( [ eq : ini ] ) , at a value of @xmath33 of order @xmath5 , evolves them at the scale @xmath6 ( taken to be of order @xmath4 ) , and then applies formula ( [ eq : factorization ] ) , using a nlo expression for the partonic cross section [ eq : sigmahat ] ( x , e , ) = ^(0)_i(x ) + ^(1)_i(x , e , ) ( ) + o(^2 ) .
for example , if the parton @xmath20 is the heavy quark itself , one gets [ eq : sigmahatq ] ( x , e , ) = ( 1-x ) + _ q^(1)(x , e , ) ( ) + o(^2 ) , where we have normalized the cross section to 1 at zeroth order in the strong coupling constant . the procedure outlined above guarantees that all terms of the form @xmath34 ( leading order ) and @xmath35 ( next - to - leading order ) , where @xmath36 is the large logarithm , are included correctly in the resummed formula .
observe that , at the nlo level , the scale that appears in @xmath31 in eqs .
( [ eq : ini ] ) and ( [ eq : sigmahat ] ) could be changed by factors of order 1 , since this amounts to a correction of order @xmath32 .
however , one can not set @xmath37 in eqs .
( [ eq : ini ] ) ( or @xmath38 in formula ( [ eq : sigmahat ] ) ) , since this amounts to a correction of order @xmath39 , and thus it would spoil the validity of the resummation formula at the nlo level .
there is essentially no room for these considerations to fail .
they are a consequence of the factorization theorem for fragmentation functions , which is quite well established @xcite . the validity of this procedure has however been questioned by kniehl et al . in ref .
@xcite . in their procedure , the heavy quark short distance cross section is replaced by [ eq : sigmahatprime ] ( x , e , ) = ( 1-x ) + ( e ) + o(^2 ) , and the initial condition by [ eq : iniprime ] ^_q(x)=(1-x ) , which is to be evolved from the scale @xmath33 to the scale @xmath6 using the nlo @xmath19 evolution equations .
this procedure differs at the nlo level from the standard procedure advocated in ref .
the difference starts to show up in the terms of order @xmath39 .
recently , we have completed a calculation of heavy quark inclusive production at order @xmath32 @xcite . using this calculation , we are in a position to verify explicitly the approach of ref .
@xcite , and thereby dismiss the approach of ref . @xcite . in the next section we describe the procedure we followed in detail .
instead of dealing with the realistic case of @xmath40 decay , we perform the calculation for a hypothetical vector boson @xmath41 that couples only to the heavy quark with vectorial coupling .
we introduce the following notation for the mellin transform : f(n ) _ 0 ^ 1 dx x^n-1 f(x ) .
we adopt the convention that , when @xmath42 appears instead of @xmath43 as the argument of a function , we are actually referring to the mellin transform of the function . this notation is somewhat improper , but it should not generate confusion in the following , since we will be working only with mellin transforms .
the mellin transform of the factorization formula ( [ eq : factorization ] ) is given by [ eq : sigman ] ( n , e , m ) = _ i _ i(n , e , ) _
i(n,,m ) , where ( n , e , m ) = _ 0 ^ 1 dx x^n-1 ( x , e , m ) , and a similar one for @xmath44 , and the mellin transform of the altarelli - parisi evolution equation ( [ eq : ap ] ) is [ eq : apn ] = _ j ( ) _ j(n,,m ) .
we want to obtain an expression for @xmath45 valid at the second order in @xmath31 .
thus , we need the solution of eq .
( [ eq : apn ] ) , with initial condition at @xmath46 , accurate at order @xmath32 .
this is easily done by rewriting eq .
( [ eq : apn ] ) as an integral equation & & _
i(n,,m ) = _ i(n,_0,m ) + [ eq : integapn0 ] & & + _ j _ _ 0^d^2 ( ^ ) _ j(n,^,m ) .
the terms proportional to @xmath32 can be evaluated at any scale ( @xmath6 or @xmath33 ) , the difference being of order @xmath47 .
factors involving a single power of @xmath31 can instead be expressed in terms of @xmath48 using the renormalization group equation ( ^)&=&(_0)-2b_0 ^ 2(_0 ) + o(^3(_0 ) ) b_0 & = & , where @xmath23 is the number of flavours including the heavy one .
equation ( [ eq : integapn0 ] ) then becomes _
i(n,,m ) & = & _ i(n,_0,m ) + _ j _ _ 0^d^2 ( _ 0 ) p_ij^(0)(n ) _
j(n,^,m ) + & + & _ j ^2(_0 ) p_ij^(1)(n ) _
j(n,_0,m ) + [ eq : integapn1 ] & - & 2b_0_j ^2(_0 ) p_ij^(0)(n ) _ j(n,_0,m)^2 .
we now need to express @xmath49 on the right - hand side of the above equation as a function of the initial condition , with an accuracy of order @xmath31 .
this is simply done by iterating the above equation once , keeping only the first two terms on the right - hand side .
our final result is then _
i(n,,m ) & = & _ i(n,_0,m ) + _ j ( _ 0 ) p_ij^(0)(n ) _
j(n,_0,m ) + & + & _ kj ^2(_0 ) p_ik^(0)(n)p_kj^(0)(n ) _ j(n,_0,m)^2 + & + & _ j ^2(_0 ) p_ij^(1)(n ) _
j(n,_0,m ) + [ eq : integapn2 ] & - & 2b_0_j ^2(_0 ) p_ij^(0)(n ) _ j(n,_0,m)^2 . since the initial condition is _
j(n,_0,m)=_jq+(_0)d^(1)_j(n,_0,m ) + o(^2(_0 ) ) , eq .
( [ eq : integapn2 ] ) becomes , with the required accuracy : _
i(n,,m ) & = & _ iq+(_0)d^(1)_i(n,_0,m ) + ( _ 0 ) p_iq^(0)(n ) + & + & _ j ^2(_0 ) p_ij^(0)(n ) d^(1)_j(n,_0,m ) + & + & _ k ^2(_0 ) p_ik^(0)(n)p_kq^(0)(n ) ^2 + & + & ^2(_0 ) p_iq^(1)(n ) + [ eq : integapn3 ] & - & 2b_0 ^2(_0 ) p_iq^(0)(n ) ^2 .
re - expressing @xmath48 in terms of @xmath6 , we get _
i(n,,m ) & = & _ iq+()d^(1)_i(n,_0,m ) + 2b_0 ^ 2()d^(1)_i(n,_0,m ) + & + & ( ) p_iq^(0)(n ) + _ j ^2 ( ) p_ij^(0)(n ) d^(1)_j(n,_0,m ) + & + & _ k ^2 ( ) p_ik^(0)(n)p_kq^(0)(n ) ^2 + ^2 ( ) p_iq^(1)(n ) + [ eq : integapn ] & + & b_0 ^2 ( ) p_iq^(0)(n ) ^2 .
the partonic cross sections are given by [ eq : partn ] _
i(n , e,)=_iq+_i+ ( ) ^(1)_i(n , e , ) + o(^2 ( ) ) , where @xmath50 vanishes unless @xmath20 is either @xmath2 , @xmath51 or @xmath52 .
thus , combining eq .
( [ eq : partn ] ) with eq .
( [ eq : integapn ] ) according to eq .
( [ eq : sigman ] ) , we obtain ( n , e , m ) & = & 1 + ( ) + & + & ^2()\{_i ^(1)_i(n , e , ) p^(0)_iq + 2b_0d^(1)_q(n,_0,m ) + & + & _ j d^(1)_j(n,_0,m ) + & + & _ k p_kq^(0)(n ) ^2 + [ eq : sigmanall ] & + & + b_0 p_qq^(0)(n ) ^2 } .
the above formula should accurately describe the terms of order @xmath53 , @xmath31 , @xmath54 and @xmath39 .
terms of order @xmath32 , without logarithmic enhancement , are not accurately given by the fragmentation formalism at nlo level , and have consistently been neglected .
the lowest order splitting functions are given by p^(0)_qq(n ) & = & c_f , + p^(0)_qg(n ) & = & p^(0)_g(n ) = c_f , + p^(0)_gq(n ) & = & t_f , where , restricting ourselves to integer values of @xmath42 , s_1(n)=_j=1^n . @xmath55 and @xmath56 are given by [ qqandqbarq ]
p_qq^(1)(n ) = p^ns_qq(n ) + p_q^q(n),p_q^(1)(n ) = p^ns_q(n ) + p_q^q(n ) , where the non - singlet components are given by ll p_qq^ns(n ) = p^c_f_qq(n)+ p^c_a_qq(n)+p^n_f_qq(n ) , & p_q^ns(n ) = p^c_f_q(n)+ p^c_a_q(n ) , + p^c_f_qq(n)=c_f^2 , & p^c_a_qq(n)=
c_f c_ap_g(n ) , + p^n_f_qq(n)=n_f c_f t_f p_nf(n ) , & p^c_f_q(n)=c_f^2 p_a(n ) , + p^c_a_q(n)=- c_f c_a p_a(n ) , & and p_q^q(n ) = -c_f t_f .
@xmath57 , @xmath58 , @xmath59 and @xmath60 were taken from the appendix of ref .
@xcite and @xmath61 is given in eq .
( 5.39 ) of ref .
we have obtained our explicit expression for @xmath62 using the equation [ pqpq ] p_q^q= , where @xmath63 is the singlet component is used for the `` sea '' component , and @xmath64 is used for the singlet one .
here we use @xmath64 for the full @xmath65 splitting function .
] , calculated in ref .
equation ( [ pqpq ] ) is easily seen to follow from eqs .
( [ qqandqbarq ] ) and from eqs .
( 2.42 ) of ref .
@xcite .
the expressions for @xmath66 and @xmath67 are respectively given in eq .
( a.12 ) and ( a.13 ) of ref .
the coefficient @xmath68 can be obtained by performing the mellin transform of the expression @xmath69 , where @xmath70 and @xmath71 are given in eq .
( 2.16 ) of ref .
g^(1)(n , e , ) & = & c_f \ { + 4 + & & - 2 } + d^(1)_g(n,_0,m)&=&p^(0)_gq(n ) . in order
to make a more detailed comparison with our fixed order calculation , we separate the @xmath72 contributions to @xmath45 according to their colour factors .
choosing for simplicity @xmath73 and @xmath74 , and using the notation ^(1)_q / g(n)=^(1)_q / g(n , e,)_=e , d^(1)_q / g(n)=d^(1)_q / g(n,_0,m)__0=m , we write ( n , e , m ) & = & 1 + ( e ) a(n , e , m ) + ^2(e ) b(n , e , m ) + a(n , e , m)&= & ^(1)_q(n)+d^(1)_q(n ) + p_qq^(0)(n ) + b(n , e , m)&=&b_c_f(n , e , m)+b_c_a(n , e , m ) + b_n_f(n , e , m)+b_t_f(n , e , m ) + b_c_f(n , e , m)&=&\ { p^(0)_qq(n ) + p^c_f_qq(n)+ p^c_f_q(n ) } + & & + ^2 ^2 + b_c_a(n , e , m)&= & + & & + c_a p^(0)_qq(n ) ^2 + b_n_f(n , e , m)&= & + & & -n_f t_f p^(0)_qq(n ) ^2 + b_t_f(n , e , m)&= & + & & + p_qg^(0)(n ) p_gq^(0)(n)^2 , [ eq : colcoef ] where the subscripts @xmath75 , @xmath76 , @xmath23 and @xmath77 denote the @xmath78 , @xmath79 , @xmath80 and @xmath81 colour components . the fixed order calculation of ref .
@xcite can be used to compute the cross section for the production of a heavy quark pair plus one or two more partons , at order @xmath32 .
we separate contributions in which four heavy quarks are present in the final state , from those where a single @xmath82 pair is present together with one or two light partons .
these last contributions were computed only in a three - jet configuration , and they are singular in the two - jet limit , that is to say , when @xmath83 . furthermore , the virtual corrections to the two - body process @xmath84 are not included in our calculation . in order to remedy for these problems , we proceed as follows . the @xmath72 inclusive cross section for @xmath85 ,
can be written symbolically in the following form = a^(0)(1-x ) + dy a^(1)(x , y ) + ^2 where @xmath86 denotes all the other kinematical variables , besides @xmath43 , upon which the final state may depend .
we assume @xmath73 , and we do not indicate , for ease of notation , the dependence upon @xmath4 and @xmath5 of the various quantities .
the term @xmath87 arises from final states with a single @xmath82 pair plus at most two light partons , while @xmath88 arises from final states with two @xmath82 pairs .
the factor of 2 in front of the @xmath88 contribution accounts for the fact that we may detect either one of the two heavy quarks .
the moments of the inclusive cross section can be written in the following way ( n)=dx x^n-1 = + dxdy ( x^n-1 - 1)a^(1)(x , y ) & & + + ^2 & & where = a^(0 ) + dxdy a^(1)(x , y ) + ^2 .
the expression for @xmath89 can now be easily computed with our program , since the factors regularize the singularities in the two - jet limit , and suppress the two - body @xmath84 virtual terms .
furthermore , in the massless limit = 1 + 2(e ) + c^2(e ) + o ( ) + o(^3 ) where @xmath90 is a constant .
in fact , the @xmath72 term does not contain any large logarithm , as long as @xmath31 is the coupling with @xmath23 flavours , including the heavy one we have@xmath91 . ] . reintroducing the energy and mass dependence
, we have ( n , e , m ) & = & 1 + ( e ) l(n , e , m)+^2(e ) m(n , e , m ) + & & + c^2(e)+ o ( ) + o(^3 ) + l(n , e , m)&= & 2 + dx dy @xmath92a^(1)(x , y ) + m(n , e , m)&= & dxdy @xmath92 a^(2)_l(x , y ) + & & + 2dx dy @xmath93 a^(2)_h(x , y ) .
we have calculated @xmath94 and @xmath95 numerically , using @xmath96 gev and @xmath97 8 , 4 , 3 , 2.5 , 2 , 1.5 , 1 , 0.6 , 0.5 , 0.4 , 0.2 gev , for a vector current coupled to the heavy quark .
we expect that , for small masses , @xmath98 should coincide with @xmath94 , and @xmath95 should differ from @xmath99 by a mass and energy independent quantity , since such term is actually beyond the next - to - leading logarithmic approximation .
we find very good agreement between @xmath98 and @xmath94 .
we present the results for @xmath95 separated into the different colour components m(n , e , m)=m_c_f(n , e , m)+m_c_a(n , e , m ) + m_n_f(n , e , m)+m_t_f(n , e , m ) . in fig .
[ fig : ca ] we have plotted our results for @xmath100 ( crosses with error bars ) and for @xmath101 ( solid lines ) .
an arbitrary ( @xmath42-dependent ) constant has been added to the curves for @xmath101 , in order to make them coincide with the numerical result for @xmath102 .
we find , as the mass gets smaller , satisfactory agreement for all moments .
notice that , as intuitive reasoning would suggest , for higher moments we need smaller masses to approach the massless limit . in figs .
[ fig : nf ] , [ fig : cf ] and [ fig : tf ] we report the analogous results for the remaining colour combinations .
again , we find satisfactory agreement . if one follows the procedure proposed by kniehl et al .
( [ eq : colcoef ] ) are modified in the @xmath79 and in the @xmath103 coefficients .
more specifically , the terms proportional to @xmath104 all disappear from the expressions of the @xmath79 and of the @xmath105 coefficients .
in fact , by inspecting formulae ( [ eq : sigmahatprime ] ) and ( [ eq : iniprime ] ) , and the derivation of eq .
( [ eq : sigmanall ] ) , we see that the only relevant difference between the two approaches is that the term @xmath106 is replaced by @xmath107 , which , using the renormalization group equation , amounts to a difference of @xmath108 , precisely what is needed to cancel the term of the same form appearing in eq .
( [ eq : sigmanall ] ) . the modified result of kniehl et al .
is also shown in figs .
[ fig : ca ] and [ fig : nf ] ( dashed lines ) .
it is quite clear that their approach does not work .
in the present work , we have verified at order @xmath32 the nlo fragmentation function approach to the computation of the heavy quark fragmentation function given in ref . @xcite . besides excluding an alternative that has been proposed in the literature @xcite , we have given a verification of several ingredients that go into the fragmentation function formalism , such as the initial conditions , computed in ref .
@xcite , the nlo splitting functions in the time - like region @xcite , and finally we have also performed a further test of the validity of the fixed order calculation of ref .
@xcite .
99 b. mele and p. nason , .
g. altarelli and g. parisi , .
mueller , .
g. curci , w. furmanski and r. petronzio , .
w. furmanski and r. petronzio , .
e. g. floratos , r. lacaze and c. kounnas , , .
j. kalinowski , k. konishi , p. n. scharbach and t. r. taylor , ; j. kalinowski , k. konishi and t. r. taylor , .
b. a. kniehl , g. kramer and m. spira , preprint desy-96 - 210 , hep - ph/9610267 .
p. nason and b. r. webber , .
p. nason and c. oleari , preprint cern - th/97 - 219 , hep - ph/9709360 . | we use a recent @xmath0 calculation of the differential cross section for the production of heavy quarks in @xmath1 annihilation to compute a few moments of the heavy quark single - inclusive production cross section .
we verify that , contrary to some recent claims , the leading and next - to - leading logarithmic terms in this cross section are correctly given by the standard nlo fragmentation function formalism for heavy quark production . * 0.9 cm on the fragmentation function forheavy quarks in @xmath1 collisions * + 0.1 cm cern ,
th division , geneva , switzerland + .5 cm * carlo oleari * + .1 cm dipartimento di fisica , universit di milano and infn , milan , italy cern - th/97 - 209 + |
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the lanterns having been lit , we all fell to work with a zeal worthy a more rational cause ; and , as the glare fell upon our persons and implements , i could not help thinking how picturesque a group we composed , and how strange and suspicious our labors must have appeared to any interloper who , by chance , might have stumbled upon our whereabouts .
_ we dug very steadily for two hours . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this is an excerpt from the edgar allan poe story _ the gold - bug _ @xcite . published in @xmath0 , it is a gripping tale about captain kidd s lost pirate treasure map and digging for booty on the carolina shore . locating the digging spots requires a little trigonometry .
poe has our heroes dig two holes but we will show that they must really overlap .
the main character of the story is william legrand , who is a nobleman in impoverished circumstances .
he lives on tiny , mostly unpopulated , charleston island off the coast of south carolina . with him
is the old , freed , slave jupiter .
it is widely believed that legrand is slightly unbalanced and jupiter takes it upon himself to look after him .
the third character is the narrator , an old friend , who accompanies them on the treasure hunt .
legrand has found a parchment half buried in the sand beside an old ship s boat .
it is written with peculiar symbols and turns out to be a cryptogram .
he translates it and finds it gives veiled directions to a secluded spot in the woods on the mainland .
( poe gives a lovely explanation of how to crack a cryptogram . ) here the three companions dig by lantern light for what they hope is treasure .
the gold - bug itself is a living beetle that legrand has found .
it seems to be made of real gold .
this gives the story a chilling air of mysticism .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ main branch seventh limb east side shoot from the left eye of the deaths - head a bee line from the tree through the shot fifty feet out _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ at the digging site in the woods there is an enormous old tulip tree .
legrand has jupiter climb it .
going up @xmath1 feet , he climbs onto the seventh limb up the east side of the tree .
out on the limb a human skull has been fastened by the pirates .
he drops the gold - bug through the right eye socket and it falls to the ground .
they then draw a line a further @xmath2 feet out from the tree in this direction and dig a hole @xmath3 feet in diameter and @xmath4 feet deep . alas , no treasure .
they then realise jupiter has dropped the bug through the wrong eye socket .
legrand moves the mark under the skull @xmath5 inches , as if it had fallen through the other eye socket .
a line is then extended @xmath2 feet through this new point away from the trunk . here
they dig a second time .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ i dug eagerly , and now and then caught myself actually looking , with something that very much resembled expectation , for the fancied treasure [w]e were again interrupted by the violent howlings of the dog .
[ h]e made furious resistance , and , leaping into the hole , tore up the mould frantically with his claws . in a few seconds he had uncovered a mass of human bones _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ some simple trigonometry shows that , under reasonable assumptions for the radius of the tree and the distance the skull is from the trunk , the two holes must have overlapped , even though the narrator says they were several yards apart . in figure 1 we have the set up .
points labeled in the figure are @xmath6 ( centre of tree ) , @xmath7 ( drop point through left eye socket ) , @xmath8 ( drop point through right eye socket ) , @xmath9 ( first digging site ) , @xmath10 ( second digging site ) .
distances are @xmath11 ( radius of tree trunk ) , @xmath12 ( distance of each eye socket from tree trunk ) , @xmath13 ( radius of first hole ) , @xmath14 ( radius of second hole ) .
all distances are in feet , as they are in the story .
the angle between @xmath15 and @xmath16 is @xmath17 .
[ width=6 in ] digfigure5.eps legrand states that points @xmath7 and @xmath8 are @xmath18 apart .
the distances @xmath19 and @xmath20 are @xmath21 .
note that @xmath22 $ ] .
the distance between the centres of the holes is @xmath23 the condition that the holes do not overlap is that @xmath24 . with this
reduces to @xmath25 we know from the story that @xmath26 .
the narrator says that @xmath14 is slightly larger .
even if we say @xmath27 , for non - overlapping holes we must have @xmath28 .
this is clearly impossible .
the tree is _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ an enormously tall tulip tree , which stood , with some eight or ten oaks , upon the level , and far surpassed them all , and all other trees which i had then ever seen , in the beauty of its foliage and form , in the wide spread of its branches , and in the general majesty of its appearance .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ and poe states that jupiter climbs @xmath29 or @xmath1 feet up to the branch with the skull
. we should take @xmath11 to be the radius of the trunk at the height of the skull branch . for there to be a branch at that height substantial enough to hold a man s weight
, @xmath11 must be at least @xmath30 inches and could easily be @xmath31 inches or more .
( a tulip tree on my campus was recently cut down .
the stump measures @xmath32 across .
it was not a particularly old tree . ) but that means @xmath33 .
this puts the skull too close to the tree trunk .
jupiter crawls out to the end of the branch so this must be more than @xmath34 feet . taking @xmath35 only makes things worse .
hence , although it s a great story , poe neglected to do his trigonometry homework and the two treasure holes must have overlapped . from the wording in the story , it is not entirely clear whether the @xmath2 feet to the centre of the two holes is measured from the eye socket drop points @xmath7 and @xmath8 , as we have taken it , or from the tree trunk .
if from the trunk then the same assumptions on @xmath13 , @xmath14 and @xmath11 as above require @xmath12 be even smaller for the holes to not overlap .
_ the gold - bug _ has been in the public domain for many years so there are a great number of editions available , both print and electronic .
excerpts here are taken from @xcite . for more on fiction that contains mathematics ,
see alex kasman s web page @xcite . in the story , poe hints the treasure is captain kidd s .
william kidd , born circa 1654 , was hanged for murder and piracy in england in 1701 .
he was certainly a privateer , which was legal in those times , and he held a letter of marque signed by king william iii authorising him to attack french and certain other vessels . it is uncertain whether he was a pirate or not . after his death a legend grew that he had left a treasure buried on the north american atlantic coast .
for example , see @xcite .
99 g. harris , _ treasure and intrigue : the legacy of captain kidd _ , toronto , dundurn , 2002 .
http://kasmana.people.cofc.edu/mathfict/default.html e.a .
poe , _ the gold - bug _ , http://www.online-literature.com/poe/32/ | the classic edgar allan poe story _ the gold - bug _ involves digging for pirate treasure . locating the digging sites requires some simple trigonometry .
# 1#20=0=0 0 - 0.20 2=00.4pt2 |
tbvo@xmath2 and tbaso@xmath2 belong to a group of compounds which undergo a structural phase transition induced by the cooperative jahn - teller effect . both tbvo@xmath2 and tbaso@xmath2 exhibit a transition from the tetragonal _ _ i__4@xmath6/_amd _ structure at high temperatures to the twinned orthorhombic _ fddd _ phase at temperatures below 35 k@xcite . at room temperature the low - lying crystal electric field levels for the tb@xmath7 ions in the tbvo@xmath8 environment consist of two singlets separated by
2.23 mev and a degenerate , non - kramers doublet midway between the singlets@xcite .
the jahn - teller phase transition changes the local tb@xmath7 environment , with a concomitant splitting of the doublet , and increase of the singlet separation to 6.32 mev@xcite .
the resulting magnetic ground state for the tb@xmath7 ions can then order magnetically at much lower temperatures@xcite .
the structural phase transitions have also been studied in disordered systems wherein the non - magnetic cations are mixed .
such disordered materials as tbv@xmath0as@xmath9o@xmath8 are characterized by the presence of quenched random strains introduced by the ionic radius mismatch between v@xmath10 and as@xmath10 , which is quite large @xmath11 20@xmath12 .
these materials and their jahn - teller phase transitions have been known and studied for some time .
for example , optical birefrigence studies of the critical properties of tbvo@xmath8 and dyvo@xmath8 showed that the former displayed mean field criticality near its jahn - teller transition , while the latter displayed 3d ising - like universality near its transition@xcite .
early x - ray experiments measured the order parameter relevant to these jahn - teller transitions , the orthorhombic strain , but were not able to give accurate critical property measurements for this family of terbium oxides@xcite .
an x - ray scattering study of the related tbv@xmath0p@xmath9o@xmath8 system@xcite has been carried out , although again , the emphasis was not on the critical regime .
the role of quenched disorder on the jahn - teller phase transitions in the tbv@xmath0as@xmath9o@xmath8 system has more recently been the subject of further optical birefrigence studies , which also show changes in criticality induced by the presence of impurities@xcite .
the change in criticality has been discussed in the context of the random field ising model@xcite and how it may be realized in a structural analogue of the more - familiar case arising in magnetism - that of the dilute ising antiferromagnet in the presence of a uniform magnetic field@xcite .
we have carried out a high resolution x - ray scattering study of the tetragonal to orthorhombic jahn - teller phase transitions in the tbv@xmath13as@xmath9o@xmath8 system , with particular emphasis on the critical regime , where the orthorhombic strain is small and fluctuations in the strain may be expected to be large .
this will enable a quantitative microscopic understanding of the critical phenomena in the pure , end members of the family , and show the extent to which the critical behaviour in the mixed system is modified by the presence of quenched disorder .
single crystals of tbvo@xmath2 and tbaso@xmath2 and their solid solutions were grown from pbo / pbf@xmath14 flux by the slow cooling method .
pre - annealed starting materials tb@xmath14o@xmath15 and x@xmath14o@xmath16 where x is either v or as were weighed and combined with pbo and pbf@xmath14 , which act as high temperature solvents . the mix was then melted in a pt crucible at 1300@xmath17c and slow cooled ( 1 - 2 deg / h ) to 900@xmath17c @xcite .
single crystals in the form of elongated rods with typical dimensions of 5x2x2 mm were separated from this melt .
the tbv@xmath18as@xmath19o@xmath8 single crystal studied was the same one previously studied by schriemer et al @xcite . in that study the relative as concentration in tbv@xmath0as@xmath9o@xmath8
is given as x=0.15 , in good agreement with the value x=0.17 which we have determined from the current x - ray measurements .
x - ray scattering measurements were performed using a rotating - anode , cu k@xmath20 x - ray source and a four circle diffractometer employing a two dimensional area detector .
the single crystals were mounted within a helium - filled sample cell which was attached to the cold finger of a closed - cycle helium refrigerator .
scattering measurements were carried out as a function of temperature with a temperature stability of @xmath11 0.005 k , appropriate for critical scattering studies .
cuk@xmath21 radiation from the 18 kw rotating anode x - ray generator was selected using a perfect single crystal ge ( 110 ) monochromator .
the diffracted beam was measured using a bruker histar area detector , mounted on the scattering arm at a distance of 0.7 m from the sample .
this configuration allowed for high resolution characterization of the bragg peaks .
the scattered intensity was measured as a function of temperature for both the ( 6,6,0 ) bragg reflection in the tetragonal phase , which splits into the ( 12,0,0 ) and ( 0 , 12 , 0 ) reflections in the twinned orthorhombic phase , as well as for the ( 8,0,0 ) bragg reflection in the tetragonal phase , which splits into ( 8 , 8 , 0 ) and ( 8 , -8 , 0 ) reflections in the same twinned orthorhombic phase . ,
tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 .
( a ) and ( b ) show the splitting of the ( 8 , 0 , 0 ) bragg peak into ( 8 , 8 , 0)/(8 , -8 , 0 ) bragg peaks in tbvo@xmath8 .
( c ) and ( d ) , ( e ) and ( f ) , and ( g ) and ( h ) show the splitting of the ( 6 , 6 , 0 ) bragg peak into the ( 12 , 0 , 0 ) and ( 0 , 0 , 12 ) bragg peaks in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 , respectively .
the data on the left ( ( a ) , ( c ) , ( e ) and ( g ) ) are above t@xmath22 while the data on the right ( ( b ) , ( d ) , ( f ) and ( h ) ) are below t@xmath22 .
the diagonal line in ( b ) indicates the cut used for the longitudinal scans in figure 5 . a vertical line would indicate a transverse scan . ]
figure 1 shows representative x - ray scattering contour maps taken for tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 above and below their appropriate phase transition temperatures .
figure 1(a ) and ( b ) shows data taken around the ( 8 , 8 , 0 ) bragg position ( in the orthorhombic phase ) in tbvo@xmath2 , while figures 1 ( c ) and ( d ) , ( e ) , and ( f ) , and ( g ) and ( h ) shows data taken around the ( 12 , 0 , 0 ) bragg positions ( in the orthorhombic phase ) in tbvo@xmath2 , tbaso@xmath2 , and tbv@xmath18as@xmath19o@xmath2 , respectively . all of these scattering data sets were acquired by integrating the bragg scattering in the vertical direction for each sample rotation angle ( @xmath23 , plotted on the y - axis of fig .
1 ) , and then plotting the resulting intensity as a function of scattering angle , 2@xmath24 . the dashed line in fig .
1(c ) indicates a longitudinal scan , while a vertical line on any of these panels would indicate a transverse scan .
the transverse splitting of the ( 8 , 0 , 0 ) tetragonal bragg peak into ( 8 , 8 , 0 ) orthorhombic bragg peaks in tbvo@xmath2 is shown in fig .
1(a ) and ( b ) , while the longitudinal splitting of the tetragonal ( 6 , 6 , 0 ) bragg peaks into twin orthogonal peaks at ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) in all three tb - based oxides is seen in figs 1(c ) - ( h ) .
these data sets can be analysed by extracting the positions for all the bragg peaks in the field of view in both scattering angle , 2@xmath24 , and sample orientation angle , @xmath23 .
the scattering angle for the single crystal peaks in both the tetragonal phase above t@xmath4 , and the twinned orthorhombic phase below t@xmath4 , is used to extract the relevant lattice parameters and orthorhombic strain , defined as the difference between the a and b lattice parameters .
the scattering angles for all of the ( 6 , 6 , 0 ) ( tetragonal phase ) and ( 12 , 0 , 0 ) ( orthorhombic phase ) bragg peaks of all three samples are shown as a function of temperature in fig .
figure 3 ( a ) shows the correponding orthorhombic strain below t@xmath4 in all three samples . figure 3 ( b ) shows the splitting in domain orientation for tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 , extracted from the difference in sample orientation , @xmath23 positions , of the bragg peaks from the principal ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) orthorhombic twin domains below t@xmath4 . figure 2 ( a ) shows the splitting of ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) bragg peaks in all three systems in absolute units of scattering angle , while fig .
2 ( b ) shows the same data , but now relative to the scattering angle of the bragg peak in the high temperature tetragonal ( htt ) phase .
this data shows quite clearly that all three samples display continuous phase transitions , as was known previously for tbvo@xmath2@xcite and tbaso@xmath2@xcite , and that their critical properties are similar . the scattering in fig . 2 ( a ) however , also shows that within the high temperature tetragonal phase , the lattice parameter ( identified by its scattering angle ) for tbv@xmath18as@xmath19o@xmath2 lies @xmath11 17@xmath12 of the way between tbvo@xmath2 and tbaso@xmath2 , as given by a linear variation in lattice parameter between the end members of the series , tbvo@xmath2 and tbaso@xmath2 .
such a linear variation is observed in the tbv@xmath0p@xmath9o@xmath8 system@xcite .
this establishes an accurate concentration for the tbv@xmath18as@xmath19o@xmath2 single crystal by microscopic means .
figure 3 ( a ) shows the corresponding orthorhombic strains , defined as the difference in lattice parameters a and b , as a function of temperature below t@xmath4 for each of tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 .
related measurements of the orthorhombic strain have been measured previously as a function of temperature for the tbv@xmath0p@xmath9o@xmath8 system@xcite . however , the present measurements are taken sufficiently close to the phase transitions such that an accurate determination of t@xmath4 and a quantitative study of the critical properties of these phase transitions is possible .
it is clear from the data that the order parameter , the orthorhombic strain , rises very sharply with decreasing temperature , such that little of this data would reside in the asymptotic critical regime , where the order parameter is small compared with its saturation value .
therefore we fit this data to a full mean field solution , eq .
1 , rather than simply to the asymptotic expression for small values of the order parameter . this gives a mean field critical exponent for the order parameter , @xmath25 = 0.5 , but the full data sets are accurately described .
this description of the order parameters as a function of temperature using the solution to the transcendental eq . 1 below
, is shown as the solid lines in fig .
3 ( a ) and ( b ) . clearly the full mean field solution provides an excellent description of this data .
@xmath26\ ] ] figure 3 ( b ) shows the evolution of the splitting between the twin domain orientation angles , @xmath23 , describing the ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) twin orthorhombic domains below t@xmath4 in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 .
again the temperature dependence of these splittings is very well described by the full mean field solution to the order parameter , eq .
1 . this plot follows the evolution of the domain orientation for both the majority and minority domains present within the field - of - view of our scattering experiments , for which typical data sets are shown in fig .
1 . for tbvo@xmath2 and tbaso@xmath2 ,
a single majority and minority domain is visible at most temperatures .
however the tbv@xmath18as@xmath19o@xmath2 data sets are qualitatively different with many domains visible even at temperatures well removed from t@xmath4 .
for example , the data set for tbv@xmath18as@xmath19o@xmath2 at t=26.3 k ( t@xmath4 - 4 k ) shows a total of 5 twin domains ; 2 majority twin domains at high scattering angle and 3 minority twin domains at low scattering angle . moreover the dynamic range in domain orientation angle spanned by the multiple twin domains is considerably larger than is found in either tbvo@xmath2 or tbaso@xmath2 below t@xmath4 .
figure 3 ( b ) quantifies this behaviour , following the splitting of the majority and minority twin domain orientation angles ( @xmath23 ) below t@xmath4 .
the extrapolated , low temperature @xmath23 splitting is very similar for tbvo@xmath2 and tbaso@xmath2 , but considerably larger for tbv@xmath18as@xmath19o@xmath2 , where the splitting between the orientation angle , @xmath23 , of the most intense minority twin and most intense majority twin domain is plotted . the order parameter data shown in fig .
3 , along with the fits to full mean field behaviour , eq . 1 , allow a precise determination of the critical temperature t@xmath4 , in these systems . fitting both the orthorhombic strain shown in fig .
3 ( a ) and the twin domain orientation splitting shown in fig .
3 ( b ) , we obtain t@xmath4 = 33.26(2 ) k , 30.32(2 ) k and 27.30(2 ) k for tbvo@xmath2 , tbv@xmath18as@xmath19o@xmath2 and tbaso@xmath2 , respectively . in itself , this is a somewhat surprising result , as it shows the t@xmath4 for tbv@xmath18as@xmath19o@xmath2 to lie midway between that of tbvo@xmath2 and tbaso@xmath2 , despite only 17@xmath12 of the v ions being replaced by as ions .
this is very likely a manifestation of the quenched impurities on the jahn - teller driven phase transition in tbv@xmath18as@xmath19o@xmath2 .
the behaviour of the twin domains below t@xmath4 in tbv@xmath18as@xmath19o@xmath2 can be qualitatively understood in terms of pinning of the twin domains by random strains due to the quenched impurities .
clearly , this does not influence the critical properties of the order parameter , the orthorhombic strain , which remains well described by mean field behaviour at all temperatures , as is the case for both tbvo@xmath2 and tbaso@xmath2 .
the influence of quenched impurities on twin domain interfaces has been investigated by light scattering techniques@xcite , which is expected to be sensitive to such large scale structure .
the present results are qualitatively consistent with these results , showing stable multi - domain twin states in tbv@xmath18as@xmath19o@xmath2 , distinct from tbvo@xmath2 and tbaso@xmath2 , and therefore a larger volume fraction taken up by domain interfaces in tbv@xmath18as@xmath19o@xmath2 . however , the change in critical properties previously reported@xcite on introduction of quenched impurities is not observed . for each of the three samples : tbvo@xmath2 ( closed symbols ) , tbaso@xmath2 ( open symbols ) and tbv@xmath18as@xmath19o@xmath2 ( crossed symbols ) .
the structural phase transition is indicated by the splitting of the single bragg intensity into two with decreasing temperature .
figure 2a shows the true @xmath27 splitting for each of the compounds , while figure 2b shows the splitting relative to the scattering in the tetragonal phase at high temperatues .
] , tbv@xmath18as@xmath19o@xmath2 , and tbaso@xmath8 .
these have been fit to the mean field solution ( equation 1 ) for the order parameter as a function of temperature , which is shown as the solid lines .
figure 3b ( lower ) shows the domain orientation @xmath23 splitting of the ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) bragg peaks and fits to the same mean field solution .
the tbv@xmath18as@xmath19o@xmath2 @xmath23 splitting was taken as the difference between the majority and minority domains with the strongest bragg intensity .
this was an issue for tbv@xmath18as@xmath19o@xmath2 alone as the twinned domains collapsed into a single dominant majority and minority domain peak within a few degrees of t@xmath4 for the pure materials . ] .
this data is taken from longitudinal cuts to the full data sets , as shown in fig . 1 ( c ) . (
a ) and ( d ) were taken well above and below t@xmath4=33.26(2 ) k , respectively , and are resolution limited .
( b ) and ( c ) are relatively close to t@xmath4 and show substantial broadening in the longitudinal direction .
the resolution - corrected longitudinal widths of the profiles were extracted and are plotted as a function of temperature in figs .
5 and 6 . ]
close examination of the scattering data sets at the ( 12 , 0 , 0)/(0 , 12 , 0 ) bragg positions in both tbvo@xmath2 and tbaso@xmath2 revealed a very interesting longitudinal broadening in the vicinity of t@xmath4 .
representative longitudinal scans near the ( 6 , 6 , 0 ) bragg position in the tetragonal phase and ( 12 , 0 , 0 ) bragg positions in the orthorhombic phase of tbvo@xmath2 are shown in fig .
very similar data was obtained for tbaso@xmath2 , but it is not displayed .
figure 4 ( a ) shows a longitudinal scan through ( 6 , 6 , 0 ) well above t@xmath4 at t=47.05 k ( t@xmath4 + 13.8 k ) , while fig . 4 ( d ) shows a longitudinal scan through ( 12 , 0 , 0 ) at t=26.75 k , 6.51 k below t@xmath4 . both of these scans have approximately the same intrinsic width , and we consider them to be resolution limited .
figures 4 ( b ) and ( c ) however , show longitudinal scans much closer to t@xmath4 .
figure 4 ( b ) shows data just above t@xmath4 at t=33.58 k ( t@xmath4 + 0.32 k ) while fig .
4 ( c ) shows the same scan just below t@xmath4 , at t=32.82 k ( t@xmath4 - 0.44 k ) .
it is clear that both of these longitudinal scans near t@xmath4 are appreciably broader than those well removed from t@xmath4 , and therefore are not resolution limited .
we associate this broadening with fluctuations in the order parameter , similar to that observed near continuous magnetic phase transitions . however , in the present case , fluctuations in the orthorhombic strain lead to a longitudinal broadening of bragg peaks of the form ( 6 , 6 , 0 ) in the tetragonal phase . rather than narrowing in angle or reciprocal space as the phase transition
is approached , these peaks broaden and decrease in peak intensity as t@xmath4 is approached from above .
broadening is also observed below t@xmath4 , but the widths of the bragg peaks fall off quickly , such that all the scattering in both tbvo@xmath2 and tbaso@xmath2 is resolution limited for temperatures more than 2 k below t@xmath4 .
data of the form shown in fig .
4 was fit to an ornstein - zernike form , eq .
2 , convoluted with the resolution function appropriate to the measurement .
we employed the scattering well above t@xmath4 , at t @xmath11 50 k , as the resolution function , assuming that all fluctuation effects are gone at such high temperatures .
the results of fitting the tbvo@xmath2 data is shown as the solid lines in fig .
4 , and this clearly provides an excellent description of the longitudinal scans at all temperatures , similar quality fits to the tbaso@xmath2 were also obtained .
the resulting intrinsic widths to the longitudinal scans are shown as a function of temperature for tbvo@xmath2 in fig . 5 ( a ) and for tbaso@xmath2 in fig . 5 ( b ) .
data is shown for both independent warming and cooling runs for both materials , although no significant history dependence to the widths of the scattering were observed . for temperatures below t@xmath4 , longitudinal scans through both majority and minority twin domains
were investigated . over the narrow range of temperature below t@xmath4 for which a finite width can be observed ,
the intrinsic widths of the majority and minority twin domain peaks differ , with the minority twin peak showing a greater intrinsic width than the majority twin domain , consistent with the interpretation of the minority twin domains as being relatively small .
a related set of data near t@xmath4 was acquired for tbv@xmath18as@xmath19o@xmath2 , however the relevant longitudinal lineshapes were more complicated than those observed in tbvo@xmath2 and tbaso@xmath2 , and these results are not discussed here .
@xmath28 the most striking and interesting result is that the intrinsic width of the bragg peaks , interpreted as an inverse correlation length , displays a pronounced peak at t@xmath4 , reminiscent of @xmath29 anomolies in the heat capacity of systems undergoing continuous phase transitions .
the temperature dependence of the divergence of the intrinsic width , or inverse correlation length , is examined on a log - log plot in fig .
6 , wherein data above t@xmath4 in tbvo@xmath2 and tbaso@xmath2 is plotted vs reduced temperature ( t - t@xmath4)/t@xmath4 .
again independent warming and cooling runs are shown , with no history dependence observed .
this plot , which employs the same t@xmath4s used to describe the critical properties of the order parameter shown in fig .
3 , clearly brings out the power law nature of the divergence of the intrinsic width of the scattering as t@xmath4 is approached from above .
figure 6 ( a ) shows this data for both tbvo@xmath2 and tbaso@xmath2 , and it is clear that both materials display a very similar power law divergence over at least one and a half decades in reduced temperature .
figure 6 ( b ) shows the intrinsic width of the longitudinal scan in tbaso@xmath2 alone , where a single power law describes its divergence over almost three decades in reduced temperature .
the exponent characterizing the divergence is x=0.45 @xmath30 0.04 .
while we have characterized the power law divergence of the inverse correlation length , or intrinsic width , at t@xmath4 , we do not have a precise interpretation for this critical phenomenon . in their high - resolution
gamma - ray diffraction experiments on tbvo@xmath2 , smith and tanner@xcite observed a broadening in the rocking - curve widths , that is the transverse width of the bragg peaks , as the phase transition was approached from above that somewhat resembles the broadening that we measured .
they did not analyze it in detail , but suggested that it arose from the strains associated with random internal stresses relaxing as the elastic constant c@xmath31 tends to zero .
while such a mechanism could contribute in our case , it is difficult to believe that it would exhibit critical behaviour over such a wide range and with such well defined power law behaviour as we observe . as mentioned previously ,
our observed longitudinal broadening of the ( 6 , 6 , 0 ) ( tetragonal ) and the ( 12 , 0 , 0)/(0 , 12 , 0 ) ( orthorhombic ) bragg peaks in fig . 5 greatly resembles heat capacity anomolies associated with continuous phase transitions .
this may be a natural explanation of the intrinsic width , as the width is inversely proportional to the fluctuating domain size , and thus proportional to the volume fraction of the material occupied at any one time by interface .
the energy of such a system would scale as the amount of interface , and a longitudinal bragg peak width scaling as a derivative of the energy , the heat capacity , seems physically resonable
. the critical exponent x may be close to that describing fluctuations near a mean field , or mean field tricritical point , for which a large heat capacity exponent characteristic of a divergence is appropriate . at present
we can not be more precise , but hope that the observation and characteriozation of this critical behaviour informs and motivates further work and understanding .
( top ) and tbaso@xmath2 ( bottom ) . below t@xmath22 open data points refer to the majority twin domain , while closed data points refer to the minority twin domain .
square and circle data points refer to independent cooling and warming cycles . ] from eq .
2 , above t@xmath4 for tbvo@xmath8 and tbaso@xmath8 are shown as a function of reduced temperature .
the same data for tbaso@xmath8 , along with a fit to a power law divergence characterized by the exponent x=-0.45 @xmath30 0.04 . ]
we can compare the present results on the order parameter and fluctuations near the jahn - teller tetragonal to orthorhombic structural phase transitions in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 , to those in related systems .
the principal characteristics of the present results are mean field critical behaviour for the orthorhombic strain , and the observation of clear fluctuation effects in the vicinity of t@xmath4 .
two topical and well studied systems which undergo tetragonal to orthorhombic structural phase transitions are the high temperature superconductors la@xmath32ba@xmath9cuo@xmath2 and la@xmath32sr@xmath9cuo@xmath2 .
critical x - ray and neutron scattering on these materials do not show mean field behaviour , but rather the orthorhombic strain goes like the square of the order parameter@xcite , which is well described by three dimensional universality .
theoretical expectations for these systems are for 3d xy critical behaviour@xcite , and this is consistent with what is observed@xcite .
even qualitatively , it is clear that the growth in the orthorhombic strain with decreasing temperature in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 is different from that observed in la@xmath32ba@xmath9cuo@xmath2 and la@xmath32sr@xmath9cuo@xmath2 . comparing to a recent high resolution x - ray scattering study on the la@xmath32ba@xmath9cuo@xmath2 system@xcite
, we can see that the orthorhombic strain grows in a relatively gentle manner with decreasing temperature compared with what is observed in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 .
figure 3 shows that the latter orthorhombic strains grow to @xmath11 half of their saturation values at 0.9 @xmath33 t@xmath4 .
the same growth in the la@xmath32ba@xmath9cuo@xmath2 system requires lower temperatures and half of saturation is only achived by @xmath11 0.7 t@xmath4 .
there are differences of course , between the high temperature superconducting copper oxides and the terbium based oxides we report here .
the cuprates are quasi - two - dimensional , for example , while tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 are three dimensional .
most notably however the magnetic cu@xmath34 ion in la@xmath32ba@xmath9cuo@xmath2 and la@xmath32sr@xmath9cuo@xmath2 is not jahn - teller active - a single hole resides in a non - degenerate d orbital .
this would appear to be the most likely source for the difference in critical behaviour .
tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 possess tb@xmath7 ions whose orbitals and the occupation of the orbitals change on going through t@xmath4 , likely generating long range strain fields with concomittant mean field behaviour . another interesting comparison is to other tb@xmath7-based transition metal oxides . here
an interesting comparison can be made with a very recent high resolution x - ray scattering study of the cubic pyrochlore tb@xmath35ti@xmath35o@xmath36 @xcite .
this geometrically - frustrated magnetic material does not undergo a structural phase transition at finite temperature , but recent x - ray scattering measurements observe broadening of the allowed structural bragg peaks at low temperatures .
these have been interpreted as fluctuations above a very low temperature jahn - teller phase transition , which is either never realized , or only realized at unattainably low temperatures .
indeed , to our knowledge , the present work is the only other study of structural phase transitions which show such critical broadening of the bragg peaks in the vicinity of t@xmath4 .
the temperature scale over which these fluctuations are observed is quite different in tb@xmath35ti@xmath35o@xmath36 as compared with tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 . in the present case
, the broadening of the bragg peaks appears as critical phenomena and the strongest fluctuation effects fall off within a few degrees of t@xmath4 . in contrast , the jahn - teller - like fluctuations in tb@xmath35ti@xmath35o@xmath36 occur coincident with the formation of a spin liquid state , and are measureable over an extended regime of temperature from 20 k to 0.3 k. finally we return to the issue of mean field criticality vs criticality associated with random fields .
we did not observe a departure from mean field criticality associated with random fields in tbv@xmath18as@xmath19o@xmath2 .
this may be because the random fields associated with the disorder are too weak to influence the critical behaviour in the range of reduced temperatures that we could access .
a measure of the strength of the disorder and associated random fields is the extent to which they depress the transition temperature .
the observed t@xmath4 for our mixed sample was 30.32 k , lower than the value 32.25 k expected from a linear variation in t@xmath4 between the two end members .
this depression in t@xmath4 confirms that disorder is relevant and that associated random fields are present . in some related systems that have been studied , dyas@xmath1v@xmath0o@xmath2@xcite and kh@xmath14as@xmath1p@xmath0o@xmath2@xcite , however , as well as dilute antiferromagnetic systems@xcite the depressions in t@xmath4 were larger and random - field critical behaviour
was clearly observed .
we have studied the critical properties near jahn - teller , tetragonal to orthorhombic structural phase transitions in tbvo@xmath2 , tbaso@xmath2 and tbv@xmath18as@xmath19o@xmath2 using high resolution x - ray scattering .
these results show the order parameter , the orthorhombic strain , in all three materials to exhibit mean field criticality .
the disordered member of this series , tbv@xmath18as@xmath19o@xmath2 is distinguished only in so far as it exhibits a wide dynamic range of majority and minority orthorhombic twin domain orientations , and these are stable to low temperatures .
we ascribe this to the pinning of the domain structure by random strains arising from the quenched disorder in tbv@xmath18as@xmath19o@xmath2 .
we also observe very interesting fluctuation effects in tbvo@xmath2 and tbaso@xmath2 in the vicinity of t@xmath4 .
these are manifest in longitudinal broadening of the ( 6 , 6 , 0 ) bragg peaks in the tetragonal phase , which split to become ( 12 , 0 , 0 ) and ( 0 , 12 , 0 ) bragg peaks in the orthorhombic low temperature phase .
this broadening exhibits a power law divergence at t@xmath4 , in tbvo@xmath2 and tbaso@xmath2 , with a critical exponent , x=0.45 @xmath30 0.04 . we hope that this latter work informs and motivates a full theoertical understanding of this fluctuation phenomena .
hutchings , r. scherm , s.h .
smith and s.r.p .
smith , j. phys .
c : solid state physics * 8 * l393 ( 1975 ) .
elliot , r.t .
harley , w. hayes , s.r.p .
smith , proc .
* a328 * 217 ( 1972 ) . g. a. gehring and k. a. gehring , reports on progress in physics * 38 * 1 ( 1975 ) .
w. sch@xmath37fer and g. will , acta cryst . *
b35 * 588 ( 1978 ) .
r. t. harley , r. m. macfarlane , j. phys .
c : solid state physics .
* 8 * l451 ( 1975 ) . g. will , h. gobel , c.f .
sampson , j.b .
forsyth , phys . lett . * 38a * 207 ( 1972 ) .
smith and b. k. tanner , j. phys .
c : solid state phys . * 11 * l717 ( 1978 ) .
y. hirano , n. wakabayashi , c .- k .
loong and l.a .
boatner , phys .
b * 67 * 014423 ( 2003 )
choo , h.p .
schriemer and d.r .
taylor , phys .
b * 61 * 11197 ( 2000 ) .
y. imry and s.k .
ma , phys .
lett . * 35 * 1399 ( 1975 ) .
belanger , in spin glasses and random fields , edited by a. p. young ( world scientific , singapore ) , p. 251
( 1998 ) g. jasiolek and h.a .
dabkowska , j. crystal growth * 79 * 534 ( 1986 ) .
p. hagenmuller , j. less common metals * 110 * 225 ( 1985 ) .
schriemer , c .- h .
choo and d.r .
taylor , phys .
b * 59 * 8351 ( 1999 ) .
w. berkhahn , h.g .
kahle , l. klein abd h.c .
schopper , phys .
b , * 55 * 265 ( 1973 ) .
d. vaknin et al .
lett . * 58 * , 2802 ( 1987 ) ; r.j .
birgeneau et al . , phys .
lett . * 59 * , 1329 ( 1987 ) ; p. boni et al .
b * 38 * , 185 ( 1988 ) ; t.r .
thurston et al .
b * 39 * , 4327 ( 1989 ) .
y. zhao , b.d .
gaulin , j.p .
castellan , j.p.c .
ruff , s.r .
dunsiger , g.d .
gu , and h.a .
dabkowska , phys .
b * 76 * , 184121 ( 2007 )
le guillou and j. zinn - justin , phys .
. lett . * 39 * , 95 ( 1977 ) ; j.c .
le guillou and j. zinn - justin , phys .
b * 21 * , 3976 ( 1980 ) .
ruff , b.d .
gaulin , j.p .
castellan , k.c .
rule , j.p .
clancy , j. rodriguez , and h.a .
dabkowska , phys .
lett . * 99 * , 237202 ( 2007 ) .
graham , j.h .
page , d.r .
taylor , phys .
b * 44 * 4127 ( 1991 ) .
k.a . reza and d.r .
taylor , phys .
b * 46 * 11425 ( 1992 ) .
taylor , j. t. love , g. j. topping , and j. g. a. dane , phys .
b * 72 * , 052109 ( 2005 ) . | the critical behaviour associated with cooperative jahn - teller phase transitions in tbv@xmath0as@xmath1o@xmath2 ( where _ x
_ = 0 , 0.17 , 1 ) single crystals have been studied using high resolution x - ray scattering .
these materials undergo continuous tetragonal @xmath3 orthorhombic structural phase transitions driven by jahn - teller physics at t@xmath4 = 33.26(2 ) k , 30.32(2 ) k and 27.30(2 ) k for _ x _ = 0 , 0.17 and 1 respectively .
the orthorhombic strain was measured close to the phase transition and is shown to display mean field behavior in all three samples .
pronounced fluctuation effects are manifest in the longitudinal width of the bragg scattering , which diverges as a power law , with an exponent given by @xmath5 , on approaching the transition from either above or below . all samples exhibited twinning ; however the disordered x = 0.17 sample showed a broad distribution of twins which were stable to relatively low temperatures , well below t@xmath4 .
this indicates that while the orthorhombic strain continues to develop in a conventional mean field manner in the presence of disorder , twin domains are easily pinned by the quenched impurities and their associated random strains . |
in contrast to the pion - nucleon interaction , the @xmath0-nucleon interaction at small momenta is attractive and rather strong .
this attraction can be seen from the fact that the @xmath0 threshold ( 1488 mev ) is situated just below the @xmath11 resonance which couples strongly to the @xmath12 channel .
initial calculations by bhalerao and liu@xcite obtained attractive s - wave @xmath12 scattering lengths . with these values , haider and liu@xcite have shown that @xmath0 can be bound in nuclei with a @xmath13 10 .
other groups have also found similar results @xcite . on the experimental side results
are meager .
the first experiments searching for @xmath0-mesic nuclei at bnl@xcite and lampf@xcite by using a missing - mass technique in the ( @xmath14 ) reaction reached negative or inconclusive results .
later it became clear that the peaks are not necessarily narrow and that a better strategy of searching for @xmath0-nuclei is required .
furthermore , the bnl experiment was in a region far from the recoilless kinematics , so that the cross section is substantially reduced@xcite .
more recently , the existence of @xmath0-mesic @xmath15 was claimed to have been observed in the reaction @xmath16 using the photon beam at mami @xcite .
it has , however , been pointed out@xcite that the data of ref.@xcite does not permit an unambiguous determination of the existence of a @xmath17-bound state . the suggestion that @xmath17 is not bound is also supported by the theoretical studies of refs.@xcite
two different methods have been applied in the search for @xmath0 bound states .
one is the study of @xmath0 production on nuclei and extraction of the @xmath0 nucleus scattering length .
the other is the direct production of the @xmath0 meson in a bound state .
this state is measured via missing mass technique .
the gem collaboration has used both methods , as will be shown in this paper .
according to the watson - migdal theory@xcite , when there is a weak transition to a system where there is a strong final state interaction ( _ fsi _ ) , one can factorize the @xmath18-wave reaction amplitude , @xmath19 , near threshold in the form @xmath20 where @xmath21 is the @xmath0 c.m . momentum , @xmath22 the complex scattering length and @xmath23 the effective range . the unperturbed production amplitude @xmath24 is assumed to be slowly varying and is often taken to be constant in the near - threshold region .
unitarity demands that the imaginary part of the scattering length be positive , _
i.e. _ , @xmath25 .
in addition , to have binding , there must be a pole in the negative energy half - plane , which requires that@xcite @xmath26 finally , in order that the pole lie on the bound- rather than the virtual - state plane , one needs also @xmath27 .
recently , two different experiments at cosy jlich measured @xmath0 production in @xmath28 reactions very close to threshold with an extremely high precision of the data@xcite .
the latter authors folded out the experimental resolution and got @xmath3{\rm{fm}}$ ] and @xmath29{\rm{fm}}$ ] for the effective range .
since the data are not sensitive to the sign of the real part of the scattering length , the quest for a bound state or an unbound pole can not be answered .
its value is @xmath30 from the model calculations it is known that binding is more probable for heavier nuclei than for lighter nuclei .
we therefore can expect the relation latexmath:[\[\label{equ : relation } study of the fsi of the @xmath0 @xmath32 system .
this is produced in the reaction @xmath33 the existing data before the gem measurement were not sufficient to extract the s - wave contribution of the cross section . in order to do
so gem made use of a tensor polarized deuteron beam@xcite .
the beam momentum of 2385.5 mev / c corresponds to an excess energy of @xmath34 mev for this reaction when an @xmath0-meson mass of @xmath35 mev/@xmath36 is used @xcite .
the big karl magnetic spectrograph @xcite employed for this study is equipped with two sets of multi - wire drift chambers ( mwdc ) for position measurement and thus track reconstruction .
two layers of scintillating hodoscopes , 4.5 m apart , led to a more accurate time - of - flight measurement than previously achieved with big karl .
they also provided the energy - loss information necessary for particle identification .
more information on the experiment are given in ref.@xcite .
in a first step the unpolarized cross section was measured .
left upper panel : differential cross section for the @xmath37 reaction .
left lower panel : analyzing power @xmath38 .
the solid curves represent a fit with four partial waves ; the dotted curves with invariant amplitudes .
right : excitation function for the total cross section for the @xmath39 reaction .
only statistical errors are shown .
, title="fig:",scaledwidth=48.0% ] left upper panel : differential cross section for the @xmath37 reaction . left lower panel : analyzing power @xmath38 .
the solid curves represent a fit with four partial waves ; the dotted curves with invariant amplitudes .
right : excitation function for the total cross section for the @xmath39 reaction .
only statistical errors are shown .
, title="fig:",scaledwidth=48.0% ] the obtained angular distribution is shown in fig .
[ fig : partial_wave_fit ] .
it can be fitted by legendre polynomials . because of the symmetric entrance channel only even polynomials contribute .
three parameters are sufficient to reproduce the data .
this indicates that there must be at least @xmath40-wave contributions .
this is to be contrasted to the lower energy anke results @xcite , where @xmath41 suffices . from the fit total cross section of @xmath42 nb , where uncertainties in the target thickness , incident flux , and acceptance introduce an additional systematic error of @xmath43 nb .
this value is shown in fig .
[ fig : partial_wave_fit ] together with the world data @xcite , @xcite , and @xcite .
it seems that the cross sections start to saturate for @xmath0 momenta above 80 mev / c .
the next step is to extract the s - wave part of the total cross section .
for this task the knowledge of polarization observables is necessary .
first the polarization of the beam @xmath44 was measured by measuring elastic backward scattering of the deuterons on protons . from the known analyzing power @xmath45
the beam polarization is obtained . in the measurement of analyzing powers
one usually compares polarized and unpolarized cross sections .
this introduces ambiguities because of different luminosity measurements . in order to avoid this
we applied another method .
we integrated the cross section over intervals in the polar angle , where on one hand we have full geometrical acceptance of the apparatus and @xmath45 practically vanishes .
we measure then only @xmath38 .
the cross section integrated over these intervals of azimuthal angle becomes simply @xmath46 , \label{dsigma_rho_axx}\end{gathered}\ ] ] where the unpolarized cross section is integrated over the same @xmath47 range .
carrying out this procedure for the two polarization states , we find that @xmath48 and hence @xmath49 the angular distribution thus obtained is also shown in fig .
[ fig : partial_wave_fit ] . from both angular distributions
we extract partial waves , one s - wave , one p - wave and two d - waves .
this yields seven parameters to be fitted
. however , strong correlations were found .
an alternative method is to employ the spin structure of the matrix element . due to the symmetry in the entrance channel three independent scalar amplitudes
are necessary to describe the spin dependence of the reaction : @xmath50 . @xmath51 and @xmath52 have no angular dependence and for @xmath53 we assume the expansion @xmath54 .
fortunately the dependencies can be decoupled @xmath55 we now extract the magnitude of the @xmath18-wave amplitude @xmath56 . from this
we obtain a spin - averaged square of the @xmath18-wave amplitude , @xmath57 through @xmath58 wroska et al.@xcite could not distinguish whether the angular distribution is due to a @xmath59 interference or a @xmath40-wave .
this puzzle can be solved by our result here .
the willis et al.@xcite data for s - wave were extracted from the measurement by just dividing by @xmath60 .
this is , however , not justified in the region of their highest point . assuming a dependence of the @xmath40-waves with @xmath61 we can extract a more correct value for the s - wave strength . from all measurements
we then obtain a scattering length of @xmath62{\rm{fm}}.\ ] ] this result can be converted into a pole position @xmath63 since the present results fulfill the conditions for a bound state - except for the sign of the real part of the scattering length - and the relation [ equ : relation ] , it may well be that @xmath64 binding is observed . for an even heavier system we may expect stronger binding .
gem has therefore studied @xmath0 production in an almost exclusive reaction @xmath65be , by measuring the recoiling be ions . since the energy of the ions is rather small , they are strongly ionizing particles and new detectors have been applied .
details are given in @xcite .
since all states of @xmath66be above 1.59 mev are particle unstable , only the two lowest states with @xmath67 contribute .
be in its ground state or first excited state .
the solid curve is phase space behavior fitted to the present measurement.,scaledwidth=50.0% ] in fig .
[ fig : pli ] the new data point at 11.3 mev excess energy is shown together with a point earlier obtained by @xcite and corrected for @xmath68 contributions .
the data can obviously be accounted for by phase space .
more data closer to threshold are necessary to see fsi effects .
the following method was successfully employed in the production of hypernuclei and pionic atoms .
a nucleon is replaced by a hyperon or a pion .
maximal cross section is obtained when the momentum transfer from the projectile to the particle to be implanted is minimized , which means that the produced particle is almost at rest in the laboratory system . in pionic atom studies a proton transfer via the @xmath69 reaction was successfully applied@xcite . however , although this reaction has reasonably large cross section it also has the disadvantage of a huge background due to the break - up of the deuterons .
the resulting protons have the beam velocity and thus the same magnetic rigidity as the @xmath15 particles .
this problem can be overcome by making use of a two - nucleon transfer reaction like @xmath70 but at the expense of a much smaller cross section .
the choice of an odd - odd target nucleus would be ideal in order to avoid nuclear excitations as much as possible .
since there exist no solid material with these properties @xmath71al was chosen as a compromise .
the experiment was done by making use of the following reaction chain : @xmath72 where the @xmath73 carries the beam momentum away .
the unobserved system @xmath6 is therefore at rest .
one of the possibilities is @xmath74 .
since the @xmath0 is also at rest it can undergo a second chain of reactions @xmath75 the final fate of that chain is the decay to a pion and a nucleon . in the case of a neutral resonance
the final state can be @xmath76 , which have to be emitted almost back to back , if we ignore fermi motion .
the @xmath73 is detected with the magnetic spectrograph big karl@xcite .
for the detection of the two decay particles a dedicated detector enstar was built .
it has cylindrical shape around the target and consists of three layers of scintillating material .
the layers were subdivided into pieces making polar and azimuthal angle measurements possible . the read - out of the pieces
was performed with scintillating fibres@xcite .
al@xmath77he@xmath78 at recoil free condition for an @xmath0.,scaledwidth=50.0% ] the experiment was performed at a beam momentum of 1745 mev / c . at this momentum final states with small binding energies can be produced with small momentum transfer .
states with binding energies from 0 mev to -30 mev will have momenta below 30 mev / c . two settings of the spectrograph were used with a large overlap . excellent particle identification in big karl
was achieved by measuring energy loss in the start detector of the tof facility in the focal plane and the tof .
the identification of a back to back proton and pion with momenta corresponding to a @xmath79(1535 ) resonance is mandatory to see structure in the missing mass spectra .
the spectra for the two settings are shown in fig .
[ fig : two_settings ] .
the shaded area may correspond to an @xmath0 bound state .
the data in the lower panel contain the unbound region with quasi - free @xmath0 production .
more details can be found in @xcite .
the two spectra were then combined .
counts were transformed into differential cross sections by making use of the integrated luminosity and detector acceptance as well as its efficiency .
missing mass spectrum converted to binding energy @xmath80 of a bound system @xmath81 as measured with the magnetic spectrograph .
all cuts and background subtraction have been applied .
the solid curve is a fit with a constant background , two gaussians and a phase space behavior for the unbound system .
dotted curve : same as solid curve but without phase space contribution .
, width=302 ] these cross sections are shown in figure [ fig : binding - spectra ] . here we have converted the missing mass scale into binding energy @xmath80 .
the data show an enhancement around @xmath82 mev .
the significance of this structure is extracted according to the two methods given in @xcite . at first
, we test the hypothesis of peak structure being a fluctuation of background , i.e. the origin of the background is taken to be independent of the signal .
the background outside the peak region , for simplicity approximated by a constant , was found to be [email protected] .
the significance is then given by @xmath83 where @xmath84 is the total counts in the region of interest , @xmath85 is the total background in this region as determined from the fit to the outside region and @xmath86 is error in the estimation of background value as taken from the fit .
this yields a value of significance which is 5.3@xmath10 .
here we have assumed gaussian errors .
for the assumption of poisson errors the background is [email protected] .
this larger value is typical for poisson distribution and hence the significance reduces to 4.9@xmath10 .
finally a gaussian on top of the background was fitted to the whole data set .
this yielded for the case of poisson statistics [email protected] for the background , [email protected] for the amplitude , [email protected] mev for the centroid and [email protected] mev for the width . in the second method ,
the statistical significance is extracted by assuming the background events as well as the peak events on top of the background being poisson distributed . again
a constant background and a gaussian was assumed . in this way
, we obtain a value of 6.20@xmath10 for the significance , assuming a simultaneous determination of amplitude , centroid and width of the signal .
the fit gives for the centroid @xmath87 mev and for the width @xmath88 mev .
these results compare favorably with those from the first method .
we , therefore , consider the present experimental results to provide a strong hint of a nuclear @xmath0 bound state .
in the following we take a different view on the data than in @xcite
. it may well be that in addition to a @xmath89 state a @xmath90 state is bound if the scattering length @xmath91 is large .
the former state is stronger bound than the later and it is expected to have a larger width than the later @xcite .
we therefore fit two gaussians on a constant background to the data .
the quasi - free region is represented by phase space behavior .
.fitted parameters of two gaussians to the data and their hypothetical assignment . [
cols="^,^,^,^",options="header " , ] the constant background was found to be 0.25 nb / sr mev and the fit parameters of the gaussians are given in table [ tab:1 ] .
the resulting curve is also shown in fig .
[ fig : binding - spectra ] .
the hypothetical @xmath89 state has a very large width and reaches into the unbound region .
however , the evidence for this peak is less than the one for the peak at 12 mev . due to the low statistics of the data , which is a result of the small cross section , the uncertainties in the fitted parameters are large .
we have measured @xmath0 production in two body final state by @xmath92 and @xmath93li@xmath94be reactions close to threshold . for both reactions angular distributions
were obtained which allow to extract total cross sections . because of making use of a tensor polarized deuteron beam an average matrix element for the s - wave could be extracted and thus allows the determination of the scattering length . from the data analysis it is found that @xmath21-waves are absent .
the data situation in the case of the beryllium final state is insufficient to draw conclusions .
we have furthermore measured @xmath95al@xmath96he@xmath97x at kinematical conditions where an @xmath0 in the elementary reaction @xmath98he is at rest .
the missing mass spectrum shows a significant peak .
there is evidence that the @xmath0 may be bound in a @xmath89 and a @xmath90 state .
however , better statistics data are mandatory in order to identify the nature of the missing mass spectrum . | the @xmath0 meson can be bound to atomic nuclei .
experimental search is discussed in the form of final state interaction for the reactions @xmath1 and @xmath2 . for the latter case tensor
polarized deuterons were used in order to extract the s - wave strength . for both reactions complex scattering lengths
are deduced : @xmath3 $ ] fm and @xmath4 $ ] fm . in a two - nucleon transfer reaction under quasi - free conditions , @xmath5 , was investigated .
the system @xmath6 can be the bound @xmath7 at rest .
when a possible decay of an intermediate @xmath8 is required , a highly significant bump shows up in the missing mass spectrum .
the data give for a bound state a binding energy of [email protected] mev and a width of @[email protected] mev . |
blazars vary violently at all wavelengths , with timescales as short as hours days .
this suggests that the injection mechanism and/or the cooling processes operate on timescales shorter than the light crossing time @xmath0 .
this implies simmetric light curves during rapid flares , as observed during particularly intensive monitoring campaigns . to interpret in detail the variability pattern at different wavelenghts we need to study the time dependent behavior of the emitting particle distribution
* the source , of typical dimension @xmath1 , embedded by a tangled magnetic field @xmath2 , moves relativistically , and the radiation is beamed with a doppler factor @xmath3 .
* relativistic electrons are injected homogeneously throughout the source for a time which can be less than @xmath0 .
* we consider synchrotron and self compton cooling , and particle escape . *
the electron distribution is found by solving the continuity equation : @xmath4 + q(\gamma)- { n(\gamma , t ) \over t_{esc}}\ ] ] where @xmath5 is the injection term , @xmath6 is the cooling term , and @xmath7 is the escape timescale of the particles , assumed to be independent of their energy .
we solve this equation numerically , according to the scheme proposed by chang & cooper ( 1970 ) .
since the cooling and injection timescales are shorter than @xmath0 , the particle distribution evolves more rapidly than the light crossing time .
_ the observer will see a convolution of different spectra , each produced in a different region of the source_. initially we see only the emission by fresh electrons located in the slice nearest to us .
after a time @xmath0 we see the entire source : the back of it with fresh electrons , and the front of it with older electrons .
in may 1994 mkn 421 underwent a x ray flare during an high state of the tev emission ( macomb et al .
1995 ) . fig .
1 shows the overall spectra taken from macomb et al .
( 1996 ) fitted by two ssc spectra calculated with our program assuming that the particle distribution reached equilibrium .
the two models differ only by the total injected power ( factor 2 ) and the @xmath8 ( factor 3 ) .
takahashi et al . ( 1996 ) studied the time lag between hard and soft x
rays ( asca data ) , finding a time delay of @xmath91 hour .
the also note that the tev flux varied with approximately the same amplitude of the x rays , while the optical flux remained quasi constant .
we qualitatively reproduce these features by assuming : i ) injection of a flat power electron distribution ( @xmath10 ) between @xmath11 and @xmath12 for a time equal to @xmath0 ; ii ) @xmath13 cm , @xmath14 g , @xmath15 , @xmath16 erg / s ; iii ) the flaring emission corresponding to the fast varying electron distribution is summed to a constant component .
2 shows the light curves at four different frequencies : note that the soft x rays lag the hard x
rays ( approximately 1 hour ) , and the optical flux remains quasi constant .
the peak of the optical emission lags the hard x
rays by @xmath93 hours , while the tev peak has a delay of 2 hours .
the seed photon for the tev emission have ir frequencies , and they lag the x rays : this is the reason for the delay of the tev emission w.r.t .
the x rays . | we calculate the time dependent electron distribution under the assumption of continouos injection of new , relativistic particles , and assuming radiative cooling ( by synchrotron and inverse compton ) and particle escape .
resulting photon spectra are calculated taking into account the time delays introduced by the different light travel times across the source .
we apply these calculations to the varying x ray spectrum of mkn 421 .
# 1 # 2 _ mem .
soc .
astron .
it . _ * # 1 * , # 2 # 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 _ astron .
astrophys .
suppl .
ser . _ * # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 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
. soc
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 epsf.sty |
mrk 273 is an ultraluminous iras galaxy ( ulirg ) at redshift @xmath14 .
it is a merging galaxy that shows a striking long tidal tail ( @xmath15 ) to the south , and a large tidal plume to the northeast ( see fig .
[ fig : halo ] ) .
b - band images uncover diffuse clumps in the northeast plume . these plumes exhibit unusual optical spectra that are consistent with excitations by the shock plus precursor mechanism ( dopita & sutherland 1995 ; xia et al .
deep narrow - band @xmath16 and @xmath8 images ( armus et al . 1990 ) reveal filaments and arc structures in the northeast plume and patchy areas of ionized gas ( extending tens of kpc ) along the southern tidal tail .
these observations of the extended structures provide evidence for shock excitations produced during merging .
the nuclear region of mrk 273 is extremely complex .
it shows double nuclei from both ground based images and hst optical ( wfpc2 ) , k - band ( nicmos ) images ( knapen et al .
1998 ; soifer et al .
2000 ; carilli & taylor 2000 ) .
the projected separation between the two nuclei in the k - band is @xmath17 .
the @xmath18 and @xmath8 maps obtained using integral field spectroscopy of mrk 273 show that there are two distinct emission regions separated by @xmath7 ( colina et al .
the northeastern region coincides with the optical and k - band nuclei ; it is strong in the @xmath18 emission but weak in the @xmath8 emission , i.e. , it exhibits liner spectral characteristics . on the other hand , the southwestern region is dominated by diffuse @xmath8 emission and colina et al .
( 1999 ) identified this region as a seyfert 2 nebula .
radio observations ( condon et al . 1991 ; knapen et al .
1997 ) show that there are three components : the northern and southwestern components are coincident with the two k - band nuclei respectively while the southeastern component has a faint blue optical counterpart which was identified as a star - cluster by scoville et al .
high resolution radio continuum and neutral hydrogen 21 cm absorption observations by merlin , very large baseline array , and very large array ( cole et al .
1999 ; carilli & taylor 2000 ) reveal a gas disk associated with the northern nucleus .
the disk has a diameter of @xmath19 with an inclination angle of @xmath20 and an average neutral hydrogen column density @xmath21 .
furthermore , the high resolution co(2 - 1 ) map by downes & solomon ( 1998 ) uncovers a bright @xmath22 co core in the nuclear molecular disk of the northern nucleus of mrk 273 .
this is the most luminous extreme compact starburst region for local ulirgs with an infrared luminosity of @xmath23 and a molecular mass of @xmath24 . in short ,
the mrk 273 nuclear region is very complex and it is not even clear how many different components there are , let alone whether various components are powered by agn and/or starbursts .
x - ray observations provide an independent probe of the physical processes in merging galaxies such as mrk 273 .
asca observation shows that there is a heavily obscured active nucleus in mrk 273 ( turner et al .
1997 , 1998 ; iwasawa 1999 ) .
hard x - ray emissions above 3kev and a narrow 6.4kev @xmath0 line were detected .
an absorbed power - law model provides an acceptable fit to the spectrum although a reflection - dominated model can not be rejected either , based on the asca data ( iwasawa 1999 ) .
however , given that asca s spatial resolution is @xmath25 , the position of the agn can not be pinpointed ; this is particularly serious because in the field of mrk 273 , rosat hri images show that there is a background x - ray source ( mrk 273x at redshift @xmath12 ) about @xmath26 away ( xia et al .
so the asca data include contributions from both mrk 273 and mrk 273x
. the sub - arcsecond resolution and high - energy sensitivity of chandra are therefore essential for understanding the nature of x - ray emissions in the mrk 273 field .
chandra observations are interesting for yet another important reason .
elliptical galaxies are known to have hot gas halos ; however , the origin of the hot gas is still under debate ( osullivan et al .
2001 ; sansom , hibbard & schweizer 2000 ) .
there is now firm evidence supporting the merger scenario for the formation of elliptical galaxies , through major and/or multiple merger ( e.g. , hernquist et al .
1996 ; borne 2000 ; bekki 2001 ; cui et al .
2001 and references therein ) .
based on the asca data for 4 ulirgs ( mrk231 , mrk273 , arp220 and ngc6240 ) , iwasawa ( 1999 ) pointed out that the soft x - ray emissions of these ulirgs are thermal with a temperature of @xmath27 k. incorporating the extended morphology of soft x - ray emissions from ngc 6240 , arp 220 and ngc 3690 observed by rosat hri , a thermal origin of the soft x - ray emission is acceptable .
furthermore , two - component fits to the soft x - ray data are better than one - component fits for these four ulirgs and the low temperature component is more extended than the high temperature component ( iwasawa 1999 ) .
a plausible explanation is that the high temperature component is from central starbursts and/or agns while the more extended low - temperature component is provided by the shock - heated gas produced in mergers .
a high - resolution soft x - ray image is therefore necessary to disentangle the x - ray emission from different parts of merging galaxies . for the reasons described above , we have observed the mrk 273 field using the acis - s3 instrument on chandra
; in this paper we report the results of these observations .
our focus is on mrk 273 due to its implications for galaxy merging and formation , although we do briefly discuss the nature of the background source mrk 273x ( see 5 ) .
the structure of this paper is as follows . in 2 , we describe the observations and data reduction . in 3 , we present the images for mrk 273 in both the soft and hard x - ray band .
these images reveal a compact hard x - ray nucleus and several soft x - ray clumps accompanied by a very extended hot halo . in 4 , we study the spatially - resolved spectral behaviors for the nuclear and extended x - ray emissions in mrk 273 . in 5 , we discuss the x - ray results for mrk 273x from the chandra observations . and
finally in 6 , we summarize and discuss our results , particularly concerning the nature of the agn in mrk 273 and the connection to the origin of hot gas in elliptical galaxies . throughout this paper
, we use a hubble constant of @xmath3 and an einstein - de sitter ( @xmath28 ) cosmology although the latter has little influence on our results due to the low - redshift ( @xmath29 ) of mrk 273 . at the redshift of mrk 273
, @xmath30 corresponds to 1.03kpc for the adopted cosmology .
mrk 273 and mrk 273x were observed using the chandra acis - s3 ccd chip on 19 april 2000 , with a total exposure time of @xmath31 47ks .
the observation was performed in the very faint ( vfaint ) mode with acis working at the focal temperature of @xmath32c .
these two objects separated by @xmath31 @xmath26 were positioned on the same chip with offsets of @xmath33 and @xmath34 from the acis - s aim - point , respectively .
the data were initially processed with the pipeline sdp version r4cu5upd2 , available when the data were first available to us ; this pipeline has an astrometric error of about @xmath35 .
the data used in this paper were reprocessed by the chandra team using the new version of the software ( sdp r4cu5upd13.2 ) in january 2001 .
the most updated calibration files ( such as acis qeu and gain map as well as observation - specific bad pixel lists ) were used in the new data products .
the data were also filtered to include only standard event grades 2 , 3 , 4 and 6 ( see chandra proposer s observatory guide , 2000 ) .
the improvements in the processing software and calibrations lead to an astrometric error of about @xmath36 . however , this position accuracy is still insufficient to align various nuclear components discovered in the optical , infrared and x - ray ( see 1 ) . by carefully examining both the chandra image and a wfpc2 hst image of mrk 273
, we found three point sources ( shown in fig .
[ fig : smoothed ] as a [ i.e. , mrk 273x ] , b , c ) that are present in both sets of images . these point sources allow us to determine the astrometry in the mrk 273 field to a much higher accuracy , @xmath37 ; this accuracy is crucial to register the x - ray nuclear components with those in other wavebands .
the pre - processed and astrometrically - aligned data products were then analyzed with the standard cxc ciao package ( version 2.0 ) .
it begins with an inspection of bad aspect times and high background times .
a background flare event was detected at the beginning of the observation ; the corresponding time intervals with large background rate ( i.e. 20% over the quiescent value ) were thus removed , yielding an effective exposure time of @xmath38 ks in the energy range 0.310kev .
all the followup spatial and spectral analysis will be restricted to this energy band ; the april 2000 release of the acis ccd calibration files ( fef ) was used .
as the standard ciao tasks do not produce arf and rmf for an extended source , we also made use of the contributed software , calcrmf , for this purpose .
[ fig : raw ] shows the broad band ( 0.310kev ) raw image of mrk 273 and mrk 273x at the full resolution ( 0.49@xmath39/pixel ) of chandra acis - s3 .
it is clear from fig .
[ fig : raw ] that the x - ray structure of mrk 273 is complex with a bright nuclear region and very extended diffuse emission . fig .
[ fig : smoothed ] displays a true color image of mrk 273 , produced by a mosaic of three x - ray bands , namely , red : 0.31kev , green : 13kev and blue : 310kev .
images in each of the three bands were adaptively smoothed using the ciao task _ csmooth _ , at 3@xmath40 significant levels .
the smoothing function we used is a scale - variable two - dimensional gaussian with a maximum scale of 10 pixels ( @xmath41 ) to avoid inducing any over - smoothing effects .
the inset at the bottom left corner is for the central @xmath42 nuclear region .
the complex x - ray emission structures are clearer in this picture .
it is obvious from fig .
[ fig : smoothed ] that the hard x - ray ( 3 - 10kev ) emission for mrk 273 is compact ( the white central region ) while the soft x - ray emission ( 0.31kev ) consists of diffuse halos and bright complex structures
. furthermore , there are obvious intermediate x - ray band emissions surrounding the hard compact region shown as yellow in fig .
[ fig : smoothed ] .
we will discuss the possible origin for these emissions in 6 .
[ fig : energyevolution ] shows more clearly the evolution of the nuclear region of mrk 273 with the x - ray energy . as can be seen from this figure , the hard x - ray ( @xmath43 3kev ) emission originates from a compact region while the soft x - ray emission is diffuse with some bright clumps .
the dominant soft x - ray bright clumps are in the northeast and southwest .
as x - rays become harder , the northeastern clumps become brighter while the southwestern clumps become fainter .
this indicates that the northeastern clumps are harder than the southwestern clumps .
note that there is also a soft x - ray faint region in the northwest direction which can be clearly seen in the 0.6 - 0.8kev and 0.8 - 1.1kev panels .
[ fig : nuclearregion ] is an adaptively smoothed 0.31kev soft x - ray image of the nuclear region .
four concentric circles indicate the @xmath44 and @xmath4 regions around the central nucleus .
the square at the center of circles indicate the position of the compact hard x - ray source .
it is clear that there are 4 soft x - ray bright clumps , labelled as n , se , sw1 and sw2 , surrounding the hard x - ray source .
the three plus signs in fig .
[ fig : nuclearregion ] indicate the three resolved radio components as given by knapen et al .
table 1 lists the positions of the hard x - ray component and soft x - ray bright clumps together with the positions of the three radio components .
the absence of soft x - ray emission in the hard x - ray compact region indicates that most soft x - ray emissions in the nucleus of mrk 273 are absorbed .
it is obvious from fig .
[ fig : nuclearregion ] and table 1 that the hard x - ray position is coincident with the northern radio component within the chandra astrometric uncertainty ( @xmath45 ) .
this northern component also has counterparts in the k - band , mid - infrared and @xmath16 ( @xmath18 ) narrow band images .
the compact hard x - ray source is also likely to be embedded in a very dense , bright core of molecular gas ( about @xmath46 ) at the center of this nucleus ( downes & solomon 1998 ) .
there is little doubt that the northern radio component of mrk 273 hosts an agn .
the optical spectrum of this nucleus is of liner - type ( colina et al .
1999 ; for more see 5 ) . on the other hand , the southwestern and southeastern radio components do not have any hard x - ray counterparts .
the southwestern radio component is coincident with the southwestern k - band nucleus , which has been proposed to be the second nucleus of mrk 273 .
however , the lack of hard x - ray emissions makes the interpretation less clear .
the left and right panels of fig .
[ fig : oiiihalpha ] shows the 0.31kev soft x - ray contours overlaying on @xmath8 and h@xmath6 narrow band image for nuclear region of mrk 273 , respectively .
the @xmath8 and h@xmath6 narrow band image are obtained by the 2.16 m telescope of beijing astronomical observatory .
the circle in the left and right panel of fig .
[ fig : oiiihalpha ] indicates the hard x - ray position .
it is clear from this figure that the southwestern soft x - ray bright clumps ( sw1 and sw2 ) are coincident with @xmath8 nebula and the hard x - ray source is within the northeastern h@xmath6 bright clump , which is identified as the northern nucleus in multi - wavelengths ( see 1 ) .
in fact , from table 1 the angular distance between the hard x - ray position and the sw2 component is about @xmath7 , the same as the angular distance between the northern nucleus and southwestern ` seyfert 2 ' nebula in colina et al .
( 1999 ) , which is dominated by the @xmath8 line emission .
the adaptively smoothed 0.31kev soft x - ray image of mrk 273 is shown in the left panel of fig .
[ fig : halo ] .
the maximum smoothing scale is @xmath47 ( 10 pixels ) and the contours shown in the middle and right panels ( see below ) are 3 , 5 , 10 , 20 , 30 , 50 , 100 and 300@xmath40 significance levels respectively .
analysis indicates that most statistically significant photons are inside in the outer ellipse in the left panel of fig .
[ fig : halo ] and its major and minor radius are @xmath48 and @xmath49 , respectively .
these correspond to a physical diameter of @xmath50 .
this is the largest extended soft x - ray halo that has been identified in an ultraluminous iras galaxy .
furthermore , as we have shown , the high - resolution observation of chandra allows us to compare fine structures with optical images at comparable resolution .
the middle panel of fig .
[ fig : halo ] shows a deep r - band image of mrk 273 ( armus et al .
1990 ) overlaid with the soft x - ray contours .
it is obvious that the soft x - ray halo encompasses the whole mrk 273 optical image , including the southern long tidal tail and northeast plume . in the right panel of fig .
[ fig : halo ] , the soft x - ray contours are overlayed on an @xmath16 narrow band image of mrk 273 obtained using the 2.16 m telescope of beijing astronomical observatory .
we can clearly see from this plot that the southern and northeastern soft x - ray structure are almost the same as the @xmath16 structure .
curiously , while the x - ray extension seems agree well with the @xmath16 extension , it shows a slight offset to the west of the optical tidal tail .
this may be a natural consequence of the different behavior of collisionless stars and collisional gas ( outflow ) during galaxy merging as seen in arp 299 ( hibbard & yun 1999 ) .
we will return to the origin of the extended halo in 6.2 .
given the sub - arcsecond resolution of chandra , we can perform not only global but also spatially - resolved spectral analysis .
the latter is particularly important for understanding the relative contribution of the agn / starburst contributions to the x - ray emission , and by implication the origin of hot gas in the descendents of mergers elliptical galaxies .
high spatial and spectral resolution of chandra also allow one to separate , for the first time , the contributions to the hard x - ray emissions from mrk 273 and mrk 273x ( see fig . [ fig : raw ] ) ; this was impossible for asca due to its limited spatial resolution . from figs .
[ fig : energyevolution ] and [ fig : nuclearregion ] it is obvious that the 3 - 10kev hard x - ray emission of mrk 273 is mainly concentrated in the central compact region which shows little soft x - ray emission . therefore , to study how the x - ray emissions evolve as one moves away from the central nucleus , we extract spectra for the four annulus within @xmath51 region , as shown in fig . [
fig : nuclearregion ] , respectively .
the spectra were binned so that each bin contains at least 15 counts .
the results are shown in fig .
[ fig : ringspectra ] .
it is clear from this figure that within the inner - most @xmath52 region , most photons are hard with energy higher than 3kev and there are few photons with energy below 0.8kev . as the radius increases ,
the counts in soft x - ray band ( 0.32kev ) increase significantly , but the counts of hard x - ray photons have no substantial increase . more specifically ,
more than 80% hard x - ray photons are from the inner 1@xmath39 region . and
when the radius is increased to @xmath35 , the soft x - ray emission ( 0.32kev ) already dominates and there is no big difference in counts within the @xmath53 and @xmath51 circles .
this result confirms the visual impression that the hard x - ray emission is mainly from the central compact region while the soft x - ray ( below 2kev ) contribution is from a more extended area .
[ fig : modelfit ] is the spectrum for the whole 10@xmath39 region of mrk 273 from 0.3 to 8kev , which is more reliably calibrated for the acis .
the spectrum is clearly complex and consists of at least three components . at energies above 3kev
, there is a heavily absorbed power - law plus a narrow 6.4 kev @xmath0 line .
the other two components are a less absorbed power - law and a mekal thermal plasma emission with many evident line features , in particular fe l and ne line complex , and other @xmath6 elements emission lines ( such as @xmath54 , @xmath55 lines , ovii and oviii lines ) .
the parameters of the spectral fitting are given in table 2 and the information for the emission lines is given in table 3 and shown in fig .
[ fig : emissionlines ] .
these components are discussed in more detail below .
it is clear from fig .
[ fig : modelfit ] and table 2 that the central hard x - ray compact source has an absorbed power - law component and has a strong 6.4kev @xmath0 emission line with an ew of 213 ev .
the neutral hydrogen column density is @xmath56@xmath57 and the photon index is @xmath58 .
the absorption - corrected x - ray luminosity is @xmath2 , which is two orders of magnitude below the far - infrared luminosity of mrk 273 .
it is worth comparing the value of neutral hydrogen column density determined by iso / sws mid - ir observation , which is @xmath59@xmath57 ( genzel et al .
the similarity of the @xmath60 values obtained using these independent methods gives strong support to this spectral model .
these fitting results indicate that mrk 273 contains an obscured moderate luminosity agn . incorporating the image information ( 3.1 ) ,
the northern nucleus of mrk 273 is a seyfert 2 nucleus with a hard x - ray compact source embedded in a bright compact molecular core . on the other hand ,
the mekal thermal plasma model yields a satisfactory fit to the energy distribution of the hot gas surrounding the central nuclear region .
its temperature is about 0.8kev , with @xmath61@xmath57 and a metallicity of about @xmath62 ( cf . table 2 ) . also , there are remarkable fe l , ne line complexes , and @xmath55 , @xmath54 , ovii and oviii emission lines that are typical thermal excitation lines in the soft x - ray band .
the absorption - corrected soft x - ray luminosity of the thermal emission is @xmath63 .
the properties of this soft x - ray thermal emission are very similar to the diffuse emission ( excluding all the detected point sources ) in the antenna galaxy ngc 4038/4039 ( fabbiano , zezas & murray 2001 ) and those in the prototypical starburst galaxy ngc 253 ( strickland et al . 2000 ) .
therefore , the thermal bright soft x - ray emission within the central 10@xmath64 of mrk 273 is very likely due to massive starbursts .
there is also clearly a less absorbed second power - law component from the spectral fitting as shown in table 2 and in fig .
[ fig : modelfit ] ( the dashed line ) .
we first identified it as the scattered light from the central nucleus since the fitting power - law index @xmath65 ( @xmath661.9 ) is similar to that for the heavily absorbed power - law component .
however , most of the scattered light below 1kev from the agn can not escape from the very dense core , due to the high neutral hydrogen column density in the nuclear region even outside the central torus ( @xmath67 , cole et al .
this suggests that the second power - law component shown in fig .
[ fig : modelfit ] may have other contributions in addition to any directly scattered light from the central nucleus .
this component may be partly contributed to by the x - ray binaries ( xrbs ) or supernova remnants .
it has become increasingly clear that they can make a significant contribution to the x - ray luminosity in starburst galaxies .
examples include ngc 253 ( strickland et al . 2000 ; pietsch et al . 2001 ) , m82 ( kaaret et al .
2001 ) , interacting galaxies ngc 4038/4039 ( fibbiano et al .
2001 ) , ulirgs arp 220 ( iwasawa et al . 2001 ) and perhaps even in the x - ray faint elliptical and s0 galaxies ( blanton et al . 2001 ) . as cappi et al .
( 1999 ) points out , the typical spectrum for the sum of low - mass xrbs and high - mass xrbs can be described by an exponentially cut - off power - law with index @xmath68 which seems to describe the observed chandra spectrum for ngc 4038/4039 .
therefore , it is plausible that the unresolved xrbs and supernova remnants give contributions to the less absorbed second power - law component in spectral fitting for the central 10@xmath39 region of mrk 273 .
the spectral fitting for the hot gas halo has been performed for the region between the outer ellipse and the inner circle shown in the left panel of fig .
[ fig : halo ] excluding all bright point sources ( shown as small circles ) in this region . fig .
[ fig : halospectrum ] gives the mekal thermal model fitting and the fitting parameters are listed in table 4 .
a thermal model with temperature of 0.62kev , @xmath69@xmath57 and metallicity of about @xmath70 yields an acceptable fit .
the model dependent soft x - ray luminosity is @xmath10 .
comparing with the central soft x - ray spectral properties , the temperature of the extended hot gas is clearly lower than that of the central starburst region .
however , the difference is not large .
the most striking difference between the central soft x - ray and the extended hot gas is the metallicity .
the central starburst region has metallicity higher than the solar metallicity while the more extended hot gas outside seems to have a very low metallicity .
we caution that we have modelled the halo as a single mekal thermal model ; in reality , the gas may be clumpy and have complex temperature structures
. a map with higher signal - to - noise ratio will be very useful to further strengthen our conclusions on metallicities .
we briefly return to this important question of chemical enrichment in merging galaxies in 6.2 .
based on its optical spectrum , mrk 273x is classified as a seyfert 2 galaxy .
rosat observations indicate that it has an x - ray luminosity , @xmath71 in the 0.1 - 2.4kev band , one of the highest in the soft x - ray band among seyfert 2 galaxies ( xia et al .
1999 ) . borne et al .
( 1999 ) used hst i - band images of the mrk 273 field to analyze the luminosity distribution of faint galaxies surrounding mrk 273x .
they suggest that mrk 273x is the brightest elliptical galaxy in a relatively poor cluster .
hard x - ray observations have been carried out by asca and bepposax . however , due to their limited spatial resolution they can not resolve mrk 273x from mrk 273 . therefore the spectrum shown by iwasawa ( 1999 ) is a composite spectrum of mrk 273 and mrk 273x . the high resolution and high - energy response of chandra allow us to investigate the x - ray properties of mrk 273 and mrk 273x separately and also test the variability for both mrk 273 and mrk 273x . as can be seen from figs . 1 and 2
, mrk 273x is a bright x - ray point source .
figure 11 shows the observed chandra spectrum of mrk 273x from 0.3 to 8kev . a power - law model with photon index @xmath72 and @xmath73@xmath57 plus a thermal component with temperature of @xmath74 kev and solar abundance give the best fit to the spectrum of mrk 273x ( the @xmath75 per degree of freedom is 41.8/56 ) .
the absorption corrected x - ray luminosity in the 0.5 - 10 kev band is @xmath76 .
the luminosity of the thermal component is @xmath77 , about 3% of the total x - ray luminosity of mrk 273x . since mrk 273x is an elliptical galaxy from its de vaucouleurs surface brightness profile ( borne et al .
1999 ) , the thermal component of x - ray emission may naturally arise from the mrk 273x host galaxy . from the spectral
fitting , there is one significant emission line at 2.66 kev ( sxvi line ) with ew@xmath78 . in contrast , the @xmath0 emission line is not convincingly detected ( the ew upper limit for a line at 6.4kev is 30ev at the 90% confidence level ) .
we reported that there was no significant variability for either mrk 273 or mrk 273x during rosat observations ( xia et al .
the timing analysis for mrk 273 and mrk 273x based on the chandra data alone also does not reveal any significant variability . in order to test for possible long term variabilities ,
we compile in table 5 the x - ray fluxes of mrk 273 and mrk 273x in various x - ray bands from the literature . from table 5 ,
the flux differences in the soft x - ray band ( 0.1 - 2.4 kev ) are about 33% , 34% between rosat and chandra observations over a span of 8 years for mrk 273 and mrk 273x , respectively . in the 0.5 - 10kev band ,
the difference for the total flux of mrk 273 and mrk 273x between the chandra observation in 2000 and the asca observation in 1996 is about 21% . in the 2 - 10kev band ,
there are three reported observations for the total flux . while the chandra and asca fluxes are within 30% of each other , the bepposax flux is lower by about a factor of 2 ( risaliti et al . 2000 ) .
to summarize , all the data except that from bepposax are consistent with no significant variability .
however , much denser time sampling is required to firmly establish the variabilities of mrk 273 and mrk 273x .
the x - ray properties of mrk 273x can be compared with those of qsos , seyfert 1 and seyfert 2 galaxies by asca and bepposax ( e.g. , reeves & turner , 2000 ; pappa et al , 2001 ) .
although mrk 273x is classified optically as a seyfert 2 galaxy , its x - ray properties are typical for a seyfert 1 galaxy .
its high x - ray luminosity of @xmath76 falls into the regime of seyfert 1 galaxies .
furthermore , similar to many seyfert 1 galaxies , it is dominated by a power - law spectrum with @xmath79 , there is no cutoff at low energies and the @xmath0 line is not detected
. objects with seyfert 2 s optical spectrum but with seyfert 1 s properties in the x - ray have been reported ( for example ngc3147 , ptak et al .
1996 ; ngc7590 , bassani et al .
1999 ; ngc4698 , pappa et al , 2001 ; ngc7679 , della ceca et al , 2001 ) .
mrk 273x is another such example , although the x - ray luminosity of mrk 273x and the ratio of the hard x - ray luminosity ( in the 2 - 10kev band ) to the oiii luminosity , @xmath80 , are one of the highest even among this subclass of agns ( the @xmath80 ratio for mrk 273x is 86 ) .
given that these sources have weak or no @xmath0 detection and high @xmath80 ratios , they could not be compton thick objects ( with @xmath81@xmath57 ) because for such objects , the ew of @xmath0 is well above 1kev and the @xmath80 ratio is less than 1 ( gilli et al .
therefore , the power - law spectrum of this type of object is not from reflection , but could be due to the lack of intrinsic absorption .
the most straightforward explanation for this type of object is that it lacks broad line regions and the seyfert 2 optical spectrum is intrinsic , but not due to heavy absorption ( pappa et al .
the existence of such objects challenges the standard agn unification model .
moreover , mrk273x has a very high x - ray to optical b - band flux ratio ( @xmath82 , xia et al .
1998 ) , which is only achieved by bl lacertae objects ( stocke et al . 1991 ) .
the strong optical emission lines and the lack of dramatic x - ray variability ( see 5.2 ) , however , do not support mrk 273x being a bl lacertae object . therefore , mrk 273x still remains a mysterious object that warrant further studies .
compared with earlier satellites , the chandra acis - s3 observation of the mrk 273 field unveils unprecedented details of the x - ray emissions in the merging galaxy mrk 273 .
these observations raise a number of important questions , which we will discuss below .
multi - wavelength high resolution imagings for the nuclear region of mrk 273 reveal intricate and puzzling details of the heavily dust - obscured nuclear region of mrk 273 ( see 1 and 3.1 ) .
some components in one waveband have no clear counterparts in other wavebands .
the northern k band putative nucleus has radio , mid - infrared and @xmath16 narrow band counterparts .
combined with the fact that there is a bright molecular core at the center , this nucleus most likely hosts a central agn .
this is further supported by the chandra x - ray observations , where we have identified a hard x - ray source coincident with this northern nucleus within the astrometric uncertainty .
the x - ray spectrum of the hard x - ray source is typical for seyfert 2 galaxies .
the ` contradiction ' is from the comparison with the integral field spectroscopy of mrk 273 by colina et al .
( 1999 ) from which the northern nucleus of mrk 273 has a liner optical spectrum .
however , as suggested by terashima et al .
( 2000 ) , a liner 1 spectrum can also be produced by photoionizations due to hard photons from a low - luminosity agn .
the northern nucleus of mrk 273 may be similar , although in this case , the low - luminosity is not intrinsic but due to the heavy attenuation of gas . the high neutral hydrogen column - density ( @xmath83 )
implies that the soft x - ray photons could not escape from the central region , and thus the photoionizations can only be due to hard x - ray photons from the mrk 273 northern nucleus .
therefore , a liner 1 spectrum can also be a result of photoionizations by hard photons from a high - luminosity agn that is heavily absorbed by a dense neutral gas .
this mechanism may be common for ulirgs where both central agns and large amount of neutral gas are present in these merging galaxies .
on the other hand , the southwestern soft x - ray clumps are coincident with a bright @xmath8 nebula with angular distance of 4@xmath64 ( projected separation of about 4 kpc ) from the northern nucleus ( see fig .
[ fig : oiiihalpha ] ) .
this region has a spectrum dominated by the @xmath8 line emission and is classified as a seyfert 2 nebula by colina et al .
the northern nucleus has an associated extended gas disk with an inclination angle of @xmath20 ( carilli & taylor 2000 ) .
the co molecular disk has a similar inclination angle from the velocity map of downes & solomon ( 1998 ) .
this orientation , combined with the relative positions of the hard x - ray compact source , bright @xmath8 nebula , southwestern radio emission ( cole et al . 1999 ) as shown in figs . 4 and 5 and the narrow profile of @xmath8 line ( @xmath84 ) ( colina et al .
1999 ) , suggest that the southwestern bright @xmath8 nebula may be an extended narrow line region : the @xmath8 emission region may reside in an ionization cone centered on the hard x - ray source .
this geometry and the strong correlation between the soft x - ray clumps and the high excitation ( e.g. @xmath8 ) optical line emission is similar to the case in the nearby seyfert 2 galaxy ngc 1068 ( young et al .
therefore , the photoionization source for both the @xmath8 and soft x - ray may be the central agn associated with the northern nucleus .
it is , however , still possible that the excitation can be produced by shocks occurring in galaxy merging ( xia et al .
1999 ) . figs . [ fig : smoothed ] and [ fig : nuclearregion ] clearly show that the soft x - ray ( 0.32kev ) image of mrk 273 is complex with several bright clumps surrounding a hard x - ray compact nucleus , irregular structure along the direction of southern tidal tail and faint structure along the northern plume .
it extends beyond the optical emission region with a projected diameter of about @xmath85 .
a mekal thermal model with temperature of 0.62kev and metallicity of about @xmath70 provides the best fit .
the model - dependent soft x - ray luminosity of this hot gas halo is @xmath10 . from tables 2 and 4 ,
the hot gas properties in the central starburst region and the faint very extended halo are very different .
the very low metallicity implies that the diffuse hot gas halo has not yet been chemically contaminated by stellar processes such as superwinds which are expected to carry a substantial amount of heavy metals ( heckman 2001 ) .
the temperature and luminosity of this very extended hot gas halo are in the range of bright elliptical galaxies ( brighenti & mathews 1997 ) and a sample of merging galaxies studied by read & ponman ( 1998 ) .
the x - ray data therefore suggest that merging can produce hot halos that resemble those seen in elliptical galaxies , which adds another piece of evidence for elliptical galaxy formation via mergers .
the low - metallicity of the gas indicates that the cooling time of the gas may have been under - estimated by a factor of a few ( e.g. , fig .
9 - 9 in binney & tremaine 1987 ) since many previous studies assumed solar - metallicities ( e.g. , sarazin 1990 ) .
this hot gas may remain x - ray luminous for more than @xmath31 1 gyr , long after the stellar components at the center have relaxed dynamically .
the low metallicity , although somewhat uncertain , suggests that part of the x - ray emission in elliptical galaxies may be due to secondary infalls in the pre - merger ( group ) environment ( e.g. , brighenti & mathews 1998 , 1999 ) , which have not yet been contaminated by the central starbursts .
this is also consistent with the recent work by bekki ( 2001 ) who studied multiple galaxy mergers using numerical simulations .
he showed that metals produced and ejected in central star formation regions are mostly mixed with the local interstellar medium .
consequently , the ism of the outer part of the merger is less metal - enriched .
the radial gradient in metallicity seen in mrk 273 is certainly consistent with the picture , where the 10 central region has a metallicity ( @xmath86 ) while the very extended hot gas halo has a much lower metallicity ( @xmath11 ) .
we caution , however , that the star formation treatment in numerical simulations is not realistic in the sense that the multi - phase medium nature of the interstellar medium is not properly taken into account . as a consequence of this inadequacy
, the region that are chemically enriched may be under - estimated .
to conclude , the high resolution observation of the mrk 273 field by chandra yielded important results for both mrk 273 and mrk 273x .
for mrk 273x , chandra observation confirm that its x - ray properties are similar to seyfert 1 galaxies while it is classified as a seyfert 2 galaxy optically .
for mrk 273 , where our primary interests lie , chandra observation , for the first time , reveals a compact hard x - ray nucleus inside a much more extended halo .
although the nuclear region of mrk 273 is complex , it is a typical seyfert 2 nucleus with extended narrow line region .
the high temperature and high - metallicity thermal emission component from the spectral fitting are obviously from the starburst region , while the much more extended hot gas halo has a much lower metallicity and may have a different origin .
the data provide further support for the evolutionary connection between galaxy merging and elliptical galaxies .
more high - resolution deep images of merging galaxies at various stages will be very useful for providing further clues of x - ray halos in merging galaxies and their descendents , elliptical galaxies .
we are grateful to the cxc team for assistance in observations and data reduction .
we also thank drs .
mo , a. pedlar , j. skelton , y. gao , m.s .
yun , x.p .
wu , d. wang , t.g .
wang and x.w .
liu for advice and helpful discussions .
thanks are also due to dr .
z.j . jiang for help in the data reduction .
this project was supported by the nsf of china and nkbrsf g19990754 .
sm gratefully acknowledges a travel grant awarded by the nsf of china .
armus , l. , heckman , t. m. , & miley , g. k. 1990 , apj , 364 , 471 bassani , l. , dadina , m. , maiolino , r. , salvati , m. , risaliti , g. , della ceca , r. , matt , g. , & zamorani , g. 1999 , apjs , 121 , 473 binney , j. j. , tremaine , s. 1987 , galactic dynamics ( princeton : princeton university press ) , p. 580 bekki , k. 2001 , apj , 546 , 189 blanton , e. l. , sarazin , c. l. , & irwin , j. a. 2001 , apj , in press ( astro - ph/0012481 ) borne , k.d .
, colina , l. , bushouse , h. , & lucas , r.a . 1999 ,
apj , 527 , 554 cappi , m. , et al .
1999 , a&a , 350 , 777 carilli , c. l. , & taylor , g. b. 2000 , apj , 532 , l95 cole , g. h. j. , pedlar , a. , holloway , a. j. , & mundell , c. g. 1999 , mnras , 310 , 1033 colina , l. , arribas , s. , & borne , k. 1999 , apjl , 527 , l13 condon , j. j. , huang , z .- p . , yin q. f. , & thuan , t. x. 1991 , apj , 278 , 65 cui , j. , xia , x .- y . , deng , z .- g . ,
mao , s. , zou , z.l .
2001 , aj , in press ( astro - ph/0104296 ) della ceca , r. , pellegrini , s. , bassani , l. , beckmann , v. , cappi , m. , palumbo , g.g.c . , trinchieri , g. , & wolter , a. 2001 , a&a in press ( astro - ph/0106444 ) dopita , m. a. , & sutherland , s. 1995 , apj , 455 , 468 downes , d. , & solomon , p. m. 1998 , apj , 507 , 615 fabbiano , g. , zezas , a. , & murry , s. s. 2001 , apj , in press ( astro - ph/0102256 ) genzel , r. , et al .
1998 , apj , 498 , 579 gilli , r. , comastri , a. , brunetti , g. , & setti , g. 1999 , newa , 4 , 45 heckman , t. m. 2001 in `` gas & galaxy evolution '' , in press ( astro - ph/0009075 ) hibbard , j. e. , & yun , m. s. 1999 , aj , 118 , 162 iwasawa , k. 1999 , mnras , 302 , 96 iwasawa , k. , matt , g. , guainazzi , m. , & fabian , a. c. 2001 , mnras , in press ( astro - ph/0103417 ) kaaret , p. , et al .
2001 , mnras , 321 , l29 knapen , j. h. , laine , s. , yates , j. a. , robinson , a. , richards , a. m. s. , doyon , r. , & nadeau d. 1997 , apj , 490 , l29 osullivan , e. , forbes , d. a. , & ponman , t. j. 2001 , mnras in press , ( astro - ph/0101271 ) pappa , a. , georgantopoulos , i. , stewart , g.c .
, & zezas , a.l .
mnras in press ( astro - ph/0104061 ) pietsch , w. , et al . 2001 , a&a , 365 , l174 ptak , a. , yaqoob , t. , serlemitsos , p.j . , & kunieda , h. 1996 , apj , 459 , 542 risaliti , g. , gill , r. , maiolino , r. , & salvati , m. 2000 , a&a 357 , 13 read , a.m. , & ponman , t. j. 1998 , mnras , 297 , 143 reeves j.n . , &
turner m.j.l .
2000 , mnras , 316 , 234 sansom , a.e . ,
hibbard , j.e .
, schweizer f. 2000 , 120 , 1946 sarazin , c. l. 1990 , in the interstellar medium in galaxies , eds .
thronson , & j. m. shull ( dordrecht : kluwer ) , 201 sarazin , c. l. , irwin , j. a. , & bregman , j. n. 2000 , apj , 544 , l101 scoville , n. z. , evans , a. s. , thompson , r. , rieke , m. , hines , d. c. , low , f. j. , dinshaw , n. , surace , j. a. , & armus , l. 2000 , aj , 119 , 991 soifer , b. t. et al .
2000 , aj , 119 , 509 stocke , j.t . ,
1991 , apjs , 76 , 813 strickland , d. k. , heckman , t. m. , weaver , k. a. , & dahlem , m. 2000 , aj , 120 , 2965 terashima , y. , ho , l. c. , & ptak , a. f. 2000 , apj , 539 , 161 turner , t. j. , george , i. m. , nandra , k. , & mushotzky , r. f. 1997 , apjs , 113 , 23 turner , t. j. , george , i. m. , nandra , k. , & mushotzky , r. f. 1998 , apj , 493 , 91 xia , x .- y . , boller , th . , wu , h. , deng , z .- g . , gao , y. , zou , z .-
mao , s. , & brner , g. 1998 , apj , 496 , l9 ; erratum , apj , 1998 , 507 , l99 xia , x .- y . , mao , s. , wu , h. , liu , x .- w . ; gao , y. , deng , z .- g . ,
1999 , apj , 524 , 746 young , a. j. , wilson , a. s. , & shopbell , p. l. 2001 , apj , in press ( astro - ph/0104027 ) llll hard x - ray & 13:44:42.12 & 55:53:12.9 & 0.28 clump
n & 13:44:42.11 & 55:53:14.4 & 1.22clump se & 13:44:42.33 & 55:53:11.5 & 2.46clump sw1 & 13:44:41.93 & 55:53:11.0 & 2.69clump sw2 & 13:44:41.70 & 55:53:11.5 & 3.90radio n & 13:44:42.1171 & 55:53:13.182 & 0radio se & 13:44:42.1677 & 55:53:12.496 & 0.81radio sw & 13:44:42.0372 & 55:53:12.144 & 1.24 cccc 6.37@xmath97 & 6.2@xmath98 & 213 & @xmath0 5.81@xmath99 & 2.7@xmath100 & 126 & ? 5.47@xmath101 & 0.4@xmath102 & 100 & ? 3.88@xmath103 & 0.5@xmath104 & 110 & ca xix 3.55@xmath103 & 0.6@xmath105 & 195 & ? 1.91@xmath106 & 1.0@xmath107 & 110 & @xmath55 1.49@xmath106 & 1.0@xmath108 & 75 & @xmath54 0.76 - 1.26 & - & - & fe & ni complex 0.69@xmath109 & 2.5@xmath110 & 65 & o viii 0.57@xmath111 & 1.7@xmath112 & 34 & o vii llllll mrk 273 & 1.25 & & & rosat & 1992 mrk 273 & 0.94 & 7.93 & 8.72 & chandra & 2000 mrk 273x & 1.10 & & & rosat & 1992 mrk 273x & 0.82 & 1.19 & 1.85 & chandra & 2000 mrk 273+mrk 273x & 2.35 & & & rosat & 1992 mrk 273+mrk 273x & 1.7 & 7.0 & 8.7 & asca & 1996 mrk 273+mrk 273x & & 3.50 & & bepposax&1998 mrk 273+mrk 273x & 1.76 & 9.12 & 10.57 & chandra & 2000 | we report x - ray observations of the field containing the ultraluminous iras galaxy mrk 273 .
the data were obtained using the acis - s3 instrument on board chandra .
the high resolution x - ray image , for the first time , reveals a compact hard x - ray nucleus in mrk 273 . its position is coincident with the northern nucleus identified in the optical , infrared , radio and in molecular co maps .
its x - ray energy distribution is well described by a heavily obscured agn spectrum with an absorbed power - law plus a narrow @xmath0 emission line at 6.4kev .
the neutral hydrogen column density is about @xmath1 , implying an absorption - corrected x - ray luminosity ( 0.110 kev ) for the nucleus of @xmath2 for @xmath3 .
the x - ray properties therefore firmly establishes the northern nucleus of mrk 273 as a seyfert 2 active galactic nucleus .
there are also bright soft x - ray clumps and diffuse soft x - ray emissions surrounding the central hard x - ray nucleus within the @xmath4 of the nuclear region .
its spectrum can be fitted by a mekal thermal model with temperature of about 0.8kev and high metallicity ( @xmath5 ) plus emission lines from @xmath6 elements and ions .
we find that a soft x - ray clump , about @xmath7 ( projected separation of about 4 kpc ) southwest of the northern hard x - ray nucleus , is coincident with a nebula with strong @xmath8 emissions .
further outside the central region , the chandra observations reveal a very extended hot halo in mrk 273 .
the x - ray halo encompasses the entire optical tidal tail and plume , with a projected diameter of about @xmath9 .
the total soft x - ray luminosity ( 0.12.4kev ) of the hot halo is @xmath10 , in the range of the soft x - ray luminosity of bright elliptical galaxies .
the temperature of the hot gas is about 0.62kev with a low metallicity ( @xmath11 ) .
we discuss the nature of the agn in mrk 273 and the implications of our results on the origin of x - ray halos in elliptical galaxies .
we also discuss the properties of mrk 273x , a background agn at redshift @xmath12 in the mrk 273 field .
the agn has an x - ray luminosity of @xmath13 in the 0.5 - 10kev band .
its x - ray properties resemble those of seyfert 1 galaxies while its optical properties are similar to seyfert 2 galaxies .
such mixed classifications in the optical and x - ray may be a challenge for the unification scheme of agns . #
1#2=.5 = .5 # 1*(smao : # 1 ) * # 1(*#1 xia ) * |
network - based dynamical systems feature agents that communicate via a dynamic graph while acting on the information they receive .
these systems have received increasing attention lately because of their versatile use in modeling social and biological systems @xcite . typically , they consist of a fixed number @xmath0 of agents , each one located at @xmath1 on the real line .
the agents positions evolve as interactions take place along the edges of a dynamical graph that evolves endogenously .
the motivation behind the model is to get a better understanding of the dynamics of collective behavior .
following @xcite , we express the system as a set of @xmath0 coupled stochastic differential equations : @xmath2 where @xmath3 is the magnitude of the noise , @xmath4 are independent wiener processes , and the influence " parameter @xmath5 is a function of the distance between agents @xmath6 and @xmath7 ; in other words , @xmath8 , where @xmath9 is nonnegative ( to create attractive forces ) and compactly supported over a fixed interval ( to keep the range of the forces finite ) .
intuitively , the model mediates the competing tension between two opposing forces : the sum in ( [ sde ] ) pulls the agents toward one another while the diffusion term keeps them jiggling in a brownian motion ; the two terms push the system into ordered and disordered states respectively . in the mean field limit , @xmath10 , equation ( [ sde ] )
induces a nonlinear fokker - planck equation for the agent density profile @xmath11 @xcite : @xmath12 the function @xmath11 is the limit density of @xmath13 , as @xmath0 goes to infinity , where @xmath14 denotes the dirac measure with point mass at @xmath15 . in the classic hegselmann - krause ( _ hk _ ) model , one of the most popular systems in consensus dynamics
@xcite , each one of the @xmath0 agents moves , at each time step , to the mass center of all the others within a fixed distance .
the position of an agent represents its `` opinion '' .
if we add noise to this process , we obtain the discrete - time version of ( [ sde ] ) for @xmath16}\left(y\right)$ ] . to be exact , the original _ hk _ model does not scale @xmath5 by @xmath17 but by the reciprocal of the number agents within distance @xmath18 of agent @xmath6 .
canuto et al .
@xcite have argued that this difference has a minor impact on the dynamics . by preserving the pairwise symmetry among the agents
, however , the formulation ( [ sde ] ) simplifies the analysis .
the _ hk _ model has been the subject of extensive investigation .
a sample of the literature includes work on convergence and consensus properties @xcite , conjectures about the spatial features of the attractor set @xcite , and various extensions such as _ hk _ systems with inertia @xcite , leaders @xcite , or random jumps @xcite .
this paper focuses on the analysis of the nonlinear fokker - planck equation for the noisy _ hk _ system , which corresponds to setting @xmath16}\left(y\right)$ ] in ( [ pde ] ) . for concreteness ,
let us denote @xmath19 $ ] , @xmath20\times\left(0,t\right)$ ] for @xmath21 , and consider the following periodic problem for the _ hk _ system : @xmath22\\ \rho & = \rho_{0 } & & \text{on}\ u\times\left\ { t=0\right\ } \end{aligned } \end{cases}\label{eq : evo_equation}\end{aligned}\ ] ] where @xmath23 and the initial condition @xmath24 is assumed to be a probability density , i.e. , @xmath25 and @xmath26 .
the positive constants @xmath27 , @xmath18 and @xmath3 are fixed with @xmath28 .
note that we have to periodically extend @xmath29 outside of @xmath30 in order to make sense of the integral above .
the periodicity of @xmath29 , together with eq .
( [ eq : evo_equation ] ) , immediately implies the normalization condition @xmath31 for all @xmath32 .
[ [ main - results . ] ] main results .
+ + + + + + + + + + + + + we establish the global well - posedness of eq .
( [ eq : evo_equation ] ) , which entails the existence , uniqueness , nonnegativity and regularity of the solution .
in addition , we prove a global stability condition for the uniform solution @xmath33 , representing the state without any clustering of opinions .
this gives a sufficient condition involving @xmath18 and @xmath3 for which no consensus can be reached regardless of the initial condition .
the paper is organized as follows . in section [ sec : apriori ] , we first derive the aforementioned global stability condition by assuming that a sufficiently smooth solution exists . more precisely , we show that as long as @xmath34 , the uniform solution @xmath33 is unconditionally stable in the sense that @xmath35 exponentially as @xmath36 for any initial data @xmath37 . an important ingredient in
the proof is a @xmath38 estimate on the solution ( lemma [ lem : l1 ] ) .
interestingly , this estimate immediately implies the nonnegativity of the solution while no arguments using maximum principles are required .
the latter may not be easy to obtain for nonlinear partial integro - differential equations , such as the one we consider here .
we close the section by discussing the physical significance of the stability result and how it relates to other works in the opinion dynamics literature . in section [ sec : existenceuniqueness ] , we prove the global existence and uniqueness of the weak solution to ( [ eq : evo_equation ] ) when the initial data @xmath37 . here , we construct approximate solutions by solving a series of linear parabolic equations obtained from ( [ eq : evo_equation ] ) by replacing @xmath39 with @xmath40 . using energy estimates ,
we find that the sequence of solutions @xmath41 forms a cauchy sequence in @xmath42 and we use this strong convergence result to simplify the existence proof . finally , in section
[ sec : regularity ] we establish improved regularity properties of the weak solution if @xmath43 for some @xmath44 .
this allows us to remove the a priori smoothness assumptions in the stability and positivity results of section [ sec : apriori ] .
the main results in this paper are then summarized in theorem [ thm : mainresult ] .
[ [ notation . ] ] notation .
+ + + + + + + + + as customary in the literature , we often treat @xmath29 ( and other functions on @xmath45 ) not as a function from @xmath45 to @xmath46 , but from @xmath47 $ ] to a relevant banach space . in this case , we define for each @xmath48 , @xmath49\left(\cdot\right):=\rho\left(\cdot , t\right).\ ] ] for a shorthand , we denote the usual @xmath50 norms on @xmath30 by @xmath51 other banach space norms will be written out explicitly .
since we are dealing with periodic boundary conditions , a subscript `` per '' will often be attached to the relevant banach space symbols to denote the subspace of periodic functions on @xmath30 , e.g. , @xmath52 for @xmath53 and @xmath54 .
they are equipped with the usual norms .
finally , we denote by @xmath55 any generic constants ( possibly depending on @xmath18 , @xmath27 or @xmath3 ) used in the various energy estimates .
their actual values may change from line to line .
when they depend on the time interval under consideration , we will indicate the dependence explicitly , e.g. , @xmath56 .
first , let us assume that there exists a sufficiently smooth solution @xmath57 to equation ( [ eq : evo_equation ] ) .
this allows us to prove a priori energy estimates , from which we can deduce the solution s nonnegativity and derive a global stability condition of the spatially uniform solution @xmath33 .
we begin by setting @xmath58 so that @xmath59
. then , @xmath60 satisfies the equation @xmath61 the other two extra terms are zero since @xmath62 for all @xmath15 .
multiplying equation ( [ eq : psi ] ) by @xmath60 and integrating by parts over @xmath30 , we have @xmath63 by the cauchy - schwarz inequality , @xmath64 first let us estimate @xmath65 .
for any @xmath15 and @xmath48 , we have @xmath66 in order to proceed with the bound on @xmath65 , we need an @xmath67 estimate of @xmath68 . [
lem : l1 ] suppose @xmath69 is a solution of ( [ eq : evo_equation ] ) with @xmath70 , then @xmath71 for all @xmath72 .
let @xmath73 and define @xmath74 this is a convex @xmath75-approximation of the absolute value function satisfying @xmath76 .
multiplying @xmath77 to equation ( [ eq : evo_equation ] ) and integrating by parts over @xmath30 , we have @xmath78^{1/2}\right\vert _ { 2}^{2}=-\int_{u}\rho g_{\rho}\chi_{\epsilon}^{\prime\prime}\left(\rho\right)\rho_{x}\ , dx.\ ] ] by cauchy - schwarz and young s inequality , @xmath79^{1/2}\right\vert _ { 2}^{2}\nonumber \\ & \leq & \left\vert \rho\left(t\right)g_{\rho}\left(t\right)\left[\chi_{\epsilon}^{\prime\prime}\left(\rho\left(t\right)\right)\right]^{1/2}\right\vert _ { 2}\left\vert \rho_{x}\left(t\right)\left[\chi_{\epsilon}^{\prime\prime}\left(\rho\left(t\right)\right)\right]^{1/2}\right\vert _ { 2}\nonumber \\ & \leq & \frac{1}{2\sigma^{2}}\left\vert \rho\left(t\right)g_{\rho}\left(t\right)\left[\chi_{\epsilon}^{\prime\prime}\left(\rho\left(t\right)\right)\right]^{1/2}\right\vert _ { 2}^{2}+\frac{\sigma^{2}}{2}\left\vert \rho_{x}\left(t\right)\left[\chi_{\epsilon}^{\prime\prime}\left(\rho\left(t\right)\right)\right]^{1/2}\right\vert _ { 2}^{2}. \end{aligned}\ ] ] therefore , @xmath80^{1/2}\right\vert _ { 2}^{2}\nonumber \\ & \leq & \frac{1}{2\sigma^{2}}\left\vert g_{\rho}\left(t\right)\right\vert _ { \infty}^{2}\left\vert \rho\left(t\right)\left[\chi_{\epsilon}^{\prime\prime}\left(\rho\left(t\right)\right)\right]^{1/2}\right\vert _ { 2}^{2}.\label{eq : l1_beforelim } \end{aligned}\ ] ] replacing @xmath60 by @xmath29 in ( [ eq : g_infty_est ] ) we have @xmath81 the term @xmath82^{1/2}\right\vert _ { 2}^{2}$ ] can be split into two integrals : @xmath83^{1/2}\right\vert _ { 2}^{2 } & = & \int_{u}\rho^{2}\chi_{\epsilon}^{\prime\prime}\left(\rho\right)\ , dx\nonumber \\ & = & \int_{u}\rho^{2}\chi_{\epsilon}^{\prime\prime}\left(\rho\right)\mathbf{1}_{\left\ { \left|\rho\right|>\epsilon\right\ } } \ , dx\nonumber \\ & & + \int_{u}\rho^{2}\chi_{\epsilon}^{\prime\prime}\left(\rho\right)\mathbf{1}_{\left\ { \left|\rho\right|\leq\epsilon\right\ } } \ , dx .
\label{eq : bef_second_one } \end{aligned}\ ] ] for @xmath84 , @xmath85 by construction , and hence the first integral above is zero .
the second integral is estimated as : @xmath86 therefore , by ( [ eq : gtilda ] , [ eq : bef_second_one ] , [ eq : second_one ] ) , eq . ( [ eq : l1_beforelim ] ) becomes @xmath87^{2}.\ ] ] applying grnwall s inequality , we get @xmath88 .
\label{eq : l1_gronwall } \end{aligned}\ ] ] since @xmath29 is continuous , the integral in the exponential is finite . therefore , taking the limit @xmath89 yields @xmath90 for every @xmath72 , as required .
incidentally , lemma [ lem : l1 ] establishes the nonnegativity of @xmath29 .
this is important because @xmath29 represents the density of opinions of individuals and , as such , is necessarily nonnegative at all times .
it is interesting that a @xmath91 estimate suffices to show nonnegativity and no arguments from maximum principles are required .
[ cor : positivity ] if @xmath69 is a solution of ( [ eq : evo_equation ] ) , with @xmath92 and @xmath26 , then @xmath93 and @xmath94 in @xmath30 for all @xmath72 . since @xmath26 ,
the normalization condition ( [ eq : normalization_cond ] ) is satisfied for @xmath95 . applying lemma [ lem : l1 ] , we have @xmath96 hence , @xmath93 . but @xmath97 these equations imply that @xmath98 , and hence , @xmath94 a.e . in @xmath30 . by continuity ,
@xmath94 in @xmath30 for all @xmath72 . with lemma [ lem : l1 ]
, it follows from ( [ eq : g_infty_est ] ) that @xmath99 next , we also have @xmath100 consequently , @xmath101 with estimates ( [ eq : est_1 ] ) and ( [ eq : est_2 ] ) , ( [ eq : energy_est1 ] ) becomes @xmath102 hence , we have @xmath103 by construction , @xmath104 for all @xmath48 .
thus , @xmath105 satisfies the poincar inequality @xmath106 for the interval @xmath19 $ ] , the optimal poincar constant is @xmath107 .
therefore , ( [ eq : energy_est2 ] ) becomes @xmath108 but @xmath109 thus we obtain the integral form of ( [ eq : energy_est2 ] ) : @xmath110 in particular , if @xmath34 , the constant factor in the exponential is negative , therefore @xmath111 as long as @xmath112 .
we summarize these results : let @xmath69 be a solution of ( [ eq : evo_equation ] ) with @xmath92 , @xmath26 , and @xmath113 .
if @xmath34 , then @xmath35 in @xmath114 exponentially as @xmath36 .
[ thm : apriori_stability ] the noisy _ hk _ model generally exhibits two types of steady - states . the first is a spatially uniform steady - state , i.e. , @xmath29 is constant .
this represents the case where individuals have uniformly distributed opinions , without any local or global consensus .
the second involves one or more clusters of individuals having similar opinions , in which case @xmath29 is a multi - modal profile . which of these two steady - states appear in the long - time limit
depends on the interaction radius @xmath18 and noise @xmath3 , as well as the initial profile @xmath115 . in this aspect ,
theorem [ thm : apriori_stability ] gives a sufficient condition for the spatially uniform solution to be the globally attractive state , irrespective of the initial profile @xmath115 .
in other words , as long as @xmath34 , any initial profile converges to the spatially uniform state .
in particular , clustered profiles do not even have local stability .
this immediately indicates a forbidden zone for consensus : when the volatility of one s opinion is too large compared to the interaction radius , there can be no clustering of opinions regardless of the initial opinion distribution .
it should be noted that this is the first result regarding the nonlinear stability of the _ hk _ system . on the other hand , it is straightforward to perform linear stability analysis of eq .
( [ eq : evo_equation ] ) at the uniform solution @xmath33 to derive a linear stability condition for the uniform solution @xcite .
the combination of these two results indicate a region where it is possible to have both clustered and uniform states as locally stable solutions ( see figure [ fig : pd ] ) . , above which the spatially uniform solution ( @xmath116 ) is unconditionally stable , i.e. no clustering of opinions is possible .
the bottom ( blue ) curve is obtained from linear stability analysis around the spatially uniform solution , and has the form @xmath117 . below this curve ,
the uniform solution loses linear stability and only clustered solutions are permitted . between these two curves
is the region for which both clustered and uniform solutions can be stable with respect to small perturbations . ]
our discussion so far has assumed the existence of a solution to ( [ eq : evo_equation ] ) . in this section ,
we prove the existence and uniqueness of the weak solution by defining a sequence of linear parabolic equations , whose solutions converge strongly to a function @xmath29 that solves a weak formulation of eq .
( [ eq : evo_equation ] ) . to begin with ,
let @xmath21 and consider a sequence of linear parabolic equations @xmath118\\ \rho_{n } & = \ , \rho_{0 } & & \text{on}\ u\times\left\ { t=0\right\ } \end{aligned } \end{cases}\label{eq : sequence_evolution}\end{aligned}\ ] ] for @xmath119 , with @xmath120 for all @xmath95 . for convenience , we assume that the initial condition satisfies @xmath121 , @xmath92 and @xmath122
. the smoothness condition will be relaxed later ( see theorem [ thm : existnece-1 ] at the end of this section ) .
consider the case @xmath123 .
since @xmath121 and both @xmath124 and @xmath125 are bounded , by standard results on linear parabolic evolution equations , there exists a unique @xmath126 satisfying ( [ eq : sequence_evolution ] ) for @xmath123 .
iterating this for @xmath127 implies that there exists a sequence of smooth functions @xmath128 satisfying ( [ eq : sequence_evolution ] ) .
next , we establish some uniform energy estimates on @xmath129 .
let @xmath21 and suppose @xmath128 satisfy ( [ eq : sequence_evolution ] ) with @xmath121 .
then , @xmath130 for all @xmath131 and @xmath119 .
[ prop : unif_l1_est ] since we know that @xmath132 for all @xmath119 and all @xmath131 , we can proceed exactly as in the proof of lemma [ lem : l1 ] . in this case , instead of ( [ eq : l1_gronwall ] ) we have @xmath133 .
\end{aligned}\ ] ] since @xmath134 is smooth , the integral in the exponential is finite , hence we take the limit @xmath89 to obtain @xmath135 for all @xmath131 .
let @xmath21 and suppose @xmath128 satisfy ( [ eq : sequence_evolution ] ) with @xmath121 , @xmath92 and @xmath26 .
then , @xmath136 and @xmath137 in @xmath30 for all @xmath131 and for all @xmath119 .
[ cor : positivity_n ] since the functions @xmath129 are all periodic , we have @xmath138 ; hence the proof is identical to corollary [ cor : positivity ] .
let @xmath21 and suppose @xmath128 satisfy ( [ eq : sequence_evolution ] ) with @xmath121 , @xmath92 and @xmath26 .
then , there exists a constant @xmath139 such that @xmath140 [ prop : unif_l2_est ] we proceed as in section [ sec : apriori ] by multiplying ( [ eq : sequence_evolution ] ) by @xmath129 and integrating by parts .
this gives us @xmath141 using proposition [ prop : unif_l1_est ] and corollary [ cor : positivity_n ] we have @xmath142 and hence ( [ eq : unif_l2_main1 ] ) becomes @xmath143 which implies , by integration , that @xmath144 for all @xmath131 , and @xmath145 let @xmath21 and suppose @xmath128 satisfy ( [ eq : sequence_evolution ] ) with @xmath121 , @xmath92 and @xmath26 .
then , there exists a constant @xmath139 such that @xmath146 [ prop : unif_l2_est_higher ] multiplying equation ( [ eq : sequence_evolution ] ) by @xmath147 and integrating by parts over @xmath30 , it follows from cauchy - schwarz , young s inequality , and ( [ grhoub ] ) that @xmath148 now , @xmath149 \nonumber \\ & = & -r\left(\rho_{n-1}\left(x+r , t\right)+\rho_{n-1}\left(x - r , t\right)\right ) \nonumber \\ & & + \int_{x - r}^{x+r}\rho_{n-1}\left(y , t\right)dy .
\label{eq : g_diff_first } \end{aligned}\ ] ] by ( [ rhoncub ] ) and morrey s inequality , @xmath150 it follows that ( [ eq : regularity_exp_1 ] ) becomes @xmath151 integrating over @xmath48 , we have @xmath152 applying the estimates in proposition [ prop : unif_l2_est ] , we find that @xmath153 with the uniform estimates above , we can now show that @xmath129 converges strongly to a limit .
let @xmath21 and suppose that @xmath128 satisfies ( [ eq : sequence_evolution ] ) with @xmath121 , @xmath92 and @xmath26 .
then there exists @xmath154 such that @xmath155 in @xmath156 .
[ lem : cauchy ] we set @xmath157 for @xmath158 . for @xmath159 ,
the evolution equation for @xmath160 reads @xmath161 let @xmath73 . multiplying the equation above by @xmath162 ( see definition ( [ eq : chi_def ] ) ) and integrating by parts yields @xmath163^{1/2}\phi_{n_{x}}\left(t\right)\right\vert _ { 2}^{2}\nonumber \\ & \leq & \int_{u } \left\vert\chi^{\prime\prime}_{\epsilon}\left(\phi_n\right ) \phi_{n_{x}}\phi_{n } g_{\rho_{n-1}}\right\vert dx + \int_{u}\left\vert\chi^{\prime}_{\epsilon}\left(\phi_n\right ) \left(\rho_{n-1}g_{\phi_{n-1}}\right)_x\right\vert dx\nonumber \\ & \leq & \frac{\sigma^{2}}{2 } \left\vert \left[\chi^{\prime\prime}_{\epsilon}\left(\phi_n\left(t\right)\right)\right]^{1/2}\phi_{n_{x}}\left(t\right)\right\vert _ { 2}^{2 } \nonumber \\ & & + \frac{1}{2\sigma^2 }
\left\vert g_{\rho_{n-1}}\left(t\right)\right\vert^{2}_{\infty } \left\vert \left[\chi^{\prime\prime}_{\epsilon}\left(\phi_n\left(t\right)\right)\right]^{1/2 } \phi_{n}\left(t\right)\right\vert^2_{2 } \nonumber \\ & & + \int_{u}\left\vert\chi^{\prime}_{\epsilon}\left(\phi_n\right ) \left(\rho_{n-1}g_{\phi_{n-1}}\right)_x\right\vert dx .
\label{eq : phi_n_main1 } \end{aligned}\ ] ] by corollary [ cor : positivity_n ] , @xmath164 . also , as in ( [ eq : second_one ] ) from the proof of lemma [ lem : l1 ] , we have @xmath165^{1/2}\phi_{n}\left(t\right)\right\vert _ { 2}^{2 } \leq c \epsilon.\ ] ] to estimate the last integral in ( [ eq : phi_n_main1 ] ) , observe that @xmath166 , hence @xmath167 but we know that @xmath168 and that @xmath169 ( see expression ( [ eq : g_diff_first ] ) )
. moreover , morrey s inequality implies @xmath170 .
hence , it follows that , as @xmath171 tends to @xmath172 , ( [ eq : phi_n_main1 ] ) becomes @xmath173 where in the last line we used proposition [ prop : unif_l2_est_higher ] and the shorthand @xmath174 now , for @xmath175 we define @xmath176 by ( [ eq : phi_n_main2 ] ) and corollary [ cor : positivity_n ] , @xmath177 moreover , the @xmath129 s coincide at @xmath178 , so @xmath179 .
thus by grnwall s inequality , @xmath180 uniformly in @xmath0 and @xmath48 .
furthermore , for each @xmath48 , @xmath181 is a bounded monotone sequence in @xmath0 , hence there exists @xmath182 such that @xmath183 , pointwise in @xmath48 . by the monotone convergence theorem , @xmath184
this result immediately implies that @xmath185 is a cauchy sequence in @xmath156 .
indeed , for @xmath73 we can pick @xmath175 such that @xmath186 .
hence , for all @xmath187 , @xmath188 therefore , @xmath185 is a cauchy sequence and there exists @xmath154 such that @xmath155 in @xmath156 .
note that we can extract from @xmath185 a subsequence that converges weakly in smaller spaces .
we denote by @xmath189 the dual space of @xmath190 .
since periodic boundary conditions allows integration by parts without extra terms , most characterizations of @xmath191 carries over to @xmath192 .
we have @xmath193 , with @xmath194 , and the estimate @xmath195 moreover , there exists a subsequence @xmath196 such that @xmath197 and @xmath198 [ lem : weakconv ] from proposition [ prop : unif_l2_est ] , we have @xmath199 next , observe that from the evolution equation of @xmath129 , we have @xmath200 hence , @xmath201 where in the last step we used proposition [ prop : unif_l2_est ] .
therefore , we have the uniform estimate @xmath202 hence , @xmath193 , with @xmath194 and they satisfy the same estimate ( [ eq : unif_3_ineq ] ) .
furthermore , there exists @xmath203 such that @xmath204 following @xcite , we can deduce from lemma [ lem : weakconv ] the following result : suppose @xmath205 with @xmath194 , then @xmath206 up to a set of measure zero .
further , the mapping @xmath207 is absolutely continuous , with @xmath208 for a.e .
@xmath131 . here
, @xmath209 denotes the pairing between @xmath192 and @xmath210 .
[ thm : evans ] the proof is identical to the proof in evans @xcite section 5.9 theorem 3 .
the only difference here is that we are considering @xmath210 and @xmath192 , instead of @xmath211 and @xmath212 . since periodic conditions
still guarantees integration by parts without extra terms , all proofs follow through . now , we are ready to prove the existence of a weak solution to equation ( [ eq : evo_equation ] ) .
we say that @xmath193 with @xmath194 is a _ weak solution _ of equation ( [ eq : evo_equation ] ) if for every @xmath213 , @xmath214 and @xmath215 .
note that since @xmath206 ( theorem [ thm : evans ] ) , the last condition makes sense as an initial condition .
[ thm : existnece](existence and uniqueness ) let @xmath121 , @xmath92 and @xmath26 .
then , there exists a unique weak solution @xmath216 , with @xmath217 , to equation ( [ eq : evo_equation ] ) with the estimate @xmath218 for each @xmath219 , we multiply equation ( [ eq : sequence_evolution ] ) ( with @xmath220 ) by @xmath221 and integrate over @xmath45 to obtain @xmath222 there are no boundary terms due to periodic boundary conditions .
now , @xmath223 we know from lemma [ lem : weakconv ] that @xmath224 in @xmath225 .
moreover , @xmath226 is uniformly bounded so that @xmath227 .
thus , @xmath228 also , @xmath229 but @xmath230 .
hence , by the strong convergence result in lemma [ lem : cauchy ] , we have @xmath231 and thus @xmath232 combining ( [ eq : conv0 ] ) , ( [ eq : conv1 ] ) and ( [ eq : conv2 ] ) , we have @xmath233 by the weak convergence results established in lemma [ lem : weakconv ] , we also have @xmath234 putting together ( [ eq : conv_1 ] ) and ( [ eq : conv_2 ] ) , we obtain in the limit @xmath235 , @xmath236 for every @xmath213 . finally , we have to show that @xmath237 .
pick some @xmath238 with @xmath239 .
then , we have from ( [ eq : weak_form ] ) that @xmath240 similarly , we also have @xmath241 where we have used the fact that @xmath242 for all @xmath243 . taking the limit @xmath235 and comparing ( [ eq : comp_1 ] ) and ( [ eq : comp_2 ] ) , we have @xmath244 since @xmath221 is arbitrary , we conclude that @xmath237 .
this completes the proof of the existence of a weak solution .
now , we prove its uniqueness .
let @xmath245 and @xmath246 be weak solutions to ( [ eq : evo_equation ] ) and set @xmath247 .
then , for every @xmath213 , we have @xmath248 adding and subtracting @xmath249 , we obtain @xmath250 but , @xmath251 and @xmath252 so that @xmath253 now , set @xmath254 , and use theorem [ thm : evans ] , we have @xmath255 since this holds for all @xmath256 , we must have @xmath257 and hence @xmath258 for a.e . @xmath131 . but @xmath259 and the @xmath29 s are continuous in time , we have @xmath260 for all @xmath131 . finally , the energy estimate is from lemma [ lem : weakconv ] . the strong convergence result ( lemma [ lem : cauchy ] ) is important here because without it , we could not have concluded that expression ( [ eq : strongconv ] ) converges to @xmath172 , because it involves a different subsequence @xmath261 .
strong convergence ensures that all subsequences converge in @xmath262 . throughout this section we assumed that the initial condition is smooth , i.e. @xmath121 .
we can in fact relax this condition to @xmath263 by mollifying the initial data .
[ thm : existnece-1](existence and uniqueness with relaxed regularity assumption on the initial condition ) let @xmath92 , @xmath26 and @xmath263 .
then , there exists a unique weak solution @xmath216 , with @xmath217 to equation ( [ eq : evo_equation ] ) with the estimate @xmath218 let @xmath73 and consider the modified problem @xmath264\\ \rho_{n}^{\epsilon } & = \rho_{0}^{\epsilon } & & \text{on}\ u\times\left\ { t=0\right\ } \end{aligned } \end{cases}\label{eq : sequence_evolution_mollify } \end{aligned}\ ] ] where @xmath265 and @xmath266 . here
, @xmath7 is a standard positive mollifier with compact support on @xmath30 and @xmath267=1 . with mollification ,
@xmath268 is now smooth and we can apply theorem [ thm : existnece ] to conclude that there exists a unique weak solution @xmath269 , with @xmath270 to equation ( [ eq : sequence_evolution_mollify ] ) with the estimate @xmath271 but for all @xmath171 , we have @xmath272 .
hence , there exists @xmath216 , with @xmath217 , satisfying @xmath273 and a sequence @xmath274 , with @xmath275 , such that @xmath276 as @xmath277 .
we now show that @xmath29 is in fact a weak solution to ( [ eq : evo_equation ] ) . since each @xmath278 solves the weak formulation of ( [ eq : sequence_evolution_mollify ] ) ( albeit with different initial data )
, we have @xmath279 using ( [ eq : weak_conv_mollify ] ) , we can replace @xmath278 by @xmath29 in the first two integrals above in the limit @xmath277 .
moreover , as in ( [ eq : conv0 ] ) , we write the last integral as @xmath280 since @xmath281 , we have @xmath282 and hence @xmath283 next , we can write @xmath284 dxdt \nonumber \\ & = & \int_{0}^{t } \int_{u } h \left(\rho^{\epsilon_k}-\rho\right ) dy dt , \label{eq : h } \end{aligned}\ ] ] where we have defined @xmath285 clearly , @xmath286 so that in particular , @xmath287 and from ( [ eq : h ] ) we obtain @xmath288 thus , we have shown that @xmath29 satisfies @xmath289 to show that @xmath237 , we again take @xmath238 with @xmath239 . since @xmath290 uniformly , we have ( c.f . expressions ( [ eq : comp_1 ] ) and ( [ eq : comp_2 ] ) ) @xmath244 since @xmath221 is arbitrary , we have @xmath237 .
the uniqueness follows from exactly the same argument in the proof of theorem [ thm : existnece ] and we omit writing it again here .
in this section , we prove improved regularity of the weak solution to ( [ eq : evo_equation ] ) .
this allows us to put the results in section [ sec : apriori ] on a rigorous footing . as in the previous section
, we always mollify @xmath24 by @xmath291 so that the resulting evolution equations ( [ eq : sequence_evolution ] ) admit smooth solutions .
this allows us to differentiate the equation as many times as required , and we take the limit @xmath89 at the end . for simplicity of notation
, we drop the @xmath171 superscripts on @xmath41 and implicitly assume that we perform the limit at the end .
first , we prove a useful estimate . [
prop : induction_estimate]let @xmath292 .
then for @xmath293 we have the estimate @xmath294 we have @xmath295 where @xmath296 denotes the @xmath297 derivative with respect to @xmath15 . applying the leibniz rule , we have @xmath298 but , @xmath299+\int_{x - r}^{x+r}v\left(y\right)dy & i=1\\ -r\left[v^{\left(i-1\right)}\left(x+r\right)+v^{\left(i-1\right)}\left(x - r\right)\right]\\ + v^{\left(i-2\right)}\left(x+r\right)-v^{\left(i-2\right)}\left(x - r\right ) & i\geq2 \end{cases}\label{eq : diff_g_high_order}\ ] ] hence we have the bound @xmath300 for @xmath301 , @xmath302 and for @xmath303 , @xmath304 keeping only the highest sobolev norms , we have @xmath305 now , we assume that @xmath306 for some @xmath307 and prove the corresponding regularity of @xmath29 .
[ thm : x_regular](improved regularity ) let @xmath308 and suppose @xmath309 with @xmath92 and @xmath26 . then the unique solution to ( [ eq : evo_equation ] ) satisfies @xmath310 with the estimate @xmath311 where @xmath312 we prove the statements by proving uniform estimates on @xmath129 by induction on @xmath243 . the base case @xmath313 is provided in proposition [ prop : unif_l2_est ] . the @xmath314 case is proposition [ prop : unif_l2_est_higher ] .
suppose for some @xmath54 , @xmath315 for all @xmath316 .
we differentiate equation ( [ eq : sequence_evolution ] ) @xmath243 times with respect to @xmath15 , multiply it by @xmath317 and integrate over @xmath30 to get @xmath318 using proposition [ prop : induction_estimate ] with @xmath319 and @xmath320 , we have @xmath321 integrating over time , we get @xmath322^{4}\nonumber \\ & \leq & c\left(\rho_{0};k+1,t\right)^{2 } \end{aligned}\ ] ] this completes the induction . taking limits , we obtain @xmath323 with the estimate @xmath324 so
far we have only considered regularity in space .
the same can also be done in the time domain .
[ thm : reg_xt](improved regularity ) let @xmath308 and suppose @xmath325 with @xmath92 and @xmath26 .
then , \(i ) for every @xmath326 , the unique solution to ( [ eq : evo_equation ] ) satisfies @xmath327 with the estimate @xmath328 where @xmath329 \(ii ) moreover , @xmath330 with the estimate @xmath331 let us prove that for all @xmath332 , @xmath333 where we have defined @xmath334 this is done by induction on @xmath335 up to @xmath243 .
the case @xmath336 is theorem [ thm : x_regular ] .
suppose we have for some @xmath337 the estimate ( [ eq : inductive_hyp-1 ] ) .
differentiating equation ( [ eq : sequence_evolution ] ) @xmath335 times with respect to @xmath48 and using the leibniz rule , we have @xmath338 where we used the shorthand @xmath339 .
thus , we have @xmath340 using proposition [ prop : induction_estimate ] with @xmath341 and @xmath342 , we have @xmath343 integrating over time then gives @xmath344 since @xmath345 , we can apply the inductive hypothesis ( [ eq : inductive_hyp-1 ] ) to conclude that @xmath346 similarly , @xmath347 this completes the induction on @xmath335 up to @xmath243 .
putting @xmath348 into ( [ eq : inductive_hyp-1 ] ) and taking limits proves part ( i ) . to prove the second part , notice that @xmath349 hence , @xmath350 taking limits then proves part ( ii ) .
[ cor : last_reg ] let @xmath21 and @xmath351 with @xmath92 and @xmath26 .
then the unique solution to ( [ eq : evo_equation ] ) satisfies @xmath352 after possibly being redefined on a set of measure zero . by theorem [ thm : x_regular ] ,
@xmath353 , i.e. @xmath354 . hence there exists a version of @xmath29 with @xmath355 , so that in particular , @xmath356 .
next , using theorem [ thm : reg_xt ] , we have @xmath357 and @xmath358 , hence by theorem [ thm : evans ] there is a version of @xmath29 so that @xmath359 .
hence we have @xmath352 up to a set of measure zero .
this result allows us to restate the results in section [ sec : apriori ] without the a priori smoothness assumption .
we summarize the main results of this paper in the following : [ thm : mainresult ] let @xmath351 with @xmath92 and @xmath26 .
then , there exists a unique weak solution @xmath29 to equation ( [ eq : evo_equation ] ) , with \(i ) ( regularity ) @xmath360
\(ii ) ( nonnegativity ) @xmath94 for all @xmath72 .
\(iii ) ( stability ) furthermore , if @xmath34 , then @xmath35 in @xmath114 exponentially as @xmath36 . existence and uniqueness follow from theorem [ thm : existnece-1 ] .
( i ) follows from corollary [ cor : last_reg ] .
having established ( i ) , ( ii ) and ( iii ) then follows from corollary [ cor : positivity ] and theorem [ thm : apriori_stability ] respectively .
this work was completed while q. jiu was visiting the department of mathematics at princeton university .
the authors are grateful for many discussions with prof . | this paper establishes the global well - posedness of the nonlinear fokker - planck equation for a noisy version of the hegselmann - krause model .
the equation captures the mean - field behavior of a classic multiagent system for opinion dynamics .
we prove the global existence , uniqueness , nonnegativity and regularity of the weak solution .
we also exhibit a global stability condition , which delineates a forbidden region for consensus formation .
this is the first nonlinear stability result derived for the hegselmann - krause model .
hegselmann - krause model , nonlinear fokker - planck equation , well - posedness , global stability 35q70 , 35q91 |
this study was supported by the e.u . strep macomufi ( 033221 ) ,
the contract femmes of the agence nationale pour la recherche .
also acknowledges financial support from the conseil gnral de lessone . the authors would like to thank h. jaffrs , m. viret , g. catalan and j. scott for fruitful discussions . | we have combined neutron scattering and piezoresponse force microscopy to study the relation between the exchange bias observed in cofeb / bifeo@xmath0 heterostructures and the multiferroic domain structure of the bifeo@xmath0 films .
we show that the exchange field scales with the inverse of the ferroelectric and antiferromagnetic domain size , as expected from malozemoff s model of exchange bias extended to multiferroics .
accordingly , polarized neutron reflectometry reveals the presence of uncompensated spins in the bifeo@xmath0 film at the interface with the cofeb . in view of these results we discuss possible strategies to switch the magnetization of a ferromagnet by an electric field using bifeo@xmath0 .
the renaissance of multiferroics @xcite , i.e. materials in which at least two ferroic or antiferroic orders coexist , is motivated by fundamental aspects as well as their possible application in spintronics @xcite .
such compounds are rare and the very few that possess simultaneously a finite magnetization and polarization usually order below about 100k @xcite .
ferroelectric antiferromagnets ( feaf ) are less scarce , and some exhibit a coupling between their two order parameters .
this magnetoelectric ( me ) coupling allows the reversal of the ferroelectric ( fe ) polarization by a magnetic field @xcite or the control of the magnetic order parameter by an electric field @xcite .
the practical interest of conventional antiferromagnets ( af ) is mainly for exchange bias in spin - valve structures .
the phenomenon of exchange bias ( eb ) @xcite manifests itself by a shift in the hysteresis loop of a ferromagnet ( fm ) in contact with an af and arises from the exchange coupling at the fm / af interface @xcite .
combining this effect with the me coupling in a feaf / fm bilayer can allow the reversal of the fm magnetization via the application of an electric field through the feaf , as reported recently at 2k in ymno@xmath0/nife structures @xcite .
to exploit these functionalities in devices one needs to resort to feaf materials with high transition temperatures .
bifeo@xmath0 ( bfo ) is a fe perovskite with a curie temperature of 1043k @xcite that orders antiferromagnetically below t@xmath1=643k ( t@xmath1 : nel temperature ) @xcite .
bfo thin films have a very low magnetization ( @xmath20.01 @xmath3/fe ) compatible with an af order @xcite , and remarkable fe properties with polarization values up to 100 @xmath4c.@xmath5 range @xcite .
recently , we reported that bfo films can be used to induce an eb on adjacent cofeb layers at room temperature @xcite .
this observation together with the demonstration of a coupling between the af and fe domains @xcite paves the way towards the room - temperature electrical control of magnetization with bfo .
however , several questions remain before this can be achieved .
key issues concern the precise magnetic structure of bfo thin films , and the mechanisms of eb in bfo - based heterostructures . in this letter , we report on the determination of the magnetic structure of bfo films by means of neutron diffraction ( nd ) , and the analysis of the eb effect in cofeb / bfo heterostructures in terms of malozemoff s model @xcite . accordingly , we find a clear dependence of the amplitude of the exchange field h@xmath6 with the size of the multiferroic domains , which provides a handle to control the magnetization of the cofeb film by an electric field .
the observation of eb and enhanced coercivity correlates with the presence of uncompensated spins at the interface between the fm and the af , as detected by polarized neutron reflectometry ( pnr ) .
bifeo@xmath0 films were grown by pulsed laser deposition @xcite , directly onto ( 001)- or ( 111)-oriented srtio@xmath0 ( sto ) or ( 001)-oriented laalo@xmath0 ( lao ) substrates , or onto 10 - 25 nm - thick metallic buffers of la@xmath7sr@xmath8mno@xmath0 ( lsmo ) or srruo@xmath0 ( sro ) @xcite .
7.5 nm - thick cofeb layers were sputtered in a separate chamber at 300k in a magnetic field of 200 oe , after a short plasma cleaning .
the samples were capped by 10 - 30 nm of au .
high resolution x - ray diffraction evidenced a cube - on - cube growth for all the perovskite layers onto the substrates .
cofeb was amorphous . while the ( 111)-oriented films were found to be rhombohedral as bulk bfo @xcite , ( 001)-oriented films were found tetragonal or monoclinic @xcite .
a first key information that is usually required to analyze eb is the magnetic structure of the af .
bulk bfo is known to have a g - type af order @xcite , with a superimposed cycloidal modulation @xcite . in view of the strong strain sensitivity of the properties of fe and magnetic oxides
, one can anticipate that the magnetic order of bfo films might be different from that of the bulk . in order to determine their magnetic structure , selected ( 001)- and
( 111)-oriented bfo films were thus characterized by nd with the triple axis 4f1 spectrometer at the orphe reactor of the laboratoire lon brillouin ( llb ) @xcite . in a g - type af
, superstructure peaks are expected to appear at [ ] -type reflections . in fig .
[ neutrons - mh]a and b , we show the diffracted intensity at the [ ] reflection in bfo films grown on ( 001)- and ( 111)-oriented sto .
clearly an af peak is present for both films . on the other hand ,
no intensity was measured at [ 0 0 ] -type or [ 0]-type reflections , characteristic of a - type and c - type antiferromagnetism , respectively .
these data thus show that both ( 001)- and ( 111)-oriented films are antiferromagnetic with a g - type order similar to that of the bulk .
in other words , neither strain nor changes in the unit - cell symmetry modify the type of magnetic order , besides destroying the cycloidal modulation ( see ref . for details on this aspect ) . within a simplistic model of eb
, the exchange field h@xmath6 depends on the interface coupling j@xmath9=@xmath10 ( j@xmath11 is the exchange parameter , s@xmath12 and s@xmath13 are the spin of the interfacial atoms in the ferromagnet and the af , respectively , and @xmath14 is the unit cell parameter of the af ) , on the magnetization and thickness of the ferromagnet m@xmath15 and t@xmath15 , and on the anisotropy and the thickness of the af k@xmath13 and t@xmath13 @xcite .
@xmath16 provided that 1/@xmath17 is smaller than 1 ( otherwise h@xmath6 is zero ) .
if t@xmath13 is large , @xmath18 and thus h@xmath19 . in a naive picture of perfect surfaces and because the af order is g - type in both ( 001 ) and ( 111 ) films , the ( 111 ) films are expected to have magnetically uncompensated surfaces , yielding s@xmath13=5/2 , while the ( 001 ) films should have compensated surfaces , yielding in average s@xmath13=0 . therefore , a finite h@xmath6 should be found for ( 111 ) films only . for ( 111 ) films , taking j@xmath11=5 10@xmath20 j @xcite ,
a=3.96 @xmath21 , s@xmath22=1/2 , s@xmath13=5/2 , m@xmath15=800 ka / m and t@xmath15=7.5 nm one can estimate j@xmath9=4 mj / m@xmath23 and the large value of h@xmath6=6.5 koe .
reflection for ( 001)- and ( 111)-oriented bfo films on sto , at 300k .
magnetic field dependence of the magnetization of bfo / cofeb ( c , d ) and lsmo / bfo / cofeb samples ( e , f ) grown on ( 001)-oriented sto ( c , e ) and ( 111)-oriented sto ( d , f ) , at 300k . ] figure [ neutrons - mh]c and d show m(h ) hysteresis loops measured at 300k for bfo(70nm)/cofeb stacks grown on sto(001 ) and sto(111 ) .
the loops are shifted towards negative magnetic field values by an exchange field h@xmath6 of -39 oe for ( 001 ) films and -19 oe for ( 111 ) films . furthermore the loops are enlarged by some tens of oe compared to those measured on cofeb single films @xcite . from m(h ) data for lsmo / bfo / cofeb samples ( see fig .
[ neutrons - mh]e and f ) , a symmetric ( i.e. not showing eb ) contribution from the lsmo films is visible ( the curie temperature of our lsmo films is about 330k ) in addition to that coming from the cofeb .
for a given thickness , the exchange field experienced by the cofeb is virtually the same irrespective of the presence of the lsmo buffer layer .
malozemoff s random field model @xcite has been proposed to resolve the long - standing discrepancy between the model of eq.([he ] ) and the experimental data @xcite .
it considers that in the presence of some atomic - scale disorder that locally creates a net magnetization in the af at the interface with the fm , the af splits into domains , which decreases considerably the interface coupling that now writes : @xmath24 l is the antiferromagnetic domain size and @xmath25 a factor depending on the shape of the domains and on the average number @xmath26 of frustrated interaction paths for each uncompensated surface spin @xcite . for hemispherical bubble domains , @xmath27 and @xmath26
is of order unity @xcite . from equations
( [ he ] ) and ( [ jeb - malo ] ) and for large af thickness ( @xmath18 ) the exchange field should vary as @xmath28 : @xmath29 and indeed , a linear dependence of h@xmath6 with 1/l was experimentally observed in several af / fm systems @xcite , providing strong support to the model . in order to determine the domain size in antiferromagnets ,
a technique of choice is x - ray photoelectron emission microscopy @xcite .
alternatively , an estimation of the average domain size can be inferred from the width of the nd peaks that reflects the coherence length in the sample @xcite . in some multiferroics like bfo , the fe and af domains
are coupled @xcite so that it is possible to infer the size and distribution of the af domains by imaging the fe domains , e.g. using piezoresponse force microscopy ( pfm ) .
we have characterized the fe domains in two sets of bfo samples by combining in - plane and out - of - plane pfm measurements @xcite .
a first set consists of @xmath265 nm - thick bfo films grown on different buffers and substrates ( see table [ tableau ] ) .
a second set consists of bfo films with varying thickness ( 5 nm @xmath30 240 nm ) grown on lsmo//sto(001 ) . in bfo , since the polarization is oriented along the @xmath31111@xmath32 directions there generally exist 8 possible polarization orientation variants .
this leads to a large number of possible domain patterns and to several types of domain walls ( dws ) depending on the angle between the polarization vectors in the adjacent domains ( 71 , 109 or 180@xmath33 ) . in principle
only 71@xmath33 and 109@xmath33 fe dws correspond to an af dw @xcite . in most of the samples considered in this study ,
the three types of dws are present , with the density of 180@xmath33 type dws being negligible . in the following
we analyze the exchange field in terms of the average fe domain size l@xmath34 that we identify to half the fe domain periodicity , and the average af domain size l@xmath13 that we identify to the coherence length in the nd experiments . even in the case of a strict correspondence between the fe and af domains ,
l@xmath34 is expected to be larger than l@xmath13 because it comprises both the domain and the dw widths while l@xmath13 mostly reflects the domain width . [ !
h ] ccccc substrate & buffer & @xmath35(nm ) & h@xmath6 & l@xmath34(nm ) + sto(001 ) & lsmo & 70 & -39 & 58 + sto(001 ) & sro & 70 & -14.5 & 98 + sto(111 ) & lsmo & 70 & -19 & 68 + sto(111 ) & sro & 70 & -39 & 48
+ lao(001 ) & lsmo & 60 & 0 & 350 + lao(001 ) & sro & 60 & -29 & 55 + 65 nm thick bfo films .
( b ) thickness dependence of the exchange field for cofeb / bfo stacks grown on sto(001 ) .
the dotted lines are guides to the eye . ]
let us first compare the values of l@xmath13 and l@xmath34 for the samples of fig .
[ neutrons - mh ] . from fig .
[ neutrons - mh]a we calculate l@xmath13=42 nm while from pfm we find l@xmath34=58 nm for this sample . for the sample of fig .
[ neutrons - mh]b , l@xmath13=56 nm and l@xmath34=98 nm . l@xmath34 is larger than l@xmath13 by some tens of nm that likely correspond to the dw width @xcite .
when l@xmath13 increases l@xmath34 increases also , as expected if the af and fe domains are correlated , as found by zhao _
et al _ @xcite . in figure
[ dep]a we plot the exchange field as a function of the inverse of the domain size for the first set of samples ( constant thickness ) .
a linear variation of h@xmath6 with @xmath28 is observed , as expected from eq .
( [ he - malo ] ) .
furthermore , there is an excellent _ quantitative _ agreement between the model and the data as illustrated by the similarity between the experimental points ( solid symbols in fig .
[ dep]a ) and the values of h@xmath6 calculated using the domain sizes ( open symbols ; the only free parameter is @xmath25 that we set to 3.2 ) . as illustrated by fig .
[ dep]b , this model also accounts for the thickness dependence of the exchange field . below a critical bfo thickness of about 10 nm ,
there is no exchange bias but at larger thickness , h@xmath6 increases abruptly and takes values of about 40 oe for t@xmath36 nm .
a similar thickness dependence was reported for other af / fm systems such as femn / nife @xcite and is expected from eq .
( [ he ] ) .
combining eq .
( [ he ] ) and eq .
( [ jeb - malo ] ) , and using the measured domain sizes one can calculate h@xmath6 for these films .
the only free parameters are @xmath25 and k@xmath13 . as shown by the open symbols in fig .
[ dep]b , a rather good agreement with the data is obtained for @xmath25=3.2 ( as before ) and k@xmath13=6.8 kj / m@xmath37 .
we note that this value of k@xmath13 is lower than the value inferred for the bulk from electron spin resonance experiments @xcite by about one order of magnitude
. this can be due to the low thickness of the bfo films and possibly to strain effects , or reflect modified magnetic properties at the bfo / cofeb interface , as will be discussed in the following .
: squares ; r@xmath38 : circles ) fitted with a three - layer model ( a ) and a four - layer model ( b ) . introducing an interface layer of @xmath22 nm carrying a moment of 1@xmath39 @xmath3/f.u . significantly improves the quality of the fit for up - up reflectivity ( see arrows ) . ] to validate the analysis of our data with malozemoff s model , we have attempted to detect the presence of a net magnetization in bfo close to the interface with cofeb using pnr .
the pnr measurements were carried out with the prism instrument of the llb at room temperature .
spin - up and spin - down reflectivities ( r@xmath40 and r@xmath38 ) were collected and the data were corrected for the polarization efficiency .
the least square fittings were made using the simul - reflec software .
the structural parameters of the layers were first determined by fitting x - ray reflectometry data ( not shown ) . in the following
we discuss the results for a au(30nm)/cofeb(7.5nm)/bfo(18nm)//sto(001 ) sample displaying an eb of 18 oe and in which the cofeb / bfo interface has a roughness of only 0.5 nm , yielding high quality reflectivity data .
figure [ refl ] shows the pnr results and the corresponding fits using two different sample models . in the first sample model ( fig .
[ refl]a ) we only consider the presence of three layers , i.e. au , cofeb and bfo .
the best fit is obtained for the structure au(31.0@xmath39nm)/cofeb(7.0@xmath39nm)/bfo(18@xmath39 nm ) with zero magnetization in the bfo and au layers , and a magnetization of 1@xmath41 @xmath3/f.u . for the cofeb . as can be appreciated ,
the fit is good but not perfect ( see arrows ) .
the fit quality is significantly improved , especially at high @xmath42 , by using a four - layer model , i.e. splitting the bfo layer in two .
the best fit is then obtained if a 2@xmath39 nm layer carrying a magnetic moment of 1@xmath39@xmath3/f.u .
is present in the bfo , at the interface with the cofeb .
this ultrathin magnetic layer inside the bfo accounts for the presence of a large density of uncompensated spins ( corresponding to a surface moment m@xmath43=31.8 @xmath3.nm@xmath44 ) at the bfo / cofeb interface , similarly to results reported on the co / lafeo@xmath0 exchange bias system @xcite .
the expected surface moment due to pinned uncompensated spins within malozemoff s model is m@xmath45=2s@xmath13/[email protected] @xmath3.nm@xmath44 , which represents only a small fraction of the surface moment measured by pnr .
the majority of uncompensated spins is thus unpinned ( as observed by x - ray magnetic circular dichroism in co / irmn @xcite ) and rotates with the cofeb , producing an increase of the coercive field , as found experimentally ( see fig .
[ neutrons - mh ] ) .
this observation suggests that two different , yet possibly related and complementary , strategies are possible to tune the magnetic switching fields of the ferromagnet electrically .
one would rely on the manipulation of the pinned uncompensated spins to modify the exchange field h@xmath6 , e.g. by changing the domain size that should be controllable by ad - hoc electrical writing procedures .
the other could consist in controlling the unpinned uncompensated spins in order to alter the coercive field h@xmath47 .
this might be achieved by modifying the effective surface anisotropy of the af , for instance by playing with the ferroelastic energy of the domains , which would change the magnetoelastic contribution to the anisotropy . in summary , we analyzed the exchange bias in the cofeb / bifeo@xmath0 system and found that the exchange field does not correlate with the type of magnetic surface of the antiferromagnet - compensated or uncompensated .
rather , the exchange field scales with the inverse of the ferroelectric and antiferromagnetic domain size in the multiferroic bifeo@xmath0 film , as expected from malozemoff s model that we extend for the first time to the case of ferroelectric antiferromagnets .
polarized neutron reflectometry measurements reveal the presence of a net magnetic moment within a @xmath22 nm slab in the bfo at the interface with the cofeb , reflecting the presence of uncompensated spins in the bifeo@xmath0 , consistent with the observation of exchange bias and enhanced coercivity . as the ferroelectric domain structure
can be easily controlled by an electric field , our results strongly suggest that the electrical manipulation of magnetization should be feasible at room temperature in bfo - based exchange - bias heterostructures . |
linear colliders offer unique opportunities to study photon - photon collisions obtained using the process of compton backscattering of laser light off electron beams from the linear collider @xcite .
this option is included now in conceptual design reports of the nlc , jlc and tesla / sblc projects @xcite .
the expected physics at the photon linear collider ( plc ) is very rich and complementary to that in @xmath3 collisions ( see _ e.g. _ @xcite and references therein ) . without the discovery of a light higgs boson at lep2 , lhc or the linear collider , the best strategy to probe the symmetry breaking sector would lie in the study of the self couplings of the @xmath2 , @xmath4 bosons .
the large cross sections for the processes involving @xmath2 s make @xmath5 colliders especially attractive tool in probing the @xmath2 self couplings .
indeed , the reaction @xmath6 would be the dominant source of the @xmath1 pairs at future linear colliders , provided that photon - photon collider option will be realized .
the born cross section of @xmath1 pair production in photon - photon collisions in the scattering angle interval @xmath7 is 61 pb at @xmath8 gev and 37 pb at 1 tev .
corresponding cross sections of @xmath1 pair production in @xmath3 collisions are an order of magnitude smaller : 6.6 pb at 500 gev and 2.5 pb at 1 tev . with more than a million @xmath9 pairs per year a photon - photon collider
can be really considered as a @xmath2-factory and an ideal place to conduct precision tests on the anomalous triple @xcite and quartic @xcite couplings of the @xmath2 bosons .
in addition , in the process of triple @xmath10 vector boson production it is possible to probe the tri - linear @xmath11 and quartic couplings @xcite as well as the @xmath12 violating anomalous @xmath11 , @xmath13 interactions @xcite . with the natural order of magnitude on anomalous couplings one needs to know the @xmath14 cross sections with a precision better than 1% to extract these small numbers . from a theoretical point of view
this calls for the very careful analysis of at least @xmath15 corrections to the cross section of @xmath1 pair production in @xmath5 collisions , which were recently calculated including virtual corrections @xcite and complete @xmath15 corrections taking into account both virtual one - loop corrections and real photon and @xmath4-boson emission @xcite .
inclusive cross section of @xmath1 pair production in photon - photon collisions to third order in @xmath16 is given by the ir - finite sum of born cross section , interference term between the born and one - loop amplitudes and cross section of @xmath1 pair production accompanied by the real photon emission . at energies above the @xmath10 threshold one should add the cross section of @xmath17 production .
@xmath18 in contrast to the corrections to the @xmath4 boson production at lep1 , as a consequence of the universality of the electric charge no power or logarithmically enhanced corrections involving @xmath19 , @xmath20 , @xmath21 , @xmath22 or @xmath23 appear in the on - shell scheme in the limit , when higgs - boson or top - quark masses are much larger than the collision energy or the fermion mass @xmath24 is much smaller than @xmath25 @xcite .
figure 1 shows total cross section of @xmath9 pair production summed over @xmath9 and @xmath26 final states and integrated over @xmath27 scattering angles in the interval @xmath28 as a function of energy for various polarizations . the bulk of the cross section originates from transverse @xmath29 pair production .
transverse @xmath2 s are produced predominantly in the forward / backward direction and the helicity conserving amplitudes are dominating .
cross sections integrated over the whole phase space are non - decreasing with energy . for a finite angular cutoff they do decrease as @xmath30 , but still they are much larger than suppressed cross sections . for the dominating @xmath31 , @xmath32 , @xmath33 helicity configurations corrections are negative and they rise with energy ranging from @xmath34 at 500 gev to @xmath35 at 2 tev .
the correction for the next to the largest cross section @xmath36 is also negative . for the other helicities radiative corrections are positive at high energies due to dominating positive contribution of real photon emission .
radiative corrections to cross sections @xmath37 , which are decreasing at born level as @xmath38 , are positive and large . for the cross sections @xmath39 and @xmath40 , which are decreasing as @xmath41 for a finite angular cutoff , corrections are even larger .
the cross section @xmath42 is decreasing at tree level as @xmath43 for a finite angular cutoff and is quite negligible at high energy .
the cross sections @xmath44 and @xmath45 vanish at the born level , so only the process of @xmath26 production contributes in figure 1 .
the cross sections @xmath42 and @xmath39 exhibit a clear clear higgs resonance peak and the interference pattern , respectively , at @xmath46 gev .
( 5.4,3 ) ( -.1,0 ) ( 2.7,0 ) .unpolarized born cross sections and relative corrections for various intervals of @xmath27 scattering angles .
corrections originating from real hard photon ( @xmath47 gev ) and @xmath4-boson emission as well as sum of soft photon and virtual boson contributions , fermion virtual corrections and total corrections are given separately .
[ cols="^,^,^,^,^,^,^ " , ] in table 1 born cross sections and relative corrections are given for several intervals of @xmath27 scattering angles . at high energies
large cancellations occur between negative virtual corrections and positive corrections corresponding to real photon or @xmath4-boson emission . consequently , although the correction originating from the @xmath10 production is completely negligible at @xmath48 tev , it is of the same order of magnitude as hard photon correction at 2 tev .
although at @xmath49 gev corrections are quite small ranging from @xmath50 to @xmath51 , depending on angular cuts , at tev energies the value of radiative corrections in the central region of @xmath1 production become quite large , so that corrections in the region @xmath52 are @xmath53 times larger than the corrections to the total cross section at @xmath54 tev .
they range from @xmath55 to @xmath56 .
thus if precision measurements are to be made at tev energy more careful theoretical analysis is needed in order to reliably predict the value of the cross section in the central region where the value of the cross section is the most sensitive to the @xmath2 anomalous couplings .
although the cross section of @xmath9 production is much larger in @xmath5 collisions , this fact itself is not to be considered as an obvious advantage of plc .
the reason is that although the anomalous contribution to the amplitude of longitudinal @xmath57 pair production is enhanced by a factor of @xmath58 both in for @xmath59 ( for @xmath60 in @xmath5 collisions anomalous coupling contribution is not enhanced at all ) and @xmath61 collisions , the @xmath14 born amplitude of @xmath57 production at plc is suppressed as @xmath62 , so that the contribution of the interference term to the cross section is decreasing as @xmath30 at plc @xcite . on the contrary , in @xmath61 collisions
the anomalous contribution to the cross section is enhanced , corresponding to non - decreasing cross section of @xmath57 production .
recently the authors of ref .
@xcite have demonstrated that enhanced coupling could still be exploited in the mode .
their clever idea is to reconstruct the non diagonal elements of the @xmath9 polarization density matrix by analyzing the distributions of the decay products of the @xmath2 s , thereby achieving the improvement over simple counting rate method of more that an order of magnitude at @xmath63 tev .
however , although the benefits from mode are really quite evident at @xmath64 gev and allowed region in the anomalous couplings parameter space shrinks considerably when results from @xmath65 are combined with ones from @xmath6 , at energies above 1 tev combining results from @xmath61 and modes does not considerably reduce the bounds obtained from @xmath66 alone @xcite .
this is especially true for fits with one anomalous coupling .
qualitatively these results can be understood considering the ratio @xmath67 as a measure of statistical significance of the anomalous coupling signal @xmath68 with respect to the @xmath14 background @xmath69 . since the total @xmath14 cross section is decreasing as @xmath30 in @xmath61 collisions and
is constant in collisions , while the enhanced anomalous cross section behaves like a constant we get , while @xmath70 .
if we take into account that anomalous couplings affect mostly the cross section in the central region , where the @xmath14 cross section behaves like @xmath71 , we get p_t , the same improvement at higher energy as for @xmath61 collisions but only for large values of @xmath72 cut @xmath73 , with which the cross section of @xmath9 production in collisions is not enhanced any more with respect to production in @xmath61 collisions .
moreover , it is this region , where radiative corrections are as large as @xmath74 . to resume , at 500 gev linear collider much stronger constraints on the anomalous @xmath2 couplings
could be obtained combining the bounds from @xmath66 and @xmath6 . at such an energy radiative corrections
are well under control and photon - photon collider option is essential for precision measurements . at energies above 1 tev , where enhancement factor @xmath58 for anomalous amplidudes starts playing a crucial role
, plc is not so much advantageous with respect to @xmath3 collider at the same integrated lumonosity .
radiative corrections are quite large and theoretical consideration beyond one - loop approximation is needed in order to safely predict @xmath14 cross section . nevertheless ,
if in the photon - photon collisions one can obtain much larger luminosity than in @xmath3 collisions @xcite , the plc potential could still be as high as at lower energies .
i am grateful to the organizers of the workshop for warm hospitality and to f. boudjema , i. ginzburg and v. telnov for fruitful discussions .
this work has been supported by the alexander von humboldt foundation .
m. baillargeon , g. blanger , f. boudjema and g. jikia , proceedings of the 3rd workshop _
@xmath3 collisions at tev energies : the physics potential , part d _ , desy , hamburg , germany , august 30 september 1 , 1995 , p. 511 . | recent progress in calculating @xmath0 electroweak corrections to @xmath1 pair production in photon - photon collisions is reviewed .
the potential of the photon linear collider to study anomalous @xmath2 couplings is discussed . |
the omnipresence of 1/f noise is one of the oldest puzzle of the contemporary physics . during more than 70 years since the first observation by johnson , long - memory processes with long - term correlations
have been observed in many types of systems from physics , technology , biology , astrophysics , geophysics and sociology ( see @xcite and references herein ) .
recently 1/f noise was discovered in human cognition @xcite , human coordination @xcite and even in distribution of prime numbers @xcite . despite of the widespread occurrence of fluctuations of signals and variables exhibiting @xmath0 ( @xmath1 ) behavior of the power spectral density @xmath2 at low frequencies in large diversity of systems no generally recognized explanation of the ubiquity of 1/f noise is still proposed .
physical models of 1/f noise in some physical systems are usually very specialized , complicated ( see @xcite and references herein ) and they do not explain the omnipresence of the processes with @xmath0 spectrum @xcite .
note also some mathematical analysis @xcite , models and algorithms of generation of the processes with 1/f noise @xcite .
these models also expose some shortcomings : they are very specific , formal ( like fractional brownian motion or half - integral of a white noise signal ) or unphysical .
they can not , as a rule , be solved analytically and they do not reveal the origin as well as the necessary and sufficient conditions for the appearance of 1/f type fluctuations .
in such a situation the simple analytically solvable model system generating 1/f noise may essentially influence in the reveal of the origin and essence of the effect . here
we present a model which generates 1/f noise in any desirably wide range of frequency .
our model is a result of the search of necessary and sufficient conditions for the appearance of 1/f fluctuations in simple systems affected by the random external perturbations initiated in @xcite and originated from the observation of the transition from chaotic to nonchaotic behavior in the ensemble of randomly driven systems @xcite .
contrary to the mcwhorter model @xcite based on the superposition of large number of lorentzian spectra and requiring a very wide distribution of relaxation times , our model contains only one relaxation rate @xmath3 and can have an exact 1/f spectrum in any desirably wide range of frequency .
the model may be used as a basis for the checking of assumptions made in the derivation of 1/f noise spectrum for different systems .
furthermore , it allows us to make a heuristic presumption for the generalizations of the theory of 1/f noise .
numerical simulations and comparisons with analytical results confirm this supposition .
in many cases , the intensity of some signal or current can be represented by a sequence of random ( however , as a rule , mutually correlated ) pulses or elementary events @xmath4 . here
the function @xmath5 represents the shape of the @xmath6 pulse having an influence to the signal @xmath7 in the region of transit time @xmath8 .
the signal or intensity of the current of particles in some space cross - section may , therefore , be expressed as @xmath9 it is easy to show that the shapes of the pulses influent mainly on the high frequency , @xmath10 with @xmath11 being the characteristic pulse length , power spectral density while fluctuations of the pulse amplitudes result , as a rule , in the white or lorentzian but not 1/f noise @xcite .
therefore , we restrict our analysis to the noise due to the correlations between the transit times @xmath8 . in such an approach
we can replace the function @xmath4 by the dirac delta function @xmath12 and the signal express as @xmath13 this model also corresponds to the flow of identical objects : electrons , photons , cars and so on . on the other hand , fluctuations of the amplitudes @xmath14 may result in the additional noise but can not reduce 1/f noise we are looking for .
the power spectral density of the current ( 2 ) is @xmath15 where @xmath16 is the whole observation time interval , @xmath17 and @xmath18 are minimal and maximal values of index @xmath6 in the interval of observation and the brackets @xmath19 denote the averaging over realizations of the process . in this approach
the power spectral density of the signal depends on the statistics and correlations of the transit times @xmath8 only .
it is well known that sequence of random , poisson , transit times generates white ( shot ) noise @xcite .
the sequence of transit times @xmath8 with random increments , @xmath20 ( where @xmath21 is the average time interval between pulses , @xmath22 denotes the sequence of uncorrelated normally distributed random variables with zero expectation and unit variance , i.e. the white noise source , and @xmath23 is the standard deviation of white noise ) results in the lorentzian spectra @xcite .
here we will consider sequences of the transit times with random increments of the time intervals between pulses , @xmath24 , where @xmath25 .
it is natural to restrict in some way the infinite brownian increase or decrease of the intervals @xmath26 , e.g. by the introduction of the relaxation to the average period @xmath27 rate @xmath3 .
so , we have the recurrent equations for the transit times @xmath28 the simplest physical interpretation of the model ( 4 ) corresponds to one particle moving in the closed contour with the period of the drift of the particle round the contour fluctuating ( due to the external random perturbations ) about the average value @xmath27 @xcite .
an advantage of the model ( 4 ) is that it may be solved analytically .
so , an iterative solution of eqs . ( 4 ) results in an expression for the period @xmath29 where @xmath30 is the initial period . the dispersion of the period @xmath26 is @xmath31 } { 2\gamma \left ( 1-\gamma /2\right ) } \simeq \left\ { \begin{array}{ll } \sigma ^2k , & 2k\gamma \ll 1 \\
\sigma ^2/2\gamma , & 2k\gamma \gg 1 .
\end{array } \right .
\eqno{(6)}\ ] ] therefore , after the characteristic transition to the stationary process time , @xmath32 , the dispersion of the period approaches the limiting value @xmath33 .
after some algebra we can also obtain an explicit expression for the transit times @xmath8 ( @xmath34 ) , @xmath35 \left ( \tau _ 0-\bar \tau \right ) + \frac \sigma \gamma \sum_{l=1}^k\left [ 1-\left ( 1-\gamma \right ) ^{k+1-l}\right ] \varepsilon _ l , \eqno{(7)}\ ] ] where @xmath36 is the initial time .
the dispersion of the transit time @xmath8 is @xmath37 + \left ( 1-\gamma \right ) ^2\frac{1-\left ( 1-\gamma \right ) ^{2k}}{1-\left ( 1-\gamma \right ) ^2}\right\}\ ] ] @xmath38 at @xmath39 eq .
( 7 ) generates the stationary time series .
the difference of the transit times @xmath40 and @xmath8 in eq .
( 3 ) for @xmath41 or @xmath42 is @xmath43 \sum_{l=1}^k\left ( 1-\gamma \right ) ^{k+1-l}\varepsilon _ l+\sum_{l = k+1}^{k+q}\left [ 1-\left ( 1-\gamma \right ) ^{k+q+1-l}\right ] \varepsilon _ l\right\ } , q\geq 0.\eqno{(9)}\ ] ] the dispersion of this times difference equals @xmath44 where @xmath45 ^2\sum_{l=1}^k\left ( 1-\gamma \right ) ^{2l}+\sum_{l=1}^q\left [ 1-\left ( 1-\gamma \right ) ^l\right ] ^2\right\ } , \quad q\geq 0.\eqno{(11)}\ ] ] summation in eq . ( 11 ) results in @xmath46 } { 1-\left ( 1-\gamma \right ) ^2 } \left\ { 2+\left [ 1-\left ( 1-\gamma \right ) ^q\right ] \left ( 1-\gamma \right ) ^{2k+1}\right\ } \right\ } .\eqno{(12)}\ ] ] at @xmath47 @xmath48 while for @xmath42 we have @xmath49 } { 1-\left ( 1-\gamma \right ) ^2 } \right ] \eqno{(13)}\ ] ] @xmath50 note that for @xmath51 we should replace in eq .
( 9)(13 ) @xmath52 by @xmath53and @xmath6 by @xmath54 .
therefore , the function @xmath55 at @xmath56 is even , i.e. @xmath57 . substituting eq .
( 9 ) into eq .
( 3 ) and replacing the summations in the exponents by the multiplications of the exponents we have the following expression for the power spectral density of the current @xmath58 \left ( 1-\gamma \right ) ^{k+1-l}\varepsilon _ l\right\ } \right . \\
\prod\limits_{l = k+1}^{k+q}\exp \left\ { i\frac{2\pi f\sigma } \gamma \left [ 1-\left ( 1-\gamma \right ) ^{k+q+1-l}\right ] \varepsilon _ l\right\ } \right\rangle .
\end{array } \eqno{(3a)}\ ] ] the average over realizations of the process coincides with the average over the distribution of the random variables @xmath59 .
using the fact that random variables @xmath59 are independent and mutually uncorrelated we can fulfill the averaging over every random variable @xmath60 independently .
therefore , eq . ( 3a ) may be rewritten in the form @xmath61 \left ( 1-\gamma \right ) ^{k+1-l}\varepsilon _ l\right\ } \right\rangle \\ \prod\limits_{l = k+1}^{k+q}\left\langle \exp \left\ { i\frac { 2\pi f\sigma } \gamma \left [ 1-\left ( 1-\gamma \right ) ^{k+q+1-l}\right ] \varepsilon _ l\right\ } \right\rangle .
\end{array } \eqno{(3b)}\ ] ] the result of the averaging of the exponent @xmath62 ( with @xmath63 being a constant ) over the normally distributed random variable @xmath59 with zero expectation and unit variance is @xmath64 therefore , after the averaging over the normal distribution of the random variables @xmath59 eq .
( 3b ) takes the form @xmath65 ^2\left ( 1-\gamma \right ) ^{2\left ( k+1-l\right ) } \right\ } \\ \prod\limits_{l = k+1}^{k+q}\exp \left\ { -\frac{2\pi ^2f^2\sigma ^2}{\gamma ^2}\left [ 1-\left ( 1-\gamma \right ) ^{k+q+1-l}\right ] ^2\right\ } .
\end{array } \eqno{(3c)}\ ] ] transition in eq .
( 3c ) from the multiplications of the exponents to the summations in the exponents and transformations in analogy with eq .
( 11 ) of the two sums summation indexes @xmath66 and @xmath67 , respectively , yield according to eq . ( 11 ) the final expression for the power spectral density @xmath68 since the expansion of the function @xmath69 in powers of @xmath70 at @xmath42 according to eqs .
( 12 ) and ( 13 ) is @xmath71 for @xmath72 and @xmath73 we can replace the summation in eq .
( 14 ) by the integration @xmath74 where @xmath75 is the averaged current .
furthermore , at @xmath76 it is sufficient to take into account only the first term of expansion ( 15 ) , @xmath77 .
integration in eq .
( 16 ) hence yields to 1/f spectrum @xmath78 dq=\bar i^2 \frac{\alpha _ h}f,\quad f_1<f
< f_2,f_\tau \eqno{(17)}\ ] ] where @xmath79 is a dimensionless constant ( the hooge parameter ) @xmath80 using an expansion of the function @xmath55 at @xmath81 according to expression ( 13 ) , @xmath82 , we obtain from eq .
( 16 ) the lorentzian power spectrum density for @xmath83 @xmath84 here @xmath85 and @xmath86 is the diffusion coefficient of the time @xmath8 according to eqs . ( 7 ) and ( 8) . the model is , therefore , free from the unphysical divergence of the spectrum at @xmath87 ; for @xmath88 we have from eq . ( 19 ) the white noise @xmath89 therefore , the model containing only one relaxation rate @xmath3 for sufficiently small parameter @xmath3 can produce an exact 1/f - like spectrum in any desirably wide range of frequency , @xmath90 . furthermore , due to the contribution to the transit times @xmath8 of the large number of the very separated in time random variables , @xmath59 ( @xmath91 ) , our model represents a long - memory random process .
eqs . ( 16)(20 ) describe quite well the power spectrum of the random process ( 4 ) . as an illustrative example in fig .
1 the numerically calculated power spectral density averaged over five realizations of the process ( 4 ) is compared with the analytical calculations according to eqs .
( 16)(20 ) .
the analytical results are in good agreement with the numerical simulations .
note , that analytical results predict not only the slope and intensity of 1/f noise but the frequency range @xmath92 of 1/f noise and intensity of the very low frequency , @xmath93 , white noise ( 20 ) as well .
this model may also be generalized for the non - gaussian and for the continuous perturbations of the systems parameters resulting in the fluctuations of the period @xmath94 .
so , for perturbations by the non - gaussian sequence of random impacts @xmath95 with zero expectations eqs .
( 1)(13 ) remain valid . only the result ( 14 ) of
the averaging over realizations of the process in the case of the non - gaussian perturbations may have different form .
consider now such a situation in more detail . the power spectral density ( 3 )
may be rewritten in the form @xmath96 where the transit times @xmath40 and @xmath8 difference is expressed as @xmath97 and the brackets denote the averaging over the time ( index @xmath6 ) and over the realizations of the process . here
@xmath98 is the averaged time interval between the subsequent transit times in the time interval @xmath99 .
note that for the slow ( diffusive - like ) fluctuations of the averaged interval @xmath100 with the change of the index @xmath6 eq .
( 22 ) is valid also when @xmath51 , i.e. @xmath101 , @xmath51 . at @xmath102
we may replace the summation in eq .
( 21 ) by the integration and do not take into account the dependence of @xmath103 on @xmath52 . in such a case
( 21 ) yields @xmath104 here the averaging over @xmath6 and over the realizations of the process coincides with the averaging over the distribution of the periods @xmath26 , i.e. @xmath105 where @xmath106 is the periods @xmath26 distribution density and @xmath107 is the characteristic function of the distribution @xmath106 . taking into account the property of the characteristic function @xmath108 we have from eqs . ( 23 ) and ( 24 ) the final expression for the power spectral density @xmath109 substituting into eq . ( 25 ) the value @xmath110 for the gaussian distribution of the periods @xmath26 we recover the result ( 17)(18 ) . since different processes result in the gaussian distribution it is likely that perturbation by the non - gausssian impacts @xmath22 in eq .
( 4 ) yields nevertheless the gaussian distribution of the periods @xmath26 . for the demonstration of such statement and validity of the approach ( 21)(25 )
we have performed numerical analysis of the model ( 1)(4 ) for different distributions of the perturbations @xmath111 .
figures ( 2 ) and ( 3 ) represent the calculated power spectral densities for the rectangular ( uniform ) and asymmetric @xmath112 distributions of the sequence @xmath95 with zero expectations and the same variances and other parameters like those for fig .
1 . we can notice only the slight dependence of the spectra on the distribution function of the perturbing impacts @xmath22 with the same expectations and variances .
these results confirm also the presumptions made in the derivation ( 21)(24 ) of the 1/f noise intensity ( 25 ) .
analysis of the exactly solvable model of 1/f noise display main features of the noise and serves as a basis for revealing of the origin and parameter dependences of the flicker noise .
this allows us to make generalizations of the model resulting in the expression for the 1/f noise intensity through the integral of the characteristic function of the distribution of the time intervals between the subsequent transition times of the elementary events , pulses or particles .
it should be noticed , however , that eq .
( 25 ) represent an idealized 1/f noise .
the real systems have finite relaxation time and , therefore , expression of the noise intensity in the form ( 23 ) is valid only for @xmath113 with @xmath114 being the relaxation time of the period s @xmath26 fluctuations . on the other hand , due to the deviations from the approximation @xmath115 at large @xmath52 , for sufficiently low frequency we can obtain the finite intensity of @xmath116 ( @xmath1 ) noise even in the case @xmath117 but for the signals with fluctuations resulting in the dense concentrations of the transit times @xmath8 .
generalizations of the approach ( 21)(25 ) and analysis of the deviations from the idealized 1/f noise expression ( 25 ) are subjects of separate investigations .
f. n. hooge , t. g. m. kleinpenning , and l. k. j. vadamme , rep .
. phys . * 44 * , 479 ( 1981 ) ; p. dutta and p. m. horn , rev .
mod . phys . * 53 * , 497 ( 1981 ) ; sh .
m. kogan , usp .
nauk * 145 * , 285 ( 1985 ) [ sov .
* 28 * , 170 ( 1985 ) ] ; m. b. weissman , rev .
* 60 * , 537 ( 1988 ) ; m. j. kirton and m. j. uren , adv . phys . * 38 * , 367 ( 1989 ) ; v. palenskis , lit .
sb . * 30 * , 107 ( 1990 ) [ lithuanian phys
. j. * 30*(2 ) , 1 ( 1990 ) ] ; f. n. hooge , ieee trans .
electron devices * 41 * , 1926 ( 1994 ) ; g. p. zhigalskii
nauk * 167 * , 623 ( 1997 ) [ phys .- usp . * 40 * , 599 ( 1997 ) ] . w. h. press , comments astrophys . * 7 * , 103 ( 1978 ) ; a. lawrence , m. g. watson , k. a. pounds , and m. elvis , nature ( london ) * 325 * , 694 ( 1987 ) ; m. gartner , sci
238 * , 16 ( 1978 ) ; s. maslov , m. paczuski , and p. bak , phys .
* 73 * , 2162 ( 1994 ) ; m. usher and m. stemmler , ibid * 74 * , 326 ( 1995 ) ; y. shi , fractals * 4 * , 547 ( 1996 ) ; v. b. ryabov , a. v. stepanov , p. v. usik , d. m. vavrin , v. v. vinogradov , and yu .
f. yurovsky , astron .
* 324 * , 750 ( 1997 ) .
j. kumiak , ann .
physik * 3 * , 207 , ( 1994 ) ; j. kumiak , in ref @xcite , p. 93 ; h. akbane and m. agu , in ref @xcite , p. 601 ; j. timmer and m. knig , astron . astrophys . * 300 * , 707 ( 1995 ) ; m. knig and j. timmer , astron .
. ser . * 124 * , 589 ( 1997 ) .
j. bernamont , ann .
physik * 7 * , 71 ( 1937 ) ; m. surdin , j. phys .
radium ( serie 7 ) * 10 * , 188 ( 1939 ) ; f. k. du pr , phys . rev . * 78 * , 615 ( 1950 ) ; a. van der ziel , physica * 16 * , 359 ( 1950 ) ; a. l. mcwhorter , in : _ semiconductor surface physics , _ edited by r. h. kingston ( univ .
philadelphia , 1957 ) , p. 207 ; f. n. hooge , in ref @xcite , p. 3 .
b. kaulakys , xxx.lanl.gov/abs/adap-org/9806004 and to be published ; b. kaulakys and t. mekauskas , 7th vilnius conf . on probab .
theory and 22nd european meeting of statisticians , vilnius , august 12 - 18 , 1998 .
abstracts ( vilnius , tev , 1998 ) , p.265 ; a. bastys and b. kaulakys , ibid , p. 144 . fig .
power spectral density vs frequency of the current generated by eqs .
( 2)(4 ) with the gaussian distribution of the random increments @xmath111 for different parameters @xmath27 , @xmath23 , and @xmath3 .
the sinuous fine curves represent the averaged over five realizations results of numerical simulations , the heavy lines correspond to the numerical integration of eq .
( 16 ) with @xmath69 from eq .
( 13 ) , and the thin straight lines represent the analytical spectra according to eqs .
( 17 ) and ( 18 ) .
2 . same as in fig . 1 but for the uniform distribution of the random increments @xmath95 .
note that for the non - gaussian distributions of the random perturbations we have no explicit expression analogous to eq .
( 16 ) for the integral representation of the noise power spectral density . | * physical review e vol .
58 , no . 6 , p. 7013 - 19 * the noise of signals or currents consisting from a sequence of pulses , elementary events or moving discrete objects ( particles ) is analyzed .
a simple analytically solvable model is investigated in detail both analytically and numerically .
it is shown that 1/f noise may result from the statistics of the pulses transit times with random increments of the time intervals between the pulses .
the model also serves as a basis for revealing parameter dependences of 1/f noise and allows one to make some generalizations . as a result the intensity of 1/f noise is expressed through the distribution and characteristic functions of the time intervals between the subsequent transit times of the pulses .
the conclusion that 1/f noise may result from the clustering of the signal pulses , elementary events or particles can be drawn from the analysis of the model systems .
pacs number(s ) : 05.40.+j , 02.50.-r , 72.70.+m |
nowadays it is well - established that a massive dark object , very possibly a massive black hole ( mbh ) with a mass of about @xmath8 , is lurking in the centre of the milky way @xcite . while there is an emerging consensus about the origin and growth of supermassive black holes ( smbh
, with masses about or larger than @xmath9 ) , mbhs with smaller masses such as the one in the galactic centre remain a ( relatively ) understudied enigma .
one of the keys to understanding the growth and evolution of mbhs in this lower mass range resides in the dynamics of stars in their vicinity .
this is the case mainly because relaxation times there are low enough that the surrounding stellar systems should have had enough time through two - body relaxation _
alone_to evolve towards a steady - state which is independent of the particular initial conditions at the time of formation .
the galactic center is thought to fulfill such condition .
it is the universality of such relaxed stellar nuclei that gives us a crucial predictive power on the expected properties of the mbh environment , on the stellar candidates for close interaction with the central mbh and on the resulting gravitational wave ( gw ) signatures . if , on the contrary , non - relaxed systems were generic , then one would need to resort to _ case - by - case modelling of each galactic nucleus_. the ideal probe for these innermost regions of galaxies is the gw radiation that is emitted by stellar bhs and other compact objects that come very close to the mbh .
one of the main channels for interaction between stars and a central mbh is the adiabatic , slow inspiral of compact remnants ( cr ) into the mbh due to the emission of gws an emri . during such an event ,
the small body effectively acts as a probe of spacetime close to the mbh as its orbit slowly shrinks due to the energy and angular momentum lost in the form of gw radiation . in case of @xmath10 mbhs , after some @xmath11 orbits in the lisa band ( @xmath12 hz and a periapsis @xmath13 , since we only consider sources which are completely embedded in the band , and not bursting sources ) , the small body eventually merges with the mbh .
the information contained in the waves will allow us to determine the parameters of these binary system with an unprecedented accuracy ( see for instance * ? ? ? * ; * ? ? ? * ) , corroborate the existence of mbhs and maybe even provide the first direct detection of an intermediate mbh ( in case the primary is @xmath14 ) .
_ lisa _ will thus scrutinize exactly the mass range about which electromagnetic observational information is currently lacking . in its most general form , the emri problem the astrophysical modelling of event rates and parameters for emris
spans many orders of magnitude . from the bulge regions at few@xmath15 pc , where the dynamics is essentially collisionless but from where single stellar bhs and binaries with crs originate
; down to the parsec scale of the nucleus itself which evolves secularly over ( local ) relaxation timescales ; and then further down to milliparsec scales where relativistic effects start to dominate the evolution .
but , however , once a steady state configuration establishes itself in the central parsec region , the emri rates are rather expected to depend strongly on the ( universal ) density distribution of crs within ( in order of magnitude ) @xmath16 from the hole .
this is indeed the region from which these inspiralling sources are expected to originate @xcite .
the dynamics in this tiny volume has been rather unexplored until the relevance of emris and sub - parsec observations of the galactic center have raised its interest . since then
, many authors have devoted a number of works to the analysis of this peculiar regime @xcite .
we discuss in this work the stellar distribution of dense stellar systems around mbhs in the lisa mass range .
realistic modeling of mass segregation which is the natural outcome for any realistic stellar population will strongly impact the expected emri rates , since it favors the accumulation of heavier objects towards the center @xcite . in section @xmath6 ,
we begin by summarizing the results obtained by @xcite that show how to calibrate the fp calculations with direct @xmath17-body simulations - body simulations compute the gravitational accelerations between particles using the exact newton s law , without introducing any approximations to compute the gravitational potential @xcite . ] ; then , still in the same section , we present new results concerning the robustness of @xmath17-body realisations of stellar cusp growth with respect to the total particle number @xmath17 . in section @xmath18
, we present new results on the growth of stellar cusps from a variety of initial conditions resulting from carving a cavity in the star s phase space distribution function .
this is motivated from a variety of astrophysical mechanisms that may lead to cusp destruction ; and these mechanisms are critically assessed in the end of the section .
we show that , under very generic circumstances , the time required for the growth of a cusp is shorter than a hubble time . therefore , quasi - steady , mass segregated , stellar cusps are expected to be common around mbhs in the lisa mass range . but see @xcite and @xcite for different perspectives .
emri detection rates for _ lisa _ are expected to peak for @xmath19 @xcite leading us to conclude that at least a sizeable fraction of these events should originate from strongly segregated cusps . finally , in section @xmath20 we present new estimates on the expected emri rates in mass segregated nuclei and conclude that our realistic modeling of mass segregation has a significant impact on these rates .
the distribution of stars around a massive black hole is a classical problem in stellar dynamics @xcite . @xcite
have shown , through a kinetic treatment that , within the radius of gravitational influence of the hole @xmath21 , in case all stars are of the same mass , this quasi - steady distribution takes the form of power laws , @xmath22 , in physical space with @xmath23 the stellar density at a radius @xmath24 and @xmath25 in energy space ( with @xmath26 the energy and @xmath27 and @xmath28 ) .
this is the so - called _ zero - flow solution _ for which the net flux of stars in energy space is _ precisely _ zero .
@xcite and @xcite were the first to demonstrate the robustness of the corresponding direct - summation @xmath17-body realizations , and have therefore validated the assumptions inherent to the fokker - planck ( fp ) approximation namely , that scattering is dominated by uncorrelated , @xmath6-body encounters and , in particular , dense stellar cusps ; it is said to be a _ cusp _ if @xmath29 . ] populated with stars of the _ same mass _ are robust against ejection of stars from the cusp .
the latter result is not trivial as for a bw @xmath27 cusp stellar densities are extremely high at the center and the fraction of stars with speeds close to the escape velocity from the cusp is quite high at all radii @xmath30 , with @xmath21 the influence radius of the mbh @xcite .
single mass models are very poor approximations of real stellar populations .
to first order of approximation , an evolved stellar population can be represented by two ( well - separated ) mass scales : one in the range @xmath31 corresponding to low mass main - sequence stars , white dwarfs ( wds ) and neutron stars ( nss ) ; another with @xmath32 representing stellar bhs .
therefore , for simplicity , here we restrict our discussion to models with two mass components and leave the more general case to another work in preparation @xcite .
when stars of two different masses are present , there is mass segregation which is a process by which the heavy stars accumulate near the center while the lighter ones float outward @xcite .
accordingly , stars with different mass get distributed with different density profiles .
@xcite , henceforth bw77 , have argued heuristically that a scaling relation @xmath33 ( where the subindices @xmath34 , @xmath35 refer to the light or heavy components ) establishes itself and depends only on the mass ratio . here , as in the single - mass case , the crucial assumption is that all components are abundant enough that they undergo enough scattering among themselves and with the other components as to stabilize into an approximate _ zero - flow _ solution .
obviously , this can not happen independently of the number fraction of the different stellar masses @xcite . in the realistic situation where the number fraction of heavy objects ( in our case , stellar bhs ) is small , a new solution coined by @xcite as _ strong mass segregation _ ( sms ) obtains with density of heavy objects scaling as @xmath36 , where @xmath37 .
the solution has two branches and can be parametrized by the parameter @xmath38 where @xmath39 and @xmath40 are the total number of light and heavy stars , @xmath41 and @xmath42 are the corresponding individual masses .
@xmath43 provides a measure of the importance of the heavy star s self - coupling relative to the light - heavy coupling ( in terms of the @xmath44 and @xmath45 order diffusion coefficients ) ; and it depends essentially on the mass _ and _ number ratios , which is one parameter more than proposed by bw77 .
the @xmath46 branch , for @xmath47 corresponds to the scaling relations found by bw77 ; while the @xmath48 branch , for @xmath49 , generalizes the bw77 solution .
stellar populations with continuous star formation and an initial mass function ( imf ) given by @xmath50 will be characterized by @xmath49 if @xmath51 and @xmath49 otherwise ; and , in particular , salpeter and kroupa s imf generate evolved stellar populations with @xmath52 @xcite .
* figure 1 .
* evolution of density profiles .
mass density profiles , @xmath53 ( left panels ) and @xmath54 ( right panels ) at the end of the integrations , after @xmath55 .
red curves are from fp calculations , green curves are from nb simulations .
the agreement between both methods is quite good .
the mass ratio between heavy and light stars is @xmath56 , representing the expected typical mass ratio between light stars ( ms stars , wds and nss ) and stellar bhs , as explained in the text ; the number fraction of heavy stars @xmath57 ( top and middle panels ) and @xmath58 ( lower panels ) , corresponding to the strong and weak segregation regimes respectively .
the initial condition is a dehnen profile with central slope @xmath59 for the top and bottom panels @xcite , @xmath60 in the middle panel ; a central mbh with @xmath61 of the total mass of the cluster and @xmath62 likewise .
the particle number is @xmath63 ( top and bottom ) and @xmath64 ( middle ) . the asymptotic slope @xmath65 decreases from @xmath66 to @xmath67 when moving from the strong to the weak branch of the solution .
the asymptotic slope @xmath68 throughout , or just slightly below this value .
the arrows point to radii @xmath21 and @xmath69 .
there is a straightforward physical interpretation for the strong branch of mass segregation . in the limit where heavy stars are very scarce
, they barely interact with each other and instead sink to the center due to dynamical friction against the sea of light stars .
therefore , a quasi - steady state develops in which the heavy star s current is _ not _ nearly zero and thus the bw77 solution does not hold exactly anymore .
indeed , in the limit where the number fraction @xmath70 of heavy stars is vanishingly small , as is the case of nuclei with realistic imfs , the stellar potential is dominated by the light component . in this case , the light stars should evolve as if in isolation and develop a @xmath71 density cusp .
the scarce heavy stars sink to the center due to dynamical friction against the background of light stars , and will not exert any significant back - reaction on them @xcite .
figure 1 displays the fp and nb evolutions of the spatial density @xmath53 and @xmath54 for models with two mass components corresponding to different initial profiles , mbh masses and total particle number @xmath17 .
the starting models are either @xmath59 or @xmath60 dehnen profiles for both components with a mbh of @xmath62 or @xmath61 of the total mass of the cluster .
dehnen density profiles are defined by @xmath72 , have total stellar mass @xmath73 , an inner ( outer ) logarithmic slope @xmath74 ( @xmath75 ) and a break radius @xmath76 ( which is larger than @xmath21 in all cases ) .
we adopt units where @xmath77 .
the density of both components reaches a quasi - steady state within @xmath78 , where @xmath79 relaxation time measured at the influence radius @xcite .
the top and middle panels display the strong mass segregation solution with @xmath80 as expected for @xmath81 ( @xmath82 ) ; while , in the bottom panel , @xmath83 ( @xmath58 ) displays the weak solution for which @xmath84 .
the former value was chosen to be close to the number fraction of stellar bhs to be expected from a standard salpeter or kroupa s imf ; the latter value is chosen to be representative of the regime of weak segregation studied by bw77 .
one can see from figure 1 that in the case of weak segregation @xmath85 everywhere due to the extremely high number of heavy objects ; in contrast , in the sms regime @xmath86 only for @xmath87 ( and the light objects dominate in number almost everywhere ) . in all cases
the asymptotic slopes are valid within @xmath88 and are totally predictable once @xmath43 is known .
these results agree pretty well with the predictions for the sms regime @xcite .
the particle number in our direct - summation @xmath0body simulations sample ranges from @xmath63 to @xmath64 ; our results do not show evidence of any dependence on total @xmath17 , nor on the initial value of @xmath74 , once the results are re - scaled appropriately ( _ i.e. _ measured in terms of the relaxation time ) .
the agreement between nb and fp methods is quite good in all cases .
we have seen that theory predicts a steady state cusp that reaches extremely high densities in the center near the mbh .
furthermore , given a normalization at , say , @xmath21 and a knowledge of the stellar mass function ( and thus of @xmath43 ) , the density profile inside @xmath21 becomes completely determined .
but observations are much more complicated to interpret .
first , one must realize that there are very few galaxies for which the influence radius @xmath21 can be resolved .
in fact , except for the nearest galaxies , @xmath21 covers an angular region in the sky which is too small to be resolved even with the hst .
the hst has an angular resolution of @xmath89 . in the case of galaxies in the virgo cluster , at @xmath90 mpc of distance
, it can only resolve regions with linear sizes @xmath91 pc .
therefore , hst would not be able to resolve sgra * s radius of influence if it were at the distance of virgo . since @xmath92
, it can only start to resolve the influence radius of virgo s mbhs that have masses @xmath93 .
second , even in the few cases for which @xmath21 can be resolved to some extent , it still is necessary to assess whether the observed stars ( only those that are bright enough to be detected ) really trace the underlying ( dynamically dominant ) invisible population .
third , given the fact that , as we have seen , stars tend to segregate by mass , there is an extra uncertainty related to the unknown stellar mass function .
moreover , there are indications that star formation events are common in galactic nuclei and furthermore that the resulting imf in these sub - parsec regions may be substantially different from that of the field stars and biased towards heavy masses @xcite .
finally , it is necessary to deproject the observations and , in the ( inevitable ) absence of complete knowledge of phase space coordinates , one must rely on kinematic assumptions regarding the ( an-)isotropy of stellar velocities and on the three dimensional shape of the stellar system .
time for cusp re - growth .
single - mass relaxation time at @xmath21 for single - mass cored models as a function of mbh mass .
the shaded area covers @xmath94$]the time for cusp re - growth if there is no hole in the initial df .
the three dashed lines above the shaded region represent the average time needed for the cusp re - growth in case one imposes an initial cavity with size @xmath95 or @xmath6 pc .
the horizontal dashed curve represents 13 gyr .
it can be seen that the time needed to re - grow a cusp around mbh with masses @xmath1 is below a hubble time so long as the initial cavity is smaller than @xmath96 pc .
nevertheless , it has come as a surprise that very recent spectroscopic observations of the galactic center revealed a core ( or even a dip ) in the surface distribution of the old stellar population ( essentially red giants ) which should have had time to relax into a cuspy density profile .
the caveat is that the detected stars are still a small fraction , of about 5% , of the stellar population as a whole and therefore do not exclude the presence of an extended dark cluster ( presumably made of stellar bhs and other crs)which would indeed agree with our theoretical expectations . to assess the likelihood that the galactic center is indeed unrelaxed , it is natural to ask : how long does it take to re - grow a cusp
if , at some point , it has been destroyed ?
a complete answer will , of course , depend on the extent to which the cusp was destroyed , _
i.e. _ how much mass was expelled from the original cusp and over which radial range . at this level , it does not matter very much which mechanism led to the destruction of the cusp .
we discuss briefly possible scenarios for cusp destruction at the end of the section . in order to investigate this question ,
we have concocted a set of initial conditions purported to mimic the outcome of a destroying cusp event such as the carving of a cavity in phase space through the ejection of stars by , say , an infalling imbh or , following a major merger , by a mbh .
we model the outcome of such an event by imposing that all stars with binding energies larger than some @xmath97 or almost equivalently , with semimajor axis smaller than @xmath98 , are not present in the initial df .
in fact , inside @xmath21 the mbh dominates the gravitational potential and @xmath99 .
we thus set up an initial dehnen model with @xmath100 for @xmath101in other words , there is an initial cavity in the phase space df , but not in physical space as the stars with lower energy still entail @xmath102 at the center , although with a smaller amplitude than the original model .
the values of @xmath97 were chosen to represent cavities of size @xmath103 pc resulting from the inspiral of an imbh / mbh .
note that these models are , by construction , isotropic in the velocity distribution . or @xmath104 maximum . ]
our fiducial model is a milky way type nucleus with @xmath105 , some @xmath106 in total stellar mass inside @xmath107 pc distributed according to an initial central density slope @xmath60 , two components with masses @xmath108 and @xmath109 , and @xmath110 of stellar bhs by number .
when the stellar distribution has no phase space cavity , this translates into having stars down to roughly @xmath111 pc .
having validated the fp models with detailed @xmath17-body simulations , we now proceed in the rest of the paper to describe results obtained with the ( much faster ) fp approach .
figure 2 shows the times for cusp re - growth computed with fp for different galactic nuclei models .
the time for cusp re - growth is defined as the time it takes for a given initial density profile ( @xmath23 in space or @xmath112 in phase space , with or without an initial cavity ) to reach its asymptotic slope , which depends on @xmath43 , down to @xmath113 .
this is indeed the scale which is resolved by recent observations of the galactic center @xcite .
the shaded region represents the time of cusp re - growth for a range of @xmath114 and @xmath70 ( all in the sms regime , @xmath115 ) for the case where @xmath112 extends to high @xmath26 without any cut .
it can be seen that , for @xmath116 , cusps grow in less than a hubble time ; in the particular case of the milky way nucleus with @xmath117 , it takes no longer than @xmath118 gyr to fully re - grow a steady - state , mass segregated , stellar cusp and only @xmath119 gyr to have it grown down to @xmath120 .
if , instead , an initial cavity is imposed at the center with size @xmath95 or @xmath6 pc in case of the milky way ( or @xmath121 or @xmath122 in case of a generic nucleus ) , times for re - growth are represented by the dashed curves above the shaded region . in this case , times for cusp re - growth increase ; in the milky way case , it becomes @xmath123 or @xmath124 gyr , respectively .
note that , in the mass range @xmath10 , the times for cusp re - growth are definitely _ much shorter _ than a hubble time even if a fairly large cavity of size comparablee to @xmath21 is hypothesized .
the full curve represents the relaxation times computed at the radius of influence @xmath21 , while the dashed curves represent the _ actual _ times for cusp re - growth as measured from fp calculations @xcite .
it is difficult to devise plausible mechanisms for the formation of such large cores in the stellar distribution .
for instance , the inspiral of an imbh of mass @xmath125 that forms an unequal - mass binary with the mbh and ejects stars through three body encounters would tend to progressively wipe out the stellar cusp . however , the core radius carved by such an event is @xmath126 pc @xcite and thus a steady inflow of such imbhs ( one every @xmath127 years for a hubble time ) would be required in order to carve a large core @xmath128 or @xmath129 larger .
such large inflow of imbhs have been proposed by @xcite .
this does not seem very likely anymore in light of the fact that such imbhs were hypothesized to be formed by runaway mergers of stars in the center of globular clusters .
however , at solar metallicities , such mechanism seems very inneficient .
mass loss due to very strong winds severely limits the growth of the stellar object being formed and the likely end result of a runaway merger is a @xmath130 wolf - rayet star . at lower metallicities ,
mass loss is lower and the remnant can be more massive @xmath131 , but in any case it will not form an imbh . in sum , it looks very unlikely that sufficient imbhs can be formed in order to generate such steady inflow to the galactic center .
another possibility would be that sgra * is a binary mbh , but this would most likely imply that there has been a , more or less recent , major merger involving the milky way .
this would contradict the apparent pure - disk nature of the galaxy , as theoretical interpretations of stellar kinematic data of the galactic bulge seem to favor that the bulge is part of the disk and not a separate component resulting from a merger @xcitey aside from the fact that there are strong constraints from the sgra * proper motion @xcite .
stars in a keplerian potential , @xmath132 , do not precess because of the @xmath133 resonance between their radial and azimuthal frequencies .
resonant relaxation ( rr ) results from the coherent torques that such stars exert on each other leading to a fast evolution of their orbital angular momenta over timescales @xmath134 , where @xmath135 is the precession timescale due to departures from an exact kepler potential @xcite .
@xcite suggest that rr , by increasing the rate of angular momentum diffusion in the near - keplerian gravitational potential around the mbh , may boost the tidal disruption rate of stars and could thus create a near - cavity ( out to @xmath136 pc ) in the stellar distribution .
it is certain that rr operates to some extent in the inner parsec , but we doubt it can completetly explain the dearth of red giant stars there or , more generally , the full destruction of a cusp including crs such as stellar bhs first , their final density distribution does not show a cavity , nor a shallow cusp profile , for @xmath137 pc ; instead , they get final slopes @xmath138 for @xmath137 pc .
this is in contrast with observations which show a decaying density for @xmath139 pc .
second , we believe that their monte - carlo calculations severely underestimate the rate at which the cusp re - grows ; in fact , they obtain a timescale which is @xmath140 times longer than that obtained with our @xmath17-body simulations ( which are free from any simplifying dynamical assumptions ) , and also from ours and their own recent fp calculations @xcite .
third , were they to use the latter rates , and given that the time rr takes to deplete the cusp is of the same order as the time we obtain for cusp re - growth , the net effect of rr on the cusp would likely be minute . moreover , stellar bhs can not be tidally disrupted , and this makes them less susceptible to be extracted from the cusp than @xmath141 stars .
given a steady state stellar bhs continue to diffuse in @xmath142-space and some of them eventually come into close interaction with the mbh . during a close interaction ,
a stellar bh can either be promptly scattered into the mbh , accompanied by a single or a few brief bursts of gws in the _ lisa _ band the so - called `` direct - plunges'' , though they are not likely detectable unless if emitted from the galactic center @xcite , or scattered outwards in the cusp . in either case
, it does not live enough to become an emri .
alternatively , it may undergo a very slow , adiabatic , inspiral without being appreciably disturbed by other stars and , in this case , it will eventually become an emri detectable by lisa .
an emri object thus has to spend very many orbits without being significantly scattered by the gravitational tugs of the other stars .
in other words , they must fullfill the following _ inspiralling criterion _ : the time @xmath143 it takes for the inspiral , due to orbital energy lost by gw emission only , must be shorter than the typical time @xmath144 it takes on average to drift in angular momentum by an amount @xmath145 which equals its orbital angular momentum .
otherwise , they will be promptly captured by the mbh before entering the lisa band .
the inspiral criterion can be stated in terms of the parameter @xmath146 being smaller than unity , @xmath147 . for @xmath148 , it is almost certain that this object has either taken an almost radial orbit and fallen into the mbh as a direct plunge or has been scattered outwards .
will be compensated by another that drifts inward by the same amount .
this balance only breaks down for those orbits that fall on to the hole , since there are obviously no stars coming out of it to keep detailed balancing . ]
it turns out that this parameter simply scales with orbital s semimajor axis : @xmath149 @xcite , which means that it is a decreasing function of @xmath150 so long as @xmath151 .
this is indeed the case in both regimes of mass segregation .
furthermore , @xcite have shown that the probability for a successful inspiral as a function of orbital semimajor axis ( or energy ) is almost a step function of semimajor axis .
if @xmath152 , it is almost certain that the stellar bh will become an emri ; it will almost certainly not become one in case the inequality sign is reversed ( and the width of the `` transition region '' is very small ) .
this crucial threshold quantity demarcates the orbits which are close enough to the mbh to sucessfully decouple from the rest of the cluster and undergo the slow , adiabatic inspiral that defines an emri from those more weakly bound orbits that will be perturbed out of the emri tracks due to scattering with other stars .
therefore , in order to estimate the emri event rate , given a steady state @xmath112 , obtained via the fp equation , one essentially counts the number of stars that populate the region of phase space for which the inspiralling criterion above is satisfied and divide it by the local relaxation time .
note that here , for simplicity , we ignore other driving mechanisms in particular , we ignore resonant relaxation . under these assumptions ,
the emri rate for stellar bhs is approximately given by @xmath153 where @xmath154 is the number fraction of sbhs in the stellar population , @xmath155 is the total number of stars per unit energy(@xmath156 , see @xcite ) , @xmath157 is the specific angular momentum of a circular orbit of energy @xmath26 , @xmath158 is the loss - cone angular momentum and @xmath159 $ ] is the relaxation time .
the log term in the denominator in ( [ emri - rate ] ) arises from the phase space ( partial ) depletion resulting from the presence of the loss cone .
the conversion between @xmath24 and @xmath26 is , for @xmath160 , @xmath161 or @xmath162 .
the critical radius @xmath163 , or energy @xmath164 , for emris is approximately @xmath165 ; and , to first order , @xmath163 is independent of @xmath166 @xcite .
the weak regime of sms , and corresponding bw solution , would lead to a fairly high intrinsic rate , per galaxy , of emris .
in fact , figure 3 shows that , for a milky way nucleus , in case @xmath47 , the intrinsic emri rate is @xmath167 per gyr .
this is , however , unrealistic as such scenario pressuposes an unrealistically high number fraction of bhs ( @xmath168 for @xmath169 ) . in the more realistic case , when @xmath170 the bw solution would entail a strong supression of the emri rate to at best a few tens of events per gyr
this is where sms solution appears to rescue us .
sms implies a higher density of bhs inside @xmath21 as compared with the @xmath27 solution , and in this way by decreasing the local @xmath171 and increasing @xmath155 close to the mbh it partially , but not completely , compensates for the small number fraction of bhs entailed by realistic mass functions .
emri rate as a function of @xmath43 .
the number of stellar bh emri events per gyr in a milky way type nucleus ( @xmath105 and @xmath172 ) as a function of the parameter @xmath43 .
this is computed from a two - component mass segregated stellar cusp ( @xmath173 and @xmath174 ) with mass ratios @xmath56 and @xmath175 obtained from fp calculations . in the case of the fiducial value @xmath176 , @xmath177 ; in those circumstances , each milky way like nucleus will produce on average @xmath4 stellar bh emris per gyr . in order to quantitatively evaluate the boost @xmath178 to the emri rates from sms , for a given @xmath43 and a fixed mass normalization at @xmath21 , one needs to estimate what would be the rate if the spatial and phase space densities were determined by the @xmath27 cusp for @xmath179 .
this is done as follows : we define analytically both @xmath23 and @xmath112 that would result from a @xmath27 inside @xmath69 @xmath180 and @xmath181 where the indices fp mean that the profile is taken from the fokker - planck calculation .
@xmath182 ( and @xmath183 is a reference radius ( energy ) chosen according to @xmath184 . * figure 4 .
* boost on the emri rates from strong segregation .
one can see that , for values of @xmath49 , there is a significant boost to the emri rates in comparison to which it would result in the case of a @xmath27 bw cusp .
in particular , for our fiducial value @xmath170 ( @xmath185 ) , the boost is of order of a factor @xmath186 with respect to a @xmath187-bw cusp with the same mass normalization at @xmath188 pc .
figure 4 shows the boost to the emri rates due to sms relative to what would be obtained from a bw profile .
going from an unrealistically high @xmath154 , as adopted by bw77 ( say @xmath189 ) , to a more realistic @xmath154 ( say @xmath190 ) , while neglecting the existence of sms , one supresses the emri rate by factors of @xmath191 ( the former would lead to @xmath192 emris per gyr ; the latter is reduced to @xmath193 few tens per gyr ) .
however , by taking into account the sms solution , for this low @xmath190 , we boost again the rates by a factor close to @xmath186 , thus partially compensating the reduction of emris ( from few tens to a few hundred per gyr ; in fact , there are @xmath4 per gyr in case @xmath190 for a milky way nucleus ) .
we conclude that the apparently inocuous and tiny change of the logarithmic slope from @xmath194 to @xmath195 can have a substantial effect ( a factor of @xmath2 ) on the expected emri rate .
* figure 5 .
* emri rates as a function of mbh mass in strongly segregated nuclei .
the emri rate depends on the mbh mass , @xmath196 . shown are curves for @xmath190 ( @xmath176 and @xmath197 ) and three different mass ratios between heavy and light stars , @xmath56 and @xmath175 .
figure 5 shows the dependence of the intrinsic emri rate on the mass of the central mbh , where the validity of the @xmath198 relation was assumed @xcite .
we see that the emri rate for stellar bhs scales as @xmath199 , independently of @xmath114 and @xmath154 .
its absolute normalization depends obviously on the number fraction @xmath154 of sbhs , in agreement with figures @xmath18 and @xmath20 .
one can make a rough conversion of the estimated intrinsic rate into lisa detection rates .
following @xcite , who made a number of assumptions regarding the local density of mbhs and its spin distribution , plus on the lisa detection capabilities , we find that according to its equation ( @xmath200 ) , lisa will see around @xmath5 emri events during a @xmath6-year or @xmath7-year mission , respectively .
note that these rates may change by factors of @xmath201 as a function of corrections to the local mbh mass density @xcite ; moreover , the uncertainties regarding the efficiency of rr and other channels may still affect the rate predictions by one or two orders of magnitude .
a lot of work still remains to be done ; nevertheless , the consequences regarding the sms regime are significant and under control .
we have considered simplified stellar models of galactic nuclei , with only two mass components , which harbor mbhs that fall into the lisa detection bandwidth . for quite generic initial conditions ,
such stellar clusters are expected to have reached a relaxed , mass segregated , steady state which is independent of initial conditions at the time of formation .
strong ( realistic ) mass segregation is a robust outcome from the growth and evolution of stellar cusps around mbhs in the mass range @xmath202 to which lisa will be sensitive .
our n - body results validate the fokker - planck description of the _ bulk _ properties of the stellar distribution .
sms boosts the emri event rates with respect to what would be implied by a shallower stellar density profile ( _ e.g. _ @xmath27 , which has been the working assumption of almost all event rate estimates in the literature so far ) that also respect the mass normalization obtained from observations of the galactic center at @xmath107 pc from the hole . in particular , our fiducial models of the galactic center are enhanced by a factor of @xmath2leading to a predicted value of @xmath4 stellar bh emris per gyr .
the fp formalism assumes two - body relaxation as the only dynamical driver present this could be a severe restriction at radii @xmath203 , inside which even the nb simulations with higher @xmath17 in our sample start to run out of particles and where rr could play an important role @xcite .
other crucial mechanisms are resonant relaxation , ( small ) triaxiality of the galactic potential , tidal separation of binaries and massive perturbers ( see e.g. * ? ? ?
* for a review ) .
these are the subject of our current research work , and the extent to which they can significantly affect the emri rates is still an open question .
it is a pleasure for pas to thank tom prince for the invitation to give a plenary talk in the symposium .
he is equally indebted with esmeralda martnez for discussions as well as with vivian j. drew for her great sense of humour and coordination of the workshop .
mp thanks lauren b. for interesting comments .
pas and mp are partially supported by dlr ( deustches zentrum fr luft- und raumfahrt ) .
the simulations have been carried out on the dedicated high - performance grape-6a clusters at the astronomisches rechen - institut in heidelberg , which was funded by project i/80041 - 043 of the volkswagen foundation and by the ministry of science , research and the arts of baden - wrttemberg ( az : 823.219 - 439/30 and /36 ) , an also at the tuffstein cluster of the aei .
s , baker j g , benacquista m j , cornish n j , larson s l , mandel i , mcwilliams s t , petiteau a , porter e k , robinson e l , vallisneri m , vecchio a ,
data challenge task force t m l , adams m , arnaud k a , baut a , bridges m , cohen m , cutler c , feroz f , gair j r , graff p , hobson m , shapiro key j , krlak a , lasenby a , prix r , shang y , trias m , veitch j , whelan j t 2010 _ classical and quantum gravity _ * 27*(8 ) , 084009+ .
bartko , h. ; martins , f. ; trippe , s. ; fritz , t. k. ; genzel , r. ; ott , t. ; eisenhauer , f. ; gillessen , s. ; paumard , t. ; alexander , t. ; dodds - eden , k. ; gerhard , o. ; levin , y. ; mascetti , l. ; nayakshin , s. ; perets , h. b. ; perrin , g. ; pfuhl , o. ; reid , m. j. ; rouan , d. ; zilka , m. ; sternberg , a. 2010 _ apj _ * 708 * , 34840 f , genzel r , alexander t , abuter r , paumard t , ott t , gilbert a , gillessen s , horrobin m , trippe s , bonnet h , dumas c , hubin n , kaufer a , kissler - patig m , monnet g , strbele s , szeifert t , eckart a , schdel r zucker s 2005 _ apj _ * 628 * , 246259 . | one of the most interesting sources of gravitational waves ( gws ) for _ lisa _ is the inspiral of compact objects on to a massive black hole ( mbh ) , commonly referred to as an `` extreme - mass ratio inspiral '' ( emri ) .
the small object , typically a stellar black hole ( bh ) , emits significant amounts of gw along each orbit in the detector bandwidth .
the slowly , adiabatic inspiral of these sources will allow us to map space - time around mbhs in detail , as well as to test our current conception of gravitation in the strong regime .
the event rate of this kind of source has been addressed many times in the literature and the numbers reported fluctuate by orders of magnitude . on the other hand , recent observations of the galactic center revealed a dearth of giant stars inside the inner parsec relative to the numbers theoretically expected for a fully relaxed stellar cusp .
the possibility of unrelaxed nuclei ( or , equivalently , with no or only a very shallow cusp , or core ) adds substantial uncertainty to the estimates . having this timely question in mind
, we run a significant number of direct - summation @xmath0body simulations with up to half a million particles to calibrate a much faster orbit - averaged fokker - planck code .
we show that , under quite generic initial conditions , the time required for the growth of a relaxed , mass segregated stellar cusp is shorter than a hubble time for mbhs with @xmath1 ( i.e. nuclei in the range of _ lisa _ ) .
we then investigate the regime of strong mass segregation ( sms ) for models with two different stellar mass components .
given the most recent stellar mass normalization for the inner parsec of the galactic center , sms has the significant impact of boosting the emri rates by a factor of @xmath2 in comparison to what would result from a @xmath3bahcall & wolf cusp resulting in @xmath4 events per gyr per milky way type galaxy
. such intrinsic rate should translate roughly into @xmath5 sbh s emris detected by lisa over a mission lifetime of @xmath6 or @xmath7 years , respectively ) , depending on the detailed assumptions regarding lisa detection capabilities . |
the advent of topological materials @xcite has ignited a tremendous interest of the realization of quantum operations in platforms immune to decoherence processes @xcite .
a prominent feature of such materials is that they exhibit quasiparticle excitations similar to the elementary particles predicted by ettore majorana @xcite and named majorana fermions ( mfs ) @xcite . in particular ,
the exoticness of such particles is that they coincide with their own anti - particles .
the enormous interest generated by majorana quasiparticles in solid state systems resides in the possibility of creating such excitations non - locally and manipulate them for quantum computation purposes @xcite .
majorana quasiparticles behave as nonlocal qubits being resistant to decoherence phenomena . by exchanging two majorana
quasiparticles a non - trivial quantum operation ( unitary transformation or braiding operation ) , is performed . indeed , braiding manipulations among majorana quasiparticles are the fundamental basis for the realization of topological quantum computation @xcite .
there are a plethora of proposals for the observation of majorana quasiparticles in a diverse of solid state setups : quantum hall states , @xmath13-superconductors , or topological insulators @xcite among others .
nevertheless , the first signatures of the occurrence of such exotic excitations was reported in quasi - one - dimensional semiconductor nanowires @xcite .
this system consists of a nanowire with strong spin - orbit interaction put in proximity to a superconductor , and exposed to a magnetic field @xcite .
the experimental evidence of such quasiparticles @xcite was realized by means of tunnel spectroscopy .
however , so far , these evidences are not totally conclusive . in these experiments ,
the appearance of a zero - bias anomaly ( zba ) in the nonlinear conductance is ascribed to the majorana physics .
nevertheless , recently it has been pointed out that other sources of zba in normal - superconductor nanowires can occur .
for instance , kondo physics or shiba states @xcite or even nearly zero energy andreev states @xcite and weak antilocalization @xcite can not be discarded as explanations of the observed zbas .
therefore , the detection of the mfs is now a dire challenge .
most of schemes for the detection up to now are exploiting the electronic transport involving the mfs . and
only very recently the thermoelectric properties of majorana fermion are examined and suggested for another way to distinguish majorana physics from other explanations @xcite .
since the mf has the particle - hole symmetry itself , thermoelectric devices involving only the mf can not generate any temperature - driven charge current nor electric - field - driven heat current .
hence , an additional quantum object which breaks the particle - hole symmetry should be incorporated .
the recent studies @xcite have coupled the mf to the simplest quantum object , that is , a spinless quantum dot ( qd ) and controlled the particle - hole symmetry by tuning the dot level .
it turns out that the mf can give rise to a non - trivial thermoelectric response upon a thermal bias across the qd @xcite .
interests for the reciprocity between heat and charge currents has been revived in systems at nanoscale @xcite .
the electric response to a temperature gradient , in nanostructures , can achieve much bigger values in efficiency compared with macroscopic samples @xcite .
the main reason for such high heat - to - electricity conversion factors resides in the intrinsic quantum property of energy level discretization for confined systems @xcite .
however , in a more fundamental basis , thermoelectric transport can reveal rather useful information about the intrinsic nature of a quantum system @xcite and about their interactions @xcite . in this respect ,
our investigation explores the thermoelectric response to characterize the half - fermionic nature of majorana nanowires . ) and to ( a ) the mf state formed at one end of a topological superconductor wire and ( b ) the spin - polarized ( sf ) ordinary fermion state by qd - mf or qd - sf tunneling rate @xmath14 . due to the helical property of the mf state , only one of the spin components ( say spin-@xmath15 )
is coupled to the mf .
in the sf case , the spin is assumed to be polarized in the spin-@xmath15 direction.,width=340 ] in the context of the mf , mostly the spinless or spin - polarized qd is chosen because an external magnetic field is necessary for inducing the mf .
however , as long as the zeeman splitting in the qd is small enough or the external magnetic field is replaced by a proximity induced effective field due to a ferromagnetic layer , one can consider the case of the spinful qd in which the coulomb interaction is highly relevant .
in fact , the interplay of the coulomb interaction and the mf in the kondo dot side - coupled to a mf [ see fig . [
fig : model](a ) ] is found to display an interesting transport behavior @xcite : the presence of the mf modifies the kondo effect profoundly by reducing the linear conductance by a factor 3/4 or by inducing an effective zeeman splitting on the qd , which leads to spin - split kondo resonance . in ref .
[ ] some of the authors of this work investigated the interplay of majorana and kondo physics in the electrical conductance . here
, the purpose is to study the thermoelectric response of the side - coupled qd - nanowire system [ see fig .
[ fig : model](a ) ] as a tool to detect majorana quasiparticles .
a temperature difference is established between the two normal contacts coupled to the qd , then the electric response is measured .
we find that thermoelectric properties in the kondo dot - mf system depend strongly on the qd - mf coupling ( @xmath5 ) , the overlap between two mf end - states ( @xmath16 ) , and the kondo temperature @xmath17 .
for the ideal case ( @xmath18 ) , due to the anti - fano resonance between kondo resonance peak and the zero - energy mf mode , the thermoelectrical conductance has a finite value at @xmath8 in the weak - coupling regime ( @xmath19 ) . in the strong - coupling regime ( @xmath20 ) , the induced zeeman splitting shifts the kondo - induced peak in the thermoelectrical conductance .
the finite overlap ( @xmath21 ) leads to the interference between majorana fermions and restores the kondo effect in a small energy scale @xmath22 and the thermoelectrical conductance becomes finite at @xmath23 whose sign is opposite to that at @xmath24 .
we also consider the case in which the mf is replaced by an ordinary spin - polarized fermion ( sf ) and find similarities and differences between two cases , eventually identifying the fingerprint of mf in the thermoelectric effect . in sec .
ii we introduce the model of hamiltonian employed to describe the qd - mf setup and the qd side - coupled to a spin polarized fermion .
electrical and thermoelectrical conductances are also derived in terms of the local spectral density for the interacting qd .
iii and iv explain our results for the temperature dependence of the electrical and thermoelectrical conductances for the qd - mf device and a comparison with the spin polarized fermion model is shown .
section v is devoted to describe our findings for the gate dependence of the thermoelectrical conductance , and we finish by enumerating the main conclusions in sec . vi .
we consider a two - terminal system in which a qd is tunnel - coupled to two normal - metal leads and to one end of a topological superconducting wire ( tsw ) which hosts two mfs at its ends [ see fig .
[ fig : model](a ) ] .
the on - site coulomb repulsion in the qd is considered .
the voltage bias and/or the thermal gradient will be applied across the qd . assuming that the superconducting gap in the tsw is large enough
, the effective low - energy hamiltonian reads @xmath25 \\ & \quad\mbox { } + \epsilon_{d } \sum_{\mu } n_\mu + u n_\up n_\down \\ & \quad\mbox { } + 2i \epsilon_{m}\gamma_{1 } \gamma_{2 } + t_{m } ( d^{\dagger}_\down\gamma_{1 } + \gamma_{1}d_\down ) .
\end{split}\end{aligned}\ ] ] here @xmath26 creates an electron with momentum @xmath27 , energy @xmath28 , and spin @xmath29 in lead @xmath30 .
two leads are assumed to share a same flat - band dispersion @xmath31 with a half bandwidth @xmath32 and density of states @xmath33 .
the operator @xmath34 creates an electron in the single - level qd with energy @xmath35 and spin @xmath29 : @xmath36 is the dot occupation for spin @xmath29 .
the coulomb repulsion in the qd is denoted by @xmath37 . as mentioned in sec .
[ sec : introduction ] the qd is considered to be spin degenerate @xcite .
this condition can be achieved by using low @xmath38 factor material for the qd or by using proximity with a magnetic insulator to induce the effective zeeman splitting in the tsw @xcite .
the tunneling between the leads and the qd is quantified by the tunneling amplitude @xmath39 , which defines the tunneling rates @xmath40 .
the low - energy hamiltonian of the tsw is governed by the two mf operators @xmath41 ( @xmath42 ) @xcite which satisfy the clifford algebra @xmath43 . in a finite - length tsw
the two mf states have a finite overlap between their wave functions , which is quantified by the overlap integral @xmath44 .
in addition , the mf operators can be written in terms of an ordinary fermion operator @xmath45 which satisfies the usual fermionic commutation relation @xmath46 as @xmath47 in terms of this @xmath45 operator , the majorana coupling term in eq .
( [ eq : h : mf ] ) can be rewritten as @xmath48 , which means that @xmath44 is nothing but the energy splitting between empty ( @xmath49 ) and filled ( @xmath50 ) states .
we assume that the qd is coupled to the nearest mf ( say @xmath51 ) and that the coupling to the other mf is negligible .
the qd - mf tunneling amplitude @xmath52 characterizes the tunneling rate latexmath:[$\gamma_{m } = \pi \rho_{\textrm{dot } } in the qd at the fermi level .
it should be noted that due to the helical property of the mf states , only one of the qd spin orientations ( say spin-@xmath15 ) hybridizes with the mfs @xcite . in order to clarify the genuine effect by the mf it is useful to compare the mf system with the non - mf system having the same configuration .
explicitly , we replace the mf operator in eq .
( [ eq : h : mf ] ) by an ordinary spin - polarized fermion operator @xmath54 [ see fig .
[ fig : model](b ) ] . in this sf case , the last line of the hamiltonian eq .
( [ eq : h : mf ] ) is changed into @xmath55 in fact , the first term is effectively same as the mf coupling term ( except a constant term ) if @xmath54 operator is replaced by @xmath45 operator .
the difference between two cases arises in the tunneling term . in the sf case , the second term in eq .
( [ eq : h : sf ] ) represents the hopping of an electron between the qd and the sf level . in the mf case ,
the tunneling term @xmath56\end{aligned}\ ] ] includes not only the electronic hopping but also the cooper pair creation / annihilation .
this superconducting coupling should distinguish the mf case from the sf case .
we are interested in measuring the charge current through the qd upon the voltage bias and/or temperature gradient between two terminals [ see fig . [
fig : model ] ] .
the current is the expectation value of the charge current operator which is given by the time derivative of the dot occupation @xmath57 . using the equation - of - motion technique , the charge current
can be expressed in terms of the retarded qd green s function @xmath58 as @xmath59 where @xmath60 is the fermi - dirac distribution with the voltage @xmath61 and the temperature @xmath62 and @xmath63 is the qd spectral density .
assuming that the temperature gradient @xmath64 and the voltage bias @xmath65 across the leads are small enough , the charge current can be linearized as @xmath66 where the electrical conductance @xmath1 and the thermoelectrical conductance @xmath0 are given by @xmath67 here @xmath68 is the equilibrium fermi - dirac distribution function at the fermi energy @xmath69 and the background temperature @xmath2 . in the low temperature limit
the two linear conductances can be analytically obtained by using a sommerfeld expansion , then @xmath70 where @xmath71 is the dot transmission .
it should be noted that @xmath0 becomes finite only if the spectral density is asymmetric with respect to the fermi level : if the spectral density has larger weight at @xmath72 in the particle ( hole ) part than in the hole ( particle ) part , @xmath0 ( in unit of @xmath73 ) becomes positive ( negative ) at temperature @xmath2 .
the nrg method @xcite is known to be highly effective in studying the effect of the strong coulomb interaction nonperturbatively .
we have already used the nrg technique for investigating the electric property of our system in ref . [ ] , finding very interesting interplay between the majorana and kondo physics . for a full review of the nrg
, one can refer to ref .
below we give a brief introduction of the nrg used in this work .
the nrg method consists in a series of consecutive exact diagonalizations of an appropriately discretized form of @xmath74 represented by a linear tight - binding chain . in each partial diagonalization
step the spectrum is truncated by retaining only those low - energy excitations that contains the strongest coupling terms .
better improvements of the nrg approach such as the density - matrix nrg @xcite , the @xmath75-averaging , and refinement of the logarithmic discretization @xcite allows us to calculate the dynamical properties at higher energies in a reliable and rather accurate way , which is essential for our thermoelectric study .
the nrg procedure also exploits the existence of symmetries for better efficiency .
note that this system has two conserved quantities : @xmath76 = [ p_\down,{{\mathcal{h } } } ] = 0 $ ] where @xmath77 and @xmath78 are charge number operator for spin-@xmath79 electrons and parity operator for the sum @xmath80 of spin-@xmath15 electrons and @xmath45 electrons , respectively .
note that the qd - mfs hopping changes @xmath80 by even numbers only . to obtain the electrical conductance @xmath1 and the thermoelectrical conductance @xmath0 , at each iteration in the backward stage
, the spin - resolved qd spectral density @xmath81 is calculated @xmath82 , \end{split}\end{aligned}\ ] ] where @xmath83 and @xmath84 are the eigenstates and the corresponding eigenenergies obtained at iteration @xmath85 .
the reduced density matrix @xmath86 is iteratively calculated in backward process , starting at the final iteration @xmath87 which is determined by the condition that @xmath88 where @xmath89 is the energy scale corresponding to iteration @xmath87 . at the final iteration ,
the reduced matrix is given by @xcite @xmath90 since we are interested in the kondo regime , throughout our work , we have used the following values of the parameters unless specified otherwise : @xmath91 , @xmath92 , and @xmath93 using the bandwidth @xmath32 as the energy unit ( @xmath94 ) .
for an ideal tsw the two mf end states have zero overlap .
the parameter that quantifies the degree of overlap between the two end states is denoted by @xmath44 that is considered null through this section . here
, we present our nrg results for both electrical and thermoelectrical conductances in this case .
indeed , the electric response in the qd - mf setup was discussed in ref .
[ ] , however for completeness we briefly explain the main findings for the electrical conductance and discuss in detail the thermoelectrical response of our device .
importantly , as we discuss below , @xmath0 exhibits different behaviors for the weak- ( @xmath96 ) and strong- ( @xmath97 ) coupling regimes . .
left and right panels correspond to the weak- ( @xmath96 ) and strong - coupling ( @xmath97 ) regimes , respectively : [ ( a),(b ) ] qd spin-@xmath79 spectral density , [ ( c),(d ) ] qd spin-@xmath15 spectral density , [ ( e),(f ) ] electrical conductance , and [ ( g),(h ) ] thermoelectrical conductance for annotated values of @xmath52 .
here we have used @xmath98 ( @xmath99 and @xmath100 ) and @xmath101 .
@xmath102 with these parameters.,width=321 ] first , we investigate the weak - coupling regime ( @xmath103 ) [ see the left panels of fig . [
fig : mf : em0 ] ] .
when the qd is uncoupled from the tsw , i.e. , when @xmath104 , the qd spin - resolved spectral densities @xmath105 [ see fig .
[ fig : mf : em0](a ) and [ fig : mf : em0](c ) ] exhibit a kondo resonance peak located at @xmath106 with a half - width @xmath107 and two side resonance peaks at @xmath108 . for a finite but weak qd - mf coupling ( @xmath109 ) , while @xmath110 is not affected by the coupling to the mf since the spin @xmath79 is not directly coupled to the mf [ see fig . [
fig : mf : em0](a ) and [ fig : mf : em0](c ) , for @xmath111 , @xmath112 , @xmath113 exhibits a half - dip whose width is comparable to @xmath5 [ see fig .
[ fig : mf : em0](c ) ] .
the half - fermionic dip is ascribed to the destructive interference between the kondo resonance and the mf states . as a result
an anti - fano resonance occurs due to the half - fermionic nature of the mf state @xcite .
the resultant linear conductance @xmath1 as a function of the temperature is shown in fig . [
fig : mf : em0](e ) . at low temperatures ( @xmath114 ) , @xmath115 , owing to the half - fermionic anti - fano resonance .
for larger temperature ( @xmath116 ) the conductance restores the kondo unitary - limit value , @xmath117 , eventually decreasing to @xmath118 for @xmath119 because of the coulomb blockade and the destruction of the kondo effect .
the thermoelectrical conductance @xmath0 displays more interesting features as shown in fig .
[ fig : mf : em0](g ) . before addressing the thermoelectric effect ascribed to the majorana physics ,
we first need to understand the behavior of @xmath0 due to the kondo effect itself . in the symmetric case , when @xmath120 , with @xmath121 electron - hole symmetry holds and the thermoelectrical conductance vanishes completely .
away from this special situation ( @xmath122 ) the kondo resonance peak in @xmath105 is always asymmetric : the kondo peak is slightly shifted toward the positive ( negative ) frequency part for @xmath123 ( @xmath124 ) , which leads to a positive ( negative ) peak in @xmath0 at @xmath125 @xcite : as shown in fig .
[ fig : mf : em0](g ) @xmath0 has a positive peak at @xmath126 since @xmath127 .
this asymmetry can be understood in terms of the slope of the spectral density at the fermi level @xmath69 . from the luttinger theorem @xcite ,
the derivative of the spectral density is given by @xmath128 where @xmath129 , @xmath130 is the real part of the retarded self - energy per spin and @xmath131 is the dot occupation per spin . for @xmath123 ( @xmath124 ) ,
@xmath131 is less ( larger ) than 1/2 so the derivative becomes positive ( negative ) and , accordingly , the kondo resonance is slightly shifted toward the positive ( negative ) frequency . at larger frequencies ,
the side peaks in @xmath105 are responsible for the two peaks in @xmath0 at @xmath132 and @xmath133 whose signs are opposite .
hence , in the kondo regime , @xmath0 experiences two sign changes as @xmath2 increases from @xmath107 to high temperatures as predicted in ref .
[ ] and reproduced in our calculations .
in the presence of the mf , on the other hand , an additional peak , whose sign is opposite to that of the kondo - induced peak , occurs at @xmath134 .
it is surely due to the particle - hole asymmetry of the half - dip in @xmath113 whose width is @xmath14 [ see fig . [
fig : mf : em0](c ) ] .
this asymmetry also originates from that of the kondo resonance peak . the destructive interference responsible for the dip is stronger at the frequency side in which the kondo peak is located . as a result ,
the dip is also slightly shifted toward the same frequency side as the kondo peak , leading to the asymmetry of the dip in the spectral density and the observed peak in @xmath0 . as @xmath135 increases
, the half - dip is widened and eventually trims the kondo resonance peak when the dip width is comparable to the kondo peak width [ see fig . [
fig : mf : em0](c ) ] .
hence , the mf - induced peak in @xmath0 disappears when @xmath136 . in the strong - coupling regime ,
the kondo effect still survives but is modified as follows : first , the kondo peak in @xmath110 shifts with increasing @xmath52 [ see fig . [
fig : mf : em0](b ) ] .
the peak moves toward positive ( negative ) frequencies for @xmath123 ( @xmath124 ) .
it is attributed to the induced zeeman splitting @xmath138 caused by the coulomb interaction and the broken time - reversal symmetry in the tsw @xcite .
while its unnormalized value for @xmath139 is given by @xmath140 for a general value of @xmath44 , in fact the renormalization due to the coupling to the leads makes it larger than @xmath141 in magnitude @xcite . on the other hand , @xmath113 is not affected by the induced zeeman splitting because it is directly coupled to the zero - energy mf and pinned to the fermi level .
the height of @xmath142 is fixed at @xmath143 [ see fig . [
fig : mf : em0](d ) ] .
since the kondo peak in @xmath110 is shifted , the linear conductance at low temperatures is smaller than the weak - coupling - limit value , @xmath144 and decreases to @xmath145 as @xmath52 increases [ see fig . [
fig : mf : em0](f ) ] .
the effect of the qd - mf coupling in the thermoelectrical conductance is the shift of the kondo - induced peak in @xmath0 with increasing @xmath52 [ see fig . [
fig : mf : em0](h ) ] .
the shift is obviously due to that of the kondo peak in @xmath110 by the induced zeeman splitting explained above .
hence , the position of the kondo - induced peak in @xmath0 should follow the induced zeeman splitting : @xmath146 .
. left and right panels correspond to the weak- ( @xmath96 ) and strong - coupling ( @xmath97 ) regimes , respectively : [ ( a),(b ) ] electrical conductance and [ ( c),(d ) ] thermoelectrical conductance for annotated values of @xmath52 .
here we have used the same values of the parameters as in fig .
[ fig : mf : em0].,width=321 ] now one should ask whether the observed mf - induced features in @xmath0 are the genuine effects of the majorana physics or not . for that purpose , we have repeated the same calculations by using the hamiltonian , eq .
( [ eq : h : sf ] ) with a zero - energy spin - polarized level ( @xmath147 ) .
first , we have obtained the qualitatively same spectral densities as in the mf case : @xmath110 features the kondo resonance peak at @xmath148 in the weak - coupling regime and its peak moves due to the effective zeeman splitting in the strong - coupling regime . @xmath113 also exhibits the dip structure due to the anti - fano resonance by the side - coupled spin - polarized level .
one and only difference between two cases is that the dip in the sf case is a full dip so that @xmath149 . as a result , the low - temperature value of the linear conductance , which is @xmath150 in the weak - coupling regime , is decreased compared to the mf case [ see figs .
[ fig : sf : em0](a ) and [ fig : sf : em0](b ) ] . however , the other features of the linear conductance is the same as those in the mf case .
the thermoelectrical conductance also exhibits the same qualitative behavior as that of the mf case [ see figs . [
fig : mf : em0](c ) and [ fig : mf : em0](d ) ] : the appearance of the mf - induced peak at @xmath24 in the weak - coupling regime and the shift of the kondo - induced peak at @xmath151 in the strong - coupling regime .
hence , for @xmath152 , while the two cases exhibit different behaviors of the electrical conductance , the measurement of the thermoelectrical conductance can not distinguish the mf state from the ordinary spin - polarized state .
the overlap between majorana states depends on the relative length between the tsw length and the localization length of the majorana states .
hence , by changing the relative length ( for example , the tsw length by tuning the position - dependent gate potential or the mf localization length by controlling the superconducting gap and/or the gate potential ) , the overlap amplitude @xmath44 can be varied .
similarly , the level of the ordinary spin - polarized state can be tuned by electric or magnetic methods . in this section
, we examine whether the mf can display a genuine thermoelectric effect for finite @xmath44 , which can not be reproduced by the sf level .
. left and right panels correspond to the weak- ( @xmath154 ) and strong - coupling ( @xmath155 ) regimes , respectively : [ ( a),(b ) ] qd spin-@xmath79 spectral density , [ ( c),(d ) ] qd spin-@xmath15 spectral density , [ ( e),(f ) ] electrical conductance , and [ ( g),(h ) ] thermoelectrical conductance for annotated values of @xmath44 .
here we have used the same values of the parameters as in fig .
[ fig : mf : em0].,width=321 ] the left panels of fig .
[ fig : mf ] show the qd spectral densities , the linear conductance , and the thermoelectrical conductance in the weak - coupling regime for @xmath153 . as for @xmath147 ,
the kondo peak in @xmath110 remains intact irrespectively of the value of @xmath44 [ see fig . [
fig : mf](a ) ] .
an interesting change occurs in @xmath113 : inside the half - dip , a central peak is developed for finite values of @xmath44 [ see fig . [
fig : mf](c ) ] , which originates from the interference between the mfs , @xmath51 and @xmath156 .
finite coupling @xmath157 between @xmath158 and @xmath159 broadens the @xmath156 spectral density with the width @xmath160 and in turn the scattering from the resonance level of @xmath156 results in the destructive interference for @xmath51 , forming a full dip in the @xmath51 spectral density .
hence , inside the dip , the anti - fano resonance which suppressed the kondo effect does not happen any longer since @xmath51 which would interfere with the qd spin has no excitation in that range of energy , and the kondo effect is restored .
as @xmath44 increases , the central peak gets wider and the full kondo resonance peak is restored when @xmath161 or @xmath162 .
the restoration of the kondo effect at low frequencies is well reflected in the linear conductance for @xmath163 [ see fig . [
fig : mf](e ) ] .
the central peak in @xmath113 due to the restoration of the kondo effect also has its trace on the thermoelectrical conductance .
figure [ fig : mf](g ) shows that an additional peak in @xmath0 at @xmath164 is formed on top of @xmath0 in the @xmath152 case .
in fact , the central peak is also asymmetric like the original kondo peak .
it should be noted that the dip in the @xmath51 spectral density is symmetric since the mf always maintains the particle - hole symmetry .
so , the asymmetry of the central peak has the same origin as the kondo peak , and the signs of the kondo - induced peak and the finite-@xmath44-induced peak in @xmath0 are always same .
as @xmath44 increases up to @xmath165 , the half - dip and the inner central peak in @xmath113 cancels out each other .
therefore , the two mf - related peaks are merged and disappear completely when @xmath166 , making @xmath167 for @xmath168 . in the strong - coupling regime
the finite @xmath44 removes the half - fermionic fano resonance at the fermi level . as a result , @xmath169 , which was pinned to 1/2 at @xmath106 for @xmath147 , becomes larger than 1/2 , restoring the kondo correlation a bit [ see fig . [
fig : mf](d ) ] . instead ,
since the mf states are now energy - split , the half - value pinning is retained in @xmath170 for @xmath171 . in the frequency windows @xmath172 we observe in @xmath113 an asymmetric peak ( with a small dip in the opposite side )
whose position is approximately at @xmath173 .
the peak in @xmath113 is located in the opposite side with respect to the zeeman - splitting peak in @xmath110 [ see fig . [
fig : mf](b ) ] .
we speculate that the asymmetric peak in @xmath113 is the outcome of the competition between the induced zeeman splitting on spin @xmath15 ( so the peak would appear at @xmath174 ) and the half - fermionic pinning at the energy - split mf level .
note that this central peak has the opposite asymmetry with respect to that in the weak - coupling regime .
hence , due to this asymmetric peak structure , an additional peak in @xmath0 arises at @xmath175 , whose sign is opposite to that of the kondo - induced peak [ see fig . [
fig : mf](h ) ] .
finally , owing to the increase in @xmath113 at low frequencies , the low - temperature value of the linear conductance @xmath1 increases a bit [ see fig . [
fig : mf](f ) ] . for @xmath176 , the mf levels which are at @xmath177 do not interfere with the kondo resonant level any longer so that the kondo physics in the qd is completely revived @xcite : the peak in @xmath0 at @xmath178 disappears . .
left and right panels correspond to the weak- ( @xmath179 ) and strong - coupling ( @xmath180 ) regimes , respectively : [ ( a),(b ) ] qd spin-@xmath15 spectral density , [ ( c),(d ) ] electrical conductance , and [ ( e),(f ) ] thermoelectrical conductance for annotated values of @xmath44 .
here we have used the same values of the parameters as in fig .
[ fig : mf : em0].,width=321 ] spectral density and ( b ) thermoelectrical conductance in the sf case for @xmath181 in the weak - coupling case .
the corresponding values of @xmath16 are annotated .
here we use the same values of parameters as in the weak - coupling regime of fig .
[ fig : sf].,width=321 ] it is obvious that the additional low - temperature features in @xmath0 of the mf case with finite @xmath44 can not be reconstructed by the sf case with zero @xmath44 . for the sf case ,
the effect of @xmath44 being finite on the qd spectral density is quite simple .
@xmath110 is not so affected as long as @xmath182 , and the full dip in @xmath113 due to the anti - fano resonance by the spin - polarized level is formed at @xmath183 [ see figs . [ fig : sf](a ) , [ fig : sf](b ) and [ fig : sf : neg](a ) ] .
the only difference between the weak- and strong - coupling regimes is whether the dip is so narrow as to be confined inside the kondo resonance peak or wide enough to suppress it .
hence , the electrical and thermoelectrical conductances are qualitatively similar to those in the @xmath152 case : compare figs .
[ fig : sf : em0](a - d ) and [ fig : sf](c - f ) , respectively .
note that for @xmath153 and in the weak - coupling regime the sf - induced peak in @xmath0 is formed at @xmath184 .
for @xmath181 , the dip moves toward the negative frequency [ see fig .
[ fig : sf : neg](a ) ] so that the sign of the sf - induced peak is reversed [ see fig . [
fig : sf : neg](b ) ] .
therefore , as reflected in figs .
[ fig : sf ] and [ fig : sf : neg ] , the sf case with finite @xmath44 can not reproduce the temperature dependence of @xmath0 in the mf case with finite @xmath44 . hence , the observed features of @xmath0 in the mf case are genuine effects only the mf can generate .
the reason why the mf and sf cases become distinguishable from each other for @xmath185 is that the mf state is particle - hole symmetric while the sf state is not . in the mf case the only source of the asymmetry
is the kondo effect while in the sf case the sf level can also make the system asymmetric . for @xmath152 this difference did not show up since the levels are at the fermi level and particle - hole symmetric
in the previous sections it is shown that the mf adds additional peaks to the temperature dependence of @xmath0 . since the mf is particle - hole symmetric , the finite @xmath0 is attributed to the asymmetry of the kondo resonance peak , or in other words , nonzero @xmath186 .
therefore , by tuning the value of @xmath186 or the gate voltage on the qd , the magnitude and the sign of @xmath0 can be controlled . in this section
we investigate the gate dependence of @xmath0 in the kondo regime . before addressing the weak- and strong - coupling regimes separately ,
we discuss the common features seen in both cases . as shown in figs .
[ fig : l : gate : wc ] and [ fig : l : gate : sc ] , the thermoelectrical conductance vanishes at @xmath187 , irrespectively of the values of the other parameters .
it is because at @xmath188 the whole system is particle - hole symmetric even with nonzero @xmath44 .
in addition , @xmath0 is always odd function upon @xmath189 .
it is attributed to the fact that the mf is particle - hole symmetric and the qd is the only source for the asymmetry .
therefore , as the temperature ( and other parameters ) changes , the sign of @xmath0 changes globally as observed in the non - interacting case @xcite . in the weak - coupling regime for [ ( a),(b ) ] long tsw case ( @xmath18 and @xmath191 ) and [ ( c),(d ) ] short tsw case ( @xmath192 and @xmath154 ) . in ( b ) and ( d ) the thermoelectrical conductances @xmath0 are drawn as functions of @xmath186 at fixed temperatures , or along the straight lines in ( a ) and ( c ) , respectively : ( 1 ) mf - induced peak ( @xmath193 ) , ( 2 ) kondo - induced peak ( @xmath194 ) , ( 3 ) charge - fluctuation - induced peak ( @xmath195 ) , ( 4 ) restored - kondo - induced peak ( @xmath196 ) , ( 5 ) mf - induced peak ( @xmath197 ) , and ( 6 ) kondo - induced peak ( @xmath198).,width=321 ] figure [ fig : l : gate : wc ] shows the thermoelectrical conductance as a function of the temperature @xmath2 and the asymmetry parameter @xmath186 ( or the qd level ) in the weak - coupling regime of both the @xmath147 and @xmath185 cases .
the main qualitative features of the temperature dependence of @xmath0 observed in the previous sections do not change with @xmath186 , as seen in figs .
[ fig : l : gate : wc](a ) and [ fig : l : gate : wc](c ) .
instead , the position and height of the peaks are gradually changed .
owing to the dependence of the kondo temperature on the qd level , the kondo - induced peaks [ see curves ( 2 ) and ( 6 ) in fig . [
fig : l : gate : wc ] ] move toward the larger temperature as @xmath199 increases .
however , the mf - related peaks , that is , the mf - induced and restored - kondo - induced peaks [ see curves ( 1 ) , ( 4 ) and ( 5 ) in fig . [
fig : l : gate : wc ] ] do not move so much upon changing @xmath186 since @xmath14 and @xmath22 are quite immune to the qd level in the weak - coupling regime [ see figs .
[ fig : l : gate : wc](a ) and [ fig : l : gate : wc](c ) ] .
the heights of the peaks usually increase with increasing @xmath199 [ see figs . [
fig : l : gate : wc](b ) and [ fig : l : gate : wc](d ) ] .
it is because the asymmetry is the larger for the larger @xmath199 . in the strong - coupling regime for [ ( a),(b ) ] long tsw case ( @xmath18 and @xmath201 ) and [ ( c),(d ) ] short tsw case ( @xmath202 and @xmath203 . in ( b ) and ( d ) the thermoelectrical conductances @xmath0 are drawn as functions of @xmath186 at fixed temperatures , or along the straight lines in ( a ) and ( c ) , respectively : ( 1 ) zeeman - splitting - induced peak ( @xmath198 ) , ( 2 ) charge - fluctuation - induced peak ( @xmath204 ) , ( 3 ) restored - kondo - induced peak ( @xmath193 ) , and ( 4 ) zeeman - splitting - induced peak ( @xmath205).,width=321 ] finally , we consider the gate dependence of the thermoelectrical conductance in the strong - coupling regime . as in the weak - coupling regime ,
the main features of @xmath0 discussed before are present for all values of @xmath186 , as can be seen in fig .
[ fig : l : gate : sc ] . for @xmath147
[ see figs .
[ fig : l : gate : sc](a ) and [ fig : l : gate : sc](b ) ] , the gate dependence of @xmath0 in strong coupling regime has similar behavior with the weak - coupling regime : compare ( 2 ) and ( 3 ) in fig .
[ fig : l : gate : wc](b ) with ( 1 ) and ( 2 ) in fig .
[ fig : l : gate : sc](b ) , respectively .
@xmath0 increases with @xmath199 . for @xmath185
[ see figs .
[ fig : l : gate : sc](c ) and [ fig : l : gate : sc](d ) ] , on the other hand , one can observe a non - monotonic behavior of @xmath0 with respect to @xmath186 [ see ( 3 ) in fig . [
fig : l : gate : sc](d ) ] .
in fact , the peak due to the restored kondo effect is shifted as @xmath186 is varied [ see fig . [
fig : l : gate : sc](c ) ] : the peak first moves toward higher temperature as @xmath199 increases from zero and at some value of @xmath199 the shift is reversed , returning back as @xmath199 approaches @xmath37 , while the change in the peak position is within the order of magnitude @xmath206 .
this shift of the peak results in the observed non - monotonic behavior of @xmath0 at a fixed temperature .
surely , in the strong - coupling regime , the majorana hybridization @xmath22 should be renormalized depending on the value of @xmath207 .
while it is not easy to quantify the renormalization , there is an intuitive way to understand the non - monotonic behavior : the restore kondo peak in @xmath208 in the strong - coupling regime [ see fig . [
fig : mf](d ) ] is related to the induced zeeman splitting @xmath138 acting on the spin @xmath15 , as mentioned before . for small @xmath199
, @xmath138 is also small [ see eq .
( [ eq : delta ] ) ] such that @xmath209 . in this case
, the peak follows the induced zeeman splitting @xmath138 , explaining the initial increase of the peak position .
however , once @xmath138 exceeds @xmath22 , the pinning by the mfs prevents the peak from moving toward higher temperature further .
instead , as @xmath199 increases , the qd density of states @xmath210 increases due to the resonance levels , and accordingly @xmath22 decreases ( note that @xmath211 ) .
hence , the peak in @xmath208 moves toward lower temperature .
this non - monotonic behavior of @xmath0 is the outcome of the interplay between the kondo and majorana physics so its experimental observation can be much stronger evidence of the existence of the mfs .
we have investigated the thermoelectric properties of strongly correlated quantum dot side - coupled to the majorana fermions .
the electrical conductance @xmath1 and the thermoelectrical conductance @xmath0 through qd are calculated using the numerical renormalization group method in both weak- and strong - coupling regimes .
the side - coupled majorana fermion causes the anti - fano resonance in weak - coupling regime and the induced zeeman splitting in strong - coupling regime .
these two effects considerably modify the shape of the kondo resonance peak in the qd spectral densities and eventually transform the structure of @xmath0 and @xmath1 . especially , with a finite overlap between mfs ( @xmath212 ) , the thermoelectrical conductance @xmath0 exhibits conspicuous characteristics additional peaks in @xmath0 and the non - monotonic dependence of @xmath0 on the gate voltage which can not be attained by an ordinary fermion .
we expect that our work can act as a guide for finding the unique signal of majorana fermion in thermoelectric devices .
this work was supported by mineco grant no .
fis2011 - 23526 , in part by the kavli institute for theoretical physics through nsf grant phy11 - 25915 , and by the national research foundation of korea ( nrf ) grants funded by the korea government ( msip ) ( no .
2011 - 0030046 ) . | we investigate the linear thermoelectric response of an interacting quantum dot side - coupled by one of two majorana fermions ( mfs ) formed at the ends of a topological superconducting wire .
we employ the numerical renormalization group technique to obtain the thermoelectrical conductance @xmath0 as well as the electrical conductance @xmath1 when the background temperature @xmath2 and the dot gate are tuned .
we distinguish two transport regimes in which @xmath0 displays different features : the weak- @xmath3 and strong - coupling @xmath4 regimes , where @xmath5 and @xmath6 are the majorana - dot coupling and the kondo temperature , respectively . for an ideal ( infinitely long ) nanowire where the majorana end states do not overlap @xmath7 ,
the thermoelectrical conductance @xmath0 in the weak - coupling regime exhibits a peak at @xmath8 .
this peak is ascribed to the anti - fano resonance between the asymmetric kondo resonance and the zero - energy mf mode .
interestingly , in the strong - coupling regime , the kondo - induced peak in @xmath0 is shifted due to the mf - induced zeeman splitting in the dot . for finite but small @xmath9
, the interference between two mfs restores the kondo effect in the dot in a smaller energy scale @xmath10 and gives rise to an additional peak in @xmath0 at @xmath11 , whose sign is opposite to that at @xmath12 . in the strong - coupling regime
this additional peak can cause a non - monotonic behavior of @xmath0 with respect to the dot gate .
finally , we examine the case in which an ordinary spin - polarized fermion is coupled to the dot and identify the fingerprint of mfs by comparing two cases . |
in the last years there has been an increasing interest in studying the galactic bulge resulting from the accumulation of more spectroscopic data . in particular , from these data a rather complex picture for the bulge formation arose : the existence of at least two main stellar populations in the bulge was suggested by several studies ( babusiaux et al . 2011 ; gonzalez et al .
2011 ; hill et al . 2011 ; bensby et al .
2011 , robin et al .
2012 , uttenthaler et al . 2012 ; bensby et al . 2013 ) .
the two populations arising from the spectroscopic data are characterized by a metal - poor one , with characteristics typical of stars belonging to a spheroid , and another one more metal - rich and with characteristics of bar kinematics .
also the [ @xmath0/fe ] ratios of these two populations seem different in the sense that the metal rich population shows lower [ @xmath0/fe ] ratios ( hill et al .
grieco et al .
( 2012 , hereafter gmpc12 ) tried to model these two populations and concluded that the metal - poor population formed on a very short timescale by means of an intense burst of star formation , whose efficiency was a factor of 20 higher than in the galactic thin disk .
the more metal - rich population instead should have formed on a longer timescale and with lower star formation efficiency and it was created by the bar evolution . while in the gmpc12 model the bulge was treated as a unique single zone with radius of 2kpc , here we present a model for the central region of the bulge , the galactic center ( gc ) . in particular we will focus on a region of 200 pc radius .
the gc region has an extremely rich environment : intense star formation with formation of a bulk of massive stars and three of the most massive young star clusters in the galaxy . in the last years
many assumptions have been made about the nature of the inner population and on the possible causes of the star formation activity in the last few hundreds myr .
genzel et al .
( 2006 ) pointed out that most of the stars in the gc seem to have formed @xmath4 @xmath3 ago , probably at the same time as the galactic bulge .
then the star formation rate ( sfr ) dropped to a minimum a few @xmath3 ago rising again in the past few million years .
the star formation history was derived in the gc from the h - r diagram of the red giants ( pfhul et al .
2011 ) , while the present time sfr was derived from the infrared luminosity of the young massive stars forming in the large reservoir of molecular gas present in the gc ( yusef - zadeh et al . 2009 ) or by counting these young massive stars ( immer et al .
one important issue relevant to the nature of the stellar populations in the gc is the possible presence of a second bar .
however , the presence of a double bar in our galaxy , similar to what can be observed in external galaxies ( laine et al . 2002 ; erwin 2004 ) is still under debate . alard ( 2001 ) , nishiyama et al .
( 2005 ) , and gonzalez et al .
( 2011 ) claim that there is an inner structure distinct from the large - scale galactic bar , with a different orientation angle which could be associated with a secondary , inner bar . on the other hand
, gerhard & martinez - valpuesta ( 2012 ) can explain this inner structure by dynamical instabilities from the disk without requiring a nuclear bar .
very recently , ryde & schultheis ( 2015 ) have measured , from high resolution infrared spectroscopy , abundances of mg , si , ca and fe in 9 m - type giants in the gc region . they found a metal - rich population at [ fe / h][email protected] dex with rather low [ @xmath0/fe ] ratios with the possible exception of ca . their metallicity distribution function ( mdf )
is peaked at a [ fe / h ] slightly higher than solar and agrees with previous data of cunha et al .
cunha et al . (
2007 ) had also measured high resolution infrared spectra for a sample of cool stars within 30 pc from the gc ; their results showed that [ o / fe ] and [ ca / fe ] are enhanced by 0.2 and 0.3 dex , respectively .
they also found that the total range in [ fe / h ] in the gc is narrower than that found for the older bulge population .
the aim of this paper is to compare our results with the data of ryde & schultheis ( 2015 ) in order to impose constraints on the formation and evolution of the gc .
the analysis of the abundance ratios , such as [ @xmath0/fe ] ratios , versus [ fe / h ] can allow us to understand the timescale of the process of formation of the gc region .
in fact , [ @xmath0/fe ] ratios can be used as cosmic clocks because of the time - delay model .
the time - delay model is the common interpretation of abundance patterns in terms of different chemical elements produced from different stars and on different timescales . in this context ,
starting from the fact that @xmath0-elements are produced on short timescales by core - collapse sne , whereas fe is produced partly in core - collapse sne but mainly in type ia sne exploding on longer timescales , we can interpret the observed abundances .
high [ @xmath0/fe ] ratios at low [ fe / h ] indicate that the star formation period was short enough to prevent the type ia sne from polluting heavily the gas with fe . on the other hand
, we always expect that the [ @xmath0/fe ] ratios are lower at high [ fe / h ] because of type ia sn contribution
. however , due to the combinationation of the star formation rate ( sfr ) and the time - delay in element production , objects where the sfr has been very intense show high [ @xmath0/fe ] ratios up to even solar [ fe / h ] values .
this is because @xmath0-elements and part of fe are produced fast by core - collapse sne , which are able to enrich substantially the ism in fe before the bulk of this element is restored by sne ia ( see matteucci , 2012 ) .
therefore , the comparison between observed and predicted abundances can tell us about the history of the chemical evolution of the gc and in particular on the origin of its stellar populations . to perform such a comparison
, we adopt a model very similar to that presented in gmpc12 , in particular the model describing the classical bulge population which formed on a short timescale as a result of an intense burst of star formation ( the population called metal - poor ) , applied to a central region of only 200 pc of radius . in order to reproduce the star formation observed at the present time , we also overimpose on the history of star formation , deriving from our model , a recent burst producing the same quantity of stars that have been observed .
the paper is organized as follows : in section 2 we describe the chemical evolution model , in section 3 we describe the stellar spectroscopic data as well as the sfr estimates . in section 4
we present our results and the comparison with data .
in section 5 we draw some conclusions .
the chemical evolution model we adopt here is the model of gmpc12 for the metal - poor ( mp ) stellar population in the galactic bulge . in other words
, it is the model assuming a strong initial burst of star formation triggered by the heavy gas infall coming from the initial collapse that gave rise to the halo and the bulge .
this model has been adapted here to the small central region around the galactic center ( gc ) .
the assumed gas accretion law is : @xmath6 where @xmath7 represents a generic chemical element , @xmath8 is an appropriate infall timescale fixed by reproducing the observed stellar metallicity distribution function , and @xmath9 is a parameter constrained by the requirement of reproducing the current total surface mass density in the central region of the galactic bulge ; in particular , we assume a present time surface mass density at @xmath10 pc , @xmath11 , with @xmath1214 @xmath3 being the lifetime of the galaxy , and @xmath13 the galactocentric distance . we have obtained this value by assuming a mass distribution in the bulge which follows a sersic ( 1968 ) profile .
the quantity @xmath14 is the abundance of the element @xmath7 in the infalling gas .
the parameter @xmath8 will be allowed to vary in order to find the best agreement with the present time observables ( abundances , sfr and stellar mass ) .
the star formation rate ( sfr ) is a schmidt- kennicutt law : @xmath15 with @xmath16 as in previous papers ( ballero et al .
2007 ; cescutti & matteucci , 2011 ; gmpc12 ) .
the quantity @xmath17 is the efficiency of star formation , namely the sfr per unit mass of gas , and for the bulge is assumed to be high relative to that normally assumed in the solar vicinity ( @xmath18 , see for example chiappini et al .
1997 ) , thus simulating a star burst .
the efficiency of star formation is in units of @xmath19 and represents the inverse of the time scale in which all the gas is consumed .
this approach was first suggested by matteucci & brocato ( 1990 ) and then proven to be required for reproducing the bulge properties in all the following papers dealing with the chemical evolution of the bulge .
here we assume several values for this parameter varying from @xmath17 from 20 to 200 @xmath20 .
we explore different initial mass functions ( imfs ) , in particular : * the one suggested by ballero et al .
( 2007 ) : @xmath21 with @xmath22 for @xmath23 and @xmath24 for @xmath25 in the mass range @xmath26 and @xmath27 . * the normal salpeter ( 1955 )
imf ( x=1.35 in eq .
( 3 ) ) in the mass range @xmath26 , with @xmath28 . *
the three - slope kroupa et al .
( 1993 ) imf : @xmath29 with @xmath30 . * the chabrier ( 2003 ) imf : @xmath31 with @xmath32 .
we note that this imf is very similar to that of kroupa ( 2001 ) .
the nucleosynthesis prescriptions adopted here are different from those in grieco et al .
( 2012 ) and are the same as in romano et al .
( 2010 ) : they take into account the most recent calculations and have been tested on the chemical evolution of the solar neighbourhood and found to be the best ones .
in particular , the prescriptions for different stellar mass ranges are : * the yields of karakas ( 2010 ) for low and intermediate mass stars ( @xmath33 , lims ) , who has recomputed karakas & lattanzio s ( 2007 ) models with updated nuclear reaction rates and enlarged the grid of masses and metallicities . *
the yields for massive stars ( @xmath34 ) are taken from kobayashi et al .
( 2006 ) , which included metallicity - dependent mass loss , plus the yields of the geneva group for c , n , o ( meynet & maeder 2002a ; hirschi et al . 2005 ; hirschi 2007 ; ekstrom et al .
2008 ) , limited to the presupernova stage , but computed with both mass loss and rotation . * the yields from type ia sne are assumed to originate from exploding white dwarfs in binary systems and are taken from iwamoto et al .
the sn ia rate is computed as in the previous papers ( ballero et al . 2007 ; cescutti & matteucci , 2011 ; grieco et al . 2012 ) by assuming the single - degenarate model , as described in greggio & renzini ( 1983 ) and matteucci & recchi ( 2001 ) .
this particular formulation , namely the single - degenerate scenario , gives results very similar to the double - degenerate scenario for the progenitors of sne ia , as shown by matteucci et al .
( 2006 ; 2009 ) .
the nine m - type giants in the galactic - center region discussed here were analysed in ryde & schultheis ( 2015 ) using high spectral resolution in the k band , centred at 2.1 @xmath35 m .
they were selected using the nishiyama et al .
( 2009 ) dataset and the 3d interstellar extinction maps of schultheis et al .
( 2014 ) to ensure that these giants are situated in the galactic center region .
the low - resolution k band spectra were used to determine the effective temperatures ( @xmath36 ) with a precision of approximately @xmath37 .
the surface gravities ( @xmath38 ) , were determined by adopting a mean distance of 8 kpc and using dereddened h and ks magnitudes from the vvv survey of minniti et al .
( 2010 ) , resulting in a precision of less than 0.3 dex .
the high - resolution spectra , centered at @xmath39 m , were used to determine the metallicity ( [ fe / h ] ) from several fe lines and the ca , si , and mg abundances from several atomic lines . to derive the abundances ,
spectral synthesis was used with the software _ spectroscopy made easy ( sme ) _ , valenti & piskunov ( 1996,2012 ) , based on marcs spherical - symmetric , local thermodynamic equilibrium ( lte ) model atmospheres ( gustafsson et al . 2008 ) . the abundance from an observed
spectral line was found by minimising the @xmath40 , calculated from the comparison between synthetic spectra and the observations .
the best fit was found , with the abundance as a free parameter , by minimising the @xmath40 .
the microturbulence of @xmath41 was used in the analysis . the uncertainties , due to the uncertainties in the fundamental parameters ( @xmath36 , @xmath38 , [ fe / h ] , and @xmath42 ) , in the derived metallicities and abundance ratios are of the order of 0.10 - 0.15 dex .
there might also be unknown systematic uncertainties , such as non - lte effects , but since the standard red - giant star , arcturus , is among the analysed stars and gives reasonable results , these uncertainties can not be large .
unknown blends might also exist , especially for metal - rich stars , and will tend to overestimate the metallicities . however , the line blending is less severe in the k band compared to the optical region .
furthermore , since several fe lines were used , this effect would rather lead to an increased dispersion in the abundances from the different lines .
the molecules in the wavelength regions are predominantly due to cn , and are properly synthesised .
the metallicity distribution of ryde & schultheis ( 2015 ) agrees well with that of cunha et al .
these authors also measured ca abundances , which also agree well with those measured by ryde & schultheis ( 2015 ) .
the understanding of the physical processes occuring in the nuclear disk of our galaxy is crucial for the insight into the formation and evolution of our own milky way .
the central molecular zone ( cmz ) is the innermost @xmath2 200 arcmin region of the milky way .
it covers about @xmath43 in longitude and @xmath44 in latitude which is within about 300pc of the galactic center .
it is a giant molecular cloud complex with an asymmetric distribution of molecular clouds ( see e.g. morris & serabyn 1996 , martin et al .
2004 and oka et al . 2005 ) .
genzel et al .
( 2006 ) pointed out that most of the stars in the gc seem to have formed @xmath45 @xmath3 ago , probably at the same time as the galactic bulge .
then the sfr dropped to a minimum a few @xmath3 ago and increased again in the past few million years .
recent herschel observations ( molinari et al . 2011 ) show that the cmz is a 100 pc elliptical and twisted ring of cold dust , with 3 @xmath46 10@xmath47 m@xmath48 of gas .
this prodigious reservoir of molecular gas is an active region of star formation , with evidence of starburst activity in the last 100.000 years ( yusef - zadeh et al .
the gas pressure and temperature are higher in the cmz than in the average disk , conditions that favour a larger jeans mass for star formation and an initial mass function biased toward more massive stars .
furthermore , the presence of strong magnetic fields , tidal shear , and turbulence demonstrate that the conditions for star formation in the cmz are significantly different from those in the galactic disk ( see serabyn & morris 1996 , fatuzzo & melia 2009 ) . from the observational point of view
, the present star formation rate can be determined by counting massive young stellar objects ( immer et al .
2012 ) . in order to estimate bolometric luminosities from their flux densities at 15 @xmath35 m
, one has to correct for interstellar dust extinction ( schultheis et al .
the total luminosity of the young stellar objects has then been converted to a total mass assuming a salpeter imf for masses in the range of 6 to @xmath49 , while the low - mass objects ( @xmath50 ) have been converted using the kroupa ( 2001 ) imf . in order to calculate the sfr , immer et al .
( 2012 ) assumed a typical life - time of 1myr for the young stellar objects which leads to an average star formation rate of @xmath51 .
however , one has to take into account that the derived sfr are based on simple assumptions which could lead to significant errors .
previous studies ( crocker 2011 and yusef - zadeh et al .
2009 ) published a range of values of @xmath52 .
this is consistent with other studies using different tracers of gas and star formation acitvity leading to a present sfr of @xmath53 ( morris & serabyn 1996 , yusef - zadeh et al .
2009 , molinari et al .
2011 and immer et al .
2012 ) with a total gas mass in the order of @xmath54 .
given the total amount of gas in the cmz , the sfr is by a factor of @xmath55 lower than expected .
this suppression of star formation could be due to the high turbulent pressure ( see e.g. kruijssen et al .
pfhul et al .
( 2011 ) presented spatially resolved imaging and integral field spectroscopy for 450 cool giants within 1 pc from sgra*. they derived the star formation rate of the nuclear cluster in the galactic center from the h - r diagram of the red giant population .
they found that the sfr dropped from an initial maximum roughly 10 @xmath3 ago to a minimum at 1 - 2 @xmath3 ago and that increased again in the last few hundred myrs .
their results imply that @xmath280% of the total stellar mass in the gc formed more than 5 @xmath3 ago .
two of the most important observational constraints to understand the most probable scenario for the bulge formation are the stellar metallicity distribution function ( mdf ) of stars and the [ @xmath0/fe ] ratios .
originally , matteucci & brocato ( 1990 ) suggested that in order to fit the mdf of the bulge one should assume that it formed very quickly .
the consequence of the fast formation is a particular [ @xmath0/fe ] trend with super - solar values being present for a large interval of [ fe / h ] . in other words , while in the solar vicinity the `` knee '' of the [ @xmath0/fe ] ratios occurs roughly at [ fe / h]@xmath2 - 1.0 , in the bulge the knee appears at [ fe / h]@xmath2 0 .
this is due to the fact that in a regime of intense star formation , core - collapse sne from massive stars enrich the interstellar medium ( ism ) very fast in fe ( although they are not the main producers of this element ) , so when sne ia which produce the bulk of fe occur , the abundance of fe in the ism is already almost solar ( see matteucci 2001 , 2012 ) .
we compare here the gc data with the chemical evolution model ( hereafter the reference model ) for a classical bulge of gmpc12 .
we computed several numerical models by varying the most important parameters : the imf , the efficiency of star formation and the timescale of the infalling gas .
the reference model parameters are : @xmath17=25 @xmath20 , @xmath8=0.1 @xmath3 and the salpeter imf in agreement with the gmpc12 g metal poor population . the data trend of figure [ mgsica_imf ] shows the effect of an intense burst of star formation coupled with the delay in the fe production from sne ia : since star formation is very intense , the bulge very soon reaches a solar metallicity because of the core collapse sn explosions , which produce part of the fe . after a time delay necessary to the evolution of the binary system progenitors of sne ia ,
these sne , which produce the bulk of fe , start to explode changing the [ @xmath0/fe ] slope at a larger [ fe / h ] than in the solar vicinity .
the models in figure [ mgsica_imf ] show also the effect of varying the imf : the ballero imf and the chabrier imf give very similar results since they are favouring massive stars compared to the kroupa et al .
( 1993 ) and salpeter ( 1955 ) imfs .
in particular , the kroupa et al .
( 1993 ) imf which has been derived for the solar vicinity does not reproduce either the [ mg / fe ] or the [ ca / fe ] ratios , although it could reproduce the [ si / fe ] abundances . on the other hand , the salpeter ( 1955 )
imf can reproduce the [ mg / fe ] and [ si / fe ] ratios .
we note that none of the adopted imfs can reproduce the high values of the [ ca / fe ] ratios .
this could be due to the uncertainties still existing in the stellar yields of specific elements , but also to observational errors .
the abundance data are , in fact , more sensitive to the yield uncertainties rather than to the imf , whereas the stellar mdf is highly sensitive to the imf . in figure [ mgsica_sfe ]
we can see the effects of a different star formation efficiency on the [ mg / fe ] , [ si / fe ] , [ ca / fe ] vs. [ fe / h ] trends : the green solid line is the reference model while the black dashed line is a model with an intense initial starburst with efficiency @xmath17=200 @xmath20 .
this very high efficiency is probably unrealistic since it predicts that the gas is consumed in 5 myr . in any case , the differences among different models with @xmath56 and 200 @xmath20 are negligible . on the other hand , a @xmath57 as in the solar vicinity ( see chiappini et al .
1997 ) would produce different results , in particular , all the curves would be translated at lower values of [ @xmath0/fe ] ratios . in figure [ mgsica_tinf ]
we show the effect of varying the infall timescale .
the differences among different models are small and from the figure we can conclude that a timescale between 0.1 and 1.25 @xmath3 can be acceptable , whereas a timescale of 0.01 @xmath3 seems too short . in figure [ mdf_sfe ] we show the predicted stellar mdf for the fiducial model and the two different values of @xmath17 . here , we can see that the effect of increasing @xmath17 results in a narrowing of the predicted mdf .
it is worth noting that the number of observed stars in the gc is very small and therefore we do not aim at reproducing exactly the observed mdf but only at reproducing the position of the peak .
the infall timescales also affect the mdf and in figure [ mdf_tinf ] we show the predicted mdf for the chabrier ( 2003 ) imf and different infall timescales ; we note that for this imf the infall timescale of 0.1 @xmath3 produces results more in agreement with observations , while longer timescales tend to predict a peak at too high metallicities .
finally , in figure [ mdf_imf ] we show the effect of varying the imf on the predicted mdf , when the infall timescale is fixed at 0.7 @xmath3 . in this case
, we note that the salpeter imf produces the best agreement and predicts the peak of the mdf at the right position .
however , given the small number of observed stars , we are not able to disentangle the degeneracy between imf and infall timescale .
what seems more clear is that an imf like kroupa et al .
( 1993 ) , good for the solar vicinity , predicts the worst agreement with data when both the abundance ratios and the mdf in the gc are taken into account .
previous models for the entire bulge ( brocato & matteucci , 1990 ; ballero et al . 2007 ; cescutti & matteucci , 2011 ; kobayashi et al .
2011 ) have supported an imf which contains more massive stars than the one for the solar vicinity . in summary
, the comparison between data and model results suggests that the abundances in the gc are quite compatible with a fast formation of this region similarly to what happened to the entire bulge . the fact that the [ @xmath0/fe ] ratios are low at high [ fe / h ] is not necessarily a sign of prolonged star formation .
such a sign instead would be revealed by low [ @xmath0/fe ] ratios at low [ fe / h ] , as happens in dwarf irregular galaxies and dwarf spheroidals where the sfr was quite mild .
a regime of low star formation , in fact , produces a situation where the gas is still poor in fe when type ia sne start producing the bulk of this element . at this point ,
the [ @xmath0/fe ] ratios start to decrease but the [ fe / h ] is still low .
the fact that the star formation was quite strong in the early phases of formation of the central region is also evident in the mdf which is peaked at high metallicity .
however , we should keep in mind that the number of stars observed in the gc is not yet statistically significant and therefore firm conclusions can not be drawn .
we recall that in this section we have ignored the fact that recent star formation has been observed in the gc region , so in the next section we will explore the case in which a second burst has recently occurred and whether this recent sfr can affect our predictions . in our standard model for the gc ,
the bulk of the stars formed 10 @xmath3 ago and then the sfr decreased steadily until the present time to a value slightly lower than @xmath58 .
therefore , to reproduce the observations indicating active star formation at the present time , we have simulated a star formation history made up of a first strong burst , where most of stars formed ( our fiducial model ) , plus a more recent starburst started 500 myr ago .
we have tested different cases and verified the effects of this second starburst on the [ @xmath0/fe ] ratios and the mdf . in order to simulate the second burst
, we simply assumed a second gas infall episode which increased suddenly the available mass of gas : @xmath59 where @xmath60 represents the cosmic time at which the second burst starts and @xmath61 is the infall timescale of the second burst ( see table 1 )
. the origin of this gas can be either extragalactic or coming from the inner regions of the galactic disk thanks to the bar in the bulge . [ cols="^,^,^,^,^,^",options="header " , ] we tested several cases but we show only the most realistic ones .
in particular , we assumed : i ) a burst triggered by a strong infall of gas , comparable to the initial infall rate , ii ) a burst fueled by a more modest infall of gas relative to the previous one and iii ) a starburst due only to a sudden increase of the star formation efficiency from 25 to 200 @xmath20 without an extra - infall of gas .
this case could simulate the effect of a nuclear bar , which simply compresses the existing gas . in the infall cases we tested both primordial and metal enriched infall .
we found that the best model is model ii ) with a modest infall of new gas , which reproduces the estimated total mass of stars formed in the last 500 myr ( @xmath62 ) as well as the observed present time sfr ( @xmath63 ) .
in fact , in case i ) the strong infall produces changes in the [ @xmath0/fe ] ratios which are not seen in the data : in particular , the infall dilutes the abundances of heavy elements thus producing an inversion in the [ @xmath0/fe ] vs. [ fe / h ] plot showing very low [ @xmath0/fe ] ratios at low [ fe / h ] , not observed in the data .
on the other hand , simply increasing the star formation efficiency ( model iii ) ) does not allow us to reproduce the observed present time sfr nor the observed gas in the gc .
this is because the gas was already low 500 myr ago when the starburst was supposed to start .
therefore , we suggest that this burst in the gc must have been triggered by accretion of new gas . the characteristics of the models i ) , ii ) and iii ) are reported in table 1 , where we show the timescale for the main infall episode ( @xmath8 ) , the star formation efficiency for this episode ( @xmath64 ) , the timescale for the infall giving rise to the second burst ( @xmath61 ) and the star formation efficiency in this second burst ( @xmath65 ) .
finally , in column 6 the values of @xmath60 ( the time at which the second infall starts ) is reported .
the best model appears to be model ii ) : it predicts a total mass of stars formed of @xmath66 and a present time sfr of @xmath67 , in good agreement with the observed values . in figure [ sfr_obs ]
we show the predicted sfr history for model ii ) and we report also the observations . in figure [ alpha_2infall ] we show the [ @xmath0/fe ] ratios resulting from model i ) and ii ) which represent the case with a burst triggered by a strong gas infall and a modest gas infall , respectively . in the same figure
is also shown an intermediate case showing that the inversion in the abundances , due to the dilution by infall , is independent of the onset time of the second strong burst .
as one can see , a starburst triggered by a modest episode of primordial gas infall ( at an average rate of @xmath68 ) , occurring in the last hundreds million years and reproducing the present time sfr , produces negligible effects on the predicted [ @xmath0/fe ] ratios . on the other hand ,
as already mentioned , a second starburst triggered by a strong infall of gas ( with an initial rate of @xmath69 ) would produce a noticeable but not observed effect in the data , even if the number of stars formed is the same .
this case is clearly unrealistic since there is no physical justification for a strong late infall of primordial gas , but we show the resulting [ @xmath0/fe ] ratios for the sake of comparison with case ii ) .
moreover , it is clear from figure [ alpha_2infall ] that none of the models with a late strong infall fits the data .
we note that in all these cases we have chosen the model parameters to obtain acceptable values of the present time sfr as well as the total mass of stars formed . to do that we varied the star formation efficiencies and the infall timescale .
the reason for the inversion in the [ @xmath0/fe ] ratios is therefore only due to the gas infall and to the way in which chemical elements are produced and on which timescales . in particular ,
when a starburst is overimposed on a very low star formation regime , triggered by a huge episode of gas accretion , the abundances present at the moment in the ism tend to be diluted by the infall , especially if the infalling gas is primordial , but at the same time they increase because of the new star formation .
therefore , according to the predominance of one or the other process , the abundances tend to increase or decrease . in our cases
the abundances of fe and the @xmath0-elements initially tend to decrease because of the infall , thus producing a reverse trend at constant [ @xmath0/fe ] , then as a result of the new production of @xmath0-elements by type ii sne , the [ @xmath0/fe ] ratios increase strongly .
this is also due to the fact that fe is not increasing immediately at the beginning of the second burst for the simple reason that fe is mainly produced on long timescales by type ia sne .
no stars seem to be observed with the [ @xmath0/fe ] and [ fe / h ] values predicted by the models with a late second starburst triggered by a huge gas infall . only a model with a second very early
starburst would not affect the [ @xmath0/fe ] ratios , as expected .
this fact suggests that a strong burst like the one we have hypothesized in model i ) can not have occurred , otherwise it would have left a clear signature in the abundance pattern , and we have verified that this is true also for enriched infall . on the other hand ,
a minor burst triggered by either primordial or pre - enriched gas ( enriched in metals ) would produce negligible effects on the stellar abundances , as we have shown in figure 8 . concerning the mdf we tested model ii ) and
no differences were produced by the second infall .
in this paper we have studied the chemical evolution of the central region of the galactic bulge ( 200 pc radius ) and compared our results with the recent data from ryde & schultheis ( 2015 ) concerning @xmath0-elements and fe and the present time star formation rate ( morris & serabyn 1996 , yusef - zadeh et al .
2009 , molinari et al . 2011 and
immer et al . 2012 ) .
we have tested different imfs , efficiencies of star formation and infall timescales .
we also assumed a recent and ongoing burst of star formation where @xmath70 of new stars formed .
our main conclusions can be summarized as follows : * a comparison between data and models suggests that the [ @xmath0(mg , si , ca)/fe ] ratios in the gc can be reproduced by a model assuming a strong star formation at the beginning such as that assumed for the whole bulge by previous models ( matteucci & brocato , 1990 ; ballero et al .
2007 ; cescutti & matteucci 2011 ; grieco et al . 2012 ) . under this assumption ,
most of the stars in the gc have formed at the beginning ( more than 10 @xmath3 ago ) and the sfr has decreased steadily until the present time . * for [ mg / fe ] and [ si / fe ] we can conclude that a classical salpeter imf can reproduce the data , whereas for ca the data always lie above the predictions and this is true also for the other assumed imfs . we conclude that for ca there could be a problem with stellar yields .
in fact , stellar yields of some elements are still uncertain : here , we have adopted the set of stellar yields , for stars of all masses , considered the best to reproduce the abundance patterns in the solar vicinity ( romano et al .
2010 , where also a discussion on the uncertainties on stellar yields can be found ) .
* we have varied the efficiency of star formation from 20 @xmath20 to 200 @xmath20 and found practically no difference in the results concerning the [ @xmath0/fe ] vs.[fe / h ] relations , whereas the mdf allows us to exclude the efficiency of 200 @xmath20 .
therefore , we conclude that the efficiency of star formation in the gc is similar to that inferred for the whole bulge ( gmpc12 ) , namely 20 - 25 @xmath20 .
* we have tested different imfs : salpeter , chabrier , ballero and kroupa93 and found that salpeter , ballero and chabrier imfs are probably the best to describe the gc , in agreement with previous results for the whole bulge suggesting an imf with more massive stars than in the solar vicinity ( matteucci & brocato , 1990 ; ballero et al .
2007 ; cescutti & matteucci , 2011 ; grieco et al . 2012 ) ; in fact , they produce the best fits of the [ @xmath0/fe ] vs. [ fe / h ] relations as well as the mdf , at the same time .
however , the salpeter imf requires an infall timescale longer ( 0.7 @xmath3 ) than the other two imfs ( 0.1 @xmath3 ) to fit the data .
given the uncertainties still existing in the data we are not able at the moment to solve this degeneracy and therefore to draw firm conclusions on this point , but we can say that the infall timescale for the formation of the bulk of gc stars was short . *
we tested also the effect of changing the infall timescale and varied this timescale from 0.01 @xmath3 ( practically a closed - box model ) to 1.25 @xmath3 .
again , with the exception of ca , for which all the model predictions lie below the data , the mg and si data are compatible with timescales in the range 0.1 - 1.25 @xmath3 , whereas the mdf is compatible with a range 0.1 - 07 @xmath3 . * finally , we tested the effects of a second very recent burst occurring in the last five hundreds million years and having a sfr and a total amount of formed stars comparable to those observed .
we have found that such a burst , if triggered by a new modest infall episode of gas of primordial composition , would not produce any noticeable effect in the [ @xmath0/fe ] relations and the mdf . on the other hand
, we tested also a case of a starburst triggered by a huge gas infall , still producing the same present time sfr and total mass of new stars , and found that in this case the [ @xmath0/fe ] are strongly affected by the infall .
in particular , we predict very low values of [ @xmath0/fe ] ratios at low [ fe / h ] , which are not observed . *
therefore , we conclude that the stars in the gc formed very quickly with a star formation efficiency of @xmath71 , and that the imf should contain a larger number of massive stars than the typical imf for the solar vicinity , such as the kroupa et al .
in particular , the recent imf of chabrier ( 2003 ) seems to be preferred .
the present time observed sfr in the gc can be explained by a second burst of star formation with the same efficiency as the previous one and the same imf , triggered by a modest infall episode .
this second burst would not produce evident effects on the abundance patterns nor on the mdf , and is compatible with gas accretion induced by the main bar in the bulge .
we acknowledge financial support from prin miur 2010 - 2011 , project the chemical and dynamical evolution of the milky way and local group galaxies , prot .
2010ly5n2 t .
nr acknowledges support from the swedish research council , vr ( project number 621 - 2014 - 5640 ) , funds from kungl .
fysiografiska sllskapet i lund .
( stiftelsen walter gyllenbergs fond and mrta och erik holmbergs donation ) .
we thank i.j .
danziger for reading the manuscript .
finally , we thank an anonymous referee for his / her important suggestions which improved the paper .
babusiaux , c. , gmez , a. , hill , v. , et al . , 2010 , a&a , 519 , a77 ballero s. k. , matteucci f. , origlia l. & rich r. m. , 2007 , a&a , 467 , 123 bensby t. , adn d. , melndez j. et al . , 2011 , a&a , 533 , a134 bensby t. , yee j. c. , feltzing s. et al . , 2013 ,
a&a , 549 , 147 bilitewski t. , & schnrich r. , 2012 , mnras , 426 , 2266 chabrier g. , 2003 , pasp , 115 , 763 chiappini c. et al . , 1997 ,
apj , 477 , 765 cescutti g. & matteucci f. , 2011 , a&a , 525 , 126 crocker r. m. , 2011 , arxiv:1103.4523 cunha k. , sellgren k. , smith v. v. et al . , 2007 , apj , 669 , 1011 genzel r. , tacconi l. j. , eisenhauer f. et al .
, 2006 , nature , 442 , 786 gerhard o. & martinez - valpuesta i. , 2012 , apjl , 744 , 8 gonzalez o.a . , rejkuba m. , zoccali m. , hill v. et al .
2011 , a&a , 530 , 54 greggio l. & renzini a. , 1983 , a&a , 118 , 217 grieco v. , matteucci f. , pipino a. & cescutti g. , 2012 , a&a , 548 , 60 gustafsson b. , edvardsson b. , eriksson k. et al . , 2008 , a&a , 486 , 951 karakas a. & lattanzio j. c. , 2007 , pasa , 24 , 103 karakas a. i. , 2010 , mnras , 403 , 1413 kufl h .- u . , ballester p. , biereichel p. et al . , 2004 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 5492 ,
society of photo - optical instrumentation engineers ( spie ) conference series , ed.a .
f. m. moorwood & m. iye , 12181227 kobayashi c. , umeda h. , nomoto k. , tominaga n. & ohkubo t. , 2006 , apj , 653 , 1145 kobayashi c. , karakas a. i. & umeda h. , 2011 , mnras , 414 , 3231 kroupa p. , tout c. a. & gilmore g. , 1993 , mnras , 262 , 545 kroupa p. , 2001 , mnras , 322 , 231 kruijssen j. m. d. , longmore s. n. , elmegreen b. g. et al . , 2014 , mnras , 440 , 3370 martin c.l . ,
walsh w.m .
, xiao k. 2004 , apjs 150 , 239 matteucci f. , brocato e. , 1990 , apj , 365 , 539 matteucci f. , 2001 , astrophysics and space science library , 253 matteucci f. & recchi s. , 2001 , apj , 558 , 351 matteucci f. , panagia n. , pipino a. et al . , 2006 , mnras , 372 , 265 matteucci f. , spitoni e. , recchi s. & valiante r. , 2009 , a&a , 501 , 531 matteucci f. , 2012 , chemical evolution of galaxies : , astronomy and astrophysics library .
isbn 978 - 3 - 642 - 22490 - 4 .
springer - verlag berlin heidelberg , 2012 meynet g. & maeder a. , 2002a , a&a , 390 , 561 minniti d. , lucas p. w. , emerson j. p. et al .
, 2010 , new a , 15 , 433 molinari s. , bally j. , noriega - crespo a. et al . , 2011 , apj , 735 , 33 moorwood a. , cuby j .-
, biereichel p. et al . , 1998 , the messenger , 94 , 7 morris m. & serabyn e. , 1996 , ara&a , 34 , 645 | in recent years , the galactic center ( gc ) region ( 200 pc in radius ) has been studied in detail with spectroscopic stellar data as well as an estimate of the ongoing star formation rate .
the aims of this paper are to study the chemical evolution of the gc region by means of a detailed chemical evolution model and to compare the results with high resolution spectroscopic data in order to impose constraints on the gc formation history.the chemical evolution model assumes that the gc region formed by fast infall of gas and then follows the evolution of @xmath0-elements and fe .
we test different initial mass functions ( imfs ) , efficiencies of star formation and gas infall timescales . to reproduce the currently observed star formation rate , we assume a late episode of star formation triggered by gas infall / accretion .
we find that , in order to reproduce the [ @xmath0/fe ] ratios as well as the metallicity distribution function observed in gc stars , the gc region should have experienced a main early strong burst of star formation , with a star formation efficiency as high as @xmath1 , occurring on a timescale in the range @xmath2 0.1 - 0.7 @xmath3 , in agreement with previous models of the entire bulge .
although the small amount of data prevents us from drawing firm conclusions , we suggest that the best imf should contain more massive stars than expected in the solar vicinity , and the last episode of star formation , which lasted several hundred million years , should have been triggered by a modest episode of gas infall / accretion , with a star formation efficiency similar to that of the previous main star formation episode .
this last episode of star formation produces negligible effects on the abundance patterns and can be due to accretion of gas induced by the bar .
our results exclude an important infall event as a trigger for the last starburst .
-.6 in [ firstpage ] galaxy : abundances galaxy : evolution |
graphene,@xcite a two - dimensional single layer of carbon atoms packed into a honeycomb lattice , has aroused enormous interest in recent years due to its intriguing properties.@xcite particular efforts have been devoted into the graphene nanoribbons ( gnrs),@xcite quasi one dimension ribbon like structures resulted from the finite termination of the graphene with two different possible edge geometries , namely armchairs and zigzags . since gnrs have a tunable band gap sensitive to the size and geometry , they are good candidates for the possible electronic and spintronic devices , such as the field effect transistor,@xcite the spin filter and spin transistor.@xcite to realize these devices , it is required to manipulate the transport through the gnrs in a controllable manner . among various schemes , to modulate the transport using different shapes of gnrs , e.g. , u - shaped , s - shaped , and t - shaped are of particular interest.@xcite in the t - shaped gnrs , the possible ( quasi-)bound states trapped at the intersection due to the t - shaped confinement@xcite provides another channel to modulate the transport through the fano effect,@xcite which stems from the interference between the path through a discrete state and the path through a continuum when the discrete state embeds in the continuum .
the localized states in t - shaped structures composed of the conventional semiconductor quantum wires have long been studied.@xcite it is shown that there is one bound state lying below and one quasi - bound state embedding in the lowest conduction band .
the existence of the quasi - bound state results in the asymmetrical dip in the energy dependence of the conductance , which is the character of the fano effect.@xcite since the gnrs have much more complex band structures which strongly depend on the geometry and ribbon size,@xcite the t - shaped structures consists of gnrs are expected to have more complex behaviors .
recently , the quasi - bound states are shown to play an important role in the transport through the zigzag gnr in the t - shaped gnr structure consisting of a zigzag gnr connected to an armchair sidearm.@xcite however , in the t - shaped gnr structure consisting of a metallic armchair gnr connected to a zigzag sidearm , which is more interesting for transport since the lowest excitations in the gapless armchair gnrs are massless dirac fermions whose spectrum is linear with the momentum , the existence of the ( quasi)-bound states have not yet been studied . if the ( quasi-)bound states exist , they have to embed in the continuum due to the gapless band structures of both the armchair gnr and the zigzag sidearm .
therefore , these states are expected to greatly affect the transport of the dirac fermions through the armchair gnr through the fano effect . in this work
, we systematically investigate the quasi - bound states and the resulting fano effect in the t - shaped gnr consisting of a metallic armchair gnr connected to a zigzag sidearm .
we list the condition of the existence of such quasi - bound states for a wide range of ribbon size and study the effect of these states on the transport .
and a zigzag gnr lead of width @xmath0 . the insets at
the top - left and top - right cornors show how the sizes along the armchair and zigzag edges are defined . in this scheme , @xmath1 and @xmath2 .
, width=294 ] the t - shaped structure we study is composed of an armchair gnr of width @xmath3 connected to a sidearm of width @xmath0 , as schematically illustrated in the fig . [ figxww1 ] .
we divide the system into four parts , the left , right and top leads as well as the central junction .
the hamiltonian of the system consists of the hamiltonians @xmath4 for each part and their couplings @xmath5 . using the tight - binding model under the nearest - neighbor approximation ,
the hamiltonian are written as @xmath6 here , @xmath7 is the hopping parameter@xcite and @xmath8 , set to 0 , is the on - site energy .
@xmath9 denotes that the sum is restricted to the nearest neighbors . within the landauer - bttiker framework,@xcite
the transmission amplitude of the electrons with energy @xmath10 between leads @xmath11 and @xmath12 is @xmath13,\label{landauer}\ ] ] where @xmath14 and @xmath15 are the retarded / advanced green s function and the tunneling rate matrix , respectively .
the retarded green s function is given by @xmath16^{-1},\label{green}\ ] ] in which , @xmath17 represents the retarded self - energy matrix .
it can be expressed as @xmath18 , with @xmath19 and @xmath20 being the retarded surface green s function of lead @xmath11 and the hopping matrix between the lead @xmath11 and the central junction @xmath21.@xcite similarly , one can write down the advanced green s function . the surface green s function can be calculated through an iterative scheme.@xcite the tunneling rate matrix can be obtained from the self - energy , @xmath22\ ; .\ ] ] the quasi - bound state is studied by diagonalizing the hamiltonian @xmath23 numerically .
the quasi - bound states can also be shown through the local density of state ( ldos ) , which can be obtained from the green s function .
and @xmath24 .
@xmath25 is the carbon - carbon bond distance.,width=321 ] we first study the existence conditions of the quasi - bound states located at the intersection of the t - shaped gnrs .
the states we are interested in are the ones whose eigen - energies are lower than the edges of the second sub - band of the leads .
they are picked out according to the following rules : ( 1 ) the probability density of the state concentrates at the intersection ; ( 2 ) the state is insensitive to the boundaries , that is , the state undergoes small changes when the lengths of the central junction ( @xmath26 or @xmath27 in fig . [ figxww1 ] ) change .
we find that the quasi - bound states exist in the structures only when the width ratio between the armchair gnr and sidearm is in certain regime .
the conditions under which the quasi - bound states exist are listed in table [ table1 ] for a wide range of the armchair gnr width @xmath3 , together with their eigenenergies . to show that these states are indeed trapped at the central junction
, we plot the probability density for the quasi - bound state in fig .
[ figxww2 ] for a typical t - shaped gnr structure with @xmath28 , @xmath24 .
one sees from the figure that for this state the electron indeed concentrates at the intersection .
moreover , the wave - function decays exponentially along the sidearm , but its amplitude along the two armchair gnrs is finite although small which implies that the state is a quasi - bound state
. the localized components of the quasi - bound states , whose energy is below the second sub - band edges , mainly come from the second sub - band of the gnrs.@xcite therefore , the probable condition for the quasi - bound state to exist is when the second sub - band edges of both the armchair gnr and the zigzag sidearm are comparable . in our setup the optimal condition for the quasi - bound states which is when @xmath0 is about @xmath29 , as it can be seen from table [ table1 ] .
lllll @xmath3&@xmath0&&@xmath3&@xmath0 + + 11 & 5 ( 1079 ) & & 23 & 10 ( 561 ) , 11 ( 527 ) + + 14 & 6 ( 898 ) & & 26 & 11 ( 500 ) , 12 ( 480 ) , 13 ( 457 ) + + 17 & 8 ( 712 ) & & 29 & 12 ( 452 ) , 13 ( 438 ) , 14 ( 418 ) + + 20 & 9 ( 626 ) , 10 ( 578 ) & & 32 & 13 ( 420 ) , 14 ( 406 ) , 15 ( 379 ) , + & & & & 16 ( 366 ) + + and different @xmath0 : @xmath30 ( blue chained curve ) , @xmath31 ( purple long dashed curve ) , @xmath24 ( red solid curve ) , @xmath32 ( green short dashed curve ) ; ( b ) and ( c ) : ldos at the transmission dips of @xmath33 for @xmath24 and @xmath34 , respectively.,title="fig:",width=321 ] and different @xmath0 : @xmath30 ( blue chained curve ) , @xmath31 ( purple long dashed curve ) , @xmath24 ( red solid curve ) , @xmath32 ( green short dashed curve ) ; ( b ) and ( c ) : ldos at the transmission dips of @xmath33 for @xmath24 and @xmath34 , respectively.,title="fig:",width=321 ] we now investigate the transport properties of t - shaped gnr structure .
we concentrate on the energy range near the dirac point . in fig .
[ figxww3](a ) we plot the transmission amplitudes between the left and right leads @xmath33 ( upper panel ) and between the left lead and the sidearm @xmath35 ( lower panel ) as the functions of the electron energy for t - shaped gnr structure with the same width of the armchair gnr ( @xmath28 ) but different sidearm widths . for @xmath28
, we have shown that the quasi - bound states exist only when @xmath31 and @xmath36 in the above section .
the quasi - bound states have strong effect on the transmission between the left and right leads .
one finds the a dip close to zero with an asymmetric line shape at @xmath37 mev for @xmath31 and a similar dip at @xmath38 mev for @xmath24 .
the position of the dips are close to the corresponding eigen - energies of the quasi - bound states obtained from the diagonalization .
the sharp dip near the eigen - energy of the quasi - bound state with an asymmetric line shape is the character of the fano effect .
therefore one can see that the fano effect strongly affects in the transmission between left and right leads .
the sharp dip disappears when there is no quasi - bound state , as in the case of @xmath30 and @xmath39 . in the energy regime shown in the figure , there are only shallow dips for @xmath30 and @xmath39 at @xmath40 mev and @xmath41 mev , which are the second sub - band edge of the horizontal gnr for @xmath42 and that of the sidearm gnr for @xmath32 , respectively . the shallow dips at the sub - band edges are strongly related to the opening of new conduction channels , which can be clearly seen from the boost in @xmath35 accompanying these dips in @xmath33 as shown in the lower panel of fig .
[ figxww3](a ) . from the lower panel
, one can also see that the quasi - bound states do not have any impact on the transmission between the left lead and the sidearm @xmath35 , since the quasi - bound states are localized along the sidearm .
additionally , the localization of the quasi - bound states is also identified by the local density of state in fig .
[ figxww3](b ) and ( c ) . , @xmath24 and different sidearm length : @xmath43 ( purple long dashed curve ) , @xmath44 ( green short dashed curve ) , @xmath45 ( blue chained curve ) .
we also plot @xmath33 for the t - shaped gnr structure of same @xmath3 and @xmath0 ( red solid curve ) for comparison .
( b ) and ( c ) are the ldos at @xmath46 mev and @xmath47 mev for t - stub structure with sidearm length @xmath48 , respectively .
, title="fig:",width=321 ] , @xmath24 and different sidearm length : @xmath43 ( purple long dashed curve ) , @xmath44 ( green short dashed curve ) , @xmath45 ( blue chained curve ) .
we also plot @xmath33 for the t - shaped gnr structure of same @xmath3 and @xmath0 ( red solid curve ) for comparison .
( b ) and ( c ) are the ldos at @xmath46 mev and @xmath47 mev for t - stub structure with sidearm length @xmath48 , respectively .
, title="fig:",width=321 ] it should be further noted that the sharp dips caused by the fano effect when the quasi - bound states exist should be distinguished from the ones due to the structure anti - resonance in the t - stub gnrs , i.e. , without the lead connected to the sidearm.@xcite in figure [ figxww4](a ) the transmission amplitudes @xmath33 for different sidearm lengths are plotted as functions of the energy for the t - stub gnr structure with @xmath28 and @xmath24 .
one sees that apart from the sharp dip at @xmath49 mev caused by the quasi - bound state trapped at the central junction for all sidearm lengths , there are dips caused by the structure anti - resonance due to the resonance states in the t - stub.@xcite unlike the dip due to the quasi - bound state , the structure anti - resonance dips are sensitive to the sidearm length . to distinguish the quasi - bound state and the anti - resonance states
, we also plot the ldos corresponding to the fano dip and a structure anti - resonance dip when @xmath48 in figs .
[ figxww4](b ) and [ figxww4](c ) , respectively .
it can be clearly seen that the quasi - bound state mainly concentrates at the intersection and while the anti - resonant state spreads over all of the entire sidearm .
in summary , we have studied the quasi - bound states and the transport properties in the t - shaped gnr composed of a metallic armchair gnr and a zigzag sidearm .
we systematically study the existence of the quasi - bound state in this structure for a wide range of the system size .
we find that there are quasi - bound states when the width of the armchair gnr is about twice of sidearm width .
the quasi - bound states are trapped at the intersection of the t - shaped junction and are truly localized along the sidearm direction .
the quasi - bound states have strong effects on the transport between the two armchair leads but have no effect on the transport between the armchair lead and the sidearm . due to these quasi - bound states
, the fano effect manifests through a characteristic dip with an asymmetrical line sharp in the energy dependence of transmission between the armchair leads .
we would like to thank m.w.wu for proposing the topic as well as directions during the investigation .
this work was supported by the national basic research program of china under grant no .
2012cb922002 and the strategic priority research program of the chinese academy of sciences under grant no . | we study the quasi - bound state and the transport properties in the t - shaped graphene nanoribbon consisting of a metallic armchair - edge ribbon connecting to a zigzag - edge sidearm .
we systematically study the condition under which there are quasi - bound states in the system for a wide range of the system size .
it is found that when the width of the sidearm is about half of the width of the armchair leads , there is a quasi - bound state trapped at the intersection of the t - shape structure .
the quasi - bound states are truly localized in the sidearm but have small continuum components in the armchair leads .
the quasi - bound states have strong effect on the transport between the armchair leads through the fano effect , but do not affect the transport between the armchair lead and the sidearm . |
the volume conjecture was proposed by kashaev and reformulated and refined by murakami and murakami as follows . for any knot @xmath0 , @xmath1 where @xmath2 is the ( normalized ) colored jones polynomial of @xmath0
, @xmath3 is the simplicial volume of the complement of @xmath0 and @xmath4 is the volume of the ideal regular tetrahedron .
recall that @xmath5 is nothing but the sum of the hyperbolic volumes of hyperbolic pieces in the jsj - decomposition of the complement of @xmath0 . in kashaev s original form ,
the knot @xmath0 is hyperbolic and the equation is in terms of the quantum dilogarithm invariant and the hyperbolic volume of the complement of @xmath0 .
the conjecture is marvellous in the sense that it reveals the topological meaning of the quantum invariants of knots which is quite unobvious from definition .
however , it also turns out to be rather hard to be proved .
till now , besides positive numerical evidences ( ref .
@xcite ) for some hyperbolic knots , only the cases of torus knots ( kashaev - tirkkonen @xcite ) and the simplest hyperbolic knot , the figure 8 knot ( ref .
@xcite ) have been verified . in view of the compatible behavior of both sides of the conjectured equation ( [ eqn : conj ] ) under connect sum @xmath6 the volume conjecture , in fact ,
may be reduced to the consideration of prime knots . by thurston s hyperbolization theorem ( ref .
@xcite ) , the prime knots further fall into three families : torus knots , hyperbolic knots and satellite knots . in this article
, we deal with the conjecture by examining a special case of the third family , the whitehead doubles of torus knots .
the approach is emphasized on the relation between the colored jones polynomial of a satellite knot and those of the associated companion knot and pattern link .
in particular , we show a technique to calculate the colored jones polynomial of satellite knots by cutting and gluing method .
a whitehead double of a knot @xmath0 is a knot obtained as follows .
remove the regular neighborhood of one component of the whitehead link from @xmath7 thus get a knot inside a torus , then knot the torus in the shape of a knot @xmath0 .
note that , when @xmath0 is nontrivial , a whitehead double @xmath8 of @xmath0 is a satellite knot whose complement contains an obvious essential torus @xmath9 .
cutting along the torus , we get @xmath10 thus @xmath11 in particular , if @xmath0 is a nontrivial torus knot , the complement of @xmath0 is seifert fibred and the complement of the whitehead link is hyperbolic , hence @xmath12 the article proceeds as follows .
first , we calculate the colored jones polynomials of the twisted whitehead links and the whitehead doubles of knots in section [ sec : tangle ] .
next , as a warming - up we prove in the next two consecutive sections the following two theorems , of which the former one is , in fact , the volume conjecture for twisted whitehead links and both extends the estimation ( [ eqn : conj ] ) to the second order .
[ thm : wl ] for every twisted whitehead link @xmath13 , we have @xmath14 as @xmath15 .
[ thm : torus ] for every nontrivial torus knot @xmath16 with @xmath17 , we have @xmath18 as @xmath15
. then we prove the main theorem in section [ sec : wd ] and show some observations in the final section .
[ thm : wd ] if @xmath0 is a whitehead double of a nontrivial torus knot @xmath16 with @xmath17 , then @xmath19 as @xmath15
. in particular , the volume conjecture is true for @xmath0 . in their proof of the volume conjecture for torus knots ,
kashaev and tirkkonen @xcite derived the following estimation @xmath20 but improving the estimation to ( [ eqn : torus ] ) requires a nonvanishing proposition on number theory ( see proposition [ prop : novanish ] ) to which both theorem [ thm : torus ] and theorem [ thm : wd ] are reduced in this article . with a technical condition @xmath17 we proved the nonvanishing proposition in section [ sec : torus ] .
a complete proof has been beyond the scope of the article .
we only mention here that our technique can be sharpened to prove the nonvanishing proposition , hence both theorems , at least for the cases that both @xmath21 are odd or one of them is a power of @xmath22 .
it is noteworthy that the coefficient @xmath23 " of the second term in the asymptotic expansion ( [ eqn : main ] ) disagrees with the observation due to hikami @xcite @xmath24 for many prime knots @xmath0 .
in this section , we calculate the colored jones polynomials of the twisted whitehead link @xmath25 and the whitehead double @xmath26 of a knot @xmath0 .
[ c]wl(@xmath27 ) [ c]wd(@xmath0,@xmath27 ) [ c]double(@xmath0 ) [ c]@xmath27 twists in the figure , @xmath28 denotes the ( 2,2)-tangle obtained by doubling the knot @xmath0 to a link with zero linking number and then removing a pair of parallel segments .
our trick is cutting the link diagrams into ( 2,2)-tangles and gluing the tangle invariants together .
[ c]twist [ c]belt [ c]double(@xmath0 ) [ c]clasp colored jones polynomial is also defined for tangles , but , instead of a laurent polynomial of @xmath29 , it is in general a module homomorphism of @xmath30 ( choose @xmath31 ) . especially , the colored jones polynomial of a ( 2,2)-tangle is a module homomorphism @xmath32 where @xmath33 is the @xmath34 dimensional irreducible representation of @xmath30 .
note that the tensor product admits the decomposition @xmath35 a straightforward calculation shows that the ( framing independent , unnormalized ) colored jones polynomials of the tangles are @xmath36 where @xmath37 combining the tangle invariants together , one has @xmath38 and @xmath39 note that , in the expression of @xmath40 , the factor @xmath41 is contributed by the companion knot @xmath0 and the other part is precisely obtained from the expression of @xmath42 by removing the factor contributed by the belt tangle .
let @xmath13 denote the twisted whitehead link @xmath25 . setting @xmath43 , we have @xmath44 where @xmath45 and @xmath46 first
, we prepare a lemma to estimate the norm factor @xmath47 .
put @xmath48 and let @xmath49 be the lobachevsky function .
[ lem : sest ] for @xmath50 we have uniform estimations @xmath51 on @xmath52 , @xmath53 on @xmath54 and @xmath55 on @xmath56 .
we have @xmath57 since @xmath58 uniformly on @xmath59 $ ] as @xmath60 , the first estimation follows as @xmath61 note that @xmath62 thus @xmath63 uniformly on @xmath64 \setminus \{0,u\}$ ] as @xmath65 .
it follows that @xmath66 uniformly on @xmath54 .
thanks to sterling series @xmath67 the second estimation holds . to see the third estimation ,
one notices that @xmath68 and @xmath69 in particular , we have @xmath70 and , by the second estimation , @xmath71 } + s_{[{\frac{n}2 } ] } = - \log n + o(1).\ ] ] therefore , @xmath72 uniformly on @xmath56 . from the second and the third estimations of above lemma , we have @xmath73 uniformly on @xmath74 , where @xmath75 the function @xmath76 has a unique critical point @xmath77 in the region @xmath78 , at which @xmath79 reaches maximum @xmath80 and expands as @xmath81 notice that the phase factor @xmath82 is also steady near @xmath83 . in
what follows , the summation ( [ eqn : wl ] ) is expected to be dominated by the summands whose index @xmath84 is near @xmath85 . indeed , this is the case as demonstrated by the next pair of lemmas .
[ lem : main1 ] for any @xmath86 there exist @xmath87 and @xmath88 such that @xmath89,[{\frac{n}4}]}\ ] ] for @xmath90 since @xmath79 has a unique critical point @xmath77 in the region @xmath78 , we have @xmath91 for @xmath92 . by ( [ eqn : fexp ] ) , there exist @xmath87 and @xmath93 such that @xmath94 therefore , by ( [ eqn : sest ] ) there exists @xmath95 such that @xmath96,[{\frac{n}4 } ] } - \epsilon n^{2\delta-1 } + c''\ ] ] for @xmath90 .
[ lem : main2 ] for any @xmath97 , @xmath98 and @xmath99 there exists a nonzero constant @xmath100 such that @xmath101,[{\frac{n}4 } ] } ( 1+o(n^{3\delta-2})).\ ] ] for simplicity , we use the notation @xmath102 , @xmath103 in the proof . note that @xmath104 by ( [ eqn : sest1 ] ) and ( [ eqn : fexp ] ) we have @xmath105,[{\frac{n}4 } ] } & = { \frac{n}{\pi}}f({\frac{n\pi}{n}},{\frac{i\pi}{n } } ) - { \frac{n}{\pi}}f(\frac\pi2,\frac\pi4 ) + o(n^{\delta-1 } ) \\ & = - { \frac{\pi}{n } } ( n'^2 + 2n'i'+2i'^2 ) + o(n^{3\delta-2 } ) \end{split}\ ] ] uniformly on @xmath106 .
moreover , on the same region we have the uniform estimation @xmath107 therefore , by ( [ eqn : gauss ] ) , @xmath108,[{\frac{n}4 } ] } ( 1+o(n^{3\delta-2 } ) ) .
\end{split}\ ] ] to conclude the lemma it suffices to choose @xmath109 it follows from lemma [ lem : main1 ] and lemma [ lem : main2 ] that , in the same notations as the lemmas , @xmath110,[{\frac{n}4 } ] } o(1 ) , \\ & \sum_{|n-{\frac{n}2}|+|i-{\frac{n}4}| < n^\delta } ( 2n+1 ) a_n^{4r-1 } s_{n , i } = n^2 s_{[{\frac{n}2}],[{\frac{n}4 } ] } e^{o(1 ) } , \end{split}\ ] ] so @xmath111,[{\frac{n}4 } ] } ) + o(1).\ ] ] from ( [ eqn : sest2 ] ) we also have @xmath112,[{\frac{n}4 } ] } = \frac{4n}{\pi } l(\frac{\pi}{4 } ) - \frac12 \log n + o(1).\ ] ] therefore , @xmath113 as @xmath15 . in the last row , we used the fact @xmath114 the colored jones polynomial of the torus knot @xmath16 was calculated in @xcite as @xmath115 we put @xmath116 note that @xmath117 so @xmath118 in @xcite , an estimation of @xmath119 was derived as @xmath120 in view of the periodicity of @xmath121 , to establish the theorem it suffices to show that @xmath122 never vanishes if @xmath17 , or equivalently by ( [ eqn : a ] ) , [ prop : novanish ] let @xmath123 be coprime integers with @xmath17 .
then for every integer @xmath34 , @xmath124 the proof of the proposition is purely arguments on elementary algebraic number theory . in the following , we write @xmath125 for each @xmath126 .
an algebraic number field means a finite extension of @xmath127 contained in @xmath128 . for any finite extension @xmath129 of field ,
one has a @xmath0-linear map @xmath130 , called the _ trace function _ , which values on @xmath131 the trace of the @xmath0-linear transformation @xmath132 given by @xmath133 .
[ lem : a1 ] let @xmath134 be a prime , @xmath135 and @xmath0 be an algebraic number field such that @xmath136
. then we have @xmath137 the field extension @xmath138 has a basis @xmath139 on which the diagonal of the matrix of @xmath140 consists of only @xmath141 if @xmath142 , or @xmath140 otherwise .
[ lem : a2 ] let @xmath134 be a prime and @xmath0 be an algebraic number field such that @xmath143 .
then @xmath144 for @xmath145 if and only if the @xmath146 s are identical .
on one hand , the field extension @xmath147 has a basis @xmath148 .
on the other hand , we have @xmath149 .
therefore , the summation vanishes if and only if the @xmath146 s are identical to @xmath150 .
thanks to the next lemma , we are able to eliminate the gaussian exponential appearing in the expression of @xmath121 . [
lem : a4 ] let @xmath134 be an odd prime , @xmath151 and @xmath0 be an algebraic number field such that @xmath152 .
assume that @xmath153 where @xmath154 is a finite subset of @xmath155 , @xmath145 and @xmath156 .
then we have @xmath157 if @xmath158 , or otherwise , @xmath159 } } c_j = 0.\ ] ] if @xmath158 , taking the trace function of @xmath160 on both sides of the equality assumed , we find from lemma [ lem : a1 ] that @xmath161 otherwise , choose @xmath162 such that @xmath163 . from the assumption we have
@xmath164 taking the trace function of @xmath165 on both sides of the equality , we get @xmath166 since @xmath134 is an odd prime , the congruence equation @xmath167 has no solution for some @xmath168 .
it follows from lemma [ lem : a2 ] that the coefficient of @xmath169 in above summation identically vanishes .
in particular , @xmath170 hence the lemma follows .
assume that @xmath171 for @xmath172 where the @xmath173 s and @xmath174 s are distinct odd primes not dividing and dividing @xmath34 , respectively .
rewrite @xmath171 as @xmath175 and choose @xmath176 so that @xmath177 under the galois action of @xmath178 , the equality becomes @xmath179 put @xmath180 } \cdots \alpha_r^{[(l_r+1)/2 ] } , \quad \beta = \beta_1^{k_1 - 1 } \cdots \beta_s^{k_s-1 } , \quad p ' = \beta_1 \cdots \beta_s.\ ] ] it follows from lemma [ lem : a4 ] that @xmath181 where @xmath182 now we apply the condition @xmath17 .
notice that @xmath183 dropping a nonzero factor , ( [ eqn : tmp1 ] ) becomes @xmath184 choose @xmath185 so that @xmath186 and @xmath187 .
then the left hand side of above equality , up to a sign , is @xmath188 therefore , we must have @xmath189 and @xmath190 . but from the choice of @xmath191
, it follows that @xmath192 .
hence @xmath193 , a contradiction .
let @xmath0 denote the @xmath27-twisted whitehead double of the torus knot @xmath16 .
then @xmath194 where @xmath195 setting @xmath43 , one notices that the denominator @xmath196 vanishes .
therefore , one has to apply lhospital s rule , i.e. take derivative of both the denominator and the numerator .
it follows that @xmath197 where @xmath198 and @xmath199 are same as section [ sec : wl ] .
below , we follow the approach used in @xcite to derive an estimation of @xmath200 and @xmath201 in the form of ( [ eqn : torus_estn ] ) . for any complex number @xmath202 with @xmath203 , one has the integral formula @xmath204 where the contour @xmath205 is given by the line @xmath206 and @xmath207 [ lem : int_est ] for @xmath208 we have @xmath209 uniformly on @xmath210 .
put @xmath211 .
we have @xmath212 uniformly on @xmath210 , since the function @xmath213 is bounded on the region @xmath214 counting the residues of the integrand at @xmath215 , @xmath216 , we also have @xmath217 therefore , @xmath218 uniformly on @xmath210 .
[ lem : torus_est ] for @xmath43 we have @xmath219 and @xmath220 uniformly on @xmath210 . from ( [ eqn : torus_int ] ) and lemma [ lem : int_est ] we have @xmath221 it follows from the periodicity of @xmath121 and the identity @xmath222 that @xmath223 , hence by ( [ eqn : a ] ) @xmath224 , identically vanishes .
so , the leading term of the right hand side of ( [ eqn : torus_est1 ] ) is zero .
then , applying lemma [ lem : int_est ] to the derivative of ( [ eqn : torus_int ] ) , one obtains the second estimation .
now we conclude the proof of the theorem .
it is clear that @xmath225 uniformly on @xmath74 . moreover , for @xmath226 , since the function @xmath227 is bounded on @xmath228 $ ] , we have @xmath229 uniformly on @xmath230 .
it follows from lemma [ lem : torus_est ] that @xmath231 and @xmath232 uniformly on @xmath233 . therefore , in the case that @xmath17 , by lemma [ lem : main1 ] , lemma [ lem : main2 ] and proposition [ prop : novanish ] , in the same notations as the lemmas we have @xmath234,[{\frac{n}4 } ] } o(1 ) , \\ & \sum_{|n-{\frac{n}2}|+|i-{\frac{n}4}| < n^\delta } a_n^{4r-1 } s_{n , i } \big ( b_{n , i } \hat{j}_{t(p , q),2n+1 } + t\frac{d}{dt } \hat{j}_{t(p , q),2n+1 } \big ) = n^{7/2 } s_{[{\frac{n}2}],[{\frac{n}4 } ] } e^{o(1 ) } , \end{split}\ ] ] hence @xmath235,[{\frac{n}4 } ] } ) + o(1 ) \\ & = 8 l(\frac{\pi}{4 } )
\cdot n + 4\pi \log n + o(1 ) \\ & = v_3{\| s^3 \setminus k \| } \cdot n
+ 4\pi \log n + o(1 ) \end{split}\ ] ] as @xmath15 .
although the proof of theorem [ thm : wd ] depends on the simple nature of the whitehead doubles of torus knots , the approach still works for the satellite knots to which the colored jones polynomials of the associated companion knot and pattern link satisfy certain mild conditions .
meanwhile , one has to deal with several problems .
for example , as shown in expression ( [ eqn : jwd ] ) , although the volume conjecture itself is only concerned with the value of the colored jones polynomial @xmath2 at the @xmath34-th root of unity , the values at other roots of unity become crucial once the satellite knots of @xmath0 are involved . a more challenging problem is due to the estimations ( [ eqn : torus_est1a ] ) and ( [ eqn : torus_est1b ] ) , which have enabled us to neglect the term @xmath236 in summation ( [ eqn : wd ] ) .
note that the derivative of the polynomial @xmath237 is related to @xmath2 via the identity @xmath238 therefore , the term @xmath239 in summation ( [ eqn : wd ] ) , in fact , plays the role of @xmath119 .
hence it is quite natural to see the term @xmath236 be suppressed . following this observation , when the whitehead doubles of general knots are considered , it is reasonable to expect that a similar suppression also happens . for every nontrivial knot @xmath0 we have @xmath240 uniformly on @xmath241 for some @xmath99 .
note that the conjecture excludes the case of unknot , for which the statement of the conjecture is obviously false .
indeed , the whitehead doubles of unknot are no longer satellite knots but the so called twist knots ( including unknot , trefoil , figure 8 , etc . ) , whose complements always admit a volume strictly smaller than that of whitehead link . in the sequel , the conjecture has an interesting implication : the colored jones polynomial detects unknot .
the author is grateful to xiao - song lin for enlightening discussions .
he also thanks lingquan kong who pointed out to the author the approach to proving the nonvanishing proposition .
999 k. hikami , volume conjecture and asymptotic expansion of @xmath242-series , exper .
( 2003 ) , 319337 .
r. m. kashaev , the hyperbolic volume of knots from quantum dilogarithm , lett .
39 ( 1997 ) , 269275 .
r. m. kashaev , o. tirkkonen , proof of the volume conjecture for torus knots , j. math .
115 ( 2003 ) , 20332036 .
h. r. morton , the coloured jones function and alexander polynomial for torus knots , proc .
( 1995 ) , 129135 .
h. murakami , j. murakami , the colored jones polynomials and the simplicial volume of a knot , acta math .
186 ( 2001 ) , 85104 .
h. murakami , j. murakami , m. okamoto , t. takata , y. yokota , kashaev s conjecture and the chern - simons invariants of knots and links , exper .
( 2002 ) , 427435 w. p.
thurston , three - dimensional manifolds , kleinian groups , and hyperbolic geometry , bull .
6 ( 1982 ) , 357381 . | a technique to calculate the colored jones polynomials of satellite knots , illustrated by the whitehead doubles of knots , is presented .
then we prove the volume conjecture for whitehead doubles of a family of torus knots and show some interesting observations .
keywords : volume conjecture , whitehead double , torus knots , whitehead link mathematics subject classification 2000 : 57m25 , 57n10 |
positive quadratic forms ( referred to as pqfs ) in @xmath7 indeterminate form a _ closed _ cone @xmath8 of dimension @xmath9 in @xmath10 , and this cone is the main object of study in our paper .
the interior of @xmath8 consists of positive definite forms of rank @xmath7 .
we abbreviate positive _ definite _ forms as pdqfs .
pdqfs serve as algebraic representations of _ point lattices_. there is a one - to - one correspondence between isometry classes of @xmath7-lattices and integral equivalence classes ( i.e. with respect to @xmath11-conjugation ) of pdqfs in @xmath7 indeterminates . for basic results of the theory of lattices and pqfs and their applications see ryshkov and baranovskii ( 1978 ) , gruber and lekkerkerker ( 1987 ) , erds , gruber and hammer ( 1989 ) , conway and sloane ( 1999 ) .
@xmath11 acts pointwise on the space of quadratic forms @xmath12 .
a _ polyhedral reduction partition _
@xmath13 of @xmath8 is a partition of this cone into open _
convex polyhedral _ cones such that : 1 .
it is invariant with respect to @xmath11 ; 2 .
there are finitely many inequivalent cones in this partition ; 3 . for each cone @xmath14 of @xmath13 and any pqf @xmath15 in @xmath7 indeterminates , @xmath15 can be @xmath11-equivalent to at most finitely many forms lying in @xmath14 . the _ partition into perfect cones _ and the _ @xmath0-type partition _ ( also referred to as the voronoi partition of the 2nd kind , or the partition into voronoi reduction domains ) are important polyhedral reduction partitions of @xmath8 .
( our usage of term _ domain _ is lax ; it should be clear from the context whether we mean the whole arithmetic class , or just one element of this class . )
these partitions have been intensively studied in geometry of numbers since times of korkin , zolotareff ( 1873 ) and voronoi ( 1908 ) , and more recently in combinatorics ( e.g. deza et al .
( 1997 ) ) , and algebraic geometry ( e.g. alexeev ( 1999a , b ) ) . _ in most previous works _ ( e.g. voronoi ( 1908,1909 ) ,
ryshkov , baranovskii ( 1976 ) , dickson ( 1972 ) ) _ the _ @xmath0__-type partition of _ _ @xmath16 _ _ , or sub - cones of _ _ @xmath16 _ _ , was constructed by refining the perfect partition . _
_ it is not an exaggeration to say that in almost any systematic study , except for engels computational investigations , @xmath0-types were approached via perfect forms .
for example , voronoi started classifying 4-dimensional @xmath0-types by analyzing the delaunay ( @xmath0-)tilings of forms lying in the 1st ( @xmath17 ) and 2nd ( @xmath18 ) perfect domains .
the same route was followed by ryshkov and baranovskii ( 1976 ) .
it was wideley believed that the @xmath0-type parition is the refinement of the perfect parition , i.e. each convex cone of the perfect partition is the union of finitely many conex cones from the @xmath0-type partition .
this conjecture is implicit in voronoi s memoirs ( 1908 - 1909 ) , and explicit in dickson ( 1972 ) , where he showed that the first perfect domain is the only perfect domain which coincided with an @xmath19-type domain .
in our paper we present the results of our study of the relashionship between these two partitions for @xmath20 .
we have proved that for @xmath20 the @xmath0-type partition is not a refinement of the partition into perfect cones , as was believed before .
this note contains only a scketch of the proof of this result , which will be later published in a longer journal paper ( about 35 - 40pp ) .
let @xmath0 be a lattice in @xmath22 .
a convex polyhedron @xmath23 in @xmath24 is called a delaunay cell of @xmath0 with respect to a positive quadratic form @xmath25 if : 1 .
for each face @xmath26 of @xmath23 we have @xmath27 ; 2 .
there is a quadric circumscribed about @xmath23 , called the _ empty ellipsoid _ of @xmath23 when @xmath25 is positive definite , whose quadratic form is @xmath25 ( in case @xmath28 , this quadric is an elliptic cilinder ) ; 3 .
no points of @xmath0 lie inside the quadric circumscribed about @xmath23 . when @xmath29 , our definition coincides with the classical definition of delaunay cell in @xmath30 .
delaunay cells form a convex face - to - face tiling of @xmath0 that is uniquely defined by @xmath0 ( delaunay , 1937 ) .
two delaunay cells are called _ homologous _ is they can be mapped to each other with a composition of a lattice translation and a central inversion with respect to a lattice point .
pqfs @xmath31 and @xmath32 belong to the same convex @xmath0-domain if the delaunay tilings of @xmath33 with respect to @xmath31 and @xmath32 are identical .
@xmath31 and @xmath32 belong to the same @xmath0-type if these tilings are equivalent with respect to @xmath34 .
the folowing proposition establishes the equivalence between the delaunay s definition of @xmath0-equivalence for lattices and the notion of @xmath0-equivalence for arbitrary pqfs , which is introduced above .
positive definite forms @xmath15 and @xmath35 belong to the same @xmath0-type if the corresponding lattices belong to the same @xmath0-type with respect to the form @xmath36 .
( voronoi ) the parition of @xmath8 into @xmath0-types is a reduction partition .
moreover , it is face - to - face .
the notions of delauny tiling and @xmath0-type are important in the study of extremal and group - theoretic properties of lattices .
for example , the analysis of delaunay cells in the leech lattice conducted by conway , sloane ( e.g. see ( 1999 ) ) and borcherds showed that 23 deep holes
( delaunay cells of radius equal to the covering radius of the lattice ) in the leech lattice correspond to 23 even unimodular 24-dimensional lattices ( niemeier s list ) that , in turn , give rise to 23 gluing constructions of the leech lattice from root lattices .
barnes and dickson ( 1967 , 1968 ) and , later , in a geometric form , delaunay et al .
( 1969 , 1970 ) proved the following the _ closure _ of any @xmath37-dimensional convex @xmath0-domain contains at most one local minimum of the sphere covering density .
the group of @xmath38-automorphisms of the domain maps this form to itself . using this approach ,
delaunay , ryshkov and baranovskii ( 1963 , 1976 ) found the best lattice coverings in @xmath39 and @xmath40 the theory of @xmath0-types also has numerous connections to combinatorics and , in particular , to cuts , hypermetrics , and regular graphs ( see deza et al .
( 1997 ) ) .
recently , v. alexeev ( 1999a , b ) found exciting connections between compactifications of moduli spaces of principally polirized abelian varities and @xmath0-types .
the @xmath0-type partition of @xmath8 is closely related to the theory of _ perfect forms _ originated by korkine and zolotareff ( 1873 ) .
let @xmath41 be a pdqf .
arithmetic minimum _ of @xmath42 is the minimum of this form on @xmath33 .
the integral vectors on which this minimum is attained are called the representations of the minimum , or the _ minimal vectors _ of @xmath43 : these vectors have the minimal length among all vectors of @xmath33 when @xmath44 is used as the metrical form .
form @xmath43 is called _ perfect _ if it can be reconstructed up to scale from all representations of its arithmetic minimum . in other words , a form @xmath43 with the arithmetic minimum @xmath45 and
the set of minimal vectors @xmath46 is perfect if the system [ cols="<",options="header " , ] where @xmath47 has a unique solution @xmath48 in @xmath49 ( indeed , uniquness requires at least @xmath50 minimal vectors ) .
pqfs @xmath31 and @xmath32 belong to the same cone of the perfect partition if they both can be written as _
strictly positive _ linear combinations of some subset of minimal vectors of a perfect form @xmath51 . @xmath31 and @xmath32 belong to the same perfect type if there is @xmath52 , equivalent to @xmath31 , such that @xmath52 and @xmath32 belong to the same cone of the perfect partition .
( voronoi ) the partition of @xmath8 into perfect domains is a reduction partition .
moreover , it is face - to - face .
each 1-dimensional cone of this partition lies on @xmath53 .
perfect forms play an important role in lattice sphere packings .
voronoi s theorem ( 1908 ) says that if a form is _ _
, a maximum of the packing density it must also be perfect ( see coxeter ( 1951 ) , conway , sloane ( 1988 ) for the proof ) .
the notion of _ eutactic form _ arises in the study of the dense lattice sphere packings and is directly related to the notion of perfect form .
the reciprocal of @xmath54 is a form whose gramm matrix is the inverse of the gramm matrix of @xmath55 the dual form is normally denoted by @xmath56 a form @xmath57 is called _ eutactic _ if the dual form @xmath58 can be written as @xmath59 where @xmath60 is the set of mutually non - collinear minimal vectors of @xmath61 and @xmath62 .
( voronoi ) a form @xmath54 is a maximum of the sphere packing density if and only if @xmath54 is perfect and eutactic .
voronoi gave an algorithm finding all perfect domains for given @xmath7 .
this algorithm is known as voronoi s reduction with perfect forms . for the computational analysis of his algorithm and
its improvements see martinet ( 1996 ) .
the perfect forms and the incidence graphs of perfect paritions of @xmath63 have have been completely described for @xmath64 voronoi ( 1908 - 09 ) proved that for @xmath65 the @xmath0-partition and the perfect partition of @xmath8 coincide .
the perfect facet @xmath4 ( the 2nd perfect form in 4 variables ) exemplifies a new pattern in the relation of these partitions .
namely , the facet @xmath4 is decomposed into a number of simplicial @xmath0-type domains like a pie : this decomposition consists of the cones with apex at the affine center of this facet over the @xmath66-faces .
these simplexes are @xmath0-type domains of two arithmetic types : type i is adjacent to the the perfect/@xmath0-type domain of @xmath67 type ii is adacent to an arithmetically equivalent @xmath0-type domain ( also type ii , indeed ) from the @xmath0-subdivision of the adjacent @xmath4 domain ( for details see delaunay et al .
( 1963 , 1968 ) ) .
voronoi also proved that for @xmath68 the tiling of @xmath8 with @xmath0-type domains refines the partition of this cone into perfect domains .
ryshkov and baranovskii ( 1975 ) proved the refinement hypothesis for @xmath69 . in his paper of 1972 t.j .
dickson proved that the perfect domain of @xmath70 , also called the _ first perfect form _
after korkine and zolotareff ( 1873 ) , is the only perfect domain that is also an @xmath0-type domain ; he was also first to explicitly mention the common believe in _
voronoi s refinement hypothesis_.
consider the following symetric sets of vectors in @xmath73 .
@xmath74,\pm \lbrack 2;-2;-1^{4}],\pm \lbrack 1;0;-1^{4}]\ } \\
\mathcal{p}_{2}=\{\pm \lbrack 2;-1;-2,-1^{3}]\times 4,\pm \lbrack 1;-1;0,-1^{3}]\times 4,\pm \lbrack 1;0^{2},-1^{3}]\times 10,\pm \lbrack 0;1,0^{4}]\times 5,\pm \lbrack 2;-1^{5}]\ }
\\ \mathcal{p}_{3}=\{\pm \lbrack 0;0;1,-1,0^{2}]\times 6,\pm \lbrack 1;0;-1^{2},0^{2}]\times 6\}\end{gathered}\ ] ] here we use a short - hand notation for families of vectors obtained from some @xmath7-vector by all circular permutations of selected subsets of its components : ( 1 ) @xmath75 stands for @xmath76 consecutive @xmath45 s , ( 2 ) square brackets @xmath77 $ ] are used to denote all vectors that can be obtained from vector @xmath78 by circular permutations in strings of symbols that are separated by commas and bordered on the sides by semicolons and/or brackets , ( 3 ) numbers between semicolumns and/or brackets are not permuted . the two sets of _ perfect vectors _ @xmath79 and @xmath80 are the minimal vectors for the metrical forms @xmath81 and @xmath82 , with _ arithmetic minimum _ @xmath45 and coefficient matrices @xmath83 , \qquad \mathbf{p}_{e_{6}^{\ast } } = \frac{m}{4}\left [ \begin{array}{rrrrrr } 16 & \;5 & \;5 & \;5 & \;5 & \;5 \\ 5 & 4 & 1 & 1 & 1 & 1 \\ 5 & 1 & 4 & 1 & 1 & 1 \\ 5 & 1 & 1 & 4 & 1 & 1 \\ 5 & 1 & 1 & 1 & 4 & 1 \\ 5 & 1 & 1 & 1 & 1 & 4 \end{array } \right ] .\ ] ] the metrical forms @xmath84 and @xmath82 are integrally equivalent to forms @xmath85 and @xmath86 from barnes s table ( 1957 ) . that @xmath84 and @xmath82 are forms for the root lattices @xmath3 and @xmath3 is established by the numbers of perfect vectors , which are given by @xmath87 and @xmath88 .
as shown below , if this scale parameter is set equal to @xmath89 , then , the geometric lattices corresponding to @xmath90 , @xmath84 are dual lattices . [ [ perfect - vectors - and - perfect - domains . ] ] perfect vectors and perfect domains .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath91 be centrally symmetric .
then define the _ domain of _
@xmath92 to be given by @xmath93 as indicated by the superscript @xmath94 , this summation is over an _
oriented subset _ of vectors one vector from each pair of opposites in the symmetric set @xmath92 .
the domains @xmath95 are _ perfect cones _ of types@xmath96 and @xmath97 , respectively ; they are defined on the perfect vectors for the perfect forms @xmath82 and @xmath84 .
@xmath98 is a facet of both@xmath99 and @xmath100 . if forms @xmath101 , @xmath102 have coefficient matrices @xmath103 , @xmath104 , define the scalar product @xmath105
. consider the form @xmath106 and the rank one form @xmath107 .
then , @xmath108 . if @xmath109 , then @xmath110 , with equality if and only if @xmath111 ; if @xmath112 , then @xmath113 , with equality if and only if @xmath114 .
it follows that the hyperplane with equation @xmath115 separates @xmath116 and @xmath117 , and that @xmath118 is a face of both these perfect domains .
that @xmath98 is a facet follows by showing that the linear span of the forms @xmath119 , @xmath120 has co - dimension one in the linear space of metrical forms .
( we omit this argument . )
[ [ eutactic - forms . ] ] eutactic forms . + + + + + + + + + + + + + + + the forms @xmath121 lie on the _ central rays _ of the perfect cones @xmath122 , @xmath123 , and have coefficient matrices given by @xmath124 , \text { } \mathbf{f}_{e_{6}^{\ast } } = \frac{m}{4}\left [ \begin{array}{rrrrrr } 10 & -5 & -6 & -6 & -6 & -6 \\ -5 & 4 & 3 & 3 & 3 & 3 \\ -6 & 3 & 6 & 3 & 3 & 3 \\ -6 & 3 & 3 & 6 & 3 & 3 \\ -6 & 3 & 3 & 3 & 6 & 3 \\ -6 & 3 & 3 & 3 & 3 & 6 \end{array } \right ] .\ ] ] these forms are related to the original @xmath82 , @xmath125 by the formulas @xmath126 , @xmath127 , where @xmath128 is given by @xmath129 .\qquad \mathbf{u}_{\ast } ^{-1}\mathbf{=}\left [ \begin{array}{rrrrrr } 0 & 1 & 1 & -1 & -1 & -1 \\ 1 & 1 & 1 & 0 & 0 & 0 \\ 1 & 1 & 0 & 0 & 0 & 0 \\ -1 & 0 & 0 & 0 & -1 & -1 \\ -1 & 0 & 0 & -1 & 0 & -1 \\
-1 & 0 & 0 & -1 & -1 & 0 \end{array } \right]\ ] ] being arithmetically equivalent , @xmath130 , @xmath82 are alternate metrical forms for the same geometric lattice , and similarly for @xmath131 , @xmath84 .
we will refer to the minimal vectors of these forms as _
short_. the short vectors for @xmath132 and @xmath133 are related to the perfect vectors for @xmath82 and @xmath84 by the formulas : @xmath134 , @xmath135 , where @xmath136 , @xmath137 , @xmath138 .
explicit form for the sets @xmath139 , @xmath140 , @xmath141 can be obtained by a direct calculation .
@xmath142,\pm \lbrack -2,0,-1^{4}],\pm \lbrack 0,1,0^{4}]\ } \\ \mathcal{s}_{2}=\{\pm \lbrack 0,1;-1,0^{3}]\times 4,\pm \lbrack 2,0;0,1^{3}]\times 4,\pm \lbrack 1;1^{2},0^{3}]\times 10,\pm \lbrack 1;1,0^{4}]\times 5,\pm \lbrack 3;1^{5}]\ } \\ \mathcal{s}_{3}=\{\pm \lbrack 0,0;1,-1,0^{2}]\times 6,\pm \lbrack 2,1;1^{2},0^{2}]\times 6\}\end{gathered}\ ] ] the four coefficient matrices satisfy the relations @xmath143 , so that when @xmath144 , @xmath145 , @xmath146 . under these circumstances the pair of forms @xmath132 , @xmath84 , and the pair @xmath147 , @xmath82 , are in duality : @xmath148 , @xmath149 , @xmath150 , @xmath151 .
dual forms correspond to dual lattices , so the single geometric lattice corresponding to the forms @xmath130 , @xmath152 is dual to the geometric lattice corresponding to @xmath147 , @xmath84 .
this pair of geometric lattices is the pair of root lattices @xmath2 and @xmath3 . the above representations for @xmath130 , @xmath133 can equally well be considered as representations for the dual forms @xmath153 , @xmath154 , which shows that @xmath155 , @xmath156 lie interior to the domains @xmath157 , @xmath123 determined , respectively , by @xmath81 , @xmath82 . since @xmath158 , @xmath149 , the original forms have _ eutaxy _
( see section 2 ) _ _ _ _ representations identicle to those for @xmath130 , @xmath133 , @xmath159 the summation again is over an oriented subset of minimal vectors for the reciprocal forms , respectively , the forms @xmath133 , @xmath132 . the symmetrical arrangement provided by the domains ,
@xmath160 , @xmath161 , is remarkable it is a rare occurance that forms on the central rays of adjacent domains correspond to dual geometric lattices . while the lattices are dual , the forms @xmath130 , @xmath147 are not .
these forms fit into the following intricate pattern : @xmath130 is dual to @xmath84 , which in turn is arithmetically equivalent to @xmath133 and , @xmath147 is dual to @xmath82 , which in turn is arithmetically equivalent to @xmath130 .
this pattern is known since the times of korkine and zolotareff ( 1873 ) .
see also coxeter ( 1951 ) .
barnes ( 1957 ) showed that there is only one arithmetic type of wall between @xmath2 and @xmath97 .
the point group for the root lattice @xmath2 is the product of the two element group generated by central inversion and the reflection group @xmath2 , and has order @xmath164 .
there is a representation @xmath165 @xmath166 that is the invariance group of the perfect form @xmath130 , with the following characterization : @xmath167 if and only if @xmath168 .
there is a second representation @xmath169 that is the invariance group of the form @xmath133 . the dual group ( representation ) @xmath170 is the invariance group of the perfect form @xmath84 , and the dual group ( representation ) @xmath171 is the invariance group of the form @xmath82 .
the subgroup @xmath172 is defined by @xmath173 where @xmath174 is the matrix for the dual transformation ; there is a similar definition for the subgroup @xmath175 . from the arithmetic equivalences @xmath176 and
@xmath177 , it follows that @xmath178 and @xmath179 , where @xmath180 was described above .
[ invariance.p]the subgroups @xmath181 are the full invariance groups of the respective sets of vectors @xmath182 , and act transitively on these sets .
the remaining sections will be devoted to the proof of the _ main theorem _ that provides for a counterexample to the refinement hypothesis : ( main)[main]the segment @xmath183 has forms of 5 @xmath0-types : @xmath184 .
the wall between the perfect domains @xmath185 and @xmath100 crosses this segment at point @xmath186 .
a convex polyhedron @xmath23 in @xmath22 is called a @xmath33-polyhedron if any face @xmath26 of @xmath23 is @xmath187 @xmath188 @xmath189 a @xmath33-tiling is a face - to - face tiling of @xmath190 by convex @xmath33-polyhedra . if @xmath191 @xmath192 are two convex tilings of @xmath22 , one can define the intersection tiling of @xmath193 and @xmath194 the ( open ) tiles of this tiling are all non - empty intersections of ( open ) tiles of @xmath195 and @xmath196 the intersection tiling of two @xmath33-tilings is not always a @xmath33-tiling ; it is a @xmath33-tiling if and only if the vertex set of the intersection tiling is @xmath33 .
let @xmath197 and @xmath198 be two @xmath33-polyhedra .
they are called commensurate if all the faces of @xmath199 are @xmath33-polyhedra .
in particular , two @xmath33-_polytopes _ @xmath197 and @xmath198 are commensurate if the vertex set of @xmath199 belongs to @xmath200 let @xmath191 @xmath192 be two @xmath33-tilings .
they are called commensurate if their intersection tiling consist of @xmath33-polyhedra only .
in particular , the vertex set of the intersection tiling of @xmath195 and @xmath201 must be a subset of @xmath33 ( possibly empty ) .
[ test on commensurability of 2 cells]assume that lattice tilings @xmath202 of @xmath33 are delaunay with respect to the pqfs @xmath203 and @xmath204
. also assume that @xmath205 , @xmath206 are closed delaunay cells for @xmath207 and @xmath208 . if @xmath209 , then @xmath210 is a closed delaunay cell for all intermediate forms @xmath211 , where @xmath212 .
[ comtest ] since @xmath205 is delaunay , there is a scalars @xmath213 and vector @xmath214 so that @xmath215 is zero - valued on the vertices of @xmath216 , but has positive values on all other elements of @xmath217 .
therefore , the equation @xmath218 is for an empty ellipsoid circumscribing @xmath205 .
there is a similar function for @xmath206 , so that @xmath219 is the equation of an empty ellipsoid circumscribing @xmath206 . for @xmath220 ,
the function @xmath221 is zero - valued on the vertices of @xmath14 , and positive on all other elements of @xmath217 .
therefore , @xmath14 is circumscribed by the empty ellipsoid with equation @xmath222 , and is delaunay with respect to the form @xmath223 .
[ commense ] if delaunay tilings @xmath224 and @xmath225 of @xmath33 with respect to pqfs @xmath15 and @xmath35 are commensurate , then their intersection @xmath33-tiling is delaunay with respect to @xmath226 for any @xmath227 conversly , if for all @xmath212 the delaunay tiling of @xmath33 with respect to @xmath228 is the same , then delaunay tilings @xmath224 and @xmath225 of @xmath229 are commensurate .
@xmath230 since @xmath224 and @xmath225 are commensurate , any cell @xmath14 of the intersection tiling of @xmath224 and @xmath225 can be written as @xmath231 where @xmath205 , @xmath206 are closed delaunay cells for @xmath207 and @xmath208 . by the above lemma ,
@xmath14 is delaunay for @xmath232 for any @xmath227 @xmath233 if @xmath14 is a delaunay cell with respect to @xmath228 , then , by a standard continuity argument , there are @xmath33-polyhedra @xmath234,@xmath235 which are delaunay relative to @xmath15 and @xmath35 respectively , such that @xmath236 and @xmath237 therefore , @xmath238 thus , all cells of the delaunay tiling defined by @xmath228 are the intersections of the delaunay cells of @xmath224 and @xmath239 .
this implies that @xmath224 and @xmath225 are commensurate .
the following proposition directly follows from lemma [ commense ] .
two forms @xmath15 and @xmath35 belong to the same @xmath0-cone ( not type ! ) if and only if their delaunay tilings @xmath240 and @xmath241 are commensurate .
in this section we consider the delaunay tilings for @xmath73 relative to the forms @xmath130 and @xmath133 .
more specifically , we study the delaunay tilings of forms lying on the segment joining @xmath130 and @xmath133of @xmath242
. these are affinely equivalent to the delaunay tilings for the root lattice @xmath2 and the dual lattice @xmath3 .
two delaunay cells are said to belong to the same homology class if one of them can be obtained from the other by the composition of a lattice translation and inversions .
our discussion is mostly descriptive and adapted specifically for studying the change of @xmath0-type along the segment joining forms @xmath130 and @xmath133 .
more information about these tilings can be found in coxeter ( 1995 ) , and baranovski ( 1991 ) . [
[ tiling - with - g - topes . ] ] tiling with g - topes .
+ + + + + + + + + + + + + + + + + + + + consider the @xmath243 @xmath244__triangles__@xmath245 _ _ _ _
@xmath246,[0;1,0^{4}],[2;0,1^{4}]\}\times 5,\quad \text { } \delta _ { g}^{2}=conv\{[3;1^{5}],[-1;-1,0^{4}],[0;1,0^{4}]\}\times 5 , \\
\delta _ { g}^{3}=conv\{[1;0,1^{2},0^{2}],[1;0^{3},1^{2}],[0;1,0^{4}]\}\times 15,\qquad \\ \text { } \delta _ { g}^{4}=conv\{[1;1^{2},0^{3}],[-1;0,-1,0^{3}],[2;0,1^{4}]\}\times 20\end{gathered}\ ] ] have a common centroid @xmath247 $ ] . in these triples of points
admissible positions of non - zero components can be determined from that the affine centroid is always at @xmath248 $ ] .
the convex hull is a polytope with @xmath249 vertices , the _ gosset polytope _ @xmath250 ( usually denoted by @xmath6 ) , which we call the _ reference g - tope_. there are @xmath251 homologous copies of @xmath250 that fit together facet - to - facet around the origin to form the star at the origin .
this arrangement can be extended to a tiling by @xmath250-topes , which is the delaunay tiling @xmath252 for @xmath73 determined by @xmath132 ( e.g. baranovskii ( 1991 ) ) .
these @xmath243 triangles inscribed in @xmath250 are @xmath244triangles because the edge vectors@xmath245 _ _
@xmath253,\pm \lbrack -2;0,-1^{4}],\pm \lbrack 0;1,0^{4}]\}\times 5 , \\ e_{g}^{2 } & = \{\pm \lbrack -4;-2,-1^{4}],\pm \lbrack 1;2,0^{4}],\pm \lbrack 3;0,1^{4}]\}\times 5 , \\
e_{g}^{3 } & = \{\pm \lbrack 0;0,-1^{2},1^{2}],\pm \lbrack -1;1,0^{2},-1^{2}],\pm \lbrack 1;-1,1^{2},0^{2}]\}\times 15 , \\
e_{g}^{4 } & = \{\pm \lbrack -2;-1,-2,0^{3}],\pm \lbrack 3;0,2,1^{3}],\pm \lbrack -1;1,0,-1^{3}]\}\times 20,\end{aligned}\ ] ] are subsets of @xmath254 .
this is the set of @xmath255 long vectors for @xmath130 , and is defined by @xmath256 @xmath257 is the _ second minimum _ for @xmath130 , which is the minimal value assumed on the non - zero elements of @xmath73 not belonging to @xmath258 .
as is indicated by the notation , these inscribed triangles belong to four @xmath259classes of size @xmath260 , and @xmath261 , where @xmath262 is the subgroup of permutations of the last five coordinates .
each long vector appears only once in the list , so the @xmath243 edge sets account for the @xmath263 long vectors for @xmath130 .
the group @xmath264 has the following useful characterization : @xmath265 @xmath166 _ _ is the stability group of the set of short vectors _ _ @xmath258 .
the following theorem gives information on @xmath264 ( see e.g. coxeter ( 1995 ) or baranovskii ( 1991 ) ) .
( coxeter ) [ t.g - equivalence]the group @xmath264 acts transitively on the @xmath266 elements of @xmath258 , the @xmath255 elements of @xmath254 , and , the @xmath251 g - topes in the star at the origin .
the triangle @xmath267 has a vertex at the origin and is inscribed in @xmath250 .
there are five such @xmath244triangles that belong to a single @xmath262 class and together have ten vertices at distance @xmath257 from the origin .
these vertices , which are given by @xmath268\times 5\},\{\mathbf{b}_{1},% \mathbf{b}_{2}, ... ,\mathbf{b}_{5}\}=\{[2;0,1^{4}]\times 5\}$ ] , are the vertices of a five - dimensional cross - polytope .
the five diagonals intersect at @xmath269 $ ] , and the ten diagonal vectors are given by @xmath270\times 5\},\ ] ] and belong to a single parity class .
if @xmath271 , which is the stability group for @xmath250 , then @xmath272 must fix the center @xmath269 $ ] and therefore belong to the stability group of the cross - polytope .
the group of linear transformations that map the cross - polytope onto itself has order @xmath273 , and can be written as the product @xmath274 ; the subgroup @xmath275 inverts arbitrary numbers of axes , and @xmath262 permutes the red vertices @xmath276 by permuting the last five co - ordinates .
the @xmath277action clearly stabilizes the set @xmath258 , and therefore @xmath278 .
the transformations corresponding to @xmath279 , \mathbf{i}_{12}=\left [ \begin{array}{rrrrrr } -1 & 2 & 2 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 & 0 & 0 \\ -1 & 1 & 1 & 1 & 0 & 0 \\ -1 & 1 & 1 & 0 & 1 & 0 \\ -1 & 1 & 1 & 0 & 0 & 1 \end{array } \right ] , \ ] ] belong to @xmath275 .
the first inverts the first axis of the cross - polytope so that @xmath280=[\mathbf{b}_{1},\mathbf{r}_{2}, ... ,% \mathbf{r}_{5};\mathbf{r}_{1},\mathbf{b}_{2}, ... ,\mathbf{b}_{5}],\ ] ] but is non - integral and therefore not an element of @xmath281 .
the second transformation inverts the first two axes so that @xmath282=[\mathbf{b}_{1},\mathbf{b}_{2},% \mathbf{r}_{3}, ... ,\mathbf{r}_{5};\mathbf{r}_{1},\mathbf{r}_{2},\mathbf{b}% _ { 3}, ... ,\mathbf{b}_{5}].\ ] ] a direct calculation shows this transformation leaves @xmath258 invariant , so @xmath283 @xmath281 . similarly , there are matrices @xmath284 , @xmath285 , @xmath286 that induce inversions of the pairs of axes indicated by the subscripts , and together these four inversions generate the subgroup @xmath287 . an arbitrary element @xmath288 inverts an even number of axes of the cross - polytope .
[ longtriangles1.p]@xmath289 , where the @xmath277action permutes the last four co - ordinates .
@xmath290 .
this follows immediately from the preceeding discussion .
[ longtriangles1.c1]@xmath264 acts transitively on the @xmath251 g - topes in the star of g - topes at the origin .
@xmath291 .
the @xmath292action maps the cross - polytope centered at @xmath293 $ ] to @xmath294 others , making @xmath251 it total .
since cross - polytopes are centrally symmetric , opposite @xmath250-topes have cross - polytopes that are translates . in particular , the cross - polytopes for @xmath250 and @xmath295
. consequently , there are @xmath249 sets of diagonal vectors , each with @xmath296 long vectors belonging to the same parity class .
these sets of are given by : @xmath297\times 5\}\times 1;\qquad \mathcal{l% } _
{ 4}=\{\pm \lbrack 1;2,0^{4}]\times 5\}\times 1 ; \\ \mathcal{l}_{2}=\{\pm \lbrack -2;0 - 1^{4}],\pm \lbrack 4;2,1^{4}],\pm \lbrack 0;0,-1^{2},1^{2}],\pm \lbrack 0;0,-1,1,-1,1],\pm \lbrack 0;0,-1,1^{2},-1]\}\times 5 ; \\ \mathcal{l}_{3}=\{\pm \lbrack 0;1,0^{4}],\pm \lbrack 2;1,2,0^{3}],\pm \lbrack 2;1,0,2,0^{2}],\pm \lbrack 2;1,0^{2},2,0],\pm \lbrack
2;1,0^{3},2]\}\times 5 ; \\ \mathcal{l}_{5}=\{\pm \lbrack 3;0,1^{4}],\pm \lbrack
-1;0,1,-1^{3}],\pm \lbrack -1;0,-1,1,-1^{2 } ] , \\
\pm \lbrack -1;0,1^{2},-1,1],\pm \lbrack -1;0,-1^{3},1]\}\times 5 ; \\ \mathcal{l}_{6}=\{\pm
\lbrack -3;-2,0,-1^{3}],\pm \lbrack -3;0,-2,-1^{3}],\pm \lbrack 1;0^{2},-1,1^{2 } ] , \\
\pm \lbrack 1;0^{2},1,-1,1],\pm \lbrack 1;0^{2},1^{2},-1]\}\times 10.\end{gathered}\ ] ] these parity classes are grouped into six @xmath277classes , as indicated by the notation .
this gives a second accounting for the @xmath255 elements of @xmath254 in geometric terms .
more directly related to our main line of argument are the following results on @xmath298 .
[ longtriangles1.c2]@xmath299 , where @xmath300 is permuation of the last four coordinates .
this statement follows from the geometric description of @xmath267 above , and the observation that @xmath301 .
[ longtriangles1.c3]there are @xmath255 @xmath244triangles @xmath302equivalent to @xmath267 , which are inscribed in g - topes in the star at the origin .
it follows from the equality @xmath303 that there are @xmath255 @xmath304triangles @xmath292equivalent to @xmath267 . that these are inscribed in g - topes in the star follows from _ corollary [ longtriangles1.c1]_. [ [ tiling - with - g - topes.-1 ] ] tiling with g*-topes .
+ + + + + + + + + + + + + + + + + + + + + the three @xmath305 _ _ triangles _ _ @xmath306,[0^{4},1,0],[0^{5},1]\ } , \\
\delta _ { t}^{2 } & = conv\{[2,0,1^{4}],[-1,0^{2},-1,0^{2}],[-1,0,-1,0^{3}]\ } , \\
\delta _ { t}^{3 } & = conv\{[0,1,0^{4}],[1,0^{3},1^{2}],[-1^{2},0^{4}]\},\end{aligned}\ ] ] have a common centroid @xmath307 $ ] , and belong to complementary @xmath308spaces .
their convex hull is a lattice polytope with nine vertices , which we denote by @xmath309 .
this _ reference _ @xmath310__tope _ _ is a tile in the delaunay tiling @xmath311 determined by @xmath133 .
all others are isometrically equivalent to @xmath309 relative to @xmath312 . in total , there are @xmath313 homology classes of @xmath314-topes , accounting for @xmath315 @xmath314-topes in the star at the origin .
these inscribed triangles are @xmath305 triangles because the edge vectors @xmath316,\pm \lbrack 0^{5},-1],\pm \lbrack 0^{4},1,0]\ } , \\
e_{t}^{2}= & \{\pm \lbrack 0^{2},1,-1,0^{2}],\pm \lbrack -3,0,-2,-1^{3}],\pm \lbrack 3,0,1,2,1^{2}],\ } , \\
e_{t}^{3}= & \{\pm \lbrack 2,1,0^{2},1^{2}],\pm \lbrack -1,-2,0^{4}],\pm
\lbrack -1,1,0^{2},-1^{2}]\},\end{aligned}\ ] ] belong to @xmath317 , the set of @xmath266 long vectors for @xmath131 .
the long vectors for @xmath133 are defined by @xmath318 where @xmath319 is the second minimum for @xmath133 .
there is a second representation @xmath320 for the point group for @xmath2 , which can be characterized as either the invariance group for the form @xmath133 or the stabilizer of @xmath258 .
the following theorem gives the essential information we will need on this action ( see e.g. baranovskii ( 1991 ) ) .
[ t.g*-equivalence]the group @xmath169 acts transitively on the @xmath251 elements of @xmath321 , the @xmath266 elements of @xmath317 , and , the @xmath255 @xmath314-topes in the star at the origin . from the arithmetic equivalences
@xmath322 and @xmath323 , it follows that @xmath324 and @xmath325 , where @xmath128 was described above .
the dual groups @xmath326 and @xmath327 , where @xmath328 is the matrix for the dual transformation . the proof of following corollary will be included into the full - length version of this paper .
[ c1.g*-equivalence]there are @xmath255 @xmath329triangles with a vertex at the origin . these are @xmath330equivalent . the @xmath331 triangle @xmath332,[0;1,0^{4}],[2;0,1^{4}]\}\ ] ]
is a 2-face of @xmath309 , and is equal to the @xmath244 triangle @xmath267 .
the edge set @xmath333 @xmath334 .
[ c2.g*-equivalence]there are @xmath335 @xmath331 triangles with a vertex at the origin .
these are @xmath308faces of @xmath309-topes , and are @xmath336 equivalent .
this is an immediate consequence of the arithmetic equivalences @xmath337 , @xmath338 , and _ corollary [ c1.g*-equivalence]_. [ [ long - triangles - inscribed - in - g - topes . ] ] long triangles inscribed in g*-topes .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the @xmath266 long vectors for @xmath133 are given by @xmath339 , where @xmath340 , and @xmath341,\pm \lbrack 1,2,0^{4}],\pm \lbrack 0^{2};1,0^{3}],\pm \lbrack 2,1;0,1^{3 } ] , \\ & & \pm \lbrack 1,-1;0,1^{3}],\pm \lbrack 3,0;2,1^{3}],\pm \lbrack 1,-1;1^{2},0^{2}]\}\end{aligned}\ ] ] the proof of the following lemma will be included into the full - length exposition of our counterexample .
[ longtriangles2.l]there are @xmath342 homology classes of @xmath343triangles .
[ longtriangles2.p]the lattice polytope @xmath344 is a g*-tope if and only if it is the convex hull of three @xmath305triangles with a common centroid .
each g*-tope is the convex hull of three @xmath305triangles with a common centroid . since there are @xmath313 homology classes of g*-topes , there are @xmath342 homology classes of @xmath305triangles inscribed in g*-topes . by _ lemma [ longtriangles2.l ]
_ this accounts for all the homology classes of @xmath305triangles , from which the proof of the proposition immediately follows .
since @xmath345,\pm \lbrack 1;0,-1^{4}],\pm \lbrack 2;-2,-1^{4}]\}$ ] is the edge set for a @xmath329triangle , this set spans a two - dimensional subspace .
let @xmath346 be the four - dimensional sublattice defined by @xmath347and let @xmath348 be the restriction of @xmath349 to @xmath350 .
the minimal vectors for @xmath348 are given by @xmath351 .
these minimal vectors determine @xmath352 , and @xmath353 .
this is sufficient to identify @xmath354 is a metrical form for the root lattice @xmath4 .
the set of long vectors for @xmath348 is given by @xmath355 which is the union of three parity classes of eight vectors each .
these classes are the diagonal vectors for three orientations of cross polytopes , which tile @xmath350 by translates .
this is the delaunay tiling @xmath356 of @xmath350 relative to the form @xmath352 .
this delaunay tiling can also be realized as the four - dimensional section @xmath357 .
[ [ the - group - mathcalg_d_4 . ] ] the group @xmath358 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath359 @xmath360 . since by _ proposition [ invariance.p ] _
@xmath264 is the stabilizer of @xmath361 @xmath362 , and @xmath169 is the stabilizer of @xmath363 @xmath364 , _ _
_ _ @xmath365 can be characterized as the subgroup of @xmath166 that stabilizes each of the sets @xmath139 , @xmath366 , @xmath367 .
the short and long vectors for @xmath348 relate to the sets @xmath368 and @xmath141 in the following way @xmath369 where the last union is disjoint .
@xmath370,\pm \lbrack 0;1,0^{4}],$]@xmath371\}$ ] is the edge set @xmath372 for the triangle @xmath267 , which is inscribed in @xmath250 .
the proof of the following proposition will be included in to a full - length paper
. [ d4.p]@xmath373 there are six @xmath244trianges with a vertex at the origin , and with edge set @xmath374 .
these fit around the origin , edge - to - edge , to form a hexagon .
the triangle @xmath267 is one of the tiles in this hexagon , and the other five are triangles homologous to it .
since @xmath375 , it follows from _ proposition [ longtriangles1.p ] _ and _ corollary [ longtriangles1.c1 ] _ that @xmath376 . however , the point group for the lattice @xmath4 has order @xmath377 , which is less than @xmath378 by a factor of two .
this discrepancy is explained by the following corollary . [
d4.c1]the restriction of @xmath379 to @xmath350 is a representation of the point group for the root lattice @xmath4 . since the point group for @xmath4 is the only lattice point group for a four - dimensional lattice of order @xmath377 , we need only show the kernel of the restriction map is a two element group .
the kernal @xmath380 is the subgroup that fixes each point in @xmath381 .
the four diagonals @xmath382 , of the cross - polytope in the long layer of @xmath250 form a basis for @xmath383 , and therefore an arbitrary element @xmath384 must fix these diagonals .
it follows that @xmath385 must then fix the parity class @xmath386 .
these two conditions can hold only if @xmath387 maps @xmath388 to @xmath389 . by the geometric discussion of the group @xmath390
, it follows that @xmath391 @xmath392 , where @xmath393 is central inversion and @xmath394 is the identity .
let @xmath395 , where @xmath396 , @xmath397 . if @xmath398 , then @xmath399 must also equal @xmath394 , and @xmath400 .
if @xmath401 , then @xmath399 must invert the four diagonals @xmath402 , which uniquely determines the element @xmath403 .
one of the ways to write @xmath399 in this case is as the product @xmath404 , where @xmath405 inverts the second and third axes of the cross - polytope , and @xmath406 inverts the fourth and fifth .
therefore , when @xmath401 , @xmath407 @xmath408 , and the matrix for @xmath409 is given by @xmath410@xmath411 .\ ] ] it follows that @xmath412 , and is a two element group .
[ [ the - faces - of - mathcald_d_4 . ] ] the faces of @xmath413 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the origin @xmath414 , and the star of delaunay tiles at the origin has @xmath415 cross - polytopes , eight belonging to each orientation .
the lower dimensional cells are @xmath416simplexes .
[ d4.c2]for the tiling @xmath417 , the star of @xmath418cells at the origin contains @xmath415 edges , @xmath419 triangles , @xmath419 @xmath420simplexes , and @xmath415 cross - polytopes . for each dimension , the cells in the star are @xmath421equivalent . by _ corollary [ d4.c1 ] _
the homorphic image of @xmath379 on @xmath422 is a group of order @xmath377 , which acts effectively on @xmath422 , and is a representation of the point group for the lattice @xmath4 .
the order of the full linear invariance group of a cross - polytope , attached to the origin in @xmath423 , is @xmath424 . since @xmath425 @xmath424 , the stability group for each of the @xmath415 cross - polytopes in the star has order @xmath426 , and the @xmath421orbit of each includes all @xmath427 cross - polytopes .
since the stability group in @xmath379 of any particular cross - polytope in the star is the full invariance group , all edges , all @xmath308faces and @xmath420faces of the cross - polytope , which are attached to the origin , are @xmath421equivalent .
the number of edges attached to the origin is eqaul to the number of short vectors , which is @xmath415 .
each cross - polytope has @xmath428 @xmath308cells attached to the origin , and each @xmath308cell belongs to @xmath429 cross - polytopes in the star .
therefore , there are @xmath430 @xmath308cells attached to the origin .
each cross - polytope has 8 facets attached to the origin , and each belongs to a pair of cross - polytopes . therefore , there are @xmath431 @xmath420cells attached to the origin .
[ [ equivalent - facets - of - phi - mathcalp_e_6ast - and - phi - mathcalp_e_6 . ] ] equivalent facets of @xmath100 and @xmath432 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + since @xmath433 are the full invariance groups of the corresponding sets of perfect vectors @xmath434 and @xmath435 , @xmath100 is invariant with respect to the action of @xmath264 , and @xmath185 is invariant with respect to the action of @xmath436 .
@xmath437 the proof of the following corollary will be included in to a full - length paper .
the perfect domain @xmath100 has @xmath243 facets that are @xmath292equivalent to @xmath438 , and , the perfect domain @xmath185 has @xmath243 facets that are @xmath336equivalent to @xmath439 .
the @xmath243 @xmath292equivalent facets of @xmath440 described in this corollary are in one - to - one correspondence with the @xmath243 possible edge sets for @xmath441triangles inscribed in @xmath250-topes , and , the @xmath243 @xmath442equivalent facets of @xmath116 are in one - to - one correspondence with the @xmath243 possible edge sets for @xmath443triangles , which are @xmath308faces of @xmath314-topes .
a @xmath309-tope homologous to the reference @xmath309-tope is called a @xmath444__-tope__. since @xmath445 this @xmath309-tope is commensurate with @xmath250 , and therefore commensurate with the delaunay tiling @xmath446 .
we will count @xmath309-topes below , and discover that there are @xmath415 homology classes of such commensurate @xmath309-tope .
there is a second _ incommensurate _ class of @xmath447tope with representative @xmath448 , which is the convex hull of the three @xmath343triangles @xmath449,[0^{4},1,0],[0^{5},1]\},\qquad \delta _ { q}^{2}=conv\{[0,1,0^{4}],[-1^{2},0^{4}],[1,0^{2},1^{3}]\ } , \\
\delta _ { q}^{3}=conv\{[2,0,1^{4}],[-1,0^{5}],[-1,0,-1,0^{3}]\},\end{gathered}\ ] ] with common centroid @xmath450 $ ] . in the counting exercise below
we will find there are @xmath451 homology classes of incommensurate @xmath309-topes , and will refer @xmath309-topes in any of these classes as a _ q - topes_. there are only two homology classes of @xmath309-tope , so the @xmath415 commensurate classes and the @xmath451 incommensurate classes sum to the required number of @xmath313 homology classes of @xmath309-tope .
the commensurate and incommensurate classes , and the @xmath427 commensurate classes there are sixteen homology classes of such @xmath250 * -topes , and we will refer to these as _ q - topes_. an application of _ proposition [ longtriangles2.p ] _ shows that @xmath448 is a @xmath250 * -tope . that @xmath448 is incommensurate with @xmath250 is established by considering the the simplicial @xmath420faces @xmath452 and @xmath453 .
the simplex @xmath454,[2,0,1^{4}],[-1^{2},0^{4}],[-1,0^{5}]\}\subset q$ ] is the convex hull of an edge from each of the triangles @xmath455 , @xmath456 , so is a @xmath420face of @xmath448 .
the centroid @xmath457 $ ] is also the centroid of the simplex @xmath458,[0^{3},1,0^{2}],[0^{4},1,0],$ ] @xmath459\}$ ] , which is a @xmath420face of @xmath250 .
these simplexes lie in complementary @xmath420spaces , so satisfy the condition that @xmath460 $ ] .
the relatively open cells of the intesection polytope @xmath461 are the intersections of relatively open cells of @xmath448 and @xmath250 .
for this reason the common centroid is a vertex of @xmath461 , and since it is non - integral , @xmath448 is incommensurate with @xmath250 .
therefore @xmath448 is incommensurate with the delaunay tiling @xmath252 .
another important class lattice polytope is the class of _ r - topes _ ; the lattice polytope @xmath462 serves as an example .
this r - tope has full dimension and eight vertices , so can be triangulated in just two ways . by taking the convex hull of @xmath463 with each @xmath308face of @xmath464 , the four _ blue simplexes _
@xmath465 are formed ; this collection is the star of simplexes in @xmath13 with the @xmath420face @xmath466 .
these blue simplexes tile @xmath13 .
the four _ yellow simplexes _ , @xmath467 , form a second star , each yellow simplex being the convex hull of @xmath468 with one of the four @xmath308faces of @xmath463 .
this is the collection of simplexes in @xmath13 with the @xmath420face @xmath468 . the yellow simplexes also tile @xmath13 .
lattice polytopes such as @xmath13 are refered to as _ repartitioning complexes _ in the literature on lattice delaunay tilings . in the counting exercise below
we will see that there are @xmath428 homology classes of r - topes .
[ [ stars - of - g - topes . ] ] stars of g - topes .
+ + + + + + + + + + + + + + + + + in order to effectively study the geometry of q - topes and r - topes we must first obtain information on the relationship between the perfect vectors @xmath469 , and the g - topes in the star of g - topes at the origin .
if @xmath470 is the vertex set for @xmath250 , then , relative to @xmath132 , the set vertices of @xmath250 that are a distance @xmath45 from the origin are given by @xmath471 , and the set vertices that are a distance @xmath257 are given by @xmath472 .
these _ short _ and _ long layers _ of vertices can also be characterized in terms of the perfect vector @xmath473 $ ] using the _ incidence relations _ for g - topes : @xmath474 the perfect vector @xmath214 determines the _ short _ and _ long layers _ of vertices , which in turn determine @xmath250 , so we write @xmath475
. this process can be reversed , and the incidence relations can be used to determine the perfect vector associated with @xmath250 : @xmath476 for this case the set @xmath477 contains the single elements @xmath478 . by _ proposition [ invariance.p ] _ , the incidence relations can be used to establish a one - to - one correspondence between the @xmath250-topes in the star at the origin , and the vectors of @xmath434 : if @xmath479 then @xmath480 is the corresponding @xmath250-tope in the star , and if @xmath481 is in the star , then @xmath482 is the corresponding perfect vector .
this association can be extended to cover arbitrary cells @xmath483 with a vertex at the origin .
if @xmath484 and @xmath485 are the short and long layers for @xmath14 , define @xmath486 when @xmath487 the set @xmath488 includes several perfect vectors .
the short and long layers , and therefore @xmath14 , can be recoverd from @xmath488 by again invoking the incidence relations : @xmath489 for a cell @xmath490 , @xmath491 is the collection of g - topes with face @xmath14 .
the following proposition gives information on g - stars when @xmath14 has a vertex at the origin .
[ g - star.p]if @xmath492 , @xmath493 .
this assertion follows from the discussion on incidence relations . [
[ the - geometry - of - r - topes . ] ] the geometry of r - topes .
+ + + + + + + + + + + + + + + + + + + + + + + + in the statement of the following lemma @xmath494 is the collection of g - topes , which have the face @xmath495 , and @xmath496 is the collection of g*-topes , which have the face @xmath497 . the proof of the following lemma will be included into the full - length paper .
[ r.l]@xmath498 , where @xmath499 are q - topes , @xmath500 , and the blue simplexes can be ordered so that @xmath501 .
similarly , @xmath502 , where @xmath503are g - topes , @xmath504 , and the yellow simplexes can be ordered so that @xmath505 .
[ r.p]@xmath13 is commensurate with both @xmath252 and @xmath506 . by _ lemma [ r.l ] _
, the intersection @xmath507 is tiled by the blue lattice simplexes , so @xmath13 is commensurate with @xmath508 .
similarly , the intersection @xmath509 is tiled by the yellow lattice simplexes , so @xmath13 is commensurate with @xmath252 . [ [ the - geometry - of - q - topes . ] ] the geometry of q - topes .
+ + + + + + + + + + + + + + + + + + + + + + + + the convex hull of @xmath510 with one edge from each of the other triangles @xmath511 has seven vertices , and is a lattice simplex in @xmath448 .
two edges can be selected in nine ways , so there are nine such simplexes , each with @xmath510 as a @xmath308face .
this is the star of simplexes in @xmath448 with the @xmath308face @xmath510 , and this collection tiles @xmath448 . in this subsection
we show how six of these simplexes are commensurate with both @xmath252 and @xmath311 .
we call them _ white . _
the other three are _ blue _ , and can be represented as the intersection of @xmath448 with three distinct r - topes .
there are two other collections of simplexes that tile @xmath448 , the star of simplexes in @xmath448 with the @xmath308face @xmath455 , and the star with the @xmath512face @xmath513 . however , both of these tilings play no role in our discussion .
the reason for this is that @xmath510 is a @xmath308cell in @xmath252 , but the other two triangles @xmath514 are not .
the proof of the following lemma will be included into the full - length paper .
[ q.l1]@xmath515 , where @xmath516 are g - topes and @xmath517 . moreover , for @xmath518 , @xmath519 is a six - dimensional lattice simplex with the property that @xmath520
. we will refer to these simplexes as white .
[ q.l2]there are three r - topes @xmath521 , such that @xmath522 , and such that @xmath523 are six - dimensional lattice simplexes with the @xmath308face @xmath510 . since these simplexes are the intersections of r - topes and a q - tope they are blue . the three simplexes @xmath524,[2,0,1^{4}],[-1^{2},0^{4}],[-1,0^{5}]\ } , \\
s_{q}^{2 } & = & conv\{[0,1,0^{4}],[2,0,1^{4}],[1,0^{2},1^{3}],[-1,0,-1,0^{3}]\ } , \\
s_{q}^{3 } & = & conv\{[-1^{2},0^{4}],[-1,0^{5}],[1,0^{2},1^{3}],[-1,0,-1,0^{3}]\},\end{aligned}\ ] ] are @xmath420faces of @xmath448 , and @xmath525 .
each is the convex hull of an edge from @xmath455 and @xmath513 .
the three simplexes @xmath526,[0^{3},1,0^{2}],[0^{4},1,0],[0^{5},1]\ } , \\ s_{g}^{2 } & = & conv\{[2,1,0,1^{3}],[0^{3},1,0^{2}],[0^{4},1,0],[0^{5},1]\ } , \\ s_{g}^{3 } & = & conv\{[-2,-1^{2},0^{3}],[0^{3},1,0^{2}],[0^{4},1,0],[0^{5},1]\},\end{aligned}\ ] ] are @xmath416simplexes , and @xmath527 .
the three centroids of the @xmath528simplexes @xmath529 , are given by @xmath457 $ ] , @xmath530 $ ] , @xmath531 $ ] , and these coincide with the centroids of the corresponding @xmath416simplexes @xmath532 .
the three @xmath416simplexes are translates of closed @xmath420cells in @xmath417 , so by proposition [ d4.c2 ] belong to @xmath533equivalent homology classes .
it follows that the three lattice polytopes @xmath534 , @xmath535 , @xmath536 , belong to @xmath421equivalent homology classes . since @xmath537 , these three lattice polytopes are r - topes . since by lemma [ q.l2 ] @xmath538 is a six - dimensional blue lattice simplex , the intersections @xmath539 , @xmath540 , are other six - dimensional blue lattice simplexes . by construction , @xmath541 .
[ q.p]the star of simplexes in @xmath448 containing @xmath510 , has six white simplexes @xmath542 , and three blue simplexes @xmath543 .
the six white simplexes are commensurate with @xmath544 and @xmath508 , and the three blue simplexes are commensurate with @xmath508 , but incommensurate with @xmath252 .
the simplexes @xmath542 , are those described in the statement of _ lemma [ q.l1]_. since @xmath545 , these white simplexes are commensurate with both @xmath252 and @xmath546 .
the simplexes @xmath547 , are those described in the statement of _ lemma [ q.l2]_. since @xmath548 , the blue simplex @xmath549 is commensurate with @xmath508 .
the simplex @xmath550 is a @xmath420face of @xmath551 , and the simplexe @xmath552 is a closed 3-cells in @xmath252 .
since the intersection @xmath553 is the centroid of each of these cells , @xmath554 and @xmath552 , the simplex @xmath551 is incommensurate with @xmath446 . [ [ t - topes - q - topes - and - r - topes . ] ] t - topes , q - topes and r - topes .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the intersection properties of any g*-tope with respect to @xmath446 , are invariant with respect to the @xmath421action .
therefore , any g*-tope @xmath421equivalent to @xmath344 is commensurate with @xmath252 , and any g*-tope @xmath555equivalent to @xmath448 is incommensurate with @xmath252 . since commensurability is a property that extends to homology classes , it is natural to make the following definitions : _ a t - tope is a lattice polytope _
@xmath421__equivalent to one homologous to _ _
@xmath344 _ _ and , a q - tope is a lattice polytope _
_ @xmath421__equivalent to a lattice polytope homologous to _ _
@xmath448 _ _ as a q - tope . _
_ it is also natural to refer to commensurate and incommensurate homology classes of g*-topes , so that homology classes of t - topes are commensurate classes and homology classes of q - topes are incommensurate . by the definitions of t - tope and q - tope , the homology classes of t - topes are @xmath533equivalent and , the homology classes of q - topes are @xmath533equivalent .
each of the triangles @xmath556 , @xmath557 , @xmath558 in @xmath344 has two edges that are long relative to @xmath130 , and a single edge that is short .
this is a property that is @xmath559invariant , so each t - tope has three inscribed triangles of this type . on the other hand ,
the triangle @xmath510 in @xmath448 has edges that are short relative to @xmath130 , and the triangles @xmath560 , @xmath513 have edges that are long .
accordingly , each q - tope has one short and two long inscribed triangles .
the set of r - topes is the other collection we must consider : _
an r - tope is a lattice polytope _
@xmath421__equivalent to a lattice polytope homologous to _
_ @xmath13 .
the homology classes of r - topes are @xmath561equivalent , and each r - tope in any of these classes has identicle intersection properties .
if @xmath562 is an arbitrary r - tope , then @xmath563 has four yellow simplexes @xmath564 , @xmath565 , @xmath566 , @xmath567 , which belong to distinct g - topes and fit around a @xmath420cell @xmath568 ; @xmath569 has four blue simplexes @xmath570 , @xmath571 , @xmath572 , @xmath573 , which belong to distinct q - topes and fit around a @xmath420cell @xmath574 .
each r - tope @xmath562 is commensurate with both @xmath544 and @xmath508 .
[ repackaging.l]the @xmath313 homology classes of g*-topes are divided between @xmath415 homology classes of t - topes , and @xmath451 homology classes of q - topes .
there are @xmath428 homology classes of r - topes .
will be included into the full - length paper [ [ repackaging - q - topes . ] ] repackaging q - topes .
+ + + + + + + + + + + + + + + + + + + + each q - tope contains six white , and three blue simplexes , and since there are @xmath451 homology classes of q - topes , there are @xmath419 homology classes of white , and @xmath575 homology classes of blue simplexes .
the @xmath428 homology classes of r - topes gives a second accounting for blue simplexes , since each r - tope can be tiled by four blue simplexes .
this allows a repackaging of the blue simplexes contained in q - topes , into r - topes .
the portion of space tiled by the @xmath451 classes of q - topes , can equally well be tiled by the @xmath419 homology classes of white simplexes , and @xmath428 classes of r - topes .
this alternate tiling has the property that each tile is commensurate with both @xmath544 and @xmath508 .
[ structure of d_r]the @xmath415 homology classes of t - topes , the @xmath428 homology classes of r - topes , and the @xmath419 homology classes of white simplexes fit together facet - to - facet to tile space .
this lattice tiling @xmath576 is commensurate with both @xmath252 and @xmath546 . since the @xmath344-topes and @xmath448-topes tile space
, the discussion immediately preceeding the proposition shows that the @xmath344-topes , @xmath13-topes and white simplexes also tile space . by _ lemmae _
[ r.l ] , _ [ q.p ] _ and _ proposition [ r.p ] _ this tiling is commensurate with both @xmath577 and @xmath508 .
the forms @xmath578 and @xmath579 lie on the central axes of their domains , and the line segment @xmath580 , @xmath581 , runs between them .
the arithmetic minimum along this segment is @xmath45 . at the end
points the minimal vectors are respectively @xmath258 and @xmath582 , but at intermediate points the minimal vectors are given by @xmath583 .
suppose that the reference @xmath13 is delaunay with respect to the metrical form @xmath102 .
then there is a scalar @xmath585 and vector @xmath586 so that @xmath587 is non - negative on @xmath73 , and zero valued on just the vertices of @xmath13 ; @xmath588 is the equation of an empty ellipsoid circumscribing @xmath13 .
the vertex sets for the component simplexes @xmath468 , @xmath463 of @xmath13 are given by @xmath589,[0^{3},1,0^{2}],[0^{4},1,0],[0^{5},1]\ } , \\
v_{q } & = & \{[0,1,0^{4}],[2,0,1^{4}],[-1^{2},0^{4}],[-1,0^{5}]\},\end{aligned}\ ] ] and have the property that @xmath590 $ ] .
since @xmath591 is zero on each vertex of @xmath13 , the metrical form @xmath102 must satisfy the condition .
@xmath592 where @xmath593 is the matrix for @xmath102 and @xmath594 .\ ] ] the final expression @xmath595 is the scalar product of forms @xmath596 and @xmath102 , where @xmath597 corresponds to @xmath598 ; if forms @xmath101 , @xmath102 have coefficient matrices @xmath103 , @xmath104 , then the scalar product @xmath599 is defined to be @xmath600 . [ computing the l - type break - point]the line segment @xmath603 , @xmath581 , pierces the hyperplane @xmath602 at the point @xmath604 , where @xmath605 .
there are twelve hyperplanes of the form @xmath606 , where @xmath562 is an r - tope , and the line segment @xmath603 , @xmath581 , pierces each of these at @xmath607 , which is common to all of them .
each r - tope @xmath562 corresponds to a form @xmath611 , which in turn determines a hyperplane @xmath606 with equation @xmath612 . if @xmath613 @xmath614 , where @xmath615 , then the corresponding form is given by @xmath616 and @xmath617 @xmath618 is the matrix dual to @xmath272 . on the other hand , if @xmath562 is homologous to @xmath13 , then @xmath619 and @xmath620 .
there are twelve such hyperplanes , as there are twelve homology classes of r - topes ( see _ lemma [ repackaging.l ] _ ) . the equality @xmath621 , which holds for all @xmath622 implies that the line segment @xmath623 pierces all of them at the same point .
recall that @xmath624 consists of 24 homology classes of @xmath344-topes , 96 homology classes of white simplexes , and 12 homology classes of @xmath13-topes ( proposition [ structure of d_r ] ) .
each @xmath344__-tope _ _ @xmath344 _ _ _ _ in @xmath576 is a delaunay cell of @xmath252 whose intersection with a @xmath250-tope of @xmath508 is @xmath625 each _ white _ simplex in @xmath624 is the intersection of a @xmath448-tope , delaunay cell of @xmath626 and a @xmath250-tope , delaunay cell of @xmath252 . by lemma [ test on commensurability of 2 cells ] all white simplexes and @xmath344__-topes _ _ are delaunay with respect to any intermediate form , including @xmath627 by lemma [ computing the l - type break - point ] and the preceeding discussion , all homology classes of @xmath13-topes are delaunay for @xmath627 [ commensurate.p]for @xmath628 , the delaunay tiling for each form @xmath629 is constant , and obtained by replacing each r - tope in @xmath624 by four yellow simplexes . for @xmath630 ,
the delaunay tiling for each form @xmath623 is also constant , and obtained by replacing each r - tope in @xmath576 by four blue simplexes .
consider the form @xmath632 and the rank one form @xmath633 .
then , @xmath634 @xmath635 . if @xmath636 , then @xmath637 , with equality if an only if @xmath638 , and , if @xmath639 , then @xmath640 , with equality if an only if @xmath641 .
it follows that the hyperplane @xmath642 , with equation @xmath643 , separates @xmath644 and @xmath100 , and contains the facet @xmath98 . the forms @xmath623 are invariant with respect to the action of @xmath652 .
since @xmath653 acts transitively on @xmath654 , which can easily be checked , the only @xmath555invariant positive forms on @xmath642 must lie on the central axis of @xmath98 .
therefore , the line segment @xmath623 , @xmath581 , pierces the hyperplane along the central axis of @xmath655 .
* proof of theorem [ main ] .
* by lemma [ d_r is delaunay for phi ] @xmath624 is the delaunay tiling for the midpoint of the segment @xmath656 by proposition [ structure of d_r ] @xmath624 is commensurate with both @xmath252 and @xmath508
. therefore by lemma [ commense ] the @xmath0-type does not change between @xmath130 and @xmath657 and between @xmath133 and @xmath658 by the results of section [ forms ] @xmath130 and @xmath131 are the centroids of two _ adjacent _ perfect domains and , as shown in the above lemma , the wall between them intersects the segment @xmath623 at @xmath659 e. p. baranovskii , partition of euclidean spaces into @xmath0-polytopes of certain perfect lattices .
( russian ) discrete geometry and topology ( russian ) .
_ trudy mat .
inst . steklov_.
* 196 * ( 1991 ) , 2746 ; translated in _ proc .
steklov inst .
_ * * 196 , ( * * 1992 ) , no .
4 , 29 - 51 conway , j. h. ; sloane , n. j. a. _ sphere packings , lattices and groups_. third edition .
grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , * 290*. springer - verlag , new york , ( 1999 ) . delone , b. n. ; dolbilin , n. p. ; rykov , s. s. ; togrin , m. i. a new construction of the theory of lattice coverings of an _ n_-dimensional space by congruent balls .
( russian ) izv .
nauk sssr ser . mat .
34 1970 p. erds , p. m. gruber , j. hammer , _
lattice points .
_ pitman monographs and surveys in pure and applied mathematics , 39 .
longman scientific & technical , harlow ; copublished in the united states with john wiley & sons , inc .
, new york , 1989 .
ryshkov , s. s. ; baranovskii , e. p. @xmath14-types of @xmath7-dimensional lattices and @xmath662-dimensional primitive parallelohedra ( with application to the theory of coverings ) .
cover to cover translation of trudy mat .
steklov * 137 * ( 1976 ) .
steklov inst . math .
1978 , no . * 4 * , 140 pp .
voronoi g. f. nouvelles applications des paramters continus la thorie des formes quadratiques , _ j. reine angew
_ , premiere mmoire , * 133 * ( 1908 ) * , * 97 - 178 * * , * * deuxime mmoire * * , 134 * * ( 1908 ) , 198 - 287 , * 136 * ( 1909 ) , 67 - 178 . | george voronoi ( 1908 , 1909 ) introduced two important reduction methods for positive quadratic forms : the reduction with perfect forms , and the reduction with @xmath0-type domains , often called domains of delaunay type .
the first method is important in studies of dense lattice packings of spheres .
the second method provides the key tools for finding the least dense lattice coverings with equal spheres in lower dimensions . in his investigations
voronoi heavily relied on that in dimensions less than 6 the partition of the cone of positive quadratic forms into @xmath0-types refines the partition of this cone into perfect domains .
voronoi conjectured implicitely and dickson ( 1972 ) explicitely that the @xmath0-partition is always a refinement of the partition into perfect domains .
this was proved for @xmath1 ( voronoi , delaunay , ryshkov , baranovskii ) .
we show that voronoi - dickson conjecture fails already in dimension 6 . short version * keywords : * positive quadratic form , perfect form , point lattice , delaunay tiling ( @xmath0-partition ) , @xmath0-type , repartitioning complex , lattice packings and coverings , lattices @xmath2 and @xmath3 , voronoi reduction , integral represenations of groups @xmath4 , @xmath2 , @xmath5 , gosset polytope @xmath6 * ams classification .
* primary 11 , 52 secondary 15 |
molecular - based magnetic materials have recently attracted considerable interest and study of the magnetic properties @xcite . a special class of the these materials , the so - called prussian blue analogs , such as ( x@xmath5mn@xmath6[cr@xmath7(cn)@xmath8.nh@xmath9o ( x@xmath10=ni@xmath10,fe@xmath10 ) @xcite and ( ni@xmath5mn@xmath11fe@xmath12)@xmath13[cr@xmath7(cn)@xmath8.nh@xmath9o @xcite which exhibit many unusual properties , for instance , occurrence of one @xcite or even two @xcite compensation points , magnetic pole inversion @xcite , the photoinduced magnetization effect @xcite , inverted magnetic hysteresis @xcite .
these ternary alloys have ferromagnetic - ferrimagnetic properties since they include mixed both ferromagnetic ( @xmath14 ) and antiferromagnetic ( @xmath15 ) superexchange interactions between the nearest - neighbor metal ions
. the theoretical investigations of these systems are difficult because of their structural complexity .
however , to obtain magnetic properties of the molecular - based magnetic materials , up to now , these systems have been studied by using effective - field theory @xcite , mean field theory @xcite and monte carlo simulation ( mc ) methods @xcite . in this study
we consider three dimensional ferro - ferrimagnetic ab@xmath0c@xmath1 ternary alloy , consisting of three different ising spins a=3/2 , b=1 , and c=5/2 , which corresponds to the prussian blue analog of the type ( ni@xmath5mn@xmath6[cr@xmath7(cn)@xmath8.nh@xmath9o @xcite . in this system ,
the coupling cr - ni is ferromagnetic and mn - cr is antiferromagnetic .
our aim , in this study , is to clarify the effects of the concentration and the interaction parameters on the magnetic behavior of the three dimensional ternary alloy model by using mc simulation method .
three dimensional ferro - ferrimagnetic ab@xmath0c@xmath1 ising model consists of two interpenetrating cubic sublattices as seen in fig.1 .
it can be assumed that the a ions are located on the first cubic sublattice and the b and c ions are randomly distributed on the second cubic sublattice with the concentration @xmath16 and @xmath17 , respectively . also , to construct a hamiltonian for this system , the ion a can be represented by spin s@xmath2 , and on the other hand , ions b and c can be represented by ising spins s@xmath3 and s@xmath4 , respectively
. if the interactions between nearest neighbors can be chosen such as a ions ferromagnetically interact with b , on the other hand , antiferromagnetically interact with c ions , thus , spins of the prussian blue analog of the type ( ni@xmath5mn@xmath6[cr@xmath7(cn)@xmath8.nh@xmath9o can be represented by this model where s@xmath2 , s@xmath3 and s@xmath4 correspond to cr , ni and mn , respectively . in this study we also consider next - nearest neighbor interactions between spins s@xmath2 . the hamiltonian of the considered system can be written in the form
@xmath18 \nonumber\\ -j_{aa}\sum_{<nnn>}s_{i}^{a}s_{k}^{a}\end{aligned}\ ] ] where s@xmath19 for a , s@xmath20 for b and s@xmath21 for c , on the other hand , @xmath22 is a random variable which takes the value of unity if there is a spin x ( s@xmath3 or s@xmath4 ) at the site @xmath23 , if it not is zero . in eq.(1 ) , the first sum is over the nearest - neighbor and the second one is over the next - nearest neighbor spins . in this hamiltonian
the nearest neighbor interactions are chosen as @xmath24 and @xmath25 , and the next - nearest neighbor interactions are chosen as @xmath26 . in order to show the effects of the concentration @xmath16 and the interaction parameters on the compensation and critical temperature of the three dimensional ternary alloy model
, we simulate the hamiltonian given by eq.(1 ) . to simulate this model , we employed metropolis monte carlo simulation algorithm @xcite to the @xmath27 three - dimensional lattice with periodic boundary conditions for @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 .
one of the cubic sublattice is fully decorated with spin s@xmath2 , and spins s@xmath3 and s@xmath4 are randomly distributed on the other cubic sublattice with the concentration @xmath16 or @xmath17 , respectively .
all initial spin states in the @xmath27 three - dimensional lattice are randomly assigned .
configurations are generated by making single - spin - flip attempts , which were accepted or rejected according to the metropolis algorithm .
to calculate the averages , data , over 20 different spin configuration , is obtained by using @xmath33 monte carlo steps per site after discarding @xmath34 steps .
the sublattice average magnetizations per site are obtained by @xmath35 @xmath36 @xmath37 where @xmath38 denotes the number of b ions @xmath39 , whilst @xmath40 represents the number of c ions @xmath41 on the same cubic lattice .
total magnetization per site is given by @xmath42
+ in this section , we have given the simulation results of the ternary alloy model ab@xmath0c@xmath1 and we have also discussed the dependence of the critical and compensation temperature on the concentration and other interaction parameters in the hamiltonian .
simulation results have been obtained for the system with lattice size @xmath28 , @xmath29 , @xmath30 , @xmath31 and @xmath32 , however , here , we have only presented the results of the model with lattice size @xmath43 .
we also note that the critical temperature of the system for the different interaction rates and concentrations have been obtained by using of the method of the finite - size scaling @xcite . in the three dimensional ternary alloy ab@xmath0c@xmath1 for different values of @xmath16
when @xmath44.,width=302,height=302 ] in a recent study @xcite it was reported that two dimensional ternary alloy model does not show a compensation temperature point when there is no next - nearest neighbor interactions term in the hamiltonian i.e. , @xmath45 . however , our simulations show that the system has a compensation point for all @xmath46 ( we set @xmath47 ) values in interval of @xmath48 at @xmath49 when @xmath45 .
this point will be considered below .
now , in order to compare with the previous results @xcite , in figs.2 and 3 we discuss the dependence of the critical temperature of the three dimensional ternary alloy model ab@xmath0c@xmath1 on interaction rate @xmath46 and concentration @xmath16 for next - nearest neighbor interactions i.e. , @xmath45 . in the three dimensional ternary alloy ab@xmath0c@xmath1 for several values of interaction ratio @xmath46
when @xmath44 .
the lines show a part of the second - order transitions separating the ferrimagnetic and paramagnetic phases.,width=302,height=302 ] in fig.2 , the critical temperature of the three dimensional ternary alloy model has been plotted as a function of @xmath46 for various values of @xmath16 when @xmath45 .
it can be seen from fig.2 that the critical temperature of the system has a linear dependence on the interaction ratio @xmath46 and there is a critical behavior at a special @xmath46 value . when @xmath50 , the critical temperature of the system has a fixed value of @xmath51 for all @xmath16 values . at @xmath52 ,
the critical temperature of the system does not change with concentration @xmath16 .
this means that neither the spin-1 ions nor spin-5/2 ions substitution to system change the critical temperature of the system at @xmath52 .
this critical behavior has been reported in theoretical and experimental studies @xcite . the value of the @xmath52 for ternary alloy ab@xmath0c@xmath1 whose spins consist of s@xmath53 , s@xmath54 and s@xmath55 has been obtained as @xmath56 in the study based on mean field approximation @xcite and as @xmath57 in the monte carlo simulation of two dimensional system @xcite .
furthermore , the experimental measurements indicate that there are prussian blue analogs at the @xmath58 have a @xmath59 which is almost independent of @xmath16 @xcite .
fig.2 also reveals that concentration @xmath16 plays an important role for the ternary alloy model ab@xmath0c@xmath1 since it determine the kinds of the spins and interactions in the system .
for example , when @xmath60 and @xmath49 , the system ab@xmath0c@xmath1 fully reduces to the ferromagnetic mixed spin-3/2 and spin-1 and ferrimagnetic mixed spin-3/2 and spin-5/2 ising system , respectively . as seen in fig.2 ,
although @xmath59 of the system is independent of @xmath16 at @xmath52 , however , the total magnetization of the system may considerably change owing to relatively small variation of the concentration @xmath16 .
indeed , for different values of @xmath16 , the dependence of critical temperature of the system on the interaction ratio @xmath46 is very different above and below of @xmath52 .
this behavior can be explained by the change of the concentration @xmath16 in the system . on the other hand
, it can be detected from fig.2 that when @xmath61 , the critical temperature of the mixed spin-3/2 and spin-5/2 system is smaller than mixed spin-3/2 and spin-1 system .
on the contrary , when @xmath62 , the critical temperature of the mixed spin-3/2 and spin-5/2 system has the highest value . on the @xmath59 lines , the critical temperature of the mixed
spin-3/2 and spin-1 ising system is equal to that of the mixed spin-3/2 and spin-5/2 ising one .
c@xmath1 vs temperature for different values of @xmath16 ( @xmath63 and @xmath64 ) .
, width=302,height=302 ] in fig.3 , the dependence of the critical temperature of the three dimensional ab@xmath0c@xmath1 system on the concentration @xmath16 has been shown for several values of @xmath46 when @xmath45 .
the lines represent part of the second - order phase transition separating the ferrimagnetic and paramagnetic .
fig.3 provides the argument that the concentration @xmath16 determines the magnetic features of the system mentioned above . indeed , fig.3
clearly shows that the critical temperature of the system is changed by the concentration @xmath16 for fixed values of @xmath46 . as seen from this figure that , when @xmath61 , the critical temperature of the system linearly increases with increasing of @xmath16 , whereas , when @xmath62 , the critical temperature of the system linearly decreases with increasing of @xmath16 for fixed values of @xmath46 .
however , when the values of @xmath46 close up @xmath52 , the critical temperature of the system more slowly , but linearly , change with increasing @xmath16 , and at the critical @xmath52 value , the critical temperature of the system denoted by triangle - line in fig.3 is independent of the concentration @xmath16 . on the other hand
, fig.3 also shows that the interaction rate @xmath46 plays an important role on the critical temperature of the three dimensional ab@xmath0c@xmath1 system .
finally we state that the critical temperature of the model are consistent with previous result @xcite .
c@xmath1 vs temperature for different values of @xmath16 ( @xmath63 and @xmath65 ) .
, width=302,height=302 ] in this study we recognize that the three dimensional ternary alloy model ab@xmath0c@xmath1 has one compensation behavior for @xmath45 , however , for @xmath66 it has one or multi compensation points , when other conditions are satisfied .
however , the appearance of the compensation temperature is strongly affected by the interaction and concentration parameters .
indeed we see in the present study that the model has not a compensation point for all values of @xmath16 and @xmath46 .
the dependence of the compensation temperature behavior on concentration and other interaction parameters has been discussed below . for discussion , although the system has been simulated in the intervals of @xmath67 and @xmath68 , the value of @xmath69 used in the present study is chosen based on previous theoretical study @xcite .
the results of simulation for @xmath64 and @xmath65 are respectively represented in figs.4 and 5 for chosen parameters .
one compensation point has been found in the intervals of @xmath70 and @xmath71 when @xmath63 .
however , it is seen that the system has not compensation behavior for the same values of parameters when @xmath72 . for @xmath64 and several values of @xmath16 ,
the compensation behavior of the system can be seen from fig.4 . on the other hand ,
as seen from fig.5 , the considered system has a multi compensation behavior at @xmath65 and @xmath73 for @xmath63 while it has one compensation point for @xmath74 .
furthermore , our simulation data introduce that the system shows compensation behavior at @xmath49 for all values of @xmath46 , when @xmath45 .
in addition , in the case @xmath45 , the compensation point has been found for @xmath75 at @xmath76 , @xmath77 ; for @xmath78 at @xmath73 ; for @xmath79 at @xmath73 ; for @xmath80 at @xmath81 , @xmath82 , @xmath83 ; for @xmath65 at @xmath81 , @xmath84 , @xmath82 , @xmath83 . whereas , it has been reported in previous study that there is no compensation point for @xmath45 in two dimensional model @xcite . on the interaction parameters in hamiltonian of ternary alloy ab@xmath0c@xmath1 for @xmath85.,width=302,height=302 ]
the effect of the interaction parameters on the compensation behavior of the three dimensional ternary model is also discussed in fig.6 .
this figure shows dependence of the compensation temperature @xmath86 of the model on interaction parameters in hamiltonian only for a fixed value of the concentration parameter @xmath16 ( @xmath85 ) . in this figure
, square - line represents the behavior of the compensation point vs @xmath69 for fixed values of @xmath87 , @xmath88 and @xmath85 , circle - line indicates the behavior of the compensation point vs @xmath89 for fixed values of @xmath63 and @xmath88 and @xmath85 , and on the other hand , the behavior of the compensation point vs @xmath90 is plotted for fixed values of @xmath63 , @xmath87 and @xmath85 with triangle - line . as seen from fig.6 that for fixed @xmath16 , @xmath87 and @xmath88 , the compensation temperature decreases slowly as the strength of the @xmath69 increases .
similarly for fixed @xmath16 , @xmath63 and @xmath88 , the compensation temperature decreases slowly with increasing of @xmath89 .
however , for fixed @xmath16 , @xmath63 and @xmath87 , the compensation temperature dramatically increases as @xmath91 increases .
these results indicate that the compensation temperature has a strong dependence on the parameter @xmath90 whereas its dependence on @xmath69 and @xmath89 is relatively weak .
the characteristic behavior of the dependence of the compensation temperature on the parameters of present model consistent with the results of two dimensional model @xcite .
in this study , we have considered the three dimensional ternary model ab@xmath0c@xmath1 whose spins consist of s@xmath53 , s@xmath54 and s@xmath55 .
we have investigated the dependence of the critical and compensation temperature behavior of the considered model on concentration and interactions by using mc simulation method .
we have observed that the behavior of the critical temperature and the existence of compensation points strongly depend on interaction and concentration parameters .
particularly , we have found that the critical temperature of the model is independent on concentration of different types of spins at a critical @xmath52 value and the model has one or two compensation temperature points in a certain range of values of the concentration of the different spins .
we concluded that magnetic properties of the system ab@xmath0c@xmath1 can be controlled by changing the relative concentration of the different species of ions . as a result
, we would like to stress that these theoretical results can be very useful for designing molecular magnets in experimental studies since the existence of compensation in the ternary alloy ab@xmath0c@xmath1 that can be setup by adjusting the proportion of compounds . | in this study , we have considered the three dimensional mixed ferro - ferrimagnetic ternary alloy model of the type ab@xmath0c@xmath1 where the a and x ( x = b or c ) ions are alternately connected and have different ising spins s@xmath2=3/2 , s@xmath3=1 and s@xmath4=5/2 , respectively .
we have investigated the dependence of the critical and compensation temperatures of the model on concentration and interaction parameters by using mc simulation method .
we have shown that the behavior of the critical temperature and the existence of compensation points strongly depend on interaction and concentration parameters .
in particular , we have found that the critical temperature of the model is independent on concentration of different types of spins at a special interaction value and the model has one or two compensation temperature points in a certain range of values of the concentration of the different spins . |
qcd calculations predict that high energy partons traversing nuclear matter lose energy through induced gluon radiation @xcite .
recent measurements of inclusive charged hadron distributions in au+au collisions at = 130 gev show a suppression of hadron yields at high in central collisions relative to peripheral collisions and scaled nucleon - nucleon interactions @xcite , consistent with the picture of partonic energy loss in a dense system .
the fragmentation products of high energy partons that have propagated through the azimuthally asymmetric system generated by non - central collisions may exhibit azimuthal anisotropy due to energy loss and the azimuthal dependence of the path length @xcite .
elliptic flow @xcite at @xmath7 gev / c measured at = 130 gev reaches the values predicted by hydrodynamical models based on the assumption of complete local thermalization @xcite . at high @xmath6 ,
azimuthal distributions of charged hadrons exhibit a structure suggestive of fragmentation of hard scattered partons as well as an anisotropy with respect to the reaction plane @xcite .
the new star data on the azimuthal anisotropy parameter @xmath8 for au+au collisions at @xmath1=200 gev extend the measurements up to @xmath4 gev / c and provide important constraints on the underlying mechanisms of high @xmath6 hadron production in nuclear collisions .
the main tracking detector in star is the time projection chamber ( tpc ) situated in a solenoidal magnetic field of 0.5 t. the tpc has wide pseudorapidity @xmath9 and complete azimuthal coverage , with excellent momentum resolution .
charged hadrons are identified by measuring specific energy loss @xmath10 for @xmath11 gev / c .
@xmath2 and @xmath3 are identified via their @xmath12-decay topology .
a ring imaging cherenkov detector ( star - rich ) @xcite , covering @xmath13 and @xmath14 and situated on the exterior of the tpc , extends charged hadron identification to high @xmath6 . 2.7 m minimum - bias triggered au+au interactions at = 200 gev were used for this analysis .
the minimum - bias data correspond to @xmath15% of the geometric cross section .
all tracks used in this analysis passed within 1 cm of primary vertex and had at least 20 measured space points .
the azimuthal anisotropy of an event in momentum space is quantified by the coefficients of the fourier decomposition of the azimuthal particle distributions , with the second harmonic coefficient @xmath8 referred to as elliptic flow @xcite .
@xmath8 is inferred from the azimuthal particle distribution with respect to the estimated reaction plane orientation , corrected for the reaction plane resolution .
the track selection and calculation of the reaction plane are described in @xcite .
the reaction plane analysis integrates all possible sources of azimuthal correlations , including those unrelated to the orientation of the reaction plane .
a four - particle cumulant method @xcite for flow measurements reduces non - flow sources ( resonance decays , ( mini)jets , final state interactions , momentum conservation , etc . ) to a negligible level @xcite .
figure [ fig:4part ] shows the @xmath6-dependence of @xmath8 from the reaction plane method and l85 mm four - particle cumulant method .
the new results from the four - particle cumulant method have much higher statistical precision than previously @xcite for centralities 5 - 70% .
similar to the results at 130 gev @xcite , @xmath8 rises linearly for @xmath11 gev / c , but saturates for @xmath16 gev / c for both methods .
the @xmath8 values obtained from the four - particle cumulant method are 20% lower than those from the reaction plane analysis .
this difference is attributed to the non - flow correlations in the reaction plane analysis , and we assign a one - sided 20% systematic uncertainty for the reaction plane @xmath8 values presented below . figure [ fig : ratio ] , upper panel , shows @xmath0 of charged hadrons obtained with the reaction plane analysis for minimum - bias collisions at 130 and 200 gev .
the functional form of @xmath0-dependence is similar at the two energies .
figure [ fig : ratio ] , lower panel , shows the ratio of @xmath8 from the 200 gev and 130 gev datasets . the elliptic flow is stronger at 200 gev for @xmath11 gev / c , suggesting a higher degree of thermalization . of charged particles for different centralities at 200 gev.,width=294 ] of charged particles for different centralities at 200 gev.,width=294 ] in the saturation region at high @xmath6 , the magnitude of the azimuthal anisotropy is the same to within 5% , in contrast to the growth of the inclusive cross section with increasing @xcite .
this is an indication of the geometrical origin of the large anisotropies observed in the hard scattering region .
figure [ fig : charged ] shows the @xmath0-dependence for different collision centralities for @xmath17 gev / c .
@xmath8 remains finite for non - central collisions , exhibiting a decrease from the saturation level at the highest measured @xmath6 for the more central events .
it is expected that the azimuthal anisotropy will vanish in the limit of very high @xmath6 .
@xmath8 values measured at @xmath18 gev / c for a restricted centrality range , corrected for contribution from non - flow effects , are consistent with the maximum expected @xmath8 from surface emission @xcite .
a proposed explanation of the saturation of @xmath8 at transverse momenta @xmath19 gev / c is the dominance of baryons in this @xmath6 region described by non - perturbative mechanisms such as baryon junctions or hydro , while pion production is dominated by the quenched pqcd spectra @xcite . a qualitatively different @xmath0 behavior is therefore predicted for baryons and mesons .
figure [ fig : rich ] compares @xmath8 of identified charged particles with pure hydrodynamical calculations @xcite for @xmath20 gev / c . in figure
[ fig : lambda ] , a similar comparison is presented for @xmath2 and of @xmath2 and @xmath3 for minimum - bias events at = 200 gev , compared to hydro calculations.,width=321 ] of @xmath2 and @xmath3 for minimum - bias events at = 200 gev , compared to hydro calculations.,width=321 ] @xmath3 . at @xmath11 gev / c ,
the hydro calculations describe the mass dependence of the elliptic flow well . at @xmath21 gev / c , @xmath8 of baryons
may slightly exceed that of mesons @xcite .
star has measured azimuthal anisotropies as a function of centrality in au+au collisions at = 200 gev of charged hadrons for @xmath17 gev / c and identified hadrons for @xmath22 gev / c .
the experimental observation of large azimuthal anisotropies at high transverse momenta is the focus of ongoing theoretical investigation .
many different approaches have been attempted to describe the data , such as combining hydrodynamical elliptic flow with perturbative qcd including jet quenching @xcite , parton cascade @xcite , minijets @xcite , and surface emission @xcite . a simultaneous quantitative description of all the experimental data on high @xmath6 production at rhic , including inclusive spectra suppression , saturation of @xmath8 , and disappearance of back - to - back jets in central collisions @xcite , is needed . however , the data clearly point in the direction of qualitatively new physics phenomena in high @xmath6 particle production in nuclear collisions at rhic energies .
99 m. gyulassy and m. plumer , phys .
b * 243 * , 432 ( 1990 ) ; x .-
n . wang and m. gyulassy , , 1480 ( 1992 ) .
m. gyulassy and x .-
wang , nucl . phys .
b * 420 * , 583 ( 1994 ) ; r. baier , y. l. dokshitzer , s. peigne and d. schiff , phys .
b * 345 * , 277 ( 1995 ) .
phenix collaboration , k. adcox , , 022301 ( 2002 ) .
star collaboration , c. adler , nucl - ex/0206011 , to appear in phys .
j. l. klay for the star collaboration , these proceedings .
wang , , 054902 ( 2001 ) ; m. gyulassy , i. vitev and x .-
wang , , 2537 ( 2001 ) .
ollitrault , , 229 ( 1992 ) .
star collaboration , k. h. ackermann , , 402 ( 2001 ) .
star collaboration , c. adler , , 182301 ( 2001 ) .
star collaboration , c. adler , nucl - ex/0206006 .
star - rich and star collaborations , b. lasiuk , nucl .
a * 698 * , 452c ( 2002 ) .
s. voloshin and y. zhang , z. phys .
c * 70 * , 665 ( 1996 ) ; a. m. poskanzer and s. a. voloshin , , 1671 ( 1998 ) . n. borghini , p. m. dinh , and j .- y .
ollitrault , , 054901 ( 2001 ) .
star collaboration , c. adler , , 034904 ( 2002 ) .
p. huovinen , phys .
b * 503 * , 58 ( 2001 ) .
e. v. shuryak , , 027902 ( 2002 )
. m. gyulassy , i. vitev , x .-
wang , and p. huovinen , phys . lett .
b * 526 * , 301 ( 2002 ) .
d. molnar and m. gyulassy , nucl .
a * 697 * , 495 ( 2002 ) .
y. v. kovchegov and k. l. tuchin , nucl .
a * 708 * , 413 ( 2002 ) .
d. h. hardtke for the star collaboration , these proceedings . | we report new results on @xmath0 for au+au collisions at @xmath1=200 gev for charged hadrons , pions , kaons , ( anti)protons , @xmath2 , and @xmath3 .
the analysis is extended to @xmath4 gev / c for charged hadrons and @xmath5 gev / c for identified particles . a comparison of the azimuthal anisotropy of charged hadrons measured at = 130 and 200 gev is presented .
the @xmath6-dependence of baryon versus meson elliptic flow is discussed . |
we first briefly describe the natural time analysis of seismicity @xcite . in a time series comprising @xmath5 earthquakes
, the natural time @xmath6 serves as an index for the occurrence of the @xmath7-th earthquake . in natural time analysis
the pair @xmath8 is studied , where @xmath9 is the energy released during the @xmath7-th earthquake of magnitude @xmath10 .
alternatively , one may study the pair @xmath11 , where @xmath12 is the normalized energy released during the @xmath7-th earthquake , and @xmath9 -and hence @xmath13- is estimated through the relation @xcite @xmath14 .
the variance of @xmath1 weighted for @xmath13 , designated by @xmath4 , is given by @xcite @xmath15 this quantity can be also considered as an order parameter for seismicity@xcite . in order to study the fluctuations of @xmath4
, we apply the following procedure @xcite . taking an excerpt of a seismic catalog comprising @xmath16 successive events ,
we start from the first eq and calculate the first 35 @xmath4 values for 6 to 40 consecutive eqs ( cf . the upper limit of this number does not markedly influence the results @xcite ) .
then we proceed to the second eq , and calculate again 35 values of @xmath4 using the 7@xmath17 to the 41@xmath18 event .
scanning , event by event , the whole excerpt of @xmath19 earthquakes , we calculate the average value @xmath20 and the standard deviation @xmath21 of the @xmath4 values .
the quantity @xmath22 is defined@xcite as the variability @xmath23 of @xmath4 for this excerpt of length @xmath19 .
usually we are interested on what happens to the @xmath23 value until just before the occurrence of each eq , @xmath24 , in the seismic catalog . to achieve this goal
, we calculate first the @xmath4 values using the _ previous _ 6 to 40 ( or 6 to @xmath19)@xcite consecutive eqs .
these 35 @xmath4 values are associated with the eq @xmath24 , but we clarify that the eq @xmath24 has _ not _ been employed for their calculation .
the @xmath23 value -corresponding to the eq @xmath24- for a natural time window length @xmath19 is computed using all the ( @xmath25 ) @xmath4 values associated with the eqs @xmath26 to @xmath24 .
the resulting value is denoted by @xmath27 , where the subscript @xmath19 shows the natural time window length , and the corresponding minimum is designated by @xmath28 .
we now summarize a few important points that emerged in our earlier studies@xcite , reviewed also in chapter 6 of ref . , when analyzing in natural time the _ long term
_ seismicity in wide areas . by using a natural time window comprising 6 to 40 consecutive events sliding through an earthquake catalog ,
the computed @xmath4 values lead to the construction of the probability density function ( pdf ) p(@xmath4 ) of a wide area .
the scaled distribution @xmath29 p(@xmath4 ) plotted versus ( @xmath30 ) of this area collapses on the same curve with the ones deduced from different wide areas .
for example , the curve obtained from the area @xmath31 in japan using the jma catalog , and the area @xmath32 in california using the southern california earthquake catalog ( scec ) @xcite .
it has been also found that this _
universal _ curve of long term seismicity exhibits strikingly similar features with the corresponding scaled distribution curves of the order parameter fluctuations in other -nonequilibrium or equilibrium- critical systems@xcite including the data@xcite we analyzed@xcite for the case of self organized criticality@xcite .
this similarity was focused on the existence of a common `` exponential tail '' ( which has in fact a double exponential form ) reflecting non - gaussian ( large , but rare ) fluctuations .
in this study we present examples of major eqs that occurred in california .
we used the united states geological survey northern california seismic network catalog available from the northern california earthquake data center , at the http address : www.ncedc.org/ncedc/catalog-search.html , hereafter called ncedc .
the earthquake magnitudes reported in this catalog are hereafter labelled with @xmath0 .
we considered all earthquakes with @xmath33 reported by ncedc , within the area @xmath34 ( the green rectangle in fig .
[ fig1 ] ) .
we have on average @xmath35 eqs per month since 31832 earthquakes occurred for the 25 year period from 1 january 1979 to 1 january 2004 . following ref . , we focus hereafter on the natural time window length @xmath36300 , which is compatible with the fact that the lead time of ses activities is around a few months ( i.e. , from 21 days to an upper limit of around 5 months , e.g. , see ref . ) . by analyzing in natural time the aforementioned ncedc data , we recently identified @xcite , for @xmath19=300 events , the dates of the minima @xmath28 as being 1 to 5 months before four out of the five mainshocks with @xmath37 that occurred within the wide area @xmath38 during the 25 year period 1 january 1979 to 1 january 2004 in california .
for the two examples of the @xmath07.4 landers eq on 28 june 1992 and the @xmath07.0 hector mine eq on 16 october 1999 that will be treated below , the dates of the precursory minima @xmath39 were reported to be 28 january 1992 and 14 may 1999 , respectively .
briefly , the method proposed in based on the following main aspect : we argue that a minimum should be observed almost simultaneously in the order parameter fluctuations of the seismicity within _ both _ the wide area and the epicentral area of the impending mainshock .
this stems from our findings in ref . where we identified the appearance of a @xmath28 ( at scales @xmath19=100 to @xmath19=400 ) in the seismicity within the wide area @xmath31 , i.e. , @xmath40 , in japan around the date ( i.e. , on 26 april 2000 ) of the initiation of the ses activity recorded@xcite at niijima island -in the vicinity of which a major seismic swarm started 2 months later with a @xmath06.5 eq on 1 july 2000 .
moreover , it was found that in areas surrounding niijima island the variability @xmath23 of the seismicity exhibited a minimum almost at the _ same _ date , i.e. , around 26 april .
hence , in the light of the above findings , the method to estimate the epicentral area of an impending mainshock -when no information for the recording of an ses activity is available- should consist of the following steps : first step : we start the investigation from the natural time analysis of the seismicity in a wide area , for example @xmath41 ( i.e. , @xmath42 ) in california . by considering a sliding natural time window comprising a number of events that would occur in a few months ( e.g. ,
we may adopt , as in ref . , @xmath19=300 events for the aforementioned wide area in california ) , we search for a minimum in the @xmath23 variability . 7.0 hector mine eq in california on 16 october 1999 .
( * a * ) the variability @xmath23 of the seismicity versus the conventional time for a natural time window length @xmath19 comprising a number of events that would occur in three months : the thick black curve minimizing on 14 may 1999 corresponds to the wide area @xmath41 .
the other eight curves correspond to those areas -out of the 38 encircled ones in fig .
[ fig1]- resulting in @xmath23 minima which occur at dates close to that of the @xmath43 minimum of the wide area .
these dates for the lower three minima are marked in ( * b * ) and for the remaining five minima in ( * c * ) .
( * b * ) : after discarding the red curve ( see the text ) , the cyan curve exhibits the lowest minimum on 21 may 1999 .
it corresponds to the area @xmath44 shown in fig .
[ fig1 ] by the cyan rectangle along with the hector mine eq epicenter marked with the neighboring red triangle ( the smaller ) .
the eqs with @xmath45 ( right scale ) are shown with vertical bars ending at solid circles . ]
second step : we investigate whether this minimum is a precursory one by examining whether it obeys the criteria mentioned in ref . .
this should necessarily include the detrended fluctuation analysis ( dfa)@xcite in order to investigate temporal correlations in the earthquake magnitude time series at the same natural time window length scale as the one mentioned above ( cf . it has been shown @xcite that @xmath23 captures the correlations between the events ) .
dfa has become the standard method when studying long - range correlated time series and can also be applied to real world non - stationary signals @xcite .
third step : we search for the epicentral area of the impending mainshock by using an area window of an appreciably smaller size , e.g. , @xmath46 for the cases studied in california , sliding ( with steps of @xmath47 in longitude and/or @xmath47 in latitude , see the green plus symbols in fig . [ fig1 ] ) through the wide area . for each location of the sliding area window , we repeat the analysis as in the wide area in order to determine the date of @xmath28 . note that , if in the aforementioned wide area of california ( in which , as mentioned , occur on average 100 events / month ) we used @xmath19=300 events , we should consider as @xmath19 in each location of the sliding area window the corresponding number of the events that would occur on average within three months .
this reflects that we can not select an area window of arbitrarily smaller size in view of the following limitation : in order to identify the date of the occurrence of the minimum of @xmath23 in an area window size with an uncertainty of around a few days , we must have at least on the average 2 events per week , i.e. , at least 24 events for an almost three month period ( 12 weeks ) . among these locations of the sliding area window
investigated , we select the one(s ) which led to a @xmath28 date which almost coincides with the date at which the @xmath39 value was observed in the wide area . in practice
, the former date differs from the latter usually by no more than a few days up to around 10 days , or so .
this way we usually find an area surrounding the future epicenter , as it results from the examples presented in the following two sections , which refer to major eqs that occurred in california .
since a major eq occurs , as mentioned , 1 to 5 months after the appearance of @xmath39 in the wide area @xcite , the applicability of the method proposed here becomes more clear when a single major mainshock occurs within a five month period .
such an example , is the case of the @xmath07.0 hector mine eq in california in 1999 .
furthermore , in a later section , we shall present a case in which two major mainshocks occurred inside the wide area within 5 months .
this is the case of the @xmath07.4 landers eq on 28 june 1992 that was preceded by the @xmath06.7 ( @xmath487.1 , e.g. see ref . )
cape mendocino eq on 25 april 1992 . as we shall see later
, it is strikingly interesting that in the latter case the method identifies two different epicentral areas as it should .
we clarify that in the next two sections , we solely focus on the third step of the method since the first two steps -referring to the identification in the wide area of the dates of the minima @xmath39 for major eqs in california and their distinction from the non precursory minima- have been already accomplished , as mentioned , in ref . .
to summarize : for the analysis of the examples presented below for california , we used the wide area @xmath38 as in ref.@xcite , and a sliding area window size @xmath46 ( with step @xmath47 ) resulting in 156 areas ( see the plus symbols in fig . [ fig1 ] ) out of which 38 areas ( encircled in fig .
[ fig1 ] ) had at least 24 events on average per three months .
hence , the investigation of the examples below in california was made by analyzing the seismic data occurring in these 38 areas .
the thick black curve in fig . [ fig2]a depicts the variability @xmath23 versus the conventional time when a natural time window length @xmath19=300 events is sliding through the wide area @xmath38 .
a clear minimum of @xmath23 is observed on 14 may 1999 , as marked in the figure .
the subsequent analysis of the 38 areas -out of the 156 depicted in fig .
[ fig1]- resulting from the locations of the sliding window of size @xmath46 lead to 38 curves . eight of these curves , depicted in fig .
[ fig2]a , were identified to exhibit a @xmath23 minimum with dates within 10 days , or so , from the aforementioned date on 14 may 1999 obtained from the wide area .
these dates are now marked , for reader s convenience , in two separate panels of this figure : in fig .
[ fig2]b -for the three lowest minima observed in fig .
[ fig2]a- and in fig .
[ fig2]c for the minima of the remaining five curves
. a closer inspection of fig .
[ fig2]b shows that the red curve in the bottom ( that is obtained from the area @xmath51 ) has a plateau extending from 8 may until 5 june which in fact comes from the absence of the occurrence of any event in the area during this almost one month period .
thus , this curve should be discarded from further analysis since we can not derive a definite conclusion on the existence or not of a minimum close to the aforementioned date 14 may 1999 . between the remaining two curves ,
the cyan has a minimum which is somewhat lower and occurs on 21 may 1999 .
it corresponds to the area @xmath44 depicted with the cyan rectangle in fig .
[ fig1 ] along with the epicenter of the hector mine eq ( the smaller red triangle ) lying at @xmath49n @xmath50w .
the epicenter is located just at the eastern edge of the area , see the cyan rectangle in fig .
the other curve , i.e. , the magenta in fig.[fig2]b , corresponds to the area @xmath52 , see the magenta rectangle in fig .
[ fig1 ] , which is only @xmath53 far from the hector mine eq epicenter . as for the five curves in fig.[fig2]c they correspond to the areas
@xmath54 , @xmath55 , @xmath56 , @xmath57 and @xmath58 lying at various distances ( roughly between @xmath47 and @xmath59 ) from the hector mine eq epicenter .
none of these areas , is adjacent to the epicenter as the aforementioned one @xmath60 which corresponds to the cyan curve exhibiting the lowest @xmath23 minimum in figs . [ fig2]a , b .
thus , in short , the cyan curve exhibiting a @xmath23 minimum on 21 may , i.e. , around a week later from that ( 14 may ) identified in the wide area ( thick black curve ) , which happens to be the lowest one among the other curves showing also minima at neighboring dates , corresponds to an area whose eastern edge lies just at the hector mine eq epicenter .
this eq , which is the largest moment magnitude ( @xmath63 ) eq that occurred in california since 1992 , has been preceded by the @xmath06.7 ( @xmath487.1 ) cape mendocino eq on 25 april 1992 with an epicenter at @xmath64n @xmath65w ( followed by two strong @xmath06.45 and @xmath06.6 aftershocks on 26 april 1992 ) .
hence , landers eq and cape mendocino eq occurred within almost 3 months being less than the maximum lead time between @xmath39 and the subsequent eq which is around 5 months @xcite .
the complexity of this case might be the origin of the following fact : the variability of the seismicity versus the conventional time in the wide area @xmath38 , depicted in fig .
[ fig3 ] with a thick black curve , although exhibited its lowest value on 28 january 1992 ( marked in fig .
[ fig3]a , its minimum is in fact very broad ( like a plateau ) lasting for about 4 weeks , i.e. , from 18 january to around 20 february , 1992 .
this must be carefully considered when the area window of @xmath46 is sliding ( with step @xmath47 ) through the aforementioned wide area .
we found that , among the 38 areas studied , eight of them led to @xmath23 minima with dates from 2 february to 19 february 1992 , thus lying within the broad range mentioned above for the @xmath23 minimum of the wide area .
the eight curves corresponding to these areas are plotted in fig .
[ fig3]a and can be classified into two types , which are depicted separately in figs .
[ fig3]b and [ fig3]c where we also mark the date of the minimum in each curve .
first , in three of these curves shown with magenta squares , black triangles and cyan circles , which correspond to the areas @xmath66 , @xmath67 and @xmath44 , a very sharp minimum appears on 19 february 1999 , see fig .
an inspection of these three areas shows that the epicenter of landers eq in fig .
[ fig1 ] ( large red triangle ) lies inside the latter area ( depicted with a cyan rectangle in fig .
[ fig1 ] ) , @xmath47 to the east of the first area and around @xmath68 to the southeast of the second area .
second , the remaining five curves , see fig .
[ fig3]c , shown in orange circles , blue squares , red crosses , thin black line and green asterisks ( starting from the curve with the lowest @xmath23 minimum ) , exhibit non - sharp minima on the following dates 2 , 2 , 2 , 16 and 17 february , 1992 , and correspond to the following five areas : @xmath69 , @xmath70 , @xmath71 , @xmath72 and @xmath73 , respectively .
the first area is depicted with the orange rectangle in fig.[fig1 ] , where we see that the epicenter of the cape mendocino eq marked by a red triangle -inside the orange rectangle- along with the neighboring two smaller red triangles showing its two aftershocks@xcite , lies inside it ( cf .
the brown rectangle depicted in fig , [ fig1 ] also results as a candidate epicentral area for the cape mendocino eq upon considering a reasonable experimental error in the determination of the magnitudes .
concerning the other areas , we find that the cape mendocino epicenter lies @xmath74 to the west of the second area , @xmath74 to the north from the third and the fourth , and around @xmath75 to the nw of the fifth area
. let us summarize : we found that eight curves minimize at dates within the broad minimum ( 18 january - 20 february ) exhibiting by the thick black curve corresponding to the wide area
. among these curves , three minimize on 19 february 1992 and five at earlier dates .
the former three curves correspond to areas related with landers eq since one of them is surrounding its epicenter , and the other two are neighboring areas , i.e. , @xmath47 and @xmath68 far from the epicenter .
the remaining five curves correspond to areas related with the cape mendocino eq since one of them ( exhibiting the lowest minimum ) is surrounding its epicenter , and the other four lie in its vicinity .
the prediction of the epicenter of an impending mainshock can be made on the basis of the ses physical properties , as mentioned in the introduction .
this of course has two prerequisites : first , that geoelectrical data are available and second , that we know from earlier experience the selectivity map @xcite of the measuring station at which the ses under investigation appeared .
the method proposed here can not be falsely considered as replacing that followed by means of ses since it does not improve its accuracy , which is of the order of 100 km , but it has the privilege that can be applied even in the lack of geoelectrical data . in conclusion
, here we proposed a method to estimate the epicentral area of an impending mainshock based on a precursory minimum observed in the fluctuations of the order parameter of seismicity .
this has been successfully applied to mainshocks in california including the landers eq on 28 june 1992 , which is the largest earthquake that occurred in california during the last three decades .
these results are found to be robust when repeating 100 trials upon considering a reasonable experimental error in the magnitude measurements of the earthquakes in the seismic catalog used .
all the seismic data taken from ncedc have been analyzed in natural time .
the seismic moment @xmath76 , which is proportional to the energy release during an earthquake and hence to the quantity @xmath9 used in natural time analysis , is calculated @xcite from the relation @xmath77const . | upon employing the analysis in a new time domain , termed natural time , it has been recently demonstrated that a remarkable change of seismicity emerges before major mainshocks in california .
what constitutes this change is that the fluctuations of the order parameter of seismicity exhibit a clearly detectable minimum .
this is identified by using a natural time window sliding event by event through the time series of the earthquakes in a wide area and comprising a number of events that would occur on the average within a few months or so . here ,
we suggest a method to estimate the epicentral area of an impending mainshock by an additional study of this minimum using an area window sliding through the wide area .
we find that when this area window surrounds ( or is adjacent to ) the future epicentral area , the minimum of the order parameter fluctuations in this area appears at a date very close to the one at which the minimum is observed in the wide area .
the method is applied here to major mainshocks that occurred in california during the recent decades including the strongest one , i.e. , the 1992 landers earthquake
. it may be considered ( e.g. refs . )
that earthquakes are non - equilibrium critical phenomena .
their complex correlations in time , space and magnitude @xmath0 have been investigated by various procedures ( e.g. , see refs .
@xcite ) including a very recent development @xcite of the complex networks approach . in this study
, we employ the analysis in a new time domain , termed natural time@xcite , @xmath1 , since it allows us to identify@xcite when a system approaches a critical point ( for a review see ref . ) .
seismic electric signals ( ses ) are low frequency ( @xmath2hz ) electric signals @xcite preceding earthquakes .
they are emitted from the future focal region probably through the following mechanism @xcite : when in the focal region the stress reaches a _ critical _
value @xmath3 , a cooperative orientation of the electric dipoles ( that have been formed due to defects@xcite ) occurs , which leads to the emission of a transient electric signal that constitutes an ses .
several such signals within a short time are called ses activity @xcite . by combining the ses physical properties , the epicenter and
the magnitude of the impending mainshock can be determined@xcite .
furthermore , by making use of the small earthquakes subsequent to the initiation of an ses activity and analyzing them in natural time ( see below ) , the occurrence time of an impending mainshock can be identified as follows : we compute the value of the order parameter @xmath4 of seismicity ( see below ) and then find@xcite that a mainshock occurs in a few days to one week after the @xmath4 value approaches 0.070 . by applying this procedure ,
several successful predictions have been issued @xcite in greece .
the challenging question raises , however , what can we do when geoelectrical data are lacking which was the case before the occurrence of the most strong earthquakes ( eqs ) worldwide during the recent decades .
it is the main objective of this paper to attempt answering this question motivated by an important fact which has been recently reported @xcite : the initiation of an ses activity is accompanied by a clearly detectable change in seismicity which constitutes an independent geophysical dataset of a different physical observable .
in particular , by analyzing in natural time the time series of earthquakes , we found that the fluctuations of the order parameter of seismicity exhibited a clearly detectable minimum approximately at the time of the initiation of the ses activity .
furthermore , we showed @xcite that these two phenomena ( ses activity , minimum of the order parameter fluctuations of seismicity ) beyond their co - location in time they were also linked closely in space . here , we take advantage of these recent findings and show the following : since the natural time analysis of the seismic data alone can lead to the identification of the minimum of the fluctuations of the order parameter of seismicity that precedes a major eq , an adequate study of the spatial distribution of those events resulting in this precursory minimum enables an estimate of the area in which the major eq will occur
. we shall present applications including the strongest eq that occurred in california ( see fig .
[ fig1 ] ) during the last three decades .
analogous applications have been completed for major eqs in japan@xcite . |
exclusion statistics was proposed @xcite as a phenomenological description of excitations with mixed statistical properties ( intermediate between fermions and bosons ) in many - body systems .
it is based on the assumption that the change of the number of available one - particle states in the system , @xmath0 , at a given volume and at fixed boundary conditions depends _ linearly _ on the change of the number of quasiparticles @xmath1 : @xmath2 . here
@xmath3 is the statistical parameter which is assumed to be a rational number , with @xmath4 and @xmath5 corresponding to the normal bosonic and fermionic statistics , respectively .
this generalized pauli s principle results in a @xmath3-dependent statistical weight for the ideal gas of quasiparticles ( usually referred as _
exclusons _ ) .
the standard methods of statistical quantum mechanics allow one to derive the equilibrium distribution function of exclusons @xcite @xmath6^{-1 } , \label{eq : distribution}\ ] ] where @xmath7 is the temperature ( we consider units where @xmath8 ) , @xmath9 is the chemical potential of the ideal gas , and @xmath10 is the real positive solution of the algebraic equation @xmath11 formally , exclusons can be regarded as composites of fermions and bosons since their thermodynamic potential can be decomposed into a sum of fermionic and bosonic potentials weighted by fractional coefficients,@xcite namely , @xmath12 \nonumber \\ & = & g\omega_f + ( 1-g)\omega_b , \label{eq : thermopot}\end{aligned}\ ] ] where @xmath13 denotes the grand thermodynamic potential for exclusons , fermions , and bosons , correspondingly , and @xmath14 . in spite of its phenomenological footing ,
exclusion statistics proved to be correct for quasiparticles in certain exactly solvable 1d models @xcite and , more importantly , it is the statistics of recently observed fractionally charged excitations in fractional quantum hall systems,@xcite with the statistical parameter determined by the landau level filling factor .
already for two decades , fractional charge is a theoretically well - grounded concept , both in quantum field theory and in solid state physics ( for reviews , see ref . ) .
it is known @xcite that the requirement of gauge invariance forces the gas of fractionally charged particles at equilibrium to obey fractional statistics . from a phenomenological point of view , exclusion statistics is a reasonable candidate for the required anomalous statistics of fractionally charged quasiparticles .
thus , the study of thermodynamic and transport properties of _ fractionally _ charged exclusons can be more than simply an academic exercise . in ref .
the persistent current of a 1d ring of fractionally charged exclusons was calculated .
it was shown that at low temperatures the fractional charge , masked by fractional statistics , does not show up in the aharonov - bohm oscillations . however
, the `` high - temperature '' properties of the system do depend on fractional charge .
unfortunately , this region is hardly accessible to present experiments since the oscillating current itself is exponentially small .
the purpose of the present paper is to go beyond thermodynamics and consider transport properties of exclusons .
we show within the landauer - bttiker formalism that the transport coefficients for exclusons are determined by the same general formulae as for the particles with normal statistics , provided the leads representing particle reservoirs are also described by exclusion statistics . in this case , even for an impurity - free 1d wire the transport coefficients depend on the statistical parameter .
it is argued however that for an ideal wire the above dependence disappears when one considers `` normal '' particle reservoirs .
let us consider an effectively infinite 1d wire of exclusons with a scattering potential described by the transmission probability @xmath15 .
we will assume at first that the wire is connected to reservoirs of _ exclusons _ in equilibrium at temperature @xmath7 and chemical potential @xmath9 .
in this case the generalization of landauer s formula for conductance is straightforward : for exclusons with fractional charge @xmath16 ( @xmath17 is an integer ) and statistical parameter @xmath18 it reads @xmath19 , \label{eq : conductance}\ ] ] where @xmath20 is the conductance quantum ( @xmath21 is the spin of exclusons ) and the distribution function is defined by eqs .
( [ eq : distribution ] ) and ( [ eq : implicit ] ) . to proceed we have to specify the properties of the scattering potential in the vicinity of the fermi energy .
if transport of exclusons through the barrier is nonresonant , the transmission probability at low temperatures ( @xmath22 ) can be approximated by an energy independent factor @xmath23 .
then , eq . ( [ eq : conductance ] )
is reduced to a simple formula @xmath24 in particular , for a perfect wire ( @xmath25 ) the conductance of fractionally charged exclusons @xmath26 coincides with that for a homogeneous luttinger liquid ( ll ) of integer charge particles with correlation parameter @xmath18 .
however , in distinction from the ll case , exclusons are supposed to be noninteracting particles and therefore their scattering by a local potential can be described by a non - renormalized transmission probability @xmath23 .
thus the conductance is finite and temperature independent up to temperatures of the order of @xmath9 . here
a clarifying comment is needed .
we are considering a gas of exclusons in a grand canonical ensemble , so @xmath9 is an input parameter , which is assumed to be positive .
it means that for any @xmath27 there exists a low temperature region where @xmath22 . yet , for @xmath28 the chemical potential @xmath29 can also be small and the only region realizable is the `` high temperature '' one @xmath30 . in this case
the conductance will be sensitive to the low energy dependence of the transmission coefficient . for a perfect wire ( @xmath25 ) the corresponding conductance will read @xmath31 . in what follows
we will assume that even for large ( but finite ) @xmath17 the excluson chemical potential is still the largest energy scale in the problem .
the case of resonant tunneling of exclusons reveals far more interesting transport properties than the homogeneous case . to show why
, we will approximate the resonant transmission coefficient @xmath15 at energies in the vicinity of the fermi level ( @xmath32 ) by the breit - wigner form @xmath33 where @xmath34 and @xmath35 are the position and the width of the resonance level , respectively .
the resonance tunneling implies that @xmath36 and the corresponding conductance at @xmath37 coincides ( at it should be ) with that for a perfect channel . at finite temperature
the resonant conductance takes the form @xmath38 .
\label{eq : resonantcond}\ ] ] the above expression determines the temperature dependence of the resonance peak height at @xmath22 .
let us place the question - how does the peak height depend on the statistical parameter @xmath18 ?
if @xmath3 is not small the resonance conductance is qualitatively the same as that for fermions ( @xmath39 ) , @xmath40 where @xmath41 \label{eq : a_m}\ ] ] and @xmath42_{x=0}. \label{eq : b_m}\ ] ] these coefficients can be evaluated as follows . with the help of eq .
( [ eq : implicit ] ) the derivative of the distribution function can be expressed in the form @xmath43}{[g+y(x)]^3 } \label{eq : derivative}\ ] ] inserting this expression into eq .
( [ eq : a_m ] ) and changing the integration variable from @xmath44 to @xmath45 one finds for any @xmath27 that the coefficient @xmath46 takes the same value as for fermions , @xmath47 ^ 2 } { ( g + y)^2 } = \frac{\pi^2}{3}. \label{eq : a_mindep}\ ] ] this is a remarkable fact - the coefficient @xmath46 does not depend on the statistical parameter . in the next section
it will be shown that this very coefficient determines the heat conductance of exclusons .
the numerical coefficient @xmath48 on the other hand , does depend on @xmath3 and takes the values : @xmath49 , @xmath50 , @xmath51 @xmath52 . for decreasing values of the statistical parameter the temperature dependence of the resonance conductance acquires new qualitative features .
namely , the crossover from low - t to high - t behavior is transformed into a wide region @xmath53 where @xmath54 for @xmath55 depends almost _ linearly _ on temperature ( see fig.1 ) . to study this region
analytically it is tempting to use the bosonic limit . by expanding the distribution function of exclusons in the vicinity of the bosonic statistics ( @xmath4 )
it is easy to show that first two terms of @xmath56-expansion of resonant conductance , eq .
( [ eq : resonantcond ] ) , take the form @xmath57\right\ } , \label{eq : intermed}\ ] ] where @xmath58 and @xmath59 ( here @xmath60 is the gamma function@xcite ) . for finite
@xmath17 the above expression holds formally until the term in the square brackets is much smaller than @xmath17 .
this restricts the temperature region where eq .
( [ eq : intermed ] ) could be at least qualitatively correct to temperatures @xmath61 . according to eq .
( [ eq : intermed ] ) , at low temperatures ( @xmath62 ) the resonance conductance is still determined by the low-@xmath7 asymptotics of eq .
( [ eq : asymptotics ] ) .
however , for @xmath55 there is a wide temperature region @xmath63 where conductance depends _ linearly _ on temperature , @xmath64 .
it is only at temperatures @xmath65 that the resonance conductance switches to the ordinary high-@xmath7 asymptotics .
notice that eq .
( [ eq : intermed ] ) is not the true asymptotic expansion because all higher terms in @xmath56 are represented by divergent integrals .
the formula is not quantitatively reliable and can be regarded at most as an estimate .
we presented it only to show how the intermediate linear dependence on temperature appears in the crossover region of eq .
( [ eq : asymptotics ] ) .
it is worth noticing that the widening of the crossover region is a specific feature of the exclusion statistics for resonant tunneling .
the above discussed linear-@xmath7 regime is pronounced only for very small values of statistical parameter @xmath3 .
the considerations above dealt with the transport of fractionally charged quasiparticles through a wire connected to reservoirs containing the particles of the same charge and statistics . although theoretically conceivable , such situation is difficult to realize experimentally . a more natural experimental setup would be a 1d system of exclusons connected to leads containing noninteracting electron . does the conductance of a perfect `` excluson wire '' connected to electron reservoirs depend on statistical parameter @xmath3 ? to solve this problem honestly one needs to introduce a microscopical model of fractional charge transport . in the phenomenological approach developed above it is reasonable to consider the case when the `` excluson wire '' is a metallic state characterized by a fermi energy @xmath66 .
when an electron passes through the ( presumably perfect ) boundary to the `` excluson wire '' it is converted into @xmath17 particles with fractional charge @xmath67 .
therefore , the fractional charge in the expression for the quasiparticle current will be canceled by the extra factor @xmath17 in the density of exclusons . at low temperatures and provided that there is no backscattering on the boundaries , a voltage applied to the electron reservoirs will result in a current through the system which will not depend on the fractional charge .
formally , the above statement can be proved by mapping the excluson model onto a ll model @xcite ( the mapping is exact at @xmath37 and for the impurity - free case ) and by recalling that the conductance of a ll constriction connecting to reservoirs of noninteracting electrons is @xmath20 .
in the previous section we showed that the landauer s formula for conductance can be applied to particles with exclusion statistics . could we expand this claim to all transport coefficients ?
since this is not evident from a general point - of - view , we at first derive the expressions for heat conductance @xmath68 and thermoelectric cross coefficients @xmath69 using the approach developed in ref . .
knowing the explicit expression for the grand thermodynamic potential of exclusons , eq .
( [ eq : thermopot ] ) , it is easy to show that the entropy of the ideal gas of exclusons takes the form @xmath70 \ln[1-gf_g(\varepsilon_j ) ] \right .
\nonumber \\ & & \left . - [ 1+(1-g)f_g(\varepsilon_j ) ] \ln[1+(1-g)f_g(\varepsilon_j ) ] \right\ } , \label{eq : entropy}\end{aligned}\ ] ] where @xmath71 is the distribution function of exclusons obtained from eqs .([eq : distribution ] ) and ( [ eq : implicit ] ) . in the limiting cases of @xmath5 and @xmath4 , eq .
( [ eq : entropy ] ) transforms into a well - known formula for the entropy of the ideal fermion and boson gases , respectively . with the entropy of exclusons at hand one can ( following ref . )
perform all the transformations leading to the expressions for the heat conductance and the coefficient @xmath69 . using eqs .
( [ eq : distribution ] ) and ( [ eq : implicit ] ) one can show that the desired quantities are determined by the same general formulae as for ordinary particles,@xcite @xmath72 .
\label{eq : thermocond}\ ] ] a similar expression [ but linear on @xmath73 stands for the thermoelectric coefficient @xmath74 .
the analysis of eq .
( [ eq : thermocond ] ) is straightforward . at low temperatures , @xmath22
, one can replace the transmission probability by the energy - independent factor @xmath23 .
then , the heat conductance reads @xmath75 where the numerical factor @xmath46 is defined by eqs .
( [ eq : a_m ] ) and ( [ eq : a_mindep ] ) . since the above coefficient
does not depend on the statistical parameter , the low-@xmath7 heat transport of exclusons is determined by the same formulae as for free fermions .
analogously , the thermoelectric coefficient of exclusons in the limit @xmath76 vanishes irrespective of the value of @xmath3 , @xmath77 we remark that for bosonic statistics the limit @xmath78 in eq .
( [ eq : thermocond ] ) should be taken before @xmath79 .
for the case of neutral bosons ( @xmath80 ) the resulting heat conductance is sensitive to the energy dependence of the transmission probability at low energies . for a perfect channel ( @xmath25 ) and spinless particles ,
we obtain @xmath81 which again coincides exactly with that for fermions .
an important characteristic of metallic systems is the ratio between the heat and electric conductances , @xmath82 , known as the lorentz number . [ for fermi liquids , @xmath83 . according to eqs .
( [ eq : condsimple ] ) and ( [ eq : thermosimple ] ) this quantity for exclusons , though temperature independent for @xmath22 , depends on the statistical parameter @xmath84 .
this result coincides exactly with that for an infinite luttinger liquid.@xcite | we consider the ballistic transport of quasiparticles with exclusion statistics through a 1d wire within the landauer - bttiker approach . we demonstrate that quasiparticle transport coefficients ( electrical and heat conductance , as well as thermopower ) are determined by the same general formulae as for particles with normal statistics . by applying the developed formalism to the ballistic transport of fractional charge
it is shown that for a wire in contact to quasiparticles reservoirs the transport coefficients depend on the fractional charge .
specific features of resonant tunneling of quasiparticles are discussed . |
a periodic structure composed of alternating elastic and piezoelectric layers , where the faces of each piezoelectric layer are electroded and connected by a circuit with the capacity @xmath0 , has been considered in @xcite .
the dispersion equation for a bilayered unit cell has been obtained in @xcite and generalized for a multilayered unit cell in @xcite .
the tunable via @xmath0 effective constants of the periodic structure were derived in @xcite .
it has been noted in @xcite that the capacity @xmath0 of the external circuit may be negative .
the present paper analyzes some unusual features of the floquet - bloch dispersion spectrum occurring at @xmath6 in such a case , using the definition of the quasistatic effective elastic constant @xmath7 @xcite , it is shown that the most interesting behavior comes about at low frequency .
this is because the value of @xmath7 becomes negative in the certain interval of @xmath1 , where @xmath7 varies from @xmath8 to zero . as a result , at @xmath9 there occurs an absolute stopband which starts from the quasistatic limit @xmath3 , while the first dispersion branch starts from nonzero @xmath10 . at certain negative @xmath0 ,
the group velocity of this branch changes from positive to negative value .
other peculiar features are infinite group velocity at the origin @xmath11 of the first dispersion branch at the pole @xmath12 and a flat band @xmath3 at @xmath13 .
consider a periodic structure whose unit cell consists of elastic layer 1 and piezoelectric layer 2 .
suppose that the faces of each piezoelectric layer are electroded and connected by a circuit with capacity @xmath14 let a time - harmonic longitudinal wave propagate along the axis @xmath15 orthogonal to the layer interfaces .
denote the density and elastic constant of the elastic layer by @xmath16 and @xmath17 and the density and elastic , piezoelectric and dielectric constants of the piezoelectric layer by @xmath18 , @xmath19 , @xmath20 , @xmath21,@xmath22 the layer thicknesses are @xmath23 and @xmath24 so that the period is @xmath25 the displacement @xmath26 and stress@xmath27 at the period edges @xmath28 and @xmath29 are related as follows @xcite
@xmath30with @xmath31 where @xmath32 is the surface area of the electrode , @xmath33 means transposition and @xmath34 in view of ( [ 9 ] ) , the dispersion equation of the floquet - bloch spectrum @xmath35 is @xmath36 where @xmath37 is the floquet wavenumber .
according to @xcite , taking low - frequency asymptotics yields the quasistatic effective elastic constant and effective density of the periodic structure in the form of @xmath38 the corresponding quasistatic effective speed is @xmath39 our purpose is to analyze the dispersion spectrum and its low - frequency limit in the case @xmath1 .
following @xcite , consider an example of the glass / pzt periodically bilayered structure where each pzt layer is electroded and connected as in the inset to fig .
let @xmath40 mm .
the material constants are given in @xcite .
the dependence ( [ a9 ] ) of the quasistatic effective elastic constant @xmath7 on @xmath0 is plotted in fig .
it is seen that @xmath7 is negative for @xmath41 where @xmath42 are pole and zero of @xmath43 .
this leads to a quasistatic absolute stopband which starts from @xmath44 and extends to a certain nonzero @xmath10 .
let us examine in detail the evolution of the dispersion spectrum with the decrease of @xmath1 .
figure 2 displays a set of dispersion spectra at different fixed values of @xmath1 which are compared with the spectrum at @xmath45 .
it is seen that the values @xmath46 first affect the second branch @xmath35 which changes the slope from negative to positive . as @xmath47 approaches @xmath48 @xmath49 , the group velocity @xmath50 of the first branch at the origin point @xmath51 tends to infinity ( along with the quasistatic phase velocity ( [ es ] ) ) .
this implies stiffening effect which can be observed in @xcite . in the interval @xmath52 ,
where @xmath9 ( see fig .
[ fig1 ] ) , there is the absolute stopband starting from the quasistatic limit @xmath3 . at certain @xmath53 ,
the first branch which starts from @xmath54 becomes flat and then changes slope from positive to negative , see fig . 2 at @xmath55 @xmath49 and @xmath56 @xmath57 .
for @xmath58 @xmath49 the quasistatic elastic constant @xmath7 and hence the quasistatic phase velocity begin to grow from zero , so that the first dispersion branch reappears at its quasistatic origin @xmath11 . further decrease of @xmath0 affects the position of the first stopband , as it is observed in fig . 2 for @xmath59 @xmath60 , ... , @xmath61 @xmath49 .
a possibility has been shown of the quasistatic absolute stopband , the infinite group velocity and the flat bands in the floquet - bloch spectrum of a periodic structure of elastic and piezoelectric layers where the faces of each piezoelectric layer are electroded and connected via a circuit with negative capacity @xmath0 .
this work has been conducted in the framework of the project mirages anr-12-bs09 - 0015 with the support of the competitiveness cluster aerospace valley .
the authors grateful to a .- c .
hladky and b. dubus for the useful discussions . 9 s. degraeve , c. granger , b. dubus , j.o .
vasseur , a .- c .
hladky , m. pham - thi .
`` contrle lectrique de la propagation dondes lastiques dans des cristaux phononiques pizoelectriques '' 21 congrs francais de mcanique . bordeaux .
26 - 30 august , 2013 . | propagation of a longitudinal wave through the periodic structure composed of alternating elastic and piezoelectric layers is considered .
the faces of each piezoelectric layer are electroded and connected via a circuit with the capacity @xmath0 .
it is shown that if @xmath1 then the floquet - bloch spectrum @xmath2 in a certain range of negative @xmath0 may possess a quasistatic absolute stopband starting at @xmath3 .
other unusual features of the spectrum occurring at certain fixed values of @xmath1 are the infinite group velocity of the first branch at the origin point @xmath3 , @xmath4 and the flat bands @xmath5 . |
[ sec : intro ] the origin of the distribution of mass and angular momentum in disk galaxies is still an open issue , despite its long history . the archetypical model of eggen , lynden - bell & sandage ( 1962 ) provides a useful framework , in which a disk galaxy ( the milky way ) forms by centrifugal support of a collapsing gas cloud .
mestel ( 1963 ) added the assumption that the specific angular momentum of each mass element , @xmath2 , is conserved during the collapse , and demonstrated that the final mass profile of the disk could then be related to the initial mass and angular momentum distribution of the gas .
the key is that the cumulative mass with specific angular momentum less than @xmath2 , @xmath13 , is preserved during the collapse .
crampin & hoyle ( 1964 ) ( also innanen 1966 ; freedman 1970 ) then realized that the observed exponential density profiles of disk galaxies in circular motions are consistent with the angular momentum distribution of a hypothetical uniform sphere in solid body rotation , @xmath14 \label{eqt : iso}\ ] ] where @xmath15 with @xmath16 the fixed angular velocity and @xmath17 the radius of the sphere , and @xmath18 the total mass of the sphere .
fall & efstathiou ( 1980 ) re - examined this question in the context of the more modern view according to which disk galaxies form from contracting gas within extended dark - matter halos ( white & rees 1978 ) .
they added the assumption that the gas and dark matter were initially well - mixed such that the distribution of @xmath2 in the disk is equal to that of the halo . rather than assuming an angular momentum profile , they implicitly deduced it from the constraint that the final surface density profile of the disk is exponential .
the implied disk scale radius , @xmath19 , is related to the total pre - collapse angular momentum of the halo , roughly @xmath20 . here , @xmath7 is the dimensionless spin parameter ( e.g. peebles 1969 ) @xmath21 where @xmath22,@xmath23 and @xmath10 are the total angular momentum , energy and mass of the system , and @xmath24 is newton s constant .
angular momentum is presumably acquired by the dark matter ( and gas ) through tidal interactions with neighboring objects ( peebles 1969 ) .
barnes & efstathiou ( 1987 ) ( following efstathiou & jones 1979 ; efstathiou & barnes 1983 ; zeldovich & novikov 1983 ) utilized cosmological n - body simulations with a cdm power spectrum to make detailed predictions for halo angular momentum structure and statistics . in particular
they found that the distribution of spin parameters is roughly log normal , with a median value of @xmath25 .
they also confirmed that the angular momentum of pre - collapse halos grows roughly linearly with time , as predicted by linear tidal - torque theory ( doroshkevich 1970 ; white 1984 ) .
several numerical and analtyical investigations have followed , which confirmed the nature of the @xmath7 distribution , addressed the alignment of the spin vectors between neighboring galaxies , the ( lack of ) dependence of halo spin on environment , and the effect of major mergers on halo spin ( e.g. , frenk et al .
1988 ; heavens & peacock 1988 ; zurek , quinn & salmon 1988 ; warren et al .
1992 ; catelan & theuns 1996 ; cole & lacey 1996 ; lemson & kauffmann 1997 ; gardner 2000 ) .
armed with this global information about the initial state of the dark matter and gas mixture , several authors have used the above disk formation scenario to predict observable quantities .
blumenthal et al . ( 1986 ) addressed the rotation curves of disk galaxies in the context of the cdm hierarchical formation scenario , assuming @xmath26 of the mass in the gas component ( blumenthal et al .
they pointed out that the non - dissipative dark halo should react to the dissipative gas infall by considerable contraction , and showed that this process can be approximated using an adiabatic invariant .
this coupling between the halo and the disk conspires " to produce a continuity between the disk - dominated and halo - dominated regions of the rotation curves , as observed ( burstein & rubin 1985 ) .
flores et al . ( 1993 ) explored the dependence of the final rotation curves on a range of assumed parameters , including the measured spread in @xmath7 .
dalcanton , spergel , & summers ( 1997 ) included a more realistic initial halo density profile ( hernquist 1990 ) and explored the properties of both high surface brightness and low surface brightness galaxies .
mo , mao , & white ( 1998a ; 1998b , 1999 ) performed a similar examination within the context of various cosmologies , assuming that the dark halos followed the density profile advocated by navarro , frenk , & white ( 1996 , 1997 ; nfw ) and compared their predicted galaxy properties to both local and high - redshift data .
van den bosch ( 1998 ; 2000 ) and van den bosch & dalcanton ( 2000 ) extended the approach to include a bulge component and investigated the origin of the hubble sequence of galaxy types and other aspects of galaxy formation .
the main commonality in all of these investigations is that the variation in observable galaxy properties depends almost exclusively on the initial total spin parameter of the halo hosting the galaxy .
however , all of the predictions rest on the implicit assumption that , if specific angular momentum is conserved , the initial angular momentum profiles of the halos produce disks that are exponential in form .
profile was that of a uniform solid body rotator , which produces disk density profiles that are similar to exponential ( ryden & gunn 1987 ) but only over about two disk scale lengths ( olivier , primack , & blumenthal , 1991 ) . ] within this simple framework for disk formation , the key missing ingredient is the actual distribution of angular momentum in halos . partially motivated by this , we set out to measure the @xmath2 profiles directly using a statistical sample of halos in a high resolution simulation , and to estimate the intrinsic scatter in these distributions .
the long - term goal is to learn what the detailed form of halo angular momentum profiles can teach us about the origin of galaxy types , disk surface brightness profiles , and bulge properties .
an additional strong motivation for this study comes from the results of hydrodynamic simulations of disk galaxy formation in a cosmological context . unlike the simplified , monolithic - collapse framework discussed above , the hydrodynamic treatments model gas cooling within the hierarchical growth of dark halos , and typically find that the resulting disks are significantly smaller than real galactic disks ( navarro & steinmetz 2000 ; weil , eke , & efstathiou 1998 ; navarro & steinmetz 1997 ; navarro , frenk , & white 1995 ) . the problem apparently arises because most of the mass is accreted through mergers with objects whose gas component has already cooled and clumped , causing a large fraction of their angular momentum to be transferred to the dark halo or transported outwards . since angular - momentum transport is likely to always be from the inside out , the simple case in which @xmath2 is conserved during the collapse , which we can study in detail with high accuracy , can serve as a useful limiting case .
this will provide a better basis for addressing the more complex physical processes required for solving the angular - momentum problem seen in the hydro simulations . in we describe our methods , including the simulation and halo finding , the measurement of the spin parameter , the associated errors , and the measurement of the angular - momentum profile . in
we discuss the measured ( universal ) mass distribution of angular momentum in halos . in
we address the spatial distribution of angular momentum , its alignment , cylindrical symmetry , and profile in spherical shells . in
we use both tidal - torque theory and an estimate of angular momentum transfer due to minor mergers to explore the origin of the angular - momentum profile . in we briefly describe the implications for the disk surface density profile and the angular momentum catastrophe " .
we discuss our results and conclude in .
an appendix is devoted to testing our results with a simulation having @xmath27 times the mass resolution , and investigating @xmath13 profiles of halos at high redshift .
large cosmological n - body simulations have reached the stage where detailed structural properties of many dark - matter halos can be resolved simultaneously . a method which allows the required force and
mass resolution is the adaptive refinement tree ( art ) method ( kravtsov , klypin , & khokhlov 1997 ) which implements successive refinements of the spatial grid and time steps in high density environments .
the simulations based on the art code provide a statistical sample of halos , covering a wide mass range , with the resolution appropriate for studying angular - momentum profiles .
we have used the art code to simulate the evolution of collisionless dark matter within the standard " low - density flat @xmath1cdm model ( @xmath28 , @xmath29 , @xmath30 , and @xmath31 at @xmath32 ) .
the simulation followed the trajectories of @xmath33 particles within a periodic box of comoving size @xmath34 from redshift @xmath35 to the present .
a basic @xmath36 uniform grid was used , and six refinement levels were introduced in the regions of highest density , implying a dynamic range of @xmath37 .
the formal resolution of the simulation is thus @xmath38 , and the mass per particle is @xmath39 .
our halo finding algorithm ( bullock et al
. 2000 ) is based on the bound density maxima technique ( klypin & holtzman 1997 ) and has been specifically designed to identify halos ( and subhalos ) in such a high resolution simulation . in the current study of angular momentum , we limit ourselves to ` distinct ' halos , which do not reside within a larger host halo , in the mass range @xmath40 . in the appendix
, we use a higher resolution simulation to extend our mass range down to @xmath41 .
a brief description of the halo finder is as follows : after finding all the maxima in the smoothed density field of the simulation , we unify overlapping maxima , define a minimum number of particles per halo ( say 50 ) , and iteratively find the center of mass of a sphere about each of the remaining maxima .
we compute the spherical density profile about each center and identify the halo virial radius @xmath42 inside which the mean overdensity has dropped to a value @xmath43 , based on the top - hat spherical infall model . for the family of flat cosmologies ( @xmath44 ) , the value of @xmath43 can be approximated by ( bryan & norman 1998 ) @xmath45 , where @xmath46 . in the model used in the current paper
, @xmath43 varies from about 180 at @xmath47 to @xmath48 at @xmath32 .
we then fit each halo density profile with a universal functional form .
we adopt the nfw profile , @xmath49 with the two free parameters @xmath50 and @xmath51 .
an equivalent pair of parameters is , for example , the virial mass @xmath52 and the concentration parameter @xmath53 . using this fit , we iteratively remove unbound particles from each modeled halo , and unify every two halos that overlap in their @xmath50 and are gravitationally bound .
the modeling of the halos with a given functional form allows us to assign to them characteristics such as a virial mass and radius , and to estimate sensible errors for these quantities .
the halo finding is 100% complete for halos of more than @xmath54 particles , @xmath55 ( see sigad et al .
2000 ) . for our purpose here , we limit the sample to halos of more than @xmath56 particles .
a first step towards measuring the distribution of specific angular momentum within each of the halos is measuring the _ global _ spin parameter of each halo .
the angular momentum of a halo of @xmath0 particles is defined by @xmath57 where @xmath58 and @xmath59 are the position and velocity of the @xmath60th particle with respect to the halo center of mass . in principle , given the spin parameter @xmath7 , the value of the global specific angular momentum , @xmath61 , can be determined by using an assumed energy content for the halo in . in practice , however , this is not a straightforward procedure .
for example , the energy of a halo in a crowded region is somewhat ambiguous because it depends on the environment .
a related difficulty arises when only a sub - volume of the virial sphere is concerned , e.g. , within the cooling radius .
we therefore define an alternative and more practical spin parameter by @xmath62 given the angular momentum @xmath22 inside a sphere of radius @xmath17 containing mass @xmath10 , and where @xmath63 is the halo circular velocity at radius @xmath17 , @xmath64 .
this spin parameter reduces to the standard @xmath7 when measured at the virial radius of a truncated singular isothermal halo .
the value of @xmath65 turns out to be robust to the choice of an outer radius ; we use below the virial radius @xmath42 , but we find a very similar distribution of @xmath65 values when using @xmath66 instead .
this allows modelers to freely scale our results using any desired outer radius appropriate for the problem at hand .
the spin parameters @xmath65 and @xmath7 are in fact very similar for typical nfw halos .
they are related by @xmath67 , where @xmath68 and @xmath69 are the energies of virialized halos with isothermal and nfw density profiles respectively , and @xmath70 $ ] ( mo et al .
1998)./[\cvir - ( 1+\cvir)\ln(1+\cvir)]^2 $ ] . ]
the function @xmath71 is about unity for typical concentrations , @xmath72 .
the distribution of @xmath65 over the @xmath73 halos in our sample is shown in .
it is well fit by a log - normal distribution , @xmath74 with best fit values @xmath75 and @xmath76 .
not surprisingly , the distribution of @xmath65 is very similar to the known distribution of @xmath7 ( barnes & efstathiou 1987 ) .
the distribution of @xmath7 values for our halos has best fit values @xmath77 and @xmath78 . to highlight a small difference between @xmath65 and @xmath7 ,
displays the values of @xmath65 versus @xmath79 for our sample of simulated halos , and shows a weak correlation .
given that @xmath7 is known to be uncorrelated with @xmath79 ( nfw ) , this correlation arises from @xmath71 in the above relation between the two spin parameters ( as demonstrated in the figure ) and the traditional spin parameter @xmath7 . ] .
the error on the measurement of @xmath80 in the simulations is of primary importance for the interpretation of the results and , here , for the specific task of fitting a functional form to the angular - momentum profile . obtaining a precise measurement of @xmath80 in halos
is difficult because the signal - to - noise is relatively small due to the fact that the local coherent velocity of the halo contributing to the angular momentum is small compared the velocity dispersion .
the variable @xmath65 itself roughly characterizes the signal to noise ( per particle ) at any radius @xmath17 , , which is typically at the level of a few percent . in order to obtain a crude estimate of the error , which is probably an upper limit
, we singled out several large halos with @xmath81 simulated particles ( @xmath82 ) , and measured their angular momentum using a series of sparsely sampled realizations of @xmath0 particles down to @xmath83 of @xmath81 , with 100 random realizations for each value of @xmath0 . the average and standard deviation of the measured value @xmath22 relative to the original value obtained with all the particles , @xmath84 , as a function of @xmath0 ,
are shown in for three halos with different values of @xmath65 .
we find that for a subsample of @xmath0 particles the standard deviation @xmath85 scales like @xmath86 this scaling relation is shown as error bars in , and it fits rather well the measured scatter .
the first term in corresponds to the general poisson sampling error and the second term reflects the noise due to velocity dispersion .
the second term typically dominates because @xmath65 is small , and thus the spin parameter itself directly influences our ability to measure @xmath22 accurately , with the number of particles needed for a fixed accuracy scaling like @xmath87 . for @xmath88 ,
even @xmath89 accuracy requires @xmath90 .
the above estimate of the error probably tends to overestimate the true error because it includes small - scale noise that is not present when a smaller number of particles is used in the actual dynamical simulation of the halo . on the other hand
, our procedure above does not mimic the error made in following the true dynamics of the smooth halo when a smaller number of particles is used an effect which tends to cause an underestimate of the error .
an accurate evaluation of the error in @xmath22 requires a direct comparison of the same halo as simulated with different resolutions , and is clearly sensitive to the @xmath0-body code used . in our analysis here
we limit ourself to the crude estimate , . in the appendix we use a simulation with @xmath27 times the mass resolution in order to check for systematic differences in the derived @xmath13 profiles , and find none .
this analysis gives us confidence that mass resolution does not affect our results in any severe way .
our goal is to determine how the angular momentum is distributed in the halo .
we wish to compute the mass distribution of specific angular momentum , @xmath13 , and also the spatial distribution of @xmath2 .
we first compute the total @xmath80 for each halo , and let it define the @xmath91 direction . then we subdivide the spherical volume of each halo into many spatial cells , each containing many particles , and compute the specific angular momentum in each cell , projected along the @xmath91 direction .
in general we use the symbol @xmath2 to refer to this @xmath91-projected angular momentum . in some instances
we also investigate the non - projected , absolute magnitude of angular momentum in each cell , which we refer to as @xmath92 .
when comparisons between the two measurements are made , we explicitly add the subscript @xmath91 to the projected value , @xmath93 .
the cell geometry and spatial distribution were designed to maximize the number of particles in each cell while at the same time sampling as much of the @xmath2 distribution as possible .
the vector character of angular momentum naturally introduces a preferred axis , convolved with the general spherical symmetry of the mass distribution .
the cells are defined using the usual spherical coordinates ( @xmath94,@xmath95,@xmath96 ) about the halo center .
each of our cells spans the full @xmath97 range in @xmath96 , and they span the range of ( @xmath98 , @xmath99 ) from ( @xmath100 ) to ( @xmath101 ) . the radial shells are spaced such that each contains the same number of particles .
the shells are then divided into three azimuthal cells of equal volume between @xmath102 and @xmath103 .
positions with the same @xmath94 @xmath99 above and below the equatorial plane belong to the same cell .
for all the @xmath13 profiles studied in detail and presented in the figures below , we adjusted the number of radial bins such that each cell contains roughly @xmath104 particles , where @xmath105 if @xmath106 and @xmath107 for @xmath108 .
we construct @xmath13 profiles for each halo by ranking the cells by their @xmath2 values , and counting the cumulative mass in cells with angular momentum less than @xmath2 . because @xmath2 corresponds to a projected component
, it is possible for a cell to have a negative value of @xmath2 .
the anti - alignment needed for a cell to have a negative @xmath2 value is rare , but when one occurs , we remove the cell completely from the ranked list used to construct the @xmath13 profile . about @xmath109 of halos have a significant amount of their total mass ( @xmath110 ) contained in negative @xmath2 cells .
we do not consider these anti - aligned halos in our @xmath13 analysis .
we assign an error to the @xmath2 in each cell using .
every halo analyzed has @xmath111 particles within its virial radius , corresponding to @xmath112 .
we have @xmath113 halos that meet our requirements . in practice
, we determined the shape of the functional form which fits @xmath13 by examining the group of ( @xmath114 ) halos with @xmath115 particles , for which the profiles are determined with greater accuracy .
we extended the mass range down to @xmath116 particles in order to determine whether more typical galactic - sized halos have a similar profile .
although these smaller halos have larger errors on their fit parameters , they seem to obey the same @xmath13 profile .
the main limitation of this analysis is mass resolution . because we need @xmath117 particles per cell in order to obtain a reasonable measure of @xmath2 , our determination of @xmath13 is only reliable down to @xmath26 for @xmath108 and @xmath118 for @xmath106 .
we find that the specific angular - momentum profiles of halos in the simulation are well - fit by the following two parameter function : @xmath119 the profile has an implicit maximum specific angular momentum @xmath120 .
it is roughly a power - law for @xmath121 , and flattens out for @xmath122 .
as can be seen by its relation to @xmath123 , the quantity @xmath5 ( @xmath124 ) acts as a shape parameter : for @xmath125 , @xmath13 is a pure power law , while @xmath126 means that only half the mass falls within the power - law regime and the bend is pronounced .
the pair of parameters @xmath5 and @xmath6 fully defines the angular momentum distribution of the halo .
the global spin parameter is related to @xmath5 and @xmath6 via @xmath127 where @xmath128 once the value of @xmath65 is known , eq . [ eq : fit ] is a one - parameter fitting function . in fitting the @xmath13 profiles for each halo , we first measured @xmath65 and then obtained the best fit @xmath13 with the constraint that @xmath5 and @xmath6 return the measured @xmath65 .
the pair @xmath129 is perhaps the most useful pair of parameters for characterizing the halo angular momentum because @xmath65 is the conventional global measure and @xmath6 can be explicitly determined from @xmath65 and @xmath5 via , while @xmath5 can not be determined explicitly from the other two parameters without iteratively solving .
alternatively , we provide the following approximation for the iterative solution of : ( b ) 1 + , which is accurate to within @xmath130 in @xmath131 over the range @xmath132 .
figure [ fig : examples ] shows @xmath13 examples for several of our halos . in the left panel
we have normalized the profiles by @xmath133 and @xmath123 in order to illustrate how different values of @xmath5 affect the distribution . as @xmath5 approaches its minimum value of @xmath103
, a larger fraction of the halo mass is spinning slowly compared to @xmath123 .
larger values of @xmath5 imply a more uniform @xmath2 distribution .
the right panel shows the same four halos , now normalized to their best - fit @xmath6 and @xmath134 values , where @xmath135 .
notice how remarkably the halos follow the characteristic functional shape .
the spread in profile shapes can be described by the distribution of @xmath131 ( @xmath136 ) .
figure [ fig : muhist ] shows the distribution of @xmath137 for all of our halos ( histogram ) along with a gaussian distribution with the same mean and standard deviation : @xmath138 , @xmath139 .
the implied @xmath140 range is @xmath141 .
although the spin parameter @xmath65 and the shape parameter @xmath5 clearly measure two different aspects of the angular - momentum distribution , they are not necessarily uncorrelated .
figure [ fig : mu.vs.lambda ] shows the joint distribution of these two parameters for our simulated halos .
we observe a weak but significant linear correlation between @xmath142 and @xmath143 , with a pearson correlation coefficient of @xmath144 ( corresponding to a probability @xmath145 for no correlation ) .
thus , high - spin systems tend to have more evenly distributed ( power - law ) @xmath13 profiles than low - spin systems , but the scatter about this trend is large .
does the angular - momentum distribution correlate with the mass - density distribution ?
we know that the global spin characterized by @xmath146 does not correlate strongly with the halo mass ( barnes & efstathiou 1987 ) and correlates only weakly with the concentration parameter ( ) .
figure [ fig : cm ] shows @xmath131 versus these mass parameters .
we detect a marginal anti - correlation with mass , with @xmath147 ( @xmath148 for no correlation ) , and an even less significant anti - correlation with @xmath79 , of @xmath149 ( @xmath150 for no correlation ) .
in the previous section we explored the distribution by mass of the specific angular momentum within each halo .
this does not tell us much about how well the angular momentum is aligned throughout the halo or how angular momentum is distributed spatially .
this section is explicitly devoted to these issues . in
we address the question of alignment , in we investigate the cylindrical symmetry of the angular - momentum distribution , and in we explore the angular - momentum profile in spherical shells . here
we address the question of how well the smoothed angular momentum is aligned throughout the halo volume . for our first alignment statistic
, we divide the particles in each halo into an inner half - mass sphere and an outer half - mass shell and measure the cosine of the angle between inner and outer angular - momentum vectors , @xmath151 .
the distribution of measured alignment values is shown in the top panel of figure [ fig : ahist ] . before we attempt to interpret these results , it is essential to keep in mind the large errors involved in determining angular momentum direction vectors . even for a halo with an intrinsically well - aligned angular momentum distribution ( @xmath152 ) , measurement errors in each of the six directional components involved in the alignment cosine will tend to drive the measured value _ towards zero _ , and ,
if large enough , will lead to a false measure of misalignment .
the distribution in the top panel can therefore serve as a lower bound for the true alignment . in order to obtain a conservative estimate of the distribution of true alignment cosines , we use to assign errors to each component of @xmath153 in both the inner and outer half - mass regions and perform a standard propagation of errors to obtain an estimate for the measurement error on @xmath151 .
we add the estimated error in each case to the measured alignment value in order to obtain a `` corrected '' distribution for @xmath151 .
this is shown in the lower panel of figure [ fig : ahist ] . as argued above
, we believe equation [ eq : er ] is an overestimate for the measurement error , so our corrected alignment cosines represent a plausible upper limit for the alignment .
the true intrinsic distribution probably lies somewhere between the corrected and uncorrected distributions .
if we make this assumption , then the two distributions allow us to conclude that between @xmath154 and @xmath155 of the halos are aligned to a greater degree than @xmath156 , although there is also a tail of significantly misaligned halos .
if the angular momentum within a halo is poorly aligned , one might expect a reflection of this to appear in its @xmath157 profile .
there is a hint for such a trend in , where the distribution of @xmath151 for all the halos is compared to the distribution for low @xmath5 halos of @xmath158 .
the latter clearly show a higher fraction of misaligned halos .
figure [ fig : avsmu ] shows the corrected @xmath151 values versus the shape parameter @xmath5 for each halo , and indeed there is a significant trend , with @xmath159 ( @xmath160 ) . and
@xmath65 ; however , this seems to be due to the correlation between @xmath22 errors and the spin .
the correlation is not apparent when we use our corrected estimate for @xmath151 . ]
the misaligned halos tend to be associated with low-@xmath5 profiles that is , profiles that deviate significantly from power laws , with a relatively larger amount of mass in the tails of their @xmath93 distributions .
this result suggests that there are common aspects to the origin of misalignment and that of the non - powerlaw nature of the @xmath13 profiles .
since halos with misaligned angular momentum distributions may be less likely to host large disk galaxies , this result should be kept in mind when modeling galaxy formation within halos with low-@xmath5 angular momentum profiles .
addresses the angular - momentum alignment in halos in another way , independent of any specific spatial symmetry .
here we show the @xmath13 profiles for four different halos , comparing the distributions of both @xmath93 and @xmath92 in cells throughout the halo volume .
we see that in most cases , the two kinds of profiles are similar ; for at least half the mass they differ by less than a factor of two .
this indicates that the angular momentum is reasonably well aligned throughout the halo .
the slight misalignment is expressed in a trend that the @xmath92 distributions systematically resemble a power - law more than the @xmath93 distributions .
this is demonstrated in , which shows the values of @xmath131 obtained using @xmath92 versus those obtained using @xmath93 .
the correlation is obvious , with @xmath161 .
the value of @xmath131 is typically about a factor of 3 higher for the @xmath92 profiles .
the fact that this factor is somewhat higher for low values of @xmath5 , reflects the correlation seen in between alignment and @xmath5 .
cylindrical symmetry is the natural symmetry in the spatial distribution of angular momentum as a result of tidal - torques or a sequence of mergers . in order to explore the degree of cylindrical symmetry
, we have again divided the halos into cells , as described in , except that we no longer explicitly impose symmetry about the equatorial plane and rotational symmetry in @xmath96 .
specifically , we define the cells as outlined in , but then , for each value of @xmath162 , we further divide the cell particles into four regions : above and below the equatorial plane , and split by rotation angle @xmath163 and @xmath164 . because each cell now contains roughly one fourth of the particles , we increased our minimum halo mass threshold by a factor of four to @xmath165 , only for this section of our analysis .
figure [ fig : map ] shows a map of the @xmath2 distribution in cells in four of our halos . except for the halo depicted in the lower - left panel , the symmetry of the angular - momentum distribution tends to be more cylindrical than spherical , with a weak dependence on @xmath95 at a constant distance from the global spin axis , @xmath162 . on the other hand
, the angular momentum of the halo in the lower - left panel tends to be concentrated in the @xmath32 plane . another way of visualizing the degree of cylindrical symmetry of halos is shown in figure [ fig : cyl ] .
it shows the values of @xmath2 in the same cells and for the same four halos displayed in figure [ fig : map ] , now plotted as a function of the distance from the angular momentum axis , @xmath166 , and distinguished by different symbols according to their average distance from the equatorial plane , @xmath167 .
the different symbol types should be well - mixed if a halo is cylindrically symmetric .
this is the case for three out of the four halos shown , except in the lower - left panel , as expected from figure [ fig : map ] . although the angular momentum distributions in most of the halos tend to qualitatively show a certain degree of cylindrical symmetry , there is some indication that , for fixed @xmath162 , mass near the equatorial plane tends to have more specific angular momentum than corresponding mass in the poles . in order to quantify the extent to which this is true
, we compare the quantity @xmath168 , defined as the ratio of specific angular momentum in cells near the equatorial plane ( @xmath169 ) to that in cells near the poles ( @xmath170 ) . for each halo
, we determine @xmath168 by averaging over all values of @xmath162 for which the two @xmath91 ranges overlap .
for example , in figure [ fig : cyl ] , @xmath168 can be estimated for each halo by comparing the average value of @xmath2 for the solid squares ( @xmath171 ) to that for the open circles ( @xmath172 ) at fixed values of @xmath162 .
figure [ fig : jrathist ] shows the histogram of this ratio for all of our halos .
the mean is @xmath173 , and the standard deviation is @xmath174 . in order to test whether our specific choice of cell geometry would bias this measure
, we also generated 500 realizations of halos with perfect cylindrical symmetry ( @xmath175 ) .
they each had nfw density profiles and @xmath13 profiles in the form of equation [ eq : fit ] .
the resulting @xmath168 histogram for the artificial catalog is shown by the shaded historgram in figure [ fig : jrathist ] .
the sharp peak near a ratio of @xmath103 indicates that the bias due to our cell geometry is much smaller than the detected deviation from cylindrical symmetry .
recall however that the average deviation from cylindrical symmetry is small , about 35% .
although , as demonstrated in , the specific angular momentum distribution in halos is closer to being cylindrically symmetric than spherically symmetric , an exploration of how @xmath2 , averaged over shells , behaves as a function of radius is of particular interest because this is the profile that is most directly addressed by theory ( see below ) .
barnes & efstathiou ( 1987 ) found in their early work an indication for a rough @xmath176 behavior .
for our sample of well - resolved halos , we indeed find that a power - law : @xmath177 , with @xmath178 roughly distributed over the halos like a gaussian : @xmath179 is a good approximation . in order to connect more closely to the mass distribution profiles discussed in previous sections , and also to allow a more direct connection with theory , we focus on characterizing the spherical profile as a function of the mass contained within the radius @xmath94 , @xmath180 .
we find that the power - law approximation describes the profiles reasonably well , j(m ) m^s , with @xmath181 roughly distributed over the halos like a gaussian : @xmath12 .
figure [ fig : sphere ] shows three examples of the spherical @xmath182 profiles along with their best - fit power - laws .
notice that although the power - laws serve as a good general characterization of the @xmath2 behavior with @xmath10 , the profiles do not always follow a power - law form in detail .
in many cases , the profiles show an upwards bend in the outer shells , reminiscent of the characteristic bend in the low-@xmath5 mass distribution , @xmath13 .
in some cases , @xmath2 is not always monotonic with @xmath10 .
so , although the @xmath181 values provide a useful characteristic of the spherical distribution of @xmath2 , some information is lost under this approximation .
shown in is the distribution of @xmath181 values , versus halo mass . despite the large scatter , there is an anti - correlation between slope and mass , with @xmath183 , ( @xmath184 ) . as we will discuss in , a simple calculation based on linear tidal - torque theory provides a hint at understanding the typical slope value @xmath185 , and predicts a mild anti - correlation between @xmath181 and the halo mass , in agreement with the simulations .
the spherical profile @xmath182 and the mass distribution @xmath13 are expected to be related . if the mass distribution is spherically symmetric and the angular - momentum distribution is cylindrically symmetric , this relation can be spelled out explicitly . in this case
the inverse of @xmath13 is the spatial profile @xmath186 in growing cylinders about the total spin axis .
a power law with a bend for @xmath13 translates to a corresponding power law with a weaker bend in the opposite direction for @xmath182 .
we thus expect a correlation between the shape parameter @xmath5 of @xmath13 and the slope @xmath181 of the spherical profile @xmath182 .
a high @xmath181 and a low @xmath5 both correspond to halos with a high fraction of mass spinning slowly relative to @xmath123 .
figure [ fig : s.vs.mu ] shows @xmath181 and @xmath5 for our halos .
an anti - correlation is evident , as expected , with @xmath187 ( @xmath188 ) .
the scatter reflects deviations from the global spatial symmetries .
it is likely , for example , that partial misalignment and pockets of low - angular momentum material at large radii contribute to the scatter . by comparing figures [ fig : s.vs.m ] and [ fig : cm ] it is evident that the scatter in the @xmath181-@xmath5 relation is large enough to wash away any mass dependence of @xmath5 that would be implied by the weak mass dependence of @xmath181 .
in this section we explore the possible origin of the angular momentum profile based on two different pictures for how specific angular momentum is acquired in halos . in the first picture ,
angular momentum is built up shell by shell , and is modeled using linear theory with no further angular momentum transfer within the halo . in the second picture ,
the angular momentum profile arises solely from angular momentum transfer from one or more satellite halo merger events .
since halo mass accretion histories typically reflect some combination of relatively quiescent mass accretion as well as more pronounced mergers ( where @xmath2 transfer is likely ) , the @xmath2 profiles of individual halos may reflect a complicated combination of the two processes explored here .
indeed the two processes are intrinsically linked , since , in the hierarchical framework , all mass accretion can be treated as the accretion of smaller objects . as shown below
, the profiles calculated from linear theory as well as those calculated using @xmath2 transfer from mergers produce a range of @xmath5 values , in qualitative agreement with what is observed in the simulations .
linear tidal - torque theory ( ttt ) provides a hint for the origin of the detected power - law - like @xmath2 profile , as follows .
ttt ( doroshkevich 1970 ; white 1984 ) implies that the angular momentum gained by a halo at time @xmath189 before its turn - around is j_i(t ) = a(t)^2
d(t ) _ ijk t_jl
i_lk , [ eq : ttt ] where the time growth is from some fiducial initial time , @xmath190 is the expansion factor at @xmath189 , @xmath191 is the linear growth factor , @xmath192 is the inertia tensor of the proto - halo at the initial time and @xmath193 is the tidal tensor at the halo center at the initial time , smoothed on the halo scale .
this is based on assuming the zeldovich approximation for the velocities inside the proto - halo , and a 2@xmath194-order taylor expansion of the potential .
we show elsewhere ( porciani , dekel & hoffman 2000 ) that the standard scaling relation of ttt should be slightly modified ; it should read j d(t_c)^3/2 ( m ) m^2/3 , [ eq : scaling ] where @xmath2 is the specific angular momentum of the halo , @xmath195 is the turn - around time of the halo , and @xmath196 is the rms density fluctuation on scale @xmath10 at the initial time . if the perturbation turned around while the cosmology was still einstein - de sitter , then @xmath197 .
dependence on the time of collapse is @xmath198 rather than @xmath199 , because the tidal part of the deformation tensor , the source of angular momentum , is actually independent of the initial density of the perturbation , which determines the collapse time .
the shear depends instead on the rms fluctuation @xmath200 , which involves an implicit mass dependence . ] to understand the general power - law behavior of the angular - momentum profile we apply .
we can assume that mass is accreted in shells and that the turn - around time of each shell is determined by @xmath201 , where @xmath202 is the mean density inside @xmath10 .
for a gaussian field , the typical density fluctuation profile about a random point scales like ( dekel 1981 ) @xmath203 , where @xmath204 is the linear two - point correlation function .
this is accurate to a few percent also around a high peak ( bardeen et al .
1996 , fig . 8) .
thus , for a power - law power spectrum @xmath205 and a flat universe , we obtain the following spin profile within each halo : j(m ) m^2/3 + ( 3+n)/3 .
this implies @xmath206 for @xmath207 , the range appropriate for @xmath208 halos in a cdm spectrum , in pleasant agreement with our finding for the simulated halos ( ) .
the tidal torque theory can be combined with the extended press - schechter formalism ( eps , bond et al .
1991 ; lacey & cole 1993 ) to obtain a more detailed model of @xmath2-distribution , as well as quantitative predictions for values of @xmath181 and @xmath5 and their mass dependence .
let us assume for simplicity that halo mass is acquired via accretion of material with specific angular momentum given by equation [ eq : scaling ] and that the direction of angular momentum of each mass shell is perfectly aligned .
we can readily estimate the slope @xmath181 of the @xmath209 distribution if we assume that the turn - around time for each mass shell is approximated by the time at which the mass was first accreted onto the halo by comparing the expected @xmath2 value at any two shells encompassing masses @xmath210 and @xmath211 , s .
[ eq : slope ] equation [ eq : scaling ] provides the required value of @xmath2 associated with each mass shell given its `` accretion time '' , @xmath212 .
considering , for example , @xmath133 and @xmath213 , we obtain s ( ) 1.44 ( 2 ^ 2/3 d[t_1/2()]^-3/2 ) , where @xmath214 is the time at which the halo of mass @xmath133 first accreted half of its mass , @xmath215 .
we adopted for all halos , by the requirement of virialization , @xmath216 ( the current age of the universe ) , and set @xmath217 .
the most probable value of @xmath218 can be estimated via eps ( lacey & cole 1993 , eq .
the predicted @xmath219 relation is added to as the thick solid line .
the agreement with the mean trend shown by the simulated halos is remarkable , especially since our model has no free parameters .
we can further elaborate the model by following the more detailed mass accretion history of halos using eps merger trees . specifically , we have used the method of somerville & kolatt ( 1999 ) to model mass growth and halo accretion histories for a random ensemble of dark matter halos formed in the @xmath1cdm model and with masses in the mass range studied in the simulation . at each epoch , @xmath220 , and time interval , @xmath221 , the model draws a mass @xmath222 accreted by a halo of mass @xmath223 using the eps probability distribution .
if we assume that this mass is accreted with specific angular momentum given by and that there is no angular momentum loss , we can integrate the specific halo mass accretion history to get the @xmath13 distribution for the final halo at @xmath32 .
figure [ fig : models ] shows @xmath13 profiles calculated using this model along with the fits of the form @xmath224 , , for three representative mass accretion histories from the ensemble of realizations .
one can see that the model reproduces the general shape of the @xmath13 profile with values of @xmath5 that are similar to those of simulated halos . for a given halo mass ,
the differences in merger histories lead to scatter in the resulting values of @xmath5 , with the range of @xmath225 .
the absence of small @xmath5 ( @xmath226 ) may result from our simplified model assumption of perfect alignment .
indeed , the range of these modeled @xmath5 values is similar to the range obtained for our simulated halos when @xmath92 profiles are considered instead of @xmath93 ( see ) .
we can also reproduce both the range of values and the trend with mass of the slope @xmath181 if we force a power - law fit to the model @xmath182 profiles .
the halos with low ( high ) @xmath5 in this model are those that accrete large fractions of their mass at early ( late ) epochs .
interestingly , the value of @xmath5 for a given halo in this model strongly depends on the halo s formation redshift , @xmath227 , defined as the redshift where the mass of the most massive progenitor was half of the final mass for the first time [ @xmath228 . in light of this expectation , it is somewhat surprising that we find no noticeable trend with redshift for @xmath5 parameters in halos of fixed mass ( see appendix ) but we are currently investigating whether a trend is evident using individual halo merger histories ( wechsler et al .
if there is a trend with formation time , it may have interesting implications for galaxy formation .
a simple toy calculation for how angular momentum is being deposited in a large halo during a minor merger seems to provide another hint for the origin of the characteristic @xmath2 profile .
this is studied in more detail using toy models and full @xmath0-body simulations by dekel & burkert ( 2000 ) .
consider a fixed halo and an incoming satellite halo of mass profiles @xmath229 and @xmath230 respectively .
assume that as it moves in , the satellite is losing mass outside a tidal radius @xmath231 , which we approximate at @xmath94 to be determined by = _ t ( r ) , ( r)= .
[ eq : tidal ] for host halos and satellite halos with power - law mass profiles ( @xmath232 ) , the above approximation implies that the total mass of a satellite halo located at a radius r of its host is @xmath233 \propto m(r)$ ] .
suppose that the satellite is spiraling in due to dynamical friction roughly along circular orbits , and that the mass and @xmath2 are deposited locally .
then the spherical @xmath234 profile is obtained by averaging over shells , 4r^2 ( r ) j(r ) = m(r)r + m(r ) r rv_c(r ) .
[ eq : jdeposit ] the two terms on the right hand side reflect angular - momentum transfer due to the slowdown of the satellite by dynamical friction and the direct tidal stripping of mass from the satellite halo , respectively . for an isothermal host and satellite , we obtain j(m ) m .
shows the expected @xmath182 profile based on this toy model for an nfw halo swallowing a satellite , using three alternative recipes for mass loss by the satellite : ( a ) @xmath235 , ( b ) a satellite with an isothermal profile , and ( c ) a satellite of fixed mass , for comparison . the general shape of @xmath236 in the outer decade of mass , when realistic mass loss ( a or b ) is considered , is reminiscent of the profiles detected in the cosmological simulation ( see ) .
full @xmath0-body simulations of mergers , spanning a range of halo and collision parameters , confirm the robust production of such characteristic profiles ( dekel & burkert 2000 ) .
the resulting @xmath182 from a sequence of minor mergers is expected to be a sum of similar contributions , each projected onto the direction of the total net angular momentum ( perhaps determined by the most major merger involving a large fraction of the final mass ) .
we now explore certain possible implications of our results on the angular - momentum distribution in the halos for the formation of galactic disks .
we follow the limiting case assumption that the specific angular momentum is conserved during gas infall into a centrifugally supported disk , and compare the resultant gas density profile to an exponential disk .
also , since the angular momentum structure of halos is governed by at least two parameters say @xmath5 , in addition to @xmath65 we specifically explore how variations in the shape parameter may affect the disk characteristics .
suppose that each dark - matter halo has a specific angular momentum profile @xmath13 given by .
given @xmath5 , @xmath13 is determined by the spin parameter via @xmath237 .
assume that initially the specific angular momentum distribution of the gas mirrors that of the halo , @xmath238 , where @xmath239 is the mass fraction of the halo that ends up as cool baryons in the disk , @xmath240 .
after infall into a disk while preserving @xmath2 , the specific angular momentum of a gas element that ends up in a circular orbit of radius @xmath94 is @xmath241 , where @xmath229 is the total mass of dm and baryons within @xmath94 .
all the gas with specific angular momentum less than @xmath234 will wind up interior to @xmath94 . using
, we obtain for the disk mass profile @xmath242 in order to estimate the implied disk structure , and to gain a qualitative understanding of how the value of @xmath5 may affect the disk size and density profile , we assume first that the dark halo does not react to the infall of gas and approximate the total mass distribution of the system to be that of an isothermal sphere , @xmath243 . in this case @xmath244 .
the final mass distribution of the disk is then @xmath245 here , @xmath246 and @xmath247 .
the implied surface density profile is @xmath248 in our simplified model , the surface density vanishes beyond @xmath249 .
several implications are apparent from the derived disk surface - density profile of .
first , the distribution is qualitatively similar to an exponential disk it falls rapidly at large @xmath94 , and flattens towards the origin .
however , in detail , there are significant differences . at large radii , @xmath250 ,
the fall off @xmath251 is slower than exponential .
near the origin , @xmath252 , the profile is steeper than exponential , @xmath253 versus @xmath254
. as expected , the scale radius @xmath255 scales linearly with @xmath65 .
however , the scale radius is also a strongly decreasing function of @xmath5 . for a fixed spin parameter @xmath146 ,
as @xmath5 varies over its @xmath256 range about the mean ( @xmath257 ) , @xmath19 varies by about a factor of @xmath258 .
this is similar to the variation of @xmath19 as a function of @xmath65 for fixed @xmath5 ( a factor of @xmath258 over the 90% range of @xmath65 ) .
figure [ fig : surf ] shows the disk surface - density profile derived using a more accurate numerical calculation . here
we assume that the dark halo initially followed an nfw density profile with @xmath259 .
the @xmath13 profile is characterized by @xmath260 and @xmath261 , and the disk mass fraction is assumed to be @xmath262 .
the disk profile is obtained by self - consistently solving for the disk circular velocity at @xmath94 and taking into account the adiabatic contraction of the halo in response to the baryonic infall ( blumenthal et al .
1986 ) . shown in comparison is the surface density profile obtained under the same assumptions regarding the halo structure and @xmath2 conservation except that the given @xmath13 distribution is replaced with the assumption that the final disk profile is exponential .
a similar result is obtained by assuming that the initial @xmath13 is that of a uniform - density sphere in solid body rotation .
figure [ fig : surf2 ] shows how the derived disk profile varies as a function of @xmath5 .
as expected , the central concentration of the disk is a decreasing function of @xmath5 .
as anticipated , compared to the exponential disk ( and solid body rotator ) the derived disk profiles have core and tail excesses .
only halos with the highest @xmath5 values ( @xmath263 , corresponding to @xmath264 of the halos ) produce surface density profiles that begin to resemble an exponential disk . in comparison with the observed light profiles of disks ,
the derived gas surface density profiles deviate in their general shape and in particular in their central concentration .
the general deviation of shape is less worrisome because the final stellar profile may not necessarily mirror the gas profile .
as lin & pringle ( 1987 ) and olivier et al .
( 1991 ) argued , viscous transport processes can act to produce exponential stellar disks under rather general assumptions about the nature of the initial gas disk .
however , since typical viscous processes would tend to amplify the central density even further , the indicated central density excess is of some concern .
this result is reminiscent of the more severe problem of excessively small disk sizes seen in hydrodynamic cosmological simulations ( e.g. , navarro & steinmetz 1997 , 2000 ) . that we find a disk central - density excess even under the assumption that angular momentum is conserved during the collapse indicates that the solution to the problem may not be trivial , because any simple angular - momentum transport mechanism would drive the angular - momentum from the gas to the halo and from the inside out and could only worsen the discrepancy .
the solution may require that the baryons somehow obtain an angular momentum distribution that is biased relative to that of the dark matter towards high-@xmath2 .
alternatively , if the resultant profiles are interpreted to represent both a bulge and a disk component , with the slowly rotating core associated with the bulge , then the central concentration problem may be resolved .
testing this hypothesis will require more detailed modeling .
in particular , it will be necessary to match the observed distribution of galaxy bulge - to - disk ratios as well as disk scale lengths .
another way out may be associated with the fact that some of the approximations adopted here may break down when the @xmath2 distribution is significantly misaligned throughout the halo , which indeed preferentially happens for low-@xmath5 halos ( ) .
we have studied the angular momentum structure within dark - matter halos in the mass range @xmath40 , using a sample of @xmath113 halos in an n - body simulation of the @xmath1cdm cosmology .
we examined in detail the more massive of these halos , for which the angular momentum is measured more accurately , and then verified that the more typical galaxy - sized halos have a similar angular momentum distribution .
our primary result is that the mass distribution of specific angular momentum in halos obeys a roughly universal form , which is well fit by the two - parameter function @xmath265 . for a fixed value of the spin parameter @xmath65
, the parameter @xmath5 determines the shape of the distribution , with high-@xmath5 corresponding to a pure power - law and a smaller contrast between the tails of the distribution .
the distribution of @xmath266 is roughly log normal , with the @xmath155 range spanning @xmath267 . compared to an idealized uniform sphere in solid - body rotation , the simulated halos tend to have more of their mass in the tails of the distribution , especially at small @xmath2 .
the shape parameter is only weakly correlated with the spin parameter , @xmath268 with @xmath269 and a correlation coefficient @xmath270 .
most halos have well - aligned angular momentum vectors throughout their volume .
between 70% and 90% have alignment cosines greater than @xmath271 between the inner and outer half masses .
however , at least 10% of the halos have significant misalignments , which may be interesting for several reasons .
first , they tend also to be halos with low @xmath5 , indicating that the direction of * j * in accreted material , or the halo merger history , plays a role in determining the shape of the @xmath13 profile . in addition , it is unlikely that the standard simple picture of disk formation could be valid within a severely misaligned halo , so perhaps these objects typically host spheroidal stellar components .
marginally misaligned halos may play a role in the formation of galactic warps ( e.g. , dekel & schlosman 1983 ) .
the spatial distribution of halo angular momentum tends to be more cylindrically symmetric than spherically symmetric . at a fixed distance from the rotation axis , mass near the equatorial plane
typically has only about @xmath272 more specific angular momentum than corresponding mass near the poles .
the mean @xmath2 is spherical shells encompassing mass @xmath10 is well - fit by a power law , @xmath273 .
the power @xmath181 is distributed like a gaussian with a mean of @xmath274 and standard deviation @xmath275 .
we pointed out two possible explanations for the origin of the universal angular - momentum profile .
the first is based on applying a corrected version of the linear tidal - torque theory ( porciani & dekel 2000 ) to extended press - schechter mass accretion histories .
the second is based on the nonlinear process of angular momentum transfer from satellite orbits in halo mergers ( see also dekel & burkert 2000 ) .
since halo mass growth arises as a combination of merger events and relatively quiescent mass accretion , the origin of the profile in individual halos may reflect some combination of the two processes .
each of these processes seems to produce @xmath2 profiles that are similar in form to those observed in simulated halos , and thus may provide a starting point for a deeper understanding of the origin of halo angular - momentum structure .
finally , we have started to explore the implications of our universal @xmath13 distribution in the context of the archetypical model of galactic disk formation , namely , adiabatic baryonic infall to a rotationally supported disk while conserving angular momentum in every mass element .
we find that the implied surface - density profiles , which vary as a function of both @xmath5 and @xmath65 , deviate significantly from an exponential disk for all but the largest values of @xmath5 .
the resultant surface profiles are more extended than exponential at large @xmath94 , and are overly concentrated at small @xmath94 .
since the observed light profiles of disks are closer to exponential , our result indicates that the disk formation process can not be fully understood within this simplified picture .
we mentioned that the general deviation from an exponential surface density at moderate radii might be overcome in this picture if a viscous transport mechanism is included ( lin & pringle 1987 ) .
as shown by olivier et al .
( 1991 ) , such a mechanism tends to produce exponential stellar distributions regardless of the initial gas profile .
however , most of the expected @xmath2 transport typically acts to increase the central mass density , so the deviation at small radii may be a significant problem . even though we have assumed that the baryons experience no angular - momentum loss , we ended up with an `` angular momentum problem '' similar to that detected in hydrodynamic simulations ( e.g. , navarro & steinmetz 1997 , 2000 , where it is associated with @xmath2 transport from the gas to the dark matter and from inside out ) .
this problem is most severe for halos with small @xmath5 , so the tendency for these halos to have misaligned angular - momentum distributions , which makes them less - likely hosts of large disk galaxies , may partly ease the problem in the limiting case where angular - momentum is conserved .
however , even if we focus on halos that tend to be well aligned , with @xmath276 , the central densities remain higher than those in exponential disks .
a solution to this problem may be to associate the derived central mass concentrations with bulges .
it is indeed possible that the central regions of disks with profiles as in are unstable to self gravity .
more detailed modeling is needed for testing the viability of this solution .
in particular , one may be worried about the model over - predicting bulge to disk ratios and in particular under - predicting the number of bulge - less galaxies . on the other hand ,
such an interpretation may shed new light on the very origin of the hubble sequence , since both @xmath5 and @xmath65 may play a role in determining disk scale lengths , surface brightnesses , and bulge fractions .
this work has been supported by grants from the us - israel bsf , the israel sf , and nasa and nsf at ucsc and nmsu .
jsb was supported by nasa ltsa grant nag5 - 3525 and nsf grant ast-9802568 .
support for avk was provided by nasa through hubble fellowship grant hf-01121.01 - 99a from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
cp was supported by a golda meir fellowship .
we thank andi burkert , stephan corteau , sandra faber , ricardo flores , savvas koushiappas , ari maller , yair sigad , rachel somerville , frank van den bosch , maya vitvitska , david weinberg , and risa wechsler for useful discussions .
since our error estimates are only approximate , it is important to carry an independent test of the validity of our @xmath13 profiles in view of the finite mass resolution of the simulation .
the test utilizes the output from another @xmath1cdm simulation with identical cosmological parameters and the same number of particles as used in our main simulation , but now in a box of side @xmath277 , namely half the original size .
the mass resolution is thus @xmath27 times higher , so a statistical comparison with halos analyzed from our lower resolution simulation at fixed mass will provide a useful test for the effects of mass resolution .
the smaller box simulation also allows us to extend our analysis down to much smaller galaxy - mass halos ( @xmath278 ) .
unfortunately , the higher resolution simulation was stopped at @xmath279 , so we can not compare directly to the @xmath32 results presented in the main paper . however
, a comparison performed at high @xmath91 can perfectly serve our purpose .
this also allows us the opportunity to check for any evolution in the @xmath13 profiles with redshift .
figure [ fig:2box ] shows the best - fit @xmath5 parameters as a function of halo mass for halos from our main simulation at @xmath280 .
the distribution of @xmath5 values is very similar to that seen at @xmath32 ( ) , so there is no indication of evolution .
figure [ fig:2box ] also shows @xmath5 versus @xmath133 for halos from the higher resolution simulation at the same time .
although the overlapping region in mass is rather small , there is no indication of any offset between the two simulations .
the halos have a similar distribution of shape parameters over about three orders of magnitude in mass , @xmath281 .
efstathiou , g. , & barnes , j. 1983 , in proc .
3d moriond astrophysics meeting , formation and evolution of galaxies and large structures in the universe , ed .
, j.audouze and j. tran thanh van ( dordrecht : reidel ) , p. 361 | we study the angular - momentum profiles of a statistical sample of halos drawn from a high - resolution @xmath0-body simulation of the @xmath1cdm cosmology . we find that the cumulative mass distribution of specific angular momentum @xmath2 in a halo of mass @xmath3 is well fit by a _
universal _ function , @xmath4 .
this profile is defined by one shape parameter ( @xmath5 _ or _
@xmath6 ) in addition to the global spin parameter @xmath7 . it follows a power - law @xmath8 over most of the mass , and flattens at large @xmath2 , with the flattening more pronounced for small values of @xmath5 ( or large @xmath6 at a fixed @xmath7 ) .
compared to a uniform sphere in solid - body rotation , most halos have a higher fraction of their mass in the low- and high-@xmath2 tails of the distribution .
high-@xmath7 halos tend to have high @xmath5 values , corresponding to a narrower , more uniform @xmath2 distribution .
the spatial distribution of angular momentum in halos tends to be cylindrical and is well - aligned within each halo for @xmath9 of the halos .
the more misaligned halos tend to have low-@xmath5 values . when averaged over spherical shells encompassing mass @xmath10 , the halo @xmath2 profiles are fit by @xmath11 with @xmath12 .
we investigate two ideas for the origin of this profile .
the first is based on a revised version of linear tidal - torque theory combined with extended press - schechter mass accretion , and the second focuses on @xmath2 transport in minor mergers .
finally , we briefly explore implications of the @xmath13 profile on the formation of galactic disks assuming that @xmath2 is conserved during an adiabatic baryonic infall .
the implied gas density profile deviates from an exponential disk , with a higher density at small radii and a tail extending to large radii .
the steep central density profiles may imply disk scale lengths that are smaller than observed .
this is reminiscent of the angular - momentum problem " seen in hydrodynamic simulations , even though we have assumed perfect @xmath2 conservation .
a possible solution is to associate the central excesses with bulge components and the outer regions with extended gaseous disks . |
we report measurements of the thermal expansion for two eu@xmath0- and two gd@xmath1-based intermetallics which exhibit ferro- or antiferromagnetic phase transitions .
these materials show sharp positive ( euagmg and gdaumg ) and negative ( euaumg and gdagmg ) peaks in the temperature dependence of the thermal expansion coefficient @xmath2 which become smeared and/or displaced in an external magnetic field . together with specific heat data we determine the initial pressure dependences of the transition temperatures at ambient pressure using the ehrenfest or clausius - clapeyron relation .
we find large pressure dependences indicating strong spin - phonon coupling , in particular for gdagmg and euaumg where a quantum phase transition might be reached at moderate pressures of a few gpa .
there has been a continuing interest in the equiatomic ternary intermetallic compounds based upon eu and gd rare earths , the noble metals ag and au , and mg.@xcite in particular through the past ten years their synthesis , chemical and physical properties have been studied in some detail.@xcite the eu - based crystallize with the orthorhombic tinisi structure , while the gd - based materials adopt the zrnial structure .
recently , the electronic structure has been determined by x - ray photoelectron spectroscopy and compared with lda+u band structure calculations.@xcite here the valency , eu@xmath0 and gd@xmath1 , was firmly established and a variety of @xmath3-conduction bands were found at the fermi level .
one unusual feature of the photoemission spectra was the localized nature of the ag and au @xmath4-bands below @xmath5 .
since both eu@xmath0 and gd@xmath1 possess large magnetic moments ( @xmath6 ) , their magnetic ordering properties , mediated by the oscillating amplitude rkky interaction , may strongly depend on spin - orbit and spin - lattice couplings .
a recent investigation has characterized many of the bulk thermodynamic and transport properties related to the ferromagnetic transitions of euagmg , euaumg and gdagmg and the antiferromagnetic one of gdaumg.@xcite information about the above - mentioned magnetoelastic behavior can be obtained from measurements of the thermal expansion . in this work we present the results of thermal expansion measured over a large temperature range ( 4 to 200 k ) in applied magnetic fields reaching 14 t. we observe sharp peaks at the magnetic phase transitions in the uniaxial thermal expansion @xmath7 , where @xmath8 is the sample length and @xmath9 its temperature - induced change . in all four compounds we observe pronounced anomalies @xmath10 at the magnetic ordering transitions . for euagmg and gdaumg
the sign of @xmath10 is positive , while it is negative for euaumg and gdagmg .
these anomalies are smeared and/or displaced upon applying the magnetic field . by comparing @xmath11 with the corresponding anomalies @xmath12 of the specific heat @xcite at @xmath13
we obtain the pressure dependences of the critical temperatures for a first- or second - order phase transition via the clausius - clapeyron or the ehrenfest relation , respectively .
the results are most interesting for the ferromagnets gdagmg and euaumg , where we find a first- and a second - order phase transition , respectively , with very large negative pressure dependences of @xmath14 . in both materials , a finite pressure of the order of a few gpa should drive @xmath14 to zero suggesting pressure - induced quantum phase transitions .
polycrystalline samples of the above compounds were synthesized , annealed and characterized as described previously.@xcite the thermal expansion was measured in a capacitance dilatometer inserted in a @xmath15he gas - flow cryostat covering a temperature range from about 2 to 300 k.@xcite by using a superconducting magnet , fields up to 14 t could be applied over the entire temperature range .
accordingly , the length changes @xmath16 could be accurately detected through the magnetic transitions with and without the external field .
( a ) thermal expansion @xmath2 , ( b ) relative length change @xmath16 , and ( c , d ) specific heat of antiferromagnetic gdaumg for different magnetic fields.,scaledwidth=85.0% ] figure [ gdau](a ) displays @xmath17 in various magnetic fields for gdaumg .
the antiferromagnetic ordering at @xmath18 k causes a step - like anomaly @xmath19 of positive sign , which is typical for a second - order phase transition without pronounced fluctuations .
a very similar anomaly is present in the specific heat as shown in figure [ gdau](c ) .
the enhanced thermal expansion below @xmath20 signals a spontaneous contraction in the magnetically ordered phase , which is also directly seen in @xmath21 displayed in figure [ gdau](c ) .
the application of a magnetic field causes a systematic shift of the anomalies @xmath10 and @xmath22 towards lower temperature , signaling a decreasing @xmath23 .
as expected the antiferromagnetic order is destabilized by a magnetic field , although the rate @xmath24
k / t in 14 t appears relatively large in view of the high @xmath20 .
thermal expansion @xmath2 of the ferromagnetic compounds ( a ) euaumg and ( b ) euagmg for various magnetic fields .
the respective insets ( c ) and ( d ) show the relative length changes @xmath16 for zero field.,scaledwidth=85.0% ] figure [ eu ] shows @xmath17 in various fields for euaumg and euagmg , which order ferromagnetically at @xmath25 k and 22 k , respectively . in zero field
we again find a step - like anomaly @xmath10 for euaumg .
however , in contrast to gdaumg the sign of @xmath10 is negative meaning that the ferromagnetic ordering in euaumg is accompanied by a spontaneous expansion as shown in figure [ eu](c ) . for euagmg the sign of the zero - field anomaly @xmath10 is again positive , i.e. a spontaneous contraction occurs , but in this case @xmath10 has a lambda - like shape as it is typical for a second - order phase transition where the fluctuations are more pronounced .
such a difference in the anomaly shapes is also present in the specific heat anomalies.@xcite for both eu - based compounds the application of a magnetic field causes a drastic broadening of the anomalies .
this is typical for a ferromagnet because a sizeable dmagnetic field induces a strong magnetization already well above the zero - field @xmath14 . as a consequence
the magnetization does not develop spontaneously below a critical temperature anymore and , strictly speaking , a ferromagnetic transition temperature can only be defined for zero magnetic field .
( a ) thermal expansion @xmath2 and ( b ) relative length changes @xmath16 ( b ) of the ferromagnet gdagmg for various magnetic fields .
the inset ( c ) shows an expanded view of @xmath2 in order to visualize the additional anomaly at 125 k.,scaledwidth=85.0% ] in the ferromagnet gdagmg we observe a huge anomaly @xmath26 of negative sign at @xmath27 k as is illustrated in figure [ gdag](a ) .
the shape of the @xmath2 anomaly corresponds to an almost jump - like change of @xmath28 as shown in figure [ gdag](b ) .
this is a clear indication for a first - order transition in gdagmg , in agreement with our conclusions from the specific heat measurements , where a very similar anomaly shape is present at @xmath14.@xcite moreover , we observe an additional smaller anomaly of @xmath2 at @xmath29 k which hardly changes in an applied magnetic field , see figure [ gdag](c ) .
in contrast , the sharp low - temperature anomaly drastically broadens and shifts upwards in temperature . in the highest field
this broadening even extends to temperatures above about 175 k , i.e. to @xmath30 .
based on an analysis of the entropy change we suspected that the complete magnetic ordering in gdagmg might be achieved via both transitions at @xmath14 and @xmath31.@xcite however , the transition at @xmath31 is not seen in our measurements of the magnetic susceptibility @xmath32 . above about 50 k
, @xmath32 shows clear curie - weiss behavior , i.e. @xmath33 with @xmath34 k and @xmath35 close to the expected values of a ferromagnet with an @xmath36 and @xmath27 k. this and the very different field dependences , in particular the fact that the field - induced broadening of the @xmath14 anomaly even exceeds to @xmath37 , makes a common origin of both anomalies very unlikely .
since @xmath31 does not change with field , one may suspect a structural origin of this anomaly , however , no low - temperature diffraction data have been recorded yet .
alternatively , the @xmath31 anomaly could arise from small amounts of impurity phases , e.g. gdag and gdmg .
although our sample shows phase purity in x - ray powder diffraction , we can not exclude the presence of a few % of such impurity phases .
both of these binary intermetallic compounds exhibit magnetic phase transitions between 96 and 130 k.@xcite in gdmg a ferromagnetic order occurs at 120 k followed by a canting transition at 96 k , while a single antiferromagnetic phase develops in gdag below 133 k. the presence of gdmg appears also to be unlikely , since even for a few % of such an impurity phase the spontaneous magnetization due to the ferromagnetic order should be visible in @xmath38 and the anomaly should broaden in a magnetic field .
in contrast , the presence of a few % of gdag can not be excluded from our data : ( i ) due to the large background of the paramagnetic majority phase the expected weak magnetization change at @xmath20 can not be resolved in @xmath38 , and ( ii ) magnetic fields up to 14 t will not affect the antiferromagnetic order since the ordering temperature is so high .
because the @xmath31 anomaly most probably arises from an impurity phase , it will not be considered in the following discussion of the pressure dependences .
the measurements of thermal expansion and specific heat at ambient pressure allow to derive the initial slope of the change of @xmath39 under finite pressure using either the clausius - clapeyron for a first - order transition @xmath40 or the ehrenfest relation for a second - order phase transition . @xmath41
the jump - like changes of the volume and entropy are obtained via integration @xmath42 and @xmath43 , respectively .
the temperature range of integration around @xmath39 reflects the broadening of the first - order phase transition , which is always finite in a real solid .
we note that one has to use @xmath44 here and in equation ( [ ehre ] ) , since the hydrostatic pressure dependence is related to the volume expansion @xmath45 , which is given by @xmath46 for a homogeneous polycrystal .
.some characteristic properties of @xmath47 mg . [ tab ] [ cols="^,^,^,^,^ " , ] the obtained pressure dependences @xmath48 together with some other characteristic properties are given in table [ tab ] for all four compounds . obviously , we obtain rather large absolute values of @xmath49 of different signs depending on the sign of the respective @xmath2 anomaly .
these large values confirm the presence of a large spin - lattice coupling in these @xmath47 mg compounds , which we attribute to the oscillatory nature of the rkky interaction .
this may also be the main reason for the variation from ferro- to antiferromagnetic order and the wide range of different transition temperatures in the different @xmath47 mg compounds , although their structures are not too different .
we find the largest pressure dependence in gdagmg and associate this with the first - order nature of the magnetic transition . in order to drive a magnetic phase transition to first order
a strong spin - lattice coupling is required and this would also provide the mechanism for a large pressure dependence @xmath49 . we emphasize that our estimates of @xmath48 based on the ehrenfest or the clausius - clapeyron relation only yield the initial slopes at ambient pressure and extrapolations of our data to finite pressure have to be treated with caution .
nevertheless , the obtained negative values of @xmath50 and @xmath51 k / gpa for euaumg and gdagmg , respectively , are so large , that a complete suppression of the ferromagnetic order may be reached for both compounds when a rather moderate hydrostatic pressure of a few gpa is applied .
therefore both compounds are interesting candidates where pressure - induced quantum phase transitions can be studied . in summary , due to the significant spin - lattice coupling we were able to probe the magnetic ordering transitions of euagmg , gdaumg , euaumg and gdagmg via thermal expansion measurements .
using also specific heat data we were able to derive the initial pressure shifts of @xmath13 , which are rather large for all four compounds .
the largest effects are observed in euaumg and gdagmg and in both materials the ferromagnetic order is expected to be completely suppressed by hydrostatic pressure of a few gpa .
direct high - pressure experiments of @xmath39 on these compounds are highly desirable .
we thank the degussa - hls ag for a generous gift of noble metals .
this work was supported by the deutsche forschungsgemeinschaft through the priority programme spp 1166 `` lanthanoidspezifische funktionalitten in molekl und material '' . |
|
tropical geometry is rapidly developing as a new study area in mathematics . in @xcite , mikhalkin counts nodal curves on toric surfaces , which is an epoch - making result in algebraic geometry .
though there are several studies concerning tropical curves corresponding to singular algebraic curves , see for instance @xcite , the study of relation between theory of tropical curves and the singularity theory is still underdeveloping in tropical geometry . in singularity theory ,
the newton diagram is an important tool to get information of singularities .
let @xmath0 be a polynomial of two variables over @xmath1 , and suppose that @xmath2 and @xmath0 has an isolated singularity at @xmath3 .
set @xmath4 , and @xmath5 , where @xmath6 is the convex hull .
the convex hull @xmath7 is called the _
newton polytope _ of @xmath0 .
let @xmath8 be the polyhedron defined by @xmath9 where @xmath10 is the closure with usual topology of @xmath11 .
the _ newton boundary _ @xmath12 of @xmath0 is the union of compact faces of @xmath13 .
the singularity @xmath14 is _ convenient _ if @xmath8 is compact .
the singularity @xmath14 is _ newton non - degenerate _
if , for any face @xmath15 in @xmath12 , the function @xmath16 has no singularity in @xmath17 .
note that , for a convenient and newton non - degenerate singularity , the milnor number of @xmath14 is computed from @xmath18 , which is the celebrated theorem of kouchnirenko @xcite . in this paper , we regard @xmath8 as a part of polyhedrons obtained as a dual subdivision of a tropical curve and give a meaning of @xmath8 from the viewpoint of tropical geometry . to state our result
, we here introduce some terminologies in tropical geometry .
let @xmath19 be a polynomial of two variables over the field @xmath20 of convergent puiseux series over @xmath1 .
the _ tropical curve _
@xmath21 is defined by the image of @xmath22 by the valuation map , which is a 1-simplicial complex in @xmath11 .
the valuation of the coefficients of @xmath19 induces a subdivision @xmath23 of the newton polytope @xmath24 and it is known that @xmath23 is dual to @xmath21 , which is so - called the _ duality theorem _ ( see 2 ) .
now we consider a union of polygons corresponding to a part of polygons of @xmath23 .
let @xmath25 be a sub - polyhedron of @xmath26 .
a subset @xmath27 of @xmath21 is called the _ tropical sub - curve with respect to @xmath25 _ if @xmath25 is a union of sub - polyhedrons of @xmath23 and @xmath27 is dual to the subdivision of @xmath25 induced by @xmath23 .
we denote it by @xmath28 .
note that if @xmath29 then @xmath30 .
see definition [ resttrop ] for the precise definition of @xmath28 . for plane curve singularities , the real morsification due to acampo @xcite and gusein - zade @xcite gives an explicit way to understand mutual positions of vanishing cycles .
our hope is that we can perform the same observation for tropical curves realized in @xmath8 .
the main theorem in this paper asserts that we can see the vanishing cycles of @xmath31 on the tropical curve in @xmath8 . for a tropical sub - curve @xmath28 ,
let @xmath32 denote the number of 4-valent vertices of @xmath28 and @xmath33 denote the number of regions bounded by @xmath34 .
[ mainthm]_for any newton non - degenerate and convenient isolated singularity @xmath14 , there is a polynomial @xmath35 $ ] such that @xmath36 and @xmath37 satisfies @xmath38 _ this result is an analogy of the following equality required for a real morsification : @xmath39 where @xmath40 is a real morsification of @xmath0 , @xmath41 is the number of double points of @xmath42 in a previously fixed small neighborhood @xmath43 of the origin , and @xmath44 is the number of bounded regions of @xmath45 . as a corollary of theorem [ mainthm ]
we have the equality @xmath46 , where @xmath47 is the number of double points of @xmath14 , see corollary [ cor ] . remark that the polynomial @xmath19 in theorem [ mainthm ] is given by a patchworking polynomial asssociated with a subdivision of @xmath26 .
a similar observation appears in a paper of shustin @xcite , where a real polynomial whose critical points has a given index distribution is constructed .
we organize the paper as follows . in section 2
, we introduce tropical curves and subdivisions of newton polytopes induced from the valuation map and state the duality theorem . in section 3 ,
we give the definition of tropical sub - curves and prove theorem [ mainthm ] .
two examples will be given before the proof .
i am grateful to masaharu ishikawa and takeo nishinou for fruitful discussions .
i would like to thank eugenii shustin for precious comments and telling me about his previous result .
i also thank nikita kalinin for giving me interesting comments .
first , we give some definitions about polytopes . a _ polygon _ in @xmath11 is the intersection of a finite number of half - spaces in @xmath11 whose vertices are contained in the lattice @xmath48 .
a polygon is called a _ polytope _ if it is compact . in this paper ,
a _ polyhedron _ means a union of polytopes which is connected and compact .
thus a polyhedron is not convex generally .
a subset in a polyhedron is a _ sub - polyhedron _ if it is a polyhedron as a subset of @xmath11 . in particular ,
if a sub - polyhedron is a polytope then the sub - polyhedron is called a _ sub - polytope_. let @xmath49 be the field of convergent puiseux series over @xmath1 , and denote the usual non - trivial valuation on @xmath20 by @xmath50 , that is , @xmath51 where @xmath52 and @xmath53 . for a reduced polynomial @xmath54\ ] ] over @xmath20 , we denote by @xmath55 and @xmath26 the _ support _ of @xmath19 and the _ newton polytope _ of @xmath19 , respectively , that is , @xmath56 and @xmath57 . throughout this paper , we assume that the newton polytope @xmath26 of @xmath19 is 2-dimensional . the valuation map @xmath58 is defined by @xmath59 , which is a homomorphism . the closure @xmath60 with usual topology on @xmath11 is called the _ ( plane ) tropical curve defined by @xmath19_. remark that a tropical curve @xmath21 has a structure of a @xmath61-dimensional simplicial complex , cf.@xcite .
we call a @xmath61-simplex an _ edge _ and a @xmath62-simplex a _ vertex _ as usual .
let @xmath19 be a polynomial over @xmath20 .
we introduce the duality theorem which gives a correspondence between the tropical curve @xmath21 defined by @xmath19 and the newton polytope @xmath26 of @xmath19 .
let @xmath63 be the 3-dimensional polygon defined by @xmath64 and @xmath65 be the function defined by @xmath66 which is a continuous piecewise linear convex function .
we can get the following three kinds of sub - polytopes of @xmath26 from @xmath67 : * linearity domains of @xmath68 : @xmath69 , * 1-dimensional polytopes : @xmath70 and @xmath71 , * 0-dimensional polytopes : @xmath72 , where a _ linerity domain _ of @xmath68 means a maximal sub - polytope @xmath73 of the domain @xmath26 such that the restriction @xmath74 is an affine linear function .
these polytopes give a subdivision of @xmath26 , which we denote by @xmath23 .
in particular , we call a 1-dimensional and a 0-dimensional polytope a _ vertex _ and a _ edge _ of @xmath23 , respectively . for edges of a tropical curve ,
a certain weight @xmath75 is defined by directional vectors of edges canonically .
we omit the definition since we do nt use it in this paper .
we can find the following claim in 2.5.1 of @xcite .
a proof in general dimension can be found in @xcite .
[ dualthm]_the subdivision @xmath23 is dual to the tropical curve @xmath21 in the following sense : + ( 1 ) the components of @xmath76 are in 1-to-1 correspondence with the vertices of the subdivision @xmath23 , + ( 2 ) the edges of @xmath21 are in 1-to-1 correspondence with the edges of the subdivision @xmath23 so that an edge @xmath77 is dual to an orthogonal edge of the subdivision @xmath23 , having the lattice length equal to @xmath78 , which is a weight of @xmath79 , + ( 3 ) the vertices of @xmath21 are in 1-to-1 correspondence with the polytopes @xmath80 so that the valency of a vertex of @xmath21 is equal to the number of sides of the dual polygon .
_ we call the subdivision @xmath23 the _ dual subdivision _ of @xmath21 . by theorem [ dualthm ]
, we can regard a tropical curve as a dual subdivision of the newton polytope of its defining polynomial .
in this section , we first introduce the precise definition of tropical sub - curves which we mentioned in the introduction .
let @xmath81 $ ] be a polynomial over @xmath20 and @xmath82 be the dual subdivision of @xmath21 . because of the structure theorem in tropical geometry , a tropical hypersurface has the structure of a polyhedral complex . in particular ,
a plane tropical curve is an embeded plane graph in @xmath11 .
let @xmath83 $ ] denote the edge of the tropical curve @xmath21 whose endpoints are 0-cells @xmath84 and @xmath85 .
set @xmath86\setminus \{v\}$ ] .
we allow that one of the endpoints is at @xmath87 . in this case , the other endpoint is contained in the 0-cells of the curve .
note that @xmath88=[u,\infty)$ ] .
let @xmath89 be a subset of @xmath23 such that @xmath90}(f)$ ] is connected , where @xmath91}(f)$ ] is the set of vertices of @xmath92 .
let @xmath25 be a sub - polyhedron of @xmath24 given as the union of @xmath93 .
[ resttrop ] a subset of @xmath21 is called the _ tropical sub - curve with respect to @xmath25 _ if it has the structure of the metric ( open ) sub - graph @xmath96 of the tropical curve @xmath21 which satisfies the following conditions : + ( 1 ) the set of vertices @xmath97 is given by @xmath98 , + ( 2 ) the set of edges @xmath99 is given by the following manners : for each @xmath83 \in e$ ] , [ resttropex ] ( 1 ) let @xmath19 be a polynomial over @xmath20 given by @xmath106 the newton polytope @xmath26 of @xmath19 is @xmath107 .
see on the left in figure 1 .
this polynomial @xmath19 is @xmath108 in theorem [ mainthm ] for the singularity of @xmath109 at the origin .
the polyhedron @xmath25 in the figure is @xmath18 for the singularity @xmath14 .
the tropical sub - curve @xmath37 with respect to @xmath18 is as shown on the right . since @xmath110 and @xmath111 , the equality @xmath112 in theorem [ mainthm ] is verified .
\(2 ) let @xmath19 be a polynomial over @xmath20 given by @xmath114 the newton polytope @xmath26 of @xmath19 is @xmath115 see on the left in figure 2 .
this polynomial @xmath19 is @xmath108 in theorem [ mainthm ] for the singularity of @xmath116 at the origin .
the polyhedron @xmath26 in the figure is @xmath18 for the singularity @xmath14 .
the tropical sub - curve @xmath37 with respect to @xmath18 is as shown on the right . since @xmath117 and @xmath118 , the equality in theorem [ mainthm ] holds .
suppose that @xmath0 is convenient .
for the lattice points @xmath119 , we define a map @xmath120 by @xmath121 where @xmath122 are non - negative strictly increasing sequences of integers .
we then extend it to the whole domain @xmath18 as a continuous piecewise linear function and obtain a map @xmath123 .
taking sufficiently large values for @xmath124 at the lattice points of the newton boundary of @xmath0 , we may assume that the other sub - polytopes are triangles with area @xmath125 .
let @xmath129 be the rectangle given by @xmath130 we consider the special subdivision of @xmath129 .
we decompose it into @xmath131 vertical rectangles @xmath132 \times \mathbb{r})\cap \hat{\delta } \subset \hat{\delta } , \ \
\ i=0 , \dots , p-1.\ ] ] the special subdivision of @xmath129 induces a special subdivision of each @xmath133 .
let @xmath134 be the segment connecting @xmath135 and @xmath136 .
we denote by @xmath137 the number of special squares in @xmath129 which intersect @xmath134 .
similarly , we denote by @xmath138 the number of special squares in @xmath133 which intersect @xmath134 .
obviously @xmath139 .
let @xmath140 be the number of special squares of @xmath141 .
notice that @xmath142 since @xmath143 .
thus it is enough to show @xmath144 . without loss of generality
, we may assume @xmath145 .
let @xmath146 , @xmath147 be integers such that @xmath148 and @xmath149 .
the segment @xmath134 can be denoted as @xmath150.\ ] ] set @xmath151 . then , @xmath138 is calculated as @xmath152 for @xmath153 and @xmath154 where @xmath155 means the largest integer not greater than @xmath156 .
thus , we obtain @xmath157 choose a polynomial @xmath19 such that the newton polytope @xmath26 coincides with @xmath158 . to decide coefficients of @xmath19 , we take the convex function @xmath159 as a linear extension of @xmath124 used in the definition of the special subdivision of @xmath18 , and define @xmath19 as the _ patchworking polynomial _
defined by @xmath160 , that is , @xmath161 in the rest of the proof , we check @xmath113 satisfies the equality in the assertion . to calculate the number of special squares , we decompose @xmath8 into two sub - polyhedrons as follows .
let @xmath126 be coprime integers .
we denote the intersection points of the newton boundary @xmath12 of @xmath0 and the lattice by @xmath162 where @xmath163 .
we set @xmath164 and define @xmath165 notice that @xmath166 and @xmath167 . for @xmath168
, we define the subset @xmath169 of @xmath8 as @xmath170 and @xmath171 then , @xmath8 decomposes as @xmath172 . for @xmath173 ,
we denote by @xmath174 the number of special squares contained in the special subdivision of @xmath175 induced by that of @xmath8 .
then , using lemma [ lem1 ] , we have @xmath176 next we will show the following equalities : @xmath177 by theorem [ dualthm ] , the correspondence between subdivisions and tropical curves , introduced in theorem [ dualthm ] , gives a 1-to-1 correspondence of parallerograms and 4-valent vertices . in our case , any 4-valent vertex corresponds to a special square .
thus , the number of special squares , @xmath178 , coincides with the number of 4-valent vertices of @xmath113 .
hence equality ( 1 ) holds .
we prove the other equality .
there is a 1-to-1 correspondence between @xmath179 and @xmath180 where @xmath181 is the set of vertices of a special square @xmath182 in @xmath8 .
moreover , by the duality theorem [ dualthm ] of tropical curves , we have a 1-to-1 correspondence between the bounded regions contained in the complement of @xmath183 and the interior lattice points @xmath184 in @xmath18 .
since @xmath185 we get @xmath186 thus equality ( 2 ) holds . in @xcite
, we have @xmath196 , where @xmath197 is the number of local irreducible components of @xmath0 at @xmath62 .
we also have @xmath198 from the argument in the proof of theorem [ mainthm ] .
thus @xmath199 as in @xcite , we can obtain the intersection form of vanishing cycles from the immersed curve of a real morsification . to study the intersection form in our tropical curve we need to fix framings " on edges of the curve in @xmath8 , though we do nt have any good way to see these framings " . | for a convenient and newton non - degenerate singularity , the milnor number is computed from the complement of its newton diagram in the first quadrant , so - called kouchnirenko s formula . in this paper
, we consider tropical curves dual to subdivisions of this complement for a plane curve singularity and show the existence of a tropical curve satisfying a certain formula , which looks like a well - known formula for a real morsification due to acampo and gusein - zade .
[ section][thm]corollary[thm]lemma[thm]proposition[thm]definition [ thm]remark[thm]example |
* overview : * recent improvements of atomistic forcefields for nucleic acids @xcite and advances in a particular family of thermodynamic integration techniques @xcite together encourage a return to what we believe should be considered , due to its combination of simplicity and intractability , as a canonical model system to benchmark computational methods for biomolecules and polyelectrolytes : the handedness - switching b to z isomerisation of dna . here
we present a new addition to the progression of free energy methods beginning from the `` einstein crystal '' of frenkel & ladd @xcite , followed by the `` einstein molecule '' @xcite , the `` confinement method '' @xcite and a recently introduced analytical reference - liquid model @xcite .
we demonstrate this new method using the example of the transition between b and z dna . in the absence of an existing umbrella term for this philosophy of free energy calculation
, it is referred to here as thermodynamic integration relative to an exact solution ( ties ) because it relies on performing a thermodynamic integration ( ti ) calculation with a realistic ( but intractable ) model at one endpoint of the integration path and an exactly solvable ( es ) model at the other . the principal technical advance within the field of ties to be presented here is in the adaptation of a recently developed reference liquid model @xcite such that it can be used for triangular water in a molecular dynamics simulation , and successful combination of this adapted liquid model with the existing einstein molecule ( em ) model for the solute . for the ` realistic ' , but intractable , end of the integration path , the amber simulation code @xcite and
selected amber forcefield parameters were used .
the details of this quantitative all - atom treatment ( with coulomb , bonded and van der waals interactions ) are given in the methods .
the results of the example calculation are found to be of groundbreaking accuracy for the example considered , although they were computationally expensive and scope for future gains in efficiency is acknowledged .
* b - z dna isomerisation : * the b - z isomerisation of duplex dna is a gross structural transition , from a right - handed double helix ( the canonical b - form ) to a left - handed double helix ( the z - form , which occurs _ in - vitro _ at high salt concentration @xcite or _ in - vivo _ with the assistance of dna - binding proteins @xcite and/or with negative supercoiling @xcite ) .
occurrence of z - dna in mammalian cells has been implicated as a causative factor for certain cancers @xcite . in low salt
the ( more compact ) z - dna is disfavoured relative to b - dna for reasons of electrostatics and solvation ; it becomes free - energetically favourable in total when a high salt concentration screens the repulsion between the charged backbone phosphate groups , also making partial solvent - exposure of the base pairs more favourable ( [ fig : structs ] ) .
the unwinding which is required to change the handedness of the double helix presents a formidable energetic barrier which must be overcome for the transition to take place @xcite , even when the ends of the duplex are free . ] @xmath0 dna , after 12ns equilibration in 1 molar nacl .
the backbone phosphate groups are closer together in the more compact z isomer , and the base - stacks have greater solvent exposure.,scaledwidth=50.0% ] the salt concentration for coexistence of these two forms is known from circular dichroism measurements to be roughly 2.8 m nacl for [ d(cg)]@xmath1 ; this varies with sequence , temperature , chain length and methylation @xcite .
the availability of this experimental data , particularly for short dodecamer sequences , has attracted several simulation studies .
definitive thermodynamic characterisation with respect to salt concentration has remained just over the horizon for some years , however recent improvements have been made : 3drism has been qualitatively successful in finding the coexistence point @xcite ( estimated at 0.8 m ) ; and targeted molecular dynamics simulations have had at least some success in describing the transition path @xcite although this last matter is still difficult to compare directly against experiment . * previous calculations of the energetics : * the history of computational studies of the b - z transition from a free energy point of view is , in microcosm , the history of developing computational methods for molecular electrostatics and solvation .
early treatments used continuum electrostatic models and rigid - body treatments of the dna , either based on static atomic structures @xcite or on geometric constructs such as grooved cylinders @xcite or double - helical chains of charged beads @xcite . a review article in 1987@xcite pointed out that results from continuum - electrostatics seemed to depend dramatically on the assumptions made for the various geometric and physical parameters .
the same review article also ruled out an all - atom simulation approach for the near future on grounds of computational expense ( a view still advanced some 23 years later by researchers from another group@xcite ) . a paper in 1997 @xcite responded to the perceived limitations of a continuum treatment of the ions , by making a hybrid approach with discrete counterions in continuum solvent .
the same paper pointed out that treatments which give a coexistence concentration without calculating the non - electrostatic free energy difference are either right for the wrong reasons ( if they ignore this contribution ) or cheating somewhat , by setting the non - electrostatic free energy change to whatever value gives the correct coexistence point ( e.g. ref .
@xcite ) . in
the ` right for the wrong reasons ' category falls one of the most accurate results , based on a classical density - functional theory ( dft ) approach , which gave qualitative agreement of 3.6 m nacl to the infinite - chain coexistence value of 2.25 m as early as 1989 @xcite , however the model used was without any detailed representation of the dna .
when the same author returned to the problem 21 years later with a similar but more detailed model taking into account the atomic structure of the water molecules and the dna , the final value for the coexistence was in fact less accurate than the older one @xcite , at 0.8 m ( compared to the dodecamer coexistence value of 2.8 m ) , although of course more insight was available from the use of the richer model .
the calculation we present gives the most accurate value for the position of this conformational equilibrium from an all - atom treatment , and also the most accurate without fitting parameters to date ( yielding a coexistence concentration of 2.5 m nacl compared to the 2.86 m observed experimentally by pohl & jovin in 1972@xcite ) . with the understanding that calculations from other methods can of course provide insight in different ways ,
the previous values reported in the literature were , in chronological order : 2.25m@xcite ( exactly right for the infinite - chain , but with na@xmath2 hard - sphere radius set as a free parameter to 4.95 vs. the crystal ionic radius of 0.12@xcite ) , 3.6 m ( vs. 2.25 m for the infinite - chain)@xcite , no transition@xcite , 3.7 m ( vs. 2.4 m given by them as the infinite - chain value)@xcite and 0.8m@xcite ( vs. 2.86 for the dodecamer ) . given the difficulty of finding the correct coexistence value , some papers have drawn attention to the gradient of the free energy with respect to concentration , comparing their results with a fit to the empirical data of : @xmath3}{c } \label{eqn : jofit}\ ] ] where @xmath4 , @xmath5 is the free energy difference per base pair and @xmath6 is the coexistence concentration @xcite .
this fit was derived based on data at the long - chain limit ( @xmath7 , @xmath8 ) .
there has not been a great deal more success from theory in matching the observed gradient @xmath9 $ ] of the free energy near to coexistence than the coexistence point itself , and the fit of a simple logarithm to the free energy has been brought into question by more recent experimental work pointing to the existence of intermediates @xcite . the gradients near to coexistence of some reported theory treatments were : @xmath10@xcite , @xmath11@xcite and @xmath10@xcite .
one theory treatment gave an elbow in the plot , with a value below coexistence of approximately @xmath10 , and above coexistence of @xmath7@xcite .
* thermodynamic integration relative to exact solution : * this established and powerful family of thermodynamic integration ( ti ) techniques has recently been extended to permit application to liquids @xcite . in brief , the method consists of defining for a given system a reference hamiltonian @xmath12 which has an analytically tractable expression for its free energy ( described in the supplementary data ) ; and also a realistic but intractable @xmath13 ( here , the amber hamiltonian ) . a third , ` mixed ' hamiltonian is then defined as a sum of the two : @xmath14 here @xmath15 is introduced as a control variable of the system such that for @xmath16 , @xmath17 , and @xmath18 ;
while for @xmath19 , @xmath20 . the helmholtz free energy @xmath21 of the system under the realistic ( but intractable ) hamiltonian @xmath22 can then be expressed in terms of an integral with respect to @xmath15 : @xmath23 where @xmath24 and @xmath25 refer to the free energy and hamiltonian of the reference system ; @xmath26 and @xmath27 are mixing functions used to control the speed with which @xmath25 is introduced in place of @xmath22 over the interval @xmath28 $ ] ; and @xmath29 refers to the ensemble average for a given value of @xmath15 .
each ` generalised force ' @xmath30 is found by collecting time - averages of the partial derivatives of @xmath31 while the system evolves under this mixed hamiltonian at fixed @xmath15 .
( here , obviously , the sampling times need to be long enough to allow for identification of the ensemble - average with the time - average . )
while the time - averaged generalised force may be slow to converge , this procedure has the advantage that calculations at each value of @xmath15 can be carried out in parallel , without any need for intercommunication .
* einstein molecule & cage - swapping liquid model : * the first presentation of ties was in the context of an @xmath25 appropriate for solids or for macromolecules in vacuum , due to frenkel and ladd @xcite with modifications by vega @xcite , and is known as the `` einstein molecule method '' ( em ) or as `` the confinement method '' .
the basic idea of the em reference model is that the particles do not interact with each other , so that the reference partition function factorizes and the reference free energy can easily be computed exactly . in the simplest such model , the particles are simply coupled to predetermined reference positions by harmonic springs .
the em model was used here for the dna chains . the reference model used for the water and counterions in this calculation , however , is still quite new , and has been slightly modified from its most recent appearance in the literature @xcite for use in conjunction with md rather than monte carlo sampling . the em model of harmonic wells can not be used unmodified as a reference for the liquid state . in the liquid
each particle is ultimately free to move throughout the volume ; so the range of the harmonic well in the em model would have to be infinite as well .
thus to sample the average in ( eq .
3 ) one would need to sample for infinite lengths of time .
instead we use a potential which is attractive at a short range from the reference position , but flat at long range ( [ fig : emcs ] ) .
the contribution of the flat part to the partition function of the mixed model is trivial , so there is no need to extensively sample the reference model outside the short range of attraction .
the use of a potential cutoff solves the problem of dealing with an infinitely - ranged well of attraction , but it introduces another problem : for a large system volume , a diffusing particle hardly ever re - visits its reference site , once it has left the attractive well .
we solve this problem by means of a monte carlo move in which we swap particle identities such that the diffusing particle has frequent opportunities ( via a metropolis acceptance rule ) to rejoin or part from its reference site .
this model is here called ` cs ' , with reference to the ` cage - swapping ' dynamical behaviour which it provides .
the ` cage - swapping ' move reflects the factorial term in the general partition function of liquids , related to the indistinguishability of particles . for a detailed explanation see schilling & schmid@xcite .
the alternative in the literature to the cs mc move is to solve the linear assignment problem of particle identities so as to minimise the particle - well interaction energy at each md timestep , as by tyka , sessions and clarke @xcite ( see also @xcite ) .
this has both advantages and drawbacks over the approach of using an mc move , and a thorough comparison might be valuable in a later publication .
the principal advantage of linear assignment appears to be that the size of the ( configuration @xmath32 identity ) space is reduced because the system always has the identity assignment which gives the mininum energy for the given configuration .
the principal advantage of mc cage - swapping would appear to be enhanced sampling , as nearby particles can exchange wells even if it is not energetically favourable to do so , with the well exchange creating an attractive force which pulls the particles to their new minima .
modifications made for the md / mc cs model used here relative to previous pure - mc treatments were to use md timesteps instead of those mc moves which moved particles spatially without changing their identities ; and to add a harmonic region near to the well - bottoms ; such that the wells were harmonic out to 1 , then constant - force out to 5 , then flat .
the small harmonic region was necessary for stability of the md .
a natural objection to the use of a swap move in conjunction with md simulation is to ask if it is required that the mc sampling should converge before each md timestep takes place ; mercifully it has been shown in other work that any combination of md and mc steps which would individually provide boltzmann sampling of the degrees of freedom to which they are applied will in total provide boltzmann sampling of the combined degrees of freedom @xcite . *
detailed motivation of the integration : * in a ti calculation , the free energy landscape of the system gradually morphs from one shape to another , in a way that is not yet easy to characterise . from an intuitive basis
, we can state qualitatively that for effective ti four sometimes - conflicting criteria are important : 1 .
to minimise the error from the numerical integration , the integration path should be as smooth as possible ; meaning that the higher - order derivatives of each generalised force @xmath33 should be as small as possible .
2 . the shapes of landscapes to be mixed should be similar to each other ; otherwise the generalised force due to @xmath22 , for instance , could sporadically take extremely large values when @xmath34 is small and @xmath25 is primarily controlling the dynamics .
3 . to minimise the convergence time per integration point
, exploration of the mixed landscapes should be as fast as possible .
4 . because the variance of the generalised force scales with the gradient of the mixing function @xmath35 , to efficiently distribute sampling between integration points
, the mixing functions should be as close to linear as is consistent with the other criteria . to meet criterion ( a )
, it is first required to avoid discontinuities in the generalised forces by making sure that the integration does not cross any first order phase transitions .
the basis of ties is a successful treatment of this issue by defining tractable models which match a given phase in terms of the symmetries expressed by their partition functions .
definition of the model partition functions is discussed in the supplementary information .
criterion ( b ) is more difficult .
the reference model that we use decouples all particles from each other ( in this way , the partition function is easily solved analytically ) .
thus at @xmath19 particles can freely move through each other , while for @xmath36 this is not possible .
therefore the landscape under @xmath25 is topologically different to that under @xmath22 .
this brings the question of appropriately changing between landscapes sharply into focus .
in order to treat cases where a cavity in the free energy landscape ( e.g. a region of infinite potential energy , such as due to the lennard - jones short range repulsion ) must be introduced , starting from a flat potential , it is valuable to gradually increase the maximum energy associated with that region to some high but finite value .
the region is only then blocked completely , by introducing the infinite potential inside the region which is already effectively forbidden by the large finite potential .
this strategy , which was attempted here by use of the ` guide hamiltonian ' discussed below , should ensure that regions of phase space which are populated for a given @xmath15 are also populated at neighbouring values of @xmath15 , while still allowing topological change of the landscape to occur . in the absence of such a technique ,
the generalised force can diverge , with an infinitesimal change in @xmath15 giving an infinite change in @xmath37 .
a further , partial , fix for problems with respect to criterion ( b ) is provided by choice of the mixing function used to introduce or remove a given hamiltonian .
careful work has been done on this in the context of alchemical ti involving lennard - jones and coulomb forces @xcite , which was broadly followed here : the essence of the strategy is to use polynomial mixing functions such that as the ability of a given hamiltonian @xmath31 to direct the system into its own minima drops with @xmath35 , the generalised force @xmath37 also drops .
the choice of mixing function must be balanced between criteria ( b ) and ( d ) .
criterion ( c ) is addressed here by defining @xmath25 , the cs hamiltonian , with few and low energetic barriers over the difficult ( @xmath38 ) part of the integration , such that exploration is relatively quick .
it has been suggested that it might also sometimes be valuable to retain barriers in the landscape as these can serve to reduce the effective dimensionality of the configurational space and actually accelerate exploration @xcite , however this effect was not observed here .
criterion ( d ) is addressed here by choosing a relatively low - order mixing function compared to some of those which have been trialled in other work @xcite . *
guide hamiltonian : * to manage the changes of the topology of the free energy landscape over the integration , a third hamiltonian , @xmath39 , was introduced in addition to @xmath40 and @xmath41 , with a mixing function such that it would be zero at both endpoints of the path . the weak short - range interparticle repulsion provided by @xmath39 controlled the topology in the sense that it was used to first introduce gently sloping energy barriers around those states which would later take on infinite energy becoming topological ` cavities ' in the free energy landscape .
the unshielded existence of such a cavity is intolerable if carrying out md , because this leads to infinite values for the forces between particles .
this use of a guide hamiltonian , introduced here , is an alternative to the practice of adding a @xmath15-dependence to the functional form of the lennard - jones ( lj ) potential such that it is finite over all configurations for @xmath42 , known as ` softcoring ' @xcite .
softcoring of the lj typically follows mixing - out of the coulomb interaction ( which also has a discontinuity at zero separation ) , in a multi - step procedure which increases computational expense .
such a process is also undesirable because it has the potential to introduce first order phase transitions : the phase diagram of water , for instance , is dramatically altered when electrostatics are removed .
initial attempts were made to softcore the coulomb and lj terms together in such a way as to preserve the shape of the energy landscape and this was found to be prohibitively complex , especially given the requirement for the sake of efficiency to treat the coulomb interactions using a fast algorithm such as the ewald sum .
the use of an @xmath39 has the advantages of solving at a stroke discontinuities of both lj and coulomb forces , of being extremely simple to implement and to parameterise , and of introducing no extra code into the complex and highly optimised non - bonded force and ewald sum routines of the molecular dynamics program .
the short - range repulsion is handled using a neighbour list and therefore adds only a small ( and linearly scaling ) cost to the calculation .
the guide hamiltonian was defined very simply as a quadratic repulsive potential between all heavy atoms , where the cutoff @xmath43 was defined as the radius giving the minimum of the van der waals interaction for the given atom pair @xmath44 , multiplied by 0.88 .
@xmath45}{r_{ij}}\right)^2\end{aligned}\ ] ] * mixing function design : * numerical smoothness of the integration over @xmath15 depends critically on the choice of mixing functions used to turn the hamiltonians @xmath22 , @xmath25 and @xmath46 off and on .
the mixing functions were defined as polynomials : @xmath47 the choice of polynomial mixing functions ( which has been advised with order 4 or greater for alchemical transformations involving lennard - jones atoms
@xcite ) gives generalised forces @xmath48 and @xmath49 which are zero for @xmath19 and @xmath16 respectively , so that ( for example ) two particles can pass through each other without causing a singularity in the generalised force with respect to @xmath22 .
formally , the use of a polynomial mixing function is enough to stabilise the integration , mixing the two topologically distinct landscapes without introduction of singularities .
unfortunately the multiplication of very large by very small numbers is numerically abhorrent , therefore the guide hamiltonian was also required . in order to allow @xmath46 to take effect before removing the non - smooth hamiltonian @xmath22 ,
the mixing function @xmath26 , was rescaled such that the endpoint of the mixing would be at @xmath50 instead of at @xmath19 ( with @xmath51 chosen arbitrarily as 0.5 ) .
the actual function used in place of @xmath26 was therefore : @xmath52 $ ] .
the three mixing functions are shown in [ fig : mixfuncs ] .
controlled the amber hamiltonian @xmath22 , @xmath53 controlled the analytically tractable reference hamiltonian @xmath25 and @xmath54 controlled the ` path ' hamiltonian @xmath46.,scaledwidth=40.0% ]
* amber setup : * molar nacl concentration was defined as the ratio of na@xmath2 ions to h@xmath0o molecules multiplied by the molarity of water , 55.55 . @xmath22 was defined using the amber99 forcefield @xcite with the barcelona corrections to nucleic acid parameters @xcite , the joung - cheatham ion parameters @xcite and the tip3p water model @xcite .
water molecules were kept rigid using the shake algorithm .
temperature was held at 300k using a langevin thermostat with a coupling of 0.1ps@xmath55 .
the seeds for the noise - generator in the langevin thermostat were set to be the same for each pair of b and z at a given @xmath15 and [ nacl ] , in the hope of reaping some benefits by covariant sampling@xcite in the estimate of the free energy difference .
the molecular dynamics timestep was 2fs .
the ` pmemd ' dynamics engine@xcite provided in amber was used ( with default parameters ) to treat the electrostatic and other terms of the forcefield efficiently .
start configurations of [ d(cg)]@xmath1 dna double helices in the b and z conformations were prepared using the nab molecular building tool @xcite .
counterions were added near to the dna using the xleap component of amber so as to neutralise backbone charge , and further counterions were then added at random to bring the salt concentration up to the required molarity assuming 9441 water molecules .
water molecules were then added from a pre - equilibrated water box , deleting those which overlapped with an ion or dna atom , until 9441 were present .
the concentrations studied were 1.11 m , 1.61 m , 2.11 m and 2.61 m nacl .
the initial configurations were energy - minimised and then equilibrated at 300k and 1 atmosphere in the npt ensemble for 2ns .
the average volume of each system 1.11 - 2.61 m nacl was taken over the interval 1 to 2 ns , and the system box sizes in each configuration were set to these average volumes , so that the ti calculations could be carried out in the nvt rather than npt ensemble .
exploratory runs for the concentrations 1.11 m to 2.61 m nacl were made out to 20ns to check stability and to estimate the important time and lengthscales of the dynamics ( see supporting information ) . although the full ti calculations were not carried out at 2.86 m nacl , an initial 10ns test equilibration run was also made at this concentration ( see supporting information ) .
molecular images were prepared using vmd@xcite .
* analytical model setup : * in the reference model of the system , all solute atoms ( including hydrogens ) were placed in harmonic wells of spring constant 5k@xmath56t@xmath57 .
counterions and water oxygen atoms were assigned to three sets of identical cs wells . because the tip3p water molecules are treated as rigid triangles , it was not neccessary to restrain the water hydrogen atoms in addition to the oxygens .
the cs wells had a spring constant of 1.112k@xmath56t@xmath57 up to @xmath58 , then a constant force of 1.11k@xmath56t@xmath55 out to @xmath59 .
the ` cage - swapping ' reassignment of cs wells was carried out using a metropolis mc algorithm .
for each liquid molecule present in the system , 5 mc well - swapping attempts were made every timestep .
each system run under a mixed hamiltonian was allowed 500ps
( 250k timesteps ) of equilibration , before collecting the generalised force time series for a minimum further 500ps , depending on convergence .
the equilibrium work done on each integration path is shown in [ tab : freeenergies ] .
.equilibrium work over each integration path , and estimated standard errors . [ cols="<,^,^",options="header " , ] * free energies : * free energies were separately calculated for z and b conformations over a range of salt concentrations ( [ tab : freeenergies ] and [ fig : freeenergies ] ) .
the crossover between the b and z regimes occurred at around 2.5 m nacl , quite near to the 2.86 m found in the experiments of pohl and jovin @xcite .
as far as the authors are aware , this is the most accurate numerical calculation of the coexistence position ( without fitting free parameters ) to date .
the shape of the free energy difference away from coexistence does not seem to match the phenomenological equation proposed by pohl@xcite(eq . 1 )
, however given the small number of points and the large errorbars it is also difficult to completely discount this equation , or to suggest an alternative more complex functional form such as more recent theory @xcite and experiment@xcite suggesting the existence of intermediates seem to require .
a fit of eq . 1 with @xmath60 and @xmath6 free gives @xmath61 and the coexistence value @xmath62 m , although , because of the questionable agreement of the shape of eq . 1 with our data
, we prefer to report coexistence as 2.5 m based simply on the y - intercept of the trace between points three and four
. a possible explanation for the apparent overestimate of the free energy penalty for disfavoured forms ( the large value of the fitting parameter @xmath60 ) relative to the experimental data , could be that the experimentally studied left - handed dna structures at low salt are not the same as the conventional z - dna structure observed at high salt , which was used as the start configuration for the simulations at low salt .
if the starting configurations used were in fact strongly metastable with respect to some lower - energy left - handed form ( as well as with respect to the still - lower b - dna ) then this would explain the observed behaviour . a wider - ranging examination of the free energy landscape would be required to discuss this possibility further .
it was found that , while the z - dna structure used was strongly metastable at low salt concentrations , the more labile b - dna structure tended to melt or fray at 2.86 m nacl and above ( see supplementary data ) . the greater flexibility of b - dna , or the sharp elbow in the free energy difference above coexistence which has been suggested@xcite , might serve to explain this .
given that the watson - crick bonded b - dna structure was unstable with respect to a disordered structure , the ties calculation of this paper would have been difficult to apply without artificially enforcing w - c base pairing .
separate to the difficulty of such a calculation , would be the question of its interpretation in a regime where it is no longer clear that b is the dominant right - handed metastable conformation .
* integration path : * the character of the integration path can be seen by examining the generalised force due to each of the three hamiltonians with respect to @xmath15 ( [ fig : genforce ] ) . the smoothness of these curves ( apart from at @xmath63 , where the integration - out of the amber hamiltonian is completed ) indicates that no first - order phase transitions were crossed during the integration path , and also serves to justify the use of a basic simpson s rule integration scheme ; carried out separately over the three terms of the generalised force and over the [ 0,0.5 ] and the [ 0.5,1.0 ] intervals . ) , the em+cs hamiltonian ( @xmath25 ) and the ` guide ' hamiltonian used to prevent particle overlaps during the mid - section of the integration ( @xmath64 ) . the guide hamiltonian _
( right axis ) _ was much smaller than the main hamiltonians _
( left & right axes)_. a trace is shown for each of the eight calculations carried out , however b and z can not be distinguished on the scales used.,scaledwidth=49.0% ]
calculations were carried out using four 2.26 ghz processor cores per integration point ( therefore 68 per measurement of @xmath65 ) .
calculations required on average 26.3 hours per 100ps block giving a total of 210,000 core - hours for the entire calculation ( 42 core - months assuming 24/7 operation ) , plus equilibration . to estimate the convergence of the individual
generalised force measurements at each integration point , a spectral analysis of the time series of generalised forces was carried out using the coda library of r functions @xcite , in order to identify the decorrelation times and effective numbers of independent samples present .
estimated standard errors of the mean ( ese ) are given as the root of the variance divided by the effective number of independent samples .
the ese at each value of @xmath15 ( [ fig : covar](b ) ) provides a description of the efficiency of the sampling in each of the ` mixed ' free energy landscapes which were visited .
was determined by variance ( left axis , red ) and by the decorrelation time @xmath66 ( right axis , green ) .
numbered labels indicate peaks in the variance or in @xmath66 which are discussed individually in the text . *
( b ) ese of the generalised force*. the ese per 500ps was much higher for @xmath67 , therefore longer runs of 1600ps were made for this section .
a separate trace is shown for each salt concentration studied .
lines are not drawn between the points at @xmath68 and @xmath69 because the generalised forces on the two intervals @xmath70 and @xmath71 had different shapes and were integrated separately .
each trace corresponds to the @xmath72 calculation for a given [ nacl].,title="fig:",scaledwidth=49.0% ] was determined by variance ( left axis , red ) and by the decorrelation time @xmath66 ( right axis , green ) .
numbered labels indicate peaks in the variance or in @xmath66 which are discussed individually in the text . *
( b ) ese of the generalised force*. the ese per 500ps was much higher for @xmath67 , therefore longer runs of 1600ps were made for this section .
a separate trace is shown for each salt concentration studied .
lines are not drawn between the points at @xmath68 and @xmath69 because the generalised forces on the two intervals @xmath70 and @xmath71 had different shapes and were integrated separately .
each trace corresponds to the @xmath72 calculation for a given [ nacl].,title="fig:",scaledwidth=49.0% ] the radically different sampling efficiencies at the different integration points derive from separately addressable causes , the signatures of which are individually labelled in [ fig : covar ] , and which are discussed in relation to the criteria for good ti design which were presented above : 1 .
* smoothness of generalised force w.r.t .
@xmath15 : * the smoothness criterion was met quite well ( [ fig : genforce ] ) .
* small variance of generalised force : * when the potential corresponding to a given term of the generalised force is very different to the potential which is controlling the dynamics , then fluctuations in that term of the force can become very large , the peak labelled ` 4 ' on [ fig : covar ] shows this effect . +
peak ` 4 ' is lower than in first attempts , thanks to the use of a polynomial mixing function , and also greatly diminished by the use of the guide hamiltonian .
peak ` 4 ' might also be further reduced by the use of different @xmath25 hamiltonians which more closely mirror the dynamics of the ` real ' @xmath22 system .
* fast exploration : * the peaks labelled ` 2 ' and ` 3 ' show slow exploration due to roughness of the potential energy landscape : sampling of the boltzmann distribution for atomistic models of biomolecules is famously difficult , due to the many traps and barriers which exist .
sampling also becomes slower in the cs hamiltonian when it is fully turned on , due to the depth of the wells becoming greater than the thermal energy .
peak ` 5 ' shows a slowing of exploration due to decoupling of particles under the cs hamiltonian : when the particles no longer interact , decorrelation over time must rely purely on the langevin thermostat rather than being assisted by the chaotic properties of the many - body hamiltonians .
although the aspect indicated by peak ` 5 ' seems to be of relatively minor importance , it should be possible to address it by altering the coupling parameter of the langevin thermostat for different values of @xmath15 , or by using an @xmath25 which has explicit interparticle forces while remaining tractable . + to address point ( c ) there is no easy escape from the need to sample the atomistic model at one end of the integration ( ` 2 ' ) . at the other end of the integration ( ` 3 ' ) , the trapping due to the cs potential
could ( and probably should ) be reduced in a future calculation simply by holding the cs potential constant for @xmath73 .
* near - linear mixing : * peak ` 1 ' arises because the gradient of the mixing function @xmath74 is large at small @xmath15 , and this large value scales not only the generalised force but its variance - this represents a tradeoff accepted in the process of mixing function selection , it stands in balance against peak ` 4 ' . to address this need to trade between the criteria ( b ) and ( d )
it seems that the best hope is to improve the guide or endpoint hamiltonians used such that the need for a polynomial mixing function is removed .
this calculation is the most quantitatively accurate numerical estimate of the salt concentration at b - z dna coexistence to date .
as recently as 2010 it was noted that an atomistic calculation of this value should be prohibitively difficult `` because the problem requires the sampling of an extraordinarily large configuration space , including water and ions , to obtain the free energy difference '' @xcite ; that the result is quantitatively correct demonstrates solid improvements in the techniques of free energy estimation suitable for biomolecular complexes , over and above the normal progress due to improved hardware . because of the reductive nature of the molecular dynamics approach ( treating , for instance , an atom in a protein in much the same way as an atom in a dna base ) it is reasonable to assume that the software and parameter sets developed can be reused widely and can be effective for any combination of protein , nucleic acid , counterions and solvent ; although there is still substantial room for further development of them . it is intended to release the software and parameter sets as a downloadable library on dr berryman s website , for easy linking against the amber simulation program or other packages . the accuracy of the final result for the coexistence serves as a validation for use of the combination of the joung - cheatham ion parameters and the parmbsc forcefield ( both of which are relatively new ) with tip3p water for simulation of dna in the fairly unusual conditions of very high salt concentration .
the work here has made minor alterations to the ` cage - swapping ' model for absolute free energy calculation of fluids and combined it with a standard einstein molecule method for the solute .
the principal novelty was in the scale and complexity of the system treated , but the introduction of a ` guide ' hamiltonian to control the path of the thermodynamic integration while not altering behaviour at the endpoints is also novel as far as the authors are aware .
there is too much active research in thermodynamic integration and free - energy perturbation methods to give a comprehensive list of ideas which could influence further work .
a brief suggestion is that absolute free energy methods could be well suited to mapping of phase diagrams of complex fluids .
there are a large number of biomolecular and soft - matter systems which exhibit rich phase behaviour that have not yet been explored . on the methodological side , recent advances in nonequilibrium ti @xcite , in correlated sampling @xcite and in hamiltonian - exchange @xcite
all offer increases in computational efficiency for future ties calculations .
it has been suggested that an efficiency gain can be made by doing away with the intermediate stages of the ti calculation entirely , instead carrying out a type of importance sampling over @xmath22 and @xmath25 such that those areas of configurational space which overlap between them receive enhanced attention @xcite . if one is interested in locating a phase transition one can , of course , use eq . 3 to compute the free energy difference between two phases directly and avoid the `` detour '' via absolute free energies ; however , eq . 3 only holds if the integration path does not cross a first order phase transition .
ties is thus more generally useful ( compared to performing ti over single or multi - legged free energy cycles between non - tractable hamiltonians ) in that states which are separated by a first order phase transition can be compared .
this is much simpler than carrying out ti calculations which integrate between phases by following a complex path designed to give consistently pseudo - critical behaviour @xcite .
a second general argument for ties , which is more germane for the specific calulation presented here , is that setting one endpoint of the integration as an analytically tractable model leaves open the possibility of a shorter and smoother integration path , with smaller variances in the generalised force at a given @xmath15 , as the analytical models are refined .
the most obvious step in further development , then , is extension and refinement of the analytical models which serve as endpoints for the calculation .
ideally the mixed energy landscapes generated by an analytically tractable model should be as similar as possible to ` softened ' versions of the initial hamiltonian .
this possibility should serve as a call to arms for any theorists who may believe that they can swiftly assemble a tractable model for some system of interest .
a supporting document is provided , giving derivations for the absolute free energies of the cs and em hamiltonians . a second supporting document showing convergence of the simulations
is also provided .
this material is available free of charge via the internet at http://pubs.acs.org .
this work was greatly assisted by discussions with charles laughton and sarah anne harris .
computing was provided by the hpc facility of the university of luxembourg and by the european soft matter infrastructure program .
prez , a. ; marchn , i. ; svozil , d. ; sponer , j. ; cheatham , t. e. ; laughton , c. a. ; orozco , m. _ biophys . j. _ * 2007 * , _ 92 _ , 38173829 eike , d. m. ; maginn , e. j. _ j. chem .
_ * 2006 * , _ 124 _ , 164503 tyka , m. d. ; sessions , r. b. ; clarke , a. r. _ j. phys . chem .
b _ * 2007 * , _ 111 _ , 95719580 cecchini , m. ; krivov , s. v. ; spichty , m. ; karplus , m. _ j. phys . chem .
b _ * 2009 * , _ 113 _ , 97289740 schilling , t. ; schmid , f. _ j. chem
. phys . _ * 2009 * , _ 131 _ , 231102 frenkel , d. ; ladd , a. j. c. _ j. chem .
phys . _ * 1984 * , _ 81 _ , 31883193 vega , c. ; noya , e. g. _ j. chem . phys . _ * 2007 * , _ 127 _ , 154113 schmid , f. ; schilling , t. _ physics procedia _ * 2010 * , _ 4 _ , 131143 case , d. a. ; darden , t. a. ; cheatham , t. e. ; simmerling , c. l. ; wang , j. ; duke , r. e. ; luo , r. ; crowley , m. ; walker , r. c. ; zhang , w. ; merz , k. m. ; wang , b. ; hayik , s. ; roitberg , a. ; seabra , g. ; kolossvry , i. ; wong , k. f. ; paesani , f. ; vanicek , j. ; wu , x. ; brozell , s. r. ; steinbrecher , t. ; gohlke , h. ; yang , l. ; tan , c. ; mongan , j. ; hornak , v. ; cui , g. ; mathews , d. h. ; seetin , m. g. ; sagui , c. ; babin , v. ; kollman , p. a. _ amber 11 _ ; university of california , san francisco , 2010 pohl , f. m. ; jovin , t. m. _ j. mol . biol . _ * 1972 * , _ 67 _ , 375396 ferreira , j. m. ; sheardy , r. d. _ biophys . j. _ * 2006 * , _ 91 _ , 33833389 oh , d .- b . ; kim , y .- g . ;
rich , a. _ proc .
usa _ * 2002 * , _ 99 _ , 1666616671 peck , l. j. ; nordheim , a. ; rich , a. ; wang , j. c. _ proc .
usa _ * 1982 * , _ 79 _ , 45604564 wang , g. ; christensen , l. a. ; vasquez , k. m. _ proc .
usa _ * 2006 * , _ 103 _ , 26772682 kastenholz , m. a. ; schwartz , t. u. ; hnenberger , p. h. _ biophys .
j. _ * 2006 * , _ 91 _ , 29762990 lee , j. ; kim , y .- g . ; kim , k. k. ; seok , c. _
j. phys .
b _ * 2010 * , _ 114 _ , 98729881 behe , m. ; felsenfeld , g. _ proc .
usa _ * 1981 * , _ 78 _ , 16191623 maruyama , y. ; yoshida , n. ; hirata , f. _ j. phys .
b _ * 2010 * , _ 114 _ , 64646471 kollman , p. ; weiner , p. ; quigley , g. ; wang , a. _ biopolymers _ * 1982 * , _ 21 _ , 19451969 misra , v. k. ; honig , b. _ biochemistry _ * 1996 * , _ 35 _ , 11151124 guron , m. ; demaret , j. ; filoche , m. _ biophys . j. _ * 2000 * , _ 78 _ , 10701083 soumpasis , d. m. _ proc . natl .
usa _ * 1984 * , _ 81 _ , 51165120 jovin , t. m. ; soumpasis , d. m. ; mcintosh , l. p. _ annu .
chem . _ * 1987 * , _ 38 _ , 521558 montoro , j. c. g. ; abascal , j. l. f. _ j. chem .
phys . _ * 1997 * , _ 106 _ , 82398253 hirata , f. ; levy , r. m. _ j. phys .
chem . _ * 1989 * , _ 93 _ , 479484 , r. d. _ acta crystallogr .
a _ * 1976 * , _ 32 _ , 751767 pohl , f. m. _ cold spring harbor symp
. quant .
_ * 1983 * , _ 47 _ , 113117 ivanov , v. i. ; karapetian , a. t. ; miniat , e. e. ; sad , i. _ mol .
( mosk . ) _ * 1993 * , _ 27 _ , 11501156 grzeskowiak , k. ; ohishi , h. ; ivanov , v. _ nucleic acids symp .
( oxf . ) _ * 2005 * , 249250 reinhard , f. ; grubmller , h. _ j. chem
. phys . _
* 2007 * , _ 126 _ , 014102 laberge , l. j. ; tully , j. c. _ chem .
_ * 2000 * , _ 260 _ , 183191 steinbrecher , t. ; mobley , d. l. ; case , d. a. _ j. chem .
phys . _ * 2007 * , _ 127 _ , 214108 mcleish , t. _ biophys . j. _ * 2005 * , _ 88 _ , 172183 beutler , t. c. ; mark , a. e. ; van schaik , r. c. ; gerber , p. r. ; van gunsteren , w. f. _ chem .
lett . _ * 1994 * , _ 222 _ , 529539 wang , j. ; cieplak , p. ; kollman , p. a.
_ j. comput .
_ * 2000 * , _ 21 _ , 10491074 joung , i. s. ; cheatham , t. e. _ j. phys .
b _ * 2008 * , _ 112 _ , 90209041 jorgensen , w. l. ; chandrasekhar , j. ; madura , j. d. ; impey , r. w. ; klein , m. l. _ j. chem . phys . _ * 1983 * , _ 79 _ , 926935 assaraf , r. ; caffarel , m. ; kollias , a. c. _ phys . rev .
lett . _ * 2011 * , _ 106 _ , 150601 duke , r. e. ; pedersen , l. g. _ pmemd 3 _ ; university of north carolina - chapel hill , 2003 macke , t. j. ; case , d. a. in _ modeling unusual nucleic acid structures _ ; american chemical society , 1997 ; chapter 25 , pp 379393 humphrey , w. ; dalke , a. ; schulten , k. _ j. mol . graph . _ * 1996 * , _ 14 _ , 3338 frank - kamenetskii , m. d. ; lukashin , a. v. ; anshelevich , v. v. _ j. biomol .
dyn . _ * 1985 * , _ 3 _ , 3542 lukashin , a. v. ; beglov , d. b. ; frank - kamenetskii , m. d. _ j. biomol
. struct .
dyn . _ * 1991 * , _ 8 _ , 11131118 plummer , m. ; best , n. ; cowles , k. ; vines , k. _ r news _ * 2006 * , _ 6 _ , 711 vaikuntanathan , s. ; jarzynski , c. _ j. chem . phys . _ * 2011 * , _ 134 _ , 054107 khavrutskii , i. v. ; wallqvist , a. _ j. chem .
theory comput . _ * 2010 * , _ 6 _ , 34273441 wilms , d. ; wilding , n. b. ; binder , k. _ phys .
* 2012 * , _ 85 _ , 056703 grochola , g. _ j. chem . phys . _
* 2004 * , _ 120 _ , 21222126 | sets of free energy differences are useful for finding the equilibria of chemical reactions , while absolute free energies have little physical meaning .
however finding the relative free energy between two macrostates by subtraction of their absolute free energies is a valuable strategy in certain important cases .
we present calculations of absolute free energies of biomolecules , using a combination of the well - known einstein molecule method ( for treating the solute ) with a conceptually related method of recent genesis for computing free energies of liquids ( to treat the solvent and counterions ) .
the approach is based on thermodynamic integration from a detailed atomistic model to one which is simplified but analytically solvable , thereby giving the absolute free energy as that of the tractable model plus a correction term found numerically .
an example calculation giving the free energy with respect to salt concentration for the b- and z - isomers of all - atom duplex dna in explicit solvent and counterions is presented .
the coexistence salt concentration is found with unprecedented accuracy . |
front propagation is an important field of study in nonequilibrium systems .
we often encounter these propagating fronts separating different phases in physics , chemistry and biology @xcite . here , in this work , we study the dynamics of the front in the reaction - diffusion system @xmath7 , @xmath8 in one dimensional lattice . at the macroscopic level , the mean field theory yields the following partial differential equation for the coarse grained concentration @xmath9 , @xmath10 where , @xmath4 is the diffusion coefficient of the particle and @xmath11 and @xmath2 are the rates of single ( @xmath7 ) and twin ( @xmath8 ) offspring production respectively . equation ( [ e1 ] ) reduces to the well known fisher equation @xcite when @xmath12 , which models the reaction diffusion equation @xmath7 .
the microscopic lattice model for @xmath7 has been studied extensively @xcite@xcite@xcite@xcite .
the mean field allows travelling wave solution of the form @xmath13 , where the velocity of an initially sharp front between @xmath14 ( stable ) and @xmath15 ( unstable ) state approaches an asymptotic velocity @xmath16 .
we consider a @xmath1 lattice composed of sites @xmath17 .
we start with the step function like distribution where , the left half is filled with @xmath5 particles while the right half is empty .
each site @xmath18 can either be empty or occupied by maximum one particle i.e. hard core exclusion is enforced .
we update the system random sequentially where @xmath19 microscopic moves correspond to one monte carlo step ( mcs ) . during each updating
we randomly select a site and the particle at the site can undergo one of the following three microscopic moves . + ( 1 ) the particle can jump to neighbouring empty site with rate @xmath4 , + ( 2 ) the particle can give birth of one particle at either of the empty neighbouring sites with rate @xmath11 , + ( 3 ) the particle can generate two new particles at both the neighbouring sites provided both are empty with rate @xmath2 . + these processes are shown in the fig .
( [ fig : moves_1d ] ) . + as time evolves , these stochastic moves result in the stochastic movement of the front . as has been argued in @xcite , the front may be identified with the rightmost @xmath5 particle . in this paper , we are interested in the dynamics of front whose evolution may be described by the following master equation@xcite .
@xmath20 here , @xmath21 is the probability distribution of finding the front particle at position @xmath22 at a time @xmath23 and @xmath24 is the joint probability that the front is at @xmath22 at time @xmath23 _ and _ the site just behind it is empty . in eq .
( [ e4 ] ) , the first term corresponds to the forward hopping of the front particle from the position @xmath25 to reach @xmath22 due to birth of a single particle or diffusion .
the second term corresponds to twin production at left and right neighbouring sites of site @xmath25 and which results in the front moving from @xmath25 to @xmath22 .
the third term corresponds to the backward hopping of the front particle from position @xmath26 due to diffusion , provided site @xmath22 is empty .
last two terms account for the possible jumps of the front particle from position @xmath22 , which leads to front moving either at @xmath25 or @xmath26 .
the dynamical properties of the front that we want to study are its velocity @xmath27 and the diffusion coefficient @xmath28 which are defined as : @xmath29 @xmath30 where , @xmath31 .
now we use the eq .
( [ e4 ] ) and normalization @xmath32 and taking @xmath33 , where @xmath34 is the probability that site @xmath25 is occupied . denoting @xmath35 , we obtain the following expression for the asymptotic velocity and diffusion coefficient of the front .
@xmath36 @xmath37 in order to obtain the velocity and diffusion coefficient we need to know @xmath38 , which is the density of site just behind the front . in @xcite
for @xmath12 , it was shown that front velocity approaches asymptotically the mean field value @xmath39 in the limit @xmath40 , while @xmath41 in the opposite limit @xmath42 .
but , when we are in between these two extreme limits we need to know @xmath38 and we expect similar features when @xmath43 .
there is no method to find this value exactly .
here we present some approximate analytic estimates for @xmath38 and hence the front velocity . in subsection a , we use fixed site representation method , where a truncated master equation is written in the frame moving with the front , as discussed in @xcite . in subsection
b , we apply two particle representation scheme proposed by kerstein @xcite while in c a mixed scheme is proposed which yields better results than either a or b. this method has been proposed in @xcite for the reaction diffusion process @xmath44 . here ,
we write a truncated master equation in the frame moving with the front .
the simplest set of states is:@xmath45 , which corresponds to the evolution of occupancy at a site just behind the front particle(@xmath46 ) . here
the rightmost @xmath47 in each state corresponds to the front particle .
these two states make transitions between each other due to the microscopic processes in the system as shown in fig .
( [ fig : moves_1d ] )
. represents the front .
( a ) diffusion of the front particle to its right site leading to transition from @xmath48 to @xmath49 with rate @xmath4 .
( b ) creation of one particle to the left of the front leads to transition from @xmath49 to @xmath48 with rate @xmath11 .
( c ) @xmath49 changes to @xmath48 due to creation of twins at both neighbouring sites of the front with rate @xmath2 .
( d ) @xmath50 , if the front takes diffusive move to its left and the second site behind the front is occupied .
this occurs with rate @xmath51 , where @xmath52 is the probability of occupancy at the second site behind the front.,width=240,height=144 ] considering all such transitions the evolution of probabilities of these two states are given by : @xmath53 here , @xmath54 is the density at the @xmath18th site behind the front and we have neglected the spatial density correlation between consecutive pairs of sites beyond the second site behind the front .
now , using eq .
( [ e9 ] ) and normalization @xmath55 , we obtain the following expression for @xmath38 : @xmath56 from eq .
( [ e10 ] ) , we note that in order to find @xmath38 we need to know @xmath52 and @xmath57 . as a crude approximation
if we assume that @xmath58 , where @xmath59 is the bulk density , we get the following value of @xmath38 .
@xmath60 now using this approximation for @xmath38 in eq .
( [ e7 ] ) , we find the estimate for the velocity which is in reasonable agreement with the simulation , as shown in the fig .
( [ fig : velocity ] ) .
the estimate for @xmath27 can be improved by including more sites in the truncated representation .
for example , for @xmath61 we study the evolution of following set of four states : @xmath62 and as expected we get improved results as shown in fig .
( [ fig : velocity ] ) . here , we notice that for larger values of @xmath2 the simulation results show nice agreement with that of analytic results .
however , as @xmath2 decreases and approaches zero , we see gradual departure of the simulation data from the analytic one .
in fact , @xmath54 differs from the bulk density significantly with decreasing value @xmath2 as shown in the fig .
( [ fig : den_profile ] ) .
that is , the approximation @xmath63 holds better for larger values of @xmath2 and hence we get better agreement with the simulation results .
the estimate for the velocity can be further improved if we include states with larger number of sites . in fig .
( [ fig : velocity ] ) we notice two interesting points : firstly , for @xmath64 , the theoretical result matches strikingly with the simulation result , secondly , we observe a power law dependence of the velocity on @xmath2 .
in fact , the first point , can be shown to be exact by noting that when @xmath64 , the front velocity from eq.([e7 ] ) is @xmath65 , which is independent of @xmath38 . in the following ,
we try to find the analytic estimates for @xmath38 using kerstein s two particles representation@xcite . in this representation
each state of the system is defined by two rightmost particles and thus we have an infinite set of states : @xmath66 . here
, the rightmost 1 represents the front particle while the leftmost 1 is the second particle behind it and 0 stands for empty site .
let us denote by @xmath67 the probability of two particle state with @xmath68 empty sites between the leading particle and next particle behind it .
these states form a closed set under transition due to microscopic processes
. we have illustrated few transitions in the fig .
( [ fig : moves_2p ] ) . considering all such transitions and denoting the probability of occupancy of site just behind the second particle by @xmath69 , we write the following rate equations for @xmath67 . @xmath70 representing front .
( a ) diffusive move of the front particle to its left leading to transition 101@xmath7111 with rate d , ( b ) when the second particle behind the front jumps to the left , provided it is empty , state changes from 101@xmath711001 with rate @xmath72 , ( c ) birth of a single particle by the second particle to its left with rate @xmath11 leads to transition 1001@xmath71101 , ( d ) 1001@xmath71101 if the second particle gives birth of two particles , provided the the site left to it is empty , with rate @xmath73.,width=211,height=172 ] in order to solve eq.([e13 ] ) we need to specify the dependence of @xmath69 on the parameters(@xmath74 ) .
following kerstein @xcite , we write @xmath75 and enforcing the condition that @xmath14 when @xmath76 , we write the following expression for @xmath69 .
@xmath77 this equation specifies the dependence of @xmath69 on the parameters implicitly through dependence of @xmath78 on @xmath74 .
here , @xmath79 is a free parameter to be evaluated as follows .
following kerstein , using the ansatz @xmath80 and eq .
( [ e14 ] ) in eq.([e13 ] ) , we get the following quartic equation in @xmath78 .
@xmath81 in order to find @xmath78 we need to fix the value of @xmath79 . for large @xmath4 and @xmath12
, it is known that the front particle moves with its mean field velocity@xcite . if we assume that this also happens when @xmath82 , then equating the mean field front velocity @xmath83 with that obtained from eq .
( [ e7 ] ) i.e. @xmath84 when @xmath4 is very large compared to other parameters , we get @xmath85 . using this value of @xmath78 in eq .
( [ e15 ] ) we find @xmath86 in the limit @xmath87 .
we solve the quartic equation ( [ e15 ] ) to get the value of @xmath88 and hence the front velocity , as shown in the figs .
( [ fig : velocity ] ) and ( [ fig : rho1_vel_et0 ] ) and marked as @xmath89 .
since we are dealing with a multiparticle interacting system it is always desirable to include as many particles as possible while studying the evolution of the system .
the simplest extension to the two particle representation is to study the evolution of the following set of states : @xmath90 , where the rightmost @xmath47 in each state denotes the front particle .
since in this representation , each state is characterized by two or three particles and hence we name it as mixed representation ( mr ) .
the rightmost @xmath47 in each state denotes the front particle .
when viewed in the frame moving with the front , each state contains the location of second particle and the occupancy of the site just behind the second particle .
we denote these states as @xmath91 or @xmath92 , representing the states having @xmath68 empty sites between the front and the second particle and the site just after the second particle is empty or occupied respectively .
for example , by ( 0 , 0 ) we mean the state @xmath93 and ( 0 , 1 ) for the state @xmath94 .
these states are making transitions among each other due to the microscopic processes and form a closed set .
we have shown some of the transitions in the fig .
( [ fig : moves_2 ] ) . representing front .
( a ) diffusive move of the particle to the right empty site with rate d.this leads to transition from the state ( 1,1 ) to ( 0,0 ) .
( b ) birth of a new particle on the right neighbouring empty site with rate @xmath11 , which changes the state ( 1,0 ) to ( 0,1 ) .
( c ) transition from ( 1,0 ) to ( 1,1 ) with rate d@xmath69 , when the third particle jumps to the right neighbouring empty site .
( d ) ( 1,0)@xmath71(0,1 ) when the second particle behind the front in ( 1,0 ) realization produces twins at the neighbouring empty sites with rate @xmath2 .
, width=211,height=172 ] assuming @xmath69 as the density of site , which is , next nearest neighbour to the second particle , we write the following rate equation for the evolution of probabilities
@xmath95 and @xmath96 , @xmath97 .
@xmath98 in order to find @xmath78 , we need to solve the above set of coupled equations
. however , one can find the analytic estimate for @xmath78 by solving rate equations for @xmath99 and @xmath100 and assuming @xmath101 .
using @xmath102 and @xmath103 , we find steady state expression for @xmath99 and @xmath100 in terms of @xmath104 and @xmath69 and then solve the equation : @xmath105 following kerstein @xcite , if we use the ansatz @xmath106 , we get the following equation .
@xmath107 where , @xmath108 the eq .
( [ e18 ] ) is in terms of two unknowns @xmath69 and @xmath78 and hence we need to know @xmath69 in order to find @xmath78 .
we specify the dependence of @xmath69 on @xmath78 similar to what we did in the case of two particle representation . using the results obtained for @xmath109 in eq.([e7 ] )
we find the estimates for @xmath27 as shown in the figs .
( [ fig : velocity ] ) and ( [ fig : rho1_vel_et0 ] ) marked as mr .
we observe good agreement with the simulation results . ,
i.e. @xmath110 ( @xmath111 and @xmath112 representing simulation and analytic results respectively ) , is plotted against @xmath2 while keeping d=0.25 and @xmath113 fixed . here
stars , open squares , open circles and filled circles correspond to the analytic estimates using @xmath61(two states ) , @xmath114 ( four states ) , two particle(2p ) and mixed representation ( mr ) respectively .
simulation profile and analytic profile using mr are essentially coincident and hence we notice almost zero error .
inset : log - log plot for the velocity versus @xmath2 , the straight line shows power law @xmath115 . ] versus @xmath4 , keeping @xmath116 and @xmath12 . the top data(filled squares ) corresponds to ref.@xcite , where in the frame moving with the front , the evolution of particles at three sites behind the front was studied and assuming the fourth site at the bulk density .
the middle data ( open circles ) corresponds to kerstein @xcite two particle self consistent representation while the bottom data ( filled circles ) is the result from the mixed representation . ] while keeping @xmath117 and @xmath113 fixed . here
, we notice that as @xmath2 decreases density profile curve shifts away from the bulk density level . ]
in fig . ( [ fig : diff_coeff ] ) , we have shown the simulation results of the front diffusion coefficient and compared it with the results obtained by using the mean field and simulation value of @xmath38 in the equation([e8 ] ) . here , we notice the following interesting features : ( 1 ) when @xmath64 , the analytical value @xmath118 matches well with the simulation result @xmath119 .
( 2 ) when @xmath120 , the @xmath121 ( 3 ) when @xmath122 , @xmath123 .
the origin of the above discrepancy between the simulation and analytical results can be traced to the master equation ( [ e4 ] ) , where we have neglected the temporal velocity correlations . the expression for the asymptotic front diffusion coefficient with temporal correlations in velocity
is given as : @xmath124 where , @xmath125 is the front diffusion coefficient by neglecting correlations as given by eq .
( [ e8 ] ) and @xmath126 is the temporal velocity correlation defined through , @xmath127 where , @xmath128 is the displacement of the front at time @xmath23 . ) and using @xmath38 corresponding to @xmath61 ) with the simulation results for different values of @xmath2 while keeping @xmath117 , @xmath11=0.025 fixed .
we note that when @xmath2=@xmath117 , the simulation result matches with the analytic one.,width=288,height=240 ] , for different values of @xmath2 while keeping @xmath117 , @xmath11=0.025 fixed .
we notice that when @xmath129 , this correlation is zero .
inset : log - normal plot for @xmath126 versus @xmath23 with @xmath130 from top to bottom . ] in fig .
( [ fig : vel_corr ] ) , we have plotted the temporal velocity correlation @xmath126 for different values of @xmath2 . for @xmath131 , we observe positive correlation while for @xmath132 , it is negative and for @xmath6 , @xmath126 seems to vanish for all @xmath23 .
thus , @xmath6 is a special case , where the front particle moves like an uncorrelated random walker . in the following ,
we explicitly show that for the special case @xmath6 , two consecutive steps of the leading particle are uncorrelated , i.e. , @xmath133 in the steady state . since at most two sites behind the front can be affected in two consecutive steps , we consider 4 states corresponding to @xmath61 , namely , @xmath134 with the rightmost @xmath135 representing the front . in order to find @xmath136 , we write @xmath137 , where @xmath138 denotes the flux @xmath139 for taking two consecutive steps as @xmath140 and @xmath141 . here
, for example , @xmath142 is the flux of two consecutive negative steps starting from the state @xmath143 .
the only way it can occur is if the front particle takes two diffusive moves to the left and thus , @xmath144 , where @xmath145 is the steady state weight of the configuration @xmath143 . considering all such two successive moves in the each state
, we write the following expression for @xmath146 and @xmath147 , @xmath148 now using eqs.([e22 ] ) , we have @xmath149 and similarly , @xmath150 using eqs.([e23 ] ) and ( [ e24 ] ) we can find the value of @xmath151 or @xmath152 by finding the probabilities of different states , which is harder to compute exactly .
however , when @xmath6 , we find that @xmath153 is independent of all the probabilities and is equal to zero , i. e. , two successive steps are uncorrelated as observed in the simulation result ( fig.[fig : vel_corr ] ) .
we also note that for this special case the above analysis does not involve any approximation i.e. it is exact .
we also notice that when @xmath154 , @xmath155 , i.e. , the front motion is correlated .
preliminary fits suggest that the the temporal velocity correlation has the form , @xmath156 .
in the present paper , we studied the reaction diffusion system @xmath157 in one dimension . treating the rightmost occupied site as a front we compute the front velocity analytically using different approximate methods .
in fixed site representation one can systematically improve upon the estimate by studying the evolution of particles at larger number of sites behind the front .
the results from two particle representation and mixed representation show excellent agreement with the simulation results .
we also observed that the velocity depends on @xmath2 as a power law .
as far as the computation of front diffusion coefficient is concerned , we notice that one needs to take into account the temporal velocity correlation .
in fact the observed temporal correlations in the front dynamics changes sign with sign of @xmath3 . for @xmath131 or @xmath132 , front moves like a positively or negatively correlated random walk and for @xmath6 ,
the temporal correlations in steps vanish and the front particle moves like a simple uncorrelated random walker .
an interesting generalization of the process would be to include the annihilation of particles as well i.e. , @xmath158 with rate @xmath159 . the case @xmath12 and @xmath160 is exactly solvable @xcite and both temporal and spatial correlations vanish . for non - zero @xmath4 , @xmath2 and @xmath159
, simulations show that temporal correlations vanish on the plane @xmath161 , although spatial correlations do not .
computation of the previous section can be extended to this case to show that consecutive steps are temporally uncorrelated .
a general proof to show that @xmath126 vanishes for all @xmath23 remains an interesting open problem .
w.van saarloos , phys .
386 , 29 ( 2003 ) .
m. bramson , p. calderoni , a. de masi , p. a. ferrari , j. l. lebowitz and r. h. schonmann , j. stat .
45 , 905 ( 1986 ) .
a. r. kerstein j. stat .
45 , 921 ( 1986 ) .
a. r. kerstein , j. stat .
53,703 ( 1988 ) .
niraj kumar and g. tripathy , phys .
e 74 , 011109 ( 2006 ) .
r. a. fisher , ann .
eugenics , 7 , 355 ( 1937 ) ; a. kolmogorov , i. petrovsky and n. piscounov , moscow uni . bull .
a1 , 1 ( 1937 ) .
n. g. van kampen , stochastic processes in physics and chemistry ( north - holland , amsterdam , 1981 ) .
d. ben - avraham , phys . lett .
a , 247 , 53 ( 1998 ) . | we study front propagation in the reaction diffusion process @xmath0 on a one dimensional ( @xmath1 ) lattice with hard core interaction between the particles . using the leading particle picture , velocity of the front in the system
is computed using different approximate methods , which is in good agreement with the simulation results .
it is observed that in certain ranges of parameters , the front velocity varies as a power law of @xmath2 , which is well captured by our approximate schemes .
we also observe that the front dynamics exhibits temporal velocity correlations and these must be taken care of in order to find the exact estimates for the front diffusion coefficient
. this correlation changes sign depending upon the sign of @xmath3 , where @xmath4 is the bare diffusion coefficient of @xmath5 particles . for @xmath6 , the leading particle and thus the front moves like an uncorrelated random walker , which is explained through an exact analysis |
the _ gonality _ @xmath4 of a smooth projective curve @xmath5 over a field @xmath0 is defined as the minimal degree of a non - constant morphism from @xmath5 to the projective line @xmath6 .
if @xmath7 is the complex numbers , @xmath5 can be considered as a compact riemann surface , and li and yau @xcite have established a lower bound on the gonality of @xmath5 over @xmath8 in terms of of the hyperbolic volume and the first eigenvalue of the laplacian of @xmath5 .
such a bound has numerous applications , of which we mention one : abramovich @xcite has combined it with a lower bound on the eigenvalue arising from the theory of automorphic forms ( of which the currently sharpest value was given by kim and sarnak @xcite ) to prove a lower bound on the gonality of modular curves for congruence groups that is linear in the genus of the curves ( or , what is the same , linear in the index of the group in the full modular group ) . in this paper , we study a nonarchimedean analogue of these results .
the first result is an inequality between the ( geometric ) gonality @xmath9 of a curve @xmath5 defined over a complete nonarchimedean valued field @xmath0 and the `` gonality '' of the reduction graphs of suitable models of the curve .
there are various complications , such as to establish a good theory for the reduction of a covering map @xmath10 .
such a map extends to the stable model , but not necessarily as a _
finite _ morphism .
this can be remedied by choosing suitable semi - stable models .
the problem was studied by liu and lorenzini @xcite , coleman @xcite and liu @xcite , and more recently in @xcite . in section
[ extension ] , we provide another ( similar ) solution , directly adapted to the applications that we have in mind .
next , we relate the gonality of the special fiber to what we call the _ stable gonality _ of the intersection dual graph . for standard graph terminology
, we refer to section [ graphs ] .
we also need the notion of an ( indexed ) finite harmonic graph morphism , for which we refer to definition [ defmor ] . given a graph @xmath11 , then another graph @xmath12 is called a _ refinement _ of @xmath11 if it can be obtained from @xmath11 by finitely often performing the two following operations : ( a ) subdivision of an edge ; ( b ) addition of a _ leaf _ , i.e. , the addition of an extra vertex and an edge between this vertex and a vertex of the already existing graph .
stable gonality _ of @xmath11 , denoted @xmath13 , is defined as the minimal degree of a finite harmonic morphism from any refinement of @xmath11 to a tree .
this relates to , but is different from previous notions of gonality for graphs as introduced by baker and norine @xcite , and caporaso @xcite ( cf .
appendix [ app ] for a discussion of these different notions and how they relate to stable gonality ) .
[ compare ] let @xmath5 be a geometrically connected projective smooth curve over a complete nonarchimedean valued @xmath0 with valuation ring @xmath14 , and @xmath15 the stable @xmath14-model of @xmath5 .
let @xmath16 be an algebraic closure of @xmath0 .
let @xmath17 denote the intersection dual graph of the special fiber @xmath18 .
then we have @xmath19 two examples ( [ examplegon1 ] and [ examplegon2 ] ) illustrate that both refinements operation are necessary .
first , the `` banana graph '' @xmath20 given by two vertices joined by @xmath21 distinct edges has stable gonality @xmath22 , although the minimal degree of a finite harmonic graph morphism from @xmath20 itself ( without any refinement ) to a tree is @xmath23 .
secondly , the minimal degree of a finite harmonic graph morphism from any subdivision of the complete graph @xmath24 to a tree is @xmath25 .
however , by adding leaves , the stable gonality can be shown to be @xmath26 .
we then prove an analogue for graphs of the upper bound on gonality from brill - noether theory for the gonality of curves over arbitrary fields ( in this generality a theorem of kleiman
laksov @xcite ) : [ bnn ] for any graph @xmath11 with first betti number @xmath27 , we have an upper bound @xmath28 the main result is a spectral lower bound for the stable gonality of a graph .
let @xmath29 denote the first non - trivial ( i.e. , smallest non - zero ) eigenvalue of the laplacian @xmath30 of @xmath11 , and let @xmath31 denote the maximal vertex degree of @xmath11 . finally , let @xmath32 denote the number of vertices of @xmath11
. then we have [ spectral ] the stable gonality of a graph @xmath11 satisfies @xmath33 an attractive feature of the formula is that the lower bound depends on spectral data for the original graph , not on all possible refinements of the graph .
also , in the bound , one may replace @xmath34 by the corresponding data @xmath35 of any refinement @xmath12 of the graph @xmath11 .
a similar result can be proven using the normalized graph laplacian , replacing @xmath32 by the `` volume '' of the graph , cf .
theorem [ genormaliseerdegrens ] .
the result can be seen as an analogue of the li yau inequality in differential geometry @xcite , which states that the gonality @xmath36 of a compact riemann surface @xmath5 ( minimal degree of a conformal mapping from @xmath5 to the riemann sphere ) is bounded below by @xmath37 where @xmath38 is the first non - trivial eigenvalue of the laplace - beltrami operator of @xmath5 , and @xmath39 denotes the volume of @xmath5 . in remark
[ lyfout ] , we will show that the strict graph theory analog of such a formula fails .
we then apply the two theorems above to drinfeld modular curves over a general global function field @xmath40 over a finite field with @xmath41 elements , and we find the positive characteristic analogue of abramovich s result . in the applications , we will write @xmath42 for the valuation corresponding to a fixed `` infinite '' place @xmath43 of degree @xmath44 of @xmath40 , we denote by @xmath45 the subring of @xmath40 of elements that are regular outside @xmath43 , and we let @xmath46 denote a rank - two @xmath45-lattice in the completion @xmath47 of @xmath40 at @xmath43 . up to equivalence , such lattices correspond to elements of @xmath48 .
let @xmath49 denote the maximal abelian extension of @xmath40 inside @xmath50 ; then @xmath51 .
in the `` standard '' example where @xmath52 is the function field of @xmath53 and @xmath54 , @xmath55 is unique up to equivalence , and @xmath56 .
congruence subgroups @xmath1 of @xmath57 ( i.e. , containing @xmath58 for some non - trivial ideal @xmath59 of @xmath45 ) act by fractional linear transformations on the drinfeld `` upper half plane '' @xmath60 , and the quotient analytic space can be compactified into a smooth projective curve @xmath61 by adding finitely many cusps .
[ dringon ] let @xmath1 denote a congruence subgroup of @xmath62 .
then the gonality of the corresponding drinfeld modular curve @xmath61 satisfies @xmath63\ ] ] where the constant @xmath64 is @xmath65 this implies a linear lower bound in the genus of modular curves of the form @xmath66 where @xmath67 is a bound that depends only on the function field @xmath40 and the degree @xmath44 of @xmath43 .
if @xmath40 is a rational function field and @xmath68 , then we can put @xmath69 . in the proof
, we use the structure of the reduction graph of the principal modular curve of level @xmath59 ( or rather , its components @xmath70 indexed by @xmath46 running through classes in @xmath48 ) .
also used in the proof is a bound of the laplace eigenvalue for this graph that follows from the ramanujan conjecture , proven by drinfeld ( in combination with the courant - weyl inequalities ) .
the proof of the genus bound is not entirely automatic , due to possible wild ramification .
the constant @xmath64 is probably not optimal , and it would be interesting to know whether it can be replaced by an absolute constant , or at least a constant depending on @xmath41 , but tending to an absolute non - zero constant as @xmath41 increases . also notice that the bound is vacuous ( since a negative number ) if @xmath71 , and that the general upper bound @xmath72 implies that any suitable constant @xmath67 should be smaller than @xmath73 .
all previously known results on gonality of drinfeld modular curves used point counting arguments modulo primes , rather than the above `` geometric analysis '' method .
the best previously known bounds , due to andreas schweizer ( @xcite , thm .
2.4 ) are not linear in the index and are established for a rational function field @xmath52 only .
an extra asset of the new method is that it works without much extra effort for a general function field @xmath40 , rather than just a rational function field .
a first application arises from the modularity of elliptic curves over function fields .
recall that any elliptic curve @xmath74 with split multiplicative reduction at the infinite place @xmath43 admits a modular parametrization @xmath75 @xcite for some suitable drinfeld modular curve @xmath76 .
this parametrization is defined over @xmath49 .
[ moddeg ] let @xmath74 denote an elliptic curve with split multiplicative reduction at the place @xmath43 , of conductor @xmath77 .
then the degree of a modular parametrization @xmath75 is bounded below by @xmath78.\ ] ] in particular , we have @xmath79 as usual , @xmath80 means that there exists a constant @xmath81 depending only on @xmath82 such that @xmath83 . for @xmath52 a rational function field ,
the final statement of the theorem confirms a conjecture of papikian @xcite , who had proven ( using spziro s conjecture for function fields and estimating symmetric square @xmath84-functions by the ramanujan conjecture ) that @xmath85 where @xmath86 is the @xmath87-invariant of @xmath88 and @xmath89 is its inseparability degree .
actually , since he and pl have also proven an upper bound we conclude that if @xmath88 is a _ strong _
weil curve over @xmath90 with square - free conductor , then @xmath91 for any @xmath92 ; cf .
remark [ pal ] for a more precise upper bound , and a discussion of the rle of the manin constant of @xmath88 .
contrary to the case of elliptic curves over @xmath93 , gekeler has proven that the modular degree always equals the congruence number of the associated automorphic form @xcite @xcite .
hence these results also hold for the congruence number .
we then give applications to rational points of bounded degree on general curves over function fields .
in its general form , the theorem gives a finiteness result for points whose degree is bounded above by `` spectral '' data associated to the combinatorics of a special fiber : [ genth ] let @xmath5 denote a curve over a global function field @xmath40 , such that its jacobian does not admit a @xmath94-morphism to a curve defined over a finite field .
let @xmath47 denote the completion of @xmath40 at a place @xmath43 , and let @xmath11 denote the stable reduction graph of @xmath95 .
let @xmath96 denote the maximal vertex degree of @xmath11 , @xmath32 the number of vertices of @xmath11 and @xmath97 the smallest non - zero eigenvalue of the laplacian of @xmath11 .
then the set @xmath98
\leq \frac{\lambda(|g|-1)-4\delta-4}{2\lambda+8\delta+8 } } x(k')\ ] ] of rational points on @xmath5 of degree at most @xmath99 is finite .
this applies in particular to various modular curves , as follows .
[ tors ] with the same notations as in theorem [ dringon ] , if @xmath61 is defined over a finite extension @xmath100 of @xmath40 , then the set @xmath101 \leq \frac{1}{2 } \left ( c_{q,\delta } \cdot [ \gamma(1):\gamma]-1\right ) } x_\gamma(l)\ ] ] is finite
. applications to uniform bounds on isogenies and torsion points follow by applying an analogue of a method of abramovich and harris @xcite and frey @xcite . recall that @xmath49 is the maximal abelian extension of @xmath40 inside @xmath47 .
[ freyan ] if @xmath102 is a prime ideal in @xmath45 , then the set of all rank two drinfeld @xmath45-modules defined over some field extension @xmath84 of @xmath40 that satisfies the degree bound @xmath103 \leq \frac{1}{2 } c_{q,\delta } \cdot |\operatorname{\mathfrak{p}}|_\infty\ ] ] that admit an @xmath84-rational @xmath102-isogeny is finite .
we also deduce the following analogue of a result of kamieny and mazur @xcite : [ km ] fix a prime @xmath102 of @xmath45 .
there is a uniform bound on the size of the @xmath102-primary torsion of any rank two @xmath45-drinfeld module over @xmath84 , where @xmath84 ranges over all extensions for which the degree @xmath104 $ ] is bounded by a given constant .
this implies that the uniform boundedness conjecture for rank - two @xmath45-drinfeld modules over @xmath40 follows from the following statement : for fixed @xmath105 , there are only finitely many @xmath102 such that there exists an @xmath84-rational @xmath102-torsion point on an @xmath45-drinfeld module over @xmath84 with @xmath106 \leq d$ ] . for a rational function field @xmath52 , the above two corollaries were proven by schweizer @xcite .
the finite bound from these two results is not effective in the _ number _ of rational points . for effective results on the number of points of low degree on some drinfeld modular curves ,
see for example armana @xcite .
no analogue of merel s theorem ( uniform boundedness of torsion ) is currently known for rank - two drinfeld modules ( compare also poonen @xcite ) . as a final remark , there has recently been a surge in the use of gonality and graph theory in arithmetic , but mainly in characteristic zero ; for example in the work of ellenberg , hall and kowalski on generic large galois image , coupling gonality to expander properties of cayley graphs embedded in riemann surfaces @xcite . also in our applications , in a rather different way , the graph expansion properties of the reduction graphs of drinfeld modular curves seem to intervene in a crucial way in establishing interesting lower bound for their gonality ( originally , over rational function fields , we deduced the bounds from natural bounded concentrator properties of subgraphs , as in the work of morgenstern @xcite ) .
more generally , the stable gonality in a family of ramanujan graphs with fixed regularity is bounded below linearly in the number of vertices ( cf .
remark [ rama ] ) .
[ ntn-0 ] let @xmath0 denote a complete nonarchimedean valued field with valuation ring @xmath14 with uniformizer @xmath107 and residue field of characteristic @xmath108 .
for any @xmath14-scheme @xmath15 we denote by @xmath109 ( respectively @xmath18 ) the generic fiber ( respectively the closed fiber ) .
we denote by @xmath110 the singular locus of @xmath15 .
let @xmath5 be a geometrically connected projective smooth curve over @xmath0 .
@xmath14-model _ of @xmath5 is a pair @xmath111 consisting of an integral normal scheme @xmath15 that is projective and flat over @xmath14 and a @xmath0-isomorphism @xmath112 .
an @xmath14-model @xmath15 of @xmath5 is said to be _ semi - stable _ if its special fiber @xmath18 is reduced with only ordinary double points as singularities .
such a model is called _ stable _ if any irreducible component of the special fiber has a finite automorphism group as a marked curve , where the marking is given by its intersection points with other components . as was shown by liu and lorenzini in @xcite , every finite morphism @xmath113 between geometrically connected projective smooth curves over
@xmath0 extends to a morphism between the stable models of @xmath5 and @xmath46 , but the resulting map is _ not necessarily finite_. ( similar problems were already encountered and studied by abhyankar in @xcite . )
however , there exists a semi - stable model admitting an extension of the map that is a finite morphism , as was shown by coleman @xcite and liu @xcite .
we need a slightly different statement , that we prove along similar lines as liu : [ theorem-21 ] let @xmath113 be a finite morphism between geometrically connected projective smooth curves over @xmath0 , and @xmath15 an @xmath14-model of @xmath5 .
then there exist a finite separable field extension @xmath114 , semi - stable @xmath115-models @xmath116 and @xmath117 of @xmath118 and @xmath119 , respectively , over the integral closure @xmath115 of @xmath14 in @xmath120 , and an @xmath115-morphism @xmath121 such that the following conditions are satisfied@xmath122 * @xmath116 dominates @xmath123 * @xmath124 is finite , surjective , and extends @xmath125 * the induced morphism @xmath126 satisfies @xmath127 .
the proof is a slight modification of the proof of proposition 3.8 in @xcite .
we first prove the theorem in the special case where @xmath128 is a finite galois covering .
let @xmath11 be the galois group of @xmath128 .
then , replacing @xmath0 by a finite separable extension if necessary , @xmath5 has a semi - stable model @xmath129 that dominates @xmath15 and admits an extension of the @xmath11-action ( see corollary 2.5 in @xcite ) .
we want to modify this to a semi - stable model with _ inversion - free _
action , as follows .
suppose an element @xmath130 of order two interchanges two components @xmath131 and @xmath132 ( possibly @xmath133 ) intersecting at a node @xmath134 .
then we blow - up @xmath129 at the closed point @xmath134 ; we do this at all such nodes .
the exceptional curves have multiplicity two .
then we replace @xmath0 by a ramified quadratic extension @xmath120 , and take the normalization to obtain a model @xmath116 of @xmath118 ; it is clear that the @xmath11-action extends to @xmath116 .
the quotient @xmath135 is a semi - stable model of @xmath119 ( see proposition 1.6 in @xcite ) , and the quotient map @xmath121 has the desired properties ; we postpone the verification of property ( c ) .
next , we treat the case where @xmath128 is separable .
let @xmath136 denote the galois closure of @xmath113 .
then , replacing @xmath0 by a finite separable extension if necessary , we may assume that @xmath136 is smooth over @xmath0 .
as in the proof of proposition 3.8 in @xcite , replacing @xmath0 furthermore by a finite separable extension if necessary , one has a semi - stable model @xmath137 of @xmath136 that dominates @xmath15 and admits an extension of the action of @xmath138 . as in the first part ,
we modify @xmath137 to an inversion - free semi - stable model @xmath139 ( after replacing @xmath40 by a finite separable extension ) . then the obvious map @xmath140 where @xmath141 , gives the desired model of @xmath128 , as we will see soon below .
in general , we decompose @xmath128 into a finite separable @xmath142 followed by a purely inseparable frobenius map @xmath143(see proposition 3.5 in @xcite ) .
the first part @xmath144 of the decomposition has an @xmath115-model @xmath145 obtained as above . setting @xmath146
, we find that the composite map @xmath147 gives the answer .
the @xmath115-morphism @xmath121 thus obtained has properties ( a ) and ( b ) . in order to show that ( c ) holds , it suffices to show that neither of the following two situations occurs : *
there exists a double point @xmath134 of @xmath148 that is mapped to a smooth point of @xmath149 ; * there exists a smooth point @xmath134 of @xmath148 that is mapped to a double point of @xmath149 .
one can see from the construction ( due to the ` inversion - free ' nature ) above that the situation ( i ) does not occur .
finally , situation ( ii ) is also excluded due to proposition 1.6 in @xcite .
[ notgr ] let @xmath11 be a connected finite graph .
in this paper , a graph can have multiple edges ( this is sometimes called a `` multigraph '' , but we will not use this terminology ) .
we denote the sets of vertices and edges by @xmath150 and @xmath151 , respectively .
we denote by @xmath32 the cardinality @xmath152 of the vertex set . by @xmath153
we denote the set of edges connecting two vertices @xmath154 , and more generally , for two subsets @xmath155 , we denote by @xmath156 the set of edges in @xmath11 that connect elements from @xmath45 to elements from @xmath157 : @xmath158 our graphs are , unless clearly indicated , undirected , i.e. , @xmath159 . in case
we have an oriented edge we will write @xmath160 for an edge with source @xmath161 and target @xmath82 .
the set of edges incident to a given vertex @xmath161 is denoted by @xmath162 .
the number of edges in @xmath162 , where , as usual , edges in @xmath163 ( viz . , loops )
are counted with multiplicity two , is called the _ degree _ or _ valancy _ of @xmath161 , and is written @xmath164 .
a graph is called @xmath0-_regular _ if @xmath165 for all @xmath166 .
a graph is called _ loopless _ if @xmath167 for all @xmath168 .
two vertices @xmath169 are called _ adjacent _ if @xmath170 , and we denote it by @xmath171 . for a subset @xmath172 the _ volume _ is defined to be @xmath173 in particular , @xmath174 .
another important invariant of a graphs is the _ genus _ , by which we mean the first betti number @xmath175 .
note that this differs from another convention in graph theory in which `` genus '' means the minimal genus of a riemann surface in which the graph can be embedded without self - intersection .
a graph of genus 0 is called a _
tree_. functions @xmath176 , are simply called `` functions on @xmath11 '' .
these form a finite dimensional vector space , equipped with the standard inner product @xmath177 denote by @xmath178 the adjacency matrix of a connected graph @xmath11 ( of which the @xmath160-entry is @xmath179 ) and by @xmath180 the diagonal matrix with the degrees of the vertices on the diagonal .
then the laplace operator is defined by @xmath181 . for any graph
, @xmath30 is a real symmetric positive - semidefinite matrix , and therefore has non - negative real eigenvalues .
the function @xmath182 , defined as being identically equal to @xmath183 on @xmath184 , is an eigenfunction of @xmath30 with eigenvalue @xmath185 .
the other eigenvalues are positive .
we order the eigenvalues @xmath186 where @xmath23 is the number of vertices of the graph .
it is the first non - zero eigenvalue which is important for us ; we denote it by @xmath187 .
[ nl ] sometimes , one uses the _
normalized laplacian _ of @xmath11 , defined as @xmath188 weighted by vertex degrees ( compare chung @xcite ) .
a graph @xmath11 is called _ stable _ if all vertices have degree at least @xmath26 .
a graph @xmath12 is called a _ refinement _ of @xmath11 if it can be obtained from @xmath11 by performing subsequently finitely many times one of the two following operations : 1 .
subdivision of an edge , 2 .
addition of a _ leaf _ ,
i.e. , the addition of an extra vertex and an edge between this vertex and a vertex of the already existing graph . one of the main tools in this paper is the notion of harmonic morphisms of graphs as developed by urakawa @xcite and baker and norine @xcite , and later generalized to harmonic indexed morphisms by caporaso @xcite .
we will use a terminology that is compatible with that of @xcite , and different from @xcite , and we will only consider `` unweighted '' graphs in the sense of @xcite . in the appendix , we will discuss the relations between different notions of gonality for graphs .
[ defmor ] let @xmath189 be two loopless graphs . 1
. a _ finite morphism between @xmath11 and @xmath12 _ ( denoted by @xmath190 ) is a map @xmath191 such that @xmath192 and for every @xmath193 , @xmath194 , together with , for every @xmath195 , a positive integer @xmath196 , the _ index _ of @xmath197 at @xmath198 .
2 . a finite morphism is called _ harmonic _ if for every @xmath199 there exists a well - defined number , @xmath200 , such that for every @xmath201 we have @xmath202 this does not make sense if @xmath203 , but then we postulate that @xmath204 can be chosen to be any positive integer .
3 . for a finite harmonic morphism
the following number , which is called the _ degree _ of @xmath197 , is independent of @xmath205 or @xmath206 ( @xcite , lemma 2.3 ) : @xmath207 from the perspective of this paper , it is natural to define the following notion of gonality ( this is different from existing notions of gonality , but we will discuss these in the appendix ) . a graph @xmath11 is called _ stably
_ @xmath105-_gonal _ if it has a refinement that allows a degree @xmath105 finite harmonic morphism to a tree .
stable gonality _ of a graph @xmath11 is defined to be @xmath208 with @xmath12 a refinement of @xmath11 and @xmath197 a finite harmonic morphism to a tree @xmath209 .
although here , finite harmonic morphisms are defined only for loopless graphs , stable gonality is defined for all graphs , as loops can be `` refined away '' by subdividing the loop edges .
alternatively , one may extend the definition of `` harmonic '' to graphs with loops , as in @xcite .
[ examplegon1 ] the `` banana graph '' @xmath20 ( see figure [ figbanana ] ) given by two vertices joined by @xmath21 distinct edges is the intersection dual graph of two rational curves intersecting in @xmath23 points .
the minimal degree of a finite harmonic morphism from @xmath20 to a tree is @xmath23 .
however , if we subdivide each edge once , the resulting graph admits such a finite harmonic morphism of degree @xmath22 to a tree , which is a vertex with @xmath23 edges sticking out ( by identifying the two original vertices ) .
hence the banana graph has stable gonality equal to @xmath22 .
this is compatible with the fact that the banana graph can be the intersection dual graph of both hyperelliptic and non - hyperelliptic ( if @xmath210 ) curves , and these are not distinguished by all subdivisions of their reduction graph .
this example occurs in nature as the stable reduction of the modular curve @xmath211 over @xmath212 , where @xmath23 is then the number of supersingular elliptic curves modulo @xmath213 .
one should observe ( @xcite , 3.6 ) that stable reduction graphs are naturally _ metric _ graphs , and as such , the stable reduction graph of @xmath211 is only equal to the ( unit - length metrized ) banana graph for @xmath214 with @xmath215 . with a finite harmonic morphism of ( minimal ) degree @xmath23 , and its subdivision , with a finite harmonic morphism of degree @xmath22 ( all indices are @xmath183).,width=188 ] [ examplegon2 ]
the minimal degree of a finite harmonic morphism from the complete graph @xmath24 to a tree is @xmath25 ( this can be checked by a somewhat tedious enumeration ) , but by adding leaves , such a morphism of degree @xmath26 can be constructed , see figure [ figk4 ] . with a finite harmonic morphism of degree @xmath25 , and a refinement ( with leaves ) of @xmath24 with a finite harmonic morphism of degree @xmath26 ( all indices are @xmath183).,width=340 ]
let @xmath15 and @xmath216 be @xmath14-models of geometrically connected projective smooth curves over @xmath0 , and @xmath217 an @xmath14-morphism . let us say @xmath124 is _ inversion - free semi - stable _ if the following conditions are satisfied : * @xmath15 and @xmath216 are semi - stable ; * @xmath124 is finite and surjective ; * @xmath218 .
theorem [ theorem-21 ] says that any finite cover @xmath113 between geometrically connected projective smooth curves over @xmath0 admits , after replacing @xmath0 by a finite separable extension , an inversion - free semi - stable model @xmath128 ; moreover , given an arbitrary @xmath14-model @xmath15 of @xmath5 , we can take such an @xmath14-model @xmath121 of @xmath128 such that @xmath116 dominates @xmath15 .
let @xmath217 be an inversion - free semi - stable model of @xmath128 .
consider the dual graphs @xmath219 and @xmath220 of the special fibers of @xmath18 and @xmath221 , respectively .
the vertices of @xmath96 ( respectively @xmath1 ) correspond to irreducible components of @xmath18 ( respectively @xmath221 ) , and two of them are connected by an edge if and only if they intersect . the morphism @xmath124 induces the following two set - theoretic maps : * since @xmath124 is finite , it maps each component of @xmath18 surjectively onto a component of @xmath221 ; in particular , it induces a map @xmath222 between the sets of vertices of the graphs ; * due to the condition ( c ) above , each double point of @xmath18 is mapped to a double point of @xmath221 ; that is , we have the map @xmath223 between the sets of edges .
thus we obtain a graph map @xmath224 we now assume that @xmath128 is _ separable _ , and define the index @xmath225 for such @xmath128 .
let @xmath226 be an edge with extremities @xmath227 .
let @xmath228 ( respectively @xmath229 ) be the components of @xmath18 ( respectively @xmath221 ) corresponding to @xmath230 ( respectively @xmath231 ) , respectively .
the maps @xmath232 and @xmath233 ramify at the intersection point @xmath134 with the same decomposition group ; then define @xmath234 to be the order of this group . in this way
, @xmath235 becomes a finite morphism of graphs in the sense of definition [ defmor ] .
[ prop-31 ] for a separable @xmath113 that admits an inversion - free semi - stable model @xmath217 , the finite graph morphism @xmath236 constructed above is harmonic of degree @xmath237 @xmath238 in the sense of definition [ defmor]@xmath239 .
let @xmath240 be a vertex , and @xmath241 ( respectively @xmath242 ) the component of @xmath18 ( respectively @xmath221 ) corresponding to @xmath243 ( respectively @xmath244 ) .
let @xmath245 be the degree of the covering map @xmath232 .
then for any edge @xmath246 emanating from @xmath244 , we have @xmath247 and hence @xmath235 is harmonic of degree @xmath237 .
[ cor-31 ] let @xmath5 be a geometrically connected projective smooth curve over @xmath0 , and @xmath15 the stable @xmath14-model of @xmath5 , and let @xmath17 denote the intersection dual graph of the special fiber .
let @xmath16 be an algebraic closure of @xmath0 .
then we have @xmath19 gonality is the minimal degree of a map @xmath248 .
since we work over an algebraically closed field , we can decompose such a map into a separable part @xmath249 and a purely inseparable part @xmath250 . since the genus is preserved by the purely inseparable part , we find that @xmath251 , too , and hence the separable part of a general map is a map of lower degree to @xmath53 .
hence we can restrict to bounding the degree of a separable @xmath128 .
the assertion now follows from proposition [ prop-31 ] and the following auxiliary observations .
\(1 ) by theorem [ theorem-21 ] , for any given finite cover @xmath252 , replacing @xmath0 by a finite separable extension , one has an inversion - free semi - stable model @xmath253 of @xmath128 such that @xmath116 dominates @xmath15 .
in particular , @xmath254 gives a graph that arises from @xmath17 by subdividing some edges ( corresponding to blowing up nodes ) and/or adding some leaves ( corresponding to blowing up smooth points ) this is exactly the notion of refinement as we have defined it .
\(2 ) by replacing the base field @xmath0 by an arbitrary finite extension @xmath120 , the base - change @xmath255 , where @xmath115 is the integral closure of @xmath14 in @xmath120 , is a semi - stable model of @xmath118 ( see section 1.5 in @xcite ) , which obviously gives the same dual graph as @xmath17 .
one can use this comparison theorem to prove the analog for stable gonality of graphs of the upper bound for the gonality of curves given by brill
noether theory : a curve of genus @xmath256 over an algebraically closed field has gonality bounded above by @xmath257 ; this was proven in general by kleiman and laksov @xcite . to prove this for graphs ,
we first show that finite harmonic morphisms can be `` refined '' , in a sense to be made precise .
for any two refinements @xmath258 and @xmath259 of a graph @xmath11 let @xmath260 be the set of all common refinements of @xmath258 and @xmath259 .
a refinement @xmath12 of a graph @xmath11 induces refinements of all of its subgraphs . if @xmath261 is an edge in @xmath11 that connect two vertices @xmath262 , denote by @xmath263 $ ] the subgraph of @xmath11 consisting of the vertices @xmath243 and @xmath264 joined by the edge @xmath198 .
similarly , for a vertex @xmath243 , we denote by @xmath265 $ ] the subgraph which consists of @xmath243 only .
denote with @xmath266 $ ] the refinement of @xmath267 $ ] in @xmath12 , and for an edge @xmath268 denote with @xmath269 $ ] the restricted refinement : @xmath270=g'[e]-(g'[v]-[v])-(g'[w]-[w]).\ ] ] [ defref ] a _ refinement of a finite harmonic morphism _
@xmath271 is a finite harmonic morphism @xmath272 such that @xmath12 ( respectively @xmath273 ) is a refinement of @xmath11 ( respectively @xmath209 ) , and such that 1 . for all @xmath199 , @xmath274 ; 2 .
for any @xmath262 and any edge @xmath261 , every refinement of @xmath263 $ ] in @xmath12 is mapped to the refinement of @xmath275 $ ] in @xmath273 , viz . , @xmath276
) = t'[\varphi(e)];\ ] ] 3 .
[ indexedge ] for every @xmath195 and for all @xmath277 $ ] , the index @xmath278 .
it follows that @xmath279 .
let @xmath280 be a finite harmonic morphism
. then 1 .
[ boompje ] for any refinement @xmath273 of @xmath209 , there exists a refinement @xmath281 of @xmath124 ; 2 .
[ lemref ] for any refinement @xmath49 of @xmath11 , there exists a refinement @xmath282 of @xmath124 such that @xmath12 is a refinement of @xmath49 . for part ( i ) , use the following recipe : 1 .
[ verfijndeboom ] replace every edge @xmath198 in @xmath11 by @xmath283 $ ] ; 2 . put indices such that conditions ( [ indexedge ] ) in definition [ defref ] is satisfied ; 3 .
extend @xmath124 in the obvious way to a finite harmonic morphism . for part ( ii )
, first choose for every edge @xmath284 in @xmath209 an element in @xmath285),\ ] ] and replace @xmath284 by this common refinement . call the resulting new graph @xmath273 , and then apply part ( i ) .
[ examplerefine ] in figure [ figrefine ] , one sees a finite harmonic morphism @xmath286 on the left , where all edges have index @xmath183 , except the indicated edge that has index two .
the middle picture is a refinement @xmath49 of the original graph @xmath11 , and the right hand picture shows the refinement @xmath287 as constructed in lemma [ lemref ] .
both morphisms have degree @xmath26 . ;
middle : a refinement @xmath49 of @xmath11 ; right : a refinement @xmath287 with @xmath12 a refinement of @xmath49 ( only indices @xmath288 are depicted).,width=491 ] [ steq ] call two graphs _ equivalent _ if they are refinements of the same stable graph .
this defines an equivalence relation on the set of all graphs of genus at least @xmath22 .
the map @xmath289 is defined on equivalence classes of graphs .
let @xmath12 be a refinement of @xmath11 .
it follows from the definition that @xmath290 . since refinement of morphisms preserves degree ,
the previous lemmas imply that the other inequality @xmath291 also holds .
[ bnn2 ] for any graph @xmath11 of genus @xmath27 , the brill noether bound holds : @xmath28 since @xmath289 is defined on the equivalence classes of graphs it is sufficient to prove the bound for one representative of each equivalence class .
it is sufficient to show that any stable graph @xmath11 of genus @xmath27 admits a refinement @xmath12 such that there exists a curve @xmath5 such that @xmath12 is the dual graph of the minimal model of @xmath5 . indeed , since the genus of @xmath5 equals the genus of @xmath12 , which equals @xmath256 ( since the genus of a graph does nt change under refinement ) , the classical bound @xmath292 holds ( cf .
laksov @xcite ) .
the result follows from @xmath293 ( theorem [ compare ] ) .
we now show the existence of such a refinement .
let @xmath11 be a stable graph of genus @xmath27 and let @xmath294 .
choose @xmath256 edges @xmath295 of @xmath11 such that @xmath296 is a tree .
replace each edge @xmath297 ( connecting two vertices @xmath298 and @xmath299 ) by two edges @xmath300 $ ] and @xmath301 $ ] , where @xmath302 and @xmath303 are new vertices not connected to any other vertex .
in this way , @xmath11 is replaced by a tree @xmath304 .
choose an embedding of @xmath304 in the bruhat - tits tree @xmath305 for @xmath306 , where @xmath41 satisfies @xmath307 .
denote the images of @xmath302 and @xmath303 in @xmath305 by the same letters .
now choose hyperbolic elements @xmath308 in @xmath309 such that each @xmath310 acts as translation along a geodesic through @xmath302 and @xmath303 , and @xmath311 .
then @xmath312 is a schottky group .
denote by @xmath313 the subtree of @xmath305 spanned by the limit set of @xmath1 .
then @xmath314 , where @xmath12 is the refinement of @xmath11 given by subdividing each of the edges @xmath295 once . also , @xmath12 is the intersection dual graph of the minimal model of the mumford curve corresponding to @xmath1 ( @xcite , page 164 )
. it would be interesting to have a purely graph theoretical proof of the above result .
more general lifting results , such as lemma 6.3 in @xcite or corollary b.3 from @xcite with theorem [ theorem-21 ] , also imply the existence of the refinement and the curve .
in this section , we prove an analogue of the li yau bound , viz . , a spectral lower bound for the stable gonality of a graph .
the basic philosophy of the proof is to find a lower bound for the first laplace eigenvalue using its variational characterization , in terms of the degree of a finite harmonic morphism @xmath315 and the minimal `` size '' of the inverse image of the two parts in which the tree gets cut by removing one of its edges .
then a dichotomy occurs : either the minimal such size is large , or there is a vertex with a large inverse image .
initially , `` large '' depends on the maximal vertex degree of the tree @xmath209 , but if this degree is too large , we change the refinement and morphism to produce a lower bound that only depends on the original graph , not the morphism or tree itself .
we start by studying such `` sizes '' on trees abstractly : a _ measured tree _
@xmath316 is a connected tree @xmath209 with a probability measure @xmath317 on @xmath318 . for an edge @xmath319
, we decompose @xmath320 into its two connected components @xmath321 and @xmath322 : @xmath323 we define the _ size _ of an edge @xmath319 ( w.r.t .
@xmath317 ) by @xmath324 let @xmath325 . call a measured tree @xmath316 _ @xmath326-thick _ if for every vertex @xmath327
, the graph @xmath328 has a connected component of measure at least @xmath326 .
[ amini ] a @xmath326-thick measured tree @xmath316 has an edge of size at least @xmath326 . for any vertex @xmath329
, choose a connected component @xmath330 of @xmath328 of measure at least @xmath326 .
orient the unique edge that connects @xmath161 to a vertex in @xmath330 in the direction of @xmath161 . by doing this for each vertex , @xmath331 different orientations
are assigned to the @xmath332 edges of @xmath209 .
hence at least one edge of @xmath209 is oriented in both directions , and such an edge has size at least @xmath326 .
[ oldcase ] if @xmath316 is not @xmath326-thick , then there exists a vertex @xmath327 with @xmath333 . indeed , since @xmath328 has @xmath164 components , all of measure less than @xmath326 , we find that @xmath334 the measure we will use counts vertices of @xmath12 that belong to the original graph @xmath11 : let @xmath11 denote a graph , and @xmath12 a refinement of @xmath11 .
the probability measure @xmath335 on @xmath336 is defined by @xmath337 [ trivialcase ] if @xmath12 is a refinement of a graph @xmath11 , and @xmath315 a finite harmonic morphism to a tree , then @xmath338 is a measured tree , and for any vertex @xmath327 , we have @xmath339 it suffices to remark that @xmath340 the next proposition says that size and degree controls the first eigenvalue of the laplacian : [ cheeger ] if @xmath12 is a refinement of a graph @xmath11 , and @xmath315 a finite harmonic morphism , then for any edge @xmath341 , we have an inequality @xmath342 if we let @xmath343 for @xmath344 then the statement to be proven is equivalent to @xmath345 first , note that the inequality is trivial if @xmath346
. now assume the minimum is non - zero .
the estimate follows from the variational characterization of @xmath29 via the rayleigh - quotient , @xmath347 where notations are as in [ notgr ] .
we construct an appropriate function @xmath128 based on the finite harmonic morphism @xmath348 and the removed edge @xmath349 , as follows : @xmath350 it is easy to check that @xmath351 , and therefore , @xmath352 we finish the proof by showing that @xmath353 .
suppose an edge @xmath354 is replaced in @xmath12 by a path , possibly of length @xmath183 .
let us describe this path as a series of edges @xmath355 , such that @xmath356 is incident to a vertex in @xmath258 , and @xmath357 is incident to a vertex in @xmath259 .
then for at least one of the @xmath297 it holds that @xmath358 . the desired inequality follows .
a lower bound for the degree of @xmath124 now follows easily if the map gives a @xmath326-thick measured tree : [ firstcase ] if @xmath12 is a refinement of a graph @xmath11 , and @xmath315 a finite harmonic morphism such that @xmath359 is @xmath326-thick , then @xmath360 immediate from lemma [ amini ] and proposition [ cheeger ] .
if the tree is _ not
_ @xmath326-thick , we know from remark [ oldcase ] that the tree has a vertex @xmath361 with `` large '' measure : @xmath362 putting , for example , @xmath363 ( where @xmath364 is the maximal vertex degree in @xmath209 ) , the previous results gives a non - trivial lower bound on the degree of @xmath124 in terms of @xmath365 and @xmath364 of the form @xmath366 however , we want to find a bound that solely depends on @xmath11 , not on @xmath209 . for this
, in the next proposition , we engineer another harmonic morphism from a different refinement of @xmath11 , but whose degree is controlled by that of @xmath124 . [ change ] let @xmath11 denote a graph with maximal degree @xmath367 .
let @xmath368 be constants such that @xmath369 .
if @xmath12 is a refinement of @xmath11 , and @xmath315 a finite harmonic morphism such that 1 .
@xmath359 is not @xmath370-thick ; and 2 .
all vertices @xmath361 have measure @xmath371 then there exists a refinement @xmath372 of @xmath11 , a tree @xmath373 , and a finite harmonic morphism @xmath374 such that 1 .
@xmath375 ; and 2 .
there exists an edge @xmath376 of @xmath373 with @xmath377 we postpone the proof to the next section , and first discuss the main corollary .
[ deafschatting ] let @xmath11 be a graph with maximal vertex degree @xmath367 and first laplace eigenvalue @xmath29 .
the stable gonality of @xmath11 is bounded from below by @xmath33 let @xmath11 be a graph and let @xmath378 be a finite harmonic morphism , and let @xmath379 be as in proposition [ change ] . if @xmath359 is @xmath380-thick , then by corollary [ firstcase ] , we find @xmath381 on the other hand , if there is a vertex @xmath361 with @xmath382 then by lemma [ trivialcase ] , we get @xmath383 in the remaining case , proposition [ change ] implies that @xmath384 where the second inequality follows from proposition [ cheeger ] .
translating the constraints @xmath385 and @xmath369 , we conclude that it always holds that @xmath386 with @xmath387 , @xmath388 and @xmath389 .
the maximum in ( [ simp ] ) is achieved for @xmath390 and @xmath391 , and plugging this back into ( [ simp ] ) gives the result .
in the proof of proposition [ change ] we will use the following concept several times to construct new graphs from old : let @xmath392 denote @xmath23 different graphs and let @xmath393 denote tuples of their vertices .
the graph @xmath49 obtained by _ gluing _
@xmath394 along these vertices is defined to be @xmath395 [ restrictieisharmonic ] let @xmath396 be a finite harmonic morphism and let @xmath397 be a connected subgraph . then the restriction of @xmath197 to any of the connected components of @xmath398 is a finite harmonic morphism to @xmath399 .
let @xmath399 be such a subgraph and let @xmath400 be the restriction of @xmath197 to one of the connected components @xmath401 of @xmath402 .
then @xmath403 is well - defined for all @xmath404 , namely , all edges @xmath405 which are mapped to an edge @xmath406 are contained in @xmath407 , so that in particular @xmath408 .
the proof consists of various steps , in which the new morphism is constructed from the original map . as the proof proceeds , we will show the constructions on an explicit non - trivial example .
the example is printed in smaller font , and the switch between example and main proof is indicated by a diamond ( @xmath409 ) .
the basic idea of the proof is this : the hypothesis of the proposition ( which is the remaining bad case from the point of view of getting useful bounds ) is that there exists a vertex in the tree ( with controlled measure ) such that all connected components of its complement have too small measure to give a useful bound .
we collect these components ( and their preimages ) into two sets , which we call the `` left '' and `` right '' parts , such that each of these parts has a larger , more useful , measure . for technical reasons
, one should discard leaves first .
we want to map each of these parts as a whole onto a `` smaller '' tree , but this will increase the degree of the map .
we control it as follows : in the new tree @xmath373 , all of these components are glued to a central vertex .
the original morphism @xmath124 is split into local parts over each of the components , and these are refined to a harmonic morphism over the left or right parts of the new tree .
then the left and right parts are glued together over the new tree , where the indices at preimages of the central point are redefined , and maybe some leaves are added , to make the result into a harmonic morphism .
one then checks that the degree of the new morphism hasnt increased too much , but that both edges sticking out of the central vertex have large enough measure to give a useful bound .
+ * example . * we start with a graph @xmath11 of the following form ( 1)(2 ) ( 1)(10 ) ( 2)(10 ) ( 2)(7 ) ( 2)(4 ) ( 2)(3 ) ( 3)(4 ) ( 3)(6 ) ( 4)(5 ) ( 5)(6 ) ( 5)(7 ) ( 6)(7 ) ( 7)(8 ) ( 7)(9 ) ( 8)(9 ) ( 8)(10 ) the original graph has @xmath410 vertices and maximal vertex degree @xmath411 .
we consider the following refinement @xmath12 of the graph @xmath11 , and the harmonic morphism @xmath124 to the tree @xmath209 as indicated in the following picture .
we use the following _ display conventions : _ the original vertices and edges are bold , contrary to subdivision vertices and vertices and edges from leaves . the label on a vertex indicates to what vertex in the tree it is mapped .
a gray square box with a number on an edge indicates that this edge has index equal to that number ; if there is no box , then the index is @xmath183 .
( 5,0 ) node[pos=.5,right ] @xmath124 ; ( 1)(2 ) ( 2)(10 ) ( 2)(4 ) ( 2)(3 ) ( 5)(7 ) ( 6)(7 ) ( 7)(8 ) ( 7)(9 ) ( 1)(11 ) ( 9)(12 ) ( 2)(13 ) ( 2)(14 ) ( 7)(15 ) ( 7)(16 ) ( q1)(q2 ) ( q2)(q4 ) ( q5)(q6 ) ( q6)(q4 ) ( q4)(q3 ) ( q4)(q7 ) ( q4)(q8 ) ( q4)(q9 ) ( q4)(q10 ) ( 1)(19 ) ( 19)(10 ) ( 2)(18 ) ( 18)(7 ) ( 3)(21 ) ( 21)(4 ) ( 3)(22 ) ( 22)(6 ) ( 5)(23 ) ( 23)(6 ) ( 8)(20 ) ( 20)(9 ) ( 4)(24 ) ( 24)(5 ) ( 8)(17 ) ( 17)(10)2 ( 19)(19a ) ( 19)(19b ) ( 19)(19c ) ( 19)(19d ) ( 19)(19e ) ( 20)(20a ) ( 20)(20b ) ( 20)(20c ) ( 20)(20d ) ( 20)(20e ) ( 22)(22a ) ( 22)(22b ) ( 22)(22c ) ( 22)(22d ) ( 22)(22e ) ( 22a)(22aa ) ( 22b)(22bb ) ( 23)(23a ) ( 23)(23b ) ( 23)(23c ) ( 23)(23d ) ( 23)(23e ) ( 23a)(23aa ) ( 23b)(23bb ) ( 21)(21a ) ( 21)(21b ) ( 21)(21c ) ( 21)(21d ) ( 21)(21e ) ( 21a)(21aa ) ( 21b)(21bb ) ( 24)(24a ) ( 24)(24b ) ( 24)(24c ) ( 24)(24d ) ( 24)(24e ) ( 24a)(24aa ) ( 24b)(24bb ) this map has degree @xmath412 .
the push - down @xmath413 of the normalized counting measure of @xmath11 to @xmath209 takes the following value on the indicated vertices : @xmath414 if we set @xmath415 then the set - up satisfies conditions ( i ) and ( ii ) of proposition [ change ] .
indeed , the push - down measure is not @xmath416-thick ( all connected components of @xmath417 are of measure at most @xmath418 ) , and all vertices have measure smaller than @xmath419 . in the proof , we construct a new refinement @xmath372 of @xmath11 and a new tree @xmath373 with a finite harmonic morphism @xmath420 of degree @xmath421 with an edge of size @xmath422 . @xmath409
+ write @xmath423 . since @xmath316 is not @xmath380-thick , we can choose a vertex @xmath424 such that all components of @xmath425 have measure @xmath426 .
let @xmath427 be the refinement of @xmath11 which only comprises the subdivided edges of @xmath11 in @xmath12 , and let @xmath428 denote its image @xmath429 .
observe that @xmath430 , and let @xmath105 denote the degree of @xmath431 in @xmath428 .
also note that the maximal degree of @xmath432 is the same as that of @xmath11 , i.e. , @xmath433 .
denote by @xmath434 the connected components of @xmath435 .
we divide the connected components into two sets of approximately the same measure .
since @xmath436 and @xmath437 for all @xmath438 , it is possible to find a partition @xmath439 , such that @xmath440 * example ( continued ) . * in the example , @xmath432 is the graph ( 1)(2 ) ( 2)(10 ) ( 2)(4 ) ( 2)(3 ) ( 5)(7 ) ( 6)(7 ) ( 7)(8 ) ( 7)(9 ) ( 1)(19 ) ( 19)(10 ) ( 2)(18 ) ( 18)(7 ) ( 3)(21 ) ( 21)(4 ) ( 3)(22 ) ( 22)(6 ) ( 5)(23 ) ( 23)(6 ) ( 8)(20 ) ( 20)(9 ) ( 4)(24 ) ( 24)(5 ) ( 8)(17 ) ( 17)(10 ) and @xmath428 is the graph ( q2)(q4 ) ( q5)(q6 ) ( q6)(q4 ) ( q4)(q3 ) ( q4)(q7 ) ( q4)(q8 ) ( q4)(q9 ) ( q4)(q10 ) we can choose @xmath431 to be the central vertex labelled `` a '' , so that all components of @xmath425 have measure @xmath441 ( indeed , they have measure @xmath185 , @xmath418 or @xmath442 ) , and divide the @xmath443 connected components into `` left '' and `` right '' as follows : @xmath444 then the total measure of the left part is @xmath445 and the total measure of the right part is @xmath445 , which is larger than @xmath446 . @xmath409 + we now show how to construct the different pieces of the new map @xmath447 .
+ * the construction of @xmath373 . * for each @xmath438 let @xmath448 be the unique vertex which is adjacent to @xmath431 in @xmath428 .
a new graph @xmath449 is obtained by gluing all @xmath450 together at the @xmath299 , and adding a leaf at the image of the @xmath299 .
call @xmath161 the new vertex of the added leaf .
take two copies @xmath451 and @xmath452 of the pair @xmath453 , and glue them together at @xmath454 and @xmath455 to obtain a tree @xmath373 .
let @xmath456 be the image of @xmath454 and @xmath455 in @xmath373 , and call the image of @xmath457 in @xmath373 the `` left part '' @xmath458 , and the image of @xmath459 in @xmath373 the `` right part '' @xmath460 .
+ * example ( continued ) . * in the example , @xmath449 is the graph given by gluing all components @xmath461 of @xmath435 along their vertex adjacent to @xmath431 , and adding a leaf .
all but the component @xmath462 are isolated vertices , so the result is a segment isomorphic to @xmath463 , connected to a new vertex @xmath161 at @xmath464 : @xmath449 is ( q5)(q6 ) ( q6)(q4 ) .
@xmath373 is the graph given by gluing two copies of @xmath453 along the common vertex @xmath161 , so @xmath373 is [ scale=0.4,every node/.style = transform shape , baseline=-222 ] ( center)(links1 ) ( links1)(links2 ) ( center)(rechts1 ) ( rechts1)(rechts2 ) .@xmath409
+ * the construction of @xmath372 . * for each @xmath438 , let @xmath465 be the subtree of @xmath428 obtained by adding to @xmath450 the ( unique ) edge in @xmath466 .
the subgraph @xmath467 might be disconnected ; let @xmath468 be the union of the connected components of @xmath469 for which the set of edges has a non - empty intersection with @xmath470 . by lemma [ restrictieisharmonic ]
the restriction @xmath471 is finite harmonic .
now observe that @xmath449 is a refinement of @xmath465 , so by lemma [ boompje ] there exists a finite harmonic refinement morphism ( in the sense of definition [ defref ] ) @xmath472 with an inclusion map @xmath473 .
+ * example ( continued ) . * for each of the seven connected components @xmath474 , we display the construction of the `` local '' components @xmath465 , @xmath468 and the refinement morphism @xmath475 in table [ ttable ] .
the construction for @xmath476 is entirely similar to the one for @xmath477 ( with the label @xmath105 replaced by @xmath478 , respectively ) , so we do nt list it in the table . @xmath409
lmmmm @xmath479 & @xmath480 & @xmath481 & @xmath482 & @xmath483 + @xmath465 & ( q6)(q4 ) & ( q5)(q6 ) ( q6)(q4 ) & ( q6)(q4 ) & ( q6)(q4 ) + + @xmath468 & ( 1)(19 ) ( 1)(2 ) ( 9)(7 ) ( 9)(20 ) & ( 2)(10 ) ( 7)(8 ) ( 10)(19 ) ( 8)(17 ) ( 17)(10 ) ( 8)(20 ) & ( 2)(18 ) ( 18)(7 ) & ( 2)(3 ) ( 3)(21 ) ( 3)(22 ) + + @xmath484 & ( 1)(19 ) ( 1)(2 ) ( 9)(7 ) ( 9)(20 ) ( 11)(1 ) ( 12)(9 ) ( 3,1 )
( 3,0 ) node[pos=.5,right ] @xmath485 ; ( q5)(q6 ) ( q6)(q4 ) & ( 2)(10 ) ( 7)(8 ) ( 10)(19 ) ( 8)(17 ) ( 17)(10)2 ( 8)(20 ) ( 3,1 ) ( 3,0 ) node[pos=.5,right ] @xmath486 ; ( q5)(q6 ) ( q6)(q4 ) & ( 2)(18 ) ( 18)(7 ) ( f)(18 ) ( 5,1 ) ( 5,0 ) node[pos=.5,right ] @xmath487 ; ( q5)(q6 ) ( q6)(q4 ) & ( 2)(3 ) ( 3)(21 ) ( 3)(22 ) ( 3)(ff ) ( 8,1 ) ( 8,0 ) node[pos=.5,right ] @xmath488 ; ( q5)(q6 ) ( q6)(q4 ) + * the construction of @xmath372 ( continued ) .
* define , for any @xmath489 , the integer @xmath490 the graph @xmath372 is obtained by gluing all @xmath491 together at @xmath492 for all @xmath493 , and , for any @xmath243 with @xmath494 , gluing an additional copy @xmath495 of @xmath453 at @xmath243 .
+ * example ( continued ) . * in the example , the vertices @xmath496 are the numbered vertices in the following display of @xmath432 : ( 1)(2 ) ( 2)(10 ) ( 2)(4 ) ( 2)(3 ) ( 5)(7 ) ( 6)(7 ) ( 7)(8 ) ( 7)(9 ) ( 1)(19 ) ( 19)(10 ) ( 2)(18 ) ( 18)(7 ) ( 3)(21 ) ( 21)(4 ) ( 3)(22 ) ( 22)(6 ) ( 5)(23 ) ( 23)(6 ) ( 8)(20 ) ( 20)(9 ) ( 4)(24 ) ( 24)(5 ) ( 8)(17 ) ( 17)(10 ) the vertices labeled 1 and 5 have only neighbouring subgraphs @xmath497 with @xmath498 , and the indices are all one , and add up to @xmath22 .
the vertices labeled 3 , 4 , 7 and 8 have only neighbouring subgraphs @xmath497 with @xmath499 , and the indices are all one , and add up to @xmath22 .
finally , the vertices labeled 2 and 6 have neighbouring subgraph @xmath500 with index in @xmath501 , and neighbouring subgraphs @xmath502 and two of @xmath497 ( with @xmath503 ) with index in @xmath504 , for which all the indices are all one .
thus , @xmath505 thus , we need to glue in one extra copy of @xmath449 at every such vertex . @xmath409 + * the construction of @xmath506 . * to define @xmath506 , it suffices to define its restriction to @xmath497 for @xmath438 , and its restriction to @xmath507 , and show that these are compatible on intersections .
define the restrictions as follows : 1 .
if @xmath498 , set @xmath508 with index @xmath509 for all @xmath510 ; 2 . if @xmath499 , set @xmath511 with index @xmath509 for all @xmath510 ; 3 .
if @xmath512 with @xmath513 , set @xmath514 with index @xmath515 for all @xmath516 ; 4 .
if @xmath512 with @xmath517 , set @xmath518 with index @xmath519 for all @xmath516 .
one checks that this glues together correctly to a finite graph morphism @xmath447 .
+ * example ( continued ) . * the extra copies of @xmath449 that are glued to vertices 1 , 3 , 4 , 5 , 7 , 8 get index 2 on every edge , but the copies that are glued to 2 and 6 get index one .
the final re - engineered map @xmath420 in our example is given as follows : ( 5,1.5 )
( 5,-0.5 ) node[pos=.5,right ] @xmath520 ; ( 1)(2 ) ( 2)(10 ) ( 2)(4 ) ( 2)(3 ) ( 5)(7 ) ( 6)(7 ) ( 7)(8 ) ( 7)(9 ) ( 1)(19 ) ( 19)(10 ) ( 2)(18 ) ( 18)(7 ) ( 3)(21 ) ( 21)(4 ) ( 3)(22 ) ( 22)(6 ) ( 5)(23 ) ( 23)(6 ) ( 8)(20 ) ( 20)(9 ) ( 4)(24 ) ( 24)(5 ) ( 8)(17 ) ( 17)(10)2 ( 1)(11 ) ( 9)(12 ) ( 2)(13 ) ( 13)(14a ) ( 7)(15 ) ( 15)(16a ) ( 19)(19a ) ( 19a)(19b ) ( 20)(20a ) ( 20a)(20b ) ( 21)(21a ) ( 21a)(21b ) ( 22)(22a ) ( 22a)(22b ) ( 23)(23a ) ( 23a)(23b ) ( 24)(24a ) ( 24a)(24b ) ( 18)(18a ) ( 3)(25 ) ( 5)(26 ) ( center)(links1 ) ( links1)(links2 ) ( center)(rechts1 ) ( rechts1)(rechts2 ) here , the vertex @xmath456 is labelled `` a '' .
the map @xmath124 has degree @xmath521 and the edges @xmath522 and @xmath523 of @xmath373 both have size @xmath524 . @xmath409 +
* the finite morphism @xmath506 is harmonic . * we check that @xmath525 is well - defined for all @xmath526 . for
all @xmath527 there is either a unique @xmath438 such that @xmath528 and then @xmath529 , or there is a unique @xmath530 such that @xmath531 and then @xmath532 . for all @xmath533 ,
it holds that @xmath534 * the edge @xmath376 .
* any of the two edges @xmath535 satisfies @xmath536 * * the degree of @xmath506.**consider a vertex @xmath537 : if @xmath243 belongs to @xmath432 , then it belongs to at most @xmath367 different @xmath491 for @xmath438 ; and if @xmath538 , then there is a unique @xmath438 such that the unique path from @xmath243 to @xmath432 is contained in @xmath539 .
it follows that for each @xmath540 , at most @xmath367 of the neighboring @xmath497 are either all sent to @xmath458 , or all to @xmath460 . hence @xmath541 and this implies @xmath542 .
the point @xmath431 ( used in the proof ) with the property that all components of @xmath425 have measure @xmath426 is in fact _
unique_. indeed , if there are two such vertices , say , @xmath431 and @xmath454 , then let @xmath543 denote any edge on a path between @xmath431 and @xmath454 .
one component of @xmath328 contains @xmath321 and one component of @xmath544 contains @xmath322 , and hence by assumption , @xmath545 and @xmath546 but @xmath547 .
hence @xmath548 , but this is impossible with @xmath549 and @xmath550 .
we now give some examples that illustrate the bound .
[ bananensplit ] for the banana graph @xmath20 , we have @xmath551 , @xmath552 and @xmath553 , so the lower bound is trivial : @xmath554 .
however , the stable gonality of @xmath20 ( for @xmath555 ) , equals @xmath22
. see figure [ figbanana ] for such a map of degree @xmath22 .
[ asm ] for the complete bipartite graph @xmath556 , we have @xmath557 , @xmath558 and @xmath559 . if @xmath23 is even , then the lower bound is @xmath560 we expect that the stable gonality of @xmath556 equals @xmath23 . a morphism which attains degree @xmath23
is given by mapping @xmath556 to the star with one central vertex and @xmath23 emanating edges in the obvious way .
for @xmath561 ( @xmath213 prime ) , the graph @xmath556 occurs as stable reduction graph of the curve @xmath562 ( seen in @xmath563 ) with @xmath564 , over a valued field @xmath565 of characteristic @xmath213 , which , as a fiber product of two projective lines , admits an obvious morphism of degree @xmath566 to @xmath53 .
the stable reduction itself consist of two transversally intersecting families of @xmath566 rational curves ( `` check board with @xmath567 squares '' ) . for more details on these curves , see for example @xcite . in the family of example [ asm ] ,
our lower bound has the same order of growth in @xmath23 as does the expected gonality , but is about 5 times as small .
this seems to be a general phenomenon ; we do nt know an interesting example where our lower bound is sharp .
[ rama ] if @xmath568 is a family of ramanujan graphs with ( fixed ) regularity @xmath105 and @xmath23 vertices ( @xmath23 increasing ) , then by the alon - boppana bound , we get inequalities @xmath569 so we find a lower bound of the form @xmath570 for @xmath23 sufficiently large with @xmath571 a constant only depending on @xmath105 . in any family of mumford curves
whose stable reduction graphs are @xmath105-regular ramanujan graphs , the gonality goes to infinity as the number of components of the stable reduction does so .
[ lyfout ] the famous li yau inequality from differential geometry @xcite states that the gonality @xmath36 of a compact riemann surface @xmath5 ( minimal degree of a conformal mapping @xmath124 of @xmath5 to the riemann sphere ) is bounded below by @xmath37 where @xmath38 is the first non - trivial eigenvalue of the laplace - beltrami operator of @xmath5 , and @xmath39 denotes the volume of @xmath5 . for graphs @xmath11 with any laplacian ( normalized or not )
, an inequality of the form @xmath572 for some constant @xmath573 fails .
a counterexample is given by the complete graph @xmath574 , which has stable gonality @xmath575 .
however , a lower bound of the form @xmath576 would be @xmath577 for the usual laplacian , and @xmath578 for the normalized laplacian ( see table [ table ] in the appendix for the data ; one deduces that the analog of the li yau inequality also fails if one uses any of the other notions of gonality from the existing literature and are outlined in the appendix . )
one sees from our result that in a graph , the constant @xmath573 needs to be roughly divided by the maximal edge degree for such an inequality to hold . as we have seen in corollary [ steq ] , stable gonality is defined on equivalence classes of graphs , in the sense that two graphs @xmath11 and @xmath12 are equivalent ( notation @xmath579 ) if they are refinements of the same stable graph .
hence the result also implies that for any graph @xmath11 with @xmath27 , we have @xmath580 @xmath581 it is tempting to consider the limiting value for the lower bound in this theorem when the graph is further and further refined . whereas it is clear how the number of vertices and the maximal vertex degree change under refinements , the change of the eigenvalue under refinements is not so well - understood ( apart from regular graphs ) . for applications in solid state physics , eichinger and martin
have developed an algorithm that computes the change in eigenvalues under refinement by applying only linear algebra to the original laplace matrix @xcite .
examples ( such as the banana graph ) suggest that ( iterated ) refinement might worsen the lower bound .
there is a similar result for the normalized laplacian .
denote with @xmath582 the first non - trivial eigenvalue of @xmath583 .
[ genormaliseerdegrens ] let @xmath11 be a graph with maximal degree @xmath367 and first normalized laplace eigenvalue @xmath584 .
the stable gonality of @xmath11 is bounded from below by @xmath585 the proof is virtually the same as theorem [ spectral ] , so instead of providing all details , we briefly outline the differences . instead of @xmath335 , we use the measure @xmath586 on @xmath587 defined for @xmath588 by @xmath589 where @xmath590 is the degree of @xmath243 in @xmath11 .
lemma [ amini ] is valid for all probability measures , and therefore also for @xmath591 . since @xmath592
counts the number of edges instead of vertices , the conclusion of lemma [ trivialcase ] changes to @xmath593 the analogue of proposition [ cheeger ] can be derived by using the test function @xmath594 proposition [ change ] does not change for this new measure .
we conclude that the proof of corollary [ deafschatting ] only changes in the step where lemma [ trivialcase ] is used . for @xmath0-regular graphs @xmath595 and
hence the lower bounds are identical for the two different laplace operators . generically ( in the sense of algebraic geometry )
, the gonality of a curve attains the brill noether bound ( cf .
for example the references in the appendix of @xcite ) .
however , curves of fixed genus and fixed stable reduction graph can have widely varying gonality ( e.g. , the banana graph @xmath20 has stable gonality @xmath22 , but its subdivisions can occur as stable reduction graph of curves whose gonality takes on all the values @xmath596 ) ; in particular , the gonality of the curve can be much higher than the stable gonality of the reduction graph .
one may try to find the most probable stable gonality of a random connected ( multi-)graph and compare it to the most probable value of the brill noether bound . in this remark
, we compute something much simpler : the difference between the expected value of the brill noether bound and the lower bound in our theorem , for the erds
rnyi random graph model with a specific connection probability in the non - sparse region . for a random graph model @xmath597 of erds - rnyi type @xcite with @xmath23 vertices and edge probability @xmath598 for some @xmath599 , the threshold for almost sure connectivity holds ,
the expected number of edges is @xmath600 , so the first betti number of @xmath601 is @xmath602 almost surely .
chung , lu and vu @xcite have shown that ( for a class of function including these @xmath603 ) the normalized eigenvalue @xmath604 tends to @xmath183 with high probability .
also , the given assumptions imply that @xmath605 in probability ( @xcite , 3.14 ) .
hence the lower bound tends with high probability to @xmath606{2g},\ ] ] which is sublinear in @xmath256 ( and for @xmath607 , tends to @xmath608 , up to a constant the actual value of the stable gonality @xmath609 for the complete graph @xmath610 of genus @xmath611 ) , whereas the brill noether bound is linear in @xmath256 ( which happens if @xmath612 ) .
there are at least two ways to interpret this heuristic observation : either the lower bound is asymptotically bad for random graphs ; or stable gonality of random graphs is significantly lower than generic gonality of curves .
we recall the main concepts and notations from the theory of general drinfeld modular curves , cf .
@xcite , @xcite .
let @xmath40 denote a global function field of a smooth projective curve @xmath5 over a finite field @xmath613 with @xmath41 elements and characteristic @xmath614 , and @xmath43 a place of degree @xmath44 of @xmath40 .
let @xmath615 denote a uniformizer at @xmath43 .
let @xmath45 denote the subring of @xmath40 of elements that are regular outside @xmath43 .
let @xmath46 denote a rank - two @xmath45-lattice in the completion @xmath47 of @xmath40 at @xmath43 .
such lattices are classified up to isomorphism by their determinant , so they are isomorphic to @xmath616 , where @xmath617 runs through a set of representatives of @xmath618 , the ideal class group of @xmath45 .
let @xmath619 denote the automorphism group of the lattice @xmath46 : @xmath620 and let @xmath1 denote a congruence subgroup of @xmath57 .
this means that @xmath1 contains a principal congruence group @xmath621 as a finite index subgroup , where @xmath622 for @xmath59 an ideal in @xmath45 .
let @xmath623 denote the center of @xmath619 .
if @xmath624 is the `` standard '' lattice , we revert to the standard notations @xmath625 and @xmath626 .
the groups @xmath1 act by fractional transformations on the drinfeld space @xmath627 , where @xmath628 is the completion of an algebraic closure of @xmath47 .
the quotient @xmath629 is an analytic smooth one - dimensional space , and is the analytification of a smooth affine algebraic curve @xmath630 , that can be defined over a finite abelian extension of @xmath40 inside @xmath47 .
it can be compactified to a drinfeld modular curve @xmath61 by adding finitely many points , called cusps .
the @xmath628-points of the ( coarse ) moduli scheme @xmath631 of rank - two drinfeld @xmath45-modules with full level @xmath59-structure ( i.e. , an isomorphism of @xmath632 with the torsion of the drinfeld module ) can be described as @xmath633 we denote such a component by @xmath634 , and its compactification by @xmath70 .
[ dringon2 ] let @xmath1 denote a congruence subgroup of @xmath62 .
then the gonality of the corresponding drinfeld modular curve @xmath61 satisfies @xmath63\ ] ] where the constant @xmath64 is @xmath65 this implies a linear lower bound in the genus of modular curves of the form @xmath66 where @xmath635 is a bound that depends only on the function field @xmath40 . if @xmath40 is a rational function field and @xmath68 , then we can put @xmath69 .
first observe that @xmath636 , so we now consider @xmath61 as a curve over @xmath50 and are in a set up where we can apply our previous results .
the remainder of the proof has various parts .
* reduction to principal congruence groups .
* first of all , we observe that it suffices to prove the bound for the groups @xmath621 . indeed ,
if @xmath637 is a morphism , then from the inclusion @xmath638 we get a composed morphism @xmath639 of degree @xmath640}{|\gamma \cap z| } \cdot \deg \varphi,\ ] ] and hence @xmath641.\ ] ] therefore , the desired inequality @xmath642\ ] ] follows from @xmath643.\ ] ] we now prove the gonality bound by invoking theorem [ spectral ] for the reduction graph of the drinfeld modular curve @xmath70 . * semistable model .
* first , we construct a semi - stable model for the reduction of @xmath70 at @xmath43 .
the groups @xmath644 also act by automorphisms on the bruhat
tits tree @xmath305 of @xmath645 @xcite .
the quotient @xmath646 is the union of a finite graph @xmath647 and a finite number of half lines in correspondence with the cusps of @xmath61 , and the curve @xmath61 is a mumford curve over @xmath47 @xcite such that the intersection dual graph of the reduction , which is a finite union of rational curves over @xmath648 intersecting transversally in @xmath648-rational points , equals the finite graph @xmath647 ( @xcite ( 2.7.8 ) ) .
in particular , the genus of the modular curve @xmath61 equals the first betti number of this graph ( compare @xcite v.a.11 ) .
now consider the @xmath1-_stable _ part @xmath649 of @xmath305 , defined to consist of those vertices and edges of @xmath305 that have trivial stabilizer for the action of @xmath1 .
since the stabilizers of cusps are non - trivial and stabilizers of edges are subgroups of stabilizers of adjacent vertices , the stable part is a tree that ends in half - edges ( i.e. , edges with only an initial vertex ) .
let @xmath650 denote the tree obtained from deleting the half - edges , and call the images in @xmath651 of the remaining vertices that were incident to the half - lines the _ boundary points _ ; say there are @xmath652 of those , see figure [ drinfeld ] . , including stable , unstable and cuspidal part.,width=377 ] we claim that the quotient @xmath653 is a semistable reduction graph for @xmath61 .
since it is a subgraph of @xmath647 , it suffices to check that these graphs have the same genus .
this can be seen as follows .
since @xmath644 is @xmath654-torsion free , @xcite ii.2.9 , ex .
2b and thm .
13(c ) imply that @xmath655 where @xmath464 is the number of cusps of @xmath1 and @xmath656 is the euler - poincar characteristic of @xmath11 . since @xmath657 and @xmath658 , we find @xmath659 now observe that @xmath660 ; indeed , @xmath464 is the number of half - lines of @xmath646 .
a priori several half - lines could be attached to the same boundary point . on the other hand ,
all paths in the unstable graph attached to boundary vertices have to be part of infinite half - lines ( so correspond to cusps ) . indeed , if there would be such a _ finite _ path @xmath661 , let @xmath662 denote a connected lift of @xmath661 to @xmath305 and @xmath663 any half - line of @xmath305 that contains @xmath662 .
then the projection of @xmath663 in @xmath646 has to be infinite , since the orders of the stabilizers are strictly increasing along @xmath663 , and hence it would have to intersect the stable part @xmath664 .
this is impossible , since in the stable part , the stabilizers are trivial . since also @xmath665
, we conclude that @xmath666 and @xmath667 .
this means that @xmath646 is @xmath668 connected via @xmath464 paths to the @xmath464 cusps .
if not indicated otherwise , choose @xmath669 and write @xmath670 .
* a lower bound on the number of vertices . * in the bruhat tits tree @xmath305 of @xmath645 ,
every vertex is ( @xmath671)-regular . let us consider the special vertex of @xmath305 corresponding to the class of the trivial rank - two vector bundle @xmath672 $ ] on @xmath5 , and let @xmath673 denote the corresponding vertex in @xmath674 .
the stabilizer of this vertex is precisely @xmath675 ( namely , an element of the stabilizer induces an automorphism of the `` star '' of the vertex , which is given by @xmath676 . )
the stabilizer intersects @xmath677 ( where @xmath678 is the center ) in the `` constant group '' @xmath679 , and the group @xmath644 ( for @xmath669 ) in the trivial group .
there is a covering map @xmath680 we conclude that @xmath681,\ ] ] since the right hand side is the number of vertices in @xmath646 above @xmath673 which are stable , since they have trivial stabilizers in @xmath1 , and @xmath682 has cardinality @xmath683 .
this estimate for the number of vertices of @xmath11 will be enough for our purposes , since it differs from the index only by a constant in @xmath41 .
but one might also count the total number of vertices of the graph . for a rational function field
@xmath52 with a place @xmath43 of degree one , this is easily done , the result being @xmath684;\ ] ] compare also with computations in @xcite ( cf .
@xcite , @xcite ) and @xcite .
it seems another proof of the lower bound on the gonality is possible by using morgenstern s result that there is a perfect matching between a very large ( constant fraction depending only on @xmath41 , not on @xmath685 ) subset of the vertices above @xmath673 in @xmath686 and vertices in the complement , but we did not pursue this , since it would give a less general and worse result .
the gonality is _ not _ always realized by the obvious map @xmath687 .
for example , set @xmath688 and let @xmath102 denote an prime of degree @xmath26 ; then the modular curve @xmath689 is hyperelliptic , but the map @xmath690 has degree @xmath691 .
also notice that for a general base field @xmath40 , the modular curve @xmath692 is not even itself a rational curve .
counting the number of cusps ( so the number of vertices above a vertex in @xmath693 corresponding to a split bundle of high degree ) is not enough to get a linear estimate in the index , since the cusps have rather large stabilizers ( of size roughly the third root of the index ) .
* vertex degrees . * since the tree @xmath305 is @xmath694-valent and @xmath664 consists of stable vertices ( that have trivial stabilizers ) , we find that all vertices in @xmath664 are @xmath694-valent . in particular , for the maximal vertex degree , we find @xmath695 the boundary vertices @xmath243 in @xmath696 have valency @xmath697 , since we have already shown that they are connected to a unique cusp .
* the first eigenvalue of the laplace operator .
* we will relate the laplace operator on the finite graph @xmath670 to a hecke operator on ( quotients of ) the bruhat - tits tree .
the hecke operator that we consider is @xmath698 , associated to the characteristic function of the double coset @xmath699 equivalently , it acts on the vertices of the bruhat - tits tree @xmath305 of @xmath645 as the adjacency operator of @xmath305 ; so @xmath700 for a function @xmath128 on @xmath701 .
now @xmath698 descends to a hecke operator on @xmath702 by @xmath703 where we set @xmath704 $ ] ( compare also @xcite ) .
now consider the adjacency operator @xmath705 of @xmath706 , and suppose that @xmath128 is a function on the vertices of @xmath11 that is an eigenfunction for @xmath705 with eigenvalue @xmath97 .
we claim that it extends uniquely to an eigenfunction @xmath707 of @xmath698 on the vertices of @xmath646 . indeed , on all non - boundary vertices of @xmath11 , @xmath708 .
then , if @xmath243 is a boundary vertex of @xmath705 , and @xmath264 is the unique vertex outside @xmath11 that is adjacent to @xmath243 , we want @xmath709 hence we should define @xmath710 .
finally , if @xmath711 are three consecutive vertices outside @xmath11 , then by computing @xmath712 , we see that we need to define @xmath713 now @xmath714 with the weight function @xmath715 forms a _ diagram _ in the sense of morgenstern ( @xcite , @xcite ) , and @xmath707 is an eigenfunction for @xmath698 in the space @xmath716 , the orthocomplement of the constant functions in the space of square integrable functions on the vertices for the measure given by the weights @xmath715 . as in theorem 2.1 of @xcite ,
the theory of eisenstein series implies that the continuous spectrum of @xmath698 belongs to the segment @xmath717 $ ] , and drinfeld s proof of the ramanujan - petterson conjecture for function fields ( in a series of papers culminating in @xcite ) shows that the discrete eigenvalues @xmath97 of @xmath698 on @xmath716 satisfy @xmath718 .
we conclude that @xmath718 holds for the eigenvalues of @xmath705 on @xmath11 .
the degree matrix of @xmath11 is given by a diagonal matrix @xmath719 where the lower block corresponds to the boundary vertices ( which are @xmath671-valent in @xmath646 , but only @xmath697-valent in @xmath11 , since the cusps are not present in @xmath11 ) .
finally , the laplacian of @xmath11 is @xmath720 the courant - weyl inequalities ( e.g. , theorem 2.1 in @xcite ) imply that @xmath721 is larger than the first eigenvalue of @xmath722 ( plus the smallest eigenvalue of @xmath157 , which is zero ) , leading to @xmath723 * conclusion of the proof of the main bound . *
since the function @xmath724 is monotonously increasing in @xmath97 , we find the result by plugging the data from equations ( [ een ] ) , ( [ twee ] ) and ( [ drie ] ) in the lower bound from theorem [ spectral ] .
* linear lower bound in the genus . *
we now show how to convert the lower bound on the gonality of @xmath61 in terms of the index @xmath725 $ ] into a lower bound that is linear in the genus , of the form @xmath726 for @xmath727 a constant depending only on the ground field @xmath40 and the degree @xmath44 of @xmath43 .
this is not entirely obvious in positive characteristic , due to wild ramification .
first of all , it is again enough to establish such a bound for a principal congruence subgroup @xmath621 .
first , recall the riemann - hurwitz formula for a galois cover @xmath728 with galois group @xmath11 : @xmath729 where @xmath730 are the higher ramification groups of any preimage of @xmath82 in @xmath5 ( see e.g. @xcite ) . applying this to the ( galois ) cover ( [ princov ] ) and using formula ( [ prinineq ] )
, it follows that @xmath731}|\gamma\cap z| \\ & \geq \frac{c'_{k,\delta } ( g(x(y,\operatorname{\mathfrak{n}}))-1)}{[\gamma:\gamma(y,\operatorname{\mathfrak{n } } ) ] } |\gamma\cap z|\\ & \geq c'_{k,\delta } ( g(x_\gamma)-1+r ) \\ & \geq c'_{k,\delta } ( g(x_\gamma)-1 ) , \end{aligned}\ ] ] where we have assumed that the desired bound holds for @xmath621 , and @xmath732 comes from formula ( [ rh ] ) applied to the cover ( [ princov ] ) .
we now establish the bound for @xmath70 .
if this curve has genus zero or one , the required bound for the gonality holds trivially .
therefore , we can assume @xmath733 .
the riemann - hurwitz formula for the cover @xmath734 implies a relation of the form @xmath735 = ( g(x(y,\operatorname{\mathfrak{n}}))-1 ) \cdot
\frac{2(q-1)}{2g(x(y))-2 + r},\ ] ] where @xmath14 is the term in equation [ rh ] applied to the galois cover @xmath734 . hence to prove our result
, it suffices to prove a bound of the form @xmath736 for some constant @xmath737 depending only on @xmath40 and @xmath44 .
we recall some information about the `` ramification number '' @xmath14 and the genus @xmath738 from @xcite ( there , the formulae are worked out for the principal component @xmath739 only , but hold in general ) .
first of all , the genus of @xmath740 depends only on @xmath40 and @xmath44 .
secondly , ramification takes place above elliptic points and cusps of @xmath740 .
let us write @xmath741 with @xmath742 the contribution from elliptic points , and @xmath743 the contribution from cusps .
the ramification above elliptic points is tame ; and the number of elliptic points depends only on @xmath40 and @xmath44 . hence @xmath742 is bounded above by a constant in @xmath40 and @xmath44 .
the ramification above the cusps is wild , but _
weak _ ; this means that the second ramification groups are trivial , and the first ramification group is just the @xmath213-sylow group of the stabilizer of the cusp ( this follows , for example , from the fact that @xmath740 are mumford curves , hence ordinary since their jacobian admits a tate uniformization , and hence has maximal @xmath213-rank , by applying a result of nakajima @xcite ) . in the end
, we need an upper bound on @xmath744 where @xmath745 , that is independent of @xmath105 ; for example , @xmath746 ( the limit of @xmath743 as @xmath105 tends to @xmath747 ) will do , and this finishes the proof . in the `` standard '' case of a rational function field @xmath52 with a place @xmath43 of degree one
, one can make all data explicit .
the cover @xmath748 is ramified tamely at the unique elliptic point , of order @xmath749 , and at the unique cusp , of order @xmath750 , where @xmath751 .
hence the riemann - hurwitz formula becomes @xmath752 \left(1-\frac{1}{q+1 } -\frac{1}{q^d(q-1 ) } - \frac{1}{q^d } \right ) \\ & \leq [ \gamma(1):\gamma(\operatorname{\mathfrak{n } } ) ] , \end{aligned}\ ] ] and it follows that one can set @xmath753 in this case .
the previous best ( non - linear ) bounds were due to schweizer ( @xcite , thm .
2.4 ) , who showed that if @xmath40 is a rational function field , then one has , for example , @xmath754^{\frac{q-1}{2q}}.\ ] ]
assume that @xmath40 is a global function field , @xmath43 a place of @xmath40 , and let @xmath88 denote an elliptic curve over @xmath40 with split multiplicative reduction at @xmath43 ( every non - isotrivial curve acquires such a place of reduction after a finite extension of the ground field @xmath40 ) . from the work of drinfeld , it follows that @xmath88 admits a _ modular parametrization _
@xmath755 ( see gekeler and reversat @xcite ) for some suitable modular curves @xmath76 .
this parametrization is defined over the maximal abelian extension @xmath49 of @xmath40 that is contained in the completion @xmath47 .
one may study the ( minimal ) degree of such a modular parametrization , called the _ modular degree_.
contrary to the case of elliptic curves over @xmath93 , in the case where @xmath52 , gekeler has proven that the modular degree always equals the congruence number of the associated automorphic form @xcite @xcite .
we first describe some of the structure of the modular curves @xmath76 . the scheme @xmath756 , ( coarsely ) representing the moduli problem of rank - two drinfeld modules with an @xmath59-isogeny , is defined over @xmath40 , but is not absolutely irreducible
if @xmath48 is non - trivial ; it decomposes over @xmath628 as @xmath757 where the components are defined over @xmath49 , and sharply transitively permuted by the galois group @xmath51 .
one may also describe the modular parametrizations for different @xmath46 simultaneously by a @xmath40-rational map @xmath758 , with @xmath756 not absolutely irreducible .
since the elliptic curve @xmath88 admits a map of degree two to @xmath53 , we find that @xmath759 since we now have a lower bound @xmath760,\ ] ] we conclude that @xmath761.\ ] ] the desired result @xmath762 follows from the following lemma .
@xmath763 \geq |\operatorname{\mathfrak{n}}|_\infty .
$ ] since both groups @xmath62 and @xmath764 contain the center @xmath678 , this index is the degree of the covering @xmath765 although the different components @xmath76 of @xmath756 and @xmath740 of @xmath766 depend on @xmath46 , they are galois conjugate by @xmath51 . therefore , the covering _ degree _ of this cover does not depend on @xmath46 .
hence we can put @xmath55 , and a standard computation then shows that there is a bijection @xmath767 and hence @xmath768 = [ \operatorname{\mathrm{gl}}(2,a):\gamma_0(\operatorname{\mathfrak{n } } ) ] = |\operatorname{\mathfrak{n}}|_\infty \cdot \prod_{\operatorname{\mathfrak{p}}\mid \operatorname{\mathfrak{n } } } ( 1+|\operatorname{\mathfrak{p}}|_\infty^{-1 } ) \geq |\operatorname{\mathfrak{n}}|_\infty,\ ] ] as was to be proven .
[ moddeg2 ] let @xmath74 denote an elliptic curve with split multiplicative reduction at the place @xmath43 , of conductor @xmath77 .
then the degree of a modular parametrization @xmath75 is bounded below by @xmath769 \geq \frac{1}{2 } c_{q,\delta } |\operatorname{\mathfrak{n}}|_\infty.\ ] ] previously , papikian @xcite had proven ( using spziro s conjecture for function fields and estimating symmetric square @xmath84-functions by the ramanujan conjecture ) that for @xmath52 a rational function field , @xmath85 where @xmath86 is the @xmath87-invariant of @xmath88 and @xmath89 is its inseparability degree .
he had also proven that @xmath770 if @xmath59 is prime and the curve is _ optimal _ , i.e. , of minimal modular degree in its isogeny class a.k.a . a strong weil curve ( @xcite , 1.3 ) .
our lower bound @xmath771 confirms a conjecture that he made in @xcite .
[ pal ] in the other direction , papikian @xcite has proven an upper bound on the modular degree of an optimal semistable elliptic curve @xmath88 with square - free conductor for a general function field , depending on the manin constant @xmath772 .
contrary to the case of elliptic curves over @xmath93 , one really needs to assume that the curve is optimal , because of the existence of isogenies of arbitrary high degree , arising from the frobenius operator .
one should also note that in @xcite the bound is given without the manin constant as a factor , since it was at first conjectured to always equal one , but pl @xcite has given examples where this is not the case .
also , pl has proven a general upper bound for @xmath772 that combines with @xcite to give an upper bound of the form @xmath773 for the degree of an optimal modular cover @xmath235 with square - free conductor @xmath59 , and such that the class number of @xmath40 is coprime to the characteristic @xmath213 .
in particular , the analogue of the _ degree conjecture _
@xmath774 for any @xmath92 holds in this case , and combines with our lower bound .
papikian expects that the @xmath87-invariant of an optimal semi - stable elliptic curve over @xmath52 is separable , and then a lower bound of the form @xmath775 can be shown to hold in many cases , where @xmath772 is the manin - constant of @xmath88 .
the results of pl imply that @xmath772 can vary essentially from @xmath183 to @xmath776 , and thus , the value of the manin constant seems to influence how good our lower bound on the modular degree is .
we first quote the positive characteristic analogue of a theorem of frey @xcite : [ freyschw ] let @xmath5 denote a curve over a global function field @xmath40 , such that its jacobian does not admit a @xmath94-morphism to a curve defined over a finite field . if @xmath105 is an integer such that @xmath777 , then the set of points of degree @xmath105 on @xmath5 is finite , i.e. , @xmath778 \leq d } x(k ' ) \right| \leq \infty .
\box\ ] ] the result was proven in @xcite ( theorem 2.1 ) under the assumption that @xmath5 has a @xmath40-rational point ( similar to a hypothesis of frey ) , but clark has shown that this hypothesis is unnecessary , cf .
@xcite , theorem 5 .
this result has now been improved into a quantitative statement over more general fields by cadoret and tamagawa @xcite , as follows : recall that gonality may alternatively be defined as the minimal @xmath105 for which there exists a non - constant morphism from a @xmath53 to the @xmath105-th symmetric power @xmath779 of the curve @xmath5 .
define the _ isogonality _ @xmath780 of @xmath5 as the minimal @xmath105 for which there exists a non - constant morphism from a @xmath40-isotrivial curve to the @xmath105-th symmetric power @xmath779 of the curve @xmath5 .
then the result from @xcite says : _ for any finitely generated field @xmath40 of positive characteristic @xmath614 , and any smooth geometrically integral curve @xmath5 over @xmath40 , if @xmath105 is a natural number with @xmath777 and @xmath781 , then the set of points of degree @xmath782 on @xmath5 is finite .
_ we will typically apply our bound in the following situation : let @xmath40 be a finitely generated field , and @xmath0 the fraction field of an excellent discrete valuation ring , with @xmath783 ; for example , @xmath40 is a global function field and @xmath50 is the completion of @xmath40 at a place @xmath43 ; then @xmath784 , so the lower bound that we obtained for @xmath9 from the stable gonality of its reduction graph applies equally well to @xmath785 .
if @xmath786 is a mumford curve over a valued field @xmath787 , then its jacobian has split reduction , and hence it admits no map to an isotrivial curve , and @xmath788 proposition [ freyschw ] and our spectral bound on gonality now immediately imply the following general finiteness result for points on curves whose degree is bounded in terms of spectral data associated to a special fiber : [ genth2 ] let @xmath5 denote a curve over a global function field @xmath40 , such that its jacobian does not admit a @xmath94-morphism to a curve defined over a finite field .
let @xmath47 denote the completion of @xmath40 at a place @xmath43 , and let @xmath11 denote the stable reduction graph of @xmath95 .
let @xmath96 denote the maximal vertex degree of @xmath11 and @xmath97 the smallest non - zero eigenvalue of the laplacian of @xmath11 .
then the set @xmath98
\leq \frac{\lambda(|g|-1)-4\delta-4}{2\lambda+8\delta+8 } } x(k')\ ] ] of rational points on @xmath5 of degree at most @xmath99 is finite .
@xmath581 consider the curve @xmath789 from example [ asm ] , with @xmath790 of negative degree in @xmath209 .
observe that @xmath789 is a mumford curve over @xmath791 .
our gonality bound from example [ asm ] implies for example that if @xmath792 , then the set of points on @xmath789 of degree @xmath793 is finite .
note that the set of points of degree @xmath566 is infinite , so the result is best up to a constant ( for varying @xmath213 and @xmath794 ) .
we now further specialise the results to the case of drinfeld modular curves : [ tors2 ] if @xmath61 is defined over a finite extension @xmath100 of @xmath40 , then the set @xmath101 \leq \frac{1}{2 } \left ( c_{q,\delta } \cdot [ \gamma(1):\gamma]-1\right ) } x_\gamma(l)\ ] ] is finite .
the curves @xmath61 are mumford curves for the @xmath43-valuation .
therefore , the conditions to apply proposition [ freyschw ] are satisfied by @xmath795 and @xmath796 .
since @xmath1 is a congruence group , the curve @xmath61 is covered by some @xmath70 , and hence the curve @xmath61 is defined over @xmath49 .
hence one may always choose @xmath797 , but @xmath100 might be chosen smaller . also , drinfeld modular curves always have @xmath49-rational points , namely , the cusps , so the refinement of result [ freyschw ] by clark is not necessary for this application .
we also deduce the following analogue of a result of kamieny and mazur @xcite : [ freyan2 ] if @xmath102 is a prime ideal in @xmath45 , then the set of all rank two drinfeld @xmath45-modules defined over some field extension @xmath84 of @xmath40 that satisfies the degree bound @xmath103 \leq \frac{1}{2 } c_{q,\delta } \cdot |\operatorname{\mathfrak{p}}|_\infty\ ] ] that admit an @xmath84-rational @xmath102-isogeny is finite .
recall that the scheme @xmath798 , coarsely representing this moduli problem , decomposes over @xmath628 as @xmath799 where the components are defined over @xmath49 , and all components have @xmath49-rational points , namely , the cusps .
now a rank - two @xmath45-drinfeld module @xmath235 over a field @xmath84 with an @xmath84-rational @xmath102-isogeny gives rise to an @xmath84-rational point of @xmath798 , and hence to an @xmath800-rational point @xmath801 \in x_0(y,\operatorname{\mathfrak{p}})(hl)$ ] for some @xmath46 .
now the above theorem implies that @xmath802 \leq \frac{1}{2}(\mathrm{gon}_{{\overline}{h}}(x_0(y,\operatorname{\mathfrak{p}})-1 ) } x_0(y,\operatorname{\mathfrak{p}})(hl)\ ] ] is finite .
now since by theorem [ dringon ] , @xmath803,\ ] ] and we have @xmath804=|\operatorname{\mathfrak{p}}|_\infty+1,\ ] ] the result follows .
[ km2 ] fix a prime @xmath102 of @xmath45 .
there is a uniform bound on the size of the @xmath102-primary torsion of any rank two @xmath45-drinfeld module over @xmath84 , where @xmath84 ranges over all extensions for which the degree @xmath104 $ ] is bounded by a given constant .
the method of proof is similar to the one in kamieny - mazur @xcite , as used in @xcite , thm . 2.4 : the moduli space @xmath805 has only finitely many @xmath806-points as soon as @xmath807/c_{q,\delta})/\log_q(|\operatorname{\mathfrak{p}}|_\infty).\ ] ] for each of the finitely many drinfeld modules @xmath235 over @xmath806 corresponding to these points , breuer @xcite has shown that the open adelic image result of pink and rtsche @xcite implies that the @xmath102-primary torsion @xmath808 $ ] of @xmath235 is bounded by @xmath809 $ ] , where @xmath241 depends on @xmath235 , @xmath40 and @xmath102 .
one may now maximize the bound as @xmath235 runs through these finitely many drinfeld modules . also , for any drinfeld module @xmath235 , @xmath810 | \leq |\operatorname{\mathfrak{p}}|_\infty^{2(e-1)}.\ ] ] the result follows .
in general , @xmath104 $ ] is bounded from above by @xmath106 $ ] ( with equality if @xmath84 and @xmath49 are linearly disjoint ) .
this shows that a bound of the form @xmath106\leq d$ ] implies a bound of the form @xmath104
\leq d$ ] . hence the uniform boundedness conjecture for rank - two @xmath45-drinfeld modules over @xmath40 @xcite follows from the following statement :
for fixed @xmath105 , there are only finitely many @xmath102 such that there exists an @xmath84-rational @xmath102-torsion point on an @xmath45-drinfeld module over @xmath84 with @xmath106 \leq d$ ] .
in this appendix , we describe various other notions of graph gonality from the literature , and discuss the relation of stable gonality to these alternatives .
we first recall the notion of graph gonality from caporaso @xcite , but we change the terminology to be compatible with @xcite and the current paper . for the convenience of the reader
, we include a dictionary between the terminology in @xcite and this paper in table [ dict ] .
a _ morphism _ between two loopless graphs @xmath11 and @xmath12 ( denoted by @xmath190 ) is a map @xmath191 such that @xmath192 , and for every edge @xmath193 , either @xmath194 or @xmath811 and @xmath812 ; together with , for every @xmath195 , a non - negative integer @xmath196 , the _ index _ of @xmath197 at @xmath198 , such that @xmath813 if and only if @xmath811 . previously , in definition [ defmor ] , we only considered _ finite _ morphisms , which are morphisms that map edges to edges .
the notions of harmonicity and degree that we introduced in definition [ defmor ] make sense for morphisms , even if they are not finite .
a harmonic morphism is called _ non - degenerate _ if @xmath814 for every @xmath199 ( this is automatic if it is finite ) .
it suffices to prove that any non - degenerate harmonic morphism @xmath817 from a graph @xmath11 to a tree @xmath209 admits a refinement @xmath281 that is a _
finite _ harmonic morphism of the same degree as @xmath124 .
thus , let @xmath818 denote an edge that is mapped to a vertex @xmath819 .
add an extra leaf @xmath820 to @xmath209 at @xmath161 , subdivide @xmath198 into two edges @xmath821 and @xmath822 , and map both @xmath356 and @xmath823 to @xmath820 .
set @xmath824 for @xmath344 .
finally , add a leaf @xmath825 to all @xmath826 , map them all to @xmath820 , and set @xmath827 [ conn ] the gonality of a graph @xmath11 is bounded below by the edge - connectivity ( viz . , the number of edges that need to be removed from the graph in order to disconnect it ) : @xmath828 if @xmath11 is a simple graph ( i.e. , without multiple edges ) , unequal to a complete graph , then @xmath829 let @xmath830 denote a harmonic non - degenerate morphism .
choose any edge @xmath319 .
since removing @xmath198 from @xmath209 disconnects it , @xmath831 is a set of edges of @xmath11 whose removal disconnects @xmath11 .
hence @xmath832 for a simple graph which is not complete , the bound @xmath833 is one of the inequalities of fiedler @xcite ( 4.1 & 4.2 ) .
the `` trivial '' spectral bound in the above proposition is not very useful in practice , since it does not contain a `` volume '' term ( like the li yau inequality ) .
also , since every graph acquires edge connectivity two or one by refinements , the lower bound in the proposition trivializes under refinements ( which are required by the reduction theory of morphisms ) . another notion of gonality of graphs @xmath11 and , more generally , of _ metric _ graphs @xmath1 was introduced by baker in @xcite , defined as the minimal degree @xmath105
for which there is a @xmath834 on @xmath1 ( in analogy to the definition from algebraic geometry ) .
following caporaso , we call this gonality of graphs _ divisorial gonality_. in @xcite , caporaso has proven a brill noether upper bound for divisorial gonality . for a fixed unmetrized graph , there is in general no relation ( inequality either side ) between gonality and divisorial gonality , cf . @xcite remark 2.7 ( or the graphs on lines 4 and 7 of table [ table ] ) . since the reduction of a stable curve is naturally a metric graph ( cf .
@xcite ) , one should not ignore the metric in connection with gonality of curves .
baker has proven that _
the gonality of a curve @xmath5 is larger than or equal to the divisorial gonality of its metric reduction graph _
( @xcite , cor .
_ the stable gonality of a graph is larger than or equal to its _ stable _ divisorial gonality ( i.e. , the minimum of the divisorial gonality of all refinements ) , but the inequality can be strict .
_ this can be seen from the last line of table [ table ] , which shows an example of luo ye ( taken from @xcite ) of stable gonality @xmath25 but divisorial gonality @xmath26 ( for the metrization in which all edges have unit length ) .
the stable gonality can be smaller than the divisorial gonality ( without refinement ) , as one can see from the sixth line in table [ table ] ; but the next line of the table shows a refinement that lowers the divisorial gonality . the banana graph @xmath20 has divisorial and stable gonality @xmath22 but edge connectivity @xmath23 ( cf .
table [ table ] ) , showing that an equality analogous to the one in proposition [ conn ] can not hold for divisorial or stable gonality .
dion gijswijt remarked that @xmath835 with josse van dobben de bruyn , he has also proven that the divisorial gonality of a graph is larger than or equal to its treewidth ( unpublished , but some preliminaries can be found in @xcite ) , but the entries in table [ table ] show that the inequality can be strict .
lower bounds on treewidth imply such bounds on divisorial gonality ( e.g. , @xcite , @xcite ) .
it seems that our notion of stable gonality of a graph coincides with the notion of gonality introduced in @xcite from the viewpoint of tropical geometry .
the connection between tropical curves and metric graphs can already be found in mikhalkin @xcite , and the notion of harmonic morphism of metric graphs in anand @xcite .
we have collected some sample values in table [ table ] .
as above , @xmath29 is the first eigenvalue of @xmath30 , and @xmath836 is the first eigenvalue of the normalized laplacian @xmath837 ; @xmath838 is the edge connectivity , @xmath367 the maximal vertex degree , @xmath839 is the volume of the graph , @xmath840 its treewidth ; @xmath841 is the gonality , @xmath842 is the divisorial gonality , and @xmath843 is the stable gonality of @xmath11 .
we leave out the lengthy but elementary calculations ( for the divisorial gonality of @xmath610 , we refer to @xcite , 3.3 ) .
christopher kumar anand , _ harmonic morphisms of metric graphs , _ in : harmonic morphisms , harmonic maps , and related topics ( brest , 1997 ) , pp . 109112 , chapman & hall / crc res . notes math .
, vol . 413 , boca raton , fl , 2000 .
gunther cornelissen , fumiharu kato , and aristides kontogeorgis , _ the relation between rigid - analytic and algebraic deformation parameters for artin - schreier - mumford curves _ , israel j. math .
* 180 * ( 2010 ) , 345370 .
henry h. kim and peter sarnak , _ refined estimates towards the ramanujan and selberg conjectures , unctoriality for the exterior square of @xmath847 and the symmetric fourth of @xmath848 _ , j. amer .
* 16 * ( 2003 ) , no . 1 , 139183 . | we present a method to control gonality of nonarchimedean curves based on graph theory .
let @xmath0 denote a complete nonarchimedean valued field .
we first prove a lower bound for the gonality of a curve over the algebraic closure of @xmath0 in terms of the minimal degree of a class of graph maps , namely : one should minimize over all so - called finite harmonic graph morphisms to trees , that originate from any _ refinement _ of the dual graph of the stable model of the curve .
next comes our main result : we prove a lower bound for the degree of such a graph morphism in terms of the first eigenvalue of the laplacian and some `` volume '' of the original graph ; this can be seen as a substitute for graphs of the li yau inequality from differential geometry , although we also prove that the strict analogue of the original inequality fails for general graphs .
finally , we apply the results to give a lower bound for the gonality of arbitrary drinfeld modular curves over finite fields and for general congruence subgroups @xmath1 of @xmath2 that is linear in the index @xmath3 $ ] , with a constant that only depends on the residue field degree and the degree of the chosen `` infinite '' place .
this is a function field analogue of a theorem of abramovich for classical modular curves .
we present applications to uniform boundedness of torsion of rank two drinfeld modules that improve upon existing results , and to lower bounds on the modular degree of certain elliptic curves over function fields that solve a problem of papikian . |
on 2009 april 2 ( mjd 54923 ) the _ _ swift__@xmath0bat ( burst alert telescope ) detected a new transient x - ray source swift j0513.46547 in the large magellanic cloud ( lmc ) ( krimm et al .
2009 ) . an archival analysis of bat data brought out that the source had been active since 2009 march 4 ( mjd 54894 ) and had reached a peak brightness of @xmath5 mcrab on 2009 march 6 .
monitoring observations with _ rxte _
( _ rossi x - ray timing explorer _ ) tracked the decline of the outburst until 2009 may 13 ( mjd 54964 ) .
the detection of strong pulsations with a period around 27.28 s during the outburst and the association of the x - ray transient with the optical source 2mass 051328266547187 , have led swift j0513.46547 to be identified as a high mass x - ray binary system ( hmxb ) ( krimm et al .
2009 , krimm et al .
2013 ) . a recent study ( coe , udalski & finger 2013 , coe et al .
2015 ) on 4.5 years of ogle iv ( optical gravitational lensing experiment phase iv ) light curve revealed regular modulation of the optical magnitude .
approximately 0.05 magnitude flickering in every 27.4 days confirms the binary nature of swift j0513.46547 , assuming that the optical modulation represents the binary period of the system .
coe et al .
( 2015 ) refined the frequency measurements of finger & beklen ( 2009 ) from _ fermi_/gbm ( gamma ray burst monitor ) detections and derived the orbital parameters of the system .
they provided an upper limit of 0.17 to the eccentricity using a model with a fixed orbital period of 27.405 d. coe et al .
( 2015 ) also studied the optical spectrum and identified the companion as a b1ve star .
the position of swift j0513.46547 on the corbet diagram ( pulse period vs. orbital period diagram , corbet 1984 ) affirms the be / x - ray binary nature .
generally x - ray outbursts of be / x - ray binary systems occur in two different forms , classified as type i or type ii ( stella , white & rosner 1986 ) .
type i outbursts ( @xmath6 erg s@xmath2 ) are correlated with the binary orbit so that ; they are recurrently observed in certain orbital phases due to enhanced mass transfer .
in contrast , type ii outbursts ( @xmath7 erg s@xmath2 ) are less frequent , irregular but giant eruptions that decay along several orbital periods .
2009 outburst of swift j0513.4 - 6547 can be identified as type ii since its peak luminosity is derived as @xmath8 erg s@xmath2 ( coe et al .
2015 ) and the outburst lasted for about two orbital cycles .
recently , the source underwent another bursting activity in 2014 . during the _ xmm - newton _ ( _ x - ray multi - mirror mission _ ) detection on 2014 august 25 ( mjd 56894 ) , the luminosity was at the order of @xmath9 erg s@xmath2 ( sturm et al .
seven follow - up _ swift _ observations until the end of the year show that the luminosity peaked twice at the same level , september 19 ( mjd 56919 ) and october 16 ( mjd 56946 ) .
these three outbursts on august 25 ,
september 19 and october 16 can be classified as type i since their occurrence times are just about the expected optical maxima . in this paper , we study the spectral and timing properties of swift j0513.46547 through its observations with _ rxte _ and _ swift _ during the 2009 outburst .
in addition , we analyse the _ swift _ data of 2014 outburst . we introduce the observations in section 2 .
we present our timing and spectral results in sections 3 and 4 . in section 5
, we discuss our results and conclude .
+ 18 target of opportunity ( too ) observations ( see table [ rxteobs ] ) with _ _ rxte__@xmath0pca ( proportional counter array ; jahoda et al .
1996 ) were carried out between 2009 april 14 ( mjd 54935 ) and 2009 may 13 ( mjd 54964 ) .
these observations witness the decline of the 2009 outburst of swift j0513.46547 .
the data are subjected to standard analysis using the tools of ` heasoft v.6.15.1 ` .
time filtering is applied with the criteria on elevation angle to be greater than 10@xmath10 , offset from the source to be less than [email protected] and electron contamination of pcu2 to be less than 0.1 .
we also exclude data within 30 minutes of the peak of south atlantic anomaly ( saa ) in order to increase signal - to - noise ratio ( snr ) . the resulting net exposure after the filtering is 24 ks .
goodxenon events are used to generate 0.1 s binned 310 kev light curves for the timing analysis , whereas standard2f mode data are considered for broad spectral analysis .
furthermore , using ` fasebin ` tool , we construct pulse phase resolved spectra from goodxenon events .
energy - resolved pulse profiles are produced from phase spectra by obtaining the count rates per phase bin with the tool ` fbssum ` .
we select only data from pcu2 during the extraction of spectral files , since it is the only pcu that is active during all of the observations .
moreover , only photons detected on the top anode layers are selected to increase snr .
background subtraction is applied for spectra and light curves by using epoch 5c background model supplied for faint sources ( i.e. @xmath11 40 cts / s / pcu ) . + after the _
_ swift__@xmath0bat discovery of swift j0513.46547 , 3 follow - up _
_ swift__@xmath0xrt ( x - ray telescope ; burrows et al .
2005 ) observations were performed on 2009 april . moreover ; when the source re - brightened on 2014 august , 7 pointing observations were completed until the end of 2014 . a log of _ _
swift__@xmath0xrt observations is given in table [ swiftobs ] .
the total exposure of 2009 and 2014 observations are 12.67 ks and 7.92 ks , respectively .
the observations of swift j0513.46547 are in photon counting ( pc ) mode which has a time resolution of @xmath122.5 s. xrt event files are filtered with the default screening criteria of ` xrtpipeline v.0.13.0 ` script of ` xrtdas v.2.9.2 ` package .
standard grade filtering ( 0 - 12 ) is applied to the data . for each observation , vignetting corrected exposure maps are generated during the pipeline processing . throughout the 2009 observations of swift j0513.46547 count rates
were just around the pile - up limit ( @xmath12 0.5 cts s@xmath2 ) , therefore the events are corrected for pile - up .
for these observations an annular source region is selected in order to exclude an inner circle of radius @xmath127 arcsec , while the outer radius of the annulus is @xmath1270 arcsec .
in contrast , the count rates of 2014 observations were low ( @xmath12 0.1 cts s@xmath2 ) , consequently a circular source extraction region with @xmath1235 arcsec radius is chosen .
the same circular source - free region with @xmath12140 arcsec radius is used for background extraction for both 2009 and 2014 observations .
light curves and spectra are extracted by filtering selected regions using ` xselect v.2.4c ` . for the overall spectral analysis
spectral files are rebinned to have at least 20 counts per energy bin , which is required for @xmath13 fitting .
however , spectra of individual observations are rebinned to have minimum of 1 count per bin and cash statistic ( cash 1979 ) is used as it is advised for low count spectra .
we use the latest response matrix file ( version v014 ) and create individual ancillary response files using the tool ` xrtmkarf v.0.6.0 ` with the exposure maps produced during the pipeline processing .
spectral analysis is performed using ` xspec v.12.8.1 g ` .
for the _ _ rxte__@xmath0pca and _ _
swift__@xmath0xrt observations in 2009 ( see tables [ rxteobs ] and [ swiftobs ] ) , we construct background subtracted lightcurves . the resulting _ rxte _ and _ swift _ lightcurves are 0.1s and 2.51s binned respectively .
we correct the lightcurves with respect to the solar system barycenter .
the lightcurves are also corrected for the orbital binary motion using the orbital parameters given by coe et al .
( 2015 ) .
the nominal pulse period is estimated , by folding _ _ rxte__@xmath0pca barycentric lightcurve on statistically independent trial periods ( leahy et al . 1983 ) .
we search for periodicity using @xmath14 test and use the template pulse profile giving the maximum @xmath14
. then we fold the _ _ rxte__@xmath0pca and _ _ swift__@xmath0xrt lightcurves with the best period obtained from _ _ rxte__@xmath0pca lightcurve .
pulse profiles are obtained with 10 and 20 phase bins from _
_ swift__@xmath0xrt and _ _ rxte__@xmath0pca , respectively .
pulse profiles are represented by their fourier harmonics ( deeter & boynton 1985 ) .
then , we obtain the pulse arrival times for each observation by cross correlating the pulses with template pulse profile .
the phase connected pulse arrival times after the binary orbital correction are shown in the top panel of fig . [ arrival ] , whereas the residuals after the removal of the quadratic trend are shown in the bottom panel of fig .
[ arrival ] . from this quadratic trend ,
spin - up rate is found to be @xmath15hz s@xmath2 .
reduced chi square of this fit is around 1.07 .
this suggest that the orbit of the system is circular which was also suggested by coe et al .
coe et al .
( 2015 ) also found an upper limit of 0.17 to the eccentricity within @xmath16 confidence level . in order to find a further constraint on eccentricity
, an elliptical orbital model is fitted to the residuals as suggested by deeter et al .
this fit gives an upper limit to the eccentricity as 0.028 in @xmath16 level .
it should also be noted that the time coverage of pulse arrival times is only 63 percent of orbital period .
full orbital coverage of arrival times is needed to obtain a better constraint on the eccentricity .
swift j0513.4 - 6547 was also observed seven times in 2014 within a time span of @xmath17 days ( see table [ swiftobs ] ) .
as for the 2009 observations , we construct background subtracted 2.51s binned lightcurves from 2014 observations . we construct pulse phase profiles for five of these observations and connect them in phase .
pulse arrival times are presented in fig .
[ arrival2014 ] . among 2014 observations listed in table
[ swiftobs ] , no significant pulse detection is obtained from the observation at mjd 56939 , and a single pulse arrival time is measured from the observations at mjd 56985 and mjd 56986 .
the pulse frequency measurement of the timing solution is consistent with the value reported by sturm et .
2014 ) within one sigma level .
furthermore a spin down rate of @xmath18 hz s@xmath2 is found during this 70 days time span . timing solutions of 2009 and 2014
are given in table [ timing ] .
+ to see if there are any energy dependent changes in the profile , we construct 37 kev , 710 kev , 1016.5 kev and 16.525 kev pulse profiles .
the energy - resolved profiles given in the left panels of fig .
[ enres ] are profiles averaged for _ _ rxte__@xmath0pca data between mjd 5493554950 .
these energy bands are selected according to their average count rates .
the source is the strongest in 37 kev band with an average count rate of 3.2 cts s@xmath2 .
other bands have approximately the same count rates as 1.7 cts s@xmath2 on average .
energy - resolved power spectra of the same energy bands are also constructed for comparison ( see the right panels of fig .
[ enres ] ) .
the second harmonic ( 0.0734 hz ) of the pulse frequency , that indicates the double - peaked structure , is evident on power spectra . in 37
kev and 710 kev bands , pulse profiles and power spectra are similar . in 1016.5 kev band , along with a significant drop in the pulsed fraction ; the double - peaked structure strengthens . the count rate of the phase between the primary and the secondary peak ( indicated with a dashed line in fig .
[ enres ] ) reaches the same level of the count rate of pulse minimum phase , which is the 0.0 phase .
consequently , the powers of the pulse and its second harmonic are balanced on the 1016.5 kev power spectrum .
the pulse diminishes in the 16.525 kev band , although the average count rate is similar to former bands .
the powers of pulse and its second harmonic are at the level of noise on the 16.5 - 25 kev power spectrum .
we also examine the lightcurves corresponding to specific energy bands and pulse profiles of swift j0513.46547 in the context of hardness ratio .
daily averaged x - ray flux and x - ray luminosity dependences of hr1 ( 710 kev/37 kev ) and hr2 ( 1016.5 kev/710 kev ) are plotted in fig .
[ hard ] .
hardness ratios in pulse phase are plotted in fig .
the top panel is the average pulse profile in the 316.5 kev energy range , which is the combination of pulsed bands in fig .
[ enres ] .
hardness ratios hr1 and hr2 are calculated as un - normalized count rate ratios of consecutive bands in fig .
[ enres ] .
the hr1 reaches its minimum between the double peaks of the pulse .
the secondary peak has the hardest ratio , whereas the primary peak has an average hr1 value .
for hr2 the hardest phase is the 0.0 phase .
a gradual softening is observed until the primary peak , where hr2 starts to increase again .
316.5 kev pulse profiles of individual _ _ rxte__@xmath0pca observations are constructed from the count rates of phase - binned spectra using the tool ` fbssum ` .
then , the pulsed fractions are calculated with the standard definition @xmath19 ; where @xmath20 and @xmath21 are minimum and maximum counts of the pulse . in fig .
[ pf ] , measured pulsed fractions are plotted as a function of 310 kev unabsorbed flux values obtained from the spectral fits . a positive correlation between x - ray flux and pulsed fraction
is evident .
+ @xmath22 @xmath23 radius of the blackbody emitting region calculated from the normalization parameter of ` bbodyrad ` model , assuming a 50.6 kpc distance to the lmc .
@xmath24 unabsorbed flux values calculated using ` cflux ` model for the energy range 0.310 kev , in units of @xmath25erg@xmath26@xmath27s@xmath2 .
the 0.310 kev
_ _ swift__@xmath0xrt spectrum of swift j0513.46547 was previously described by a simple absorbed power law model using the 2009 observations of the source ( krimm et al .
2009 , coe et al .
our reanalysis of the _ swift _ 2009 spectrum alone gives consistent results , however a trial for a simultaneous fit of _ _ swift__@xmath0xrt and _ _ rxte__@xmath0pca data from 2009 showed that this basic model is not enough to fully describe the spectrum of swift j0513.46547 ( @xmath28=1.53 , see table [ spepar ] for best fit parameters ) .
energy ranges selected during the simultaneous fit of 12.7 ks xrt and 17.9 ks pca spectra are 0.39.5 kev and 316.5 kev respectively .
we exclude the energies above 16.5 kev , due to the low statistical significance of the pca spectral bins .
a constant factor is involved in the model , in order to mimic the normalization uncertainty between the two instruments .
we make an effort to improve the simultaneous spectral fit by including extra model components .
we achieve similar improvements by adding a blackbody component ( ` bbodyrad ` ) or a high energy cut - off ( ` highecut ` ) .
the best fit plot is given in fig .
[ simspe ] and the resulting best fit parameters are listed in table [ spepar ]
. both models result in an acceptable @xmath29 ( = 0.96 ) . during 2009 ,
a total of 21 individual observations ( 18 _ rxte _ and 3 _ swift _ ) of swift j0513.46547 cover a time span of about @xmath1230 days .
we investigate the change of spectral parameters throughout this time by fitting the best fit model ( ` phabs*pow*highecut ` ) to the individual spectra of each observation .
the spectra of _ rxte _ observations which are on the same days are summed for a daily spectrum .
snr of the individual spectra are low due to both low exposure times ( @xmath12 1 - 2 ks ) and low count rates , therefore we fix the cut - off model parameters ( i.e. @xmath30 and @xmath31 ) to their best fit values given in table [ spepar ] . moreover , the hydrogen column density ( @xmath32 ) is preset to its best fit value , since we examine that the low @xmath32 along the line of sight does not show any significant change .
therefore we only evaluate the variability of photon index and unabsorbed flux . 3 - 10 kev fluxes
are measured using the ` cflux ` model in ` xspec ` .
daily spectral results for 3 xrt and 14 pca spectra are plotted over observation mid - time in fig .
[ spe2009 ] .
the upper panel shows the decline of the 2009 outburst of swift j0513.46547 . for observations after mjd 54955
, we are only able to measure an upper limit to the flux . the lower panel of fig .
[ spe2009 ] demonstrates the increase in photon index by decreasing flux .
the photon indexes measured from _ swift _ ( filled circles ) imply a slightly softer spectra when compared to values obtained from _ rxte _
( empty circles ) , however this might be an instrumental effect due to the different energy bandpass of detectors . in fig .
[ spe2009 ] , it is observed that as the flux decreases photon index increases which indicates that softer photons emerge from the source as the x - ray flux reduces . the 2014 brightening of swift j0513.46547 is sparsely monitored by seven pointing _ swift _ observations ( each @xmath121 ks ) between mjd 56916 and 56986 .
the overall spectrum produced from 2014 observations provide similar results with _
swift _ 2009 spectrum when described by a simple absorbed power law model ( see table [ spepar ] ) .
the absorption along the line of sight is estimated just the same with 2009 whereas spectral index appear to be a little higher , with a higher uncertainty due to the statistical quality of the low count spectrum .
nevertheless , the higher photon index in 2014 may also arise from the fact that the average x - ray flux of the source is about 20 per cent of the average 2009 flux , since we observe a flux dependence of the photon index for swift j0513.46547 ( see fig .
[ spe2009 ] ) .
0.3 - 10 kev unabsorbed flux measurements of individual _ swift _ observations in 2014 are given in fig .
[ 2014flux ] . in this plot , the _ xmm - newton _
measurement on mjd 56894 is also shown ( sturm et al .
sturm et al . ( 2014 ) suggested that the 2014 brightening of swift j0513.46547 might be explained by recurrent type i outbursts , since the detections with _ xmm - newton _ and _ swift _ are close to the time of expected optical maxima .
therefore , we calculate consequent optical maxima using the ephemeris given in coe et al .
( 2015 ) and indicate them with arrows on fig .
[ 2014flux ] .
it is hard to distinguish whether the brightening is due to a single type ii outburst or a series of type i outbursts , with the very few data points .
one more supporting evidence for the type i scenario is that , we measure another flux maximum corresponding exactly to the next optical maximum on mjd 56946 .
furthermore , using the lmc distance of 50.6 kpc ( bonanos et al .
2011 ) , the peak flux on 2014 correspond to a luminosity of @xmath33 erg s@xmath2 which is about two orders of magnitude lower than the maximum luminosity on 2009 .
the order of the x - ray luminosity also supports the type i scenario for the 2014 brightening of swift j0513.46547 . in order to investigate the variation of spectral parameters with pulse phase ,
we analyze _ _ rxte__@xmath0pca observations between mjd 5493554950 using the tool ` fasebin ` .
timing and orbital parameters for swift j0513.46547 are appended into the timing files ( i.e. ` psrtime.dat ` and ` psrbin.dat ` ) and 16 phase - binned spectra are constructed for the analysis .
the exposure time of each bin turns out to be @xmath121.1 ks .
316.5 kev spectra are modeled with the high energy cut - off model as described in section 4.2 . in fig .
[ phasespe ] , 310 kev unabsorbed flux and photon index values are plotted over pulse phase .
the photon index exceeds its average value 0.93 in some certain phases .
the steepest power law with index 1.48 is acquired in - between the double peaks of the pulse , whereas the peaks have lower indexes such as 0.7 and 0.8 .
the spectral parameters of lowest count rate bins ( i.e. phase 0.0 and 0.0625 ) are not constrained well , an average index is found with a large uncertainty for these phases .
the x - ray luminosity of 2009 outburst was initially at its peak value of @xmath34erg s@xmath2 ( coe et al .
it was found to decrease below @xmath35erg s@xmath2 towards the end of the outburst .
the source disappeared for @xmath36 days and then reappeared in 2014 with @xmath37erg s@xmath2 .
our frequency measurement from 2009 observations which correspond to the final stage of the outburst ( see table [ timing ] ) complies with the general spin up trend of the initial stage of the outburst presented by coe at al .
( 2015 ) . moreover , it is important to note that the arrival times bolster the coe et al .
( 2015 ) assumption that the orbital and optical periods are the same .
the source is found to be spinning up with a rate of @xmath38 hz s@xmath2 during the final stage of the outburst of the 2009 outburst .
comparison with the values found for the initial stages of the 2009 outburst ( @xmath39hz s@xmath2 , coe et al .
2015 ) suggests that the spin rate of the source is correlated with the x - ray luminosity . on the other hand , from table [ timing ]
, it is also found that the long - term ( @xmath40 days long ) average spin - down rate of the source during the quiescence between 2009 and 2014 is @xmath1 hz s@xmath2 which is about two order higher than the rate measured during the re - brightening on 2014 .
this shows that , when the x - ray luminosity of the source increased in 2014 , its spin - down rate is found to be smaller ( i.e. spin rate value becomes greater ) .
thus , whether the source spins up or spins down , its spin rate increases with the x - ray flux .
qualitatively , this trend suggests in general that the source spin rate is correlated with the x - ray luminosity .
observations from future outbursts of the source will be useful to better understand the spin rate @xmath0 x - ray luminosity relation of the source .
it is important to note that the long - term spin - down rate of @xmath1 hz s@xmath2 corresponds to a state where the source is quiescent .
although this spin - down rate is comparable to the long - term spin - down rates of persistent accretion powered x - ray pulsars such as gx 1 + 4 ( gonzales - galan et al . 2012 ) , it is not typical within the quiescent states of the transient sources .
although some transient sources are either found to be spinning up or spinning down in their quiescent states in between outbursts ( e.g. gs 0834 - 430 , a 0535 + 262 , see bildsten et al .
1997 ) , only a few of them such as sxp 7.78 ( klus et al .
2014 ) exhibits long - term ( @xmath41 days long ) quiescent states for which the spin - down rates have the same order of magnitude . if we assume that the neutron star is in propeller state throughout the whole quiescent time interval , magnetic dipole ( @xmath42 where b is the surface dipole magnetic field and r is the radius of the neutron star ) can be estimated ( alpar 2001 ) as @xmath43 where @xmath44 is the moment of inertia , @xmath45 is the spin rate , @xmath46 is the mass and @xmath47 is the frequency of the neutron star . using equation [ eqn1 ] , for a neutron star of mass @xmath48 , moment of inertia of @xmath49 g @xmath50 and a radius of @xmath51 cm , surface dipole magnetic field
is estimated to be @xmath3 gauss which is in good agreement with the magnetic field estimation by coe et al .
( 2015 ) during the accretion phase using spin - up rate @xmath0 luminosity relation .
we confirm that swift j0513.46547 has a circular orbit by refining the upper limit for the orbital eccentricity as 0.028 in @xmath16 confidence level .
although many of the be / x - ray pulsar binary systems ( especially the ones that have transient nature ) have eccentric orbits , some of the transient be / x - ray pulsar systems such as ks 1947 + 300 ( negueruela et al . 2003 ) , 4u 1901 + 03 ( liu et al .
2003 ) , 2s 1553 - 542 ( kelley et al . 1983 ) and rx j0520.5 - 6932 ( kuehnel et al . 2014 )
have orbits with small eccentricities . as seen in fig .
[ enres ] , it is evident that the pulse profile of swift j0513.46547 is double - peaked .
the count rate of one peak significantly exceeds the other , which we might call the primary peak . regardless of the energy band , the peaks have similar phase coverage , where the mid - point between primary and secondary peaks is at @xmath52 phase . as seen in fig .
[ hard ] , hardness ratios of the source correlates with the x - ray luminosity when 3 - 10 kev unabsorbed flux values are less than @xmath53 ergs @xmath54 s@xmath2 , while for higher flux values hardness ratios remain constant . with an assumed source distance of 50.6 kpc
, the flux value separating the correlating hardness values and constant hardness values corresponds to a luminosity of @xmath4 erg s@xmath2 .
postnov et .
2015 ) found similar hardness ratio dependence on x - ray luminosity in six transient x - ray pulsars ( exo 2030 + 375,gx 304 - 1 , 4u 0115 + 63 , v 0332 + 63 , a 0535 + 26 and mxb 0656 - 072 ) .
postnov et al .
( 2015 ) explains this `` saturation '' behaviour of the hardness ratio as a result of @xmath55 surpassing a critical value that leads to an increase in the height of the accretion so that the contribution of the reflected component to the total emission starts decreasing .
but their corresponding critical luminosity is nearly one order higher than our value , being @xmath56 erg s@xmath2 .
maximum value of the pulse fraction obtained from 3 - 16.5 kev _ _ rxte__@xmath0pca observations ( see fig .
[ pf ] ) is @xmath57 which is consistent with the previously reported value ( krimm et al .
2009 ) . in fig .
[ pf ] , the pulsed fraction shows a clear correlation with the source flux , given that higher fractions are observed for higher flux . for higher accretion rates ,
the material pressure of the accreting mass allows the plasma to move in and induces a closer interaction between the plasma and the magnetosphere . consequently , increased efficiency of accretion onto the magnetic poles results in a higher pulsed fraction .
we found that overall _ _ swift__@xmath0xrt and _ _ rxte__@xmath0pca spectrum of swift j0513.46547
is well described with either a power law model with a high energy cut - off at @xmath58 kev or a model consisting of a black body peaking at @xmath59 kev and a power law .
for the overall spectrum , both of these models fit equally well . if the model consisting of a blackbody component is considered
; not only a quite high value of blackbody temperature ( @xmath59 kev ) , but also the unusually high value of blackbody emitting radius ( @xmath60 km ) makes the model not favorable .
usually such high kt s are observed in the spectra of low luminosity ( @xmath61 erg s@xmath2 ) accreting x - ray pulsars , although there might be exceptions such as sax j 2103.5 + 4545 and swift j045106.8 - 694803 with a source luminosity and black body kt of the order of those of swift j0513.46547 ( bartlett et al .
2013 ; inam et al .
moreover the fraction of blackbody flux contributing to the 0.3 - 10 kev flux is rather low as about 30 per cent of the unabsorbed flux . on the other hand ,
if the model including a power law with high energy cut - off is considered , high energy cut - off is a usual spectral feature that is observed in many of the hmxbs .
even if the cut - off energy @xmath62 kev measured for swift j0513.46547 is at the lowest end of the range of energies measured for other sources ( e.g. for 4u 162667 @xmath63 kev , see coburn et al . 2002 ) , we still prefer to represent the individual spectra with this model . from fig .
[ spe2009 ] , there is a marginal evidence that the decrease in x - ray flux leads to a softening of the x - ray spectrum .
similar spectral dependence which might be an indication of accretion geometry changes due to mass accretion rate variations was also found for the x - ray pulsars sax j2103.5 + 4545 ( baykal et al .
2002 ) , 2s 1417 - 62 ( inam et al . 2004b ) , xmmu j054134.7 - 682550 ( inam et al .
2009 ) , her x-1 ( klochkov et al . 2011 ) and swift j1626.6 - 5156 ( icdem , inam and baykal 2011 ) .
we also studied pulse phase resolved spectra of the source using _ _
rxte__@xmath0pca observations of 2009 .
as seen from fig .
[ phasespe ] , the peaks have similar photon index values of @xmath64 .
however , the midpoint between primary and secondary peaks at @xmath52 phase has a steeper ( softer ) spectrum with a power law index of 1.48 .
the softening for this phase is also evident in hr1 hardness ratio shown in fig .
the hardness ratio involving higher energy bands ( i.e. hr2 ) reveals that the spectrum of the primary peak is softer than that of the secondary peak .
this may be due to the fact that the higher energy bands have less count rates , so even modest variations in these energy bands do not affect the power law indices significantly .
we acknowledge support from tbitak , the scientific and technological research council of turkey through the research projects tbag 109t748 and mfag 114f345 .
alpar , m. a. , 2001 , apj , 554 , 1245 bartlett , e. s. ; coe , m. j. ; ho , w. c. g. 2013 , mnras , 436 , 2054 baykal , a. ; stark , m. j. ; swank , j. h. 2002 , apj , 569 , 903 bildsten , l. et al .
1997 , apjs , 113 , 367 bonanos , a.z . ;
castro , n. ; macri , l.m . ; kudritzki , r. 2011 , apjl , 729 , 9 burrows , d. n. ; hil , j. e. ; nouse , j. a. et al .
2005 , space sci .
, 120 , 165 cash , w. 1979 , apj , 228 , 939 coburn , w. ; heindl , w. a. ; rothschild , r. e. et al .
2002 , apj , 580 , 394 coe , m. j. ; finger , m. ; bartlett , e. s. ; udalski , a. 2015 , mnras , 447,1630 coe , m. j. ; udalski , a. ; finger , m. 2013 , atel , 5511 corbet , r. h. d. 1984 , a&a , 141 , 91 deeter , j. e. , boynton , p. e. , pravdo , s. h. , 1981 , apj , 247 , 1003 deeter j. e. , boynton p. e. , 1985 , in hayakawa s. , nagase f. , eds , proc .
inuyama workshop : timing studies of x - ray sources .
nagoya univ . , nagoya , p. 29
finger , m. h. , beklen , e. 2009 , atel 2023 gonzalez - galan , a. , kuulkers , e. , kretschmar , p. , larsson , s. , postnov , k. , kochetkova , a. , finger , m. h. , 2012 , a & a , 537 , 66 idem , b. ; inam , s. . ;
baykal a. , 2011 , mnras , 415 , 1523 inam , s. . ; baykal , a. ; swank , j. ; stark , m. j. 2004a , apj , 616 , 463 inam , s. . ; baykal a. , scott d. m. , finger m. , swank j. , 2004b , mnras , 349 , 173 inam , s. . ; townsend l. j. , mcbride v. a. , baykal a. , coe m. j. , corbet r. h. d. , 2009 , mnras , 395 , 1662 jahoda , k. ; swank , j. h. ; giles , a. b. et al .
1996 , proc .
spie , 2808 , 59 kelley , r. l. ; rappaport , s. ; ayasli , s. 1983 , apj , 274 , 765 klochkov d. , santangelo a. , staubert r. , rothschild r. e. , 2011 , in proc .
8th integral workshop , the restless gamma - ray universe , pos(integral 2010)061 ( arxiv:1105.3547 ) klus , h. ; ho , w. c. g. ; coe , m. j. ; corbet , r. h. d. ; townsend , l. j. 2014 , mnras , 437 , 3863 krimm , h. a. ; corbet , r. h. d. ; evans , p. a. et al .
2009 , atel 2011 krimm , h. a. ; holland , s. t. ; corbet , r. h. d. et al .
2013 , apjs , 209 , 14 kuehnel , m. ; finger , m. h. ; fuerst , f. ; pottschmidt , k. ; haberl , f. ; wilms , j. 2014 , atel , 5856 liu , q. z. ; van paradijs , j. ; & van den heuvel , e. p. j. 2006 , a & a , 455 , 1165 negueruela , i. ; israel g. l. , marco , a. ; norton , a. j. ; speziali , r. 2003 , a & a , 397 , 739 postnov , k.a . ;
gornostaev , m.i . ; klochkov , d. ; laplace , e. ; lukin , v.v . ; shakura , n.i . 2015 , arxiv:1506.07082 [ astro-ph.he ] stella , l. ; white , n. e. ; rosner r. 1986 , apj , 308 , 669 sturm , r. ; carpano , s. ; haberl , f. ; maggi p. ; vasilopoulos , g. 2014 , atel 6483 | we present timing and spectral analysis of _ _ swift__@xmath0xrt and _ _
rxte__@xmath0pca observations of the transient be / x - ray pulsar swift j0513.46547 during its outburst in 2009 and its rebrightening in 2014 . from 2009 observations , short term spin - up rate of the source after the peak of the outburst is found to have about half of the value measured at the peak of the outburst by coe et al . when the source is quiescent between 2009 and 2014 , average spin - down rate of the source
is measured to be @xmath1 hz s@xmath2 indicating a surface dipole magnetic field of @xmath3 gauss assuming a propeller state . from 2014 observations , short term spin - down rate of the source
is measured to be about two orders smaller than this long - term spin - down rate .
the orbit of the source is found to be circular which is atypical for transient be / x - ray binary systems .
hardness ratios of the source correlate with the x - ray luminosity up to @xmath4 erg s@xmath2 in 3 - 10 kev band , whereas for higher luminosities hardness ratios remain constant .
pulsed fractions are found to be correlated with the source flux .
overall _ _ swift__@xmath0xrt and _ _ rxte__@xmath0pca energy spectrum of the source fit equally well to a model consisting of blackbody and power law , and a model consisting of a power law with high energy cut - off . from the pulse phase resolved spectra and pulse phase resolved hardness ratios obtained using _ _
rxte__@xmath0pca , it is shown that spectrum is softer for the phases between the two peaks of the pulse .
[ firstpage ] x - rays : binaries , pulsars : individual : swift j0513.46547 , stars : neutron , accretion , accretion discs |
the explicit expression for the driving - induced term in the power spectrum of fluctuations of the oscillator reads @xmath97 \rangle}+ \phi_f^{(2)}(\omega).\end{aligned}\ ] ] this expression follows from eqs .
( 1 ) and ( 2 ) of the main text .
the first term gives the contribution of the fluctuations of the linear susceptibility .
the second term gives the contribution from the nonlinear susceptibility , @xmath98\nonumber\\ & \times \left[{\langle \chi_2(t , t+\tau , t+\tau ' ) q_0(0 ) \rangle } + { \langle q_0(t)\chi_2(0,\tau,\tau ' ) \rangle}\right].\end{aligned}\ ] ] this term describes the correlation between fluctuations of the second - order susceptibility and thermal fluctuations in the absence of periodic driving .
we emphasize that , for a resonantly modulated underdamped oscillator , it is pronounced at frequencies @xmath32 close to the driving frequency @xmath1 , not @xmath99 .
equation ( [ eq : phi_2 ] ) describes , in particular , the contribution to the spectrum from the nonlinear susceptibility of a nonlinear oscillator .
it is especially convenient in the case of weak nonlinearity , where the oscillator spectrum @xmath57 is broadened primarily by the decay rather than by frequency fluctuations due to the interplay of the nonlinearity and the amplitude fluctuations . in this case
the term @xmath100 gives the main contribution to @xmath43 .
the theory of a nonlinear oscillator will be discussed in a separate publication .
equation ( 3 ) of the main text for the susceptibility of a linear underdamped oscillator with fluctuating frequency can be found in a standard way by changing from the fast oscillating variables @xmath101 to slow complex oscillator amplitude @xmath102\exp(-i\omega_ft)/2 $ ] .
if the equation of motion in the lab frame is markovian , @xmath103q = f\cos\omega_f t + f(t)$ ] , where @xmath39 is the dissipation - related thermal noise , as in the example discussed in the main text , the equation for @xmath104 in the rotating wave approximation reads @xmath105u
-i \frac{f}{4\omega_0 } + f_{u}(t).\ ] ] here , @xmath106 is the detuning of the driving frequency from the oscillator eigenfrequency ; @xmath107\exp(-i\omega_0t)$ ] . equation ( [ eq : dot_u ] ) applies on the time scale that largely exceeds @xmath108 . on this scale @xmath109
is @xmath0-correlated even where in the lab frame the oscillator dynamics is non - markovian , cf @xcite . solving the linear equation ( [ eq : dot_u ] )
, one immediately obtains eq .
( 3 ) of the main text for the oscillator susceptibility @xmath110 .
we disregard corrections @xmath111 ; in particular in eq .
( [ eq : dot_u ] ) for convenience we replaced @xmath112 with @xmath113 ; similarly , in the expression for @xmath114 we replaced @xmath115 with @xmath116 .
we note that the noise @xmath117 drops out from the moments @xmath118 @xcite .
this can be used to characterize the statistics of the frequency noise . in this paper
we consider the change of the conventionally measured characteristic , the power spectrum , and the extra spectral features related to the interplay of the driving and frequency noise .
it is convenient to rewrite eq .
( [ eq : phi_f ] ) for the spectrum @xmath31 near its maximum in the form that explicitly takes into account that , when the expression for the susceptibility is substituted into eq .
( [ eq : phi_f ] ) , the fast - oscillating terms in the integrands can be disregarded .
this gives @xmath119\nonumber\\ & \times \int_{-\infty}^t dt'\int_{-\infty}^0 dt_1'\bigl\langle\chi_{\rm sl}(t ,
t')[\chi_{\rm sl}^*(0,t_1')-{\langle \chi_{\rm sl}^*(0,t_1 ' ) \rangle}]\bigr\rangle,\nonumber\\ & \chi_{\rm sl}(t , t')=e^{-(\gamma - i\delta\omega_f)(t - t ' ) } \exp\left[-i\int\nolimits_{t'}^tdt '' \xi(t'')\right].\end{aligned}\ ] ] here , function @xmath120 gives the slowly varying factor in the fast - oscillating time - dependent oscillator susceptibility @xmath110 .
function @xmath121 gives the standard ( average ) susceptibility @xmath122 the mean forced displacement of the oscillator in the linear response theory is @xmath123 .
averaging over @xmath36 in eqs .
( [ eq : phi_f_convenient ] ) and ( [ eq : lrt ] ) can be done using the noise characteristic functional ( cf .
@xcite ) , @xmath124=\left\langle\exp\left[i\int dt\,k(t)\xi(t)\right ] \right\rangle.\ ] ] as seen from eq .
( [ eq : phi_f_convenient ] ) , function @xmath125 is determined by @xmath126 $ ] with @xmath127 if @xmath128 and @xmath129 otherwise . for @xmath0-correlated noise , where @xmath54=\exp[-\int dt\,\mu(k(t))]$ ] , taking into account that @xmath130 and @xmath131 , we obtain @xmath132 \right)(t - t')\bigr ] , \nonumber\\ & \chi(\omega_f ) = ( i/2\omega_0)\left[\tilde\gamma - i(\omega_f-\tilde\omega_0)\right]^{-1}\end{aligned}\ ] ] with @xmath133 and @xmath134 .
thus , frequency noise leads to the broadening of the conventional susceptibility @xmath135 and the effective shift of the oscillator eigenfrequency by @xmath136 .
we note that the noise can be considered @xmath0-correlated when its spectrum is flat not just on the scale @xmath137 , but on the scale @xmath138 , which itself depends on the noise intensity . at the same time
, the noise spectrum is assumed to be much narrower than @xmath10 . as seen from eq .
( [ eq : phi_f_convenient ] ) the noise components oscillating at frequencies much higher than @xmath139 are averaged out ; frequency noise with frequencies @xmath140 was disregarded in eq .
( [ eq : dot_u ] ) . when writing eq .
( [ eq : dot_u ] ) we also assumed that noise at frequencies close to @xmath141 is very weak and can be disregarded .
if this were not the case , one would have to take into account the effects of nonlinear friction that come from the coupling to the source of the noise , cf .
@xcite . averaging the term @xmath142 in eq .
( [ eq : phi_f_convenient ] ) comes to calculating @xmath143\right\rangle \nonumber\\ & \equiv\left\langle\exp\left[i\int_{-\infty}^{\infty } dt_2k(t_2)\xi(t_2)\right]\right\rangle.\end{aligned}\ ] ] here @xmath144 and @xmath145 . clearly , in this equation @xmath146 .
for @xmath147 we have @xmath148 if @xmath149 and @xmath150 if @xmath151 ; for @xmath152 we have @xmath153 , if @xmath154 and @xmath150 , if @xmath155 ; otherwise @xmath156 .
for a @xmath0-correlated noise the averaging using the explicit form of @xmath54 $ ] and integration over @xmath157 gives eq .
( 5 ) of the main text . for a stationary gaussian noise
the characteristic functional is expressed in terms of the noise correlator @xcite , @xmath124=\exp\left[-\frac{1}{2}\int dt\,dt'\,\langle \xi(t)\xi(t')\rangle k(t)k(t')\right].\ ] ] if the correlator @xmath158 or equivalently , the power spectrum @xmath46 , are known , using the values of @xmath159 given below eq .
( [ eq : k(t)_definition ] ) one can perform the averaging in eq .
( [ eq : phi_f_convenient ] ) and then perform integration over time to find the power spectrum @xmath43 .
this was done to obtain the results shown in fig . 2 of the main text . for slowly varying frequency noise on the scale of the oscillator relaxation time @xmath160 , the evaluation of the susceptibility following the prescription given in the main text leads to expression @xmath161^{-1}.\ ] ]
whereas the expression for the driving - induced term in the power spectrum reads @xmath162\bigr\rangle . \ ] ] these expressions can be used for numerical calculations if the statistics of the noise @xmath36 is known . in the limit of weak slow noise , @xmath163 , eq .
( [ eq : slow_noise ] ) goes over into the result for such noise obtained in the main text ; note that in eq .
( 4 ) of the main text one should replace @xmath164 with @xmath165 in the slow - noise limit , since function @xmath46 is concentrated in the range of small @xmath166 . for the broad - band noise
, on the other hand , the weak - noise limit discussed in the main text corresponds to @xmath167 . in this case
the noise power spectrum is flat and @xmath168 . generally , the weak noise condition used to obtain eq .
( 4 ) of the main text certainly holds for @xmath169 .
it is important that , for slow noise , the condition is less stringent and can be met by increasing the detuning @xmath170 , allowing one to read the slow - noise power spectrum directly off the oscillator power spectrum . both the standard susceptibility @xmath30 and the power spectrum in the absence of driving @xmath57 are affected by frequency noise . in the considered case
they are related by the fluctuation - dissipation relation , @xmath171 . for a non - white frequency noise
the spectrum @xmath57 becomes non - lorentzian .
the explicit expressions for the susceptibility in the limiting cases of fast and slow frequency noise were given above , eqs .
( [ eq : susceptibility_delta_correlated ] ) and ( [ eq : susceptibility_slow ] ) .
a simple explicit expression for @xmath30 follows from eqs .
( [ eq : phi_f_convenient ] ) and ( [ eq : lrt ] ) also in the case of weak noise . here
, the susceptibility becomes @xmath172,\end{aligned}\ ] ] where @xmath46 is the frequency noise power spectrum and @xmath173 .
importantly , the noise - induced correction just slightly distorts the susceptibility .
for example , a sharp low - frequency peak of @xmath46 does not lead to a narrow peak in @xmath30 and , respectively , in the power spectrum @xmath57 .
this should be contrasted with the narrow peak in @xmath31 , which emerges in this case .
the device consists of a carbon nanotube contacted by source and drain electrodes and suspended over a gate electrode .
details of the fabrication and the geometry of the device can be found in ref .
we measure power spectra of displacement fluctuations using the experimental setup sketched in fig .
[ figexp]a .
displacement fluctuations induce conductance fluctuations .
we parametrically down - convert these conductance fluctuations by applying an ac voltage @xmath174 between source and drain at a non - resonant frequency @xmath175 , resulting in current fluctuations at frequencies @xmath176 khz .
+ the spectrum shown in fig .
3a of the main text is obtained in the presence of a near resonant oscillating electrostatic force @xmath177 .
this force is created by applying an oscillating voltage @xmath178 at a frequency @xmath179 hz , with @xmath180 hz , and an amplitude @xmath181 v. in this experiment , a dc gate voltage @xmath182 v and an ac source - drain voltage of amplitude @xmath183 v are used . the amplitude @xmath184 is kept below the threshold beyond which the variance of displacement of the nanotube increases with @xmath184 ( as in ref .
the mode temperature is 1.2 k. the integration time is 32 s. + it is important to verify that applying @xmath185 does not result in an increase in the mode temperature .
we consider the case @xmath186 where an increase of temperature , if any , should be most pronounced .
two mechanisms are liable to increase the mode temperature : ( i ) dissipated power related to the work done by the oscillating resonant force from the gate electrode @xmath177 , and ( ii ) joule heating related to the current , flowing through the nanotube , that is induced by the time - varying capacitance between the nanotube and the gate electrode .
we now discuss the effects of these mechanisms . , which in turn result in fluctuations of the conductance of the nanotube .
applying an oscillating voltage @xmath174 between source s and drain d results in current fluctuations @xmath187 , which are converted into voltage fluctuations across a resistor @xmath188 .
these voltage fluctuations are amplified and cross - correlated , yielding the variance of current fluctuations @xmath189 .
modulation at frequency @xmath190 is obtained by applying an oscillating gate voltage of amplitude @xmath191 .
( b ) inverse of the quality factor as a function of @xmath192 .
parameters used are @xmath182 v , @xmath193 v , and integration time 32 s. the resonant frequency is @xmath180 hz .
( c ) narrow band frequency noise spectrum .
it is obtained by fitting the broad band frequency noise part @xmath194 of the experimental spectrum in fig . 3 to a lorentzian , and then by subtracting this fit from the experimental spectrum .
the red line is a fit to @xmath195 , where @xmath196 . ]
a. from the work of a resonator subject to an oscillating force @xmath197 , the time average power reads : @xmath198 where @xmath199 is the quality factor and @xmath200 is the effective mass of the mode .
the amplitude of the oscillating force is @xmath201 , where @xmath202 is the derivative of the gate capacitance with respect to a small displacement ( we assume that the whole length of the nanotube is at a single , well - defined potential ) . from coulomb blockade measurements
, we estimate that @xmath203 f / m as detailed in ref . @xcite .
we estimate the mass @xmath204 kg from the diameter and the length of the nanotube . in figs .
3b , c of the main text , the maximum amplitude @xmath191 is @xmath205 v. using @xmath206 , @xmath182 v , and @xmath180 hz , we find that the maximum dissipated power is @xmath207 w. this is a minuscule power .
+ using a thermal conductance of @xmath208 w / k , this dissipated power translates into a temperature increase @xmath209 k , a truly insignificant increase .
this thermal conductance is inferred from two published measurements at liquid helium temperature .
the thermal conductance for a multi - wall carbon nanotube with a length of 2.5 @xmath210 m and a diameter of 14 nm was measured to be @xmath211 w / k @xcite .
the thermal conductivity of aligned single - wall nanotubes was measured to be @xmath212 wm@xmath213k@xmath213 @xcite .
these two measurements indicate that the thermal conductance is in the range @xmath214 w / k for a nanotube with a diameter of 1 nm and a length of 2 @xmath210 m .
b. as the nanotube vibrates , the distance that separates it from the gate electrode is modulated , and so is the gate capacitance @xmath215 .
the driving of @xmath215 results in a current at the driving frequency that flows through the nanotube . on resonance
, this current reads : @xmath216 where @xmath217 is the driving amplitude of @xmath215 .
note that @xmath218 also has components proportional to @xmath219 , but these have amplitudes that are several orders of magnitude smaller than @xmath220 .
the time average dissipated power related to joule heating reads @xmath221 where @xmath222 is the resistance of the nanotube .
we estimate @xmath223 f , using the resonant displacement @xmath224 m as an approximation of the motional amplitude .
hence , the dissipated power is @xmath225 w. here again , the induced temperature increase can be neglected . confirming these estimates , fig .
[ figexp]b shows that the inverse of the quality factor @xmath226 does not vary as @xmath227 increases .
since an increase in temperature would result in an increase in @xmath226 , this further indicates that @xmath185 does not affect the mode temperature .
+ the spectral feature at @xmath190 , which we associate to a narrow band frequency noise , is not related to the phase noise of the source used to supply @xmath185 . indeed , the phase noise of our source @xmath228 hz away from @xmath190 is @xmath229 db@xmath230hz , which would result in side bands of amplitude @xmath231 a@xmath232 .
these side bands would then be 4 orders of magnitude smaller than the spectral feature we associate with narrow band frequency noise .
since the narrow - band frequency noise in the nanotube is comparatively weak , one can interprete the results using the weak - noise expression for the spectrum eq . ( 4 ) of the main text
. then the shape of the resonator spectrum gives the shape of the noise power spectrum .
as seen from fig .
[ figexp]c , the spectrum is of @xmath233 type .
our data indicate that @xmath234 is close to 1/2 . shown in fig .
3a of the main text . here
the separation between the light and dark green shaded areas is obtained by approximating the dark green shaded area by a lorentzian of the same shape as a lorentzian that approximates the spectrum without driving ( the blue line in fig .
3a ) . ] to highlight spectral features that we associate to frequency noise ( light and dark green shaded areas in fig .
3a of the main text ) , we exclude the @xmath0-peak at driving frequency @xmath190 . to this end , we observe that the response of our signal analyzer to a voltage oscillating at a given frequency is a delta peak that consists of 3 points above the background .
similarly , the @xmath0-peak at @xmath190 displayed as a black trace in fig .
3a of the main text consists of 3 data points above the background signal .
we remove those 3 points from the measured spectra to estimate the spectral areas plotted in figs .
3b , c. + in fig .
3a of the main text , the separation between the light and dark green shaded areas is subject to some uncertainty because of the noise in the measurement . in fig .
[ fig : exper_lorentz ] we separate these two areas by approximating the broad peak by a lorentzian .
this also leads to a linear dependence of the areas of the peaks on @xmath72 , with the difference in the slopes @xmath73% compared to the results in fig .
3b , c of the main text .
because of the narrow - band noise , the power spectrum without modulation actually differs from a lorentzian , but this is hard to reveal by measuring just this spectrum alone ( or the absorption spectrum ) .
we now consider the area @xmath235 of the driving induced spectral peak for @xmath32 close to @xmath236 ; note that this peak may have several maxima , as seen from fig . 2 of the main text .
we define the area as an integral over positive frequencies , @xmath237 . keeping in mind that @xmath31 is small for large @xmath238 [ in fact , eq .
( [ eq : phi_f_convenient ] ) does not apply for such @xmath32 ] , we obtain @xmath240 this expression describes the dependence of the area of the driving - induced spectrum on the parameters and statistics of the frequency noise . from eq .
( [ eq : area ] ) , the area @xmath235 becomes zero in the absence of frequency noise , since @xmath241 in this case .
the area @xmath235 linearly increases with the frequency noise intensity for weak noise , as seen from eq .
( 4 ) of the main text .
an explicit expression for @xmath235 can be obtained for white frequency noise . from eq .
( 5 ) of the main text , @xmath242 from eq .
( [ eq : area_white ] ) , @xmath235 linearly increases with the characteristic noise strength @xmath243 where it is small , but once the noise becomes strong , @xmath235 decreases with increasing @xmath244 , with @xmath245 for @xmath246 . for weak narrow - band frequency noise , from eq .
( 4 ) of the main text one obtains @xmath235 in terms of the noise variance @xmath247 as @xmath248 ^ 2}.\ ] ] an explicit expression for @xmath235 can be obtained also for a strong gaussian noise .
we will assume that the noise correlator @xmath249 is not fast oscillating and , respectively , the noise spectrum @xmath46 does not have narrow peaks or dips . for the noise variance @xmath247 much larger than @xmath250 , and the squared reciprocal noise correlation time @xmath251 , from eq .
( [ eq : area ] ) @xmath252^{-1/2}.\ ] ] the variation of @xmath235 with the varying frequency - noise intensity and bandwidth is shown in fig .
[ fig : area ] , which refers to the exponentially correlated gaussian noise . as seen from this figure
, @xmath235 displays a maximum as a function of the noise intensity @xmath253 .
the dependence on the noise bandwidth @xmath65 is more complicated ; @xmath235 can have two maxima as a function of @xmath65 for sufficiently strong noise intensity . a convenient scaling factor for the distribution @xmath31 and for the area @xmath235 is provided by the area @xmath254 of the @xmath0-peak in the oscillator power spectrum at the driving frequency . as seen from eq .
( 2 ) of the main text , @xmath255 .
if @xmath256 is known from the measured power spectrum in the absence of driving with the invoked fluctuation - dissipation theorem , scaling by @xmath254 allows one to avoid the actual measurement of the force @xmath80 , which requires knowledge of the coupling to the driving field .
the expression for @xmath254 simplifies if the frequency noise can be thought of as a sum of a weak narrow - band noise @xmath257 and a broad - band ( @xmath0-correlated in slow time ) noise @xmath258 , which is not weak , generally , and is statistically independent from the narrrow - band noise . in this case , combining eqs .
( [ eq : susceptibility_delta_correlated ] ) and ( [ eq : susceptibility_slow ] ) and expanding to the leading order in the weak narrow - band noise , we obtain @xmath259^{-1}\nonumber\\ & \times\left[1-\frac{\tilde\gamma^2 - ( \omega_f-\tilde\omega_0)^2}{\pi[\tilde\gamma^2+(\omega_f-\tilde\omega_0)^2]^2}\int d\omega \,\xi_{\rm nb}(\omega)\right].\end{aligned}\ ] ] here , @xmath260 is the power spectrum of the narrow - band noise ; the variance of the narrow - band noise is @xmath261 .
the correction that contains @xmath262 can be directly read off the area of the narrow peak @xmath263 in the spectrum @xmath31 , which is due to the narrow - band noise . for weak narrow - band noise , this peak
is described by eq .
( 4 ) of the main text if one replaces in this equation @xmath47 with @xmath264 , @xmath10 with @xmath265 , and @xmath52 with @xmath260 .
this can be seen from eq .
( [ eq : phi_f_convenient ] ) . indeed , in the expression for @xmath120 in eq .
( [ eq : phi_f_convenient ] ) one can write @xmath266 . to find @xmath263
, one should integrate over the range of @xmath267 given by the reciprocal bandwidth of the narrow - band noise .
since it largely exceeds @xmath268 , the contributions of the broad - band noise to @xmath120 and @xmath269 are statistically independent .
therefore the averaging over the broad - band noise in these susceptibilities can be done independently . if this averaging is denoted by @xmath270 , @xmath271 for @xmath272 . here
@xmath273(t - t')}.\nonumber\ ] ] since function @xmath260 quickly falls off with increasing @xmath274 , in the denominator of eq .
( 4 ) of the main text one can replace @xmath32 with @xmath1 .
one then sees that , to the leading order in the narrow - band noise strength , the ratio of the area @xmath275 of the narrow peak @xmath263 to the area of the @xmath0-peak in the spectrum is @xmath276 equations ( [ eq : area_delta_peak ] ) and ( [ eq : narrow_band_area ] ) can be used to scale the area of the broader peak of @xmath31 by @xmath254 with account taken of the effect of the narrow - band frequency noise .
along with the onset of a narrow peak in @xmath43 , this noise leads to the change of the shape and area of the broad peak . where the narrow - band noise is weak , the leading - order correction can be found by replacing @xmath277 with @xmath278 in eq .
( [ eq : phi_f_convenient ] ) for @xmath120 and then expanding to second order in @xmath257 .
the result is particularly simple in the considered here case where the broad - band frequency noise is @xmath0-correlated in slow time and the broad peak of @xmath31 is described by eq .
( 5 ) of the main text .
one just has to replace in this equations @xmath265 with @xmath279 , expand in @xmath257 to the second order , and average @xmath280 .
the corresponding expression for the area @xmath281 of the broad peak of @xmath31 reads @xmath282 ^ 2}\nonumber \\ & \times \left[1-\frac{\tilde\gamma^2 - 3(\omega_f-\tilde\omega_0)^2)}{2\pi[\tilde\gamma^2+(\omega_f-\tilde\omega_0)^2]^2}\int d\omega\ , \xi_{nb}(\omega ) \right].\end{aligned}\ ] ] equations ( [ eq : narrow_band_area ] ) and ( [ eq : broad_band_corrected ] ) lead to eq .
( 6 ) of the main text , which shows the contribution of the frequency noise to the width of the broad peak of the spectrum .
the results of the simulations presented in figs . 2 and 4 of the main text were obtained in a standard way .
we integrated the stochastic differential equation ( [ eq : dot_u ] ) using the heun scheme @xcite .
for a nonlinear oscillator this equation has the extra term @xmath283 in the right - hand side @xcite . for a nonlinear oscillator
, we verified that the values of the modulating field amplitude @xmath80 were in the range where the driving - induced term in the power spectrum was quadratic in @xmath80 .
as seen from the inset in fig . 4 of the main text , the simulations are in excellent agreement with analytical results @xcite in the absence of driving .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _
( , ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , ( , ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1021/nl403936a [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) , ed . ,
@noop _ _ ( , ) in @noop _ _ , vol . , ( , ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/nature08967 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1103/physrevlett.87.215502 [ * * , ( ) ] link:\doibase 10.1063/1.127079 [ * * , ( ) ] @noop * * , ( ) | we study the spectral effect of the fluctuations of the vibration frequency .
such fluctuations play a major role in nanomechanical and other mesoscopic vibrational systems . we find that , for periodically driven systems , the interplay of the driving and frequency fluctuations results in specific spectral features .
we present measurements on a carbon nanotube resonator and show that our theory allows not only the characterization of the frequency fluctuations but also the quantification of the decay rate without ring - down measurements .
the results bear on identifying the decoherence of mesoscopic oscillators and on the general problem of resonance fluorescence and light scattering by oscillators .
the spectrum of response and the power spectrum of an oscillator is a textbook problem that goes back to lorentz and einstein @xcite .
it has attracted much attention recently in the context of nanomechanical systems . here ,
the spectra are a major source of information about the classical and quantum dynamics @xcite .
this is the case also for mesoscopic oscillators of different nature , such as superconducting cavity modes @xcite and optomechanical systems @xcite .
mesoscopic oscillators experience comparatively large fluctuations . along with dissipation
, these fluctuations determine the shape of the vibrational spectra .
a well - understood and most frequently considered @xcite source of fluctuations is thermal noise that comes from the coupling of an oscillator ( vibrational system ) to a thermal reservoir and is related to dissipation by the fluctuation - dissipation theorem .
dissipation leads to the broadening of the oscillator power spectrum and the spectrum of the response to external driving .
spectral broadening can also come from fluctuations of the oscillator frequency , which play an important role in mesoscopic oscillators . for nanomechanical resonators , frequency fluctuations can be caused by tension and mass fluctuations , fluctuations of the charge in the substrate , or dispersive intermode coupling @xcite , whereas for electromagnetic cavity modes they can come from fluctuations of the effective dielectric constant @xcite .
identifying different broadening mechanisms is a delicate task that has been attracting much attention @xcite . in this paper
we study the combined effect of periodic driving and frequency fluctuations on the power spectra of nanomechanical vibrational systems . for a linear oscillator with no frequency fluctuations ,
driving leads to a @xmath0-like peak at the driving frequency @xmath1 @xcite , because here the only effect of the driving is forced vibrations linearly superimposed on thermal motion .
frequency fluctuations make forced vibrations random .
as we show , this qualitatively changes the spectrum leading to characteristic new spectral features .
we observe these features in a carbon - nanotube resonator and use them to separate the energy relaxation rate from the overall broadening of the power spectrum in the absence of driving , as well as reveal and explore the narrow - band frequency noise .
we predict that for fluctuating nonlinear vibrational systems , too , even weak driving leads to a very specific extra spectral structure . .
panels ( a ) and ( b ) refer to large and small correlation time of the frequency noise @xmath2 compared to the oscillator relaxation time @xmath3 , respectively , i.e. , to narrow- and broad - band frequency noise .
the blue ( lower ) line shows the spectrum of thermal fluctuations in the absence of driving ; it is centered at the oscillator eigenfrequency @xmath4 . in the presence of driving there
is added a @xmath0-peak at the driving frequency @xmath1 .
the green areas show the spectral features from the interplay of the driving and fluctuations of @xmath5 .
bottom panels : @xmath5 for @xmath6 ( a ) and @xmath7 ( b ) . ] for a linear oscillator , the spectral features resulting from the interplay of driving and frequency noise are sketched in fig . [
fig : sketch ] .
the two limiting cases shown in fig .
[ fig : sketch ] correspond to the long and short correlation time of the frequency noise @xmath2 compared to the oscillator relaxation ( decay ) time @xmath3 . for @xmath6 ( panel a ) the oscillator frequency @xmath5 slowly fluctuates about what can be called the eigenfrequency @xmath4 .
one can then think of slow fluctuations of the oscillator susceptibility @xmath8 , which depends on the detuning of the driving frequency @xmath1 from @xmath5 .
the associated slow fluctuations of the amplitude and phase of forced vibrations at frequency @xmath1 lead to a finite - width spectral peak centered at @xmath1 .
this is a frequency - domain analog of the einstein light scattering due to spatial susceptibility fluctuations @xcite . for @xmath9 ( panel b ) ,
driving - induced random vibrations quickly lose the memory of the driving frequency .
they become similar to thermal vibrations .
however , their amplitude is determined by the driving , not the temperature .
this leads to a spectral peak centered at the oscillator eigenfrequency @xmath10 , with the height quadratic in the driving amplitude . in the quantum picture
, one can think that , as a result of pumping by a driving field , the oscillator emits energy quanta .
for the familiar example of an oscillating charge driven by an electromagnetic field these quanta can be photons , and one can speak of light scattering and fluorescence by an oscillator .
a quantum is emitted over time @xmath3 after the absorption event .
for @xmath7 the frequency of the quantum is uncorrelated with the excitation frequency @xmath1 .
this is a fluorescence - type process .
the energy difference @xmath11 comes from the frequency noise . for @xmath6 emission
occurs at frequencies close to @xmath1 . in the both cases the spectrum is qualitatively different from just a @xmath0-like peak in the absence of frequency fluctuations @xcite . to describe the power spectrum of a nonlinear system one has to go beyond the approximation implied above , where only the linear susceptibility is fluctuating . if driving is described by the term @xmath12 in the oscillator hamiltonian , where @xmath13 is the oscillator coordinate and @xmath14 is the driving force , to obtain terms @xmath15 in the power spectrum one should keep terms @xmath16 and @xmath15 in the response , @xmath17 here @xmath18 is thermal displacement in the absence of driving . equation ( [ eq : causality ] ) does not include averaging , @xmath19 and @xmath20 are the fluctuating linear and nonlinear susceptibilities .
the standard linear susceptibility is @xmath21 , it is a function of @xmath22 . for a harmonic oscillator , which is the central topic of this paper , @xmath23 .
the conventionally measured oscillator power spectrum is @xmath24 , where @xmath25 indicates statistical averaging and averaging with respect to @xmath26 over the driving period @xmath27 . for weak driving @xmath28 this spectrum
is sketched in fig .
[ fig : sketch ] .
function @xmath29 is the power spectrum in the absence of driving , a resonant peak associated with thermal vibrations of the oscillator .
the @xmath0-peak at the driving frequency in eq .
( [ eq : correlator_defined ] ) and in fig .
[ fig : sketch ] describes average forced oscillator vibrations , @xmath30 is the fourier transform of @xmath21 over @xmath22 . of primary interest to us
is the term @xmath31 , shown by the envelope of the green area in fig . [
fig : sketch ] .
it describes the interplay of frequency fluctuations and the driving .
we consider it for @xmath32 close to @xmath1 assuming a high quality factor , @xmath33 , typical for mesoscopic systems , and resonant driving , @xmath34 . for a harmonic oscillator with fluctuating frequency , @xmath35 , where @xmath36 is zero - mean noise .
we assume that the noise is weak compared to @xmath10 and that its correlation time @xmath37 .
the noise then does not cause parametric excitation of the oscillator @xcite .
the most simple model of the oscillator dynamics is described by equation @xmath38q = f\cos\omega_f t + f(t)$ ] , where @xmath39 is thermal noise and @xmath40 is the relaxation rate .
both @xmath39 and the direct frequency noise @xmath36 lead to fluctuations of the oscillator phase . separating their contributions by measuring the commonly used allan variance ( cf .
@xcite ) is complicated .
however , these two types of noise have different physical origin , and our results show how they can be separated using the power spectrum ; a different approach , which however may not be implemented with a standard spectrum analyzer , was proposed in @xcite . in the standard rotating wave approximation ,
see appendix , the fluctuating linear susceptibility of a damped harmonic oscillator is , @xmath41 ( @xmath42 ) .
equation ( [ eq : harmonic_chi_1 ] ) often applies even where the oscillator dynamics in the lab frame is non - markovian .
explicit expressions for @xmath19 and @xmath31 can be obtained from eq .
( [ eq : harmonic_chi_1 ] ) in the limiting cases . for weak frequency noise
, one can expand @xmath19 in @xmath36 . to the leading order ,
the spectrum @xmath43 is proportional to the noise power spectrum @xmath44 , @xmath45}\frac{\xi(\omega - \omega_f)}{\gamma^2+(\omega-\omega_0)^2}.\end{aligned}\ ] ] this expression provides a direct means for measuring the frequency noise spectrum . it already shows the peculiar features qualitatively discussed above . if @xmath46 peaks at zero frequency and is narrow on the scale @xmath47 ( as for @xmath48-type noise , for example ) , @xmath31 has a peak at @xmath1 , cf .
fig .
[ fig : sketch]a .
the shape of this peak coincides with that of @xmath46 .
if , on the other hand , @xmath46 is almost flat on the frequency scale @xmath49 ( broad - band noise ) , @xmath31 has a lorentzian peak at @xmath10 , cf .
fig .
[ fig : sketch]b . to describe the effect of a narrow - band , but not necessarily weak frequency noise
, one can replace @xmath50 in eq .
( [ eq : harmonic_chi_1 ] ) with @xmath36 .
this corresponds to the instantaneous " slowly fluctuating susceptibility @xmath51 $ ] . the resulting narrow spectrum @xmath31 is determined by the spectrum and statistics of the frequency noise .
the simple relation ( [ eq : weak_noise ] ) between @xmath31 and @xmath52 follows from this analysis for @xmath53 .
importantly , this condition can be achieved by tuning @xmath1 somewhat away from @xmath10 .
the case of flat @xmath46 , i.e. , of @xmath36 being @xmath0-correlated on time scale @xmath3 , can be analyzed for an arbitrary noise strength using that the characteristic functional of a @xmath0-correlated noise is @xmath54=\langle\exp[i\int dt k(t)\xi(t)]\rangle = \exp[-\int \mu(k(t ) ) dt]$ ] , where function @xmath55 is determined by the noise statistics .
then from eq .
( [ eq : harmonic_chi_1 ] ) @xmath56/\gamma}{8\omega_0 ^ 2[\tilde\gamma ^2 + ( \omega_f -\tilde\omega_0)^2 ] } \frac{\tilde\gamma}{\tilde\gamma^2 + ( \omega-\tilde\omega_0)^2}.\end{aligned}\ ] ] the spectrum ( [ eq : white_noise ] ) and the spectrum @xmath57 in the absence of periodic driving have the same shape given by the last factor in ( [ eq : white_noise ] ) : a lorentzian centered at the noise - renormalized oscillator eigenfrequency @xmath58 with halfwidth @xmath59 .
however , in contrast to @xmath57 , the area of @xmath60 is independent of the intensity ( @xmath61 ) of the dissipation - related noise .
instead it is proportional to the frequency - noise characteristic @xmath62 .
equation ( [ eq : white_noise ] ) suggests how to separate the noise - induced broadening of the oscillator spectrum from the decay - induced broadening , see below . .
the noise intensity is @xmath63 .
panels a and b : the full spectrum .
the color coding is the same as in fig .
[ fig : sketch ] , @xmath64 .
panel c : the driving - induced term .
the solid lines and dots show the analytic theory and simulations ; the consecutive curves are shifted by 0.25 along the ordinate . ]
function @xmath43 can be found in a closed form for a gaussian noise @xmath36 .
the results are shown in fig .
[ fig : theory_1 ] for the noise power spectrum with bandwidth @xmath65 , @xmath66 .
they illustrate how the shape of @xmath31 changes from a peak at @xmath1 for a narrow - band noise ( @xmath67 ) to a peak at @xmath10 for a broadband noise ( @xmath68 ) .
the overall area of the spectrum @xmath43 nonmonotonically depends on the frequency noise intensity : it is linear in the noise intensity for weak noise , cf .
eq .
( [ eq : weak_noise ] ) , but for a large noise intensity it decreases , since the decoherence rate of the oscillator increases . to corroborate the theory , we measured the spectrum of a modulated carbon nanotube resonator at @xmath69 1.2k . for such @xmath70 and weak driving , low - lying flexural modes of the resonator are well described by harmonic oscillators .
the driving was applied as an ac voltage @xmath71 on the gate electrode , and the power spectrum of the mechanical vibrations was probed by measuring the noise in the current flowing through the nanotube @xcite .
figure 3a compares the power spectra of the nanotube vibrations obtained with and without the oscillating force .
the spectrum without modulation ( blue trace ) is close to a lorentzian , as expected for thermal vibrations .
the spectrum with modulation ( green trace ) displays a narrow peak centered at the modulation frequency and a much broader peak of the same shape as the spectrum without modulation .
the areas of the both modulation - induced parts of the spectrum scale as @xmath72 ( fig .
[ fig : exper]b , c ) , in agreement with eq .
( 2 ) .
the separation between the parts is subject to some uncertainty because of the measurement noise in fig .
[ fig : exper]a , see appendix . the resulting uncertainty in the slopes in fig . [ fig : exper]b , c is @xmath73% .
the change of the spectrum is not a heating effect associated with the modulation , since we estimate temperature increases as @xmath74 k. the spectral feature at @xmath1 is not related to the phase noise of the source used for the modulation ; indeed , the phase noise of our source @xmath75 hz away from @xmath1 could only account for @xmath76 % of the measured power spectrum . through a driven carbon nanotube .
the measurement bandwidth is 4.7 hz . the eigenfrequency of the studied flexural mode is 6.3 mhz .
the driving frequency is 100 hz below the resonance frequency .
the blue line refers to the power spectrum without driving ; the green area shows the driving - induced spectral change .
this change is separated into the broad peak ( darker green ) , narrow peak ( lighter green ) , and a delta - spike at the modulation frequency .
this spike lies within 3 bins , within our experimental resolution , and is represented by the black vertical lines .
the separation of the broad and narrow peaks is done by the straight line that interpolates the broad peak .
shown in the lower panels is the dependence of the lighter green area ( b ) , the darker green area ( c ) , and the area under the @xmath0-peak ( d ) on the squared amplitude of the modulating gate voltage ; as expected from the theory , it is close to linear . ]
the driving - induced spectral change provides a simple means for estimating the intrinsic relaxation rate of the resonator @xmath47 from our experimental data .
this can be done using the areas @xmath77 and @xmath78 of the narrow and broad peaks in fig .
[ fig : exper ] , respectively . comparing them to the area under the driving - induced @xmath0-peak @xmath79
, we eliminate @xmath80 and obtain from eq .
( [ eq : white_noise ] ) @xmath81;\ ] ] we used @xmath82 . with @xmath83 hz read out from the collected spectra ( such as the one in fig .
3a ) , along with @xmath84 and @xmath85 measured from figs .
3b - d , we obtain @xmath86 , which gives @xmath87 hz . therefore , the broad - band fluctuations of the resonant frequency account for @xmath88% of the measured mechanical linewidth . because of the noise in the measurement in fig .
3a , the uncertainty in @xmath47 is @xmath73% .
the narrow - band frequency noise can also be characterized from the measurements .
its power spectrum is @xmath89 with @xmath90 , see appendix . obtaining the power spectra @xmath43 for several values of @xmath91 should allow separating the low - frequency part of the frequency noise spectrum @xmath52 even where it is not weaker than the broad - band part , and reading it directly off the data on the power spectrum using eq .
( [ eq : weak_noise ] ) .
an important contribution to spectral broadening of mesoscopic oscillators can come from their nonlinearity @xcite . because the vibration frequency of a nonlinear oscillator depends on the vibration amplitude , thermal fluctuations of the amplitude lead to frequency fluctuations .
this makes the power spectrum non - lorentzian and asymmetric even in the absence of driving @xcite .
the shape of the spectrum is determined by the interrelation between the frequency uncertainty @xmath47 due to the oscillator decay and the width @xmath92 of the frequency distribution due to thermal distribution of the vibration amplitude . in the presence of driving , superimposed on this effect
is the frequency shift due to the driving - induced vibrations .
zhangy_fig4_sept.eps the strong dependence of the driving - induced spectral change on the ratio @xmath93 is illustrated in fig .
[ fig : nonlinear_osc ] for the important model where the nonlinear term in the oscillator energy is @xmath94 , so that @xmath95 . for small @xmath93 the major effect of the driving is the shift of the spectrum ; then @xmath96 has a characteristic dispersive shape , changing sign at @xmath10 . with increasing @xmath93 the shape of @xmath31
becomes more complicated .
generally it still has positive and negative parts , in dramatic difference from the case of a harmonic oscillator with fluctuating frequency . also , in contrast to a harmonic oscillator , keeping terms @xmath15 in the power spectrum of a nonlinear oscillator is justified only for weak modulating fields . a detailed theory of @xmath43 for nonlinear oscillators based on eq .
( [ eq : causality ] ) will be presented elsewhere .
the above results show that the interplay of driving and frequency noise qualitatively changes oscillator spectra compared to the spectra with no frequency fluctuations @xcite .
the change sensitively depends on the frequency noise intensity and power spectrum .
the possibility to separate contributions from different parts of the frequency - noise spectrum is of significant interest , as they may come from physically different sources , like two - level fluctuators and dispersive coupling to other modes , to mention but a few .
the results suggest a way of discriminating between three major factors of the broadening of the oscillator spectra : decay ( energy relaxation ) , frequency fluctuations induced directly by the noise that modulates the eigenfrequency , and frequency fluctuations due to the oscillator nonlinearity . for linear oscillators ,
our simple procedure yields the decay rate without the need of an actual ring - down measurement that is often difficult to implement , in particular for nanotube mechanical resonators .
we also find that the nonlinearity - induced spectral change has a qualitatively different shape from that due to frequency noise in linear oscillators .
the analysis of driven linear oscillators with a fluctuating frequency immediately extends to the quantum regime , which is attracting much interest in nano- and optomechanics @xcite . for nonlinear oscillators
, the nonequidistance of the energy levels can bring in new features compared to the classical limit .
this research was supported in part by the us army research office ( w911nf-12 - 1 - 0235 ) , us defense advanced research agency ( fa8650 - 13 - 1 - 7301 ) , the erc - carbonnems project ( 279278 ) , the grant mat2012 - 31338 from the spanish state , and the grant agaur - sgr of the catalan government . |
coloring problems are among the most studied problems in algorithmic graph theory . in its classical form
, the @xmath1-coloring problem asks if there is an assignment of @xmath1 colors to the vertices of a graph in such a way that no edge is monochromatic .
many generalizations and variations of the classical coloring problem have been defined over the years .
one of such variations is the clique @xmath1-coloring problem , in which the vertices are colored so that no maximal clique is monochromatic . in this article
we study the star and biclique coloring problems , which are variations of the coloring problem similar to clique colorings .
a biclique is a set of vertices that induce a complete bipartite graph @xmath9 , while a star is a biclique inducing the graph @xmath10 . in the star ( biclique )
@xmath1-coloring problem , the goal is to color the vertices with @xmath1 colors without generating monochromatic maximal stars ( bicliques ) . the clique coloring problem has been investigated for a long time , and it is still receiving a lot of attention .
recently , the clique @xmath1-coloring problem was proved to be @xmath11-complete @xcite for every @xmath12 , and it remains @xmath11-complete for @xmath13 even when the input is restricted to graphs with no odd holes @xcite . the problem has been studied on many other classes of input graphs , for which it is was proved to be -complete or to require polynomial time ( e.g. @xcite ) . due to the close relation between cliques and bicliques , many problems on cliques have been translated in terms of bicliques ( e.g. @xcite ) . however , there are some classical problems on cliques whose biclique versions were not studied until recently @xcite .
clique colorings are examples of such problems ; research on biclique colorings begun in 2010 in the master thesis of one of the authors @xcite whose unpublished results are being extended in the present article .
it is worth mentioning that , despite its youthfulness , at least two articles on biclique colorings were written : @xcite develop a polynomial time algorithm for biclique coloring some unichord - free graphs , and @xcite determines the minimum number of colors required by biclique colorings of powers of paths and cycles .
the list coloring problem is a generalization of the coloring problem in which every vertex @xmath14 is associated with a list @xmath15 , and the goal is to color each vertex @xmath14 with an element of @xmath15 in such a way that no edge is monochromatic .
function @xmath2 is called a list assignment , and it is a @xmath1-list assignment when @xmath16 for every vertex @xmath14 .
a graph @xmath0 is said to be @xmath1-choosable when it admits an @xmath2-coloring with no monochromatic edges , for every @xmath1-list assignment @xmath2 .
the choosability problem asks whether a graph is @xmath1-choosable . in the same way as
the coloring problem is generalized to the clique ( star , biclique ) coloring problem , the choosability problem is generalized to the clique ( star , biclique ) choosability problem .
that is , a graph @xmath0 is clique ( star , biclique ) @xmath1-choosable when it admits an @xmath2-coloring generating no monochromatic maximal cliques ( star , bicliques ) , for every @xmath1-list assignment @xmath2 .
the choosability problems seem harder than their coloring versions , because a universal quantifier on the list must be checked .
this difficulty is reflected for the @xmath1-choosability and clique @xmath1-choosability problems in the facts that the former is @xmath17-complete for every @xmath18 @xcite , whereas the latter is @xmath19-complete for every @xmath12 @xcite . in
@xcite it is proven that every planar graph is clique @xmath20-choosable . however , contrary to what happens with the clique coloring problem , there are not so many results regarding the complexity of the clique coloring problem for restricted classes of graphs .
in this paper we consider the star and biclique coloring and choosability problems , both for general graphs and for some restricted classes of graphs .
the star and biclique coloring and choosability problems are defined in section [ sec : preliminaries ] , where we introduce the terminology that will be used throughout the article . in section
[ sec : general case ] , we prove that the star @xmath1-coloring problem is @xmath11-complete for @xmath12 , and that it remains @xmath11-complete even when its input is restricted to @xmath21-free graphs . clearly , every maximal biclique of a @xmath22-free graph is a star .
thus , we obtain as a corollary that the biclique @xmath1-coloring problem on @xmath21-free graphs is @xmath11-complete as well .
the completeness proof follows some of the ideas by marx @xcite for the clique coloring problem . in section [ sec : choosability ] we show that the star @xmath1-choosability problem is @xmath19-complete for @xmath12 , and that it remains @xmath19-complete for @xmath21-free graphs .
again , the @xmath19-completeness of the biclique @xmath1-coloring problem on @xmath21-free is obtained as a corollary . as in @xcite , we require a structure to force a color on a vertex .
the remaining sections of the article study the star and biclique coloring problems on graphs with restricted inputs .
these graphs are related to the graph @xmath0 that is generated to prove the @xmath11-completeness of the star coloring problem in section [ sec : general case ] .
the aim is to understand what structural properties can help make the problem simpler . in section [ sec : small forbiddens ]
, we discuss the star and biclique coloring and choosability problems on @xmath23-free , @xmath24-free , @xmath25-free , and @xmath26-free graphs . for @xmath23 , @xmath24 and @xmath25 ,
the star coloring and star choosability problems are almost trivial and can be solved in linear time . on the other hand ,
both problems are as hard as they can be for @xmath26-free graphs , even when the input is a co - bipartite graph . in section [ sec : diamond - free ] we prove that the star coloring problem is -complete for diamond - free graphs and that the star choosability problem is @xmath17-complete for a superclass of diamond - free graphs . if no induced @xmath27 is allowed for a fixed @xmath28 , then the biclique coloring and the biclique choosability problems are also -complete and @xmath17-complete . in section [ sec : split ] , the star coloring and the star choosability problems on split graphs
are proved to be -complete and @xmath17-complete , respectively .
finally , sections [ sec : threshold ] and [ sec : block ] show that the star coloring and the star choosability problems are equivalent for both threshold and net - free block graphs , and both can be solved in linear time .
table [ tab : results ] sums up the results obtained in the article for the star coloring and star choosability problems .
.complexity results obtained in this article . [ cols="<,^,^",options="header " , ] besides the @xmath29-keepers , list switchers , and clusters , we use forth kind of connection that can be used to force the color of a given vertex when an appropriate list assignment is chosen .
this connection is called the _ forcer _ and , contrary to the other connections , it connects only one vertex .
[ def : k - forcer connection ] let @xmath0 be a graph and @xmath30 .
say that @xmath31 is a _
@xmath1-forcer connecting @xmath14 _
( @xmath12 ) when @xmath32 can be partitioned into sets of leafed vertices @xmath33 and @xmath34 , and cliques @xmath35 for @xmath36 and @xmath37 in such a way that @xmath38 , @xmath39 , @xmath35 is a @xmath1-switcher connecting @xmath40 , and there are no more edges incident to vertices in @xmath41 .
let @xmath2 be a @xmath1-list assignment of @xmath0 and @xmath32 be a @xmath1-forcer connecting @xmath30 .
we say that @xmath42 is _
@xmath2-admissible for @xmath14 _ when there is an @xmath2-coloring @xmath43 of @xmath0 in which @xmath44 and no monochromatic maximal star has its center in @xmath32 .
clearly , if @xmath45 is @xmath2-admissible for @xmath14 , then any @xmath2-coloring @xmath43 of @xmath46 in which @xmath44 can be extended into an @xmath2-coloring of @xmath0 in such a way that no monochromatic maximal star has its center in @xmath32 .
the main properties of forcers are summed up in the following lemma ( see figure [ fig : switcher - forcer ] ( b ) ) .
[ lem : forcer properties ] let @xmath0 be a graph and @xmath32 be a @xmath1-forcer connecting @xmath30 ( @xmath12 ) . then
, a. no induced @xmath7 or @xmath6 of @xmath0 contains a vertex in @xmath32,[lem : forcer properties : forbidden ] b. for every @xmath1-list assignment @xmath2 of @xmath0 there is an @xmath2-admissible color for @xmath14 , and[lem : forcer properties : extension ] c. every @xmath1-list assignment @xmath2 of @xmath46 can be extended into a @xmath1-list assignment of @xmath0 in which @xmath14 has a unique @xmath2-admissible color.[lem : forcer properties : color ] let @xmath33 , @xmath34 , and @xmath35 be as in definition [ def : k - forcer connection ] , and define @xmath47 .
statement ( [ lem : forcer properties : forbidden ] ) follows from the fact that no pair of vertices in @xmath48 have a common neighbor .
( [ lem : forcer properties : extension ] ) let @xmath8 be the complete bipartite graph with bipartition @xmath49 , and @xmath50 be a @xmath1-list assignment of @xmath8 where @xmath51 for every @xmath52 . in @xcite it is proven that @xmath8 admits a vertex @xmath50-coloring @xmath53 .
define @xmath43 as any @xmath2-coloring of @xmath0 in which @xmath54 for @xmath55 , where @xmath1-switchers and leaves are colored according to lemma [ lem : switcher choosability ] and observation [ obs : leafed vertex ] , respectively .
the coloring of the @xmath1-switchers is possible because @xmath53 is a vertex @xmath50-coloring . by observation [ obs : leafed vertex ]
, no vertex in @xmath41 is the center of a maximal star , while by lemma [ lem : switcher choosability ] , no vertex in @xmath35 is the center of a maximal monochromatic star for @xmath56 and @xmath37 .
that is , @xmath57 is @xmath2-admissible for @xmath14 .
( [ lem : forcer properties : color ] ) extend @xmath2 into a @xmath1-list assignment of @xmath0 such that 1 .
@xmath58 for every pair of vertices @xmath59 , 2 .
@xmath60 is a family of different subsets included in @xmath61 such that @xmath62 for every @xmath56 and @xmath37 , and 3 .
@xmath63 for every @xmath56 , and @xmath37 .
define @xmath8 and @xmath50 as in statement ( [ lem : forcer properties : extension ] ) . by statement ( [ lem : forcer properties : extension ] )
, there is an @xmath2-coloring @xmath43 of @xmath0 that contains no monochromatic maximal star with center in @xmath32 .
since @xmath35 is a block of @xmath0 , it follows that @xmath64 by observation [ obs : block coloring ] .
then , since no maximal star with center in @xmath35 is monochromatic , it follows that @xmath65 for every @xmath56 , @xmath37 .
thus , if @xmath53 is the coloring such that @xmath66 for every @xmath52 , then @xmath53 is a vertex @xmath50-coloring of @xmath8 .
consequently , as proven in @xcite , @xmath67 is the unique color of @xmath15 that belongs to the subset of @xmath68 .
a problem @xmath69 is @xmath19when the problem of authenticating a negative certificate of @xmath69 is @xmath11 @xcite . for the star @xmath1-choosability problem , a @xmath1-list assignment of @xmath0
can be taken as the negative certificate .
using arguments similar to those in section [ sec : general case:2-stp ] for star @xmath1-colorings , it is not hard to see that it is a @xmath11problem to authenticate whether a graph @xmath0 admits no @xmath2-colorings for a given @xmath1-list assignment @xmath2 . therefore , star @xmath1-choosability is @xmath19 . in this section
we establish the hardness of star @xmath1-choosability .
for @xmath13 we reduce the complement of an instance of @xmath70 into an instance of star @xmath29-choosability .
then , we proceed by induction showing how to reduce an instance of star @xmath1-choosability into an instance of star @xmath71-choosability for every @xmath12 .
the proof for the case @xmath13 is , in some sense , an extension of theorem [ thm : stcol-2 ] .
the goal is to force the true literals of @xmath72 variables to have the same color as @xmath73 , so that a monochromatic maximal star centered at @xmath73 appears when the formula is false .
[ thm : stchose-2 ] star @xmath29-choosability is @xmath19-hard , and it remains @xmath19-hard even when its input is restricted to @xmath74-free graphs .
the hardness of star @xmath29-choosability is obtained by reducing the complement of @xmath70 .
that is , given a @xmath75-dnf formula @xmath76 with @xmath77 clauses @xmath78 , and @xmath79 variables @xmath80 , @xmath81 , @xmath82 , we build a graph @xmath0 that is @xmath29-list - choosable if and only if @xmath83 is true .
for the sake of simplicity , in this proof we use @xmath28 , @xmath84 , @xmath85 , and @xmath86 to refer to values in @xmath87 , @xmath88 , @xmath89 , and @xmath90 , respectively .
graph @xmath0 is similar to the graph in theorem [ thm : stcol-2 ] .
its vertex set is again divided into connection , inner , and leaf vertices . in turn , _ connection _ vertices are divided into a set @xmath91 , a set @xmath92 , sets @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , and @xmath98 with @xmath77 vertices each , and two vertices @xmath99 .
let @xmath100 , @xmath101 , and @xmath102 .
_ inner _ vertices are those vertices included in @xmath29-keepers , list switcher , clusters and @xmath29-forcers .
the following rules define inner vertices and the edges between connection vertices .
* edges : * @xmath73 is adjacent to all the vertices in @xmath69 , and if @xmath103 ( resp .
@xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 ) is a literal of @xmath109 , then @xmath110 is adjacent to @xmath111 $ ] ( resp .
@xmath112 $ ] , @xmath113 $ ] , @xmath114 $ ] , @xmath115 $ ] , @xmath116 $ ] ) .
* keepers : * @xmath73 and @xmath117 are connected by a @xmath29-keeper . *
list switchers : * there are list switchers connecting \{@xmath118 , @xmath111 $ ] } and \{@xmath112 $ ] , @xmath111 $ ] } ( for every @xmath119 ) , @xmath120 , -z_f[h]\}$ ] ( for every @xmath121 ) , and @xmath122 ( for every @xmath85 ) . * clusters : * there are @xmath77-clusters connecting @xmath123 , @xmath124 , and @xmath125 .
* forcers : * there are @xmath29-forcers connecting each vertex of @xmath97 ( for every @xmath86 ) and each vertex of @xmath126 . finally , every connection vertex other than @xmath73 is leafed
this ends up the construction of @xmath0 ( see figure [ fig : choosing ] ) , which can be easily computed from @xmath76 in polynomial time .
arguments similar to those in theorem [ thm : stcol-2 ] are enough to conclude that @xmath0 is @xmath74-free .
in theorem [ thm : stchose-2 ] ; for the sake of simplicity , we omit the edges from @xmath69 to @xmath127 .
square vertices represent vertices connected to a @xmath29-forcer . ]
we now show that @xmath128 is true if and only if @xmath0 is @xmath29-choosable .
we first show that if @xmath128 is true , then @xmath0 admits a star @xmath2-coloring @xmath43 for any @xmath29-list assignment @xmath2 .
the @xmath2-coloring @xmath43 is obtained by executing the following algorithm .
step 1 : : : for every @xmath129 connected to a forcer , let @xmath130 be @xmath2-admissible for @xmath129 .
such a color always exists by statement ( [ lem : forcer properties : extension ] ) of lemma [ lem : forcer properties ] .
suppose , w.l.o.g .
, that @xmath131 and let @xmath132 be a valuation of @xmath133 such that @xmath134 if and only if @xmath135 .
step 2 : : : by hypothesis , @xmath136 can be extended to include @xmath137 so that @xmath138 is true .
if @xmath139)$ ] or @xmath140 ) \neq l(-x_i[h])$ ] for some @xmath85 , then : + step 2.1 : ; ; let @xmath141 ) \neq \rho(-x_i[h])$ ] in such a way that @xmath141 ) = 1 $ ] if and only if @xmath142)$ ] and @xmath143 , while @xmath144 ) = 1 $ ] if and only if @xmath145)$ ] and @xmath146 .
step 2.2 : ; ; let @xmath147)$ ] .
step 2.3 : ; ; let @xmath148 ) \neq \rho(x_i)$ ] and @xmath149 ) \neq \rho(-x_i[k])$ ] for every @xmath150 .
+ if @xmath151 , then : + step 2.4 : ; ; let @xmath152 if and only if @xmath153 and @xmath143 , and @xmath154 .
+ note that , whichever case gets executed , @xmath155 and @xmath144 ) \neq \rho(x_i[h])$ ] for every @xmath85 .
step 3 : : : let @xmath156 ) \neq \rho(z_f[h])$ ] and @xmath157 .
step 4 : : : let @xmath43 for @xmath73 , leaves , and inner vertices be as in observation [ obs : leafed vertex ] , and lemmas [ lem : keeper choosability ] , [ lem : list switcher properties ] , [ lem : cluster choosability ] and [ lem : forcer properties ] .
observe that this is always possible .
in particular , observe that every pair of vertices connected by a list switcher have different colors , while every vertex connected to a forcer has an @xmath2-admissible color . we claim that @xmath43 is a star @xmath2-coloring of @xmath0 .
let @xmath158 be any maximal star of @xmath0 . by observation [ obs : leafed vertex ] and lemmas [ lem : keeper choosability ] , [ lem : list switcher properties ] , [ lem : cluster choosability ] and
[ lem : forcer properties ] , @xmath158 is not monochromatic when @xmath159 .
suppose , for the rest of the proof , that @xmath160 and @xmath161 is monochromatic .
by lemma [ lem : keeper choosability ] , this implies that @xmath162 , thus @xmath163 .
also , by lemma [ lem : cluster choosability ] , @xmath164 intersects no @xmath1-cluster , thus @xmath165 .
moreover , by statement ( [ lem : cluster properties : stars ] ) of lemma [ lem : cluster properties ] , @xmath166 equals either @xmath93 or @xmath94 , @xmath167 equals either @xmath95 or @xmath96 , and @xmath168 is either @xmath97 or @xmath98 , for every @xmath28 , @xmath84 , and @xmath86 .
extend @xmath136 to @xmath169 so that @xmath170 if and only if @xmath171 . by hypothesis , @xmath172 , thus there is some clause @xmath109 whose literals are all true according to @xmath136 . if @xmath110 has some neighbor in @xmath96 , then @xmath170 , thus @xmath171 and @xmath173 . if @xmath110 has some neighbor @xmath174 , then @xmath134 which means , by the way @xmath136 is defined for @xmath133 in step 1 , that @xmath135 . consequently , by step 3 , @xmath175 , i.e. , @xmath176 .
similarly , if @xmath110 has some neighbor in @xmath97 , then @xmath177 which means that @xmath178 .
thus , there must exist at least one vertex @xmath179 with @xmath180 .
then , since @xmath181 , it follows that @xmath182 . finally , if @xmath110 has a neighbor in @xmath94 , then @xmath143 , thus @xmath144 ) \neq 1 $ ] for some @xmath85 by either step 2.1 or step 2.4 .
hence , @xmath183 , thus @xmath184 .
analogously , @xmath110 has no neighbors in @xmath185 . summing up ,
since @xmath186 is an independent set , it follows that @xmath110 has no neighbors in @xmath164 , thus @xmath187 and @xmath188 is not monochromatic by step 3 .
for the converse , suppose @xmath0 is star @xmath29-choosable , and consider any valuation @xmath136 of @xmath133 .
define @xmath2 to be a @xmath29-list assignment of @xmath0 so that @xmath189 is the unique color admissible for all the vertices in @xmath97 , @xmath190 is the unique color admissible for all the vertices in @xmath191 , and @xmath192 for every vertex not connected to a @xmath29-forcer . by statement ( [ lem : forcer properties : color ] ) of lemma [ lem : forcer properties ] , such list assignment @xmath2 always exists .
let @xmath43 be a star @xmath2-coloring of @xmath0 and extend @xmath136 to include @xmath193 in its domain so that @xmath194 .
note that @xmath195 can take any value from @xmath196 , so it is enough to prove that @xmath197 .
define @xmath198 , @xmath199 , and @xmath200 , and let @xmath201 . as in theorem
[ thm : stcol-2 ] , it can be observed that ( i ) @xmath164 is a monochromatic star and ( ii ) every vertex in @xmath202 is either adjacent or equal to a vertex in @xmath164 .
thus , since @xmath43 is a star @xmath2-coloring of @xmath0 , there must be some vertex @xmath110 adjacent to no vertex in @xmath203 .
moreover , such vertex @xmath110 corresponds to some clause @xmath109 whose literals are all true by the way @xmath136 is defined .
the proof for @xmath5 is by induction , i.e. , we reduce star @xmath1-choosability into star @xmath71-choosability for every @xmath12 . roughly speaking ,
the idea of the reduction is to insert a vertex @xmath72 that forbids every vertex of the reduced graph to have the same color as @xmath72 .
[ thm : stchose - k ] star @xmath1-choosability is @xmath19-complete for every @xmath12 , and it remains @xmath19-complete when the input is restricted to @xmath204-free graphs .
the proof is by induction on @xmath1 .
the base case @xmath205 corresponds to theorem [ thm : stchose-2 ] . for the inductive step ,
we show how to transform a @xmath204-free graph @xmath206 into a @xmath207-free graph @xmath208 so that @xmath206 is star @xmath1-choosable if and only if @xmath208 is star @xmath71-choosable .
the vertices of @xmath208 are divided into connection and inner vertices .
connection vertices comprise a set @xmath209 inducing @xmath210 and a vertex @xmath72 .
inner vertices are included in @xmath71-forcers or @xmath71-switchers connecting connection vertices .
there is a @xmath71-forcer connecting @xmath72 , and a @xmath71-switcher connecting @xmath211 for every @xmath212 .
let @xmath213 be the @xmath71-switcher connecting @xmath214 , i.e. , @xmath215 and @xmath216 are cliques of @xmath208 . by statement ( [ lem : forcer properties : forbidden ] ) of lemmas [ lem : switcher properties ] and [ lem : forcer properties ] , @xmath208 is @xmath217-free .
suppose @xmath206 is star @xmath1-choosable .
let @xmath218 be a @xmath71-list assignment of @xmath208 , and @xmath219 be @xmath2-admissible for @xmath72 . recall that @xmath220 always exists by statement ( [ lem : forcer properties : extension ] ) of lemma [ lem : forcer properties ] .
define @xmath221 as a @xmath1-list assignment of @xmath222 $ ] such that @xmath223 for @xmath212 . by hypothesis
, there is a star @xmath221-coloring @xmath43 of @xmath222 $ ] .
define @xmath224 to be the @xmath218-coloring of @xmath208 such that @xmath225 for @xmath212 and @xmath226 .
inner vertices are colored according to lemma [ lem : switcher choosability ] and statement ( [ lem : forcer properties : extension ] ) of lemma [ lem : forcer properties ] .
clearly , if @xmath158 is a maximal star of @xmath208 and @xmath129 is a connection vertex , then either @xmath227 or @xmath158 includes a maximal star of @xmath228 $ ] .
whichever the case , @xmath158 is not monochromatic , i.e. , @xmath224 is a star coloring of @xmath208 . for the converse ,
let @xmath221 be an @xmath1-list assignment of @xmath222 $ ] and take a color @xmath229 .
define @xmath218 as any @xmath71-list assignment of @xmath208 such that @xmath45 is the unique @xmath218-admissible color for @xmath72 , and @xmath230 for every @xmath212 .
such list assignment always exists by statement ( [ lem : forcer properties : color ] ) of lemma [ lem : forcer properties ] .
let @xmath224 be a star @xmath218-coloring of @xmath208 . by construction , @xmath231 , and by lemma [ lem : switcher properties ] , @xmath232 .
hence , the restriction @xmath43 of @xmath224 to @xmath209 is an @xmath221-coloring of @xmath222 $ ] .
moreover , if @xmath158 is a maximal star of @xmath222 $ ] , then @xmath233 is a maximal star of @xmath208 , for every @xmath234 .
since @xmath213 is a block of @xmath208 and @xmath235 , it follows that @xmath236 . hence
, @xmath158 is not monochromatic .
the previous sections dealt with time complexity of the star and biclique coloring and choosability problems . the remaining of the article is devoted to these problems in other restricted classes of graphs . as discussed in section [ sec : chordal ] , we are interested in classes of graphs that are related to chordal graphs or can be defined by forbidding small induced subgraphs . in this section ,
we study the classes of @xmath8-free graphs , for every graph @xmath8 on three vertices .
there are four graphs with exactly three vertices , namely @xmath23 , @xmath24 , @xmath25 , and @xmath26 .
the following theorem shows that @xmath23-free graphs are star @xmath29-choosable .
[ thm : stcol triangle - free ] every @xmath23-free graph is star @xmath29-choosable .
furthermore , for any @xmath29-list assignment , a star @xmath2-coloring can be obtained in linear time .
let @xmath2 be a @xmath29-list assignment of a @xmath23-free graph @xmath0 , @xmath237 be a rooted tree subgraph of @xmath0 with @xmath238 , @xmath239 be the root of @xmath237 , and @xmath240 be the parent of @xmath14 in @xmath237 for each @xmath241 .
define @xmath43 to be an @xmath2-coloring of @xmath0 where @xmath242 and @xmath243 for every @xmath241 . since @xmath0 is @xmath23-free
, @xmath244 is a maximal star of @xmath0 for @xmath30 only if @xmath245 , hence @xmath244 is not monochromatic .
observe that a bfs traversal of @xmath0 is enough to compute @xmath43 , thus @xmath43 is computed in linear time from @xmath0 . as a corollary
, we obtain that @xmath246-free graphs are biclique @xmath29-choosable also .
however , this corollary can be easily strengthened so as to include those @xmath23-free graphs that are _ biclique - dominated_. a graph @xmath0 is when every maximal biclique is either a star or has a false dominated vertex .
some interesting classes of graphs are @xmath23-free and biclique - dominated , including hereditary biclique - helly graphs @xcite .
every @xmath23-free graph that is biclique - dominated is biclique @xmath29-choosable .
furthermore , for any @xmath29-list assignment , a biclique @xmath2-coloring can be computed in polynomial time .
let @xmath2 be a @xmath29-list assignment of a @xmath23-free graph @xmath0 that is biclique - dominated .
the algorithm for biclique @xmath2-coloring @xmath0 has two steps .
first , apply theorem [ thm : stcol triangle - free ] on @xmath0 so as to obtain a star @xmath2-coloring @xmath43 of @xmath0 .
second , traverse each vertex @xmath129 and , for each @xmath14 that is false dominated by @xmath129 , change @xmath57 with any color in @xmath247 .
( it is not important if @xmath57 or @xmath130 are later changed when other vertices are examined . )
the coloring thus generated is a biclique @xmath2-coloring .
indeed , if a maximal biclique contains a false dominated vertex @xmath14 , then it also contains the vertex @xmath129 such that @xmath57 was last changed in the second step while traversing @xmath129 . since
false domination is a transitive relation , it follows that @xmath248 when the second step is completed . on the other hand , if @xmath164 is a maximal biclique with no false dominated vertices , then @xmath164 is a star . since the colors of the vertices of @xmath164 are not affected by the second step , we obtain that @xmath164 is not monochromatic .
it is not hard to see that the algorithm requires polynomial time .
coloring a connected @xmath24-free graph is trivial because the unique connected @xmath24-free graph @xmath0 with @xmath249 vertices is @xmath250 .
thus , @xmath251 . if @xmath0 is a connected @xmath24-free graph with @xmath249 vertices , then @xmath251 .
the case of @xmath25-free graphs , examined in the next theorem , is not much harder . if @xmath0 is a @xmath25-free graph with @xmath1 universal vertices , then @xmath252 .
let @xmath253 be the set of universal vertices of @xmath0 .
clearly , @xmath253 is a block of @xmath0 , thus @xmath254 and @xmath255 by observation [ obs : block coloring ] . for the other bound , let @xmath2 be a @xmath1-list assignment of @xmath0 , and @xmath256 be the sets of vertices that induce components of @xmath257 .
define @xmath43 as an @xmath2-coloring of @xmath0 such that @xmath258 and @xmath259 for @xmath260 .
note that @xmath261 is a set of false twin vertices ( @xmath262 ) because @xmath261 is a clique of @xmath263 .
thus , every maximal star or biclique @xmath164 is formed by two vertices of @xmath253 or it contains a set @xmath261 for some @xmath260 .
whichever the case , @xmath164 is not monochromatic , thus @xmath43 is a star and biclique @xmath2-coloring .
the remaining class is the class of @xmath26-free graphs . by definition ,
if @xmath0 is @xmath26-free , then every maximal star and every maximal biclique of @xmath0 has @xmath264 vertices .
thus , it takes polynomial time to determine if an @xmath2-coloring of @xmath0 is a star or biclique coloring , for any @xmath1-list assignment @xmath2 .
hence , when restricted to @xmath26-free graphs , the star and biclique @xmath1-coloring problems belong to , while the star and biclique @xmath1-choosability problems belong to @xmath17 .
the next theorem shows that , when @xmath18 , the choosability problems are @xmath17-complete even when the input is further restricted to co - bipartite graphs .
[ thm : co - bipartite choosability ] star @xmath1-choosability and biclique @xmath1-choosability are @xmath17-complete for every @xmath18 when the input is restricted to co - bipartite graphs .
the proof is obtained by reducing the problem of determining if a connected bipartite graph with no false twins is vertex @xmath1-choosable , which is known to be @xmath17-complete @xcite .
let @xmath0 be a connected bipartite graph with no false twins , @xmath265 be a bipartition of @xmath0 , and @xmath266 .
define @xmath8 to be the bipartite graph obtained from @xmath0 by inserting , for every @xmath267 , the stars @xmath268 ( @xmath269 ) with @xmath270 and the edges @xmath271 , @xmath272 , @xmath273 , @xmath274 , @xmath275 , and @xmath276 ( see figure [ fig : co - bip choose ] ) .
we claim that @xmath0 is vertex @xmath1-choosable if and only if @xmath277 is star ( resp .
biclique ) @xmath1-choosable . in theorem [ thm : co - bipartite choosability ] ; each independent set has @xmath278 vertices . ]
suppose first that @xmath0 is vertex @xmath1-choosable , and let @xmath2 be a @xmath1-list assignment of @xmath8 and @xmath50 be the restriction of @xmath2 to @xmath279 . by hypothesis , @xmath0 admits a vertex @xmath50-coloring @xmath43 .
define @xmath224 to be any vertex @xmath2-coloring of @xmath8 so that @xmath280 for @xmath30 , and @xmath281 for every @xmath267 and every @xmath269 .
it is not hard to see that such a coloring always exists . clearly , every maximal star ( resp .
biclique ) @xmath164 of @xmath277 is formed by two twins of @xmath277 or it contains two vertices that are adjacent in @xmath8 . in the latter case @xmath164
is not @xmath224-monochromatic because @xmath224 is a vertex coloring of @xmath8 , while in the former case @xmath164 is not @xmath224-monochromatic because both of its vertices must belong to @xmath282 , as @xmath0 has no false twins , for some @xmath267 and some @xmath269 .
for the converse , suppose @xmath277 is star ( resp .
biclique ) @xmath1-choosable , and let @xmath50 be a @xmath1-list assignment of @xmath0 . define @xmath224 to be a star ( resp .
biclique ) @xmath2-coloring of @xmath277 , for the @xmath1-list assignment @xmath2 of @xmath8 where @xmath283 for every @xmath267 with @xmath284 , and every @xmath285 with @xmath269 .
suppose , to obtain a contradiction , that @xmath286 for some @xmath267 with @xmath284 and @xmath287 .
then , for every @xmath288 ( @xmath269 ) , we obtain that @xmath289 because @xmath290 is a maximal star ( resp .
biclique ) of @xmath277 .
hence , since @xmath282 is a block of @xmath277 , we obtain by observation [ obs : block coloring ] that @xmath291 for every @xmath292 . consequently , since @xmath293 is a maximal star ( resp .
biclique ) for every @xmath294 and every @xmath295 , it follows that @xmath296 .
analogously , @xmath297 for every @xmath269 .
but then , @xmath298 is a monochromatic maximal biclique that contains a maximal star , a contradiction .
therefore , @xmath299 for every edge @xmath300 of @xmath279 , which implies that the restriction of @xmath224 to @xmath279 is a vertex @xmath50-coloring of @xmath0 .
let @xmath0 be a @xmath23-free graph with no false twins , and define @xmath8 as the @xmath23-free graph that is obtained from @xmath0 as in theorem [ thm : co - bipartite choosability ] . by fixing the list assignment that maps each vertex to @xmath301 in the proof of theorem [ thm : co - bipartite choosability ]
, it can be observed that @xmath0 admits a vertex @xmath1-coloring if and only if @xmath277 admits a star ( resp .
biclique ) @xmath1-coloring , for every @xmath18 .
the problem of determining if a connected @xmath23-free graph with no false twins admits a vertex @xmath1-coloring is known to be -complete @xcite .
hence , the star and biclique @xmath1-coloring problems are @xmath302-complete when restricted to @xmath26-free graphs , for every @xmath18 .
[ thm : co - k3 coloring ] star @xmath1-coloring and biclique @xmath1-coloring are -complete for every @xmath18 when the input is restricted to @xmath26-free graphs .
the graph @xmath0 defined in theorem [ thm : stcol-2 ] contains a large number of induced diamonds . for instance , to force different colors on a pair of vertices @xmath14 and @xmath129 , a @xmath1-switcher @xmath303 connecting @xmath304 is used .
such switcher contains @xmath305 diamonds , one for each edge of @xmath303 .
an interesting question is , then , whether induced diamonds can be excluded from theorem [ thm : stcol-2 ] .
the answer is no , as we prove in this section that the star coloring problem is -complete for diamond - free graphs . by taking a deeper look at @xmath0
, it can be noted that every diamond of @xmath306 has a pair of twin vertices . in order to prove that the star coloring problem is -complete for diamond - free graphs ,
we show that the problem is even for the larger class of graphs in which every diamond has two twin vertices .
this class corresponds to the class of \{@xmath307 , dart , gem}-free graphs ( cf .
below ) , and it is worth to note that its graphs may admit an exponential number of maximal stars .
we also study the biclique coloring problem on this class , for which we prove that the problem is when there are no induced @xmath27 for @xmath308 . at the end of the section
, we study the star and biclique choosability problems , which turn to be @xmath17-hard for \{@xmath6 , dart , gem}-free graph .
let @xmath0 be a graph .
say that @xmath30 is if every pair of adjacent vertices @xmath309 not dominating @xmath14 are twins in @xmath310 $ ] .
the following lemma shows that \{@xmath307 , dart , gem}-free graphs are precisely those graphs in which every induced diamond has twin vertices , and they also correspond to those graphs is which every vertex is block separable .
this last condition is crucial in the coloring algorithms .
[ thm : w4dartgemequivalence ] the following statements are equivalent for a graph @xmath0 .
a. @xmath0 is \{@xmath307 , dart , gem}-free .
b. every induced diamond of @xmath0 contains a pair of twin vertices . c.
every @xmath30 is block separable .
\(i ) @xmath311 ( ii ) if @xmath312 induces a diamond with universal vertices @xmath313 and there exists @xmath314 , then @xmath315 induces a @xmath307 , a dart , or a gem in @xmath0 depending on the remaining adjacencies between @xmath316 and the vertices of @xmath317 .
\(ii ) @xmath311 ( iii ) suppose @xmath30 is not block separable , thus @xmath318 $ ] contains two adjacent vertices @xmath129 and @xmath72 not dominating @xmath14 that are not twins in @xmath319 $ ] ; say @xmath320 .
then , @xmath14 and @xmath72 are the universal vertices of a diamond containing @xmath129 and a vertex in @xmath321 , i.e. , @xmath0 contains an induced diamond with no twin vertices .
\(iii ) @xmath311 ( i ) the @xmath307 , dart , and gem graphs have a vertex of degree @xmath20 that is not block separable .
note that if @xmath14 is block separable , then @xmath318 $ ] can be partitioned into sets @xmath322 where @xmath323 and each @xmath261 is a block of @xmath319 $ ] .
moreover , no vertex in @xmath261 is adjacent to a vertex in @xmath324 , for @xmath325 .
we refer to @xmath322 as the of @xmath14 . by definition
, @xmath244 is a maximal star of @xmath0 with @xmath326 if and only if @xmath327 , @xmath328 and @xmath329 for @xmath330 . by theorem [ thm : w4dartgemequivalence ]
, every vertex of a \{@xmath307 , dart , gem}-free graph admits a block separation , hence the next result follows .
let @xmath0 be a \{@xmath307 , dart , gem}-free graph with a coloring @xmath43 .
then , @xmath43 is a star coloring of @xmath0 if and only if * @xmath331 for every block @xmath34 of @xmath0 , and * for every @xmath30 with block separation @xmath322 , there exists @xmath261 such that @xmath332 .
it is well known that the blocks of a graph @xmath0 can be computed in @xmath333 time .
hence , it takes @xmath334 time obtain the block separation of a block separable vertex @xmath14 , and , consequently , the star @xmath1-coloring and the star @xmath1-choosability problems are in and @xmath17for \{@xmath307 , dart , gem}-free graphs , respectively .
[ thm : np w4-dart - gem ] star @xmath1-coloring is when the input is restricted to \{@xmath307 , dart , gem}-free graphs .
[ thm : ptp w4-dart - gem ] star @xmath1-choosability is @xmath17when the input is restricted to \{@xmath307 , dart , gem}-free graphs .
we now consider the biclique coloring problem .
the algorithm for determining if a coloring @xmath43 is a biclique coloring of @xmath0 is divided in two steps .
first , it checks that no monochromatic maximal star is a maximal biclique .
then , it checks that @xmath0 contains no monochromatic maximal biclique @xmath335 with @xmath336 .
for the first step , suppose @xmath43 is a coloring of @xmath0 where @xmath337 for every block @xmath34 of @xmath0 .
let @xmath14 be a vertex with a block separation @xmath322 . as discussed above , @xmath244 is a maximal star if and only if @xmath327 , @xmath328 , and @xmath329 for every @xmath330 .
if @xmath244 is not a maximal biclique , then there exists @xmath338 $ ] adjacent to all the vertices in @xmath164 .
observe that @xmath129 has at most one neighbor in @xmath261 with color @xmath45 , for each color @xmath45 . otherwise , taking into account that twin vertices have different colors ,
@xmath339 would induce a diamond with no twin vertices , for @xmath340 . therefore , at most one monochromatic maximal star with center @xmath14 is included in a biclique containing @xmath129 , for each @xmath341 .
thus , to check if there is a monochromatic maximal biclique containing @xmath14 we first check whether @xmath342 .
if negative , then @xmath43 is not a biclique coloring of @xmath0 .
otherwise , all the monochromatic maximal stars with center in @xmath14 are generated in polynomial time , and for each such star @xmath244 it is tested if there exists @xmath338 $ ] adjacent to all the vertices in @xmath164 . [
lem : w4-dart - gem stars ] if a \{@xmath307 , dart , gem}-free graph @xmath0 and a coloring @xmath43 are given as input , then it takes polynomial time to determine if there exists a monochromatic maximal biclique @xmath244 with @xmath30 .
for the second step , suppose @xmath164 is an independent set with at least two vertices , and let @xmath343 .
note that if @xmath344 are adjacent , then they are twins in @xmath0 because @xmath345 are the universal vertices of any induced diamond formed by taking a pair of vertices in @xmath164 .
hence , @xmath346 can be partitioned into a collection @xmath347 of blocks of @xmath0 where no vertex in @xmath261 is adjacent to a vertex in @xmath324 , for @xmath325 .
thus , @xmath348 is a maximal biclique of @xmath0 if and only if no vertex of @xmath349 is complete to @xmath346 and @xmath350 for every @xmath330 .
that is , @xmath0 has a monochromatic maximal biclique @xmath348 if and only if @xmath351 , each block of @xmath346 has a vertex of color @xmath352 , and @xmath353 .
[ lem : w4-dart - gem bicliques ] let @xmath0 be a \{@xmath307 , dart , gem}-free graph . if an independent set @xmath164 and a coloring @xmath43 of @xmath0 are given as input , then it takes polynomial time to determine if @xmath0 has a monochromatic maximal biclique @xmath348 with @xmath354 .
if @xmath0 is @xmath27-free for some constant @xmath28 , then every biclique @xmath348 of @xmath0 with @xmath355 has @xmath356 .
thus , to determine if @xmath43 is a biclique coloring of @xmath0 , it is enough traverse every independent set @xmath164 of @xmath0 with @xmath357 vertices and to check that there exists no @xmath358 such that @xmath348 is a monochromatic maximal biclique . by lemmas [ lem : w4-dart - gem stars ] and [ lem : w4-dart - gem bicliques ] , it takes polynomial time to determine if there exists @xmath237 such that @xmath348 is a monochromatic maximal star .
since there are @xmath359 independent sets with at most @xmath28 vertices , the algorithm requires polynomial time .
we thus conclude that biclique @xmath1-coloring and biclique @xmath1-choosability are respectively and @xmath17when the input is restricted to \{@xmath27 , @xmath307 , dart , gem}-free graphs .
[ thm : np w4-dart - gem - kii ] biclique @xmath1-coloring is when the input is restricted to \{@xmath27 , @xmath307 , dart , gem}-free graphs , for @xmath360 .
biclique @xmath1-choosability is @xmath17when the input is restricted to \{@xmath27 , @xmath307 , dart , gem}-free graphs , for @xmath360 . in the rest of this section ,
we discuss the completeness of the star and biclique coloring and choosability problems . as in sections
[ sec : general case ] and [ sec : choosability ] , only one proof is used for each problem because @xmath6-free graphs are considered . for the reductions ,
two restricted satisfiability problems are required , namely nae - satand nae@xmath361sat .
a valuation @xmath136 of a cnf formula @xmath362 is a when all the clauses of @xmath362 have a true and a false literal .
the formula @xmath363 is when @xmath364 admits a nae - valuation , while @xmath365 is when every valuation of @xmath364 is a .
nae - satis the -complete problem ( see @xcite ) in which a cnf formula @xmath362 is given , and the goal is to determine if @xmath362 admits a nae - valuation .
analogously , nae@xmath361satis the @xmath17-complete problem ( see @xcite ) in which a cnf formula @xmath366 is given , and the purpose is to determine if @xmath367 is nae - true .
we begin discussing the completeness of the star coloring problem . in order
to avoid induced diamonds , we define a replacement of long switchers . [
def : diamond k - switcher ] let @xmath0 be a graph and @xmath368 be an independent set of @xmath0 with @xmath369 .
say that @xmath370 is a ( @xmath12 ) when @xmath164 can be partitioned into a vertex @xmath371 , a set of leafed vertices @xmath372 , a family @xmath373 of @xmath1-keepers , and a clique @xmath303 with @xmath1 vertices in such a way that @xmath374 is a clique , @xmath375 is a star , @xmath376 connects @xmath377 for @xmath378 , and there are no more edges adjacent to vertices in @xmath164 .
a diamond @xmath1-switcher is depicted in figure [ fig : diamond switcher ] .
the main properties of diamond switchers are given in the next lemma .
-switcher connecting @xmath379 . ]
[ lem : diamond switcher properties ] let @xmath0 be a graph and @xmath164 be a diamond @xmath1-switcher connecting @xmath380 ( @xmath12 ) . then
, a. no induced @xmath6 , diamond , or @xmath7 of @xmath0 contains a vertex of @xmath164,[lem : diamond switcher properties : forbidden ] b. @xmath381 for any star @xmath1-coloring @xmath43 of @xmath0 , and [ lem : diamond switcher properties : colors ] c. any @xmath1-coloring @xmath43 of @xmath382 in which @xmath381 can be extended into a @xmath1-coloring of @xmath0 in such a way that no monochromatic maximal star has its center in @xmath164.[lem : diamond switcher properties : extension ] let @xmath368 , @xmath383 , @xmath384 ,
, @xmath385 , and @xmath303 be as in definition [ def : diamond k - switcher ] .
statement ( [ lem : diamond switcher properties : forbidden ] ) follows by statement ( [ lem : keeper properties : forbidden ] ) of lemma [ lem : keeper properties ] , observing that @xmath386 is an independent set and that no induced diamond can contain a vertex in a @xmath1-keeper .
( [ lem : keeper properties : colors ] ) let @xmath43 be a star @xmath1-coloring of @xmath0 . since @xmath303 is a block of size @xmath1 , it contains a vertex @xmath45 with color @xmath387 by observation [ obs : block coloring ] .
then , taking into account that @xmath388 is a maximal star and @xmath389 by statement ( [ lem : keeper properties : colors ] ) of lemma [ lem : keeper properties ] for @xmath378 , it follows that @xmath390 for some @xmath391 .
( [ lem : keeper properties : extension ] ) to extend @xmath43 , first set @xmath392 and @xmath393 ( @xmath378 ) , and then iteratively extend @xmath43 to color the leaves and the @xmath1-keepers according to observation [ obs : leafed vertex ] and statement ( [ lem : keeper properties : extension ] ) of lemma [ lem : keeper properties ] . we are now ready to prove the -completeness of the star - coloring problem .
[ thm : w4dartgemnpc ] star @xmath1-coloring is @xmath302-complete when the input is restricted to \{@xmath6 , diamond , @xmath7}-free graphs for every @xmath12 . by theorem [ thm : np w4-dart - gem ] ,
star @xmath1-coloring is for \{@xmath6 , diamond , @xmath7}-free graphs . for the hardness part
, we show a polynomial time reduction from nae - sat .
that is , given a cnf formula @xmath362 with @xmath77 clauses @xmath394 and @xmath249 variables @xmath395 , we define a \{@xmath6 , diamond , @xmath7}-free graph @xmath0 such that @xmath362 admits a nae - valuation if and only if @xmath0 admits a star @xmath1-coloring .
the vertices of @xmath0 are divided into and vertices . for each @xmath260
there are two connection vertices @xmath396 the literals @xmath103 and @xmath104 , respectively . also , there are @xmath397 connection vertices @xmath398 .
let @xmath399 , @xmath400 , and @xmath401 for @xmath402 .
inner vertices are the vertices included in diamond @xmath1-switchers connecting connection vertices . for each @xmath403 and each @xmath404
there is a diamond @xmath1-switcher connecting @xmath405 .
also , for each @xmath406 there is a diamond @xmath1-switcher connecting @xmath407 .
finally , there is a diamond @xmath1-switcher connecting @xmath408 for every @xmath402 .
observe that @xmath409 is an independent set of @xmath0 .
thus , by statement ( [ lem : diamond switcher properties : forbidden ] ) of lemma [ lem : diamond switcher properties ] , @xmath0 is \{@xmath6 , diamond , @xmath7}-free .
suppose @xmath362 has a nae - valuation @xmath410 , and let @xmath43 be a @xmath1-coloring of the connection vertices such that @xmath411 and @xmath412 for @xmath260 and @xmath413 .
clearly , every color or valuation @xmath1-switcher connects a pair of vertices that have different colors .
also , since @xmath136 is a nae - valuation , every set @xmath408 ( @xmath402 ) has two vertices representing literals @xmath414 and @xmath415 of @xmath109 with @xmath416 .
hence , @xmath408 is not monochromatic , thus every clause diamond @xmath1-switcher connects a non - monochromatic set of vertices .
therefore , by statement ( [ lem : diamond switcher properties : extension ] ) of lemma [ lem : diamond switcher properties ] , @xmath43 can be iteratively extended into a star @xmath1-coloring of @xmath0 . for the converse , suppose @xmath0 admits a star @xmath1-coloring @xmath43 . by applying statement ( [ lem : diamond switcher properties : colors ] ) of lemma [ lem : diamond switcher properties ]
while considering the different kinds of diamond @xmath1-switchers , we observe the following facts .
first , by the color diamond @xmath1-switchers , @xmath417 and @xmath418 . then , we can assume that @xmath419 and @xmath420 .
hence , by the valuation diamond @xmath1-switcher connections , we obtain that @xmath421 for every @xmath406 .
thus , the mapping @xmath410 such that @xmath422 is a valuation .
moreover , by the clause diamond @xmath1-switcher connections , @xmath408 is not monochromatic for @xmath402 .
consequently , @xmath136 is a nae - valuation of @xmath362 .
observe that the graph @xmath0 defined in theorem [ thm : w4dartgemnpc ] is not chordal .
however , as discussed in section [ sec : chordal ] , every edge @xmath423 such that @xmath424 are connected by a @xmath1-keeper ( inside the diamond @xmath1-switchers ) can be subdivided so as to eliminate all the induced holes of length at most @xmath28 , for every @xmath308 .
we now deal with the star choosability problem .
recall that long switchers are not well suited for the star choosability problem because they contain keepers , and vertices connected by keepers need not have the same colors in every list coloring .
keepers are also present inside diamond switchers , thus it is not a surprise that diamond @xmath1-switchers are not star @xmath1-choosable .
for this reason , as in section [ sec : choosability ] , the proof is by induction , using list switchers for @xmath13 .
since list switchers contain induced diamonds , the @xmath11-hardness will be obtained for \{@xmath307 , gem , dart}-free graphs , and not for diamond - free graphs .
unfortunately , we did not find a way to avoid these diamonds .
moreover , some kind of forcers are required as well ; our forcers have induced diamonds that we were not able to remove either .
the hardness proof for @xmath13 is , in some sense , a combination of the proofs of theorems [ thm : stchose-2 ] and [ thm : w4dartgemnpc ] . roughly speaking
, the idea is to force the colors of the universal variables of the input formula as in theorem [ thm : stchose-2 ] , while a nae - valuation is encoded with colors as in theorem [ thm : w4dartgemnpc ] .
[ thm : w4-dart - gem ptp complete ] star @xmath29-choosability is @xmath17-hard when its input is restricted to \{@xmath6 , dart , gem , @xmath425}-free graphs .
the hardness of star @xmath29-choosability is obtained by reducing nae@xmath361sat .
that is , given a cnf formula @xmath426 with @xmath77 clauses @xmath78 , and @xmath427 variables @xmath80 , @xmath428 , we build a \{@xmath6 , dart , gem , @xmath425}-free graph @xmath0 that is star @xmath29-choosable if and only if @xmath429 is nae - true . for the sake of simplicity , in this proof we use @xmath28 , @xmath86 , and @xmath85 as indices that refer to values in @xmath87 , @xmath88 , and @xmath89 .
graph @xmath0 is an extension of the graph in theorem [ thm : w4dartgemnpc ] for @xmath205 ( replacing diamond switcher with list switchers ) .
it has a vertex @xmath118 ( resp .
@xmath430 , @xmath431 , @xmath432 ) @xmath103 ( resp .
@xmath104 , @xmath107 , @xmath108 ) for each @xmath28 ( and each @xmath86 ) , and two vertices @xmath117 , @xmath433 .
let @xmath434 , @xmath435 , @xmath436 , and @xmath437 .
graph @xmath0 also has which are the vertices in list switchers and @xmath29-forcers connecting connection vertices
. there are @xmath29-forcers connecting each vertex of @xmath438 , and list switchers connecting : @xmath439 and @xmath440 for each @xmath28 ; @xmath441 for each @xmath86 ; and @xmath442 and @xmath443 for each @xmath85 .
let @xmath2 be a @xmath29-list assignment of @xmath0 , and suppose @xmath444 is nae - true .
define @xmath43 as an @xmath2-coloring of the connection vertices satisfying the following conditions .
a. @xmath57 is any color @xmath2-admissible for @xmath445 .
such a color always exists by statement ( [ lem : forcer properties : extension ] ) of lemma [ lem : forcer properties ] .
suppose , w.l.o.g .
, that @xmath131 and @xmath446 , and define @xmath132 as a valuation of @xmath133 such that @xmath134 if and only if @xmath447 .
b. by hypothesis , @xmath136 can be extended into a nae - valuation of @xmath426 . if @xmath448 , then @xmath449 .
otherwise , @xmath450 .
similarly , @xmath451 if @xmath452 , while @xmath453 otherwise . c. @xmath454 .
it is not hard to see that @xmath43 can always be obtained .
observe that @xmath455 , @xmath456 only if @xmath457 , and @xmath458 is monochromatic only if @xmath459 .
therefore , @xmath43 can be extended into an @xmath2-coloring of @xmath0 by statement ( [ lem : list switcher properties : extension ] ) of lemma [ lem : list switcher properties ] and statement ( [ lem : forcer properties : extension ] ) of lemma [ lem : forcer properties ] .
for the converse , suppose @xmath0 is star @xmath29-choosable , and consider any valuation @xmath136 of @xmath133 .
define @xmath2 to be a @xmath29-list assignment of @xmath0 such that @xmath190 is the unique color admissible for @xmath117 , @xmath460 is the unique color admissible for @xmath433 , @xmath189 is the unique color admissible for @xmath431 , and @xmath461 for @xmath462 . by statement ( [ lem : forcer properties : color ] ) of lemma [ lem : forcer properties ]
, such a list assignment always exists .
let @xmath43 be a star @xmath2-coloring of @xmath0 . by repeatedly applying statement ( [ lem : list switcher properties : colors ] ) of lemma [ lem : list switcher properties ]
, it can be observed that none of @xmath463 , @xmath464 , @xmath465 , and @xmath466 are monochromatic .
therefore , @xmath136 is a nae - valuation of @xmath426 .
note that if @xmath303 is a @xmath1-switcher , then every induced diamond that contains a vertex @xmath467 also contains a twin of @xmath468 .
hence , by theorem [ thm : w4dartgemequivalence ] , if @xmath32 is a @xmath1-forcer connecting @xmath14 , then no vertex in @xmath32 belongs to an induced dart or gem .
consequently , if @xmath206 is a \{@xmath6 , dart , gem , @xmath7}-free graph , then the graph @xmath208 defined in the proof of theorem [ thm : stchose - k ] is \{@xmath6 , dart , gem , @xmath469}-free .
that is , a verbatim copy of the proof of theorem [ thm : stchose - k ] can be used to conclude the following .
star @xmath1-choosability is @xmath17-complete when its input is restricted to \{@xmath6 , dart , gem , @xmath7}-free graphs for every @xmath12 .
in this section we consider the star coloring and star choosability problems restricted to split graphs .
the reason for studying split graphs is that they form an important subclass of chordal graphs , and also correspond to the class of \{@xmath470 , @xmath6 , @xmath471}-free graphs @xcite . a graph @xmath0 is when its vertex set can be partitioned into an independent set @xmath472 and a clique @xmath473 .
there are @xmath474 maximal stars centered at @xmath30 , namely @xmath475 and @xmath476 for @xmath477 .
thus , the star coloring and star choosability problems on split graphs are and @xmath17 , respectively . in this section
we prove the completeness of both problems .
we begin observing that @xmath0 admits a star coloring with @xmath478 colors , where @xmath479 is the size of the maximum block .
indeed , each block @xmath34 of @xmath473 is colored with colors @xmath480 , while each vertex of @xmath472 is colored with color @xmath478 .
we record this fact in the following observation .
if @xmath0 is a split graph whose blocks have size at most @xmath479 , then @xmath0 admits a star coloring using @xmath478 colors .
furthermore , such a coloring can be obtained in linear time .
computing a star coloring of a split graph using @xmath478 colors is easy , but determining if @xmath479 colors suffice is an @xmath302-complete problem .
the proof of hardness is almost identical to the one in section [ sec : diamond - free ] for \{@xmath6 , diamond , @xmath7}-free graphs .
that is , given a cnf formula @xmath362 we build a graph @xmath0 using _ split _ switchers in such a way that @xmath362 admits a nae - valuation if and only if @xmath0 admits a star @xmath1-coloring .
as switchers , switchers force a set of vertices to have at least two colors in a star @xmath1-coloring .
the difference is that split switchers do so in a split graph ( see figure [ fig : split - switcher ] ) .
-switcher connecting a set @xmath209 . ]
[ def : nae connection ] let @xmath0 be a split graph and @xmath481 with @xmath482 .
say that @xmath483 is a when @xmath164 can be partitioned into two sets @xmath484 and two vertices @xmath485 in such a way that @xmath486 , @xmath487 is a clique for every @xmath488 , @xmath489 and @xmath490 are cliques , there are no more edges between vertices in @xmath409 and vertices in @xmath472 , and there are no more edges incident to @xmath491 and @xmath492 .
the properties of split @xmath1-switchers are summarized in the following lemma .
[ lem : split switcher properties ] let @xmath0 be a split graph and @xmath164 be a split @xmath1-switcher connecting @xmath493 .
then , a. @xmath494 for every star @xmath1-coloring @xmath43 of @xmath0 , and[lem : split switcher properties : colors ] b. for every @xmath1-list assignment of @xmath0 , any @xmath2-coloring @xmath43 of @xmath382 in which @xmath494 can be extended into an @xmath2-coloring of @xmath0 in such a way that no monochromatic maximal star has its center in @xmath164.[lem : split switcher properties : extension ] let @xmath495 , @xmath496 , @xmath491 , and @xmath492 be as in definition [ def : nae connection ] .
( [ lem : split switcher properties : colors ] ) let @xmath497 be a star @xmath1-coloring of @xmath0
. by definition , @xmath495 and @xmath496 are blocks of @xmath0 , thus @xmath498 by observation [ obs : block coloring ] .
let @xmath499 and @xmath500 be the vertices with color @xmath45 in @xmath495 and @xmath496 , respectively . since the maximal star @xmath501 is not monochromatic , it follows that @xmath502 .
therefore , @xmath494 .
( [ lem : split switcher properties : extension ] ) to extend @xmath43 to @xmath164 , define @xmath498 , and @xmath503 .
let @xmath504 be a maximal star with @xmath505 . if either @xmath506 or @xmath284 and @xmath507 , then @xmath508 , thus @xmath504 is not monochromatic . otherwise , if @xmath284 and @xmath509 , then @xmath510 , thus @xmath504 is not monochromatic as well . by replacing diamond @xmath1-switchers with split @xmath1-switchers in the proof of theorem [ thm : w4dartgemnpc ]
, the @xmath302-completeness of the star @xmath1-coloring problem for split graphs is obtained . for the sake of completeness , we sketch the proof , showing how to build the graph @xmath0 from the cnf formula .
[ thm : splitnpc ] star @xmath1-coloring restricted to split graphs is @xmath302-complete for every @xmath12 .
split graphs have @xmath333 maximal stars , hence star @xmath1-coloring is @xmath302 for split graphs . for the hardness part ,
let @xmath362 be a cnf formula with @xmath511 clauses @xmath512 and @xmath249 variables @xmath395 .
define @xmath0 as the split graphs with and vertices , as follows . for each @xmath260
there are two connection vertices @xmath396 the literals @xmath103 and @xmath104 , respectively . also , there are @xmath397 connection vertices @xmath398 .
let @xmath399 , @xmath400 , and @xmath401 for @xmath402 .
inner vertices form the split @xmath1-switchers connecting connection vertices . for each @xmath403 and each @xmath404
there is a split @xmath1-switcher connecting @xmath405 . also , for each @xmath406 there is a split @xmath1-switcher connecting @xmath407 .
finally , there is a split @xmath1-switcher connecting @xmath408 for every @xmath402 .
clearly , @xmath0 is a split graph , and , as in the proof of theorem [ thm : w4dartgemnpc ] , @xmath0 admits star @xmath1-star coloring if and only if @xmath362 admits a nae - valuation . for the star choosability problem of split graphs ,
the idea is to adapt the proof of theorem [ thm : w4-dart - gem ptp complete ] , providing a new kind of forcer .
this new forcer is called the _
split @xmath1-forcer _ and it is just a @xmath1-forcer where its @xmath1-switchers form a clique . for the sake of completeness , we include its definition .
[ def : k - split - forcer connection ] let @xmath0 be a split graph and @xmath513 .
say that @xmath514 is a _
split @xmath1-forcer connecting @xmath14 _ ( @xmath12 ) when @xmath32 can be partitioned into sets @xmath515 and @xmath516 for @xmath517 and @xmath37 in such a way that @xmath38 , @xmath518 , @xmath519 and @xmath520 are cliques , there are no more edges between vertices in @xmath521 and vertices in @xmath472 , and there are no more edges incident to vertices in @xmath41 .
let @xmath2 be a @xmath1-list assignment of a split graph @xmath0 and @xmath32 be a split @xmath1-forcer connecting @xmath30 .
as in section [ sec : choosability ] , we say that @xmath42 is _
@xmath2-admissible for @xmath14 _ when there is an @xmath2-coloring @xmath43 of @xmath0 such that @xmath44 and no monochromatic maximal star has its center in @xmath32 .
the following lemma resembles lemma [ lem : forcer properties ] .
[ lem : split forcer properties ] let @xmath0 be a split graph and @xmath32 be a split @xmath1-forcer connecting @xmath513 .
then , a. for every @xmath1-list assignment @xmath2 of @xmath0 there is an @xmath2-admissible color for @xmath14 , and[lem : split forcer properties : extension ] b. every @xmath1-list assignment @xmath2 of @xmath46 can be extended into a @xmath1-list assignment of @xmath0 in which @xmath14 has a unique @xmath2-admissible color .
the lemma can be proven with a verbatim copy of the proof of lemma [ lem : forcer properties ] .
in particular , observe that a split @xmath1-forcer can be obtained from a @xmath1-forcer by inserting the edges between @xmath35 and @xmath522 , for every @xmath523 and @xmath524 . the hardness of the star choosability problem is also obtained by adapting the proof of theorem [ thm : w4-dart - gem ptp complete ] .
we remark , however , that in this case no induction is required , because split @xmath1-switchers are @xmath1-choosable by statement ( [ lem : split switcher properties : extension ] ) of lemma [ lem : split switcher properties ] . the proof is sketched in the following theorem .
[ thm : split stchose ] star @xmath1-choosability restricted to split graphs is @xmath17-complete for every @xmath12 .
star @xmath1-choosability is @xmath17for split graphs because split graphs have a polynomial amount of maximal stars .
let @xmath426 be a cnf formula with @xmath77 clauses @xmath78 , and @xmath427 variables @xmath80 , @xmath428 . use @xmath28 , @xmath86 , @xmath85 , and @xmath525 to denote indices in @xmath87 , @xmath88 , @xmath89 , and @xmath526 , respectively .
define @xmath0 as the split graph that has a vertex @xmath527 ( resp .
@xmath430 , @xmath431 , @xmath432 ) @xmath103 ( resp .
@xmath104 , @xmath107 , @xmath108 ) , and @xmath1 vertices @xmath117 , @xmath433 , @xmath528 .
let @xmath529 , @xmath400 , @xmath435 , @xmath436 , and @xmath437 .
graph @xmath0 also has that are the vertices of split @xmath1-switchers and split @xmath1-forcers connecting connection vertices .
there are split @xmath1-forcers connecting @xmath431 , @xmath117 , @xmath433 , and split @xmath1-switchers connecting : @xmath439 , @xmath440 , @xmath530 , @xmath531 , @xmath532 , and @xmath533 for every @xmath534 .
following the proof of theorem [ thm : w4-dart - gem ptp complete ] , it can be observed that , for @xmath205 , @xmath0 admits a star @xmath2-coloring , for a @xmath1-list assignment @xmath2 of @xmath0 , if and only if @xmath367 is nae - true . for @xmath5 , observe that if @xmath367 is nae - true , then a star @xmath2-coloring @xmath43 is obtained if @xmath535 is taken so that @xmath536 and @xmath537 .
conversely , if @xmath0 admits a star @xmath2-coloring for the @xmath1-list assignment in which @xmath538 for every connecting vertex @xmath539 , then @xmath540 , thus a nae - valuation of @xmath366 is obtained . to end this section , consider the more general class of @xmath541-free graphs . by definition
, @xmath244 is a star of a graph @xmath0 if and only if @xmath164 is a maximal independent set of @xmath310 $ ] . in @xcite
, it is proved that @xmath310 $ ] has @xmath334 maximal independent sets when @xmath0 is @xmath541-free .
thus , @xmath541-free graphs have @xmath542 maximal stars , which implies that the star coloring and star choosability problems on this class are and @xmath17 , respectively
. [ thm:2k2 coloring ] star @xmath1-coloring and star @xmath1-choosability are respectively -complete and @xmath17-complete for every @xmath12 when the input is restricted to @xmath541-free graphs .
threshold graphs form a well studied class of graphs which posses many definitions and characterizations @xcite .
the reason for studying them in this article is that threshold graphs are those split graphs with no induced @xmath543 s .
equivalently , a graph is a threshold graph if an only if it is @xmath544-free . in this section
we develop a linear time algorithm for deciding if a threshold graph @xmath0 admits a star @xmath1-coloring .
if affirmative , then a star @xmath1-coloring of @xmath0 can be obtained in linear time . if negative , then a certificate indicating why @xmath0 admits no coloring is obtained .
we prove also that @xmath0 is star @xmath1-choosable if and only if @xmath0 admits a star @xmath1-coloring .
thus , deciding whether @xmath0 is star @xmath1-choosable takes linear time as well .
it is worth noting that threshold graphs can be encoded with @xmath545 bits using two sequences of natural numbers ( cf .
we begin this section with some definitions on such sequences .
let @xmath546 be a sequence of natural numbers .
each @xmath547 is called an of @xmath164 . for @xmath266 ,
we write @xmath548}$ ] and @xmath549}$ ] to respectively indicate that @xmath550 and @xmath551 for every index @xmath28 .
similarly , we write @xmath552}$ ] when @xmath553}$ ] , i.e. , when @xmath554 for some index @xmath28 .
note that @xmath164 could be empty ; in such case , @xmath548}$ ] and @xmath549}$ ] for every @xmath266 . for indices
@xmath555 , we use @xmath556 $ ] to denote the sequence @xmath557 . if @xmath558 , then @xmath556 = \emptyset$ ] .
similarly , we define @xmath559 = s[i+1,j]$ ] , @xmath560 $ ] , and @xmath561 $ ] .
a is a pair @xmath562 of sequences of natural numbers such that @xmath563 .
let @xmath564 and @xmath546 .
each threshold representation defines a graph @xmath565 whose vertex set can be partitioned into @xmath566 blocks @xmath567 with @xmath568 and @xmath239 independent sets @xmath569 with @xmath570 such that , for @xmath571 , the vertices in @xmath376 are adjacent to all the vertices in @xmath572 .
it is well known that @xmath0 is a connected threshold graph if and only if it is isomorphic to @xmath565 for some threshold representation @xmath562 @xcite
. the following observation describes all the maximal stars of @xmath565 .
[ obs : stars threshold ] let @xmath562 be a threshold representation and @xmath14 be a vertex of @xmath565 .
then , @xmath573 is a maximal star of @xmath565 if and only if there are indices @xmath574 of @xmath575 such that @xmath576 , and @xmath577 for some vertex @xmath578 . for @xmath266
, we say that index @xmath28 of @xmath575 is for the threshold representation @xmath562 when either @xmath579 or @xmath580 and there exists some index @xmath581 such that @xmath582 , @xmath583}$ ] and @xmath584}$ ] .
the next theorem shows how to obtain a star @xmath1-coloring of @xmath0 when a threshold representation is provided .
[ thm : threshold coloring ] the following statements are equivalent for a threshold representation @xmath562 . 1 .
@xmath585 is star @xmath1-choosable .
2 . @xmath585 admits a star @xmath1-coloring .
no index of @xmath575 is not @xmath1-forbidden for @xmath562 .
\(i ) @xmath311 ( ii ) is trivial .
\(ii ) @xmath311 ( iii ) .
suppose @xmath565 admits a star @xmath1-coloring @xmath43 and yet @xmath575 contains some @xmath1-forbidden index @xmath28 . since @xmath376 is a block of @xmath565 , then @xmath586 by observation [ obs : block coloring ] .
hence , @xmath580 and there exists an index @xmath581 such that @xmath582 , @xmath587}$ ] , and @xmath588}$ ] .
let @xmath589 be the unique vertex in @xmath590 for @xmath591 . by observation [ obs : block coloring ] ,
both @xmath376 and @xmath592 have at least one vertex of each color @xmath593 , while for each index @xmath594 there exists a color @xmath595 such that @xmath596
. consequently , there are indices @xmath597 in @xmath598 such that @xmath599 and @xmath600 for every index @xmath601 .
indeed , it is enough to take @xmath602 as the maximum index with @xmath603 and @xmath604 as the minimum index such that @xmath605 . therefore , if @xmath606 and @xmath607 are the vertices of @xmath608 and @xmath609 with color @xmath45 , respectively , then @xmath610 is a monochromatic maximal star by observation [ obs : stars threshold ] , a contradiction .
\(iii ) @xmath311 ( i ) .
let @xmath2 be a @xmath1-list assignment of @xmath585 and define @xmath611 as any vertex of @xmath612 for each index @xmath28 of @xmath346 .
for each index @xmath28 of @xmath575 , define @xmath613 as the minimum index such that @xmath614 < { [ k]}$ ] .
let @xmath43 be an @xmath2-coloring of @xmath565 that satisfies all the following conditions for every index @xmath28 of @xmath575 :
1 . @xmath615 , 2 .
if @xmath616 and @xmath617 , then @xmath618 and @xmath619 do not belong to @xmath620 , 3 . if @xmath621 and @xmath622 , then @xmath623 , and 4 . if @xmath624 and @xmath617 , then @xmath625 . a coloring satisfying all the above conditions can obtained iteratively , by coloring the vertices in @xmath626 before coloring the vertices in @xmath627 for every pair of indices @xmath628 .
we claim that @xmath43 is a star @xmath2-coloring of @xmath565 .
to see why , let @xmath573 be a maximal star of @xmath565 . by observation [ obs : stars threshold ] , there are two indices @xmath574 of @xmath575 such that @xmath576 and @xmath629 for some @xmath578 . if @xmath630 , then @xmath248 by ( 1 ) .
if @xmath631}$ ] , then @xmath590 is not monochromatic by ( 4 ) . if @xmath616
, then @xmath632 is not monochromatic by ( 2 ) . if @xmath580 and @xmath633 , then @xmath634 , thus @xmath635 is not monochromatic by ( 2 ) . finally , if @xmath636 and @xmath584}$ ] , then there exists index @xmath85 such that @xmath637 ; otherwise @xmath28 would be a @xmath1-forbidden index of @xmath562
then , by ( 3 ) , @xmath638 which implies that @xmath209 is not monochromatic .
summing up , we conclude that @xmath585 has no monochromatic maximal star .
theorem [ thm : threshold coloring ] has several algorithmic consequences for a threshold representation @xmath562 of a graph @xmath0 . as mentioned , @xmath0 is a split graph where @xmath639 and @xmath640 , thus @xmath641 is either @xmath1 or @xmath642 , for @xmath643 . while deciding
if @xmath1 colors suffice for a general split graph is an -complete problem , only @xmath644 time is needed to decide whether @xmath645 when @xmath562 is given as input ; it is enough to find a @xmath1-forbidden index of @xmath575 .
furthermore , if @xmath646 , then a @xmath1-forbidden index can be obtained in @xmath644 time as well . also , if @xmath645 , then a star @xmath1-coloring @xmath43 of @xmath0 can be obtained in linear time by observing rules ( 1)(4 ) of implication ( iii ) @xmath311 ( i ) .
to obtain @xmath43 , begin traversing @xmath575 to find @xmath647 for every index @xmath28 of @xmath575 . then , color the vertices of each block @xmath376 with colors @xmath648 .
following , color the vertices @xmath649 in that order , taking the value of @xmath650 into account for each index @xmath28 of @xmath164 .
if @xmath651 , then @xmath652 ; otherwise , @xmath619 is any value in @xmath653 . finally , color the vertices in @xmath654 with color @xmath190 for each index @xmath28 of @xmath164 . to encode @xmath43
only two values are required for each index @xmath28 of @xmath575 , namely , @xmath650 and @xmath619 .
thus , @xmath43 can be obtained in @xmath644 time as well .
finally , @xmath562 can be obtained in @xmath333 time from @xmath0 when @xmath0 is encoded with adjacency lists @xcite .
thus , all the above algorithms take @xmath333 when applied to the adjacency list representation of @xmath0 .
we record all these observation in the theorem below .
let @xmath562 be a threshold representation of a graph @xmath0 .
the problems of computing @xmath641 , @xmath655 , a @xmath1-forbidden index of @xmath575 for @xmath646 , and a @xmath641-coloring of @xmath0 can be solved in @xmath644 time when @xmath562 is given as input . in turn , all these problems take @xmath333 time when an adjacency list representation of @xmath0 is given .
in this section we study the star coloring and star choosability problems on block graphs .
a graph is a if it is chordal and diamond - free .
we develop a linear time algorithm for deciding if a net - free block graph @xmath0 admits a star @xmath1-coloring .
the algorithm is certified ; a star @xmath1-coloring of @xmath0 is obtained in the affirmative case , while a forbidden induced subgraph of @xmath0 is obtained in the negative case . as threshold graphs , net - free block graphs admit an @xmath545 space representation using weighted trees ( cf . below ) .
the certificates provided by the algorithms are encoded in such trees , and can be easily transformed to certificates encoded in terms of @xmath0 .
we begin describing the weighted tree representation of net - free block graphs .
a is a pair @xmath656 where @xmath237 is a tree and @xmath253 is a from @xmath657 to @xmath658 .
the graph @xmath659 by @xmath656 is obtained by inserting a clique @xmath660 with @xmath661 vertices adjacent to @xmath14 and @xmath129 , for every @xmath662 .
it is well known that a connected graph is a netblock graph if and only if it is isomorphic to @xmath659 for some netblock representation @xmath656 @xcite . by definition ,
two vertices @xmath663 of @xmath659 are twins if and only if 1 .
@xmath664 for some @xmath662 , or 2 .
@xmath316 is a leaf of @xmath237 adjacent to @xmath665 and @xmath666 , or 3 .
both @xmath316 and @xmath667 are leafs of @xmath237 ( in which case @xmath659 is a complete graph ) .
the following observation describes the remaining maximal stars of @xmath659 .
[ obs : blockbicliques ] let @xmath656 be a netblock representation and @xmath14 be a vertex of @xmath659 .
then , @xmath244 is a maximal star of @xmath659 with @xmath668 if and only if @xmath14 is an internal vertex of @xmath237 and @xmath164 contains exactly one vertex of @xmath669 for every @xmath670 . let @xmath656 be a netblock representation . for @xmath266 , a of @xmath656 is a subtree of @xmath237 formed by edges of weight @xmath1 ; a @xmath1-subtree is a @xmath1-subtree whose edge set is not included in the edge set of another @xmath1-subtree .
a vertex of @xmath656 is a vertex @xmath14 that has some neighbor @xmath671 such that @xmath672 .
the following theorem characterize those net - free block graphs that admit a star @xmath1-coloring .
[ thm : block graph coloring ] the following statements are equivalent for a netblock representation @xmath656 .
a. @xmath659 is star @xmath1-choosable .
b. @xmath659 admits a star @xmath1-coloring . c. @xmath673 for every @xmath674 and every maximal @xmath675-subtree of @xmath656 contains a @xmath675-exit vertex of @xmath656 .
\(i ) @xmath311 ( ii ) is trivial .
\(ii ) @xmath311 ( iii ) .
suppose @xmath659 admits a star @xmath1-coloring @xmath43 . by observation [ obs : block coloring ] ,
@xmath673 for every @xmath674 and , moreover , @xmath676 when @xmath677 incides in a leaf .
let @xmath678 be a maximal @xmath675-subtree of @xmath656 , and consider a maximal path @xmath679 of @xmath678 such that @xmath680 for @xmath681 .
note that if @xmath682 , then @xmath683 , thus ( 1 ) @xmath684 .
we claim that @xmath685 is a @xmath675-exit vertex . indeed ,
if @xmath685 is a leaf of @xmath237 and @xmath129 is its unique neighbor in @xmath237 , then @xmath686 by ( 1 ) and the fact that @xmath687 is a block .
consequently , by the maximality of @xmath69 , it follows that @xmath688 , i.e. , @xmath685 is a @xmath675-exit vertex . on the other case , if @xmath685 is an internal vertex of @xmath237 , then , by observation [ obs : blockbicliques ] , @xmath689 is a maximal star of @xmath659 for every @xmath164 that contains exactly one vertex of @xmath690 for each @xmath670 .
therefore , there exists @xmath691 such that ( 2 ) @xmath692 . by ( 1 ) , @xmath693 , thus @xmath688 by the maximality of @xmath678 and @xmath69 . moreover , @xmath694 by ( 2 ) , thus @xmath685 is a @xmath675-exit vertex .
\(iii ) @xmath311 ( i ) . for
every maximal @xmath675-subtree @xmath678 of @xmath656 , let @xmath695 be a @xmath675-exit vertex of @xmath678 .
for every @xmath696 , define @xmath697 .
observe this definition is correct , because maximal @xmath675-subtrees are vertex - disjoint . let @xmath2 be a @xmath1-list assignment of @xmath698 , and define @xmath43 as an @xmath2-coloring of @xmath659 satisfying the following properties for every @xmath662 .
1 . @xmath248 , 2 .
@xmath699 and @xmath700 , and 3 . if @xmath701 and @xmath129 belongs to the unique path from @xmath14 to @xmath702 in @xmath237 , then @xmath703 . a coloring satisfying the above conditions
can obtained by first coloring the vertices of @xmath237 , and then coloring the vertices in @xmath704 for every @xmath674 .
observe , in particular , that if @xmath701 , then @xmath705 , thus ( 3 ) is always possible .
we claim that @xmath43 is a star @xmath2-coloring of @xmath659 .
let @xmath244 be a maximal star of @xmath659 .
suppose first that @xmath706 , thus either @xmath707 for @xmath708 or @xmath14 is a leaf of @xmath237 and @xmath709 for some @xmath710 .
in the former case @xmath711 by ( 2 ) . in the latter case , by ( 2 ) , either @xmath711 or @xmath712 .
if @xmath712 , then @xmath701 and @xmath129 is not a leaf of @xmath237 ; otherwise one of @xmath713 and @xmath714 would induce be a maximal @xmath675-subtree without @xmath675-exit vertices
. hence , @xmath711 by ( 3 ) .
suppose now that @xmath668 , thus @xmath14 is an internal vertex of @xmath237 by observation [ obs : blockbicliques ] . by hypothesis
, @xmath14 has some neighbor @xmath671 such that either @xmath715 ( when @xmath716 ) or @xmath129 belongs to the path of @xmath237 between @xmath14 and @xmath702 ( when @xmath717 ) . by observation [ obs : blockbicliques ]
, @xmath164 contains a vertex @xmath316 of @xmath660 . if @xmath716 , then @xmath715 , thus @xmath711 by ( 1 ) and ( 2 ) .
on the other case , if @xmath717 , then @xmath701 , thus @xmath718 by ( 3 ) .
the algorithmic consequences of theorem [ thm : block graph coloring ] are analogous as those observed for threshold graph in section [ sec : threshold ] .
if @xmath656 is a netblock representation and @xmath1 is maximum among the sizes of the blocks of @xmath719 , then @xmath720 equals either @xmath1 or @xmath642 .
when @xmath656 is given , it takes @xmath721 time to find all the @xmath675-subtrees and its @xmath675-exit vertices , if they exist .
thus , deciding if @xmath645 takes @xmath721 time when @xmath656 is given as input .
furthermore , if @xmath646 , then it takes @xmath721 time to compute a maximal @xmath675-subtree of @xmath656 with no @xmath675-exit vertices .
such a subtree can be transformed into an induced subgraph of @xmath0 in @xmath333 time if required .
also , a @xmath641-star - coloring @xmath43 of @xmath0 can be obtained in linear with rules ( 1)(3 ) of implication ( iii ) @xmath311 ( i ) .
first apply a bfs traversal of @xmath237 to color the vertices of @xmath237 with rule ( 1 ) , and then color the remaining vertices of @xmath0 following rules ( 2 ) and ( 3 ) .
finally , observe that @xmath656 can be obtained in @xmath333 time from @xmath0 when @xmath0 is encoded by adjacency lists .
thus , all the discussed algorithms take linear time when applied to the adjacency list representation of @xmath0 .
we record all these observation in the theorem below .
[ thm : block graph complexity ] let @xmath656 be a netblock representation of a graph @xmath0 .
the problems of computing @xmath641 , @xmath655 , a maximal @xmath675-subtree of @xmath656 with no @xmath675-exit vertices for @xmath646 , and a @xmath641-coloring of @xmath0 can be solved in @xmath721 time when @xmath656 is given as input . in turn , all these problems take @xmath333 time when an adjacency list representation of @xmath0 is given .
in this paper we investigated the time complexity of the star and biclique coloring and choosability problems . in this section
we discuss some open problems that follow from our research .
theorem [ thm : stcol-2 ] states that the star @xmath1-coloring problem is @xmath11-complete even when the input is restricted to \{@xmath6 , @xmath7}-free graphs . in section [ sec : chordal ]
we discussed how to generalize this theorem to include \{@xmath722 , @xmath7}-free graphs , for every @xmath308 .
an interesting question is what happens when @xmath28 grows to infinity , i.e. , what happens when chordal graphs are considered . by theorem [ thm :
splitnpc ] , we know that the star @xmath1-coloring problem is at least -hard on chordal graphs .
is it -complete or not ?
similarly , by theorem [ thm : split stchose ] , the star @xmath1-choosability problem on chordal graphs is at least @xmath17-hard ; is it @xmath17-complete ? to prove the @xmath11-completeness of the star @xmath1-coloring problem we showed how to transform a formula @xmath366 into a graph @xmath0 .
graph @xmath0 has many connection vertices that are joined together by @xmath1-keepers and @xmath1-switchers .
the purpose of the @xmath1-keepers is to force two vertices to have the same colors , while @xmath1-switchers are used to force different colors on a pair of vertices .
both @xmath1-keepers and @xmath1-switchers contain blocks of size @xmath1 . by taking a close examination at lemmas [ lem : keeper properties ] and [ lem : switcher properties ]
, it can be seen that these blocks play an important role when colors need to be forced .
an interesting question is whether these blocks can be avoided .
keepers and switchers not only have blocks of size @xmath1 ; when a @xmath1-keeper or a @xmath1-switcher connects two vertices @xmath14 and @xmath129 , a clique of size @xmath642 containing @xmath14 and @xmath129 is generated . we know that the star @xmath1-coloring problem is @xmath11-complete for @xmath7-free graphs , but what happens when @xmath724-free graphs are considered ?
the answer for the case @xmath13 is given by theorem [ thm : stcol triangle - free ] , i.e. , the star @xmath29-coloring problem is easy on @xmath23-free graphs . and for larger values of @xmath1 ? in section [ sec : choosability ] , the @xmath19-completeness of the star @xmath1-choosability problem on @xmath6-free graphs is proved by induction . for the case
@xmath13 , a graph @xmath727 is built from a dnf formula using the same ideas that we used to prove the hardness of the star @xmath29-coloring problem .
then , for the case @xmath728 , the graph @xmath729 is transformed into a graph @xmath210 .
the reason for splitting the proof in two cases is that long @xmath1-switchers can no longer be included into @xmath206 .
so , to avoid the inclusion of induced @xmath6 s , list switchers are used to build @xmath727 , while @xmath1-switchers are used to transform @xmath729 into @xmath210 for @xmath5 . this way ,
each generated graph @xmath206 is @xmath6-free .
we did not find a way to extend the holes in @xmath206 as it is done in section [ sec : chordal ] for the star @xmath1-coloring problem .
thus , @xmath206 contains @xmath471 as an induced subgraph for every @xmath5 , while @xmath727 contains @xmath730 as an induced subgraph .
is the problem simpler when such holes are avoided ?
the star coloring problem is easier when any of the graphs on three vertices does not appear as an induced subgraph , as discussed in section [ sec : small forbiddens ] .
similarly , the biclique coloring problem is simpler when it contains no induced @xmath24 , @xmath25 , or @xmath26 , or when it is @xmath23-free and biclique - dominated .
also , by theorem [ thm : np w4-dart - gem - kii ] , the biclique coloring problem on @xmath23-free graphs is `` only '' when it contains no induced @xmath27 for @xmath308 .
the star and biclique coloring problems are also simplified when diamonds are forbidden , as seen in section [ sec : diamond - free ] . by theorem
[ thm : np w4-dart - gem ] , the star @xmath1-coloring problem is for diamond - free graphs , while the biclique @xmath1-coloring problem is for \{diamond , @xmath27}-free graphs ( @xmath308 ) by theorem [ thm : np w4-dart - gem - kii ] .
additionally , if holes and nets are forbidden , then the star @xmath1-coloring problem can be solved easily by theorem [ thm : block graph complexity ] .
this leaves at least two interesting questions .
mrcia r. cerioli and andr l. korenchendler .
clique - coloring circular - arc graphs . in _
lagos09v latin - american algorithms , graphs and optimization symposium _ , volume 35 of _ electron .
notes discrete math .
_ , pages 287292 .
elsevier sci .
b. v. , amsterdam , 2009 .
l. lovsz .
coverings and coloring of hypergraphs . in _ proceedings of the fourth southeastern conference on combinatorics , graph theory , and computing ( florida atlantic univ . , boca raton , fla . , 1973 ) _ , pages 312 , winnipeg , man . , 1973
utilitas math .
hlio b. macdo filho , simone dantas , raphael c. s. machado , and celina m. h. de figueiredo .
biclique - colouring powers of paths and powers of cycles . in _ proc . of the 11th cologne - twente workshop on graphs and combinatorial optimization ,
ctw 2012 _ , pages 134138 .
arxiv:1203.2543 .
hlio b. macdo filho , raphael c. s. machado , and celina m. h. de figueiredo .
clique - colouring and biclique - colouring unichord - free graphs . in david fernndez - baca , editor , _ latin 2012 : theoretical informatics _ , volume 7256 of _ lecture notes in computer science _ , pages 530541 .
springer berlin heidelberg , 2012 . | a biclique of a graph @xmath0 is an induced complete bipartite graph .
a star of @xmath0 is a biclique contained in the closed neighborhood of a vertex .
a star ( biclique ) @xmath1-coloring of @xmath0 is a @xmath1-coloring of @xmath0 that contains no monochromatic maximal stars ( bicliques ) .
similarly , for a list assignment @xmath2 of @xmath0 , a star ( biclique ) @xmath2-coloring is an @xmath2-coloring of @xmath0 in which no maximal star ( biclique ) is monochromatic .
if @xmath0 admits a star ( biclique ) @xmath2-coloring for every @xmath1-list assignment @xmath2 , then @xmath0 is said to be star ( biclique ) @xmath1-choosable . in this article
we study the computational complexity of the star and biclique coloring and choosability problems .
specifically , we prove that the star ( biclique ) @xmath1-coloring and @xmath1-choosability problems are @xmath3-complete and @xmath4-complete for @xmath5 , respectively , even when the input graph contains no induced @xmath6 or @xmath7
. then , we study all these problems in some related classes of graphs , including @xmath8-free graphs for every @xmath8 on three vertices , graphs with restricted diamonds , split graphs , threshold graphs , and net - free block graphs . star coloring , biclique coloring , star choosability , biclique choosability . |
during the last decade , mounting evidence has been found supporting the idea that soft gamma - ray repeaters ( sgrs ) and anomalous x - ray pulsars ( axps ) are isolated neutron stars with peculiar properties resulting from the presence of an ultra strong magnetic field , b@xmath010@xmath110@xmath2 g. the magnetar model @xcite , initially developed to explain the sgrs , has been quite successfully applied to both classes of sources , which indeed show many commonalities .
actually , the distinction between sgrs and axps might well be only a semantic heritage linked to the way these objects were first observed : the sgrs , discovered as sources of short ( @xmath31 s ) repeating bursts of hard x - rays @xcite have persistent counterparts practically indistinguishable from the axps , while several axps , discovered as spinning down x - ray pulsars without evidence of companions stars @xcite , have been observed to emit short bursts very similar to those of the sgrs . for a recent review on axps and sgrs
. here we report on new x ray observations of two sources , and , obtained as target of opportunity requests to study their recent outbursts .
the results give further evidence for the similarities between axps and sgrs . among the small group of known magnetars ( about 15 sources ) the two objects discussed here stand out for having the shortest spin periods : 2.6 s for and 2.1 s for .
both sources are transients and lie in supernova remnants .
curiously , they are also located almost in the same direction of the galactic plane , hence their `` twins '' nickname .
, width=529 ] , width=377 ]
was discovered in 1998 , during a bursting state that lasted about six weeks @xcite .
its x - ray counterpart , identified at the time of the outburst , had a luminosity of @xmath4 erg s@xmath5 ( for d=11 kpc ) . in the following years its x ray luminosity monotonically decreased @xcite , until the lowest flux was observed in february 2008 with xmm - newton ( see fig .
[ fig : lc1627 ] ) .
this flux corresponds to a luminosity of only @xmath6 erg s@xmath5 , the lowest ever observed for a sgr @xcite .
the long - term flux decay can be interpreted as the cooling of the neutron star , assuming that the star was significantly heated during the outburst .
the modelling of the long term light curve can provide information on the mechanism for ( and location of ) the heating , as well as on the neutron star structure @xcite .
however , this is complicated by the fact that the measurements carried out over several years were obtained with different satellites .
this introduces some uncertainties due to the relative calibrations .
furthermore , the emitted fluxes are also difficult to estimate precisely , owing to the high absorption of this source ( several 10@xmath7 @xmath8 ) and the limited spectral information .
in fact a reanalysis of all the available data @xcite showed that some features of the the long term light curve , from which properties of the cooling model had been inferred , are not statistically significant .
in may 2008 started a new period of bursting activity .
its x ray luminosity reached a level higher than that observed in 1998 ( fig .
[ fig : lc1627 ] ) and several short bursts were also detected .
thanks to a series of swift observations we could monitor in detail the outburst evolution @xcite .
the x ray flux initially showed a rapid decrease , later followed by a shallower phase consistent with a power law of index @xmath00.2 . in fig .
[ fig : decays ] the light curve of the 2008 outburst is compared with that of the previous outburst from this source , and with the behavior seen after the outbursts of a few other axps / sgrs @xcite .
this figure shows that , whenever early data are available , they indicate that a single power law decay can not reproduce the source fading .
this is due to the presence of a steeper initial phase in the first days after the outburst and suggests the presence of two different mechanisms at play .
one possibility is that the steep phase be due to magnetospheric currents dissipation while the later phase reflect the effect of crustal cooling .
it is also possible that x - rays emitted during the initial bright burst , delayed by interstellar dust scattering , contribute to the initial steep phase ( see below ) . due to visibility constraints
, xmm - newton could not observe until september 2008 , thus the brightest part of the outburst could not be observed with this satellite .
a target of opportunity observation was performed on 2008 september 27 - 28 , and despite the low source flux @xmath9 erg @xmath8 s@xmath5 , the large effective area of the epic instrument allowed us to collect enough counts to perform a meaningful timing analysis .
this led to the discovery of the long - sought pulsations @xcite .
the spin period is 2.6 s , one of the shortest among magnetar candidates . the x ray pulse profile , characterized by two peaks of different intensity , is shown in fig . [ fig : fol1627 ] .
the deep xmm - newton observation also showed , for the first time , diffuse x ray emission from the vicinity of ( fig .
[ fig : ima1627 ] ) .
this consists of two spectrally distinct components with different spatial extent .
the harder emission is a spatially resolved source located about 1.5 arcmin south of the sgr .
its high absorption and evidence for a redshifted fe line suggest that it might be a cluster of galaxies .
the softer emission is more extended and most likely related to the supernova remnant / hii region complex ctb 33 @xcite .
the transient x ray source was discovered with the _ einstein observatory _ almost 30 years ago @xcite in the supernova remnant g327.240.13 . it attracted little interest until new x - ray and optical studies ruled out more standard interpretations and led to propose it as an axp candidate @xcite .
this suggestion was confirmed by the subsequent discovery @xcite of radio pulsations with @xmath10 s and period derivative @xmath11 s s@xmath5 . in october 2008 started an outburst with the emission of several short bursts and a significant increase in its x - ray flux @xcite .
80 kev with the anti - coincidence system of the spi instrument on board integral on january 22 , 2009 .
the initial spike of the longest burst had a duration of @xmath00.3 s and reached a peak flux greater than 2 10@xmath12 erg @xmath8 s@xmath5 ( 25 kev - 2 mev ) .
a modulation at 2.1 s , reflecting the neutron star rotation period , is clearly visible in the burst tail .
[ fig : burst1547 ] ] 1 arcmin .
two outer rings , produced by closer dust layers are also visible .
the ring dimensions were seen to increase in later observations , as expected for scattering by narrow dust layers of the x ray flux emitted during the strong bursting activity that took place around 6:48 ut of january 22 .
[ fig : ring ] ] the bursting activity from culminated on 2009 january 22 , when more than 200 bursts were detected in a few hours .
some of these bursts were particularly bright , reaching a peak flux above 2@xmath1310@xmath12 erg @xmath8 s@xmath5 at e@xmath1425 kev .
while most of the bursts had durations of few hundreds milliseconds , as typical sgr bursts , two bright events lasted several seconds and showed a clear modulation at the neutron star spin period .
the light curve of the most interesting one is displayed in fig .
[ fig : burst1547 ] : the burst is composed of a very bright , short ( @xmath00.3 s ) initial spike , followed by a @xmath08 s long pulsating tail @xcite .
these features are typical of giant flares from sgrs @xcite .
however , the analysis of integral data obtained with the spi anticoincidence shield indicates that the energy released in this event was only of the order of a few times 10@xmath15 erg ( for an assumed distance of 10 kpc ) , which is smaller than that of the three historical giant flares .
the energetics of the strongest bursts and flares from sgrs / axps are compared in fig .
[ fig : ene ] . as mentioned above , giant flares are characterized by an initial spike followed by a pulsating tail : the plot axis refer to these two features . note that in some cases only lower limits to the emitted energy could be derived due to instrument saturation .
the two points for sgr 180620 are for the generally assumed distance of 15 kpc and for the more recent estimate d=8.7 kpc @xcite .
the values reported in fig .
[ fig : ene ] clearly indicate that there is a rather continuous distribution of intensities , from the typical short bursts up to the brightest giant flares .
it is also noteworthy that extended pulsating tails have been detected not only for the three giant flares , but also after less intense bursts @xcite .
conversely , also a few examples of pulsating tails apparently without a bright initial hard spike have been observed @xcite .
this is possibly an indication that the spike emission is non - isotropic , a fact that adds a some uncertainty to proper estimates of the involved energy . immediately after the discovery of the strong bursting activity of january 22 , several follow - up pointings of were carried out with the swift satellite . during the first swift / xrt observations ,
the imaging mode could not be used because the source was too bright .
the first data providing full imaging ( fig .
[ fig : ring ] ) were obtained on january 23 at @xmath015:30 ut and showed the presence of remarkable dust scattering rings around the source position @xcite .
further observations carried out with swift , xmm - newton and chandra clearly showed that the angular size of the three rings increased with time .
dust scattering x - ray halos around bright galactic sources were predicted well before their observations with the first x - ray imaging instruments @xcite .
their study gives information on the properties and spatial distribution of the interstellar dust .
when the scattered radiation is a short burst / flare and the dust is concentrated in a relatively narrow cloud , an expanding ring ( instead of a steady diffuse halo ) appears , due to the difference in path - lengths at different scattering angles .
x - ray expanding rings due to dust scattering have been observed in a few gamma - ray bursts , and through their study accurate distances of the scattering dust clouds in our galaxy could be determined @xcite .
the dust scattering rings around are the brightest ever observed and the first ones for an axp / sgr . by fitting their expansion law
it is possible to determine the burst emission time , which is found to coincide with the interval of highest activity including the bright event of @xmath06:48 ut shown in fig .
[ fig : burst1547 ] . a comprehensive spectral analysis of all the available x ray data of the expanding rings around will allow us to determine the distances of the source and of the three dust layers @xcite . ,
j.l . , boer , m. , hurley , k. , niel , m. , vedrenne , g. , fenimore , e.e . ,
klebesadel , r.w . ,
laros , j.g . ,
kuznetsov , a.v . ,
sunyaev , r.a . ,
terekhov , o.v . ,
kouveliotou , c. , cline , t. , dennis , b. , desai , u. , orwig , l. 1987 , _ apj _ , 320 , l105 | we report on recent results obtained thanks to target of opportunity observations of the two galactic sources sgr 162741 and 1e 15475408 . these two transient sources present several similarities which support the interpretation of anomalous x
ray pulsars and soft gamma - ray repeaters as a single class of strongly magnetized neutron stars .
address = inaf - iasf milano , v. e.bassini 15 , 20133 milano , italy address = inaf - iasf milano , v. e.bassini 15 , 20133 milano , italy address = inaf - iasf milano , v. e.bassini 15 , 20133 milano , italy , altaddress = infn - pavia , via a. bassi 6 , 27100 pavia , italy address = inaf - iasf milano , v. e.bassini 15 , 20133 milano , italy address = inaf - iasf milano , v. e.bassini 15 , 20133 milano , italy , altaddress = iuss - istituto universitario di studi superiori , viale lungo ticino sforza 56 , 27100 pavia , italy address = cea saclay , dsm / irfu / service dastrophysique , orme des merisiers , bt .
709 , 91191 gif sur yvette , france address = mpi fr extraterrestrische physik , giessenbachstrasse , postfach 1312 , d-85741 garching , germany , altaddress = mpi halbleiterlabor , otto - hahn - ring 6 , 81739 muenchen , germany address = mpi fr extraterrestrische physik , giessenbachstrasse , postfach 1312 , d-85741 garching , germany address = inaf - osservatorio astronomico di roma , via frascati 33 , 00040 monteporzio catone , italy address = inaf - osservatorio astronomico di roma , via frascati 33 , 00040 monteporzio catone , italy address = astronomical institute `` anton pannekoek '' , university of amsterdam , kruislaan 403 , 1098 sj amsterdam , the netherlands address = dipartimento di fisica , universit degli studi di padova , via f. marzolo 8 , 35131 padova , italy address = mssl , university college london , holmbury st .
mary , dorking , surrey rh5 6nt , uk |
a major achievement of the renormalization - group ( rg ) theory is the proof that critical phenomena can be divided into distinct _ universality classes _ ( for a review see , e.g. , @xcite ) .
they are characterized by the spatial dimension @xmath2 and the symmetry of the ordered state which , for simplicity , we assume in the following to be @xmath10 symmetric with an @xmath11 component order parameter .
( for other universality classes see , e.g. , @xcite . ) within a given @xmath12 universality class , all bulk systems ( with finite - range interactions and with subleading long - range interactions of the van der waals type ) have the same critical exponents and the same thermodynamic functions near criticality in terms of universal scaling functions that are obtained after a rescaling of two amplitudes : that of the singular part of the bulk free energy density @xmath13 and that of the field @xmath14 conjugate to the order parameter .
this is summarized in the asymptotic ( small @xmath15 , small @xmath14 ) scaling form ( below @xmath16 dimensions ) @xmath17 with universal critical exponents @xmath18 and the universal scaling function @xmath19 above @xmath20 and below @xmath21 @xmath3 .
once the universal quantities are known one knows the asymptotic thermodynamic critical behavior of all members of the universality class provided that only the _ two _ nonuniversal amplitudes @xmath22 and @xmath23 are specified .
( for the application to real systems , additional experimental information is necessary to identify the order parameter and the appropriate thermodynamic path tangential to the coexistence line . )
we refer to this property as _ thermodynamic two - scale factor universality_. here universality means the independence of all microscopic details such as lattice structure , lattice spacing , and the specific form and magnitude of the finite - range or subleading long - range interaction .
this implies that both fluids and anisotropic solids within the same universality class have the same scaling function @xmath24 . [
cols="^,^,^,^,^ " , ] this important concept of scaling and thermodynamic two - scale factor universality was extended to the distance @xmath25 dependence of bulk correlation functions @xcite and to the size @xmath26 dependence of quantities of confined systems @xcite ( for reviews see , e.g. , @xcite ) .
it is this _ extended _ hypothesis which is in the focus of the present paper .
we shall present results for the finite - size critical behavior of the free energy above , at and below @xmath3 that demonstrate a considerable degree of diversity within a given @xmath12 universality class primarily due to spatial anisotropy in lattice systems with non - cubic symmetry , but also due to the lattice spacing @xmath27 in systems with cubic symmetry and due to the bare four - point coupling @xmath28 of the @xmath1 theory even in the isotropic case . in this context
we also discuss nonuniversal effects related to the cutoff and to subleading long - range ( van der waals type ) interactions .
this diversity suggests to distinguish subclasses of interactions within a given universality class where the subclasses have different bulk amplitude relations , different bulk correlation functions , and , for given geometry and boundary conditions ( b.c . ) , different finite - size scaling functions .
all of these nonuniversal differences exist in the _ asymptotic _ critical region @xmath29 , @xmath30 , and @xmath31 where corrections to scaling in the sense of wegner @xcite are negligible .
a summary of these properties is given in table i. the basic framework of rg theory is fully compatible with this diversity of critical behavior .
spatially anisotropic systems such as magnetic materials , alloys , superconductors @xcite , and solids with structural phase transitions @xcite represent an important class of systems with cooperative phenomena .
one may distinguish between long - range anisotropic interactions ( such as dipolar , rkky , and effective elastic interactions ) and short - range anisotropic interactions which include the dzyaloshinskii - moriya - type antisymmetric exchange @xcite and the spatially anisotropic heisenberg exchange interactions which , in the long - wavelength limit , are described by a @xmath32 anisotropy matrix @xmath33 @xcite .
we shall confine ourselves to a detailed study of the latter type of systems but the general aspects of our results have an impact also on the former type of anisotropic systems and on systems of other universality classes @xcite , for example on the range of validity of universality for anisotropic spin glasses @xcite or for anisotropic surface critical phenomena @xcite .
a characteristic feature of spatial anisotropy with non - cubic symmetry is the fact that there exists no unique bulk correlation - length amplitude but rather @xmath2 different amplitudes @xmath34 in the directions @xmath35 of the @xmath2 principal axes .
such systems still have a single correlation - length exponent @xmath36 provided that @xmath37 .
( we do not consider _ strongly _ anisotropic systems with critical exponents different from those of the usual @xmath38 universality classes , see e.g. @xcite . )
non - cubic anisotropy effects in crystals with cubic symmetry can be easily generated by applying shear forces . in earlier work on two - scale factor universality @xcite , isotropic systems with a single bulk correlation length @xmath39
were considered and important universal bulk amplitude relations were derived that depend on only two nonuniversal parameters .
recently some of these relations were reformulated for anisotropic systems within the same universality class @xcite . in sect .
iii of the present paper we give a derivation of these and other relations above and below @xmath3 and express them in terms of universal scaling functions .
the physical quantities entering these relations depend , in general , on @xmath40 nonuniversal parameters
. we also present the appropriate formulation of the bulk scattering intensity of anisotropic systems near criticality in terms of the eigenvalues of the anisotropy matrix and discuss the nonuniversal properties of bulk correlation functions . for _ confined _ systems with a characteristic length @xmath8 the hypothesis of two - scale factor universality
is summarized by the asymptotic ( large @xmath8 , small @xmath41 , small @xmath14 ) scaling form for the singular part of the free energy density ( divided by @xmath42 @xcite @xmath43 where @xmath44 , for given geometry and b.c . , is a universal scaling function and where the two constants @xmath45 and @xmath46 are universally related to the bulk constants @xmath22 and @xmath23 of ( [ 1a ] ) . for simplicity
we shall confine ourselves to a hypercubic shape with volume @xmath47 and with periodic b.c .. calculations of @xmath48 for this case were carried out within the spherical model @xcite which supported the form of ( [ 1b ] ) . for @xmath6 the scaling form ( [ 1b ] ) was discussed in the framework of the @xmath5 expansion @xcite .
no theoretical prediction for the function @xmath49 is available up to now for finite @xmath11 in cubic geometry , except in the large-@xmath11 limit @xcite .
monte carlo ( mc ) simulations @xcite for three - dimensional ising models with nearest - neighbor ( nn ) couplings on different lattices of cubic symmetry were consistent with the universality of the amplitude @xmath50 .
these models belong to the subclass of ( asymptotically ) isotropic systems .
it was already noted in @xcite that lattice anisotropy is a marginal perturbation in the rg sense , thus it was not obvious a priori to what extent two - scale factor universality is valid in the presence of anisotropic couplings @xcite .
it was also known that , for most anisotropic systems , ( asymptotic ) isotropy can be restored by an anisotropic scale transformation @xcite ( for further references see @xcite ) .
recently it was pointed out @xcite that , in systems with anisotropic interactions of non - cubic symmetry , the scaling function @xmath51 is indeed affected by anisotropy . in particular , it was shown @xcite that by means of an appropriate rescaling of lengths a transformation to an ( asymptotically ) isotropic system is always possible provided that the anisotropy matrix @xmath52 is positive definite and that the rescaling is performed along the @xmath2 _ nonuniversal _ directions of the _ principal _ axes which , in general , differ from the symmetry axes of the system .
this rescaling is equivalent to a shear transformation which distorts the shape , the lattice structure , and the boundary conditions in a nonuniversal way ( e.g. from a cube to a parallelepiped , from an orthorhombic to a triclinic lattice , and from periodic b.c . in rectangular directions to those in non - rectangular directions ) .
this nonuniversality is reflected in a dependence of the scaling function @xmath51 on the anisotropy matrix @xmath52 , in addition to the dependence on @xmath45 and @xmath46 .
specifically , on the basis of the results of renormalized perturbation theory in sects .
iv - vi , we propose that , for anisotropic systems with the shape of a cube , ( [ 1b ] ) is to be replaced by @xcite @xmath53 with @xmath54 , @xmath55 , and with the reduced anisotropy matrix @xmath56 , @xmath57 .
the nonuniversal constants @xmath58 and @xmath59 will be specified in sect .
vi in terms of the asymptotic amplitudes @xmath60 and @xmath61 of the second - moment bulk correlation lengths for @xmath62 , @xmath63 and for @xmath64 , @xmath65 , respectively , of the transformed isotropic system .
the free energy density @xmath66 of the parallelepiped with the volume @xmath67 and with @xmath68 ( isotropy ) then attains the scaling form @xmath69 where the characteristic length @xmath70 determines the overall size of the parallelepiped and @xmath71 equation ( [ 1d ] ) has the structure of the isotropic privman - fisher scaling form ( [ 1b ] ) with a rescaled length @xmath72 and with only two nonuniversal constants @xmath73 and @xmath74 which , superficially , appears to be in agreement with two - scale factor universality .
the remaining @xmath75 nonuniversal parameters , however , are hidden in the index `` iso , @xmath76 '' .
this index is the notation for a system with the shape of a parallelepiped whose interaction @xmath77 is ( asymptotically ) isotropic and whose @xmath78 angles and @xmath79 length ratios are determined by the @xmath80 nonuniversal parameters of the reduced anisotropy matrix @xmath76 .
these parameters appear in the calculation of @xmath81 via the summation over the discrete @xmath82 vectors in the fourier space of the parallelepiped system since the @xmath82 vectors depend explicitly on @xmath76 , unlike the @xmath83 vectors of the cubic system .
thus for the calculation of @xmath84 the same nonuniversal information is required as for the calculation of @xmath85 . for general @xmath33
the function @xmath86 was presented in @xcite for @xmath87 in the large-@xmath11 limit .
furthermore , quantitative predictions were made for the nonuniversal dependence of the critical binder cumulant @xcite @xmath88_{y=0 } \;\ ] ] for @xmath89 both in three @xcite and two @xcite dimensions .
mc simulations @xcite for the anisotropic three - dimensional ising model indeed showed nonuniversal anisotropy effects which , however , did not agree with the theoretical prediction .
more accurate mc simulations @xcite for the anisotropic two - dimensional ising model demonstrated the nonuniversality of the critical binder cumulant but no comparison with a quantitative theoretical prediction was available for this two - dimensional case . thus the anisotropic finite - size theory of @xcite is as yet unconfirmed . in sects .
iv - vi of this paper we derive the finite - size scaling function @xmath90 of the singular part of the excess free energy density @xmath91 at @xmath92 above , at , and below @xmath3 for the @xmath6 universality class in @xmath4 dimensions on the basis of the anisotropic @xmath1 lattice model . for the isotropic case at @xmath3
we find excellent agreement with the mc data of mon @xcite .
slightly below @xmath3 we find a minimum of the scaling function that is similar to the minimum of the scaling function of the critical casimir force for the @xmath93 ising model in slab geometry with periodic boundary conditions @xcite . for future tests of our theory by mc simulations we consider both three- and two - dimensional anisotropies . in both cases
we predict a measurable dependence of the minimum on the anisotropy parameters , thus demonstrating the nonuniversality of the finite - size scaling function of the excess free energy density .
the magnitude of this anisotropy effect is predicted to be considerably larger than that on the binder cumulant .
we believe that a similar nonuniversal dependence can be derived for the critical casimir force by means of our perturbation approach . for @xmath94 and at @xmath64 ,
the nonuniversality of the casimir amplitude due to anisotropy was already demonstrated within the @xmath1 theory in @xcite .
this suggests that , for given geometry and b. c. , the existing theoretical @xcite and mc @xcite results for the casimir force scaling function are not universal within the _ entire _ universality class but are restricted to the subclass of isotropic systems .
extensions to the subclass of anisotropic systems are , in general , not straightforward and can not be obtained just by transformations but require new nonuniversal input , new analytical and numerical calculations , and new mc simulations
. experiments , e.g. , in anisotropic superconducting films @xcite , could , in principle , demonstrate the nonuniversality in real systems .
our present results for the @xmath1 theory can not be applied directly to two - dimensional critical phenomena .
nevertheless we are able to study two - dimensional anisotropy effects within a three - dimensional model . for the purpose of a comparison with the two - dimensional mc data @xcite
, we consider ( in sect .
viii ) a three - dimensional @xmath1 lattice model with the same two - dimensional anisotropy in the horizontal planes as in the two - dimensional model ising model studied by selke and shchur @xcite .
our theory agrees quantitatively with the non - monotonicity of the binder cumulant as a function of the anisotropiy _
ferromagnetic _ next - nearest neighbor ( nnn ) coupling found in @xcite .
we also predict a non - monotonicity for small _ anti - ferromagnetic _ couplings and the existence of a lifshitz point at a larger value of this coupling .
very recent preliminary mc data by selke @xcite for the _ two_-dimensional ising model indeed reveal such a non - monotonicity that was not yet detected in @xcite .
we predict a similar anisotropy effect for the excess free energy density of the anisotropic two - dimensional ising model .
this effect can become quite large if one of the eigenvalues of @xmath76 approaches zero , in particular if a lifshitz point is approached ( sect .
viii ) .
an important property of the scaling form ( [ 1c ] ) is that it depends on @xmath76 but not on other nonuniversal parameters such as the bare four - point coupling , the lattice spacing , and the cutoff of @xmath1 field theory .
this is a kind of _ restricted universality _ since it implies that the same finite - size scaling functions exist for the large variety of those systems within a universality class that have the same reduced anisotropy matrix @xmath76 ( and the same geometry and boundary conditions ) . in sect .
ix we propose two examples for testing this hypothesis of restricted finite - size universality by mc simulations for spin models with _
anisotropic _ interactions . for recent tests of finite - size universality of two - dimensional ising models with ( asymptotically )
_ isotropic _ interactions see @xcite ( see also table 10.1 of @xcite ) . unlike the bulk scaling function @xmath95 , ( [ 1a ] ) , that is valid in the entire range @xmath96 of the scaling argument @xmath97 , the finite - size scaling functions such as @xmath98 are valid only in a limited range of @xmath99 and @xmath100 , above the shaded region in fig .
1 . in the shaded region ,
nonuniversal nonscaling effects become nonnegligible and even dominant for sufficiently large @xmath101 and @xmath102 for both short - range and subleading long - range interactions . in this region , not only the correlation lengths are relevant but also additional nonuniversal length scales such as the lattice spacing @xmath27 , the inverse cutoff @xmath103 , the length scale @xmath104 set by the four - point coupling , and the van - der - waals interaction - length @xmath105 , as discussed in sect .
x. plane at @xmath92 for the anisotropic @xmath106 theory in a cubic geometry with periodic boundary conditions . in the central finite - size region ( above the dashed lines ) , the lowest mode must be separated whereas outside this region ordinary perturbation theory is applicable . above the shaded region ,
finite - size scaling is valid but with scaling functions that depend on the anisotropy parameters @xmath107 , see ( [ 1c ] ) . in the large - @xmath72 regime at @xmath108
( shaded region ) finite - size scaling and universality are violated for both short - range and subleading long - range interactions and for both isotropic and anisotropic systems .
a similar plot is valid for the @xmath109 plane at @xmath64.,width=302 ] we briefly comment on the methodological aspects of our finite - size calculations .
as far as the field - theoretic @xcite renormalization of the @xmath1 lattice model is concerned we use the minimal subtraction scheme @xcite not within the @xmath7 expansion but at fixed dimension @xmath2 , as introduced in @xcite and further developed in @xcite . as far as finite - size theory
is concerned we further develop earlier approaches @xcite that have been successfully used to solve several finite - size problems in the past @xcite .
after the transformation from the anisotropic to an isotropic system , the same renormalization constants ( @xmath110 factors ) and the same fixed - point value @xmath111 of the renormalized four - point coupling are obtained as for the standard isotropic @xmath1 hamiltonian .
for this reason , the same fixed - point hamiltonian and the same critical exponents govern isotropic and ( weakly ) anisotropic systems - they belong to the same universality class . the crucial point , however , is that not only the fixed - point value @xmath111 but also the orientation of the eigenvectors ( principal axes ) of the fixed - point hamiltonian relative to the orientation of the given boundaries of the _ confined _ anisotropic system determine the finite - size scaling functions .
this is a physical fact that introduces a source of nonuniversality that can not be eliminated by transformations and that makes anisotropic confined systems distinctly different from isotropic confined systems within the same universality class .
the main result for @xmath112 will be obtained in the central finite - size regime ( above the dashed lines in fig .
1 ) where the finite - size effects are most significant and where it is necessary to separate the lowest - mode from the higher modes . in this regime finite - size scaling is valid in the form of ( [ 1c ] ) .
we compare the result of our fixed - @xmath2 perturbation approach @xcite with that of the @xmath7 expansion approach .
the advantage of the former approach is that it keeps the exponential structure of the order - parameter distribution function unexpanded .
this leads to a result at @xmath3 in excellent agreement with the mc data in the isotropic case @xcite and lends credibility also to the quantitative features of our predictions of anisotropy effects .
the @xmath7 expansion result at @xmath3 turns out to be in less good agreement .
plane at @xmath92 for anisotropic bulk systems . above the shaded region
, there exists a universal scaling function @xmath113 of the bulk correlation function @xmath114 , ( [ 3n ] ) .
the scaling argument , however , contains the spatial variable @xmath115 , ( [ 3x ] ) , that depends on the anisotropy matrix @xmath116 with @xmath117 nonuniversal parameters . in the large - @xmath118 regime at @xmath108 ( shaded region ) ,
scaling and universality are violated for both short - range and subleading long - range interactions and for both isotropic and anisotropic systems .
a similar plot is valid in the @xmath119 plane at @xmath120 , with @xmath121.,width=302 ] the separation of the lowest mode is inadequate in the limit of large @xmath122 at fixed @xmath9 . in order to capture the exponential structure of finite - size effects for large @xmath72 we complement ( in sect .
x ) our results by ordinary perturbation theory outside the central finite - size regime ( below the dashed lines in fig .
this includes a small but finite region where finite - size scaling is violated ( shaded region in fig .
1 ) . there exists diversity rather than universality of finite - size critical behavior in this region depending on all microscopic details of the interactions such as the lattice spacing , the bare four - point coupling , the cutoff of the @xmath1 theory , and the amplitude of subleading long - range interactions .
this diversity can be traced back to a corresponding diversity of the large - distance @xmath123 behavior of the _ bulk _ order parameter correlation function @xmath114 @xcite where @xmath118 is the distance in the transformed isotropic bulk system , as discussed in sect .
iii . for @xmath114
there exists a region of the @xmath124 plane ( shaded region in fig .
2 ) that is the analogue of the shaded region of fig .
1 . in the isotropic case ,
this region is of physical relevance for fluids with van der waals interactions @xcite .
we start from the o@xmath0 symmetric @xmath1 lattice hamiltonian ( divided by @xmath125 ) @xmath126 , \;\end{aligned}\ ] ] @xmath127 , @xmath128 with @xmath129 , @xmath130 .
the variables @xmath131 are @xmath11-component vectors on @xmath132 lattice points @xmath133 of a @xmath2-dimensional bravais lattice with the finite volume @xmath134 with the characteristic length @xmath135 where @xmath136 is the volume of the primitive cell .
the components @xmath137 of @xmath138 vary in the continuous range @xmath139 .
the couplings @xmath140 and the temperature variable @xmath141 have the dimension of @xmath142 whereas the variables @xmath138 have the dimension of @xmath143 such that @xmath144 is dimensionless .
the free energy per unit volume divided by @xmath145 is [ 2c ] f(t , h , l ) = - v^-1 z , [ 2d ] z ( t , h , l ) = ( - h ) where @xmath110 is the dimensionless partition function .
the total excess free energy density is defined as [ 2d-1 ] f^ex(t , h , l ) = f(t , h , l ) - f_b(t , h ) where @xmath146 is the bulk free energy density . following @xcite we shall decompose @xmath147 , for large @xmath8 , into singular and non - singular parts [ 2dd ] f(t , h , l ) = f_s(t , h , l ) + f_ns(t , l ) where @xmath148 has a regular @xmath41 dependence around @xmath149 . in earlier work on finite - size effects
it was supposed @xcite that , _ for periodic boundary conditions _
, one can assume that there exists no @xmath8 dependence of the nonsingular part @xmath150 .
adopting this assumption leads to a misinterpretation @xcite of the singular part @xmath151 of the free energy density and of the casimir force in the presence of a sharp cutoff of @xmath1 field theory . here
we shall not exclude the possibility of an @xmath8 dependent nonsingular part @xmath152 even for periodic boundary conditions if long - range correlations are present . as will be shown in sect .
x , this will reconcile the earlier results @xcite with the concepts of finite - size scaling . for periodic b.c . , the fourier representations are @xmath153 and @xmath154 where the summations @xmath155 run over the @xmath132 discrete vectors @xmath83 of the first brioulluin zone of the reciprocal lattice .
we assume finite - range interactions @xmath156 with a finite value @xmath157 . in terms of the fourier components the hamiltonian reads @xmath158
\hat \varphi({\mathbf k } ) \hat \varphi({-{\mathbf k}})- h \hat \varphi({\mathbf 0 } ) \nonumber\\ + \;u_0 v^{-3 } \sum_{{\mathbf{kp}}{\mathbf q } } [ \hat \varphi({\mathbf k } ) \hat \varphi({{\mathbf p } } ) ] [ \hat \varphi({{\mathbf q } } ) \hat \varphi({-{\mathbf k}-{\mathbf p}-{\mathbf q } } ) ] \qquad\end{aligned}\ ] ] where @xmath159 $ ] . in perturbation
theory , @xmath160 plays the role of an inverse propagator .
the hamiltonian @xmath144 is isotropic in the vector space of the @xmath11-component variables @xmath138 and @xmath161 but may be anisotropic in real space and @xmath83 space . a variety of anisotropies may arise both through the lattice structure and through the couplings @xmath156 .
they manifest themselves on macroscopic length scales via the @xmath32 anisotropy matrix @xmath162 and the anisotropy tensor @xmath163 that appear in the long - wavelength form @xmath164 odd powers of @xmath165 are excluded because of inversion symmetry of the bravais lattice .
( for the case of non - bravais lattices see the discussion in section ii.c of ref .
@xcite . ) for cubic symmetry , @xmath33 has the isotropic form @xmath166 while the @xmath167 terms of cubic systems differ from those of isotropic systems . in sects .
iii and x we shall consider the model ( [ 2 g ] ) also in a fully isotropic form with the short - range interaction @xmath168 including a finite cutoff @xmath169 and , for @xmath6 , with the subleading long - range interaction @xcite @xmath170 with @xmath171 .
the second term of the interaction ( [ 2hh ] ) is usually classified as `` irrelevant '' in the renormalization - group sense @xcite since it leaves _ some _ ( but not all ) of the universal quantities unchanged : critical exponents and bulk thermodynamic scaling functions . this terminology is somewhat misleading as the term @xmath172 changes not only the leading _ finite - size _ critical behavior at @xmath173 ( in the shaded region of fig.1 ) but it also destroys the universality of the leading _ bulk _ critical behavior of the order - parameter correlation function @xmath114 ( and of other bulk correlation functions ) : @xmath114 attains an interaction - dependent power - law structure @xcite in the large - distance regime at @xmath173 ( in the shaded region of fig .
2 ) whereas systems with purely short - range interaction in the same universality class have an _ exponentially _ decaying @xmath114 in this regime ( this decay has , in addition , a nonuniversal exponential tail , see sect .
the expression for @xmath174 and @xmath175 in terms of the microscopic couplings @xmath156 is given by the second moments @xcite @xmath176 and the fourth - order moments of @xmath156 , respectively .
they have been classified and studied in the context of the bulk correlation function in ref .
the symmetric matrix @xmath33 depends only on the lattice structure and on the pair interactions @xmath156 and is independent of the boundary conditions and the geometry of the system .
its eigenvalues @xmath177 and eigenvectors @xmath178 are determined by the eigenvalue equation @xmath179 with @xmath180 . in order to have an ordinary critical point of the usual @xmath12 universality classes we assume positive eigenvalues @xmath181 , and that the fourth - order moments @xmath182 enter only the corrections to scaling .
the critical point occurs at @xmath183 and at @xmath184 corresponding to some critical value @xmath185 that is defined implicitly by @xmath186 where @xmath187 is the bulk susceptibility for @xmath188 .
this implies that @xmath189 depends on the lattice structure , on @xmath136 , on @xmath28 , and on all couplings @xmath156 .
the matrix @xmath33 affects the observable bulk critical behavior : the eigenvalues @xmath190 enter the amplitudes of the bulk correlation lengths @xmath191 in the direction of the _ principal axes _ ; the latter are determined by the eigenvectors @xmath178 of @xmath33 which provide the reference axes for the spatial dependence of the anisotropic bulk order - parameter correlation function @xmath192 where @xmath193 .
correspondingly , the matrix @xmath33 determines the anisotropy of the @xmath83 dependence of the fourier transform @xmath194 which is proportional to the observable scattering intensity .
the _ principal _ axes must be distinguished from the _ symmetry _ axes of the bravais lattice .
the latter depend only on the lattice points @xmath195 but not on the couplings @xmath156 .
below an example is given where the principal axes differ from the symmetry axes .
the long - wavelength approximation takes into account only the leading @xmath196 term of @xmath197 . in real space
this is equivalent to using the @xmath1 continuum hamiltonian for the vector field @xmath198 @xmath199\end{aligned}\ ] ] with some cutoff @xmath169 .
various types of anisotropies may result not only from pair interactions on rectangular lattice structures but also from nonrectangular lattice structures , from effective many - body interactions as well as from distortions of the lattice structure , e.g. , due to external shear forces .
the semi - macroscopic continuum model ( [ 2n ] ) is expected to be of general significance in that it provides a complete long - wavelength description of a large class of real systems near criticality whose nonuniversal properties can be condensed into the @xmath200 parameters of the anisotropy matrix @xmath33 , in addition to the nonuniversal parameters @xmath201 .
the quantities @xmath174 depend on all microscopic details ( lattice structure , electronic structure , many - body interactions ) which , in general , are not known a priori for a given material . thus the matrix elements @xmath174 represent phenomenological parameters of a truly nonuniversal character . consequently , physical quantities depending on @xmath174 ( such as @xmath202 , the binder cumulant @xmath203 , and the critical casimir amplitude ) are nonuniversal as well . for an appropriate formulation of the bulk order - parameter correlation function
( see sect .
iii ) and of finite - size scaling functions ( see sect .
vi ) it will be important to employ the reduced anisotropy matrix @xmath204 , @xmath205 with the eigenvalues @xmath206 and with @xmath207 .
the matrix @xmath76 is independent of the kind of variables @xmath138 on the lattice points and independent of the number @xmath11 of components of @xmath138 .
it is well defined , e.g. , also for models with fixed - length spin variables @xmath208 with @xmath209 and for ising models with discrete spin variables @xmath210 instead of the continuous vector variables @xmath138 .
thus the xy and ising hamiltonians @xmath211 and @xmath212 have the same reduced anisotropy matrix @xmath76 and the same reduced eigenvalues @xmath213 as the @xmath1 lattice hamiltonian if these models are defined on the same lattice points @xmath195 and if the couplings @xmath214 are proportional to @xmath156 . as an illustration we consider an @xmath215 simple - cubic lattice model ( fig .
3 ) with a lattice constant @xmath27 and with the following couplings : nearest - neighbor ( nn ) couplings @xmath216 along the cubic symmetry axes , next - nearest - neighbor ( nnn ) couplings @xmath217 only in the @xmath218 , @xmath219 , @xmath220 directions [ but not in the @xmath221 directions ] , and a third - nn coupling @xmath222 only in the diagonal @xmath223 direction ( fig .
3 ) . of the primitive cell ( cube ) of the anisotropic simple - cubic lattice model ( [ 2a ] ) and ( [ 2 g ] ) with @xmath224 .
solid , dashed , and dotted lines indicate the nn couplings @xmath225 , the nnn couplings @xmath226 , and the third - nn coupling @xmath227.,width=302 ] the corresponding anisotropy matrix is obtained from ( [ 2i ] ) as @xmath228 with the diagonal elements @xmath229 , @xmath230 , @xmath231 . for quantitative analytical and numerical studies this model with seven different couplings would , of course , be much too complicated .
we shall present explicit quantitative results only for two nontrivial cases : \(i ) model with _ three - dimensional anisotropy _ : isotropic ferromagnetic nn couplings @xmath232 and three equal anisotropic nnn couplings @xmath233 .
mc simulations for three - dimensional ising models with this type of anisotropy ( with @xmath234 ) have been performed by schulte and drope @xcite and by sumour et al . @xcite .
the corresponding reduced anisotropy matrix ( with @xmath235 ) is @xmath236 which depends only on the single anisotropy parameter @xmath237 the eigenvalues of @xmath33 and @xmath76 are @xmath238 , @xmath239 , and @xmath240 , @xmath241 , respectively .
the eigenvectors @xmath242 defining the principal axes are not parallel to the cubic symmetry axes .
the possible range of @xmath243 consistent with @xmath244 is @xmath245 . in the limit @xmath246 at fixed
@xmath247 corresponding to @xmath248 the model describes a system of variables @xmath138 on decoupled one - dimensional chains with nn interactions @xmath227 . in the previous version @xcite of this model with @xmath234 the range of @xmath243
was restricted to @xmath249 .
a vanishing of @xmath250 and @xmath251 occurs for @xmath252 . at some value @xmath253 near @xmath254 ( corresponding to @xmath255 ) our model
is predicted to have a lifshitz point with a wave - vector instability in the direction of @xmath256 , i.e. , in the @xmath257 direction ( see also sect .
\(ii ) model with _ two - dimensional anisotropy _ : an anisotropic nnn coupling @xmath258 is taken into account only in the @xmath259 planes of the three - dimensional sc lattice whereas all other anisotropic couplings @xmath260 and @xmath227 vanish .
this model is of interest for comparison with the mc data by selke and shchur @xcite for the _ two - dimensional _ anisotropic ising model as will be discussed in sect .
for further recent studies of the anisotropic two - dimensional ising model see also @xcite .
the bulk critical behavior of both models ( i ) and ( ii ) belongs to the same @xmath93 universality class as that of the isotropic model with @xmath261 and @xmath262 provided that @xmath263 . in order to derive an appropriate representation of the anisotropic bulk order - parameter correlation function ( see sect .
iii ) , to develop an appropriate formulation of finite - size perturbation theory ( see sect .
iv ) , and to treat the anisotropic hamiltonian @xmath144 by rg theory ( see sect .
v ) it is necessary to first transform @xmath144 such that the @xmath264 terms of @xmath265 attain an isotropic form .
this is a shear transformation that consists of a rotation and rescaling of lengths in the direction of the principal axes @xcite .
the rotation is provided by the orthogonal matrix @xmath266 with matrix elements
@xmath267 , @xmath268 where @xmath269 denote the cartesian components of the eigenvectors @xmath178 .
the rescaling is provided by the diagonal matrix @xmath270 with diagonal elements @xmath271 . in @xmath83 space
the transformation is @xmath272 such that the @xmath273 term of @xmath274 is brought into an isotropic form with @xmath275 , @xmath276 in real space the transformed lattice points are @xmath277 .
this transformation leaves the scalar product @xmath278 invariant .
thereby the volume of the primitive cells is changed to @xmath279 .
correspondingly the total volume of the transformed system is @xmath280 with the characteristic length @xmath70 .
our transformation is defined such that the values of the couplings @xmath281 on the transformed lattice as well as the temperature variable @xmath141 including the values of @xmath282 and @xmath41 are invariant [ see also eq .
( [ 4zz ] ) below ] . this requires us to perform the additional transformations @xmath283 and @xmath284 in terms of the fourier
transform @xmath285 the transformed lattice hamiltonian reads @xmath286
\hat \varphi'({\mathbf k ' } ) \hat \varphi'({-{\mathbf k ' } } ) } \nonumber \\ & + \;u'_0 v'^{-3 } \sum_{{\mathbf{k'p'}}{\mathbf q ' } } [ \hat \varphi'({\mathbf k ' } ) \hat \varphi'({{\mathbf p ' } } ) ] [ \hat \varphi'({{\mathbf q ' } } ) \hat \varphi'({-{\mathbf k'}-{\mathbf p'}-{\mathbf q ' } } ) ] \nonumber \\ & - h ' \hat \varphi'({\mathbf 0})\;.\end{aligned}\ ] ] of the primitive cell ( parallelepiped ) of the transformed lattice model ( [ 2p ] ) and ( [ 2q ] ) with the volume @xmath287 .
solid , dashed , and dotted lines indicate the nn couplings @xmath225 , the nnn couplings @xmath226 , and the third - nn coupling @xmath227 .
the couplings are the same as in fig . 3 but @xmath275 ( compare figs . 1 and 2 of @xcite ) . , width=302 ]
we illustrate this transformation by the example of the simple - cubic lattice model shown in fig .
3 . the primitive cell with the volume @xmath288 of the transformed system is shown in fig .
4 . it has the shape of a parallelepiped whose lengths and angles are determined such that the transformed second - moment matrix @xmath289 is the unity matrix although there are still the _ same _ nn couplings @xmath225 , nnn couplings @xmath226 , and third - nn coupling @xmath290 as in the simple - cubic lattice model of fig .
3 . working with @xmath291 rather than @xmath144 will be of advantage in the context of _ bulk _ properties and bulk renormalizations in sect .
v. this is not the case for the confined system .
although the @xmath292 term of @xmath293 in ( [ 2o ] ) and ( [ 2p ] ) looks quite simple , namely @xmath294 with a trivial anisotropy matrix @xmath295 , the summations @xmath296 are nontrivial . for concreteness
consider the simplified example ( i ) with the matrix ( [ 32a ] ) and the eigenvectors ( [ 2pp ] ) .
while the @xmath83 vectors of the @xmath297 lattice have the simple form @xmath298 with the integer numbers @xmath299 , , the @xmath82 vectors are considerably more complicated , @xmath300 these @xmath301 vectors reflect the shape and lattice structure of the transformed system .
thus the price paid for transforming @xmath302 to @xmath289 is to work with more complicated @xmath301 vectors .
this example demonstrates that the effect of anisotropy can not be eliminated for _ confined _ systems . in our applications
the summations in finite - size perturbation theory will be performed in the simpler @xmath83 space whereas _ bulk _ integrals ( with infinite cutoff ) are simplified in @xmath82 space . in real space
the hamiltonian @xmath291 reads @xmath303.\end{aligned}\ ] ] by substituting the transformations defined above one easily verifies @xmath304 the measure for the temperature distance from criticality @xmath305 is the same for both @xmath144 and @xmath291 .
defining the free energy density @xmath306 ( divided by @xmath125 ) as @xmath307 @xmath308 \exp \left ( - h ' \right),\end{aligned}\ ] ] one obtains the exact relations [ 2tt ] z ( t , h , l ) = ( ) ^- n n / ( 2 d ) z(t , h , l ) , @xmath309 \ln ( \det { \bf a } ) \ ; .\ ] ] the last term is a bulk contribution , i.e. , independent of @xmath8 .
furthermore it is independent of @xmath41 , i.e. , a non - singular bulk contribution , thus the singular bulk parts of @xmath310 and of @xmath311 as well as the total singular parts of @xmath312 and of @xmath313 are related by @xmath314 @xmath315 the bulk correlation function of the transformed system is @xmath316 where @xmath317 denotes the average with the weight @xmath318 exp @xmath319 .
it is related to @xmath114 by @xmath320 the corresponding relation between the fourier transforms is @xmath321 ( in the arguments of eqs .
( [ 2tt ] ) - ( [ 2x ] ) we have , for simplicity , not indicated explicitly the additional transformations of @xmath322 and of @xmath323 . ) in terms of the transformed field @xmath324 the hamiltonian ( [ 2n ] ) attains the form of the standard isotropic landau - ginzburg - wilson hamiltonian @xmath325\end{aligned}\ ] ] where @xmath326 with a transformed cutoff .
before turning to finite - size theory of anisotropic systems it is necessary to first discuss the bulk critical behavior of anisotropic systems and its relation to that of isotropic systems .
we are not aware of such a discussion in the literature .
it is well known that anisotropic systems and isotropic systems have the same critical exponents ( in a limited range of the anisotropy , see , e.g. , @xcite , and references therein ) . within the @xmath1 theory
this is immediately seen from the relation of dimensionally regularized bulk integrals ( at infinite cutoff ) such as @xmath327 @xmath328 provided that @xmath329 .
we see that the @xmath33 dependence is completely absorbed by the coupling @xmath330 and that the @xmath331 pole term @xmath332 does not depend on the matrix @xmath33 .
this leads to identical field - theoretic functions for anisotropic and isotropic systems ( as functions of the renormalized couplings @xmath333 and @xmath334 , respectively ) and yields the same critical exponents and fixed point value @xmath335 for anisotropic and isotropic systems ( see sect .
( the @xmath336 pole of the integral ( [ 3-a ] ) which has nothing to do with the critical behavior in @xmath337 dimensions can be incorporated in the geometric factor @xmath338 @xcite which is finite in @xmath339 dimensions [ see ( [ 5c1 ] ) below ] . ) for this reason not much attention has been paid to the role played by anisotropy in bulk critical phenomena .
this is not justified , however , in the context of the important feature of two - scale factor universality @xcite .
its validity has been established by the rg theory only for _ isotropic _ systems at @xmath92 with short - range interactions @xcite .
a brief derivation was also given by privman and fisher @xcite and by privman et al .
@xcite using scaling assumptions at @xmath340 .
their ansatz for the order - parameter correlation function , however , is not valid for anisotropic systems since they assumed the existence of a _ single _ bulk correlation length @xmath39 .
recently it has been pointed out that two - scale factor universality is absent in anisotropic systems @xcite and that anisotropy has an effect on several universal bulk amplitude combinations @xcite but no derivation was given . in particular , two important universal amplitude relations derived by privman and fisher @xcite [ eqs .
( [ 3 g ] ) and ( [ 3h ] ) below ] have not been discussed in the context of anisotropic systems .
furthermore , the bulk order - parameter correlation function of anisotropic systems was discussed only for @xmath92 and @xmath341 @xcite . here
we extend this discussion to @xmath342 and @xmath343 and provide an appropriate formulation of the order - parameter correlation function and of the scattering intensity in terms of both the eigenvalues @xmath190 and the reduced eigenvalues @xmath213 of the anisotropic system .
we also present the derivation of several amplitude combinations for anisotropic systems in terms of universal scaling functions .
all of the _ thermodynamic _ bulk relations given in the following subsections a and b remain valid also in the presence of subleading long - range interactions of the type ( [ 2hh ] ) .
this is not the case , however , for bulk correlation functions at @xmath344 , @xmath92 and @xmath345 , @xmath64 in the large - distance regime corresponding to the shaded region in fig .
first we summarize the bulk critical behavior of systems described by the ( asymptotically ) isotropic lattice hamiltonian @xmath291 and the continuum hamiltonian @xmath346 .
near @xmath3 the bulk free energy density can be decomposed uniquely into singular and non - singular parts as @xmath347 where @xmath348 has a regular @xmath41 dependence .
it is well established that @xmath349 has the asymptotic ( small @xmath41 , small @xmath350 ) scaling form below @xmath16 dimensions latexmath:[\[\label{3a } f'_{s , b } ( t , h ' ) = a_1 ' |t|^{d \nu } \ ; w_\pm ( a'_2 h ' because of isotropy it is justified to define a _ single _
( second - moment ) bulk correlation length @xmath356 above and below @xmath3 , respectively . in ( [ 3b ] )
we have assumed sufficiently rapidly decaying correlations , i.e. , general @xmath11 for @xmath357 but @xmath358 for @xmath359 .
we assume , in the asymptotic region @xmath360 and for @xmath361 and small @xmath350 , the asymptotic _ isotropic _ scaling form @xmath362 and @xmath363 .
we use the normalization @xmath364 , thus @xmath365 above @xmath3 .
the length @xmath366 will be needed as a reference length in the formulation of renormalized finite - size theory in sect . v. the corresponding scaling form of the fourier transform @xmath367 of ( [ 3c ] ) is @xmath368 @xmath369 the three amplitudes @xmath370 , @xmath371 , and @xmath366 in eqs .
( [ 3c ] ) - ( [ 3da ] ) are nonuniversal .
the basic content of two - scale factor universality is that all of these amplitudes are universally related to the two thermodynamic amplitudes @xmath354 and @xmath372 .
the relations read @xmath373 @xmath374 @xmath375 in ref .
@xcite the universal constants @xmath376 and @xmath377 were denoted by @xmath378 and @xmath379 . in order to conform with refs . @xcite and to avoid confusion we reserve the notation @xmath378 and @xmath379 for the _ different _ universal constants in eqs .
( [ 3k ] ) and ( [ 3l ] ) below . for the sake of clarity we present the explicit expressions for @xmath380 and @xmath381 in terms of universal scaling functions in appendix a. an equivalent formulation of eq .
( [ 3f ] ) is @xmath382 \ ; = \ ; q_1 ( d , n ) = universal \ ; .\ ] ] the validity of ( [ 3f ] ) and ( [ 3f1 ] ) has been established by the rg theory @xcite . furthermore
, the following amplitude ratios @xmath383 @xmath384 have been proposed @xcite to be universal . the constants @xmath385 and @xmath61 are defined as follows : @xmath386 where @xmath387 , and @xmath388 . the length @xmath389 with the amplitude
@xmath61 is a natural reference length of finite - size theory at @xmath149 , @xmath390 [ see ( [ 6-l ] ) below ] .
@xmath391 is the asymptotic ( small @xmath82 ) amplitude of @xmath392 . alternatively , eq .
( [ 3l ] ) can be formulated as @xmath393 , or , equivalently , @xmath394 where @xmath395 is the asymptotic ( large-@xmath396 ) amplitude of @xmath397 .
the derivation of eqs .
( [ 3k ] ) - ( [ 3l1 ] ) is sketched in app .
a. again , all of the constants on the left - hand sides of ( [ 3k ] ) - ( [ 3l1 ] ) are universally related to @xmath354 and @xmath355 .
below @xmath3 at @xmath398 we have , for @xmath358 , @xmath399 previously the bulk relations ( [ 3f ] ) - ( [ 3e ] ) were expected to be universal for all systems _ within a given universality class _ @xcite .
consistency with the universality of ( [ 3k ] ) and ( [ 3l ] ) was found @xcite for two - dimensional ( square and triangular ) ising lattices and three - dimensional ( sc and bcc ) ising lattices with _ isotropic _
nearest - neighbor interactions ( see also @xcite ) .
all of these systems , however , belong to the subclass of asymptotically isotropic systems with an anisotropy matrix @xmath400 or @xmath401 and with an isotropic scattering intensity @xcite .
also the honeycomb - lattice ising model considered in @xcite belongs to that subclass , with a constant @xmath402 different from that for the triangular lattice or the square lattice . as will be shown in subsect .
b below , eqs .
( [ 3f ] ) and ( [ 3f1 ] ) - ( [ 3e ] ) must be reformulated for anisotropic systems with noncubic anisotropy at @xmath403 whereas ( [ 3 g ] ) and ( [ 3h ] ) remain valid also for anisotropic systems provided that @xmath404 and @xmath371 are transformed appropriately .
it has been shown @xcite that the universal scaling form ( [ 3c ] ) , ( [ 3d ] ) is not valid in the regime @xmath405 above @xmath3 .
the same can be shown for @xmath6 in the regime @xmath406 below @xmath3 .
note that this regime is part of the _ asymptotic _ critical region @xmath407 and @xmath408 corresponding to the shaded area in the @xmath409 - @xmath410 plane in fig .
2 . in this regime ,
corrections to scaling in the sense of wegner @xcite are still negligible .
one must distinguish at least four cases : ( i ) for the @xmath1 lattice model with short - range interactions , the exponential decay above @xmath3 depends explicitly on the lattice spacing @xmath27 via the _ exponential correlation length _
@xmath411 @xcite ; in sect .
x we shall show that it also depends on the bare four - point coupling @xmath28 .
( ii ) for the @xmath1 continuum theory with a _ smooth _ cutoff @xmath169 , the exponential decay depends explicitly on @xmath169 via @xmath412 @xcite ( see also sect .
( iii ) for the @xmath1 continuum theory with a _
cutoff @xmath169 , a nonuniversal oscillatory power - law decay dominates the exponential decay @xcite .
( iv ) in the presence of subleading long - range interactions of the type ( [ 2hh ] ) , the power law @xmath413 @xcite dominates the exponential short - range behavior .
for @xmath414 this has been shown explicitly for the mean spherical model where the asymptotic structure for @xmath415 at @xmath92 is @xcite @xmath416 \nonumber\\\end{aligned}\ ] ] with @xmath417 . in all cases
( i ) - ( iv ) , scaling in the sense of ( [ 3c ] ) , ( [ 3d ] ) and two - scale factor universality are violated in the regime @xmath418 ( shaded area in fig .
2 ) because , in addition to the reference length @xmath419 , the nonuniversal lengths @xmath27 , @xmath420 , @xmath103 , and @xmath105 govern the leading large @xmath421 behavior . [
[ absence - of - two - scale - factor - universality - in - anisotropic - bulk - systems ] ] absence of two - scale factor universality in anisotropic bulk systems ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ now we turn to the anisotropic system . according to ( [ 2oo ] ) , ( [ 2v ] ) and
( [ 3a ] ) , the asymptotic scaling form of @xmath422 is given by ( [ 1a ] ) with the nonuniversal amplitudes @xmath423 in order to represent the order - parameter correlation function ( [ 2l ] ) in an appropriate asymptotic scaling form it is necessary to employ both of the diagonal matrices @xmath424 and @xmath425 with diagonal elements @xmath190 and @xmath426 . using ( [ 2y ] ) , ( [ 2x ] ) , ( [ 3c ] ) and ( [ 3da ] ) we write @xmath114 and @xmath427 as @xmath428
@xmath429 here we identify the spatial variable @xmath118 in the scaling argument of @xmath430 used in fig . 2 as @xmath431 which
, for given @xmath432 , depends on all of the @xmath117 parameters contained in @xmath33 .
while the simple transformations ( [ 3 mm ] ) and ( [ 3n1 ] ) follow immediately from the transformations of @xmath433 , and @xmath434 , the transformation of @xmath435 is less trivial . using eqs .
( [ 3 mm ] ) , ( [ 3nn ] ) , and ( [ 3n1 ] ) we find that the universal amplitude relations ( [ 3 g ] ) and ( [ 3h ] ) of isotropic systems remain valid also for anisotropic systems : @xmath436 @xmath437 with the same universal constants @xmath376 and @xmath377 as in ( [ 3 g ] ) and ( [ 3h ] ) . eq .
( [ 3n ] ) differs from the representation of @xmath114 of @xcite at @xmath92 where , instead of @xmath435 , an overall amplitude @xmath438 was employed .
the latter representation is inappropriate as @xmath439 is not universally related to @xmath22 and @xmath23 .
the relations ( [ 3 g ] ) and ( [ 3n2 ] ) follow from the sum rule ( see app .
a ) @xmath440 less obvious are the relations ( [ 3h ] ) and ( [ 3n3 ] ) .
their derivation is , in fact , based on an additional assumption about the _ unsubtracted _ order - parameter correlation function ( see app .
the physical significance of ( [ 3n3 ] ) is that , at criticality , the bulk correlation function and its fourier transform , if expressed in term of @xmath425 , @xmath441 @xmath442 have an overall amplitude @xmath435 that is universally determined by the _ thermodynamic _ amplitudes @xmath22 and @xmath23 of the bulk free energy @xmath422 . unlike in isotropic systems , however , the spatial dependence of @xmath114 and the @xmath83 dependence of @xmath427 at criticality are governed by the @xmath2 reduced nonuniversal eigenvalues @xmath213 ( with @xmath443 independent parameters ) .
in addition , the knowledge of @xmath444 nonuniversal parameter is needed in order to specify the orthogonal matrix @xmath445 , i.e. , to specify the directions @xmath446 of the principal axes relative to the symmetry axes of the system .
thus @xmath447 nonuniversal parameters are needed at @xmath184 and @xmath183 , and @xmath448 nonuniversal parameters at finite @xmath14 .
these parameters can be measured by elastic - scattering experiments at bulk criticality of anisotropic solids .
now we discuss the temperature and @xmath14 dependence of @xmath114 away from criticality . along the direction @xmath178 of the principal axis @xmath449
the spatial dependence of ( [ 3n ] ) is , for @xmath450 , @xmath451 with @xmath452 and , because of ( [ 2oo ] ) and ( [ 3d ] ) , latexmath:[\ ] ] apart from nonasymptotic corrections that depend on @xmath1390 and vanish for @xmath1391 at fixed finite @xmath1392 .
the bulk integrals in ( [ c3 ] ) follow by differentiation with respect to @xmath681 .
fisher , _ rev .
phys . _ * 46 * , 597 ( 1974 ) .
e. vicari , proceedings of science ( lattice 2007 ) 023 ; arxiv:0709.1014 [ hep - lat ] .
d. stauffer , m. ferer , and m. wortis , _ phys .
_ * 29 * , 345 ( 1972 ) .
fisher , in _ critical phenomena _ , proceedings of the 1970 international school of physics `` enrico fermi '' , course 51 , edited by m.s .
green ( academic , new york , 1971 ) , p. 1 .
v. privman and m.e .
fisher , _ phys .
* b 30 * , 322 ( 1984 ) .
v. privman , _ phys .
* b 38 * , 9261 ( 1988 ) .
v. privman , in _ finite size scaling and numerical simulation of statistical systems _ , edited by v. privman ( world scientific , singapore , 1990 ) , p.1 .
v. privman , a. aharony , and p.c .
hohenberg , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic , new york , 1991 ) , vol .
14 , p. 1 .
x.s . chen and v. dohm ,
j. _ * b 10 * , 687 ( 2000 ) .
chen and v. dohm , _ eur .
j. _ * b 15 * , 283 ( 2000 ) .
chen and v. dohm , _ phys .
* e 66 * , 016102 ( 2002 ) .
chen and v. dohm , _ phys .
* e 70 * , 056136 ( 2004 ) . v. dohm ,
_ j. phys . _ * a 39 * , l 259 ( 2006 ) .
b. widom , _ j. chem
. phys . _ * 41 * , 74 ( 1964 ) ; r.f .
kayser and h.j .
ravech , _ phys .
rev . _ * a 29 * , 1013 ( 1984 ) .
d. dantchev and j. rudnick , _ eur .
j. _ * b 21 * , 251 ( 2001 ) .
d. dantchev , _ eur .
j. _ * b 23 * , 211 ( 2001 ) .
h. chamati and d.m .
dantchev , _ eur .
j. _ * b 26 * , 89 ( 2002 ) .
chen and v. dohm , _ physica _ * b 329 - 333 * , 202 ( 2003 )
. f. wegner , _ phys .
* b 5 * , 4529 ( 1972 ) .
t. schneider , in _ the physics of conventional and unconventional superconductors _ , edited by k.h .
bennemann and j.b .
ketterson ( springer - verlag , berlin ) , in press . ; cond - mat/0204236 .
bruce and r.a .
cowley , _ structural phase transitions _ ( taylor & francis ltd . ,
london , 1981 ) .
salje , _ phase transitions in ferroelastic and co - elastic crystals _ ( cambridge univ .
press , 1990 ) ; _ phys .
rep . _ * 215 * , 49 ( 1992 ) .
liu , _ phys .
lett . _ * 31 * , 459 ( 1973 ) ; d. belitz , t.r .
kirkpatrick , and a. rosch , _ phys .
* b 73 * , 054431 ( 2006 ) ; m. bode , m. heide , k. von bergmann , p. ferriani , s. heinze , g. bihlmayer , a. kubetzka , o. pietzsch , s. blgel , and r. wiesendanger , _ nature _ * 447 * , 190 ( 2007 ) .
katzgraber , m. krner , and a.p .
young , _ phys .
* b 73 * , 224432 ( 2006 ) .
a. drewitz , r. leidl , th.w .
burkhardt , and h.w .
diehl , _ phys .
* 78 * , 1090 ( 1997 ) ; d.b .
abraham and p.j .
upton , _ phys .
lett . _ * 85 * , 2541 ( 2000 ) .
tonchev , _ phys .
* e 75 * , 031110 ( 2007 ) , and references therein .
f. wegner , in _ phase transitions and critical phenomena _ , edited by c. domb and m.s .
green ( academic , new york , 1976 ) , vol .
a. aharony , _ phys .
* b 9 * , 2107 ( 1974 ) .
gerber , _ j. phys .
_ * a 8 * , 67 ( 1975 ) .
hohenberg , a. aharony , b.i .
halperin , and e.d .
siggia , _ phys .
* b 13 * , 2986 ( 1976 ) .
s. singh and r.k .
pathria , _ phys .
* b 31 * , 4483 ( 1985 ) .
j. rudnick , h. guo , and d. jasnow , _
phys . _ * 41 * , 353 ( 1985 ) .
h. guo and d. jasnow , _ phys .
* b 35 * , 1846 ( 1987 ) ; * b 39 * , 753(e ) ( 1989 ) .
mon , _ phys .
lett . _ * 54 * , 2671 ( 1985 ) .
mon , _ phys .
rev . _ * b 39 * , 467 ( 1989 ) .
mon , in _ finite size scaling and numerical simulation of statistical systems _ , edited by v. privman ( world scientific , singapore ) , p. 261 .
a. aharony , in _ phase transitions and critical phenomena _ , edited by c. domb and m.s .
green ( academic , new york , 1976 ) , vol .
6 , p. 357 .
cardy , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic , new york , 1987 ) , vol .
11 , p. 55 .
g. kamieniarz and h.w.j .
blte , _ j. phys .
_ * a 26 * , 201 ( 1993 ) .
a corresponding scaling form @xmath1393 should be valid for the correlation length in an anisotropic system with the shape of an @xmath1394 cylinder considered in @xcite .
k. binder , _
b : condens . matter _ * 43 * , 119 ( 1981 ) .
m.schulte and c. drope , _ int .
_ * c 16 * , 1217 ( 2005 ) .
sumour , d. stauffer , m.m .
shabat , and a.h .
el - astal , _ physica _ * a 368 * , 96 ( 2006 ) . w. selke and l.n .
shchur , _ j. phys .
_ * a 38 * , l 739 ( 2005 ) .
d. dantchev and m. krech , _ phys .
* e 69 * , 046119 ( 2004 ) .
o. vasilyev , a. gambassi , a. maciolek , and s. dietrich , _ europhys .
lett . _ * 80 * , 60009 ( 2007 ) .
m. krech and s. dietrich , _ phys .
rev . _ * a 46 * , 1886 ( 1992 ) ; * a 46 * , 1922 ( 1992 ) ; m. krech , _ the casimir effect in critical systems _
( world scientific , singapore , 1994 ) .
m. krech , _ phys .
* e 56 * , 1642 ( 1997 ) .
r. zandi , j. rudnick , and m. kardar , _ phys .
lett . _ * 93 * , 155302 ( 2004 ) .
d. dantchev , h.w .
diehl , and d. grneberg , _ phys .
rev . _ * e 73 * , 016131 ( 2006 ) .
diehl , d. grneberg , and m.a .
shpot , _ europhys .
lett . _ * 75 * , 241 ( 2006 ) ; d. grneberg and h.w .
diehl , _ phys . rev . _
* b 77 * , 115409 ( 2008 ) .
d. dantchev , f. schlesener , and s. dietrich , _ phys .
* e 76 * , 011121 ( 2007 ) .
r. zandi , a. shackell , j. rudnick , m. kardar , and l.p .
chayes _ phys .
_ * e 76 * , 030601(r ) ( 2007 ) .
z. borjan and p.j .
upton , arxiv:0804.2340 [ cond-mat.stat-mech ] ( 2008 ) .
a. hucht , _ phys .
lett . _ * 99 * , 185301 ( 2007 ) .
williams , _ phys
lett . _ * 92 * , 197003 ( 2004 ) . w. selke , private communication .
w. selke , _ eur .
j. _ * b 51 * , 223 ( 2006 ) .
w. selke , _ j. stat .
* p * 04008 ( 2007 ) .
e. brzin , j.c .
le guillou , j. zinn - justin , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic , new york , 1976 ) , vol . 6 , p. 125 ; g. parisi , _ j. stat .
* 23 * , 49 ( 1980 ) .
g. t hooft and m. veltman , _ nucl .
* b 44 * , 189 ( 1972 ) ; g. t hooft , _ ibid .
_ * 61 * , 455 ( 1973 ) .
v. dohm , _ z. phys . _
* b 60 * , 61 ( 1985 ) ; * b 61 * , 193 ( 1985 ) .
r. schloms and v. dohm , _ nucl .
* b 328 * , 639 ( 1989 ) ; r. schloms and v. dohm , _ phys .
_ * b 42 * , 6142 ( 1990 ) .
e. brzin and j. zinn - justin , _ nucl .
* b 25 [ fs 14 * , 867 ( 1985 ) .
a. esser , v. dohm , m. hermes , and j.s .
wang , _ z. phys . _ * b 97 * , 205 ( 1995 ) .
a. esser , v. dohm , and x.s .
chen , _ physica _ * a 222 * , 355 ( 1995 ) .
chen , v. dohm , and n. schultka , _ phys .
lett . _ * 77 * , 3641 ( 1996 ) .
chen , v. dohm , and talapov , _ physica _ * a 232 * , 375 ( 1996 ) .
chen and v. dohm , _ physica _ * a 235 * , 555 ( 1997 ) ; _ int . j. mod
* b 12 * , 1277 ( 1998 ) .
chen , v. dohm , and d. stauffer _ eur .
j. b _ * 14 * , 699 ( 2000 ) ; x.s .
chen , v. dohm , and a. esser , _
i france _ * 5 * , 205 ( 1995 ) ; w. koch , v. dohm , and d. stauffer , _ phys .
lett . _ * 77 * , 1789 ( 1996 ) ; w. koch and v. dohm , _ phys .
_ * e 58 * , r1180 ( 1998 ) .
m. campostrini , a. pelissetto , p. rossi , and e. vicari , _ phys .
* e 57 * , 184 ( 1998 ) .
zandvliet , _ europhys .
lett . _ * 73 * , 747 ( 2006 ) ; a. nubaumer , e. bittner , and w. janke , _ epl _ * 78 * , 16004 ( 2007 ) ; r. schumann and s. kobe , _ physica _ * 7a 386 * , 187 ( 2007 ) .
bruce , _ j. phys . _
* c 7 * , 2089 ( 1974 ) .
tarko and m.e .
fisher , _ phys .
* b 11 * , 1217 ( 1975 ) .
liu and m.e .
fisher , _ physica _ * a 156 * , 35 ( 1989 ) .
fisher and r.j .
burford , _ phys .
rev . _ * 156 * , 583 ( 1967 ) .
chen and v. dohm , _ phys .
* e 67 * , 056127 ( 2003 ) .
hohenberg and b.i .
halperin , _ rev .
phys . _ * 49 * , 435 ( 1977 ) .
diehl , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic , london , 1986 ) , vol .
10 , p. 76 ; _ int
* b 11 * , 3503 ( 1997 ) .
s. dietrich , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic , new york , 1988 ) , vol .
v. dohm , _ z. phys . _
* b 75 * , 109 ( 1989 ) . m. strsser and v. dohm , _ phys .
* e 67 * , 056115 ( 2003 ) ; _ phys .
* e 70 * , 069902(e ) ( 2004 ) .
p.m. morse , h. feshbach , _ methods of theoretical physics _
( mc graw - hill , new york , 1953 ) .
e. brzin , _ j. phys .
( paris ) _ * 43 * , 15 ( 1982 ) .
h. kleinert and b. van den bossche , _ phys .
* e 63 * , 056113 ( 2001 ) .
larin , m. mnnigmann , m. strsser , and v. dohm , _ phys .
* b 58 * , 3394 ( 1998 ) .
krause , r. schloms , and v. dohm , _ z. phys . _ * b 43 * , 119 ( 1981 ) .
chen and v. dohm , _ physica _ * b 284 * , 45 ( 2000 ) .
chen and v. dohm , _ eur .
j. _ * b 5 * , 529 ( 1998 ) .
the curve in our fig . 8 for @xmath1395 is the same as that in fig . 1 of @xcite .
in ( 55 ) of @xcite , @xmath1396 and @xmath1397 should read @xmath1398 and @xmath1399 , respectively .
blte , e. luijten , and j.r .
heringa , _ j. phys .
_ * a 28 * , 6289 ( 1995 ) .
we have been informed by w. selke that the preliminary simulations of @xcite for @xmath1400 were performed at @xmath1401 and @xmath1402 .
diehl and m. shpot , _ phys .
* b 62 * , 12338 ( 2000 ) .
the interpretation in @xcite that finite - size scaling is restored if the variable @xmath1403 is employed instead of @xmath1404 is not correct .
the reason is that @xmath1160 must be represented in its complete nonuniversal non - asymptotic form , i.e. , as a non - power - law function of @xmath41 with the nonuniversal parameters @xmath1405 and @xmath1406 and different powers @xmath1407 and @xmath1408 according to ( [ viir ] ) , ( [ xd ] ) , and ( [ xh ] ) .
this violates the simple structure of the universal power - law scaling form ( [ 1b ] ) .
furthermore , two different arguments @xmath1409 and @xmath1410 would be needed above and below @xmath3 whose ratio @xmath1411 is nonuniversal because the non - asymptotic corrections must not be neglected in the exponential arguments .
j.f . nicoll and p.c .
albright , _ phys .
* b 34 * , 1991 ( 1986 ) .
d. dantchev , m. krech , and s. dietrich , _ phys .
_ * e 67 * , 066120 ( 2003 ) .
g. parisi , _ statistical field theory _
( addison - wesley , reading , ma , 1992 ) .
wilson and j. kogut , _ phys .
* 12 c * , 75 ( 1974 ) .
see , e.g. , d.j .
amit , _ field theory , the renormalization group , and critical phenomena _
( mcgraw - hill , new york , 1978 ) ; j.j .
binney , n.j .
dowrick , a.j .
fisher , and m.e.j .
newman , _ the theory of critical phenomena _ ( clarendon press , oxford , 1992 ) .
similar anomalies exist in the @xmath1 theory with a sharp cutoff above the upper critical dimension @xmath331 as found in x.s .
chen and v. dohm , _ phys .
* e 63 * , 016113 ( 2001 ) , and references therein .
b. kastening and v. dohm , unpublished in eq .
( 52 ) of ref .
@xcite the term @xmath1412 $ ] should be replaced by @xmath1413 .
i thank b. kastening for pointing out this correction to me . in eq .
( 14 ) of ref .
@xcite the term @xmath1414 should read @xmath1413 . , edited by m. abramowitz and i.a .
stegun ( dover , new york , 1972 ) . | we study spatial anisotropy effects on the bulk and finite - size critical behavior of the o@xmath0 symmetric anisotropic @xmath1 lattice model with periodic boundary conditions in a @xmath2-dimensional hypercubic geometry above , at and below @xmath3 .
the absence of two - scale factor universality is discussed for the bulk order - parameter correlation function , the bulk scattering intensity , and for several universal bulk amplitude relations .
the anisotropy parameters are observable by scattering experiments at @xmath3 . for the confined system , renormalization - group theory within the minimal subtraction scheme at fixed dimension @xmath2 for @xmath4 is employed .
in contrast to the @xmath5 expansion , the fixed-@xmath2 finite - size approach keeps the exponential form of the order - parameter distribution function unexpanded . for the case of cubic symmetry and for @xmath6
our perturbation approach yields excellent agreement with the monte carlo ( mc ) data for the finite - size amplitude of the free energy of the three - dimensional ising model at @xmath3 by mon [ phys .
rev .
lett . * 54 * , 2671 ( 1985 ) ] .
the @xmath7 expansion result is in less good agreement .
below @xmath3 a minimum of the scaling function of the excess free energy is found .
we predict a measurable dependence of this minimum on the anisotropy parameters .
the relative anisotropy effect on the free energy is predicted to be significantly larger than that on the binder cumulant .
our theory agrees quantitatively with the non - monotonic dependence of the binder cumulant on the ferromagnetic next - nearest neighbor ( nnn ) coupling of the two - dimensional ising model found by mc simulations of selke and shchur [ j. phys .
* a 38 * , l739 ( 2005 ) ] .
our theory also predicts a non - monotonic dependence for small values of the _ antiferromagnetic _ nnn coupling and the existence of a lifshitz point at a larger value of this coupling .
the nonuniversal anisotropy effects in the finite - size scaling regime are predicted to satisfy a kind of restricted universality .
the tails of the large-@xmath8 behavior at @xmath9 violate both finite - size scaling and universality even for isotropic systems as they depend on the bare four - point coupling of the @xmath1 theory , on the cutoff procedure , and on subleading long - range interactions . |
soft x - ray emission of hot gas in and around galaxies depends heavily on the properties of metals ( i.e. , elements heavier than helium ) . due to the concentrated enrichment of type ia supernovae ( ia sne ) , the primary iron provider of the universe , hot gas in galactic stellar spheroids ( galactic bulges and elliptical galaxies ) , in particular ,
is expected to have a super - solar iron abundance .
away from such a stellar spheroid , the iron - rich hot gas might be diluted by in - falling low - abundance gas .
thus a monotonically decreasing iron abundance profile as a function of galactic radius is also expected ( e.g. , @xcite ) and is indeed observed on scales of the intragroup and intracluster media . on smaller scales ( within individual stellar spheroids and their immediate vicinity ) , however , x - ray - inferred iron abundances of the hot gas seem to be at odds with the predictions . the abundances are typically sub - solar for low- and intermediate - mass spheroids with log(@xmath0 to about solar for more massive ones ( e.g. , @xcite ) , substantially less than what are expected from the ia sn enrichment .
furthermore , a number of galaxies show a significant iron abundance drop toward their central regions , where the stellar feedback should be the strongest . examples of this abundance drop include m87 @xcite , ngc 4472 , ngc 5846 @xcite , and ngc 5044 @xcite .
the inward drop of the iron abundance and its low value in general are so far not understood , which may indicate that only a small portion of metals produced by ia sne is observed , while the rest is either expelled or in a state that the present x - ray data are not sensitive to .
@xcite have explored the possibility that the iron produced by ia sne may have radiatively cooled and may then have avoided the detection . based on a 1-d modeling of ia sn remnant ( snr ) evolution
, they have demonstrated that the iron - rich ejecta may cool rapidly to low temperatures because of its large radiative emissivity .
this scenario assumes that the cooling occurs within a critical time scale for the ejecta to be mixed microscopically with the ambient medium to an iron abundance below about 100 solar .
this critical time scale is not determined , but should partly depend on the uncertain magnetic field strength and configuration , which can substantially affect the effectiveness of the iron diffusion .
the mixing , of course , can also be dynamic in the violent environment of galactic stellar spheroids .
we have studied the ia snr evolution in a more realist setting .
we have conducted 3-d hydrodynamic simulations of hot gas in a typical galactic stellar spheroid of an intermediate - mass ( i.e. , a galactic halo mass @xmath1 ) in a relatively isolated environment .
the hot gas is expected to be in a galaxy - wide outflow , either supersonic or subsonic , depending largely on the feedback history of the galaxy @xcite .
@xcite have presented simulations for the supersonic outflow case and have discussed several important 3-d effects : ( 1 ) sn ejecta is not well mixed with the ambient medium , and soft x - ray emission arises primarily from relatively low temperature gas shells associated with sn blastwaves ; ( 2 ) the inhomogeneity in the gas temperature substantially alters the emission spectral shape , which can lead to an artificially lower iron abundance ( by a factor of 2 - 3 ) in a spectral fit with a simplistic thermal plasma model ; ( 3 ) the average 1-d and 3-d simulations give substantially different radial temperature profiles ( e.g. , the inner temperature gradient in the 3-d simulation is positive , mimicking a `` cooling flow '' ) ; ( 4 ) the inhomogeneity also enhances the diffuse x - ray luminosity by a factor of a few . in the present work , we focus on the evolution of sn ejecta and its effect on the average iron abundance measurements of the hot gas in both supersonic and subsonic outflow cases .
we use the 3-d simulations to study the dilution of ia sn iron ejecta by the stellar mass loss and the dynamic mixing with the ambient medium as well as the bulk radial outflows at abundance - dependent velocities .
we demonstrate that the dropout of the iron ejecta is unlikely to be important because of the effective dynamic mixing and dilution .
but the differential outflows and their gradual mixing can naturally result in an iron abundance profile qualitatively similar to what is observed .
the non - uniformity of the iron abundance in the flow could also lead to an underestimate of the globally averaged iron abundance in x - ray spectral measurements .
these effects are in addition to the bias that is caused by the broad temperature distribution of the hot gas considered in @xcite , where only supersonic outflows are simulated and a uniform solar abundance is assumed for mock x - ray spectral analysis . the rest of the paper is organized as follows : we first describe in 2 our model inputs and simulation setups and then present in 3 our new results ; we give simple explanations for a few key phenomena in 4 and
summarize the work in 5 . throughout of the paper
, we adopt the solar metal abundances as listed in @xcite ; in particular , our quoted iron mass abundances ( i.e. , , ) are relative to the solar value @xmath2 .
lr present mass loss rate ( @xmath3 ) & 0.075 + present energy input rate ( @xmath4 ) & @xmath5 + bulge stellar mass @xmath6 ( m@xmath7 ) & @xmath8 + bulge scale @xmath9 ( kpc ) & 0.7 + halo mass @xmath10 ( m@xmath7 ) & @xmath11 + [ tab : model ] the present investigation is based primarily on two 3-d simulations of hot gas in galaxy - wide outflows driven by ia sne in galactic stellar spheroids .
the supersonic case corresponds to the bulk outflow that reaches a sonic point ( at radius @xmath12 kpc ) , whereas the subsonic one does not .
the basic setup for these two cases is the same ; the model parameters are listed in table [ tab : model ] . the only difference between the two is the setup of the initial conditions : the supersonic case starts from the 1-d solution , similar to what is presented in @xcite , whereas the subsonic case starts from the final output of a 1-d simulation , which includes the evolution of the circum - galactic medium in a more realistic context of galaxy evolution ( model ve ; @xcite ) . by comparing results from the two cases , we intend to illustrate the dependence of the iron ejecta evolution on the global outflow speed .
both the gas mass and the mechanical energy injections are assumed to follow the same smooth distribution as the stars , which is a hernquist profile @xcite with the characteristic radius of @xmath9 .
the uncertainties in the mass and energy injection rates are a factor of @xmath13 , depending on specific empirical calibrations , potential mass - loading from cool gas , etc .
( e.g. , @xcite and references therein ) .
our chosen energy rate is on the lower side of the uncertainty range , whereas the mass rate is on the higher side .
these choices result in on average a lower gas temperature , a greater soft x - ray emissivity , and a lower iron abundance , which are more close to what are directly inferred from x - ray observations .
the 1-d models also assume an instantaneous mixing between the ejecta and stellar mass injection and hence a uniform iron abundance of 2.7 solar .
we have ignored any feedback from the supermassive black hole expected to be present at the center of a spheroid .
this feedback , likely occurring in bursts with certain preferential directions ( e.g. , in form of jets ) , can occasionally result in significant disturbances in global hot gas distributions , as reflected by the asymmetric x - ray morphologies observed in some elliptical galaxies @xcite .
but , averaged over the time , ia sne are energetically more important for a low- or intermediate - mass spheroid with @xmath14 @xcite .
the simulations are performed with flash @xcite , an eulerian astrophysical hydrodynamics code with the adaptive mesh refinement ( amr ) capability .
the 3-d simulated box , 128 kpc on a side , is centered on the spheroid and has the so - called outflow ( sometimes called zero - gradient ) boundary condition . as in @xcite , only one octant of the box
is simulated at full resolution ( down to about 4 pc ) , while the resolution is degraded by a factor of four in the rest of the grid , except for regions where snrs seeds have just been embedded .
the difference in the resolution between the regions of the same simulation allows us to check the resolution effect ( e.g. , on the x - ray luminosity ) , which appears to be rather small . in the 3-d simulations ,
the ia sn injection follows the star distribution only statistically @xcite . for each `` randomly '' generated sn ,
its remnant seed is extracted from a library of 1-d simulated radial density , temperature , and velocity profiles and is scaled appropriately before being planted into the 3-d simulations @xcite .
we evolve the two simulations till the averaged radial profiles of the density , pressure , and abundance have reached a statistically steady state inside the galactic radius of 10 kpc . for any quantitative analysis , we will use only the data within this radius to avoid any potential artifacts introduced by the chosen boundary condition , in the high - resolution octant , and during the final stabled period . for the calculation of the x - ray emission and the cooling of the hot gas
, we use the mekal plasma model @xcite , extracted from the xspec package .
the cooling function of the gas is divided into the iron - free part @xmath15 and the pure iron rate @xmath16 , which are shown in fig .
[ f : coolfuns ] ( see also @xcite ) .
[ f : global ] shows a sample snapshot of our simulated subsonic outflow , with perspectives on the density , temperature , pressure , and iron mass fraction distributions across one plane cutting through the simulated box .
the highly structured distributions of the hot gas are similar to those seen in the supersonic case ( 1 ; @xcite ) , although features in the subsonic case tend to be more compact because of its higher average gas density .
time - sequenced close - ups of a region near the center of the subsonic outflow is presented in fig .
[ f : ironejecta ] , demonstrating the evolution of several snrs as well as the bulk motion .
the feature labeled with `` a '' in the upper left panel , for example , represents the iron core of a newly embedded snr .
as indicated in subsequent panels , the iron core is gradually diluted by the local mass injection from evolved stars . in the mean time , the shapes of the core as well as the outer shell are strongly affected by the impacts of nearby snrs ( lower row ) . because of both the dilution and the mixing with the surrounding medium , the iron abundance of the core is reduced to about 10 solar after about 1.5myr ( right panel ) .
this time scale is much too short to allow for significant cooling of the ejecta .
[ f : zfegas ] shows various averaged number density , temperature , and pressure profiles extracted from the two simulations . as expected , the profiles for the subsonic outflow are much flatter than those for the supersonic ones .
the difference in the density , for example , is about one order of magnitude at large radii .
the profiles of the hot gas components in four iron abundance ranges show interesting differences and similarities . in general , the higher the abundance of a gas component is , the smaller the number density or the higher the temperature is . except for the component with the highest abundance ,
the other components all have similar pressure profiles .
this means that the bulk of the gas is approximately in a pressure equilibrium , although the temperature and density distributions may still be highly structured .
similar phenomena are also reported for simulations of the sn dominated interstellar medium in the galaxy @xcite and in starburst regions @xcite .
[ f : zfegas_vr ] demonstrates that gas with higher iron abundance generally has a larger net outward radial velocity .
the velocity of the gas component with @xmath17 fluctuates greatly , because it represents young snrs which may or may not have been fully accelerated by the buoyancy ( see 4.2 ) .
the gas component with @xmath18 , representing fully - accelerated and evolved snrs with moderate dilution , has the largest velocity .
[ f : ironfractions ] shows the iron mass fractions of the individual gas components in the different abundance ranges .
the fractions in iron - rich components decrease considerably faster with increasing radius in the subsonic case than in the supersonic one .
this is because the outflow velocity is much slower in the former case , leading to more dilution and mixing of the iron ejecta before moving to large galactic radii .
[ f : zfegradient ] presents the radial profiles of the averaged iron abundance , weighted by three means . the mass - weighted iron abundance , defined as @xmath19 where @xmath20 and @xmath21 is the and mass density in a simulated sub - volume @xmath22 , while the sum is over an interested region ( e.g. , a shell ) .
@xmath23 is thus directly proportional to the iron mass in the region .
similarly , we can define the emission measure - weighted iron abundance as @xmath24 where the electron number density @xmath25 is related to the hydrogen number density @xmath26 ( hence @xmath21 ) and the @xmath27 in the same way as described by @xcite .
@xmath28 probably represents the simplest way to estimate the equivalent abundance that can be compared with an actual measurement based on x - ray emission .
an alternative model proxy to the x - ray emission - inferred value is the emission - weighted iron abundance , @xmath29 in our simulations , the @xmath30 might be overestimated at large radii because sn iron cores are not adequately resolved ( i.e. , the spatial resolution in regions of @xmath31kpc is reduced by a factor of four ) , which numerically mixes the iron ejecta over a larger volume and effectively results in a higher @xmath32 ( hence stronger emission ) .
[ f : zfegradient ] shows that the radial trends of these differently weighted iron abundances are qualitatively the same : a positive gradient with a minimum value at the center !
the lower limit of the abundance is solar , assumed for the stellar mass loss .
here we attempt to give a physical account of several key phenomena observed in the simulations presented above .
we find no evidence for significant cooling of the hot gas , especially the iron ejecta , in our simulated region . to have a better understanding of this issue , it is instructive to first consider a simple 1-d simulation of a typical snr in a hot gas environment , as in @xcite , but in the context of an intermediate - mass spheroid considered here @xcite . as mentioned in 2 , the 1-d simulated snr is also used to generate the seeds for the 3-d simulations .
[ f : snejecta ] demonstrates the simulated snr evolution in a uniform ambient medium of temperature @xmath33k and electron density @xmath34 , appropriate for a region near the center of the spheroid .
the core that encloses 0.7@xmath35 reaches a size of 15pc at the age of @xmath36 years and increases only slightly after that .
this is also roughly the time scale for the ejecta to be fully thermalized by the converging reverse shock .
then the core turns into a tenuous hot bubble , with a very low density and high temperature . outside
the iron core is an envelope of the rest of the sn ejecta ( assumed to be 0.7m@xmath7 ) .
the swept - up material can also be heated to more than @xmath37k only initially by the strong forward blastwave ; the averaged temperature of the swept - up material is considerably lower because of the quick weakening of the expanding blastwave and the adiabatic cooling of the snr . the radiative cooling time scale ( fig .
[ f : snr_cooltime ] ) for such an snr is much too long to be important , compared to the outflow time of the hot gas from a spheroid ( @xmath38 yrs ) . in particular , the x - ray emission from the ejecta is largely undetectable .
we now consider additional processes that an snr should experience in a spheroid .
one is the dilution due to mass injection from evolved stars ( or evaporation of cloud - lets ) .
as the dilution takes place , @xmath32 increases while @xmath16 drops .
the net result is that the cooling rate peaks when the iron abundance drops to @xmath39 solar @xcite .
further dilution tends to decrease the cooling rate .
take the above 1-d snr as an example .
it takes about 2.7myr for the snr core to accumulate an additional 5.4 @xmath40 , which dilutes the ejecta to an iron abundance of 100 solar , the value to avoid rapid radiative cooling @xcite . in comparison ,
the time for the snr to flow this radius to 2kpc is about 5myr .
thus the dilution is important , although the core should still have a high iron abundance when being transported out of the spheroid .
another process is the mixing of the iron ejecta with the ambient medium .
the density of bulk of the snr interior is always smaller than the surrounding medium , as demonstrated in the 1-d modeling , and is likely subjected to the rayleigh - taylor instability ( e.g. , @xcite ) .
the passage of blastwaves produced by nearby sne also tends to accelerate the mixing process , especially in the inner spheroid region where the stellar density is very high .
our euler - based hydrodynamical simulations account for this cooling rate effect as well as the dilution and mixing processes .
we conclude that the radiative cooling of the iron ejecta is unlikely to be important in the type of spheroids we have considered .
but we can not rule out the possibility that ia sn ejecta may cool and drop off from hot gas in a more quiescent environment . in a giant elliptical galaxy considered by @xcite
, hot gas density can be rather high , substantially increasing the cooling , and the bulk motion may not be significant .
unfortunately , a 3-d simulation of the hot gas in such an environment is still very difficult to conduct , because of both the large dynamic range over a big volume and the difficulty to handle the boundary condition .
one of our findings is that the differential radial outflows play an important role in determining the iron abundance profile of the hot gas .
the relatively large outflow speed of the iron - rich gas is apparently caused by the presence of the buoyancy force . a reverse - shock heated sn ejecta bubble , with a density that is one to two orders of magnitude less than that of the ambient medium , is subject to a buoyancy - driven acceleration as @xmath41 where @xmath42 is the gravitational acceleration while @xmath43 and @xmath44 are the average densities of the bubble and its ambient medium .
for our adopted model spheroid , @xmath45 at 0.5kpc , for example , and is larger near the galactic center . assuming a typical @xmath46 , the hot bubble can be accelerated over a period of 0.2myr to about 150 @xmath47 , a typical velocity difference between the iron - rich gas and the bulk of the outflow .
this velocity is also about the equilibrium value that can be reached when the buoyancy force is balanced by a drag : @xmath48 where @xmath49 , @xmath50 , and @xmath51 are the maximum cross - section area , mass , and velocity of the hot bubble , respectively .
the coefficient @xmath52 is suggested to be @xmath53 ( e.g. , @xcite ) .
the above crude estimates are consistent with the results from our detailed examination of individual hot bubbles in the simulations , which naturally account for such complications as the changing sizes and shapes as well as the decreasing density contrasts relative to the ambient medium .
we find that the positive radial gradient of the iron abundance shown in fig .
[ f : zfegradient ] is due largely to the differential outward velocities of the gas ( fig .
[ f : zfegas_vr ] ) .
the relative fast outward motion of the iron - rich gas naturally leads to a reduced iron mass fraction in the bulk outflow . for an illustration ,
suppose that a model hot gas contains only two components : the first one ( sub - scripted with 1 hereafter ) is iron - free , whereas the second one ( sub - scripted with 2 ) is pure iron .
these two components have outward mass fluxes at radius @xmath54 : @xmath55 where @xmath56 and @xmath57 represent the average mass fraction and outward velocity of each component , while @xmath58 is the total mass density of the gas .
the iron abundance is then @xmath59 where @xmath60 is adopted because the total iron mass is no more than a few percent of the total gas mass .
the ratio @xmath61/@xmath62 is the same as the iron mass fraction of the injection from the stellar mass loss , independent of the spatial lumpiness in the iron distribution .
[ eq : feabundef ] thus illustrates that the faster the iron - rich gas component moves relative to the rest of the outflow , the lower its mean iron abundance should be .
we have examined the evolution of ia sn iron ejecta in the hot gas undergoing outflows from a typical intermediate - mass galactic stellar spheroid .
3-d hydrodynamic simulations of this hot gas are conducted for both supersonic and subsonic cases .
these simulations show no evidence for the iron ejecta to significantly cool and drop out of the hot gas . the largely inhomogeneous enrichment and heating , inherited from sporadic ia sne , can significantly affect the x - ray measurement of the mean iron abundance of the hot gas in two ways : 1 .
hot and low - density iron bubbles , generated by ia sne , tend to move outward substantially faster than the ambient medium .
this differential motion , produced mainly in the inner galactic region under the strong buoyancy force , effectively reduces the mean iron abundance of the hot gas .
the reduction decreases with the increasing galactic radius as the iron bubbles are gradually mixed with the ambient medium and are diluted by fresh mass injection from evolved stars .
because of their high temperature and low density , the iron - rich bubbles hardly radiate , contributing little to the x - ray emission of the hot gas @xcite .
the bulk of the emission arises from the ambient medium swept up by snr forward blastwaves , which has higher densities and lower temperatures .
this iron segregation as well as the density and temperature distributions , most apparent again in the inner galactic region , can lead to a significant underestimate of the actual mean iron abundance of the hot gas , when characterized with a simplistic plasma model ( e.g. , with one- or two - temperature components ) .
these two effects together provide a natural explanation for the low iron abundance and its positive radial gradient , as inferred from existing x - ray measurements of hot gas in various galactic stellar spheroids .
this work together with those presented in @xcite demonstrates that x - ray observations of hot gas in and around spheroids can potentially be used as a unique tool to probe the dynamics of the feedback from evolved stars as well as smbhs , which represents a very poorly understood part of galaxy formation and evolution theories . to reach this goal ,
more work is clearly needed to understand the hot gas , both theoretically and observationally .
particularly useful will be spatially - resolved high - resolution spectroscopy of the hot gas , as could be provided by the future international x - ray observatory , as well as careful analysis of existing high - quality x - ray imaging data . but
a large parameter space of the modeling also needs to be explored to facilitate quantitative comparisons with the existing and future observations .
we thank the referee for useful comments . the software used in this work was in part developed by the doe - supported asc / alliance center for astrophysical thermonuclear flashes at the university of chicago .
simulations were performed at the pittsburgh supercomputing center supported by the nsf .
we also acknowledge the support by nasa through grants nnx06ai18 g and tm7 - 8005x ( via sao / cxc ) .
anders e. , & grevesse n. , 1989 , geochimica et cosmochimica acta , 53 , 197 borgani s. , fabian d. , tornatore l. , schindler s. , dolag k. , diaferio a. , 2008 , space sci .
review , 134 , 379 brighenti fabrizio , & mathews william g. , 2005 , apj , 630 , 864 buote d. a. , 2000a , apj , 539 , 172 buote d. a. , 2000b , mnras , 311 , 176 buote d. a. , lewis a. d. , brighenti f. , & mathews w. g. 2003 , apj , 595 , 151 cioffi , d. f. , mckee , c. f. , & bertschinger , e. 1988 , apj , 334 , 252 david l. p. , jones c. , forman w. , vargas i. m. , nulsen p. , 2006 ,
apj , 653 , 207 de avillez m. a. , & breitschwerdt d. , 2005 , a&a , 436 , 585 diehl , s , & statler , t. s. 2008 , apj , 680 , 897 fabian d. , tornatore l. , borgani s. , saro a. , dolag k. , 2008 , mnras , 386 , 1265 fryxell b. , et al . , 2000 , apjs , 131 , 273 gastaldello f. , & molendi s. 2002 , apj , 572 , 160 hernquist l. , 1990 , apj , 356 , 359 humphrey p. j. , & buote d. a. , 2006 , apj , 639 , 136 ji j. , irwin j. a. , athey a. , bregman j. n. , & lloyd - davies e. j. 2009 , apj , 696 , 2252 jones t. w. , ryu d. , & tregillis i. l. , 1996 , apj , 473 , 365 joung m. k. , mac low m .- m . , bryan g. l. , 2008 , apj , 704 , 137 joung m. k. , & mac low m .- m . , 2006 , apj , 653 , 1266 liedahl d. a. , osterheld a. l. , & goldstein w. h. , 1995 , apj , 438 , l115 loewenstein m. , mathews w. g. , 1987 , apj , 319 , 614 mac low m .- m . , balsara d. s. , kim j. , de avillez m. a. , 2005 , apj , 626 , 864 mathews , w. g. , & brighenti , f. 2003 , ara&a , 41 , 191 mewe r. , gronenschild e. h. b. m. , & van den oord g. h. j. 1985 , a&as , 62 , 197 raley e. a. , shelton r. l. , plewa t. , 2007 , apj , 661 , 222 sedov l. i 1959 , similarity and dimensional methods in mechanics , translation from 4th russian edition , academic press new york and london strickland d. k. , & stevens i. r. , 1998 , mnras , 297 , 747 sutherland r. s. , dopita m. a. , 1993 , apjs , 88 , 253 tang s. , & wang q. d. 2005 , apj , 628 , 205 tang , s , & wang , q. d. 2009 , mnras , 397 , 2106 tang s. , wang q. d. , lu y. , mo h. j. , 2009a , mnras , 392 , 77 tang s. , wang q. d. , mac low m .-
joung m. r. 2009b , mnras , 398 , 1468 trinchieri , g. , et al .
2008 , apj , 688 , 1000 wang c. , & chevalier r. a. , 2001 , apj , 549 , 1119 | type ia supernovae ( ia sne ) provide a rich source of iron for hot gas in galactic stellar spheroids .
however , the expected super - solar iron abundance of the hot gas is not observed . instead
, x - ray observations often show decreasing iron abundance toward galactic central regions , where the ia sn enrichment is expected to be the highest .
we examine the cause of this missing iron problem by studying the enrichment process and its effect on x - ray abundance measurements of the hot gas .
the evolution of ia sn iron ejecta is simulated in the context of galaxy - wide hot gas outflows , in both supersonic and subsonic cases , as may be expected for hot gas in galactic bulges or elliptical galaxies of intermediate masses .
sn reverse - shock heated iron ejecta is typically found to have a very high temperature and low density , hence producing little x - ray emission .
such hot ejecta , driven by its large buoyancy , can quickly reach a substantially higher outward velocity than the ambient medium , which is dominated by mass loss from evolved stars .
the ejecta is gradually and dynamically mixed with the medium at large galactic radii .
the ejecta is also slowly diluted and cooled by _
insitu _
mass injection from evolved stars .
these processes together naturally result in the observed positive gradient in the average radial iron abundance distribution of the hot gas , even if mass - weighted .
this trend is in addition to the x - ray measurement bias that tends to underestimate the iron abundance for the hot gas with a temperature distribution . |
random numbers are important for an array of applications from encryption and authentication systems @xcite , to monte carlo simulations for molecular dynamics , nuclear reactors , and others @xcite . as a result , a variety of classical methods ( computational pseudo - random number generators , sampling stochastic physical processes , etc . ) to generate random number sequences have been developed .
an attendant host of tests to certify that a given data sequence is `` random '' has been also been created @xcite . while pseudo - random numbers are useful for many of these applications , including simulations and encryption with suitably high quality sources , their inherent determinism means that any encryption or authentication scheme is in principle breakable with sufficient computational power .
this principle applies to any deterministic system , including processes described by classical physics . on the other hand ,
the only nondeterministic physical theory with experimentally accessible applications is quantum mechanics @xcite .
the additional security provided by non determinism is a requirement for quantum key distribution , for instance , whose security proofs often rely on the concept of true , nondeterministic randomness in order to guarantee successful secret key sharing @xcite .
thus , a wide array of so - called quantum random number generators ( qrng ) have been developed . from radioactive decay @xcite to quantum optical techniques @xcite , a host of methods involving photon arrival time @xcite and vacuum noise measurements @xcite have been demonstrated . despite the prevalence of qrngs and their acknowledged need , many implementations use extractors ( such as hashes ) to remove large amounts of bias computationally , exposing a potential weakness in their physical implementations .
for instance , if an adversary is able to computationally reverse the extractor function that a given qrng implements in order to achieve random number uniformity and the underlying ( `` physical '' ) distribution is strongly biased then he or she will have a best - guess strategy against the qrng device .
therefore , one s ability to detect and remove bias before applying an extractor function improves the qrng s security .
one of the major sources of bias in qrngs , aside from environmental noise , is the lack of knowledge of precise values of the qrng s physical parameters .
the best one may do is to estimate the parameters statistically .
but because the estimates are statistical they are intrinsically noisy , and thus assigning a single value to a parameter can lead to errors and bias .
nevertheless , parameter estimator errors are usually ignored in qrng design and simple point estimators are used . here ,
we show that using point estimators may introduce possible binning bias .
we argue that using a bayesian statistical inference method removes this type of bias and propose a binning scheme that extracts the optimum number of bits possible for a given entropy from a given physical random number distribution .
when used as a diagnostic for qrngs in combination with maximum likelihood estimators ( mle ) , uniform distributions can be generated from sources of quantum randomness .
using bayesian hypothesis updating techniques , our scheme allows for a test of the quantum model that produced a given set of numbers , potentially allowing for a fast , on - line quantum test of randomness .
this technique has applications to high bit rate qrngs which need testing and verification to ensure the device remains bias - free during use .
let @xmath0 be a continuous random variable with probability density function ( _ pdf _ ) @xmath1 , where @xmath2 is a fixed- ( but unknown- ) value parameter . the particular form and parametric dependence of @xmath1
is determined by the experimental setup at hand .
our goal in this section is to introduce a typical problem of physical random number generation that can be formulated as follows : provided @xmath3 independent samples of @xmath0 , @xmath4 are measured in an experiment , convert , if possible , each measurement outcome @xmath5 into a discrete random variable @xmath6 with the probability mass function ( _ pmf _ ) @xmath7 and corresponding domain @xmath8 .
a uniform distribution @xmath9 is often important in applications and here we will also concentrate on the case of @xmath10 . then the problem essentially reduces to constructing a surjection @xmath11 from the set @xmath12 onto the set @xmath13 . traditionally , the problem is solved by dividing the domain of @xmath14 , @xmath15 , into @xmath16 mutually non - intersecting bins such that @xmath17@xcite . when bins are selected such that the probability of the random variable @xmath0 to fall into the @xmath18-th bin
@xmath19 is @xmath20 then the surjection @xmath21 can be constructed by following a simple rule : if a measurement result @xmath22 for some @xmath18 then we assign @xmath23 . of course
this mapping works only if the value of the model parameter @xmath2 is known .
since it is usually not the case in the majority of experimental situations , the first order of business is to find a good estimate of the value of @xmath2 . in many cases ,
the number of possible ways to construct an estimator that provides an unbiased estimate of @xmath2 is infinite @xcite .
moreover , it is not always possible to find an estimator that has minimal uncertainty , and often one is forced to choose one from a set of almost optimal candidates . in practice
the maximum likelihood estimator ( mle ) is a common choice .
given a set @xmath24 of independent samples of @xmath0 , @xmath25 , we can introduce the _ likelihood _ function , @xmath26 the likelihood @xmath27 indicates which values of @xmath2 are more likely given measurement data @xmath24 .
we can also compute , at least numerically , the value of @xmath2 that maximizes @xmath27 , provided the likelihood function is convex .
the resulting estimator is mle , i.e. @xmath28 .
using @xmath29 as the `` true '' parameter value for binning purposes in eq.([eq : binprobability1 ] ) might at first appear as a reasonable choice , and this approach is a mainstay in qrng design .
but what happens if instead of @xmath29 one uses some other estimate @xmath30 that differs from @xmath29 only slightly in the value of the likelihood , i.e. @xmath31 ?
choosing @xmath30 over @xmath29 will have an effect on the size of bins @xmath19 generated via eq.([eq : binprobability1 ] ) .
we illustrate this situation in fig .
[ fig : binningexample1 ] , where the random variable @xmath0 follows gamma distribution @xmath32 ( i.e. @xmath33 ) and we are interested in converting each measurement outcome @xmath34 into a uniformly distributed discrete random variable k that can take on values \{0,1,2,3}. we fit the same measurement data using two slightly different values of the parameter @xmath2 .
the red solid line represents the fit with @xmath35 and the blue solid line has @xmath36 .
the vertical dashed blue ( red ) lines represent bins calculated using eq.([eq : binprobability ] ) with @xmath37 and @xmath38(@xmath39 ) .
the green circle is a particular measurement outcome @xmath40 that we would like to assign a discrete value @xmath41 to . according to our previous discussion ,
@xmath42 and @xmath43 if we use values @xmath44 and @xmath45 respectively .
now imagine that @xmath45 and @xmath44 are such that the likelihood function does not provide a reliable differentiation between them , i.e. @xmath46 .
which value of @xmath41 , if any , should we then adopt ?
there are four possible options : * choose @xmath39 when @xmath45 is the true estimate ( @xmath47 ) .
* choose @xmath38 when @xmath44 is the true estimate ( @xmath48 ) .
* choose @xmath39 when @xmath44 is the true estimate ( @xmath47 ) .
* choose @xmath38 when @xmath45 is the true estimate ( @xmath48 ) .
the first two choices are trivial since they obviously result in a uniform pmf @xmath49 . the last two choices , however , generate a bias that distorts the uniformity of @xmath50 . to see that we calculate the probability of @xmath0 occupying the @xmath18-th bin provided that @xmath39 is chosen when @xmath44 is the true estimate , @xmath51^{\frac{\theta_{1}}{\theta_{2 } } } - [ \frac{n - i-1}{n}]^{\frac{\theta_{1}}{\theta_{2}}},\label{eq : binprobunder}\end{aligned}\ ] ] where @xmath37 , @xmath52 , and @xmath53 .
we notice that , by definition , @xmath54 .
similarly , if we choose @xmath38 when @xmath45 is the true estimate then the @xmath18-th bin probability @xmath55 reads , @xmath56^{\frac{\theta_{2}}{\theta_{1 } } } - [ \frac{n - i-1}{n}]^{\frac{\theta_{2}}{\theta_{1 } } } , \label{eq : binprobover}\end{aligned}\ ] ] where @xmath57 . finally , the plot of pmfs @xmath50 and @xmath58 in fig .
[ fig : binprobability ] , calculated using eqs.([eq : binprobunder ] ) and ( [ eq : binprobover ] ) respectively for @xmath37 , illustrates the effect of parameter under(over)-estimation on the uniformity of the random numbers generated using the continuous distribution binning method .
the horizontal axis represents the bin number @xmath41 where a measurement outcome @xmath40 is placed as the result of binning .
the vertical axis is the probability for different values of @xmath41 to occur .
ideally , if the value of @xmath2 was known exactly , the probability of @xmath59 or @xmath60 would be the same at @xmath61 .
this situation is represented by the solid blue line .
when the value of @xmath2 is overestimated , @xmath58 the corresponding pmf in eq.([eq : binprobover ] ) depicted by green crosses , exhibits a bias towards placing measurement outcomes into the first two bins . in a similar fashion ,
@xmath50 in eq.([eq : binprobunder ] ) , represented by red circles , corresponds to the situation when the parameter @xmath2 is underestimated and demonstrates bias towards @xmath48 . to quantify the amount of introduced bias we compute values of kullback - leibler ( kl ) divergence @xmath62 and @xmath63 between the bin pmf @xmath58 ( @xmath50 ) and the ideal uniform pmf @xmath64 respectively . by definition ,
kl divergence measures the information lost when the uniform pmf @xmath65 is used to approximate @xmath58 or @xmath50 .
we find that @xmath66 bits and @xmath67 bits .
this example shows that discrete random number generation procedures relying on binning a continuous probability distribution with a parametric dependence potentially introduces bias .
this happens because the point parameter estimation approach is prone to over(under)-estimating the true value of the parameter .
hence , the question arises : is there a binning method that does not introduce bias ?
the short answer is yes , and such a method will be introduced in section [ sec : bayesianinference ] . in the next section , a slightly different approach to binning is shown in order to motivate the discussion .
a measurement outcome @xmath34 does not depend on the value of the pdf parameter @xmath2
. however , the probability of the outcome does .
as we have already seen , this means that the size of the bins also depends on @xmath2 , which makes binning procedure problematic .
the reverse situation would be more practical , in which the bin size is fixed ( independent of @xmath2 ) but the measurement outcome depends on the pdf parameter .
of course , this does not remedy the problem of bias discussed earlier , but it will be useful in formulating a solution in the next section . for a given fixed value @xmath2 , the probability @xmath68 that the continuous random variable @xmath0 is less than @xmath69 reads , @xmath70 where we have assumed that @xmath71 .
by definition @xmath72 $ ] and @xmath73 can be interpreted as a uniform continuous random variable on the @xmath74 $ ] interval provided @xmath68 is a continuous function of @xmath69 .
the proof is straightforward and can be found elsewhere @xcite . on the other hand , if the value of @xmath69 is fixed , e.g. @xmath75 , and the value of @xmath2 is unknown then @xmath76 is clearly a function of @xmath2 with the range @xmath74 $ ] .
if we divide the @xmath74 $ ] interval into @xmath16 uniform bins , each of the size @xmath77 , then for every measurement outcome @xmath34 a discrete random number @xmath78 can be generated by finding @xmath41 such that @xmath79 .
this is exactly what we were looking for . by replacing the random variable @xmath0 with @xmath80 using the integral transform in eq.([eq : inttrans ] )
we switched from having bins that explicitly depended on the model parameter @xmath2 to having constant bin size .
the parametric dependence is now shifted to the random variable that we bin , i.e. @xmath81 , and now we need to figure out a way to assign a value to @xmath81 which does not create bias .
we could try to fix the value of @xmath81 by using an estimate of @xmath2 ( e.g. mle ) as was done previously in section [ sec : directbinning ] . however , this approach is inherently flawed because any finite data sample estimator though it can be very close to the true parameter value will over(under)-estimate the true parameter value . however , the concept of likelihood , or , more precisely , the concept of treating the distribution parameter @xmath2 as an unknown ( but not random ) variable given a set of measurements @xmath25 can be inverted using bayesian inference to compute the probability @xmath81 of occupying a given bin @xmath18 .
indeed , the bayesian approach treats @xmath2 as a quantity whose variation is described by a probability distribution @xmath82 usually referred to as the _
prior_. the prior is a subjective distribution determined by experimenter s personal beliefs and knowledge about the system of interest prior to any observations on the system .
once @xmath82 is formulated , an observation on the system is made .
the prior is then updated with the result of the observation using bayes rule and the next measurement is taken with the updated prior , often called _ posterior _ , as the new prior . if the sampling distribution , i.e. the distribution we draw measurement outcomes from , is @xmath1 ( the pdf to observe @xmath83 as a result of our measurement , given the parameter value @xmath2 ) and the measurement result is @xmath83 then the posterior distribution is given by @xmath84 where @xmath85 is the marginal distribution of @xmath0 : @xmath86 the posterior distribution can be subjectively interpreted ( since it does depend on the choice of the prior ) as a conditional distribution ( conditioned on the observed sample ) for the parameter @xmath2 . on the other hand , we know that @xmath81 is a function of @xmath2 given the measurement outcome @xmath83 .
therefore , @xmath81 can also be interpreted as a random variable on @xmath74 $ ] with distribution function @xmath87 that can be computed using @xmath88 , @xmath89 where the plus ( minus ) sign is taken when @xmath81 is an increasing ( decreasing ) function of @xmath2 , @xmath80 is assumed to be continuous , and @xmath90 has a continuous first derivative .
now we are fully equipped to calculate the probability that a measurement outcome @xmath34 converts into an integer @xmath41 ( @xmath91 $ ] ) .
it is equivalent to the probability that the random variable @xmath92 falls into the interval @xmath93 given by , @xmath94 this means that we now can assign a bin to a measurement outcome using a simple acceptance / rejection test : we accept @xmath40 into the @xmath41-th bin if @xmath95 and reject @xmath40 in the @xmath41-th bin otherwise . here
@xmath96 is the user - defined acceptance probability .
the binning bias can be completely eliminated by setting the value of @xmath96 high ( @xmath97 ) .
this means that only the measurement outcomes that have more than 95% of their distribution function @xmath98 localized within a certain bin will be accepted and converted into a discrete random number .
all other measurements will be rejected . on the other hand ,
if @xmath96 is set too low , say , @xmath99 then less measurements will be rejected . however
, this may lead to conflicting situations when a measurement outcome could be placed into two or more different bins which , in turn , may lead to binning bias .
let us consider an example depicted in figure [ fig : binvsparam ] where two distribution functions @xmath100 ( red solid line ) and @xmath101 ( green solid line ) for two independent samples @xmath102 and @xmath103 are plotted .
we are interested in converting each measurement outcome @xmath69 into an integer value @xmath104 . using our acceptance / rejection test with @xmath105
we conclude that @xmath103 is an acceptable measurement that can be converted to @xmath43 . on contrary , @xmath103 will be rejected and no integer value will be assigned to it .
we finally summarize our approach to qrng data processing as the following 5 step algorithm : 1 .
run qrng and collect @xmath3 independent samples @xmath106 from the distribution @xmath1 defined by the qrng .
2 . construct a prior @xmath82 for all possible values of @xmath2 .
3 . update the prior @xmath3 times using the bayes rule eq.([eq : bayesrule ] ) .
compute the posterior @xmath107 .
4 . for each measurement outcome @xmath40 compute the correspondent distribution @xmath98 using eq.([eq : gofu ] ) and eq.([eq : inttrans ] ) . set the acceptance probability value @xmath108 5 .
use the proposed acceptance / rejection test to convert the measured sequence @xmath24 into integer values .
it is worth mentioning that alternatively , instead of waiting to collect a measurement record @xmath24 , one could choose to update the prior on - line i.e. after each measurement . in this case
it is likely that a few first measurement results will be discarded as we accumulate information about the qrng device at hand .
however , after enough information is received to narrow down the parameter distribution , it will be possible to convert upcoming measurements into random bit values .
to illustrate how our approach works in an experiment we consider two physical implementations of qrngs .
we first introduce mathematical models to describe the qrngs of interest in the section [ sec : physmodels ] and then proceed with the analysis of the experimental data and numerical simulations results in section [ sec : randd ] .
let us first consider a qrng based on measuring time - of - arrival statistics of a coherent light source .
our experimental setup consists of a tapered amplifier , emitting spontaneously and subsequently attenuated to a coherent state , that continuously illuminates the surface of a free - running single photon counting module with 80 ns dead time @xcite . using a gated fpga essentially acting as a time to digital converter , we measure the time interval @xmath109 between two consecutive photodetection events .
the time interval @xmath109 plays the role of a physical random variable @xmath0 that we would like to convert into a discrete uniform random variable . to determine statistical properties of @xmath109 , a quantum model of phododetection process is needed . for this purpose
we introduce a positive - operator valued measure ( povm ) @xmath110 , where @xmath111 is a projection operator that corresponds to `` no - click '' measurement of the detector and @xmath112 represents a `` click '' detection event .
note that @xmath113 .
then the detector click rate ( i.e the click probability per unit time ) reads , @xmath114,\ ] ] here @xmath115 is the density operator of the laser field and @xmath2 describes the overall detection efficiency .
therefore , the probability to get a click in a short time interval @xmath116 is @xmath117 $ ] . on the other hand ,
the probability to detect no click in the same time interval is @xmath118 .
next consider the time interval @xmath109 between two consecutive detector clicks . we can model the absence of clicks during time @xmath109 by a sequence of @xmath16 successful `` no - click '' measurements each of the duration @xmath119 .
hence , the probability to observe no clicks during time @xmath109 reads , @xmath120 where we introduced @xmath121 $ ] . in the limit of large @xmath16
we obtain , @xmath122 we now can compute the conditional probability to detect a click at @xmath123 given a click was detected at @xmath124 , @xmath125 finally , the probability density @xmath126 for the random variable @xmath109 can be obtained by taking a derivative of @xmath127 , @xmath128 two main assumptions were made in the derivation of eq.([eq : toapdf ] ) .
first , the detection events are independent and identically distributed .
this assumption is justifiable in case of moderate laser powers .
second , we have assumed noiseless detection .
the later assumption is , unfortunately , not very realistic .
avalanche photodiode detectors usually introduce two main sources of noise that affect the value of @xmath109 afterpulsing and timing jitter .
afterpulsing is a false detection event in which electrons that were trapped by quenching in a previous detector gate are rereleased in subsequent detector gates , usually occuring after a true click due to a photon absorption event .
the time interval @xmath129 between a true detection and an afterpulse event can be well characterized experimentally and the raw data can be filtered to remove the afterpulsing events by only accepting measurements with @xmath130 .
the filtering procedure effectively results in rescaling of the probability density in eq.([eq : toapdf ] ) , @xmath131 where @xmath129 is a characteristic afterpulsing time .
the time jitter is a small error in the measurement of @xmath109 .
the recorded time interval between two sequential clicks @xmath132 is a sum of two random variables @xmath133 , where @xmath109 is the `` true '' time interval with pdf @xmath134 given in eq.([eq : toapdfrenorm ] ) and @xmath135 is a time jitter random @xmath136 variable .
one can show that the probability density for @xmath132 reads , @xmath137\end{aligned}\ ] ] where @xmath138 denotes the error function .
notice that if the time jitter is small ( @xmath139 ) , eq.([eq : toapdfnoise ] ) coincides with eq.([eq : toapdfrenorm ] ) .
since the observed time jitter is indeed small we will model the time - of - arrival qrng using the probability density in eq.([eq : toapdfrenorm ] ) with one parameter @xmath140 .
let us also discuss how to implement the qrng data processing algorithm described earlier for this model .
the model pdf is given in eq.([eq : toapdfrenorm ] ) .
an obvious choice for the prior is a non - informative ( uniform ) prior @xmath141 that assigns constant weight to all values of the parameter @xmath140 .
it turns out that in this case the posterior distribution @xmath142 after @xmath143 measurements can be calculated even analytically ( instead of standard numerical updating ) as , @xmath144 where @xmath145 , and we assume that the characteristic afterpulsing time @xmath129 is known ( not a parameter ) and @xmath146 denotes the gamma distribution function . using eq.([eq : inttrans ] ) we introduce @xmath143 random variables @xmath147 , where @xmath148 and compute their probability distribution @xmath149 using eq.([eq : gofu ] ) and eq.([eq : posteriortoa ] ) , @xmath150^{n}(1-u_{i})^{\frac{1}{t(\tau_{i}-\tau_{a})}-1}}{n!(t(\tau_{i}-\tau_{a}))^{n+1}}.\ ] ] and finally we calculate the probability @xmath151 that @xmath152 falls into the @xmath153-th bin ( @xmath154 $ ] ) @xmath155,\end{aligned}\ ] ] where @xmath156 denotes the lower incomplete gamma function . applying the acceptance / rejection test to @xmath157 for all pairs @xmath158 will convert measurement outcomes @xmath159 into a sequence of uniformly distributed integers on @xmath160 $ ] .
the second system that we consider here is a popular qrng implementation based on vacuum quadrature measurement .
quantum vacuum fluctuations of the electromagnetic field are measured routinely at optical wavelengths using homodyne detection techniques @xcite .
a typical homodyne detector consists of a beam splitter with two input ( @xmath161 ) and two output ports ( @xmath162 ) .
suppose that the input port @xmath163 carries a laser field described by a density operator @xmath164 and the port @xmath165 carries the vacuum . by placing a photodetector in each of the output ports we measure the photon number difference operator @xmath166 between @xmath167 and @xmath168 , @xmath169a^{\dagger}a + [ ( \eta_{1}r)^2 - ( \eta_{2}t)^2]b^{\dagger}b + \nonumber \\ & & rt(\eta_{1}^2 + \eta_{2}^2)[a^{\dagger}b + ab^{\dagger}],\end{aligned}\ ] ] where @xmath170 ( @xmath171 ) are the transmittance ( reflectance ) of the beam splitter , @xmath172 are detector 1,2 detection efficiencies and @xmath173 ( @xmath174 ) are creation / annihilation operators for the input port @xmath163 ( @xmath165 ) .
therefore , in a general experimental situation , @xmath166 will depend on three parameters @xmath175 ( note that @xmath176 ) and the laser field @xmath164 .
but since we only perform a numerical simulation of an experiment here and thus can `` control '' the parameters perfectly , we will assume that we have a 50/50 beam splitter ( @xmath177 ) and 100 percent efficient detectors ( @xmath178 ) .
we will also assume that the laser field is in a coherent state , i.e. @xmath179 .
therefore , the expectation value of @xmath166 , @xmath180 is proportional to the expectation value of the vacuum quadrature operator @xmath181 . setting @xmath182
we conclude that by measuring the photon number difference in the output ports @xmath167 and @xmath168 we effectively measure the vacuum @xmath0 quadrature , and hence , a particular measurement outcome in a normal random variable @xmath183 with pdf @xmath184
. in reality measurement results are always affected by electronic noise .
the noise is usually model by a normal distribution @xmath185 and thus the outcome of the quadrature measurement is a sum of the `` true '' quadrature random variable and the noise i.e. @xmath186 .
since @xmath183 and @xmath187 are independent and normally distributed , their sum @xmath188 is also a normally distributed random variable with pdf @xmath189 .
therefore , we will model the output of vacuum quadrature measurement based qrng as a continuous random variable @xmath188 with the distribution function @xmath190 , @xmath191 where @xmath192 is an unknown parameter . with the qrng model at hand
we can now discuss how to apply the data processing algorithm developed in section [ sec : bayesianinference ] to the vacuum quadrature measurement qrng .
once again we start by choosing a prior .
we propose to use a non - informative prior @xmath193 as in the previous example .
the posterior distribution @xmath194 after @xmath143 measurements can then be calculated analytically and reads , @xmath195 where @xmath196 and @xmath197 denotes the gamma function .
the next step in our procedure is to introduce @xmath143 random variables @xmath198 that later on will be binned . unlike the previous example , where eq.([eq : inttrans ] ) was used for that purpose
, we will rely on _ box - muller _
transform @xcite here . recall that @xmath199 and @xmath200 , two independent uniform(0,1 ) random variables , can be converted into two independent normal @xmath201 random variable @xmath0 and @xmath202 using the following transformation , @xmath203 on the other hand , a pair of measurement outcomes @xmath204 and @xmath205 can be converted into two random variables @xmath206 and @xmath207 @xmath208 [ 0,1 ] . since @xmath209 does not depend on the parameter @xmath210 ( it is constant for a given pair @xmath211 ) , it can be immediately placed into the @xmath153-th bin that satisfies @xmath212 . as to @xmath213 which indeed is a function of @xmath210
, we can derive its probability distribution function @xmath214 using the posterior distribution in eq.([eq : posteriorhqrng ] ) , @xmath215 finally , the probability @xmath216 that @xmath217 falls into the @xmath153-th bin ( @xmath154 $ ] ) @xmath218,\end{aligned}\ ] ] where @xmath156 is the lower incomplete gamma function . applying the acceptance / rejection test to @xmath219 for all
@xmath153 will convert @xmath217 into a uniformly distributed integer on @xmath160 $ ] .
therefore , a pair of normally distributed outputs of the vacuum homodyne measurement @xmath220 and @xmath221 converts into two uniformly distributed integer random numbers @xmath222 and @xmath223 .
we collected a sample containing 256,000 measurements of the time interval between two consecutive detection events @xcite .
the raw data was filtered and all entries @xmath224 were removed from the sample to mitigate the effect of detector afterpulsing .
the resulting filtered sample consisted of 221,890 measurements .
we binned the filtered data into 100 bins of equal size @xmath225 and calculated the probability of each bin .
the corresponding probability distribution is depicted in fig .
[ fig : toadatapdf ] with red circles . based on the qrng model discussed in section [ sec : physmodelstoa ] we calculated @xmath226 , mle for the parameter @xmath140 .
we used @xmath227 in conjunction with the probability density function in eq.([eq : toapdf ] ) to fit the experimental data .
the result is depicted on fig .
[ fig : toadatapdf ] with the solid blue line .
not surprisingly , given the number of measurements , the ml curve fits the data well .
next we applied our data processing algorithm to the filtered data .
we set the acceptance probability @xmath105 and proceeded to convert the data into a set of 4-bit random numbers ( i.e. measurement results are binned among @xmath228 bins ) .
the number of measurements that passed acceptance / rejection criterion ( @xmath229 ) , and were assigned a bin value ( @xmath230 ) , was 215,538 ( out of 221,890 ) .
the resulting bin probability distribution is depicted in fig .
[ fig : toa4bitpdf ] using green triangles .
the solid blue line corresponds to the ideal 4-bit uniform distribution and the red crosses represent a 4-bit probability distribution obtained from the same data set using the conventional fixed - parameter binning technique with @xmath231 .
both methods generate a visually uniform distribution .
the uniformity is also confirmed by the values of shannon entropy per bit
@xmath232 for each distribution . for the conventional binning @xmath233 bits and
the entropy of the distribution generated by our binning method is @xmath234 bits . in conventional bin assignment methods ,
once the distribution parameter value is estimated from a given set of measurements , the number of random bits that can be generated per single measurement is , in principle , only limited by the number of measurements @xcite .
this is because the mean error ( standard deviation ) of the parameter estimator is ignored in conventional binning .
however , if the parameter estimation error is greater than the width of the bin where the measurement result is placed then such a bin assignment is erroneous and this measurement must be ignored and removed from the data .
but this is exactly what our bin assignment method with the acceptance probability @xmath105 does .
it effectively requires that the bin width should be greater than 4 standard deviations of the random variable @xmath152 .
if this requirement is not fulfilled the @xmath18-th measurement can not be assigned a bin reliably and the measurement is discarded . hence , in contrast to the conventional binning our approach reduces the overall number of measurements .
therefore , for a given initial set of data , the number of random bits per measurement is naturally less in our method . in other words
bayesian updating provides a more conservative estimate of randomness of a qrng when compared to ad - hoc binning .
to illustrate this we generated 7- and 8-bit random number distributions from the same filtered data that we used for the 4-bit distribution above and the acceptance probability @xmath105 .
the resulting distributions are depicted with green triangles on fig .
[ fig : toa7bitpdf ] and fig . [ fig : toa8bitpdf ] . as
before the solid blue line corresponds to the ideal 7(8)-bit uniform distributions and the red crosses represent 7(8)-bit probability distribution obtained from the same size data sets using the conventional fixed - parameter binning technique with @xmath231 .
the number of measurements that have passed acceptance / rejection criteria ( @xmath229 ) is 172,736 ( 122,927 ) for the 7-(8-)bit distribution .
we also calculated shannon entropy for the conventional binning , @xmath235 bits , and the entropy of our binning method is @xmath236 bits . on the other hand
the entropy in the 8-bit case for conventional binning is @xmath237 bits and for the proposed binning method is @xmath238 bits . as previously suspected , we observe a drop in the entropy of the 8-bit distribution generated using our technique .
this implies that the collected data can reliably be converted into random bit sequences up to 7 bits long .
note that the conventional binning method does not provide us with such a conclusion .
we simulated vacuum homodyne measurements using a pseudo random number generator .
two independent sets of 50,000 random numbers were created by sampling the normal distributions @xmath239 and @xmath240 respectively .
the first set with @xmath241 represents noiseless vacuum quadrature measurement whereas the second set with @xmath242 corresponds to the electronics noise .
thus , the sum of the sets simulates the vacuum homodyning based qrng that we previously modeled using eq.([eq : hqrngpdf ] ) .
we used the data to produce sets of 6- and 7-bit random numbers implementing both the conventional ( mle based ) and proposed ( bayesian ) binning methods .
the resulting distributions are depicted in fig .
[ fig : vachomodyne6bitpdf ] and fig .
[ fig : vachomodyne7bitpdf ] .
the green triangles correspond to the probability distributions generated using our technique ( @xmath105 ) , the red circles depict the results of the conventional mle based binning and black crosses represent conventional binning with the `` true '' value of the parameter @xmath243 .
examining fig .
[ fig : vachomodyne6bitpdf ] and fig .
[ fig : vachomodyne7bitpdf ] visually we observe that our method fails to produce a uniform 7-bit distribution indicating that the maximum number of random bits per measurement outcome can not exceed 6 for the simulated data sample .
this is also confirmed by the values of shannon entropy @xmath244 versus @xmath245 . of course
, generating a larger sample of measurements would allow a higher number of bits per measurement outcome as was the case in the previous section .
this illustrates the interplay between the number of measurement in a sample , acceptance probability , and the number of random bits that can be extracted from the sample .
in this manuscript we have demonstrated a new binning technique for qrngs , as well as a formalized approach to characterize traditional binning methods . in particular , ad - hoc binning approaches are shown to result in possible bias when the model of the physical qrng system is not taken into account .
using bayesian hypothesis updating , a physical model can be used to quickly characterize experimental data .
this has implications for new types of quantum statistical tests for randomness in a potentially more accessible manner than loop - hole - free bell inequality violation tests .
p. l. would like to thank bing qi , ryan bennink and travis humble for useful discussions .
this work was performed at oak ridge national laboratory , operated by ut - battelle for the u.s .
department of energy under contract no .
de - ac05 - 00or22725 .
increasing the number of bins @xmath16 for a fixed number of measurements @xmath3 also increases fluctuations in the probability of occurrence of a given bin , ultimately reducing the entropy of the distribution to zero . | the majority of quantum random number generators ( qrng ) are designed as converters of a continuous quantum random variable into a discrete classical random bit value . for the resulting random bit sequence to be minimally biased
, the conversion process demands an experimenter to fully characterize the underlying quantum system and implement parameter estimation routines .
here we show that conventional approaches to parameter estimation ( such as e.g. _ maximum likelihood estimation _ ) used on a finite qrng data sample without caution may introduce binning bias and lead to overestimation of the randomness of the qrng output . to bypass these complications
, we develop an alternative conversion approach based on the bayesian statistical inference method .
we illustrate our approach using experimental data from a time - of - arrival qrng and numerically simulated data from a vacuum homodyning qrng .
side - by - side comparison with the conventional conversion technique shows that our method provides an automatic on - line bias control and naturally bounds the best achievable qrng bit rate for a given measurement record . |
at high energies the soft and the semihard components of the scattering amplitude are closely related @xcite , and it becomes important to distinguish between semihard gluons , which participate in hard parton - parton scattering , and soft gluons , emitted in any given parton - parton qcd radiation process .
a class of models based on qcd incorporate soft and semihard processes in the treatment of high - energy hadronic interactions using a formulation compatible with analyticity and unitarity constraints @xcite . in this talk we present a qcd - based model @xcite in which the coupling @xmath0 is constrained by the value of the so called `` dynamical gluon mass '' ,
whose existence is strongly supported by recent qcd lattice simulations .
this frozen coupling , obtained by means of the pinch technique in order to derive a gauge invariant schwinger - dyson equation for the gluon propagator @xcite , has been adopted in many phenomenological studies @xcite .
more specifically , we discuss some recent applications of the model to hadron - hadron collisions , gap survival probability calculations , and soft gluon resummation techniques .
these results indicate a smooth transition from nonperturbative to perturbative behaviour of the qcd .
in the qcd - based ( or mini - jet " ) models the increase of the total cross sections is associated with semihard scatterings of partons in the hadrons @xcite .
the energy dependence of the cross sections is driven especially by gluon - gluon scattering processes , where the behaviour of the gluon distribution function at small @xmath1 exhibits the power law @xmath2 .
since gluon - gluon subprocesses are potentially divergent at small transferred momenta , it becomes important to regulate this behaviour by introducing a mass scale which separates the perturbative from the non - perturbative qcd region .
in the so called `` dgm '' model ( dynamical gluon mass model ) @xcite , the dynamical gluon mass , as well as the infrared finite coupling constant associated to it @xcite , are the natural regulators for the cross sections calculations . in this eikonal model , the total cross section ,
the ratio @xmath3 of the real to the imaginary part of the forward scattering amplitude , and the differential elastic scattering cross section are given by @xmath4 , \label{degt1}\end{aligned}\ ] ] @xmath5 \}}{\textnormal{im } \
{ i \int b\ , db\ , [ 1-e^{i\chi ( b , s ) } ] \ } } , \label{degthyj1}\end{aligned}\ ] ] and @xmath6\ , j_{0}(qb ) \right|^2 , \label{degthyj3}\end{aligned}\ ] ] respectively , where @xmath7 is the square of the total cm energy , and @xmath8 is the ( complex ) eikonal function , which is written as a combination of an even and odd eikonal terms related by crossing symmetry . in terms of the proton - proton ( @xmath9 ) and antiproton - proton ( @xmath10 ) scatterings , this combination reads @xmath11 .
the even eikonal is written as the sum of gluon - gluon , quark - gluon , and quark - quark contributions : @xmath12 , \label{final4}\end{aligned}\ ] ] where @xmath13 is the overlap function at impact parameter space and @xmath14 are the elementary subprocess cross sections of colliding quarks and gluons ( @xmath15 ) .
the odd eikonal @xmath16 , that accounts for the difference between @xmath9 and @xmath10 channels , is parametrized as @xmath17 where @xmath18 is the dynamical gluon mass and the parameters @xmath19 and @xmath20 are constants to be fitted .
the factor @xmath21 is defined as @xmath22 , with the dynamical coupling constant @xmath23 set at its frozen infrared value .
the gluon - gluon eikonal contribution , that dominates at high energy and determines the asymptotic behaviour of the total cross sections , is written as @xmath24 , where @xmath25 here @xmath26 is the convoluted structure function for pair @xmath27 , @xmath28 is the subprocess cross section and @xmath29 is a fitting parameter . in the above expression
it is introduced the energy threshold @xmath30 for the final state gluons , assuming that these are screened gluons @xcite . in the expression ( [ sloh1 ] )
the gluon - gluon subprocess cross section @xmath28 is calculated using a procedure dictated by the dynamical perturbation theory ( dpt ) @xcite .
it means that the effects of the dynamical gluon mass in the propagators and vertices are retained , and the sum of polarizations is performed for massless ( free - field ) gluons . as a result ,
since the dynamical masses go to zero at large momenta , the elementary cross sections of perturbative qcd in the high - energy limit are recovered @xcite .
the result of the fit to @xmath31 for both @xmath9 and @xmath10 channels is displayed in figs . 1 , together with the experimental data .
( solid curve ) and @xmath10 ( dashed curve ) scattering . ]
the study of the survival probability @xmath32 of large rapidity gaps ( lrg ) is currently a subject of intense theoretical and experimental interest .
its importance lies in the fact that systematic analyses of lrg open the possibility of extracting new physics from hard diffractive processes . on the theoretical side , its significance is due to the reliance of the @xmath32 calculation on subtle qcd methods .
since rapidity gaps can occur in the case of higgs boson production via fusion of electroweak bosons , we have focused on @xmath33 fusion processes @xcite .
the inclusive differential higgs boson production cross section via @xmath34 fusion is given by @xmath35 where @xmath36 is the overlap function at impact parameter space of the @xmath34 bosons .
this function represents the effective density of the overlapping @xmath34 boson distributions in the colliding hadrons .
the cross section for producing the higgs boson and having a large rapidity gap is given by @xmath37 where @xmath38 is the probability that the two initial hadrons have not undergone a inelastic scattering at the parton level .
in the dgm model this probability is given by @xmath39 , where the imaginary part @xmath40 of the eikonal function receives contributions of parton - parton interactions .
therefore , the factor @xmath38 suppresses the contribution to the higgs boson cross section where the two initial hadrons overlap and there is soft rescatterings of the spectator partons .
hence in our model we can write down the survival factor @xmath32 for higgs production via @xmath34 fusion as @xmath41 where we have used the normalization condition @xmath42 .
the sensitivity of the survival probability @xmath32 ( for @xmath9 collisions ) to the gluon dynamical mass is shown in figure 1 for some cm energies .
the @xmath32 value decreases with the increase of the energy of the incoming hadrons , in line with the available experimental data for lrg .
( central values ) as a function of the dynamical gluon mass @xmath18 . ]
it is possible to apply the soft resummation mechanism to the calculation of overlap functions in which the coupling @xmath0 is constrained by the value of the dynamical gluon mass .
the main steps of this calculation can be summarized as follow : in qed the soft photon resummation in the energy - momentum @xmath43 can be obtained order by order as @xcite @xmath44 @xmath45 = \int_{0}^{e } \frac{d^3k}{2k_{0}}\ , \left| j_{\mu}(k,\{p_{i}\})\right|^{2 } ; \end{aligned}\ ] ] here @xmath46 is the four - dimensional probability distribution for soft massless quanta emitted by a semiclassical source , @xmath47 is the maximal energy allowed to single photon emission and @xmath48 is the momenta of the emitting fields . for strong interactions
the resummed transverse momentum distribution is given by @xmath49 in qcd applications @xmath50 can be written as @xcite @xmath51 , \end{aligned}\ ] ] with the integral of the function @xmath50 , which describes the relative transverse momentum distribution induced by soft gluon emission from a pair of initially collinear partons , downed to infrared momentum modes .
the attenuation of the rise of the total cross sections comes from soft gluon @xmath52-emission .
these emissions break collinearity between the colliding partons , hence changing the overlap function @xmath53 ( matter distribution ) , given by @xmath54 note that the high - energy dependence of the total hadronic cross section ( large @xmath55 value ) is intrinsically related to the small @xmath52 value of @xmath0 .
thus the behaviour of @xmath0 in the infrared limit plays a central role . in order to calculate @xmath53
we have adopted the frozen - strong coupling @xmath23 obtained by cornwall @xcite , @xmath56 } , \hspace{0.5truecm } \textnormal{where}\end{aligned}\ ] ] @xmath57^{- 12/11 } .\end{aligned}\ ] ] we have compared our result with those obtained using the usual phenomenological prescription to the strong coupling based on richardson potential @xcite : @xmath58}.\end{aligned}\ ] ] the overlap functions @xmath59 and @xmath60 , calculated using the expressions ( 15 ) and ( 17 ) , respectively , are displayed in the fig .
3 . and @xmath61 . ]
the frozen coupling @xmath23 avoids small @xmath52 divergences in partonic subprocesses .
thus it provides an useful phenomenological tool to the study of high - energy strong interactions .
the general picture indicates a smooth transition from nonperturbative to perturbative behaviour of the qcd .
interestingly enough , by means of a triple - regge analysis which explicitly accounts for absorptive corrections , this smooth transition can also be inferred from recent determinations of the bare triple - pomeron vertex @xcite . | in this talk we introduce the main features of a qcd - based model in which the coupling @xmath0 is constrained by an infrared mass scale . we show recent applications of this model to hadron - hadron collisions , gap survival probability calculations , and soft gluon resummation techniques .
these results indicate a smooth transition from nonperturbative to perturbative behaviour of the qcd . |
_ proof _ : this proposition holds if the first - order derivative @xmath110 with @xmath111 . according to the formula in eq .
( 3 ) of the main text , we have @xmath112>0.\end{aligned}\ ] ] the details for illustrating the positivity of eq .
( 1 ) are as follows .
the inverse hyperbolic tangent function has the form @xmath113,\end{aligned}\ ] ] and the last two terms in eq .
( 1 ) can be simplified as @xmath114 thus the first - order derivative is @xmath115 in which @xmath116,\nonumber\\ & & \mathcal{t}_2=\mbox{ln}[\sqrt{x}/(1-\sqrt{1-x})],\nonumber\\ & & \mathcal{t}_3=2\sqrt{1-x}\mbox{ln}[\sqrt{x}/(1-\sqrt{1-x})]+\mbox{ln}(x/4),\end{aligned}\ ] ] respectively . due to @xmath117 , it is obvious that the term @xmath118 . for the term @xmath119 , we have @xmath120 and @xmath121 , which results in @xmath122 therefore , the second term @xmath119 in eq .
( 4 ) is positive . for the third term
, we have @xmath123 ^ 2<0,\end{aligned}\ ] ] where , in the first inequality , we use the property @xmath124 , and the last inequality is satisfied due to @xmath125 . since @xmath118 , @xmath126 and @xmath127 , the first - order derivative in eq .
( 1 ) is positive .
combining the fact with that @xmath128 corresponds to the minimum @xmath129 and @xmath130 corresponds to the maximum @xmath131 , we get that @xmath14 is a monotonically increasing function of @xmath132 , which completes the proof of the proposition .
the detailed derivation for the above result is as follows . in eq .
( 8) , when the parameter @xmath136 , the factor @xmath135 $ ] is positive . in this case , the positivity of @xmath137 is equivalent to @xmath138 where @xmath139 $ ] . in order to analyze the sign of @xmath140 , we study its monotonic property .
the first - order derivative of @xmath140 has the form @xmath141\nonumber\\ & = & \eta_1\cdot \eta_2 , \end{aligned}\ ] ] where the parameters are @xmath142,\nonumber\\ & & \eta_2=3+\mbox{ln}(\frac{x}{4}),\end{aligned}\ ] ] respectively .
since @xmath143 , we have the factor @xmath144 .
therefore , according to eq .
( 10 ) , the function @xmath145 increases monotonically when the factor @xmath146 , and it decreases monotonically for the case @xmath147 .
this property is shown in fig.1 , and the maximal value of @xmath145 corresponds to the critical point @xmath148 .
now , we analyze two endpoints of @xmath140 for @xmath128 and @xmath130 .
when @xmath128 , we can deduce @xmath149\}\nonumber\\ & = & \lim_{x\to + 0}[\mbox{ln}(\frac{x}{4})-\mbox{ln}(\frac{1+\sqrt{1-x}}{\sqrt{x}})\times ( -2)]\nonumber\\ & = & \lim_{x\to + 0}\mbox{ln}[\frac{(1+\sqrt{1-x})^2}{4}]\nonumber\\ & = & 0,\end{aligned}\ ] ] where in the second equation we have used the property @xmath150 . for the other endpoint @xmath130
, we can get @xmath151 . combining it with the monotonic properties of @xmath140
, we can find @xmath152 , which is equivalent to @xmath133 in the region @xmath117 .
furthermore , we analyze the value of second - order derivative @xmath137 at the endpoints .
when @xmath128 , we can get @xmath153}{2(2 - 5x)\sqrt{1-x}(\mbox{ln}2)^2}\nonumber\\ & = & \frac{1}{4(\mbox{ln}2)^2}\lim_{x\to + 0}\{[\mbox{ln}(\frac{4}{x})-3]\nonumber\\ & & \times\mbox{ln}(\frac{1+\sqrt{1-x}}{\sqrt{x}})\}\nonumber\\ & = & \infty.\end{aligned}\ ] ] on the other hand , when @xmath130 , we have @xmath154\mbox{ln}(\frac{1+\sqrt{1-x}}{\sqrt{x } } ) } { 2(2 - 5x)\sqrt{1-x}(\mbox{ln}2)^2}\nonumber\\ & = & \lim_{x\to1}[\frac{3-\mbox{ln}4-\mbox{ln}(\frac{1}{x})-2\sqrt{1-x}\mbox{ln } ( \frac{1+\sqrt{1-x}}{\sqrt{x}})}{6x(5x-4)(\mbox{ln}2)^2}]\nonumber\\ & = & \frac{3-\mbox{ln}4}{6\times(\mbox{ln}2)^2}\nonumber\\ & \approx & 0.55979.\end{aligned}\ ] ] thus , we have shown the second - order derivative @xmath133 in the whole region @xmath155 $ ] , and then complete the proof of proposition ii . in fig.2 ,
the derivative is plotted as a function of @xmath132 , which illustrates our result . according to the proposition @xmath156 in the letter
, we know that the squared entanglement of formation @xmath21 is monotonically increasing as a function of the squared concurrence @xmath22 . combining this property with the monogamy relation of concurrence in eq .
( 1 ) of the main text , we can derive the first inequality in eq .
( 5 ) of letter @xmath157 here , it should be emphasized that the composite system @xmath158 is in a pure state @xmath23 .
the reason is that , in a generic mixed state @xmath159 , the relation between the entanglement of formation @xmath160 and the squared concurrence @xmath161 can not be characterized by eq .
( 3 ) of letter since the subsystem @xmath109 is not equivalent to a logic qubit . furthermore , according to the proposition @xmath162 that the squared entanglement of formation @xmath21 is convex as a function of @xmath22 , we can derive @xmath163 a schematic diagram for this inequality is shown in fig.3 . due to the monotonic and convex property of @xmath164 ) , we have the relations of gradients @xmath165 and @xmath166 , which give rise to @xmath167 where @xmath168 , @xmath169 , and @xmath170 , respectively . by iterating the property of gradients @xmath171 and @xmath172
, we finally have the inequality @xmath173 combining eqs .
( 18 ) and ( 15 ) , we can obtain the inequalities in eq .
( 5 ) of letter . in the letter , the second term in eq .
( 9 ) has the form @xmath174 and here we prove that it is non - negative . for any pure state component @xmath175 in eq .
( 7 ) of the main text , the monogamy property of @xmath14 is satisfied , and we have @xmath176 for the two arbitrary pure state components @xmath175 and @xmath177 , we can get @xmath178 where in the second inequality we have used the relation of the perfect square trinomials @xmath179 after taking the square root on both sides of eq .
( 21 ) , we have @xmath180 since @xmath175 and @xmath177 are two arbitrary components in eq . ( 7 ) of the letter , the above relation of eq . ( 23 ) is also satisfied for any other components .
therefore , we can obtain @xmath181 which completes our proof .
we next show the sufficiency .
it is a fact that when the indicator @xmath182 is zero , there is _ at most _ one nonzero two - qubit concurrence in the three - qubit pure state .
this is because that , if @xmath187 and @xmath188 , we have @xmath189 which is contradictory to the premise @xmath190 . here ,
in the first inequality we use proposition @xmath156 , and in the third inequality we use proposition @xmath162 with @xmath191 ( as shown in fig.3 ) which results in the nonzero value of the indicator in the case of @xmath187 and @xmath188 .
when both the two - qubit concurrences are zero , we have the entanglement @xmath192 , which corresponds to the product states @xmath193 and @xmath194 . in the following
, we will prove that the three - qubit pure state is a bipartite product state when the indicator is zero with one nonzero two - qubit concurrence .
without loss of generality , we assume that the concurrence @xmath195 is nonzero . in this case , we can obtain @xmath196 according to the definition of the entanglement of formation @xcite , we further have @xmath197 where @xmath198 is the reduced density matrix of subsystem @xmath7 and the minimum runs over all the pure state decompositions @xmath199 with @xmath200 .
on the other hand , we known that the von neumann entropy is a strictly concave function of its input @xcite , @xmath201 where the reduced density matrix @xmath202 .
note that the equality holds if and only if all the state @xmath203 are identical @xcite .
combining eqs .
( 28 ) and ( 29 ) , we can obtain that , if @xmath8 is a mixed state , all the pure state components @xmath204 in an arbitrary pure state decomposition should have the identical reduced state @xmath205 now we prove that @xmath8 can not be a mixed state by analyzing the generic form of three - qubit pure states . under local unitary operations ,
the standard form of three - qubit pure state can be written as @xcite @xmath206 where the real number @xmath12 ranges in @xmath46 $ ] with the condition @xmath207 , and the relative phase @xmath208 changes in @xmath209 $ ] .
its two - qubit reduced state of subsystem @xmath210 can be expressed as @xmath211 with the non - normalized pure state components being @xmath212 from the requirement in eq .
( 30 ) plus the nonzero value of @xmath213 in the premise , we conclude that for both components the entropy should satisfy @xmath214 . however , in eq .
( 33 ) , the second component @xmath215 is a product state and we have @xmath216 due to @xmath217 .
this means that the probability for @xmath215 is zero and then @xmath8 can be written as @xmath218 being a two - qubit pure state .
therefore , in three - qubit pure states , when the indicator @xmath190 with the nonzero concurrence @xmath195 the composite system has the form @xmath219 similarly , for the case @xmath220 , we can derive that @xmath221 .
this completes the proof of the lemma in three - qubit pure states . _
proof_. for three - qubit mixed states , the tripartite entanglement indicator is defined as @xmath222 where the minimum runs over all the pure state decompositions @xmath223 . for the biseparable state in the lemma
, we can obtain @xmath190 , which proves the necessity of the lemma .
next , we analyze the sufficiency of the lemma .
when the indicator is zero , according to the previous lemma a in three - qubit pure states , we known that there must exist an optimal pure state decomposition for @xmath68 in which every pure state component is bipartite product and has the forms shown in eq .
( 25 ) . in the general case
, this kind of three - qubit mixed state can be written as @xmath224 which is just the biseparable entangled state and does not contain the genuine tripartite entanglement @xcite .
this completes the proof for the lemma .
[ [ vi .- proof - of - lemma-2-and - its - application - to - an - n - qubit - mixed - state - without - two - qubit - concurrence - and - n - tangles ] ] vi .
proof of lemma 2 and its application to an @xmath1-qubit mixed state without two - qubit concurrence and @xmath0-tangles ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _ lemma b_. in @xmath1-qubit pure states , the multipartite entanglement indicator @xmath226 is zero if and only if the quantum state is a bipartite product state in the forms @xmath227 where @xmath228 and @xmath229 are the complementary sets of qubits @xmath230 and @xmath231 with @xmath232 , respectively .
we next show the sufficiency . when the entanglement indicator @xmath234 is zero
, we can obtain that there is _ at most _ one nonzero two - qubit concurrence @xmath235 , which is due to the monotonic and strictly convex property of @xmath14 as a function of squared concurrence @xmath22 ( propositions @xmath156 and @xmath162 ) . if two concurrences are nonzero , the indicator is inevitably nonzero ( similar to the three - qubit case in eq .
26 ) , which is contradictory to the premise of zero indicator value . in the case of all two - qubit concurrences @xmath236 ( @xmath237 )
being zero , we have the entanglement @xmath238 and then the @xmath1-qubit pure state has the form @xmath239 . in the following , we will prove that the @xmath1-qubit pure state has the form @xmath240 when the indicator is zero with one nonzero concurrence @xmath235 . without loss of generality ,
we assume that the concurrence @xmath241 is nonzero . in this case , we have @xmath242 and @xmath243 . due to the strictly concave property of the von neumann entropy @xcite
, we obtain that when @xmath244 is a mixed state , all its pure state components @xmath245 in an arbitrary pure state decomposition should have the identical reduced state @xmath246 next , we prove that @xmath244 can not be a mixed state .
we assume that , in the spectral decomposition , the two - qubit mixed state can be written as @xmath247 where the reduced state of each orthogonal component has the same form @xmath248=\rho_{a_1}$ ] with @xmath249 , respectively .
according to the reduction interpretation of @xmath244 , the mixed state comes from a larger pure state via the partial trace of environment subsystem .
different measurement on the environment subsystem results in different pure state decomposition .
thus , for the @xmath1-qubit pure state @xmath250 , the qubits @xmath251 are the environment subsystem and equivalent to two logic qubits @xmath252 . the spectral decomposition of @xmath244 in eq . ( 40 ) can be obtained by the measurement @xmath253 on logic qubits @xmath252 , and the enlarged pure state can be expressed as @xmath254
when we measure the logic qubit @xmath255 on the basis @xmath256 , the mixed state of subsystem @xmath257 can be written as @xmath258 where the probabilities are @xmath259 and @xmath260 , and two components have the forms @xmath261 with @xmath262 , @xmath263 , @xmath264 , and @xmath265 , respectively .
the first component @xmath266 is equivalent to a three - qubit pure state , and , under local unitary transformation , it can be expressed as @xcite @xmath267 when we measure the logic qubit @xmath268 in the basis @xmath269 , the two pure state components of @xmath270 are @xmath271 due to the component @xmath272 being a product state , its probability is zero according to the condition in eq . ( 39 ) which requires that each component in an arbitrary pure state decomposition should have the equal entanglement .
therefore , the three - qubit state @xmath273 can be rewritten in the form @xmath274 combining eqs . ( 47 ) and ( 43 ) , we can obtain that , for the two components @xmath275 and @xmath276 in the spectral decomposition of @xmath244 , there exists only _
one_. this is because the three - qubit state @xmath273 is a product state in the partition @xmath277 according to eq .
if both the two components @xmath275 and @xmath276 exist , then the state @xmath273 in eq .
( 43 ) will be an entangled state which is contradictory to the expression in eq .
( 47 ) . without loss of generality , we assume that the component @xmath275 exists . similarly , for the three - qubit state @xmath278 in eq .
( 44 ) , we can obtain that there is only one component between @xmath279 and @xmath280 . here
, we assume the component @xmath279 exists . in this case , the two - qubit state @xmath244 in eq .
( 40 ) is a rank-2 state , and has _ at most _ two components @xmath275 and
@xmath279 in its spectral decomposition .
thus , the four - qubit state in eq . ( 41 ) is equivalent to a three - qubit state and can be rewritten as @xmath281 where the logic qubit @xmath282 represents the environment subsystem @xmath251 . when the entanglement indicator @xmath234 is zero with the nonzero @xmath241
, the quantum state @xmath283 has the following form according to the previous lemma a @xmath284 note that the forms @xmath285 are excluded since they are contradictory to the condition that @xmath241 is nonzero . combining eqs .
( 49 ) and ( 48 ) , we can obtain that the two - qubit quantum state @xmath244 has only one component in its spectral decomposition , and then it is a two - qubit pure state @xmath286 .
therefore , when the indicator @xmath234 is zero with a nonzero @xmath241 , the @xmath1-qubit pure state has the form @xmath287 similarly , when the nonzero concurrence is @xmath235 with @xmath288 , we can get the @xmath1-qubit pure state @xmath289 which completes the proof of lemma b. _ lemma c_. in @xmath1-qubit mixed states , the multipartite entanglement indicator @xmath290 is zero if and only if the quantum state is biseparable in the form @xmath291 where @xmath228 and @xmath229 are the complementary sets of qubits @xmath230 and @xmath231 , respectively . _
proof_. in @xmath1-qubit mixed states , the multipartite entanglement indicator is defined as @xmath292,\end{aligned}\ ] ] where the minimum runs over all the pure state decomposition @xmath293 . for the biseparable state in eq .
( 53 ) , it is easy to obtain that the indicator @xmath290 is zero , which proves the necessity of the lemma .
next , we consider the sufficiency of the lemma .
when the entanglement indicator @xmath290 being zero , according to the previous lemma b in @xmath1-qubit pure state , we can obtain that there must exist an optimal pure state decomposition for @xmath294 in which each pure state component is bipartite product and has the forms shown in eq .
( 38 ) . in the general case
, this kind of @xmath1-qubit mixed state can be written as the form shown in eq .
( 53 ) , which completes the proof of lemma c. according to lemmas b and c , we can obtain that , when the indicator @xmath75 is zero , there is _ at most _ two - qubit entanglement in the partition @xmath17 . on the other hand ,
whenever the multiqubit entanglement exists in this partition @xmath17 , the entanglement indicator @xmath75 is surely nonzero .
_ lemma 2_. in @xmath1-qubit mixed states , the multipartite entanglement indicator @xmath295 is zero if and only if the quantum state is @xmath77-separable in the form @xmath296 , which has at most two - qubit entanglement with the superscript @xmath79 being all permutations of the @xmath1 qubits . _
proof_. when the @xmath1-qubit quantum state is in the form shown in this lemma , it is easy to obtain that all the indicators @xmath297s are zero for @xmath298 according to the lemma b. thus the multipartite entanglement indicator @xmath80 is zero , which proves the necessity of lemma 2 . next , we analyze the sufficiency .
when the indicator @xmath80 is zero , we can obtain that , for each component @xmath299 in the optimal pure state decomposition of @xmath300 , the qubit @xmath301 is entangled with at most one other qubit @xmath302 with @xmath303 . in this case , the pure state component is a tensor product state of at most @xmath304 two - qubit entangled states .
note that the two - qubit tensor product state @xmath305 is a special form of pure state @xmath306 . after considering all the pure state components ,
the @xmath1-qubit mixed state can be written in a general @xmath77-separable form @xmath307 which only contains two - qubit entanglement with the superscript @xmath79 being all permutations of the @xmath1 qubits .
this completes the proof of lemma 2 in the letter .
next , as an application case , we will analyze an @xmath1-qubit mixed state written as @xcite @xmath308 where the parameter @xmath309 and the @xmath1-qubit @xmath41 state being @xmath310 . in ref .
@xcite , it was pointed out that this quantum state is entangled but without two - qubit concurrence and multiqubit @xmath0-tangles with @xmath311 , where the @xmath0-tangle is a kind of multipartite entanglement measure based on the monogamy property of squared concurrence . to reveal the critical entanglement structure
, we will utilize the newly presented multipartite entanglement indicator @xmath80 in eq .
( 55 ) . due to the quantum state
@xmath312 being symmetric under qubit permutations , we can get the following relation @xmath313,\nonumber\end{aligned}\ ] ] where the minimum runs over all the pure state decompositions of @xmath312 . after some analysis
, we obtain that the optimal decomposition is @xmath314 with the two components being @xmath315 and @xmath316 , which means the decomposition in eq .
( 57 ) is optimal
. therefore , the entanglement indicator in eq .
( 58 ) is @xmath317,\nonumber\end{aligned}\ ] ] where we use the symmetry property of @xmath318 , and the two concurrences are @xmath319 and @xmath320 .
the indicator is nonzero and detects the existence of genuine multiqubit entanglement .
since the optimal component @xmath318 is not product under any partition , the multipartite entanglement is the genuine @xmath1-qubit entanglement . in table
i , we give the values of multiqubit entanglement indicator @xmath321 for the cases @xmath322 . [ [ vii .- the - calculation - of - indicator - tau_sef2rho_c_1c_2r_1c_1-in - the - dynamics - of - cavity - reservoir - systems ] ] vii .
the calculation of indicator @xmath323 in the dynamics of cavity - reservoir systems ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in the letter , the multipartite entanglement indicator of the system @xmath324 has the form @xmath325 which is used to detect the genuine tripartite entanglement in the dynamical procedure . here , a crucial point is to calculate the bipartite entanglement @xmath102 , because the subsystem @xmath100 is not equivalent to a logic qubit and then the formula in eq .
( 3 ) of letter does not work . from the koashi - winter formula @xcite
, this entanglement is related to quantum discord @xmath326 where the minimum runs over all the positive operator - valued measures ( povms ) .
et al _ presented an effective method for calculating the quantum discord and choosing the optimal measurement @xcite .
after some analysis , we obtain that the optimal measurement is @xmath327 , and then the entanglement of formation is @xmath328 where @xmath329 with @xmath330^{1/2}$ ] . in fig.2 of the main text , the squared entanglements @xmath331 , @xmath332 and @xmath333 are plotted as functions of the time evolution @xmath104 and the initial amplitude @xmath105 , which characterize the bipartite entanglement distribution in the multipartite system .
in particular , the indicator @xmath323 in the figure detects the genuine tripartite entanglement in the dynamical procedure .
[ [ viii .- the - indicator - tau_sef2-not - being - monotonic - under - local - operations - and - classical - communication ] ] viii .
the indicator @xmath334 not being monotonic under local operations and classical communication ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the entanglement indicator @xmath334 in eq .
( 16 ) of letter is used to detect the multipartite entanglement not stored in two - qubit pairs . in the following
, we will prove that this indicator is not an entanglement monotone under local operations and classical communication ( locc ) . as a counter - example , we consider the tripartite entanglement indicator @xmath335 , in which @xmath336 is the reduced state of @xmath337 in eq .
( 19 ) of the main text .
when the initial state parameter is chosen as @xmath338 and the time evolution is @xmath339 , the value of the indicator is @xmath340 it is known that any local protocol can be decomposed into a sequence of two - outcome povms involving only one party @xcite . for the quantum state @xmath336 , we consider a local povm performed on the subsystem @xmath341 , which has the form @xcite @xmath342 and satisfies the relation @xmath343 . here
we choose the povm parameters to be @xmath344 and @xmath345 , respectively .
after the local operation , the average entanglement indicator is @xmath346 where @xmath347 with @xmath348 $ ] and @xmath349 , respectively . by a direct comparison , we have @xmath350 which means that the indicator @xmath335 can increase under the locc .
therefore , we obtain that the multipartite entanglement indicator @xmath334 in eq .
( 16 ) of letter is not an entanglement monotone .
finally , we want to point out that @xmath334 is still an effective indicator for multipartite entanglement .
the situation is similar to that of residual tangle for the squared concurrence which is also increasing under the locc @xcite but can be served as a quantifier for multipartite entanglement @xcite .
r. horodecki , p. horodecki , m. horodecki , and k. horodecki , rev .
* 81 * , 865 ( 2009 ) .
w. k. wootters , phys .
lett . * 80 * , 2245 ( 1998 ) .
m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , england , 2000 ) , p516 - 517 .
a. acn , a. andrianov , l. costa , e. jan , j. i. latorre , and r. tarrach , phys .
lett . * 85 * , 1560 ( 2000 ) .
m. huber , f. mintert , a. gabriel , and b. c. hiesmayr , phys .
lett . * 104 * , 210501 ( 2010 ) . y .-
k . bai , m .- y . ye , and z. d. wang , phys . rev .
a * 78 * , 062325 ( 2008 ) .
m. koashi and a. winter , phys .
a * 69 * , 022309 ( 2004 ) .
q. chen , c. zhang , s. yu , x. x. yi , and c. h. oh , phys . rev . a * 84 * , 042313 ( 2011 ) . w. dr , g. vidal , and j. i. cirac , phys .
a * 62 * , 062314 ( 2000 ) .
bai , d. yang , and z. d. wang , phys .
a * 76 * , 022336 ( 2007 ) .
l. amico , r. fazio , a. osterloh , and v. vedral , rev .
* 80 * , 517 ( 2008 ) . | we prove exactly that the squared entanglement of formation , which quantifies the bipartite entanglement , obeys a general monogamy inequality in an arbitrary multiqubit mixed state . based on this kind of exotic monogamy relation ,
we are able to construct two sets of useful entanglement indicator : the first one can detect all genuine multiqubit entangled states even in the case of the two - qubit concurrence and @xmath0-tangles being zero , while the second one can be calculated via quantum discord and applied to multipartite entanglement dynamics . moreover , we give a computable and nontrivial lower bound for multiqubit entanglement of formation . for multipartite quantum systems ,
one of the most important properties is that entanglement is monogamous @xcite , which implies that a quantum system entangled with another system limits its entanglement with the remaining others @xcite . for entanglement quantified by the squared concurrence @xcite , coffman , kundu , and wootters ( ckw ) proved the first quantitative relation @xcite for three - qubit states , and osborne and verstraete proved the corresponding relation for @xmath1-qubit systems , which reads @xcite @xmath2 similar inequalities
were also generalized to gaussian systems @xcite and squashed entanglement @xcite .
as is known , the monogamy property can be used for characterizing the entanglement structure in many - body systems @xcite . a genuine three - qubit entanglement measure named three - tangle "
was obtained via the monogamy relation of squared concurrence in three - qubit pure states @xcite .
however , for three - qubit mixed states , there exists a special kind of entangled state that has neither two - qubit concurrence nor three - tangle @xcite .
there also exists a similar case for @xmath1-qubit mixed states @xcite . to reveal this critical entanglement structure other exotic monogamy relations beyond the squared concurrence may be needed . on the other hand , from a practical viewpoint , to calculate the entanglement measures appeared in the monogamy relation is basic .
unfortunately , except for the two - qubit case @xcite , this task is extremely hard ( or almost impossible ) for mixed states due to the convex roof extension of pure state entanglement @xcite .
quantum correlation beyond entanglement ( e.g. , the quantum discord @xcite ) has recently attracted considerable attention , and various efforts have been made to connect quantum discord to quantum entanglement @xcite .
it is natural to ask whether or not the calculation method for quantum discord can be utilized to characterize the entanglement structure and entanglement distribution in multipartite systems . in this letter , by analyzing the entanglement distribution in multiqubit systems , we prove exactly that the squared entanglement of formation @xcite is monogamous in an arbitrary multiqubit mixed state .
furthermore , based on the exotic monogamy relation , we construct two sets of useful indicators overcoming the flaws of concurrence , where the first one can detect all genuine multiqubit entangled states and be utilized in the case when the concurrence and @xmath0-tangles are zero , while the second one can be calculated via quantum discord and applied to a practical dynamical procedure .
finally , we give a computable and nontrivial lower bound for multiqubit entanglement of formation . _
general monogamy inequality for squared entanglement of formation .
_ the entanglement of formation in a bipartite mixed state @xmath3 is defined as @xcite , @xmath4 where the minimum runs over all the pure state decompositions @xmath5 and @xmath6 is the von neumann entropy of subsystem @xmath7 .
for a two - qubit mixed state @xmath8 , wootters derived an analytical formula @xcite @xmath9 where @xmath10 is the binary entropy and @xmath11 is the concurrence with the decreasing nonnegative @xmath12s being the eigenvalues of the matrix @xmath13 . a key result of this work is to show exactly that the bipartite entanglement quantified by the squared entanglement of formation @xmath14 obeys a general monogamy inequality in an arbitrary @xmath1-qubit mixed state , _
i.e. _ , @xmath15 where @xmath16 quantifies the entanglement in the partition @xmath17 ( hereafter @xmath18 for qubit cases ) , and @xmath19 quantifies the one in the two - qubit system @xmath20 . under two assumptions , a qualitative analysis on three - qubit pure states
was given in ref .
@xcite . before showing the general inequality ,
we first give the two propositions , whose analytical proofs are presented in the supplemental material @xcite .
_ proposition i _ : the squared entanglement of formation @xmath21 in two - qubit mixed states varies monotonically as a function of the squared concurrence @xmath22 . _ proposition ii _ : the squared entanglement of formation @xmath21 is convex as a function of the squared concurrence @xmath22 .
we now analyze the monogamy property of @xmath14 in an @xmath1-qubit pure state @xmath23 . according to the schmidt decomposition @xcite ,
the subsystem @xmath24 is equal to a logic qubit @xmath25 .
thus the entanglement @xmath26 can be evaluated using eq .
( 3 ) , leading to @xmath27 where we have used the two propositions , with the details presented in @xcite . at this stage ,
most importantly , we prove that the squared entanglement of formation @xmath14 is monogamous in an arbitrary @xmath1-qubit mixed state @xmath28 . in this case , the analytical wootters formula in eq .
( 3 ) can not be applied to @xmath29 , since the subsystem @xmath24 is not a logic qubit in general .
but , we can still use the convex roof extension of pure state entanglement as shown in eq .
( 2 ) .
therefore , we have @xmath30 where the minimum runs over all the pure state decompositions @xmath31 .
we assume that the optimal decomposition for eq .
( 6 ) takes the form @xmath32 under this decomposition , we have @xmath33 where the @xmath34 is the average entanglement of formation under the specific decomposition in eq . ( 7 ) and the parameter @xmath35 $ ] .
then we can derive the following monogamy inequality @xmath36 where , in the second equation , the first term is non - negative because the @xmath14 is monogamous in pure state components , and the second term is also non - negative from a rigorous analysis shown in the supplemental material @xcite , justifying the monogamous relation . on the other hand , for the two - qubit entanglement of formation , the following relation is satisfied @xmath37 since the @xmath38 is a specific average entanglement under the decomposition in eq .
( 7 ) , which is greater than @xmath39 in general . combining eqs .
( 9 ) and ( 10 ) , we can derive the monogamy inequality of eq .
( 4 ) , such that we have completed the whole proof showing that the squared entanglement @xmath14 is monogamous in @xmath1-qubit mixed states .
_ two kinds of multipartite entanglement indicator . _ lohmayer _ et al _ @xcite studied a kind of mixed three - qubit states composed of a @xmath40 state and a @xmath41 state @xmath42 where @xmath43 , @xmath44 , and the parameter @xmath45 ranges in @xmath46 $ ] .
they found that , when the parameter @xmath47 with @xmath48 and @xmath49 , the mixed state @xmath50 is entangled but without two - qubit concurrence and three - tangle .
the three - tangle quantifies the genuine tripartite entanglement and is defined as @xcite @xmath51 $ ] .
it is still an unsolved problem on how to characterize the entanglement structure in this kind of states , although an explanation via the enlarged purification system was given @xcite .
based on the monogamy inequality of @xmath14 in pure states , we can introduce a kind of indicator for multipartite entanglement in an @xmath1-qubit mixed state @xmath28 as @xmath52,\ ] ] where the minimum runs over all the pure state decompositions @xmath53 .
this indicator can detect the genuine three - qubit entanglement in the mixed state specified in eq . (
11 ) .
after some analysis , we can get the optimal pure state decomposition for the three - qubit mixed state @xmath54 where the pure state component @xmath55 and the parameter @xmath56 with @xmath57 .
then the indicator is @xmath58 where @xmath59 and @xmath60 . in fig.1 , we plot the entanglement indicators @xmath61 , @xmath62 and @xmath63 in comparison to the indicators @xmath64 , @xmath65 and @xmath66 calculated originally in ref .
@xcite . as seen from fig.1
, although the three - tangle @xmath64 is zero when @xmath67 , the nonzero @xmath61 indicates the existence of the genuine three - qubit entanglement .
this point may also be understood as a fact that the three - tangle @xmath64 indicates merely the @xmath40-type entanglement while the newly introduced indicator @xmath61 can detect _ all _ genuine three - qubit entangled states . for three - qubit mixed states ,
a state @xmath68 is called genuine tripartite - entangled if any decomposition into pure states @xmath69 contains at least one genuine tripartite - entangled component @xmath70 with @xmath71 and @xmath72 corresponding to the states of a single qubit or a couple of qubits @xcite . for the tripartite entanglement indicator @xmath73
, we have the following lemma and the proof can be found in the supplemental material @xcite .
_ lemma 1 _ : for three - qubit mixed states , the multipartite entanglement indicator @xmath73 is zero if and only if the quantum state is biseparable , _
i.e. _ , @xmath74
. when the three - qubit mixed state @xmath68 is genuine tripartite entangled , its optimal pure state decomposition contains at least one three - qubit entangled component . according to the lemma
, we obtain that @xmath73 is surely nonzero . for @xmath1-qubit mixed states , when the indicator @xmath75 in eq .
( 12 ) is zero , we can prove that there exists _ at most _ two - qubit entanglement in the partition @xmath17 ( see lemmas b and c in @xcite ) and we further have the following lemma : _ lemma 2 _ : in @xmath1-qubit mixed states , the multipartite entanglement indicator @xmath76 is zero if and only if the quantum state is @xmath77-separable in the form @xmath78 , which has at most two - qubit entanglement with the superscript @xmath79 being all permutations of the @xmath1 qubits . according to lemma 2 , whenever an @xmath1-qubit state contains genuine multiqubit entanglement , the indicator @xmath80 is surely nonzero . thus this quantity can serve as a genuine multiqubit entanglement indicator in @xmath1-qubit mixed states .
the analytical proof of this lemma and its application to an @xmath1-qubit mixed state ( without two - qubit concurrence and @xmath0-tangles ) are presented in the supplemental material @xcite . in general , the calculation of the indicators defined in eqs .
( 12 ) and ( 15 ) is very difficult due to the convex roof extension . here ,
based on the monogamy property of @xmath14 in mixed states , we can also introduce an alternative multipartite entanglement indicator as @xmath81 which detects the multipartite entanglement ( under the given partition ) not stored in pairs of qubits ( although this quantity is not monotone under local operations and classical communication @xcite ) . from the koashi - winter formula @xcite ,
the multiqubit entanglement of formation can be calculated by a purified state @xmath82 with @xmath83 , @xmath84 where @xmath85 is the quantum conditional entropy with @xmath86 being the von neumann entropy , and the quantum discord @xmath87 is defined as @xcite @xmath88 with the minimum running over all the povms and the measurement being performed on subsystem @xmath89 .
recent studies on quantum correlation provide some effective methods @xcite for calculating the quantum discord , which can be used to quantify the indicator in eq .
( 16 ) . for all partitions
, we may introduce a partition - independent indicator @xmath90 .
we now apply the indicator @xmath91 to a practical dynamical procedure of a composite system which is composed of two entangled cavity photons being affected by the dissipation of two individual @xmath1-mode reservoirs .
the interaction of a single cavity - reservoir system is described by the hamiltonian @xcite @xmath92 .
when the initial state is @xmath93 with the dissipative reservoirs being in the vacuum state , the output state of the cavity - reservoir system has the form @xcite @xmath94 where @xmath95 with the amplitudes being @xmath96 and @xmath97^{1/2}$ ] . as quantified by the concurrence , the entanglement dynamical property was addressed in refs .
@xcite , but the multipartite entanglement analysis is mainly based on some specific bipartite partitions in which each party can be regarded as a logic qubit .
when one of the parties is not equivalent to a logic qubit , the characterization for multipartite entanglement structure is still an open problem . for example , in the dynamical procedure , although the monogamy relation @xmath98 is satisfied , the entanglement @xmath99 is unavailable so far because subsystem @xmath100 is a four - level system and the convex roof extension is needed . fortunately , in this case , we can utilize the presented indicator @xmath101 to indicate the genuine tripartite entanglement , where @xmath102 can be obtained via the quantum discord @xmath103 @xcite .
this indicator detects the genuine tripartite entanglement which does not come from two - qubit pairs . in fig.2 , the indicator and its entanglement components
are plotted as functions of the time evolution @xmath104 and the initial amplitude @xmath105 , where the nonzero @xmath106 actually detects the tripartite entanglement area and the bipartite components of @xmath14 characterize the entanglement distribution in the dynamical procedure . by analyzing the multipartite entanglement structure
, we can know that how the initial cavity photon entanglement transfers in the multipartite cavity - reservoir system , which provides the necessary information to design an effective method for suppressing the decay of cavity photon entanglement .
_ discussion and conclusion .
_ the entanglement of formation is a well - defined measure for bipartite entanglement and has the operational meaning in entanglement preparation and data storage @xcite .
unfortunately , it does not satisfy the usual monogamy relation . as an example
, its monogamy score for the three - qubit @xmath41 state is @xmath107 . in this letter
, we show exactly that the squared entanglement @xmath14 is monogamous , which mends the gap of the entanglement of formation .
furthermore , in comparison to the monogamy of concurrence , the newly introduced indicators can really detect all genuine multiqubit entangled states and extend the territory of entanglement dynamics in many - body systems .
in addition , via the established monogamy relation in eq .
( 4 ) , we can obtain @xmath108 which provides a nontrivial and computable lower bound for the entanglement of formation .
in summary , we have not only proven exactly that the squared entanglement of formation is monogamous in @xmath1-qubit mixed states but also provided a set of useful tool for characterizing the entanglement in multiqubit systems , overcoming some flaws of the concurrence .
two kinds of indicator have been introduced : the first one can detect all genuine multiqubit entangled states and solve the critical outstanding problem in the case of the two - qubit concurrence and @xmath0-tangles being zero , while the second one can be calculated via quantum discord and applied to a practical dynamical procedure of cavity - reservoir systems when the monogamy of concurrence loses its efficacy . moreover , the computable lower bound can be utilized to estimate the multiqubit entanglement of formation .
_ acknowledgments . _
this work was supported by the rgc of hong kong under grant nos .
hku7058/11p and hku7045/13p .
y .- k.b . and
y .- f.x .
were also supported by nsf - china ( grant no .
10905016 ) , hebei nsf ( grant no .
a2012205062 ) , and the fund of hebei normal university .
_ note added . _ recently , by using the same assumptions as those made in @xcite , a similar idea on the monogamy of squared entanglement of formation was presented in @xcite , but the claimed monogamy for mixed states was not proven in that paper @xcite , in contrast to what we have done in the present work .
99 r. horodecki p. horodecki , m. horodecki and k. horodecki , rev .
mod .
phys . * 81 * , 865 ( 2009 ) . c. h. bennett , h. j. bernstein , s. popescu , and b. schumacher , phys .
rev .
a. * 53 * , 2046 ( 1996 ) .
w. k. wootters , phys .
rev .
lett . * 80 * , 2245 ( 1998 ) .
v. coffman , j. kundu , and w. k. wootters , phys .
rev .
a. * 61 * , 052306 ( 2000 ) .
t. j. osborne and f. verstraete , phys .
. lett . * 96 * , 220503 ( 2006 ) .
g. adesso and f. illuminati , new j. phys . *
8 * , 15 ( 2006 ) . t. hiroshima , g. adesso , and f. illuminati , phys
. rev .
lett . * 98 * , 050503 ( 2007 ) . m. koashi and a. winter , phys .
rev .
a * 69 * , 022309 ( 2004 ) .
m. christandl and a. winter , j. math .
phys .
* 45 * , 829 ( 2004 ) .
y .- k .
bai , d. yang , and z. d. wang , phys .
rev .
a * 76 * , 022336 ( 2007 ) ; y .- k .
bai and z. d. wang , phys .
rev .
a * 77 * , 032313 ( 2008 ) .
r. lohmayer , a. osterloh , j. siewert , and a. uhlmann , phys .
rev .
lett . * 97 * , 260502 ( 2006 ) .
y .-
k . bai , m .- y . ye , and z. d. wang , phys . rev .
a * 78 * , 062325 ( 2008 ) . c. h. bennett , d. p. divincenzo , j. a. smolin , and w. k. wootters , phys .
rev .
a * 54 * , 3824 ( 1996 ) .
h. ollivier and w. h. zurek , phys .
rev .
lett . * 88 * , 017901 ( 2001 ) . l. henderson and v. vedral , j. phys .
a * 34 * , 6899 ( 2001 ) .
k. modi , a. brodutch , h. cable , t. paterek , and v. vedral , rev .
mod .
phys . * 84 * , 1655 ( 2012 ) . s. hill and w. k. wootters ,
phys .
rev .
lett . * 78 * , 5022 ( 1997 ) .
y .-
k .
bai , n. zhang , m .- y . ye , and z. d. wang , phys . rev .
a * 88 * , 012123 ( 2013 ) .
supplemental material .
a. peres , _ quantum theory : concepts and metheds _ ( kluwre academic publishers , dordrecht , 1995 ) .
s. luo , phys .
rev .
a * 77 * , 042303 ( 2008 ) .
m. d. lang and c. m. caves , phys .
rev .
lett . * 105 * , 150501 ( 2010 ) .
p. giorda and m. g. a. paris , phys .
rev .
lett . * 105 * , 020503 ( 2010 ) .
g. adesso and a. datta , phys .
rev .
lett .
* 105 * , 030501 ( 2010 ) .
m. ali , a. r. p. rau , and g. alber , phys .
rev .
a * 81 * , 042105 ( 2010 ) .
y. huang , phys .
rev .
a * 88 * , 014302 ( 2013 ) .
l .-
x .
cen , x. q. li , j. shao , and y. j. yan , phys .
rev .
a * 83 * , 054101 ( 2011 ) .
q. chen , c. zhang , s. yu , x. x. yi , and c. h. oh , phys . rev . a * 84 * , 042313 ( 2011 ) . m. shi , c. sun , f. jiang , x. yan , and j. du , phys .
rev .
a * 85 * , 064104 ( 2012 ) . c. e. lpez , g. romero , f. lastra , e. solano , and j. c. retamal , phys .
rev .
lett .
* 101 * , 080503 ( 2008 ) .
y .-
k . bai , m .- y . ye , and z. d. wang , phys . rev .
a * 80 * , 044301 ( 2009 ) ; w. wen , y .- k .
bai , and h. fan , eur .
phys .
j. d * 64 * , 557 ( 2011 ) .
t. r. de oliveira , m. f. cornelio , and f. f. fanchini , phys .
rev . a * 89 * 034303 ( 2014 ) .
note that the wootter s formula of eq .
( 3 ) can not be directly applied to the entanglement of formation @xmath29 because the subsystem @xmath109 is not a logic qubit in general . |
the first randall - sundrum(rs1 ) brane - world scenario@xcite was designed to solve the gauge hierarchy problem which is one of the longstanding puzzle in physics . to examine this problem they have introduced two branes located at the boundary of the compactified fifth dimension .
the second randall - sundrum(rs2 ) scenario@xcite is followed from rs1 by remoting one of the brane to infinity .
the most remarkable feature of rs2 scenario is that it leads to a massless graviton localized on the 3-brane at the linearized fluctuation level@xcite . in this paper
we will explore the localized rs graviton problem at rs1 and rs2 from the viewpoint of the singular quantum mechanics .
in addition to its good features on hierarchy and localized graviton problem , rs picture supports a non - static cosmological solution@xcite which leads to the conventional friedmann equation if one introduces the bulk and brane cosmological constants and imposes a particular fine - tuning condition between them .
furthermore , rs scenario is also applied to the cosmological constant hierarchy@xcite and black hole physics@xcite .
the bulk spacetime of rs scenario is two copies of @xmath3 glued in a @xmath4-symmetric way along a boundary which is interpreted as the 3-brane world - volume .
it is explicitly seen by examing the line elements ; @xmath5 where @xmath6 and @xmath7 .
the parameter @xmath8 is a radius of the compactified fifth dimension .
the anology of rs scenario to @xmath9/cft@xcite enables us to explore the finite temperature effect in rs brane - world scenario by extending @xmath3 bulk spacetime to schwarzschild-@xmath3@xcite .
inserting the small fluctuation equations @xmath10 to @xmath11 einstein equation one can derive a gravitational fluctuation equation @xmath12 where @xmath13 represents @xmath14 rs scenario .
for each rs scenario the potential becomes @xmath15 \\
\nonumber v_2(z)&=&\frac{15 k^2}{8(k|z| + 1)^2 } - \frac{3}{2 } k \delta(z)\end{aligned}\ ] ] where @xmath16 . the function @xmath17 is related to the linearized gravitational field @xmath18 as follows @xmath19 where @xmath20 and @xmath21 since all components are same , lorentz indicies @xmath22 and @xmath23 are suppressed in eq.([relation1 ] ) .
when deriving the linearized fluctuation equation ( [ grafluc ] ) , we have used the rs gauge choice @xmath24 for each rs scenario .
this gauge choice , however , generally generates a non - trivial bending effect on the brane@xcite .
the bending effect usually makes the linearized fluctuation equation ( [ grafluc ] ) to be non - homogeneous form , _
i.e. _ @xmath25 .
thus , inclusion of the bending effect makes the stroy to be more complicated . in this paper
we will not consider the bending effect for simplicity .
the linearized fluctuation equation ( [ grafluc ] ) looks like usual schrdinger equation . from the purely mathematical point of view the hamiltonian operator @xmath26 in eq.([grafluc ] ) is a singular operator due to the singular @xmath27-function potential in @xmath28 . in the path - integral framework@xcite
the @xmath29 @xmath27-function potential was treated by schulman about one and half decades ago as follows@xcite .
let us consider @xmath29 hamiltonian @xmath30 where @xmath31 it is well - known that the euclidean propagator @xmath32 $ ] for @xmath33 obeys the following integral equation @xmath34 = g_v[x_1 , x_2 : t ] - v \int_0^t ds \int dx g_v[x_1 , x ; t - s ] \delta(x ) g[x , x_2 ; s]\ ] ] where @xmath35 $ ] is an euclidean propagator for @xmath36 .
the euclidean propagator @xmath32 $ ] is related to the usual feynman propagator(or kernel ) @xmath37 $ ] as follows ; @xmath38 = g[x_1 , x_2 ; it].\ ] ] taking a laplace transform @xmath39 to both sides of eq.([inteq ] ) yields @xmath40 = \hat{g}_v[x_1 , x_2 ; e ] - v \hat{g}_v[x_1 , 0 ; e ] \hat{g}[0 , x_2 ; e]\ ] ] which supports a solution @xmath41 = \frac{\hat{g}_v[0 , x_2 ; e]}{1 + v \hat{g}_v[0 , 0 , e]}.\ ] ] inserting eq.([example4 ] ) into eq.([example3 ] ) again completes schulman s procedure ; @xmath42 = \hat{g}_v[x_1 , x_2 ; e ] - \frac{\hat{g}_v[x_1 , 0 ; e ] \hat{g}_v[0 , x_2 ; e ] } { \frac{1}{v } + \hat{g}_v[0 , 0 ; e]}.\ ] ] the usual energy - dependent green s function @xmath43 $ ] which is a fourier transform of @xmath37 \theta(t)$ ] , where @xmath44 is a step function , is also evaluated from the corresponding fixed - energy amplitude @xmath45 $ ] by a relation @xmath46 = -i \hat{g}[x_1 , x_2 ; -e]\ ] ] where @xmath47 in @xmath48 is a usual euclidean nature .
of course , one can compute the feynman propagator by taking an inverse laplace transform to @xmath45 $ ] and using a relation ( [ kerneldef ] ) .
extension of schulman s procedure to higher dimensional cases is not straightforward due to the infinity arising from origin . in these cases we have to modify eq.([schulman ] ) appropriately to escape the ultraviolet divergence@xcite .
especially , @xmath49 case is very interesting because a lot of non - trivial effects are involved in @xmath49 @xmath27-function potential such as scale anomaly and dimensional transmutation . in ref.@xcite
jackiw explored the @xmath49 @xmath27-function potential system by making use of the physically - oriented coupling constant renormalization and the mathematically - oriented self - adjoint extension@xcite .
he also derived the relation of the renormalized coupling constant to the self - adjoint extension parameter .
his result is generalized within a path - integral or green s function formalism in ref.@xcite .
the purpose of this paper is to examine the property of the localized gravity in rs1 and rs2 scenario by treating eq.([grafluc ] ) as a schrdinger equation . in this paper
we adopt @xmath9/cft setting , _ i.e. _ single copy of @xmath3 spacetime with a singular brane on the boundary .
the @xmath9/cft setting generates non - trivial constraints . for rs1 and
rs2 it generates @xmath29 box(@xmath50 ) and half - line(@xmath51 ) constraints respectively .
these constraints makes the fixed - energy amplitude for @xmath26 to be crucially dependent on the boundary conditions(bcs ) .
the combination of these constraints with a singular @xmath27-function potential makes the situation to be complicated .
the fascinating fact is that dirichlet bc requires a coupling constant renormalization to lead a non - trivial fixed - energy amplitude although our case is one - dimensional singular quantum mechanics .
the paper is organized as follows . in section 2
we will consider the free particle case with an half - line constraint and @xmath27-function potential at the boundary as a toy model of rs2 case . in this section
we will show how bcs play important roles in this simple singular quantum mechanics .
also we will show why coupling constant renormalization is necessary to lead a non - trivial modification in the fixed - energy amplitude at dirichlet bc .
in section 3 we will compute the fixed - energy amplitude for rs2@xcite which depends on a free parameter @xmath1 where @xmath2 and @xmath52 correspond respectively to pure dirichlet and pure neumann bc cases .
we will derive in this section the general criterion in the parameter space for the localized graviton on the 3-brane to be massless .
we will also show that the massless graviton at @xmath53 is followed via a coupling constant renormalization . in section 4 we will consider the free particle case with an @xmath29-box constraint and @xmath27-function potentials at the both boundaries as a toy model of rs1 .
the final expression is dependent on the two free parameters @xmath54 and @xmath55 which parametrize the bcs arising at both boundaries of @xmath29 box . in section 5 we will compute the fixed - energy amplitude for rs1 which depends on two free parameters @xmath54 and @xmath55 .
we will show in this section that there is no localized massless graviton on both branes . for positive - tension brane ,
however , the massless graviton can appear when the width of @xmath29 box is infinity , which is essentially identical to rs2 .
for the negative - tension brane our numerical calculation indicates there is no localized massless graviton regardless of the size of @xmath29 box . in final section a brief conclusion
in this section as a toy model of rs2 we will examine green s function for the free particle system defined on a half - line(@xmath56 ) with @xmath27-function potential whose hamiltonian is @xmath57 where @xmath58 is a free particle hamiltonian with the half - line constraint , _ i.e. _ @xmath59 of course the main problem in this model is how to compute the fixed - energy amplitude for @xmath58 .
once this is completed , one can derive a fixed - energy amplitude for @xmath60 by employing the schulman procedure described in the previous section .
we start with a fixed - energy amplitude @xmath61 $ ] for free particle without any constraint @xmath62 = \frac{e^{-\sqrt{2e } |x - y|}}{\sqrt{2e}}.\ ] ] then , the fixed - energy amplitude for @xmath58 can be computed as follows from @xmath61 $ ] .
first , we have to note that the fixed - energy amplitude for @xmath58 is dependent on bc at @xmath63 arising due to the half - line constraint . the usual dirichlet or neumann bcs at @xmath63
are properly incorporated into the path - integral formalism using @xmath27- and @xmath64-function potentials with infinite coupling constant@xcite ; @xmath65&=&\hat{g}_f[a , b ; e ] - \frac{\hat{g}_f[a , 0 ; e ] \hat{g}_f[0 , b ; e]}{\hat{g}_f[0^+ , 0 ; e ] } \\ \nonumber \hat{g}_f^n[a ,
b ; e]&=&\hat{g}_f[a , b ; e ] - \frac{\hat{g}_{f , b}[a , 0 ; e ] \hat{g}_{f , a}[0 , b ; e]}{\hat{g}_{f , ab}[0^+ , 0 ; e]}.\end{aligned}\ ] ] where the superscripts @xmath66 and @xmath67 stand for dirichlet and neumann respectively .
the explicit calculation shows @xmath68&=&\frac{e^{-\sqrt{2e } |x - y|}}{\sqrt{2e } } - \frac{e^{-\sqrt{2e}(|a| + |b|)}}{\sqrt{2e } } \\
\nonumber \hat{g}_f^n[a , b ; e]&=&\frac{e^{-\sqrt{2e } |x - y|}}{\sqrt{2e } } + \frac{\epsilon(a ) \epsilon(b ) e^{-\sqrt{2e}(|a| + |b|)}}{\sqrt{2e}}.\end{aligned}\ ] ] one can show easily @xmath69 and @xmath70 satisfy the following bcs ; @xmath71&=&\hat{g}_f^d[0 , b ; e ] = 0 \\
\nonumber \hat{g}_{f , b}^n[a , 0 ; e]&=&\hat{g}_{f , a}[0 , b ; e ] = 0.\end{aligned}\ ] ] then , the general fixed - energy amplitude for @xmath58 can be obtained by linearly combining @xmath69 and @xmath70 ; @xmath72 = \xi \hat{g}_f^n[a , b ; e ] + ( 1 - \xi ) \hat{g}_f^d[a , b ; e]\ ] ] where @xmath73 is real parameter parametrizing the bcs at the origin .
of course @xmath2 and @xmath52 represent the pure dirichlet and pure neumann bcs respectively .
another interesting case is @xmath74 , in which the contribution of neumann and dirichlet have an equal weighting factors .
since the fixed - energy amplitude @xmath75 is in general expressed in terms of eigenvalues @xmath76 and eigenfunctions @xmath77 of @xmath58 as follows @xmath78 = \sum_n \frac{\phi_n(a ) \phi_n^{\ast}(b)}{e - e_n},\ ] ] the @xmath79 case should correspond to the free particle case without any constraint at the origin . following schulman procedure
one can calculate the fixed - energy amplitude @xmath80 for @xmath60 from @xmath75 as follows ; @xmath81 \equiv \hat{g}^{\xi}[a , b ; e ] - \hat{g}_f^{\xi}[a , b ; e ] = \frac{4 \xi^2}{\frac{\sqrt{2e}}{v } - 2 \xi } \frac{e^{-\sqrt{2e}(|a| + |b|)}}{\sqrt{2e}}.\ ] ] at @xmath82 and @xmath79 the fixed - energy amplitudes are simply reduced to @xmath83&= & \frac{e^{-\sqrt{2e } |a -b|}}{\sqrt{2e } } + \frac{\frac{\sqrt{2e}}{v } + 2}{\frac{\sqrt{2e}}{v } - 2 } \frac{e^{-\sqrt{2e } ( |a| + |b|)}}{\sqrt{2e } } \\
\nonumber \hat{g}^{\xi=\frac{1}{2}}[a , b ; e]&= & \frac{e^{-\sqrt{2e } |a -b|}}{\sqrt{2e } } + \frac{e^{-\sqrt{2e } ( |a| + |b|)}}{\sqrt{2e } \left(\frac{\sqrt{2e}}{v } - 1\right ) } \end{aligned}\ ] ] and the corresponding bound state energies
@xmath84 arising due to the @xmath27-function potential are @xmath85 finally , let us consider @xmath53 case . in this case eq.([toy1modi ] ) shows that the modification term @xmath86 vanishes .
this means the @xmath27-function potential in eq.([toy1hamil1 ] ) does not play any important role .
in fact this is obvious if we consider the fact that at @xmath53 the hamiltonian @xmath87 describes the free particle system plus @xmath88 which makes the half - line constraint .
thus , the @xmath27-function potential in eq.([toy1hamil1 ] ) is absorbed to @xmath87 . even in this case , however , one can derive a non - trivial fixed - energy amplitude under the assumption that @xmath89 is infinite bare coupling constant by adopting the coupling constant renormalization . to show this explicitly we re - express the modification term @xmath86 as folllows ; @xmath90 = \lim_{\epsilon \rightarrow 0^+ } \frac{\hat{g}_f^{\xi = 0}[a , \epsilon ; e ] \hat{g}_f^{\xi = 0}[\epsilon , b ; e ] } { \frac{1}{v } - \hat{g}_f^{\xi = 0}[\epsilon , \epsilon ; e]}.\ ] ] expanding the denominator and numerator separately one can conclude @xmath91 = \frac{2}{\sqrt{2e } - v^{ren } } e^{-\sqrt{2e}(|a| + |b|)}\ ] ] where the renormalized coupling constant @xmath92 is defined as @xmath93 it is easy to show that @xmath92 has a same dimension with the bare coupling constant @xmath89 .
following the philosophy of renormalization we regard @xmath92 as a finite quantity . combining eq.([toy1dn ] ) and eq.([toy1dmodi2 ] )
we get finally @xmath94 = \frac{e^{-\sqrt{2e } |a - b|}}{\sqrt{2e } } + \frac{\frac{\sqrt{2e}}{v^{ren } } + 1 } { \frac{\sqrt{2e}}{v^{ren } } - 1 } \frac{e^{-\sqrt{2e}(|a| + |b|)}}{\sqrt{2e}}\ ] ] whose bound state energy is @xmath95 . in the next section
we will apply the analysis in this toy model to the rs2 scenario .
recently , one of the present authors computed the fixed - energy amplitude for rs2 at ref.@xcite which will be reviewed in this section briefly .
furthermore we will derive the general condition in the parameter space for the appearance of the localized massless graviton .
the hamiltonian for rs2 can be read from eq.([grafluc ] ) and ( [ potential ] ) easily ; @xmath96 of course , we can obtain the exact rs2 hamiltonian by letting @xmath97 , @xmath98 and @xmath99 , where @xmath100 is the radius of @xmath3 . in this section , however , we do not require them from the beginning . in other words we will compute the fixed - energy amplitude for arbitrary @xmath101 , @xmath102 , and @xmath89 when the half - line constraint ( @xmath103 ) is imposed .
this will give us the general condition for the localized graviton on the brane to be massless .
the half - line constraint makes @xmath104 in eq.([rs2hamil1 ] ) to be a following simple form ; @xmath105 where @xmath106 .
thus our half - line constraint @xmath103 is changed into @xmath107 . if @xmath108 , the euclidean propagator @xmath109 $ ] and the corresponding fixed - energy amplitude @xmath110 $ ] for hamiltonian ( [ rs2hamil2 ] ) are given at ref.@xcite ; @xmath111&=&\frac{\sqrt{a b}}{t } e^{-\frac{a^2 + b^2}{2 t } } i_{\gamma}\left ( \frac{a b}{t } \right ) \\
\nonumber \hat{g}_{>0}[a , b ; e]&= & 2 \sqrt{a b } i_{\gamma}\left ( \sqrt{\frac{e}{2 } } [ ( a + b ) - |a - b| ] \right ) k_{\gamma}\left ( \sqrt{\frac{e}{2 } } [ ( a + b ) + |a - b| ] \right)\end{aligned}\ ] ] where @xmath112 and @xmath113 are the usual modified bessel functions , and @xmath114 .
the main problem for the computation of the fixed energy amplitude for @xmath104 in eq.([rs2hamil2 ] ) is how to adopt an asymmetric constraint @xmath107 in terms of @xmath115 .
however , this is already explained at the previous section by introducing an infinite energy barrier . in the asymmetric barrier
the fixed - energy amplitude will be dependent on the bc at @xmath116 .
hence , the final form will be one parameter family type @xmath117 = \xi \hat{g}_0^n[a , b ; e ] + ( 1 - \xi ) \hat{g}_0^d[a , b ; e]\ ] ] where @xmath118 and @xmath119 are the fixed - energy amplitudes which obey the neumann and dirichlet bcs respectively at @xmath116 .
of course , @xmath118 and @xmath119 can be calculated from @xmath110 $ ] introducing @xmath27- and @xmath64-functions as we did in the previous section ; @xmath120&=&\hat{g}_{>0}[a , b ; e ] - \frac{\hat{g}_{>0}[a , c ; e ] \hat{g}_{>0}[c , b ; e ] } { \hat{g}_{>0}[c^+ , c ; e ] } \\ \nonumber \hat{g}_0^n[a , b ; e]&=&\hat{g}_{>0}[a , b ;
e ] - \frac{\hat{g}_{>0 , b}[a , c ; e ] \hat{g}_{>0 , a}[c , b ; e ] } { \hat{g}_{>0 , ab}[c^+ , c ; e]}.\end{aligned}\ ] ] inserting eq.([rs2schul ] ) into eq.([rs2gdgn ] ) one can derive the explicit forms of @xmath119 and @xmath118 ; @xmath121&=&\hat{g}_{>0}[a , b ; e ] - 2 \sqrt{a b } \frac{i_{\gamma}(\sqrt{2e } c)}{k_{\gamma}(\sqrt{2e } c ) } k_{\gamma}(\sqrt{2e } a ) k_{\gamma}(\sqrt{2e } b ) \\ \nonumber \hat{g}_0^n[a , b ; e]&=&\hat{g}_{>0}[a , b ; e ] + 2 \sqrt{a b } \frac{f_i(c , e)}{f_k(c , e ) } k_{\gamma}(\sqrt{2e } a ) k_{\gamma}(\sqrt{2e } b)\end{aligned}\ ] ] where @xmath122 the useful relation which will be used frequently is @xmath123 following schulman procedure it is straightforward to derive a fixed - energy amplitude for @xmath124 ; @xmath125 = \hat{g}_0^{\xi}[a , b ; e ] + \frac{\hat{g}_0^{\xi}[a , c ; e]\hat{g}_0^{\xi}[c , b ; e ] } { \frac{1}{v } - \hat{g}_0^{\xi}[c^+ , c ; e]}\ ] ] where @xmath126 is given in eq.([rs2onepara ] ) . the most convenient form of @xmath127 is @xmath128&= & \hat{g}_0^d[a , b ; e ] + \frac{\sqrt{a b}}{c v } \frac{k_{\gamma}(\sqrt{2e } a ) k_{\gamma}(\sqrt{2e } b ) } { k_{\gamma}^2(\sqrt{2e } c ) } \\
\nonumber & & \hspace{3.0 cm } \times \left [ \left ( \frac{\gamma - \frac{1}{2}}{2\xi c v } - 1 \right ) + \frac{\sqrt{2e}}{2 \xi v } \frac{k_{\gamma - 1}(\sqrt{2e } c ) } { k_{\gamma}(\sqrt{2e } c ) } \right]^{-1}.\end{aligned}\ ] ] since @xmath119 satisfies the usual dirichlet bc , _
i.e. _ @xmath129 = \hat{g}_0^d[c , b ; e ] = 0 $ ] , the fixed - energy amplitude on the brane is simply reduced to @xmath130 = \left [ \left ( \frac{\gamma - \frac{1}{2}}{2 \xi c } - v \right ) + \frac{\sqrt{2e}}{2 \xi } \frac{k_{\gamma - 1}(\sqrt{2e } c ) } { k_{\gamma}(\sqrt{2e } c ) } \right]^{-1}.\ ] ] if @xmath131 , @xmath132 $ ] becomes @xmath133 = \delta_0 + \delta_{kk}\ ] ] where @xmath134 of course , @xmath135 and @xmath136 represent the zero mode and the higher kaluza - klein excitations respectively .
this means that the condition for the localized graviton to be massless is @xmath137 at the rs limit @xmath138 , @xmath139 and @xmath140 this condition really holds at @xmath141 .
but for other valus of @xmath1 except @xmath53 one can also obtain the massless graviton by changing @xmath102 , @xmath101 , and @xmath89 appropriately to obey eq.([rs2condi ] ) . at the pure dirichlet bc case(@xmath53 ) eq.([rs2condi ] ) can not hold unless @xmath142 .
thus in this case we can get the massless graviton via a coupling constant renormalization as we did in the previous section . to show explicitly we have to re - write eq.([rs2schulman ] ) by introducing a positive infinitesimal parameter @xmath143 ; @xmath144 = \hat{g}_0^d[a , b ; e ] + \lim_{\epsilon \rightarrow 0^+ } \frac{\hat{g}_0^d[a , c + \epsilon ; e ] \hat{g}_0^d[c + \epsilon , b ; e ] } { \frac{1}{v } - \hat{g}_0^d[c + \epsilon , c + \epsilon ; e]}.\ ] ] explicit calculation shows that even in this case one can derive a massless graviton when @xmath145 , where @xmath92 is a renormalized coupling constant defined as @xmath146 the coupling constant renormalization procedure in rs2 is in detail explained in ref.@xcite . in this case the fixed - energy amplitude and the corresponding gravitational potential is exactly same with that of the original rs result when @xmath145 .
this result may provide us the compromise of the massless graviton with a small cosmological constant@xcite .
in this section as a toy model of rs1 we will examine green s function for the free particle system in a @xmath29 box(@xmath147 ) with @xmath27-function potentials at both end points .
the hamiltonian for this system is @xmath148 where @xmath149 the main problem in this toy model is of course to derive a fixed - energy amplitude @xmath150 for @xmath151 .
once @xmath150 is obtained , the fixed - energy amplitude @xmath152 for the total hamiltonian @xmath153 is straightforwardly obtained by performing the schulman procedure twice ; @xmath154 = \hat{{\cal g}}_0^{box}[a , b ; e ] - \frac{\hat{{\cal g}}_0^{box}[a , l ; e ] \hat{{\cal g}}_0^{box}[l , b ; e ] } { \frac{1}{v_2 } + \hat{{\cal g}}_0^{box}[l , l^- ; e]}\ ] ] where @xmath155 = \hat{g}_0^{box}[a , b ; e ] + \frac{\hat{g}_0^{box}[a , 0 ; e ] \hat{g}_0^{box}[0 , b ; e ] } { \frac{1}{v_1 } - \hat{g}_0^{box}[0^+ , 0 ; e]}.\ ] ] the fixed - energy amplitude @xmath150 for @xmath151 is also obtained directly from that for free particle on half - line , _ i.e. _ @xmath156 $ ] in eq.([toy1gen1 ] ) .
of course , the parameter @xmath54 represents the type of bc at @xmath63 .
then , the fixed - energy amplitude @xmath150 can be computed by introducing an infinite barrier at @xmath157 to the half - line constraint system . as we commented in section 2
the infinite barrier is introduced by @xmath27- and @xmath64-functions potential at @xmath157 with assumption that the coupling constant is infinite .
thus the final form of @xmath150 is dependent on the two parameters as follows ; @xmath158 = \xi_2 \hat{g}_0^{\xi_1 , n}[a , b ; e ] + ( 1 - \xi_2 ) \hat{g}_0^{\xi_1 , d}[a , b ; e]\ ] ] where @xmath159&=&\hat{g}_f^{\xi_1}[a , b ; e ] - \frac{\hat{g}_f^{\xi_1}[a , l ; e ] \hat{g}_f^{\xi_1}[l , b ; e ] } { \hat{g}_f^{\xi_1}[l , l^- ; e ] } \\ \nonumber \hat{g}_0^{\xi_1 , n}[a ,
b ; e]&=&\hat{g}_f^{\xi_1}[a , b ; e ] - \frac{\hat{g}_{f , b}^{\xi_1}[a , l ; e ] \hat{g}_{f , a}^{\xi_1}[l , b ; e ] } { \hat{g}_{f , ab}^{\xi_1}[l , l^- ; e]}.\end{aligned}\ ] ] of course the parameter @xmath55 in eq.([toy2box1 ] ) parametrizes the various bcs at @xmath157 .
explicit calculation shows @xmath160&= & \frac{1 } { \sqrt{2e } [ \xi_1 \cosh \sqrt{2e } l + ( 1 - \xi_1 ) \sinh \sqrt{2e } l ] } \\
\nonumber & & \times \bigg [ \xi_1 \bigg\{\sinh \sqrt{2e } ( l - |a - b| ) - \sinh \sqrt{2e } ( ( a + b ) - l ) \bigg\ } \\
\nonumber & & \hspace{1.0 cm } + ( 1 - \xi_1 ) \bigg\{\cosh \sqrt{2e } ( l - |a - b| ) - \cosh \sqrt{2e } ( ( a + b ) - l ) \bigg\ } \bigg ] \\ \nonumber \hat{g}_0^{\xi_1 , n}[a ,
b ; e]&= & \frac{1 } { \sqrt{2e } [ ( 1- \xi_1)\cosh \sqrt{2e } l + xi_1 \sinh \sqrt{2e } l ] } \\
\nonumber & & \times \bigg [ \xi_1 \bigg\{\cosh \sqrt{2e } ( l - |a - b| ) + \cosh \sqrt{2e } ( ( a + b ) - l ) \bigg\ } \\
\nonumber & & \hspace{1.0 cm } + ( 1 - \xi_1 ) \bigg\{\sinh \sqrt{2e } ( l - |a - b| ) + \sinh \sqrt{2e } ( ( a + b ) - l ) \bigg\ } \bigg].\end{aligned}\ ] ] inserting eq.([toy2explicit ] ) into eq.([toy2box1 ] ) we get @xmath161&= & \frac{1}{\sqrt{2e } } \bigg [ \left ( \mu(\xi_1 , \xi_2 ) + \mu(1 - \xi_1 , 1 - \xi_2 ) \right ) \cosh \sqrt{2e } ( l - |a - b| ) \\
\nonumber & & \hspace{1.0 cm } + \left ( \mu(\xi_1 , \xi_2 ) - \mu(1 - \xi_1 , 1 - \xi_2 ) \right ) \cosh \sqrt{2e } ( ( a + b ) - l ) \\
\nonumber & & \hspace{1.0 cm } + \left ( \nu(\xi_1 , \xi_2 ) + \nu(1 - \xi_1 , 1 - \xi_2 ) \right ) \sinh \sqrt{2e } ( l - |a - b| ) \\
\nonumber & & \hspace{1.0 cm } + \left ( \nu(\xi_1 , \xi_2 ) - \nu(1 - \xi_1 , 1 - \xi_2 ) \right ) \sinh \sqrt{2e } ( ( a + b ) - l ) \bigg]\end{aligned}\ ] ] where @xmath162 it is interesting to note the following special cases ; @xmath163&= & \frac{\cosh \sqrt{2e } ( l - |a - b| ) - \cosh \sqrt{2e } ( ( a + b ) - l ) } { \sqrt{2e } \sinh \sqrt{2e }
l } \\
\nonumber \hat{g}_0^{nn}[a , b ; e]&= & \frac{\cosh \sqrt{2e } ( l - |a - b| ) + \cosh \sqrt{2e } ( ( a + b ) - l ) } { \sqrt{2e } \sinh \sqrt{2e } l}\end{aligned}\ ] ] where the superscript @xmath164(or @xmath165 ) stands for dirichlet - dirichlet(or neumann - neumann ) bcs at @xmath63 and @xmath157 .
similar results to eq.([toy2special ] ) are found at ref.@xcite .
bound state energy spectrum is obtained from poles of @xmath166 and @xmath167 which indicates @xmath168 where @xmath169 is integer .
another interesting case is @xmath170 case where @xmath150 is simply reduced to the free particle case without any constraint , _ i.e. _ @xmath171 . inserting eq.([toy2box2 ] ) into eq.([toy2schul2 ] ) and subsequently eq.([toy2schul1 ] ) we get the final form of @xmath152 for the hamiltonian @xmath153 .
since final expression is too long , we do not describe it explicitly in this paper .
instead we will consider two special cases .
the first case we will consider is @xmath170 . in this case eq.([toy2schul1 ] ) and eq.([toy2schul2 ] ) yield @xmath172&= & \frac{e^{-\sqrt{2e}|a - b|}}{\sqrt{2e } } + \left ( \frac{\sqrt{2e}}{v_1 } - 1 \right)^{-1 } \frac{e^{-\sqrt{2e}(a + b)}}{\sqrt{2e } } \\
\nonumber & & - \left [ \left ( \frac{\sqrt{2e}}{v_1 } - 1 \right ) \left ( \frac{\sqrt{2e}}{v_2 } + 1 \right ) + e^{-2 \sqrt{2e } l } \right]^{-1 } \\
\nonumber & & \times \bigg [ \left ( \frac{\sqrt{2e}}{v_1 } - 1 \right ) \frac{e^{\sqrt{2e}(a + b)}}{\sqrt{2e } } + \frac{e^{\sqrt{2e}(a - b)}}{\sqrt{2e } } + \frac{e^{-\sqrt{2e}(a - b)}}{\sqrt{2e } } + \\
\nonumber & & \hspace{6.0 cm } \left ( \frac{\sqrt{2e}}{v_1 } - 1 \right)^{-1 } \frac{e^{-\sqrt{2e}(a + b)}}{\sqrt{2e } } \bigg].\end{aligned}\ ] ] the second case we will consider is @xmath173 . in this case @xmath150 is @xmath166 in eq.([toy2special ] ) .
as expected @xmath166 satisfies the usual dirichlet - dirichlet bcs ; @xmath174 = \hat{g}_0^{dd}[a , 0 ; e ] = \hat{g}_0^{dd}[l , b ; e ] = \hat{g}_0^{dd}[a , l ; e ] = 0.\ ] ] if , therefore , @xmath175 and @xmath176 are finite , we arrive at a conclusion @xmath177 = \hat{g}_0^{box}[a , b ; e]$ ] .
if however , @xmath175 and @xmath176 are infinite and unphysical bare quantities , one can arrive at different conclusion via the coupling constant renormalization as we have seen in section 2 and 3 . to adopt the coupling constant renormalization we introduce the infinitesimal positive constant @xmath143 as follows ;
@xmath178&\rightarrow&\hat{g}_0^{box}[\epsilon^- , \epsilon , e ] = 2 \epsilon - 2 \sqrt{2e } \coth \sqrt{2e } l \epsilon^2 + { \cal o } ( \epsilon^3 ) \\
\nonumber \hat{g}_0^{box}[l , l ; e]&\rightarrow&\hat{g}_0^{box}[l-\epsilon^- , l-\epsilon , e ] = 2 \epsilon - 2 \sqrt{2e } \coth \sqrt{2e } l \epsilon^2 + { \cal o } ( \epsilon^3 ) \\
\nonumber \hat{g}_0^{box}[0 , l ; e]&\rightarrow&\hat{g}_0^{box}[\epsilon , l-\epsilon , e ] = \frac{2 \sqrt{2e}}{\sinh \sqrt{2e } l } \epsilon^2 + { \cal o } ( \epsilon^3 ) \\
\nonumber \hat{g}_0^{box}[l , 0 ; e]&\rightarrow&\hat{g}_0^{box}[l-\epsilon , \epsilon , e ] = \frac{2 \sqrt{2e}}{\sinh \sqrt{2e } l } \epsilon^2 + { \cal o } ( \epsilon^3 ) \\ \nonumber \hat{g}_0^{box}[a , 0 ; e]&\rightarrow&\hat{g}_0^{box}[a , \epsilon , e ] = \frac{2 \sinh \sqrt{2e } ( l - a)}{\sinh \sqrt{2e } l } \epsilon + { \cal o } ( \epsilon^3 ) \\ \nonumber \hat{g}_0^{box}[a , l ; e]&\rightarrow&\hat{g}_0^{box}[a , l-\epsilon , e ] = \frac{2 \sinh \sqrt{2e } a}{\sinh \sqrt{2e } l } \epsilon + { \cal o } ( \epsilon^3 ) \\ \nonumber \hat{g}_0^{box}[0 , b : e]&\rightarrow&\hat{g}_0^{box}[\epsilon , b ; e ] = \frac{2 \sinh \sqrt{2e } ( l - b)}{\sinh \sqrt{2e } l } \epsilon + { \cal o } ( \epsilon^3 ) \\
\nonumber \hat{g}_0^{box}[l , b ; e]&\rightarrow&\hat{g}_0^{box}[l-\epsilon , b ; e ] = \frac{2 \sinh \sqrt{2e } b}{\sinh \sqrt{2e } l } \epsilon + { \cal o } ( \epsilon^3).\end{aligned}\ ] ] inserting eq.([toy2ddepsi ] ) into eq.([toy2schul2 ] ) and eq.([toy2schul1 ] ) , and defining the renormalized constants @xmath179 one can arrive at the following long expression after tedius calculation ; @xmath180&=&\hat{g}_0^{dd}[a , b ; e ] \\
\nonumber & & + 2 ( v_1^{ren } + \sqrt{2e } \coth \sqrt{2e}l)^{-1 } \frac{\sinh \sqrt{2e } ( l - a ) \sinh \sqrt{2e } ( l - b)}{\sinh^2 \sqrt{2e } l } \\
\nonumber & - & 2 \left [ ( v_1^{ren } + \sqrt{2e } \coth \sqrt{2e } l ) ( v_2^{ren } - \sqrt{2e } \coth \sqrt{2e } l ) + \frac{2e}{\sinh^2 \sqrt{2e } l } \right]^{-1 } \\
\nonumber & \times & \bigg[(v_1^{ren } + \sqrt{2e } \coth \sqrt{2e } l ) \frac{\sinh \sqrt{2e } a \sinh \sqrt{2e } b}{\sinh^2 \sqrt{2e }
l } \\
\nonumber & & + \sqrt{2e } \frac{\sinh \sqrt{2e } a \sinh \sqrt{2e } ( l - b ) + \sinh \sqrt{2e } ( l - a ) \sinh \sqrt{2e } b } { \sinh^3 \sqrt{2e } l } \\
\nonumber & & + 2e ( v_1^{ren } + \sqrt{2e } \coth \sqrt{2e } l)^{-1 } \frac{\sinh \sqrt{2e } ( l - a ) \sinh \sqrt{2e } ( l - b ) } { \sinh^4 \sqrt{2e } l } \bigg ] .
\end{aligned}\ ] ] in the next section we will apply the analysis in this toy model to the rs1 scenario .
in this section we will examine the fixed - energy amplitude for rs1 whose linear gravitational fluctuation is given in eq.([grafluc ] ) and eq.([potential ] ) .
the hamiltonian for rs1 can be read from these equations easily @xmath181 of course , the exact rs1 hamiltonian can be obtained by letting @xmath97 , @xmath98 , @xmath182 and @xmath16 . as we did in section 3 , however , we will try to examine the fixed energy amplitude for arbitrary parameter as much as possible .
next we impose @xmath183 is non - negative .
this means we use the single copy of @xmath3 as a bulk spacetime . in this sense
we have a same setting with that of @xmath9/cft . in this
setting hamiltonian @xmath104 in eq.([rs1hamil1 ] ) becomes @xmath184 where @xmath185 .
of course , the main problem is to compute the fixed - energy amplitude @xmath186 $ ] for @xmath104 in eq.([rs1hamil2 ] ) . from @xmath186
$ ] it is simple to derive the fixed energy amplitude for @xmath187 by applying the schulman procedure twice ; @xmath188 = \hat{{\cal g}}_0[a , b ; e ] - \frac{\hat{{\cal g}}_0[a , l ; e ] \hat{{\cal g}}_0[l , b ; e ] } { \frac{1}{v_2 } + \hat{{\cal g}}_0[l , l^- ; e]}\ ] ] where @xmath189 = \hat{g}_0[a , b;e ] + \frac{\hat{g}_0[a , c ; e ] \hat{g}_0[c , b ; e ] } { \frac{1}{v_1 } - \hat{g}_0[c^+ , c ; e]}.\ ] ] the fixed - energy amplitude for @xmath104 is also straightforwardly obtained from @xmath190 in eq.([rs2onepara ] ) by introducing an infinite barrier at @xmath157 again .
then , the amplitude is dependent on the two parameters @xmath54 and @xmath55 which represent the various bcs at @xmath63 and @xmath157 respectively ; @xmath191&\equiv & \hat{g}_0^{\xi_1 , \xi_2}[a , b ; e ] \\
\nonumber & = & \xi_2 \hat{g}_0^{\xi_1 , n}[a , b ; e ] + ( 1 - \xi_2 ) \hat{g}_0^{\xi_1 , d}[a , b ; e]\end{aligned}\ ] ] where @xmath192&=&\hat{g}_0^{\xi_1}[a , b ; e ] - \frac{\hat{g}_0^{\xi_1}[a , l ; e ] \hat{g}_0^{\xi_1}[l , b ; e ] } { \hat{g}_0^{\xi_1}[l , l^- ; e ] } \\ \nonumber \hat{g}_0^{\xi_1 , n}[a ,
b ; e]&=&\hat{g}_0^{\xi_1}[a , b ; e ] - \frac{\hat{g}_{0 , b}^{\xi_1}[a , l ; e ] \hat{g}_{0 , a}^{\xi_1}[l , b ; e ] } { \hat{g}_{0 , ab}^{\xi_1}[l , l^- ; e]}.\end{aligned}\ ] ] explicit calculation shows @xmath193&=&2\sqrt{ab } \bigg[i_{\gamma}(\sqrt{2e } min(a , b ) ) k_{\gamma}(\sqrt{2e } max(a , b ) ) \\
\nonumber & & \hspace{3.0 cm } + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } a ) k_{\gamma}(\sqrt{2e } b ) \bigg ] \\ \nonumber \hat{g}_0^{\xi_1 , d}[a ,
b ; e]&=&\hat{g}_0^{\xi_1}[a , b ; e ] - 2\sqrt{ab } g_d(\xi_1 , e ) [ i_{\gamma}(\sqrt{2e } a ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } a ) ] \\ \nonumber & & \hspace{3.0 cm } \times [ i_{\gamma}(\sqrt{2e } b ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } b ) ] \\
\nonumber \hat{g}_0^{\xi_1 , n}[a , b ; e]&=&\hat{g}_0^{\xi_1}[a , b ; e ] - 2\sqrt{ab } g_n(\xi_1 , e ) [ i_{\gamma}(\sqrt{2e } a ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } a ) ] \\
\nonumber & & \hspace{3.0 cm } \times [ i_{\gamma}(\sqrt{2e } b ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } b)]\end{aligned}\ ] ] where @xmath194 and , @xmath195 and @xmath196 are defined at eq.([fkfi ] ) . inserting eq.([rs1explicit ] ) into eq.([rs1twopara ] ) one can obtain the fixed - energy amplitude for hamiltonian ( [ rs1hamil2 ] ) ; @xmath197&=&\hat{g}_0^{\xi_1}[a , b ; e ] - 2 \sqrt{a b } g_{box}(\xi_1 , \xi_2 , e ) [ i_{\gamma}(\sqrt{2e } a ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } a ) ] \\
\nonumber & & \hspace{3.0 cm } \times [ i_{\gamma}(\sqrt{2e } b ) + g_1(\xi_1,e ) k_{\gamma}(\sqrt{2e } b)]\end{aligned}\ ] ] where @xmath198 thus inserting eq.([rs1xi1xi2 ] ) into eq.([rs1schul2 ] ) and subsequently eq.([rs1schul1 ] ) we can derive the fixed - energy amplitude @xmath199 $ ] .
the expression is too long to describe it here .
so , we rely on the numerical computation to check the occurrence of the localized massless graviton .
first , we will check the possibility for the appearance of the massless graviton at the brane located in @xmath116 .
1 shows @xmath200 $ ] when @xmath201 and @xmath202 .
of course we have taken rs limit , _
i.e. _ @xmath203 , @xmath204 , and @xmath205 . in order for the massless graviton to appear on the brane we need a pole in @xmath206 $ ] at @xmath207 .
this means the numerical value of @xmath200 $ ] should be non - zero and finite at @xmath208 .
1 indicates that the zero mass graviton appears only when l is infinitely large .
in fact , this limit is effectively rs2 scenario .
numerical calculation shows that there is no massless graviton on the brane located at @xmath157 regardless of @xmath209 if one chooses @xmath170 .
one may conjecture that the condition ( [ rs2condi ] ) for the appearance of the massless graviton in rs2 may be modified to @xmath210 for the appearance of the massless graviton on negative - tension brane .
2 shows @xmath211 $ ] where @xmath212 and @xmath55 is determined from eq.([rs1wrongcondi ] ) .
2 shows again that there is no massless graviton .
although we have not tested all kinds of possibility , our numerical results strongly suggest that there is no room for the appearance of the massless graviton in negative - tension brane regardless of @xmath54 , @xmath55 , and @xmath209 . of course
, one can derive a fixed - energy amplitude for @xmath173 case via the coupling constant renormalization in principle .
however , long expression for @xmath213 seems to make the calculation too tedious .
so , we do not describe the result of this case in this paper .
in this paper we have examined the localized gravity on the brane in rs brane - world scenario from the singular quantum mechanics . choosing a single copy of @xmath3 as a bulk spacetime we have shown that the fixed - energy amplitude for rs1 and rs2 are non - trivially dependent on the bcs . as a result
the fixed - energy amplitude for rs2 is dependent on the free parameter @xmath1 , which parametrize the bc at @xmath214 .
computing the fixed - energy amplitude explicitly one can derive the general criterion ( [ rs2condi ] ) for the appearance of the localized massless graviton on the brane when @xmath1 is arbitrary but non - zero .
when @xmath53 , the massless graviton is obtained via the coupling constant renormalization . in rs1 scenario
the final expression of the fixed - energy amplitude is dependent on the two free parameters @xmath54 and @xmath55 , which parametrize the various bcs at the end - points of @xmath29 box .
the appearance of the massless graviton is numerically tested by examing the pole at @xmath207 . for the positive - tension brane
our numerical test indicates that there is no massless graviton if the length of @xmath29 box is finite .
however , the infinite length of @xmath29 box makes the graviton localized on the positive - tension brane to be massless , which is effectively identical to the rs2 scenario . for the negative - tension brane
our numerical test shows that there is no massless graviton regardless of the length of @xmath29 box and choice of bcs .
we can consider the various extension for this paper .
firstly , one may include the bending effect of the brane in the computation . in this case , however , the final expression of the linearized fluctuation does not seem to be like schrdinger equation .
thus , we think the method used in ref.@xcite is more convenient than the technique of singular quantum mechanics to treat the bending effect
. one can extend the method presented in this paper to the higher - dimensional rs scenario@xcite . if one can find a singular brane solution in the higher - dimensional case
, one can apply the self - adjoint extension or a coupling constant renormalization to treat the higher - dimensional @xmath27-function potential .
of course , it is very interesting if we can find a singular solution in six dimension because two - dimensional @xmath27-function potential has various non - trivial properties such as scale anomaly and dimensional transmutation@xcite .
one may extend the present paper to the moving brane picture@xcite .
but it is unclear for us whether or not the path - integral solution is in this case analytically obtainable .
we think the most interesting problem is to understand the reason why there is no massless graviton in rs1 scenario .
this means that the gauge hierarchy problem is not compatible with the massless graviton problem .
thus , it seems to be important to compromise these two distinct phenomena .
99 l. randall and r. sundrum , large mass hierarchy from a small extra dimension , phys .
* 83 * ( 1999 ) 3370 [ hep - ph/9905221 ] .
l. randall and r. sundrum , an alternative to compactification , phys .
* 83 * ( 1999 ) 4690 [ hep - th/9906064 ] .
j. garriga and t. tanaka , gravity in the randall - sundrum brane world , phys .
* 84 * ( 2000 ) 2778 [ hep - th/9911055 ] .
m. j. duff and j. t. liu , complementarity of the maldacena and randall - sundrum pictures , phys .
lett . * 85 * ( 2000 ) 2052 [ hep - th/0003237 ] .
p. binetruy , c. deffayet and d. langlois , non - conventional cosmology from a brane universe , nucl . phys .
* b565 * ( 2000 ) 269 [ hep - th/9905012 ] . c. cski , m. graesser , c. kolda , and j. terning , cosmology of one extra dimension with localized gravity , phys
b462 * ( 1999 ) 34 [ hep - ph/9906513 ] . j. m. cline , c. grojean , and g. servant , cosmological expansion in the presence of an extra dimension , phys .
* 83 * ( 1999 ) 4245 [ hep - ph/9906523 ] .
j. e. kim , b. kyae , and h. m. lee , model for self - tuning the cosmological constant , phys .
lett . * 86 * ( 2001 ) 4223 [ hep - th/0011118 ] .
s. alexander , y. ling , and l. smolin , a thermal instability for positive brane cosmological constant in the randall - sundrum cosmologies [ hep - th/0106097 ] .
a. chamblin , s. w. hawking , and h. s. reall , brane - world black holes , phys . rev .
* d 61 * ( 2000 ) 065007 [ hep - th/9909205 ] .
r. emparan , g. t. horowitz , and r. c. myers , exact description of black holes on branes , jhep * 0001 * ( 2000 ) 007 [ hep - th/9911043 ] .
s. b. giddings , e. katz , and l. randall , linearized gravity in brane backgrounds , jhep * 0003 * ( 2000 ) 023 [ hep - th/0002091 ] .
j. maldacena , the large - n limit of superconformal field theories and supergravity , adv .
* 2 * ( 1998 ) 231 [ hep - th/9711200 ] .
d. k. park , h. s. kim , y. g. miao , and h. j. w. mller - kirsten , randall - sundrum scenario at nonzero temperature , phys .
* b 519 * ( 2001 ) 159 [ hep - th/0107156 ] .
d. k. park , h. s. kim , and s. tamaryan , nonvanishing cosmological constant of flat universe in brane - world scenario , to appear in phys .
b [ hep - th/0111081 ] .
r. p. feynman and a. r. hibbs , _ quantum mechanics and path integrals _ ( mcgraw - hill , new york , 1965 ) .
l. s. schulman , _ techniques and applications of path integrals _ ( wiley , new york , 1981 ) .
l. s. schulman , in _ path integrals from mev to mev _ , edited by m. c. gutzwiller , a. inomata , j. r. klauder , and l. streit ( world scientific , singapore , 1986 ) .
d. k. park , greens - function approach to two- and three - dimensional delta - function potential and application to the spin-1/2 aharonov - bohm problem , j. math .
* 36 * ( 1995 ) 5453 [ hep - th/9405020 ] .
r. jackiw , in _ m. a. bg memorial volume _ , edited by a. ali and p. hoodbhoy ( world scientific , singapore , 1991 ) .
a. z. capri , _ nonrelativistic quantum mechanics _
( benjamin / cummings , menlo park , 1985 ) .
s. albeverio , f. gesztesy , r. hoegh - krohn , and h. holden , _ solvable models in quantum mechanics _
( springer , berlin , 1988 ) . c. grosche , @xmath27-function perturbations and boundary problems by path integration , ann .
physik * 2 * ( 1993 ) 557 [ hep - th/9302055 ] . c. grosche , @xmath64-function perturbations and neumann boundary - conditions by path integration , j. phys . *
a28 * ( 1995 ) l99 [ hep - th/9402110 ] .
d. k. park , proper incorporation of self - adjoint extension method to green s function formalism : one - dimensional @xmath64-function potential case , j. phys .
* a29 * ( 1996 ) 6407 [ hep - th/9512097 ] .
d. k. park and s. k. yoo , propagators for spinless and spin-1/2 aharonov - bohm - coulomb systems , ann phys .
* 263 * ( 1998 ) 295 [ hep - th/9707024 ] .
d. k. park and s. tamaryan , compromise of localized graviton with a small cosmological constant in randall - sundrum scenario , phys .
b532 * ( 2002 ) 305 [ hep - th/0108068 ] . c. grosche and f. steiner , _ handbook of feynman path integrals _ ( springer , berlin , 1998 ) .
f. leblond , r. c. myers , and d. j. winters , brane world sum rule and @xmath9 soliton [ hep - th/0107034 ] . c. p. burgess , j. m. cline , and n. r. constable , and h. firouzjahi , dynamical stability of six - dimensional warped brane - worlds , jhep * 0201 * ( 2002 ) 014 [ hep - th/0112047 ] .
g. kofinas , new perspectives on moving domain wall in @xmath215 space , nucl .
* b 622 * ( 2002 ) 347 [ hep - th/0103045 ] . | the graviton localized on the @xmath0-brane is examined in randall - sundrum brane - world scenario from the viewpoint of one - dimensional singular quantum mechanics . for the randall - sundrum single brane scenario the one - parameter family of the fixed - energy amplitude
is explicitly computed where the free parameter @xmath1 parametrizes the various boundary conditions at the brane . the general criterion for the localized graviton to be massless
is derived when @xmath1 is arbitrary but non - zero . when @xmath2 , the massless graviton is obtained via a coupling constant renormalization .
for the two branes picture the fixed - energy amplitude is in general dependent on the two free parameters .
the numerical test indicates that there is no massless graviton in this picture . for the positive - tension brane ,
however , the localized graviton becomes massless when the distance between branes are infinitely large , which is essentially identical to the single brane picture .
for the negative - tension brane there is no massless graviton regardless of the distance between branes and choice of boundary conditions . |
consider a system evolving with a hamiltonian @xmath0 for a time @xmath1 , and then backwards with a slightly different hamiltonian @xmath2 : a ` loschmidt echo ' @xcite experiment ( see , in the present context , @xcite ) .
a particular case is when at time @xmath1 , the system is subjected to a rapid ` kick ' with hamiltonian @xmath3 , and subsequently is evolved back for a time @xmath1 with @xmath4 . in classical mechanics , the returning trajectory is the time - reversed of the initial trajectory , except that its beginning has been perturbed .
it is clear then that if the system is chaotic , and the time @xmath1 is long enough , the return point differs from the initial point by a magnitude @xmath5 , where @xmath6 is the lyapunov exponent that does not depend on the nature of the kick ( for @xmath7 ) , and @xmath8 is a constant that does .
an analogous result holds for a quantum system if one prepares it in a localized state , and evolves it with a hamiltonian and back with a slightly perturbed hamiltonian .
if one measures appropriately the overlap between the initial and the returning wavefunction , there is a regime where this difference scales exponentially in time , with the ( quantum lyapunov ) exponent independent of the perturbation between the outgoing and ingoing hamiltonians @xcite .
several years ago , jalabert and pastawski @xcite showed that for a semiclassical system , and for a given range of perturbation and times , this quantum exponent corresponds to the classical one ( although an ` annealed ' version thereof , see discussion in @xcite ) .
consider a hermitean operator @xmath9 , which we shall assume is concentrated in some region of phase space , as seen , for example , in its coherent - state or wigner representation .
we introduce a measure of the ` fidelity ' @xmath10 $ ] , comparing the operator @xmath9 with the one @xmath11 transformed by loschmidt echo trajectory . for a ` kick ' perturbation , we have : @xmath12=\mbox{tr } \left\{\left [ e^{i\frac{t}{\hbar}h } e^{i \frac{\delta}{\hbar } b } e^{-i\frac{t}{\hbar}h } \right ] a \left[e^{i\frac{t}{\hbar}h } e^{i \frac{\delta}{\hbar } b } e^{-i\frac{t}{\hbar}h } \right]^\dag a \right\ } = \mbox{tr } \left\ { a e^{i\frac{t}{\hbar}h } e^{i \frac{\delta}{\hbar } b } e^{-i\frac{t}{\hbar}h } a e^{i\frac{t}{\hbar}h } e^{-i \frac{\delta}{\hbar } b } e^{-i\frac{t}{\hbar}h } \right\}\ ] ] if @xmath13 we clearly get @xmath14 $ ] . putting @xmath15 and developing in powers of @xmath16 ( linear response ) @xmath17
we get : @xmath18 - \mbox{tr } [ a_{tran}a ] = - \frac{\delta^2}{\hbar^2 } \mbox{tr } \left\ { b(t ) a b(t ) a \right\ } + \frac{\delta^2}{\hbar^2 } \mbox{tr } \left\ { b^2(t ) a^2 \right\}= -\frac{\delta^2}{2\hbar^2 } \mbox{tr } \left\ { [ b(t),a]^2 \right\}\ ] ] the measure we are seeking is thus @xcite @xmath19 ^ 2 \right\ } \label{crochet}\ ] ] which is positive since it is minus the trace of an anti - hermitean operator squared .
we need the operator @xmath9 to be concentrated , in order to localize our measure in a region of phase space , and to make the trace converge .
two choices have been proposed .
a traditional one is is to consider a localized wavefunction @xmath20 @xcite and : @xmath21 ( here in the ` kick ' version , although more commonly with a continuous perturbation @xmath22 ) .
a more recent one @xcite , that lends itself better for the study of an extensive system at temperature @xmath23 , is obtained by choosing a smooth ( possibly unlocalized ) hermitean operator @xmath24 and to render it concentrated on an energy shell , at least in the thermodynamic limit , by putting in ( [ crochet ] ) : @xmath25 the symbol @xmath26 means that @xmath9 is normalized by @xmath27 .
the second term only contributes a two - time function and its precise form is not relevant beyond canceling constants , we keep it for convenience here .
( the regularization strategy of splitting the equilibrium density operator in factors here four of them is standard in chemical physics , see @xcite . ) for this form , in some cases there is a lyapunov regime : @xmath28 in this context , it has been shown that there is a bound to chaos @xcite of purely quantum nature : @xmath29
in the semiclassical case , where there is no apparent limitation , the question immediately arises of how does quantum mechanics intervene .
we start by considering a hamiltonian system of the form @xmath30 in the semiclassical limit @xmath31 . for the bound to be effective ,
we need the product @xmath32 to be finite , and this may only be achieved in the semiclassical limit for very low temperatures , corresponding to the lowest classical energies of the system .
it is clear that if the classical ground state does not have dimension larger than two , there will be no chaos : the system will sit on the classical ground state and perform vibrations and quantum fluctuations around it . in order to render the system interesting from our point of view
, we consider a potential such that the minimum of @xmath33 is a manifold of dimension @xmath34 .
thus , even at the lowest energies , the system will have long - distance motion ` at the bottom of the well ' .
since we are interested in such a motion , we may concentrate on it by working with the quantum mechanics in the curved space of minima of the potential ( a ` quantum wire ' @xcite or ` quantum conducting surface ' problem ) .
the hamiltonian restricted to that manifold is constructed as follows : we set up a system of @xmath34 curvilinear coordinates @xmath35 of the manifold of minima , and the associated momenta @xmath36 , @xmath37 @xcite .
let the metric tensor be @xmath38 so that length element @xmath39 .
the hamiltonian is proportional to the laplace beltrami operator @xcite @xmath40 where @xmath41 and @xmath42 .
the r.h.s .
shows that the spectrum is positive semidefinite , including zero if and only if the system is bounded
. the imaginary - time evolution with hamiltonian ( [ hamquan ] ) is just diffusion on the constrained surface .
the lowest eigenvalues of ( [ hamquan ] ) indeed correspond to the inverse thermalization times associated with this diffusion process , the eigenvalue zero to the flat invariant measure .
equation ( [ lyap2 ] ) can be viewed as an expectation of various observables ( correlations and responses ) of a process with four diffusion stretches lasting @xmath43 .
a two - dimensional version of the constant negative curvature surface has been extensively studied as the simplest , and one of the few solvable chaotic systems , as was discussed in detail by balazs and voros @xcite , @xmath44 \label{hyper}\ ] ] where @xmath45 is the ` radius ' of the hyperboloid .
the classical lyapunov exponent is @xmath46 .
expectation values may be computed analytically , because eigenbases are known .
we shall not do this interesting calculation here , but keep the example in mind throughout .
* classical limit * in the classical limit , the hamiltonian is given by the kinetic energy of the constrained motion : @xmath47 and the lagrangian : @xmath48 where @xmath49 is the inverse of @xmath50 .
the geodesic lengths are then given by the extrema of : @xmath51 because @xmath52 is a constant of motion , the lagrange equations for geodesics and for the action lead to proportional equations .
we hence conclude that the classical trajectories are _ geodesics _ in the coordinate space of the @xmath35 , traversed with a velocity that is proportional to the square root of the ( purely kinetic ) energy .
if the system is bounded , we may compute the equilibrium quantities via : @xmath53 @xmath54 is the energy per degree of freedom .
it is interesting to observe that in going from ( [ cosa ] ) to its low - energy version ( [ hamcla ] ) , a kind of reparametrization invariance appears . given two points in _ configuration _ space and two times @xmath55 and @xmath56 ,
there is a trajectory in space that will join the points in time @xmath55 and another in @xmath56 just by running over the same path at two different speeds .
note that this property was absent in the original system ( [ cosa ] ) .
time - reparametrization plays an important role in quantum glass models that saturate the bound ( [ bound ] ) @xcite * classical lyapunov exponent and geodesic deviation * let us now assume that the ground state manifold is such that the geodesics are _ chaotic_. as mentioned above , an example is the constant negative curvature surface in two dimensions @xcite .
two nearby trajectories _ in phase space _ @xmath57 and @xmath58 will separate exponentially @xmath59(t ) } \propto \;\ ; e^{\lambda t}\ ] ] it is easy to write a differential equation for the evolution of the phase - space lyapunov vector @xmath60 , which will be as usual first order in time . in this particular case ,
one may also easily eliminate @xmath61 and obtain a _
second order _ differential equation for the @xmath62 ( the ` geodesic deviation ' ) exclusively .
we need here only to remark that it is of the form @xcite @xmath63 , with @xmath64 quadratic in @xmath65 and linear in both the riemann curvature tensor @xmath45 and @xmath66 . in a chaotic system
, the purely spatial separation is expected to scale exponentially with the geodesic length @xmath67 along the trajectory ( fig .
[ remsketch ] ) @xmath68 where @xmath69 , the _ lyapunov length _ , is the typical distance travelled for the separation of two nearby trajectories to be multiplied by @xmath70 .
we expect that in a system with a good thermodynamic limit , @xmath69 scales with the size as @xmath71 with @xmath72 a length that has a finite thermodynamic limit .
we may understand this scaling as follows : if we consider the cartesian product of two independent chaotic systems with the same lyapunov length , and consider pairs of trajectories for both systems starting from separations @xmath73 .
after a length @xmath72 travelled by _ each _ , their separations will achieve @xmath74 for both .
but this happens in a combined length travelled that is now @xmath75 , while the total separation has increased only by @xmath70 .
consider now the usual lyapunov exponent associated with time .
it is clear that the same pair of nearby geodesics traversed at twice the speed will diverge twice as fast . because the energy is quadratic in the speed , we conclude then that : @xmath76 the right hand side confirms that @xmath72 has a good thermodynamic limit if @xmath6 also does .
( the largest lyapunov value is expected to be @xmath77 in @xmath78 , although there have been some debate about how general this is , see @xcite . )
a quantum system will have a thermal de broglie length defined as : @xmath79 this is a quantity _ per degree of freedom_. the wavelength along a trajectory in full phase - space will be @xmath80 times that , i.e. @xmath81 . let us compare lyapunov and debroglie lengths : @xmath82 rearranging , we find that both lengths are comparable if : @xmath83 where we omitted the numerical factors in the last , dimensional comparison .
the bound ( [ bound ] ) has hence the meaning that when the thermal debroglie length is of the order or larger than the characteristic ` chaos ' length @xmath84 , the semiclassical estimate breaks down and chaoticity is bounded , see figure [ remsketch ] .
it is hard to scatter a wavefront along a distance shorter than its wavelength .
let us make this argument more precise .
in order to do this , we make a semiclassical calculation in a way that closely resembles the one of jalabert and pastawski @xcite .
we sketch the calculation here and leave the details for the appendix .
we have to compute ( [ lyap2 ] ) , the first term of which is @xmath85 we rescale quantities using : @xmath86 the value of @xmath87 is the typical distance traveled by a classical particle at temperature @xmath88 , it is much longer than the thermal length . to recap , we have three quantities , @xmath72 , @xmath89 and @xmath90 : ` spatial divergence ' , thermal debroglie and trajectory lengths , respectively
. we shall now write the real and imaginary time propagations as : @xmath91 } = e^{-\frac{i\hat t } { \ell } \bar h } \hspace{.7 cm } ; \hspace{.7 cm } \mbox{imag .
= e^{-{\frac \beta 4 } h}=e^{+
\frac{\ell^2}{32\pi^2 } g^{-\frac12 } \partial_i g^{ij } g^{\frac12 } \partial_j } = e^{-d}\ ] ] which implicitly define @xmath92 and @xmath93 .
we recognize the real time propagations as semiclassical , with @xmath89 playing the role usually played by @xmath94 . as to the imaginary propagations , it is diffusion in curved space with ` noise ' of small amplitude @xmath95 .
let us introduce a basis : @xmath96 we now use the fact that the real time propagations are semiclassical , and that the diffusions are over very small @xmath97 lengths ( fig .
[ trajectory ] ) .
we first write the short diffusion episodes introducing @xmath77 lengths @xmath98 : @xmath99 and compute ( appendix ) , to leading order in @xmath89 : @xmath100 and similarly for @xmath101 ( one may omit constant prefactors , as they appear in the denominator too ) .
note that the @xmath102 play the role of velocities in the lagrangian @xmath52 .
for the integral , we have : @xmath103 next , we introduce the van vleck formula for the semiclassical propagator : @xmath104 we have to take into account the shifts of @xmath105 and @xmath106 in the brackets . using the fact that the derivative of the classical action @xmath107 with respect to the coordinate gives the momentum , we may put : @xmath108 where @xmath109 denote the classical momentum at the time corresponding to @xmath110 and @xmath111 , respectively .
the shifts @xmath98 in the @xmath112 are negligible , since they are of @xmath97 .
we then have : @xmath113 we next calculate the integrals over @xmath114 by saddle point evaluation .
derivatives of the action @xmath107 with respect to @xmath115 give the momenta at the end and at the beginning of the subsequent classical path , in such a way that the classical trajectory ` bounces back ' three times , retracing its steps , see fig . [
trajectory ] . _
note that the saddle point does not fix the endpoint of the trajectory ( or , equivalently , the momentum ) , which is a zero - mode at the classical level .
_ on the other hand , we are not yet integrating over the initial position . denoting @xmath116 , @xmath117 , and @xmath118 , we compute the gaussian integral around the saddle point , taking care of zero - modes and of jacobians associated with changes of variables . after a few steps ( see the appendix ) ,
we get @xmath119 \right\ } \label{exp0}\end{aligned}\ ] ] where we have defined : @xmath120_{ij } ( q_i , q_f , { \hat t } ) \label{lambda}\ ] ] the leading term is cancelled by the second term in ( [ lyap2 ] ) .
the expectations with two @xmath121 contribute only to connected two - time correlations that have decayed in the time regime that we are studying because chaos leads to cancellations ( see the appendix ) .
we are left with the expectation containing four @xmath121 , which , after rearrangement , may be expressed in terms of classical poisson brackets . assembling all the pieces together , we get the final , purely classical , result : @xmath122 where we have defined the classical adimensional hamiltonian @xmath123 the meaning of this formula is clear
. we are summing over trajectories of different ( rescaled ) speeds , starting at all points , and weighted with the gibbs measure .
the time is the rescaled time @xmath90 , which accounts for the absence of the temperature in the exponent , so that this expression is , for given @xmath90 , purely geometrical .
this is as far as we can go for a system with few degrees of freedom in the canonical framework .
* canonical vs. micro - canonical * for systems with few degrees of freedom , quantities have thermal fluctuations , and there is no single velocity associated with all trajectories in ( [ exp1 ] ) .
let us consider for simplicity the case in which the measure is concentrated on a single energy , as in the thermodynamic limit .
introducing a delta function and rescaling with the energy density @xmath124 @xmath125 we have : @xmath126 ( we have dropped the subindex @xmath127 in the arguments @xmath128 ) . putting all together we may write : @xmath129+n{\cal{s } } } \;\ ; { \bf \langle } \hspace{-.07cm}{\bf \langle } \{b_{f},w_i\}^2{\bf \rangle}\hspace{-.07 cm } { \bf \rangle}(2 \pi \hat t ) } { \int d { \cal{e}}_o e^{-n { \cal{e}}_o+\frac{n-1}2 \ln [ 2 { \cal{e}}_o]+n{\cal{s } } } } \end{aligned}\ ] ] where @xmath130 is a purely geometric contribution to the entropy , and @xmath131 the delta function imposes unit speed .
it is now clear that if the number of degrees of degrees of freedom is large , and the system has a thermodynamic limit , a single energy density @xmath132 will dominate and the argument of the poisson bracket in ( [ vas ] ) will simply be @xmath133 . here
the limit of large @xmath78 has to be made _ before _ the limit of large @xmath1 , otherwise the poisson bracket term will distort the thermodynamic measure . for this value , we define the reference lyapunov exponent @xmath134 , a purely geometric quantity @xmath135 which implies that @xmath136 , the spatial geodesic separation we defined above , and @xmath137 which is the classical estimate we made at the beginning , i.e. the expected result that semiclassical and classical lyapunov exponents coincide . .
[ trajectory1 ] @xmath138 * quantum length @xmath89 and characteristic length(s ) of the geometry * the situation we have is sketched in fig [ trajectory1 ] . for short de broglie wavelengths ,
the adimensional lyapunov exponent @xmath139 is proportional to @xmath140 .
if the system has a single characteristic length @xmath45 , as for example the case of the pseudosphere , whose only parameter is its ` radius ' @xmath45 , then the classical lyapunov exponent scales with this length , and there is only the question of how does the thermal debroglie length compare with the geometric one . as we consider longer quantum wavelengths , the semiclassical approximation becomes less good , and in the regime where the wavelength becomes large with respect to all the geometrical features completely breaks down .
it is not even clear that a lyapunov regime exponentially dependent on @xmath1 and independent of @xmath141 and @xmath24 exists for the loschmidt echo , and one would expect that chaoticity becomes smaller . in the example of the motion on the pseudosphere ,
the hamiltonian has a continuous spectrum starting from a lowest eigenvalue @xcite @xmath142 the same system bounded to a large region , will have a similar spectrum ( this time not continuous ) , plus a ` flat ' state of eigenvalue zero @xmath143 .
let us compare the thermal energy with the lowest nonzero eigenvector : @xmath144 we see what happens when @xmath89 becomes larger than @xmath72 : the factor in the exponent of @xmath145 projects onto the lowest , non - propagating mode .
we may think of the evolution as diffusion over a long time @xmath146 , larger than the ergodic time of the diffusive system .
we may estimate that the result of @xmath147 for @xmath148 is , up to exponentially small @xmath149 : @xmath150 where @xmath151 denotes flat averages over the manifold .
again , it seems doubtful that there is chaos in this regime @xmath152 ( the gray region ) , and it is not even clear whether a lyapunov exponent independent of @xmath153 , governing an exponential relaxation , still exists .
is it possible to have chaos all the way down to zero temperature ?
it is easy to construct a ( very artificial ) model that does .
for this we have to use the fact , already mentioned above , that the lyapunov exponent of the cartesian product of two weakly or non - interacting systems is just the largest of the exponents of the individual systems .
consider then a @xmath154 dimensional system composed of @xmath78 hyperbolic systems like ( [ hyper ] ) , each with typical ` radius ' @xmath155 .
the total lyapunov of the combined system will be composed of the envelope of the highest of the curves of fig .
[ trajectory1 ] , each one with a characteristic @xmath156 .
in such a case , one may imagine that as one moves into the deeper quantum regime ( longer thermal lengths ) , the wave propagation ignores the shorter geometric lengths but is scattered by the longest . in a system with regions with all levels of curvature , the quantum lyapunov
would then be governed by the smaller curvature of the order of the inverse debroglie length , and this is compatible with @xmath139 stabilizing as the thermal length becomes longer and longer .
the example of critical opalescence comes to mind : at exactly the critical point there are fluctuations of density and hence of refractive index at all scales , and the fluid looks white because it scatters all wavelengths .
it is tempting to compare the situation with systems like syk that saturate the bound at exactly zero temperature and have a critical , gapless point there .
the example of a particle moving on a curved surface , or of a ` bosonic ' system with a classical ground - state manifold of high dimension , is a context where one may picture the action of the quantum bounds on chaos in an intuitive way
. it would be interesting to discuss the effect of ` driving ' such a system , and computing transport and dissipation coefficients , and the limits quantum mechanics places upon them .
another interesting exercise would to obtain a bound that applies to the loschmidt echo in the more conventional setting of a wave packet with typical energy @xmath157 , rather than in the canonical ensemble .
it seems clear that some such bound should exist , and would be useful for systems with few degrees of freedom .
we next calculate the integrals over @xmath114 by saddle point evaluation .
we get : @xmath161 as mentioned above , derivatives of the action @xmath107 with respect to @xmath115 give the momenta at the end and at the beginning of the subsequent classical path , in such a way that the classical trajectory ` bounces back ' three times , retracing its steps see fig .
[ trajectory ] .
the speed is the same in modulus in the four trajectories , but is as yet undetermined . or , in other words , the saddle point does not fix the endpoint of the trajectory , which is then a zero - mode at the classical level . denoting @xmath116 , @xmath117 , and @xmath118 ,
we compute the integral around the saddle point . rather than expressing it as a path integral
, we write an expression for the gaussian fluctuations around each trajectory . for this
we expand @xmath162 \nonumber \\
a^+\left ( q_1-\ell \frac{y_2}2,q_8 \right ) & \sim & a^+(q_i , q_f)=a^-(q_f , q_i ) \end{aligned}\ ] ] where , as usual , linear terms will cancel at the saddle point , and second derivatives are evaluated on the saddle - point values @xmath163 .
we have kept the leading order in the @xmath112 .
note that @xmath164 = |a(q_i , q_f)|^2\ ] ] we now develop the classical action up to second order in @xmath165 , and @xmath166 in the exponent .
we also need to integrate over the multiple saddle points , each with a given @xmath167 . the gaussian fluctuations naturally contain zero - modes associated to the multiplicity of saddles , we must supress them with a factor @xmath168 , otherwise we would be double - counting the integral over @xmath167 .
all in all , we get : @xmath169 \right\ } \end{aligned}\]]note the cancellations of terms with second derivatives with respect to a single vector @xmath170 , etc .
, and also the absence of terms with @xmath171 due to the @xmath168 .
we now integrate over @xmath105 and @xmath106 , which transforms @xmath159 into the adimensional hamiltonian ( [ adim ] ) in the exponent , to obtain : @xmath172 \right\ } \end{aligned}\ ] ] the final step is to change integration variables from @xmath173 to @xmath174 : @xmath175 so that : @xmath176 \;\ ; \exp \left\{i{\hat t}\left[\lambda^{-1}_{ab } \hat q_{4a } \hat q_{5b } \right ] \right\ } \label{exp}\end{aligned}\ ] ] note that the normalization is a constant independent of @xmath177 .
@xmath179 expectations of terms of the form @xmath181 ( comma denotes derivative ) are proportional to @xmath182 .
they are zero at long times for classical reasons : the basis that diagonalizes @xmath183 varies wildly from trajectory to trajectory in a chaotic system .
this is how the two - time connected correlation function vanishes at the times we are interested in .
similar two - time contributions come from the second term in ( [ lyap2 ] ) , and are negligible for the same reason .
@xmath179 finally , the relevant contribution , of @xmath97 comes from : @xmath184 up to a total derivative in @xmath185 and @xmath167 this yields , neglecting rapidly oscillating terms @xmath186 where we have used @xmath187 i would like to thank c. bachas , d bernard , g biroli and d. reichman for helpful discussions , and j maldacena for relevant references .
this work was supported by a grant from the simon foundation ( no 454943 ) .
100 maldacena , juan , stephen h. shenker , and douglas stanford .
`` a bound on chaos . ''
arxiv preprint arxiv:1503.01409 ( 2015 ) .
hayden , patrick , and john preskill . `` black holes as mirrors : quantum information in random subsystems . ''
journal of high energy physics 2007.09 ( 2007 ) : 120 .
sekino , yasuhiro , and leonard susskind . `` fast scramblers . ''
journal of high energy physics 2008.10 ( 2008 ) : 065 .
campisi , michele , and john goold .
`` thermodynamics of the quantum butterfly effect . ''
arxiv preprint arxiv:1609.07848 ( 2016 ) . for experimental evidence ,
see : levstein , patricia r. , gonzalo usaj , and horacio m. pastawski .
`` attenuation of polarization echoes in nuclear magnetic resonance : a study of the emergence of dynamical irreversibility in many - body quantum systems . ''
the journal of chemical physics 108.7 ( 1998 ) : 2718 - 2724 .
miyaji , masamichi .
`` butterflies from information metric . ''
arxiv preprint arxiv:1607.01467 ( 2016 ) .
peres , asher .
`` stability of quantum motion in chaotic and regular systems . ''
physical review a 30.4 ( 1984 ) : 1610 .
jalabert , rodolfo a. , and horacio m. pastawski . `` environment - independent decoherence rate in classically chaotic systems . '' physical review letters 86.12 ( 2001 ) : 2490 .
see : gorin , thomas , et al .
`` dynamics of loschmidt echoes and fidelity decay . ''
physics reports 435.2 ( 2006 ) : 33 - 156 , and references therein .
vanicek , jiri , and eric j. heller .
`` semiclassical evaluation of quantum fidelity . ''
physical review e 68.5 ( 2003 ) : 056208 .
silvestrov , p. g. , j. tworzydo , and c. w. j. beenakker .
`` hypersensitivity to perturbations of quantum - chaotic wave - packet dynamics . ''
physical review e 67.2 ( 2003 ) : 025204 .
berenstein , david , and antonio m. garca - garca .
`` universal quantum constraints on the butterfly effect . ''
arxiv preprint arxiv:1510.08870 ( 2015 ) .
jensen , h. , and h. koppe .
`` quantum mechanics with constraints . ''
annals of physics 63.2 ( 1971 ) : 586 - 591 .
a. kitaev , a simple model of quantum holography .
`` http://online.kitp.ucsb.edu/online/entangled15/kitaev/,http : //online.kitp.ucsb.edu / online / entangled15/kitaev2/. talks at kitp , april 7 , 2015 and may 27 , 2015 .
maldacena , juan , and douglas stanford . ' ' comments on the sachdev - ye - kitaev model . " arxiv preprint arxiv:1604.07818 ( 2016 ) .
balazs , n. l. , and a. voros .
`` chaos on the pseudosphere . ''
physics reports 143.3 ( 1986 ) : 109 - 240 .
let us note in passing that there is some arbitrariness in terms of order @xmath94 that depend on how we fix the constraints , here for simplicity we choose the laplace beltrami operator : see ikegami , mitsuhiro , et al . ``
quantum mechanics of a particle on a curved surface comparison of three different approaches . ''
progress of theoretical physics 88.2 ( 1992 ) : 229 - 249 .
sachdev , subir , and jinwu ye .
`` gapless spin - fluid ground state in a random quantum heisenberg magnet . '' physical review letters 70.21 ( 1993 ) : 3339 .
+ parcollet , olivier , and antoine georges .
`` non - fermi - liquid regime of a doped mott insulator . '' physical review b 59.8 ( 1999 ) : 5341 .
+ polchinski , joseph , and vladimir rosenhaus .
`` the spectrum in the sachdev - ye - kitaev model . ''
arxiv preprint arxiv:1601.07768 ( 2016 ) .
livi , r. , a. politi , and s. ruffo .
`` distribution of characteristic exponents in the thermodynamic limit . ''
journal of physics a : mathematical and general 19.11 ( 1986 ) : 2033 .
hartle , james b. gravity : an introduction to einstein s general relativity .
shenker , stephen h. , and douglas stanford . `` black holes and the butterfly effect . ''
arxiv preprint arxiv:1306.0722 ( 2013 ) .
miller , william h. , steven d. schwartz , and john w. tromp . `` quantum mechanical rate constants for bimolecular reactions . ''
the journal of chemical physics 79.10 ( 1983 ) : 4889 - 4898 . | we discuss the quantum bound on chaos in the context of the free propagation of a particle in an arbitrarily curved surface at low temperatures .
the semiclassical calculation of the lyapunov exponent can be performed in much the same way as the corresponding one for the ` loschmidt echo ' .
the bound appears here as the impossibility to scatter a wave , by effect of the curvature , over characteristic lengths smaller than the debroglie wavelength . in a recent paper
a bound to chaos , imposed by quantum dynamics , was derived .
although the original context of the discussion was that of black holes and string theory , the result itself is a general elementary property of quantum mechanics , and the proof only relies on shrdinger s equation plus reasonable ` clustering ' properties of operators @xcite ( see also @xcite ) . in this note ,
i discuss the appearance of this result in a setting in which it is an intuitive property of a wave dynamics in a scattering medium . |
let @xmath3 be a killed markov process with law @xmath4 , taking its values in @xmath5 , where @xmath6 is a cemetery point .
we denote by @xmath7 the killing time of @xmath3 . a probability measure @xmath8 on @xmath9
is called a * quasi - stationary distribution * ( qsd ) if , for all @xmath10 , the distribution of the process @xmath11 , initially distributed with respect to @xmath8 and conditioned to be not killed before time @xmath12 , is still @xmath8 at time @xmath12 , that is @xmath13 for every @xmath14 and @xmath15 . without loss of generality , we suppose that @xmath6 is an absorbing point , so that @xmath16 .
let @xmath17 be a probability measure on @xmath9 . if it exists and provided it is a probability , the limiting conditional distribution @xmath18 is called the yaglom limit for @xmath17 , from the russian mathematician a.m. yaglom .
he showed in @xcite that the limiting conditional distribution of the number of descendants in the @xmath19 generation of a galton - watson process always exists in the subcritical case .
the existence or uniqueness of such invariant conditional distributions have been proved in a host of contexts .
when @xmath9 is finite , it is proved in @xcite that there exists a unique qsd @xmath8 and that the yaglom limit converges to @xmath8 independently of the initial distribution . in @xcite , the case of a birth and death process on @xmath20 is studied .
for this process , the set of qsds is either empty , or a singleton , or a continuum indexed by a real parameter and given by an explicit recursive formula .
this is an exception : most of the known results on qsds are related with existence or uniqueness problems . in @xcite , the existence of a quasi - stationary distribution for a continuous time markov chain on @xmath20 killed at @xmath2 is proved under conditions on moments of the killing time , using an original renewal dynamical approach . in @xcite , the case of @xmath0-dimensional diffusion on @xmath21 with @xmath22 drift and
killed at @xmath2 are studied , with the assumption that @xmath23 is a natural boundary .
the dependence between the initial measure and the yaglom limit is explored in @xcite ( for a brownian motion with constant drift killed at @xmath2 ) and @xcite ( for the orstein - uhlenbeck process killed at @xmath2 ) . in @xcite ,
the case of @xmath0-dimensional diffusions with general killing on the interior of a given interval is investigated . in @xcite ,
the authors study the existence and uniqueness of the qsd for @xmath0-dimensional diffusions killed at @xmath2 and whose drift is allowed to explode at the boundary , which is the case under study in the present paper .
see @xcite for a regularly updated extensive bibliography on qsd . in this paper
we are concerned with @xmath0-dimensional it diffusions with values in @xmath10,+\infty[\cup\{\partial\}$ ] killed at @xmath2 and defined by the stochastic differential equation @xmath24 where @xmath25 is a standard @xmath0-dimensional brownian motion and @xmath260,+\infty[)$ ] . in @xcite , the yaglom limit of this process is studied and the authors give some conditions on the drift @xmath27 , which are sufficient for the existence and the uniqueness of the qsd . in particular , they allow the drift to explode at the origin .
as explained in the paper , this diffusion is closely related with some markov mortality models .
such applications need the computation of the process qsd , but the tools used in @xcite are based on spectral theory s arguments and do nt allow us to get explicit values .
our aim is to give an easily simulable approximation s method of this qsd .
the problem of qsd s approximation has been already explored in @xcite , @xcite when @xmath9 is a bounded open set of @xmath28 and @xmath11 is a brownian motion killed at the boundary of @xmath9 .
the authors proved an approximation s method exposed in @xcite , which is based on a fleming - viot type system of interacting particles whose number is going to infinity . in @xcite , it is proved that this method works well for a continuous time markov chain in a countable state space under suitable assumption on the transition s rates ( moreover , the existence of a qsd is a consequence of the approximation s method ) .
new difficulties arise from our case with unbounded drift .
for instance , the interacting particle process introduced in @xcite is nt necessarily well defined .
to avoid this difficulty , we begin by proving that one can approximate our qsd by the qsd s of diffusions with bounded drifts .
let us denote by @xmath29 the law of a diffusion with values in @xmath1\epsilon,1/\epsilon[$ ] , defined by the stochastic differential equation @xmath30 and killed when it hits @xmath31 or @xmath32 . in @xcite
, it is proved that the yaglom limit associated with @xmath29 exists and is its unique qsd .
we will denote it by @xmath33 . in the first part of this paper ,
we give some conditions on @xmath260,+\infty[)$ ] for the family @xmath34 to be tight and to converge , when @xmath35 , to a qsd for the law @xmath36 .
we point out the fact that this result remains valid in the case of an unbounded drift diffusion with values in a bounded interval . in a second part , we prove an approximation method for each probability measure @xmath33 , based on the interacting process introduced in @xcite .
fix @xmath37 and let us describe the interacting particle process of size @xmath38 : each particle moves independently in @xmath1\epsilon,1/\epsilon[$ ] , each one with law @xmath29 until one of them hits the boundary . at this time , the killed particle jumps on the position of an other particle , chosen uniformly between the @xmath39 remaining one .
then the particles evolve independently , until one of them is killed and so on ( see figure [ fig1 ] ) .
one has to prove that the particles do nt degenerate at the boundary . in @xcite ,
the authors prove a non - degeneracy result with arguments based on a construction of the @xmath40-dimensional brownian motion due to it , where @xmath41 .
it seems that this tool ca nt be easily generalized to other diffusions . to prove such results under our settings
, we build an original coupling between the interacting particle process and an independent particle system of the same size .
this coupling is valid for all drifted brownian motions with continuous bounded drift , killed at the boundary of a bounded interval of @xmath42 .
it will be used in each step of the proof .
we conclude the paper by two numerical applications . at first
, we treat the case of the logistic feller diffusion introduced in @xcite and studied in @xcite with values in @xmath10,+\infty[$ ] , driven by the stochastic differential equation @xmath43 where @xmath25 is a @xmath0-dimensional brownian motion and @xmath44 are two positive constants . clearly , @xmath2 is an absorbing state for this diffusion . in a second time
we study in detail the case of the wright - fisher diffusion on @xmath10,1[$ ] conditioned to be killed at @xmath0 ( see @xcite ) .
this diffusion takes values in @xmath10,1[$ ] , and is defined by @xmath450,1[,\ ] ] where @xmath25 is a @xmath0-dimensional brownian motion .
this diffusion is absorbed at @xmath0 .
let @xmath46 be the law of a diffusion process taking its values in @xmath10,+\infty[\cup\{\partial\}$ ] , killed when it hits @xmath2 and defined by the stochastic differential equation ( sde ) @xmath47 where @xmath25 is a @xmath0-dimensional brownian motion .
the drift @xmath27 is taken in the set of real valued continuously differentiable functions @xmath480,+\infty[)$ ] .
we denote by @xmath49 the infinitesimal generator associated with @xmath36 .
we define , @xmath500,+\infty[$ ] , @xmath51 @xmath52 and @xmath53 for all @xmath540,1/2[$ ] , we define @xmath55 as the law of the diffusion taking its values in @xmath1\epsilon,1/\epsilon[$ ] , defined by the sde @xmath56\epsilon,1/\epsilon[\ ] ] and killed when it hits the boundary @xmath57 .
let @xmath58 be the infinitesimal generator @xmath59 with the dirichlet boundary condition on @xmath57 .
@xmath60 has a simple real eigenvalue @xmath61 ( see ( * ? ? ?
* theorem kr ) ) at the bottom of its spectrum .
the corresponding eigenfunction @xmath62 is positive and belongs to @xmath63)$ ] .
we choose it so that @xmath64 let us recall some results of @xcite : [ th9 ] the yaglom limit associated with @xmath29 exists for all initial distributions @xmath65 , @xmath66\epsilon,1/\epsilon[$ ] , and does nt depend on @xmath67 .
this limit is a qsd , which we denote by @xmath33 .
furthermore , we have @xmath68 in fact , @xmath33 is the unique qsd of the process , as proved in lemma [ le3 ] below , but we wo nt use it in this section .
the aim of this section is to study the asymptotic behaviour of @xmath69 when @xmath31 goes to @xmath2 . from now , @xmath700,+\infty[)$ ] denotes the space of probability measures on @xmath10,+\infty[$ ] equipped with the weak topology .
the following hypotheses arise naturally in the proof of theorem [ th1 ] , which is based on a compactness - uniqueness method .
@xmath71 is bounded below by @xmath72 , where @xmath73 is a positive constant .
moreover , @xmath74 when @xmath75 .
@xmath76 @xmath77 [ th1 ] assume that hypotheses ( h1 ) and ( h2 or h3 ) are satisfied
. then @xmath780,+\infty[),\ ] ] where @xmath8 is a qsd for @xmath36 , which is equal to the yaglom limit @xmath79 , @xmath500,+\infty[$ ] .
the hypotheses ( h1 ) and ( h2 or h3 ) are the assumptions that are made in @xcite to prove the existence of the yaglom limit .
if a process satisfies the hypotheses of theorem [ th1 ] , then it is killed in finite time a.s or it is never killed a.s . indeed , assume that the process can be killed in finite time with a positive probability . then @xmath80 ( see @xcite ) and @xmath81 ( as a consequence of ( h1 ) and ( h2 or h3 ) ) . but this two conditions are fulfilled if and only if the process is killed in finite time almost surely ( see ( * ? ? ?
* theorem 3.2 p.450 ) ) .
[ re1 ] the existence of a qsd for @xmath36 can be seen as a consequence of theorem [ th1 ] .
the existence of the yaglom limit is proved in ( * ? ? ?
* theorem 5.2 ) . in part
[ par3 ] , we give the counterpart of theorem [ th1 ] for diffusions with values in a bounded interval .
the end of the section is devoted to the proof of theorem [ th1 ] .
this part is devoted to the proof of the following result , [ pr1 ] assume that the hypotheses ( h1 ) and ( h2 or h3 ) are satisfied .
then the family @xmath34 is tight .
moreover , every limit point is absolutely continuous with respect to the lebesgue measure .
we know that @xmath83 , @xmath84)$ ] , and @xmath62 satisfies the differential equation @xmath85 with the boundary conditions @xmath86 define @xmath87 . by , we know that : @xmath88 we have @xmath89 with the boundary conditions @xmath90 [ le5 ] assume that the hypothesis ( h1 ) is fulfilled . then @xmath91 is uniformly bounded above and the family @xmath92 is tight .
_ proof of lemma [ le5 ] : _ from the differential equation satisfied by @xmath93 , we have @xmath94 integrating by parts and looking at the boundary conditions ( [ eqc2 ] ) , @xmath95 where @xmath93 is normalized in @xmath96 .
that implies @xmath97 the eigenvalue @xmath61 of @xmath60 is given by ( see for instance ( * ? ? ?
* , part 8) ) @xmath98\epsilon,1/\epsilon[\right)}{(l^{\epsilon}\phi,\phi)_{\mu}},\nonumber\\ & = & \inf_{\phi\in c^{\infty}_0\left(]\epsilon,1/\epsilon[\right)}{(l^0\phi,\phi)_{\mu}},\label{eq69}\end{aligned}\ ] ] where @xmath99\epsilon,1/\epsilon[\right)$ ] is the vector space of infinitely differentiable functions with compact support in @xmath1\epsilon,1/\epsilon[$ ] and
we deduce from it that @xmath61 increases with @xmath31 and is uniformly bounded above by @xmath101 .
we have then @xmath102 looking at the boundary conditions ( [ eqc2 ] ) , we obtain , for all @xmath66\epsilon,1/\epsilon[$ ] , @xmath103 then , applying the cauchy - schwarz inequality to the right term above , @xmath104 from ( [ eqc3 ] ) , the integral product is bounded by @xmath105 , thus @xmath106 , independent from @xmath31 , such that @xmath107 where @xmath108 for all @xmath660,+\infty[$ ] , thanks to hypothesis ( h1 ) .
that implies the first part of lemma [ le5 ] .
let us prove that the family @xmath109 is tight .
fix @xmath110 .
we have to find a compact subset @xmath111 in @xmath10,+\infty[$ ] such that @xmath1120,+\infty[\setminus k_{\delta}}{v_{\epsilon}^2(x)dx}\leq\delta,\ ] ] for all @xmath540,1/2[$ ] .
thanks to , we have @xmath113 , then @xmath114 from the second part of hypothesis ( h1 ) , @xmath115 such that @xmath116 for all @xmath117 .
that implies @xmath118 where the last inequality is due to .
finally , the compact set @xmath119 $ ] satisfies .
@xmath120 [ le6 ]
assume that ( h1 ) is satisfied .
then @xmath121 is uniformly bounded below by a constant @xmath122 .
_ proof of lemma [ le6 ] : _ assume that @xmath121 is nt uniformly bounded below : one can find a sub - sequence @xmath123 , where @xmath124 , which tends to @xmath2 . from lemma [ le5 ] , @xmath125 is uniformly bounded , so that @xmath126 .
the family @xmath127 being tight , one can find ( after extracting a sub - sequence ) a positive map @xmath128 such that , for all continuous and bounded @xmath129 , @xmath130 indeed , @xmath131 being uniformly bounded , all limit measure is absolutely continuous with respect to the lebesgue measure .
in particular , @xmath132 then @xmath133 but @xmath134 is continuous and positive on @xmath42 , so that @xmath128 vanishes almost every where .
finally , by the convergence property ( [ eqc5 ] ) applied to @xmath135 equal to @xmath0 almost everywhere , we have @xmath136 what is absurd .
thus , one can define @xmath137
. @xmath120 [ le7 ] assume that ( h1 ) and ( h2 ) are satisfied .
then the family @xmath138 is tight .
_ proof of lemma [ le7 ] : _ by , we have @xmath139 for all @xmath140 , using cauchy - schwarz inequality , we get on one hand @xmath141 on the other hand , @xmath142 thanks to ( h2 ) , both terms are going to @xmath2 uniformly in @xmath31 , when @xmath143 and @xmath144 tend respectively to @xmath2 and @xmath23 . as a consequence ,
the family @xmath138 is tight .
@xmath120 [ le8 ] assume that ( h1 ) and ( h3 ) hold .
then the family @xmath138 is tight .
_ proof of lemma [ le8 ] : _ the first part of the hypothesis ( h3 ) is the same as ( h2 ) s one , then @xmath145 when @xmath144 goes to infinity , uniformly in @xmath31 .
moreover , there exists a constant @xmath146 such that , for any @xmath660,1]$ ] and any @xmath540,1/2]$ ] , @xmath147 this is a consequence of ( * ? ? ?
* proposition 4.3 ) whose proof is still available under our settings .
this inequality allows us to conclude the proof of lemma [ le8 ] .
@xmath120 thanks to equality ( [ eqb1 ] ) and lemmas [ le6 ] , [ le7 ] and [ le8 ] , the first part of proposition [ pr1 ] is proved .
moreover , @xmath33 has a density with respect to the lebesgue measure which is bounded on every compact set , uniformly in @xmath37 . thus every limit point is absolutely continuous with respect to the lebesgue measure .
[ pr11 ] assume that hypotheses ( h1 ) and ( h2 or h3 ) are fulfilled and let @xmath8 be a probability measure which is the limit of a sub - sequence @xmath148 , where @xmath124 when @xmath149 .
then @xmath8 is a qsd with respect to @xmath36 .
_ proof of proposition [ pr11 ] : _ from proposition [ pr1 ] , the family @xmath82 is tight .
let @xmath8 be a limit point of the family @xmath82 .
there exists a sub - sequence @xmath150 which converges to @xmath8 , where @xmath151 is a decreasing sequence which tends to @xmath2 .
we already know that @xmath8 is absolutely continuous with respect to the lebesgue measure .
that implies that , for all open intervals @xmath152c , d[\subset\mathbb{r}_+$ ] , @xmath153 and , for all bounded maps @xmath135 continuous on @xmath42 , @xmath154 let @xmath155 ( resp .
@xmath156 ) be the distribution at time @xmath12 of a diffusion with law @xmath157 ( resp .
@xmath158 ) , conditioned to be not killed until time @xmath12 , that is @xmath159 and @xmath160 the probability measure @xmath33 being a qsd for @xmath29 , we have @xmath161 for all @xmath162 .
we want to show that @xmath163 for all @xmath162 , we have then to prove the following convergence result : @xmath164c , d[\subset\mathbb{r}_+,\ \nu_{\epsilon_k , t}(d){\smash{\mathop{\longrightarrow}\limits_{k\rightarrow+\infty}^{}}}\nu_{t}(d).\ ] ] indeed , suppose that ( [ eq68 ] ) holds , then on the one hand , @xmath165 . in the other hand @xmath166 .
we have then @xmath167c , d[)=\nu(]c , d[)$ ] , @xmath168c , d[\subset\mathbb{r}_+$ ] and this conclude the proof of proposition [ pr11 ] .
let us prove ( [ eq68 ] ) . by definition , @xmath169\epsilon_k,1/\epsilon_k[}{\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)d\nu_{\epsilon_k}(x ) } } { \int_{]\epsilon_k,1/\epsilon_k[}{\mathbb{p}^{\epsilon_k}_x(\tau_{\partial}>t)d\nu_{\epsilon_k}(x)}}.\ ] ] the numerator is equal to @xmath170\epsilon_k,1/\epsilon_k[}{\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)d\nu_{\epsilon_k}(x)}=\int_{]\epsilon_k,1/\epsilon_k[}{\mathbb{p}^0_x(\omega_t\in d)d\nu_{\epsilon_k}(x)}\\ + \int_{]\epsilon_k,1/\epsilon_k[}{\left[\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)-\mathbb{p}^0_x(\omega_t\in d)\right]d\nu_{\epsilon_k}(x)}\end{gathered}\ ] ] for all @xmath162 , the map @xmath171 is continuous and bounded , then , by the convergence property ( [ eqc11 ] ) , @xmath172\epsilon_k,1/\epsilon_k[}{\mathbb{p}^0_x(\omega_t\in d)d\nu_{\epsilon_k}(x)}\rightarrow\int_{]0,\infty[}{\mathbb{p}^0_x(\omega_t\in d)d\nu(x)},\ ] ] assume that ( h2 ) is fulfilled , then , similarly to ( [ eq71 ] ) and ( [ eq72 ] ) , we have , for all bounded continuous functions @xmath1730,+\infty[\rightarrow\mathbb{r}$ ] and all @xmath174 , @xmath175 and , for all @xmath176 , @xmath177 replacing @xmath178 by @xmath179 , which is decreasing to @xmath2 when @xmath149 , and by monotone convergence theorem , we have @xmath1720,m[}{\left[\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)-\mathbb{p}^0_x(\omega_t\in d)\right]d\nu_{\epsilon_k}(x)}{\smash{\mathop{\longrightarrow}\limits_{k\rightarrow+\infty}^{}}}0\ ] ] and @xmath172m,+\infty[}{\left[\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)-\mathbb{p}^0_x(\omega_t\in d)\right]d\nu_{\epsilon_k}(x)}{\smash{\mathop{\longrightarrow}\limits_{k\rightarrow+\infty}^ { } } } 0.\ ] ] finally , the density of @xmath180 being bounded above in every compact set @xmath181 $ ] , uniformly in @xmath182 , the same argument of monotone convergence gives us @xmath172\epsilon_k,1/\epsilon_k[}{\left[\mathbb{p}^{\epsilon_k}_x(\omega_t\in
d)-\mathbb{p}^0_x(\omega_t\in d)\right]d\nu_{\epsilon_k}(x)}{\smash{\mathop{\longrightarrow}\limits_{k\rightarrow+\infty}^ { } } } 0.\ ] ] with similar arguments , the same holds under ( h3 ) .
finally , we obtain @xmath172\epsilon_k,1/\epsilon_k[}{\mathbb{p}^{\epsilon_k}_x(\omega_t\in d)d\nu_{\epsilon_k}(x)}\rightarrow\int_{]0,+\infty[}{\mathbb{p}^0_x(\omega_t\in d)d\nu(x)}.\ ] ] thanks to ( * ? ? ?
* lemma 5.3 and theorem 2.3 ) , the map @xmath1830,+\infty[)$ ] is continuous , and @xmath184 is increasing to @xmath2 when @xmath149 .
thus the denominator can be treated in the same way .
@xmath120 we can now conclude the proof of theorem [ th1 ] : [ pr12 ] assume that ( h1 ) and ( h2 or h3 ) hold .
the limit measure @xmath8 in the statement of proposition [ pr11 ] is unique .
moreover @xmath8 is the yaglom limit associated with @xmath46 , @xmath500,+\infty[$ ] .
_ proof of proposition [ pr12 ] : _ the proof of proposition [ pr11 ] implies that @xmath185 the probability measure @xmath33 being a qsd for @xmath29 , we have @xmath186 thanks to ( [ eq69 ] ) , @xmath61 is decreasing to @xmath1870,+\infty[\right)}{(l^0\phi,\phi)_{\mu}}$ ] when @xmath31 goes to @xmath2 . as a consequence , @xmath188 in this case , the density of @xmath8 with respect to @xmath189 is an eigenfunction of @xmath49 with eigenvalue @xmath190 , where @xmath191 is the adjoint operator of @xmath49 ( this is a consequence of the spectral decomposition proved in ( * ? ? ? * theorem 3.2 ) ) .
as defined , @xmath192 is at the bottom of the spectrum of @xmath191 .
thanks to ( * ? ? ?
* theorem 3.2 ) , this eigenvalue is simple .
moreover , ( * ? ? ?
* theorem 5.2 ) states that this qsd is equal to @xmath193 , what concludes the proof .
@xmath120 theorem [ th1 ] is stated for @xmath0-dimensional diffusions with values in @xmath10,+\infty[$ ] .
however , most of the proofs can be easily adapted to diffusions with values in a bounded interval @xmath1a , b[$ ] , where @xmath194 , defined by the sde @xmath195a , b[,\ ] ] and killed when it hits @xmath196 or @xmath197 . here
@xmath25 is a standard brownian motion and @xmath26a , b[)$ ] .
more precisely , let us denote by @xmath36 the law of such a diffusion .
for each @xmath37 , define @xmath29 as the law of a diffusion with values in @xmath1a+\epsilon , b-\epsilon[$ ] , driven by the sde @xmath56a+\epsilon , b-\epsilon[,\ ] ] and killed when it hits @xmath198 or @xmath199 . as proved in @xcite
, there exists a unique qsd @xmath33 associated with @xmath29 . we define @xmath200 and @xmath201 .
the counterpart of theorem [ th1 ] under these settings is [ th10 ] assume that the following hypotheses are fulfilled : @xmath71 is uniformly bounded below by @xmath72 , where @xmath73 is a positive constant .
@xmath202 or @xmath203 is integrable on a neighbourhood of @xmath196 .
@xmath202 or @xmath204 is integrable on a neighbourhood of @xmath197 .
then the family of qsd @xmath69 is tight as family of measures on @xmath1a , b[$ ] .
moreover , every limit point of the family @xmath82 is a qsd for @xmath36 .
our aim is nt to develop this part , but we point out that to show that proposition [ pr12 ] remains valid , most of the arguments used to prove the key results ( * ? ? ?
* theorem 3.2 ) and ( * ? ? ? * theorem 5.2 ) can be adapted to these settings .
we are interested in proving an approximation method for the qsd associated with @xmath29
. it will be sufficient to prove it for any diffusion @xmath3 taking its values in @xmath10,1[$ ] in place of @xmath1\epsilon,1/\epsilon[$ ] , defined by the stochastic differential equation ( sde ) @xmath205 and killed when @xmath206 hits the boundary @xmath207 . here
@xmath25 is a real brownian motion and @xmath208)$ ] .
the law of @xmath11 will be denoted by @xmath4 . from @xcite ,
the qsd of @xmath11 is unique and equals the yaglom limit .
it will be denoted by @xmath8 . for notational convenience ,
new notations have been defined for this section , which is totally independent of the previous one .
fix @xmath38 and let us define formally the interacting particle process with @xmath209 particles described in the introduction .
let @xmath210 be @xmath209 independent brownian motions and @xmath2110,1[^n$ ] be the starting point of the process .
* for each @xmath212 , the particle @xmath213 evolves in @xmath10,1[$ ] and satisfies the sde @xmath214 ( and then it is independent of the others ) until @xmath215 . * at time @xmath216 , the path of a particle , denoted by @xmath217 ( it is unique ) , has a left limit equal to @xmath2 or @xmath0 . *
a particle @xmath218 is chosen in @xmath219 .
the particle @xmath217 jumps on the position of the particle @xmath218 : we set @xmath220 . * after time @xmath221 , each particle @xmath213 evolves in @xmath10,1[$ ] with respect to the sde @xmath214 until @xmath222 . at time
@xmath221 , all the particles are in @xmath10,1[$ ] , so that we have @xmath223 for all @xmath212 almost surely . * at time @xmath224 ( which is then strictly bigger than @xmath221 ) , a unique particle @xmath225 has a path whose left limit is equal to @xmath2 or @xmath0 . * a particle @xmath226 is chosen in @xmath227 .
the particle @xmath225 jumps on the position of the particle @xmath226 : we set @xmath228 . * after time @xmath229 , the particles evolve independently from each other and so on . following this way , we define the strictly increasing sequence of stopping times @xmath230 , the time @xmath231 and the interacting particle system @xmath232 for all @xmath233 .
the law of @xmath234 will be denoted by @xmath235 .
[ [ section ] ] we can now state the main result of this section : [ th2 ] @xmath234 is well defined , that means @xmath236 almost surely .
it is geometrically ergodic , with unique stationary distribution @xmath237 .
let @xmath238 be the empirical stationary measure of the interacting particle process with @xmath209 particles , that is the empirical measure of a random vector @xmath2390,1[^n$ ] distributed with respect to the stationary measure @xmath237 of the process @xmath234 .
the sequence of random measures @xmath240 converges in law to the deterministic measure @xmath8 , qsd of the process @xmath11 .
subsection [ parts21 ] is devoted to prove a coupling which ensures the non - degeneracy of the particles at the boundary .
a consequence will be that @xmath236 almost surely . in subsection [ parts22 ] ,
the process is studied in finite time .
we prove that the empirical measure of the process @xmath234 at time @xmath12 converges , when @xmath209 goes to infinity , to the distribution of @xmath206 conditioned to not be killed until time @xmath12 , which is @xmath241 ( @xmath242 denotes the limit of the empirical measure of the process at time @xmath2 ) .
we prove in subsection [ parts23 ] that the interacting particle system @xmath234 is geometrically ergodic .
we conclude by showing the convergence of the empirical stationary measure to the qsd @xmath8 .
one of the most important fact when studying the interacting particle system is that the particles do nt degenerate at the boundary .
this is an evidence if the particles are independent . in our case
, we will prove a coupling between @xmath234 and an other process @xmath243 whose components are independent identically distributed and do nt degenerate at the boundary and such that , for all @xmath212 , @xmath244 with this construction , the process @xmath234 does nt degenerate at the boundary , because each of its particles @xmath213 is contained in @xmath245 $ ] .
this coupling will be useful in each step of the proof .
define @xmath2460,1[}{|q(x)|}$ ] ( we have @xmath247 by hypothesis ) and fix @xmath212 .
the process @xmath248 is defined with values in @xmath249 $ ] by the sde @xmath250 with @xmath2 and @xmath251 as reflecting boundary ( see @xcite for the definition of a reflected diffusion ) .
the coupling inequality is fulfilled at time @xmath252 .
the brownian motion @xmath253 will depend on @xmath254 and on the position of @xmath213 . if @xmath255 belongs to @xmath249 $
] , then @xmath256 and the second part of the coupling inequality is satisfied , independently of the choice of @xmath253 .
we only need to ensure that @xmath255 stays bigger than @xmath257 .
if @xmath255 belongs to @xmath258 , we only need to ensure that it is smaller than @xmath259 .
if it belongs to @xmath11/3,2/3[$ ] , the coupling inequality is obviously fulfilled , thanks to the reflection of @xmath248 on @xmath251 .
assume that @xmath255 is in @xmath10,1/3]$ ] and that the coupling inequality is fulfilled at time @xmath12 .
we have @xmath260 , with @xmath261 . if we choose @xmath253 so that @xmath262 , then @xmath263 is increasing with time almost surely and the coupling inequality remains fulfilled . in a similar way ,
if @xmath255 is in @xmath258 , we have to choose @xmath253 so that @xmath264 .
we will see in the proof of the coupling inequality that the jumps of @xmath213 and the reflexion of @xmath248 on @xmath2 do not play any role in the coupling inequality ( see figure [ fig2 ] ) .
let us define the brownian motion @xmath253 in an explicit form .
first , we build a sequence of strictly increasing times @xmath265 such that , for each @xmath266 , @xmath2670,2/3[$ ] for all @xmath268 and @xmath2671/3,1[$ ] for all @xmath269 .
define @xmath270,\ x^i_t\in]0,1/3]\}},\\ t^i_1=\inf{\{t\in[t_0,+\infty],\ x^i_t\in[2/3,1[\}},\end{aligned}\ ] ] and , for @xmath271 , @xmath272,\ x^i_t\in]0,1/3]\}},\\ t^i_{2n+1}=\inf{\{t\in[t^i_{2n},+\infty],\ x^i_t\in]2/3,1]\}}.\end{aligned}\ ] ] depending on the position of @xmath213 , which is stated by the sequence @xmath273 , we define @xmath253 by @xmath274,\end{aligned}\ ] ] and , for all @xmath266 , @xmath275\\ \label{eq10 } w^i_t = w^i_{t_{2n+1}}-(b^i_t - b^i_{t_{2n+1}})\ \mbox{for}\ t\in[t_{2n+1},t_{2n+2}].\end{aligned}\ ] ] if @xmath276 , then @xmath253 can be extended by continuity on @xmath21 , so that @xmath257 is well defined for all @xmath277 .
the sequence @xmath278 is a sequence of stopping times for the natural filtration of the process @xmath234 .
conditionally to the sequence @xmath279 ( the particles which are chosen at each successive jump ) , @xmath234 only depends on @xmath280 .
one can then apply the strong markov property to @xmath280 at time @xmath281 , @xmath282 and @xmath283 . as a direct consequence of the symmetry of a brownian motion s law and of the strong markov property applied to the @xmath209-dimensional brownian motion @xmath284 ,
@xmath253 is a brownian motion .
note that the sequence of stopping times depends clearly on the position of other particles and then on the @xmath285 s , for @xmath286 .
the brownian motions @xmath287 are independent processes . indeed
, the sequence @xmath288 does nt play any role in the law of @xmath253 and , to be convinced of that , one can compute the covariance matrix of the gaussian variables @xmath289 , for any @xmath277 , which is clearly a diagonal one .
[ pr2 ] the coupling inequality is fulfilled for all @xmath233 .
_ proof of proposition [ pr2 ] : _ define the time @xmath290 and let us work conditionally to @xmath291 , then we have , by right continuity of the two processes , @xmath292 a.s .
we first show that @xmath293 is a jump time for the particle @xmath294 .
assume the converse .
if @xmath295 , then @xmath296 and , if @xmath297 , then @xmath213 and @xmath248 are continuous in a neighbourhood @xmath298 $ ] of @xmath293 , where @xmath299 , and we have @xmath300 , for all @xmath301 , so that , by left continuity @xmath302 , and then @xmath303 . therefore , we have @xmath3040,1/3]$ ] and @xmath305 such that @xmath306 , and , for @xmath307 small enough , one can assume that @xmath308 . then , for all @xmath309 $ ] , @xmath310 , where @xmath311 and @xmath312 is an increasing process due to the reflecting property of the boundary @xmath251 for @xmath248 ( note that @xmath313 , so that , for @xmath299 small enough , @xmath314 for all @xmath315 ) . then @xmath316 stays non - negative between times @xmath293 and @xmath317 , what contradicts the definition @xmath293
. the time @xmath293 is then a jump time for the particle @xmath294 . if @xmath318 , then , by definition of @xmath293 , @xmath319 and , by left continuity of the process @xmath248 , @xmath320 , so that @xmath321 , what is impossible .
therefore @xmath322 , and , by existence of left limits for the two processes , @xmath323 such that @xmath324 .
define @xmath325 .
we have then , conditionally to @xmath291 , @xmath326 . by symmetry , conditionally to the event @xmath327 , we have @xmath328 , then @xmath291 and therefore @xmath328 .
finally , @xmath329 almost surely .
@xmath120 [ pr3 ] for all @xmath38 , the interacting particle system @xmath234 is well defined , that is @xmath236 almost surely .
_ proof of proposition [ pr3 ] : _ let @xmath38 be the size of the interacting particle system and fix arbitrarily its starting point @xmath660,1[^n$ ] .
we define the event @xmath330 .
conditionally to @xmath331 , the total number of jumps is equal to @xmath23 .
there is a finite number of particles , then at least one particle , say @xmath332 , makes an infinite number of jump before @xmath333 . at each jump of @xmath332 ,
a particle is uniformly chosen in @xmath334 . by the law of large numbers ,
each particle is chosen infinitely often before @xmath333 .
assume that a particle , say @xmath335 , remains all the time in @xmath1\epsilon,1-\epsilon[$ ] , with @xmath37 .
@xmath332 will jump on the position of @xmath335 infinitely often .
then it will come back from @xmath1\epsilon,1-\epsilon[$ ] to the boundary infinitely often in finite time , what is impossible .
we deduce that , conditionally to @xmath331 , all particles of the interacting particle system are going near to the boundary , that is @xmath336 for each particle @xmath337 . using the coupling inequality of proposition [ pr2 ]
, we deduce from ( [ eq59 ] ) that @xmath338 then , conditionally to @xmath331 , @xmath339 and @xmath340 are independent reflected diffusions with bounded drifts , which hit @xmath2 at the same time .
this occurs for two independent reflected brownian motions with probability @xmath2 , and then for @xmath339 and @xmath340 too , by the girsanov s theorem .
that implies @xmath341 .
finally , we have @xmath236 almost surely . one could hope to apply this method directly to the process with law @xmath36 studied in the first part of this paper .
unfortunately , it can be very difficult to show the existence of the process or the non - degeneracy at the boundary : the drift being unbounded , the law of the reflected diffusion used in this proof is nt absolutely continuous with respect to the law of the reflected brownian motion and stay at @xmath2 all the time after hitting it . for all @xmath342 ,
we define the open set @xmath343r,1-r[$ ] .
let @xmath344 ( resp .
@xmath345 ) be the empirical measure of the system of particles @xmath346 ( resp .
@xmath347 ) , that is @xmath348 we will suppose that , at time @xmath2 , the sequence of empirical measures @xmath349 satisfies the following non - degeneracy property , which ensures that the mass of @xmath350 does nt degenerate at the boundary , uniformly in @xmath209 : the family of random probabilities @xmath351 is said to verify the non - degeneracy property if , for any @xmath37 , @xmath352 where @xmath3530,r]\cup[1-r,1[$ ] . from definition of @xmath354 ,
the non - degeneracy of @xmath355 is the consequence of the non - degeneracy of @xmath350 .
the end of the section is devoted to prove the following proposition , which states that the non - degeneracy property is maintained over time : [ pr4 ] assume that @xmath356 satisfies the non - degeneracy property , then , for all @xmath357 and all @xmath37 , @xmath358}{\mu^n(t , d_r^c)}>\epsilon)}}=0.\end{aligned}\ ] ] _ proof of proposition [ pr4 ] : _ fix @xmath357 and @xmath37 .
because of the non - degeneracy of @xmath355 , one can find @xmath359 such that @xmath360 when @xmath209 goes to @xmath361 .
we want to apply a law of large numbers , but the @xmath362 are nt independent and depend on the number of particles @xmath209 .
let us define the diffusion @xmath363 with values in @xmath249 $ ] , defined by the sde @xmath364 with @xmath2 and @xmath251 as reflecting boundary . here
the @xmath365 are independent brownian motions .
the random processes @xmath366 are independent , identically distributed with values in @xmath367,\mathbb{r})$ ] .
one can apply to them the following law of large numbers , proved in @xcite : @xmath368}\left|\frac{1}{n}\sum_{i=1}^n{\mathbf{1}_{z^i_t\in d_r^c}}-e\left(\mathbf{1}_{z^i_t\in d_r^c}\right)\right| { \smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{prob } } } 0.\ ] ] we have @xmath369 , which tends uniformly to @xmath2 when @xmath370 .
we deduce from it that @xmath371}{\frac{1}{n}\sum_{i=1}^n{\mathbf{1}_{z^i_t\in d_r^c}}}\geq \epsilon/2\right)=0.\ ] ] for each number of particles @xmath38 , one can easily find a coupling between @xmath363 and @xmath248 , where @xmath372 implies @xmath373 for all @xmath374 $ ] . with such a coupling
, we have @xmath375 by adding the contribution of the @xmath248 which starts in @xmath10,a[$ ] , we get @xmath376 the limits and allow us to conclude the proof .
fix @xmath357 .
this section is devoted to the proof of the following proposition , which states that the empirical measure process converges to the distribution of the process @xmath11 conditioned to not be killed . [ pr5 ]
assume that @xmath377 converges in law to the random probability measure @xmath378 with respect to the weak topology and satisfies the non - degeneracy property .
then , @xmath379 , the measure processes @xmath380}$ ] converge in law to @xmath381}$ ] in the skorokhod space @xmath367,{\cal m}_1(]0,1[))$ ] when @xmath382 . here
@xmath700,1[)$ ] denotes the space of probability measures on @xmath10,1[$ ] equipped with the weak topology .
_ proof of proposition [ pr5 ] : _ for all maps @xmath383)$ ] vanishing on @xmath207 , one can apply the it s formula to the semimartingale @xmath384 ( see ( * ? ? ?
* theorem 27.1 ) ) , whose number of jumps in @xmath385 $ ] is finite almost surely : @xmath386 let us denote by @xmath387 the increasing sequence of jump times of the particle @xmath294 .
we have @xmath388 because @xmath389 and @xmath390 .
that implies @xmath391 by summing over @xmath212 , we obtain @xmath392 where @xmath393 is the continuous martingale @xmath394 and @xmath395 is the pure jump martingale @xmath396 now , we interpret each jump as a killing .
then we introduce a loss of @xmath397 of the total mass at each jump : we look at the measure process @xmath398 decreased by a factor @xmath399 at each jump .
more precisely , we set @xmath400 where @xmath401 denotes the total number of jumps before time @xmath12 .
[ led1 ] the sequence of measure processes @xmath402 converges in law to @xmath403 in the skorokhod topology @xmath367,{\cal m}(]0,1[))$ ] .
_ proof of lemma [ led1 ] : _ applying the it s formula to the semimartingale @xmath404 , we deduce from that @xmath405 where we have @xmath406 with @xmath407 and @xmath408 then @xmath409 that implies @xmath410 we deduce that for all smooth functions @xmath411 vanishing at the boundary @xmath412 because @xmath413 a.s . and by the doob s inequality , we have @xmath414 } \left| \int_0^t{\left(\frac{n-1}{n}\right)^{a^n_{s\minus}}d{\cal m}^c(\psi(t,.),s ) } \right|^2\right ) \leq \frac{1}{n } t \|\frac{\partial \psi}{\partial x}\|_{\infty}^2.\end{aligned}\ ] ] note that the jumps of the martingale @xmath415 are smaller than @xmath416 , then @xmath417\\ & \leq \frac{4}{n^2}\|\psi\|_{\infty}^2 e\left[\sum_{0\leq \tau_n \leq t}{\left(\frac{n-1}{n}\right)^{2 a_{\tau_n\minus}}}\right]\\ & \leq \frac{4}{n}\|\psi\|_{\infty}^2\end{aligned}\ ] ] by the doob s inequality , we have then @xmath418 } \left|\sum_{0\leq\tau_n\leq t}{\left(\frac{n-1}{n}\right)^{a^n_{\tau_n\minus}}\left({\cal m}^j(\psi(\tau_n,.),\tau_n)-{\cal m}^j(\psi(\tau_n\minus,.),\tau_n\minus)\right ) } \right|^2\right ) \leq \frac{4}{n}\|\psi\|_{\infty}^2,\ ] ] define @xmath419 , where @xmath420)$ ] vanishes on @xmath207 , and @xmath421 is the semigroup associated with the diffusion @xmath11 defined by . from kolmogorov s equation ( see ( * ? ? ?
* proposition 1.5 p.9 ) ) , @xmath422 we deduce from , and , that @xmath423 } \left| \nu^n(t , f)-\int_0 ^ 1{p_{t}f(x)}d\mu^n(0,x ) \right|^2\right ) \leq \frac{1}{n } c(f),\end{aligned}\ ] ] where @xmath424 is a positive constant , which only depends on @xmath425 . for each map @xmath426)$
] , one can set @xmath427 , with @xmath342 , where @xmath428)$ ] is equal to @xmath0 on @xmath12r , 1 - 2r[$ ] and vanishes on @xmath10,r[\cup]1-r,1[$ ]
. then @xmath429 where ( see proposition [ pr4 ] ) and are going to @xmath2 when @xmath430 tends to @xmath2 , uniformly in @xmath209 , and @xmath431)$ ] vanishes on @xmath207 .
then @xmath432 } \left| \nu^n(t , g)-\int_0 ^ 1{p_{t}g(x)}d\mu^n(0,x ) \right| { \smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{l^2 } } } 0.\end{aligned}\ ] ] in particular , @xmath433 converges in law to @xmath403 .
@xmath120 [ [ section-1 ] ] let us conclude the proof of proposition [ pr5 ] . from lemma [ led1 ] , @xmath4340,1[),\nu^n(t , dx))_{t\in[0,t ] } { \smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{law}}}(\mathbb{p}_{\mu(0,dx)}(x_t\in]0,1[),\mathbb{p}_{\mu(0,dx)}(x_t\in dx))_{t\in[0,t]}\ ] ] in the skorokhod topology @xmath367,\mathbb{r}\times{\cal m}_1(]0,1[))$ ] .
that means @xmath435}{\smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{law}}}(\mathbb{p}_{\mu(0,dx)}(x_t\in]0,1[),\mathbb{p}_{\mu(0,dx)}(x_t\in dx))_{t\in[0,t]}.\ ] ] the process @xmath4360,1[)$ ] never vanishing almost surely and the limit process being continuous almost surely , we have @xmath437}{\smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{law}}}\left(\mathbb{p}_{\mu(0,dx)}(x_t\in dx)/\mathbb{p}_{\mu(0,dx)}(x_t\in]0,1[)\right)_{t\in[0,t]}\ ] ] in the skorokhod topology @xmath367,{\cal m}_1(]0,1[))$ ] .
that means @xmath438 } { \smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{law } } } ( \mathbb{p}_{\mu(0,dx)}(x_t\in dx|x_t\neq\partial))_{t\in[0,t]}\ ] ] in the skorokhod topology @xmath367,{\cal m}_1(]0,1[))$ ] .
the proof of proposition [ pr5 ] is then complete . for each @xmath38
, we say that the interacting particle process @xmath234 is exponentially ergodic , if there exists a probability measure @xmath237 on @xmath10,1[^n$ ] such that , @xmath4390,1[^n,\ \forall t\in\mathbb{r}_+,\ ] ] where @xmath440 is finite , @xmath441 and @xmath442 is the total variation norm . in particular , @xmath237 is a stationary measure for the process @xmath234 .
when @xmath237 exists , we denote by @xmath238 the empirical stationary measure associated with @xmath237 , that is a random probability which is distributed as @xmath443 , where @xmath444 is a random vector in @xmath10,1[^n$ ] distributed with respect to @xmath237 . in a first time
, we prove that for all @xmath445 , the interacting particle process with @xmath209 particles @xmath234 associated with @xmath11 is exponentially ergodic .
we conclude by proving that @xmath446 converges in law to the unique qsd of @xmath11 . here
@xmath445 is fixed .
we are interested in proving the following result , which is the first part of theorem [ th2 ] [ pr9 ] the interacting particle process @xmath234 with law @xmath235 is exponentially ergodic .
_ proof of proposition [ pr9 ] : _ we focus on the @xmath0-skeleton of the interacting particle process with @xmath209 particles , which is the markov chain @xmath447 .
thanks to ( * ? ? ?
* theorem 5.3 p.1681 ) , exponential ergodicity of @xmath234 will be obtained as soon as the associated @xmath0-skeleton is geometrically ergodic , which means that it exists a probability measure @xmath448 on @xmath10,1[^n$ ] such that @xmath4490,1[^n,\ \forall n\in\mathbb{n},\end{aligned}\ ] ] where @xmath450 is finite and @xmath451 . to prove the geometrical ergodicity of the 1-skeleton , let us introduce the following definition : @xmath4520,1[^n$ ]
is said to be a * small set * for the markov chain @xmath447 if , for some nontrivial probability measure @xmath453 and some @xmath271 , @xmath37 , the @xmath454-step transition probability kernel @xmath455 satisfies , for all @xmath456 , @xmath4570,1[^n).\ ] ] [ le4 ] all compact set @xmath458^n$ ] ( with @xmath342 ) is a small set for the @xmath0-skeleton .
moreover , @xmath459 so that @xmath460 where @xmath461 is the return time to @xmath73 .
_ proof of lemma [ le4 ] : _ fix @xmath342 and let @xmath462 be the event `` the process @xmath234 has no jumps between times @xmath2 and @xmath0 '' . define @xmath463^n}\mathbb{p}^{ipp}_x({\cal f})$ ] .
thanks to the coupling with @xmath243 , we have @xmath464 . conditionally to the event @xmath465 , the particles of the interacting particle process are independent from each other .
let us study @xmath466 .
the law of @xmath467 conditionally to @xmath465 is the same as the law of @xmath11 conditioned to not jump , because , given this last event , the process @xmath467 does nt depend on the other particles .
thus the probability of `` @xmath468 '' is @xmath469 .
the law of @xmath470 has a density @xmath471 with respect to the lebesgue s measure and @xmath471 depends continuously on @xmath472 and @xmath473 .
it only vanishes when @xmath474 or @xmath0
. then @xmath475\times[r,1-r]}{p_1(x_1,y)}>0\ ] ] denoting this minimum by @xmath476 , we have , for all @xmath477 $ ] , @xmath478}(x ) dx$ ] . conditionally to @xmath465 , the particles are independent from each other , so that @xmath479 where @xmath480 is greater than @xmath481 and then greater than @xmath482
. finally , we have @xmath483^n}(y_1, ... ,y_n ) dy_1 ... dy_n,\end{aligned}\ ] ] so that @xmath484^n$ ] is a small set . for all @xmath660,1[^n$ ] , and all @xmath485 , the probability of being in @xmath73 at time @xmath486 starting from @xmath67 at time @xmath454
is bounded below by the probability @xmath487 for @xmath243 to enter @xmath73 at time @xmath486 , starting from @xmath2 at time @xmath454 .
hence , at each time @xmath485 , @xmath234 returns to @xmath73 at time @xmath486 with a probability greater than @xmath487 .
that implies that the return time to @xmath73 for the 1-skeleton of the interacting particle process with @xmath209 particles is bounded above by a time of geometrical law , independent of the starting point @xmath660,1[^n$ ] , and then satisfies condition .
@xmath120 [ [ section-2 ] ]
the chain @xmath447 is aperiodic .
moreover , if the lebesgue measure of a subset @xmath4880,1[^n$ ] is strictly positive , then @xmath489 for all @xmath660,1[^n$ ] , where @xmath490 is the first hitting time on @xmath491 for the chain @xmath447 .
thanks to ( * ? ? ?
* theorem 2.1 p.1673 ) , if such a markov chain has a small set which satisfies , then it is geometrically ergodic . as a consequence , lemma [ le4 ] allows us to conclude the proof of proposition [ pr9 ] .
we are interested in proving the following result , which is the second part of theorem [ th2 ] [ pr10 ] the sequence of random measures @xmath492 converges in law to the deterministic measure @xmath8 , qsd of the process @xmath11 .
_ proof of proposition [ pr10 ] : _ for each @xmath4930,1/4[$ ] , we define @xmath494 as a non - negative continuous bounded function from @xmath10,1[$ ] to @xmath495 , equal to @xmath0 on @xmath496 and equal to @xmath2 on @xmath497 .
we have @xmath4980,1/4[.\ ] ] thanks to proposition [ pr9 ] , the sequence of random measures @xmath499 converges in law to @xmath238 when @xmath12 tends to @xmath23 .
that implies @xmath5000,1/4[.\ ] ] we denote by @xmath345 the empirical measure of @xmath501 .
let us choose @xmath494 monotone on @xmath1r,2r[$ ] and @xmath11 - 2r,1-r[$ ] . from the coupling inequality , @xmath502 for all @xmath503 and @xmath342 .
then @xmath504 which tends to @xmath2 when @xmath12 tends to @xmath23 and @xmath430 to @xmath2 , uniformly in @xmath209 . as a consequence , @xmath505 where @xmath506 is the deterministic measure defined by @xmath507 , for all measurable set @xmath491
. that yields @xmath508 uniformly in @xmath209 .
the family of intensity measures @xmath509 is then tight .
this is a sufficient condition for the family of random variables @xmath510 to be tight , as shown in ( * ? ? ?
* corollary 2.2 ) .
we conclude that it exists a sub - sequence @xmath511 which converges in law to a random probability measure @xmath512 .
choose @xmath513 .
the non - degeneracy property is fulfilled .
thanks to proposition [ pr5 ] , @xmath514 } { \smash{\mathop{\longrightarrow}\limits_{n\rightarrow\infty}^{law } } } ( \mathbb{p}_{\cal x}(x_t\in dx|x_t\neq \partial))_{t\in[0,t]},\ \forall t>0\ ] ] in the skorokhod topology @xmath367,{\cal m}_1(]0,1[))$ ] .
the limiting process @xmath515}$ ] being almost surely continuous , @xmath516 with respect to the weak topology of @xmath700,1[)$ ] . by stationarity ,
the random probability measures @xmath517 and @xmath518 have the same law .
making @xmath209 tend to @xmath361 , we deduce that @xmath519 and @xmath520 have the same law too .
this looks like a qsd , but @xmath520 is _ a priori _ a random measure and we need the following result to conclude .
[ le3 ] for all @xmath5210,1[)$ ] , @xmath522 where @xmath523 is the killing time of the process @xmath11 and @xmath8 its unique qsd .
_ proof of lemma [ le3 ] : _ let @xmath128 be a probability measure on @xmath10,1[$ ] .
@xmath524 , @xmath525 and @xmath526 two continuous maps vanishing on @xmath2 and @xmath0 such that , for all @xmath5270,1[$ ] ( see r.g .
pinsky s explanations ( * ? ? ?
* hypotheses 2 and 3 ) ) : @xmath528 here @xmath529 is uniformly bounded above in the variables @xmath12 and @xmath67 ( see ( * ? ? ?
* proof of the equality 7.2 , p27 ) ) , then , by dominated convergence , one can integrate with respect to @xmath128 under the limit in , @xmath530 the same holds for : @xmath531 then , by fubini s theorem , @xmath532 that is , from ( [ eq40 ] ) and ( [ eq41 ] ) with @xmath5330,1[$ ] , @xmath534 which is nothing else but @xmath535 ( see ( * ? ? ?
* proposition 1.10 ) ) .
@xmath120 [ [ section-3 ] ] thanks to lemma [ le3 ] , @xmath519 converges almost surely to the yaglom limit when @xmath536 , and so do @xmath512 .
finally , @xmath512 is the unique qsd of the process and the proof is complete .
we apply our result to the logistic feller diffusion with values in @xmath10,+\infty[$ ] , driven by the stochastic differential equation @xmath537 and killed when it hits @xmath2 . here
@xmath25 is a @xmath0-dimensional brownian motion and @xmath538 are two positive constants . as proved in @xcite , ( h1 ) and ( h3 )
are fulfilled in this case .
thanks to theorem [ th1 ] and denoting by @xmath8 the yaglom limit associated with @xmath36 , we have @xmath541 in the numerical simulations below , we set @xmath31 equal to @xmath542 . by theorem [ th2 ] , we have @xmath543 where @xmath238 is the empirical measure of the system studied in section [ par2 ] . in the numerical simulations , we set @xmath544 and , because of the randomness of @xmath238 , we approximate @xmath506 using the ergodic theorem : we compute @xmath545 .
the graphic below ( see figure [ fig10 ] ) shows this approximation for different values of @xmath430 and @xmath546 .
we illustrate the result of section [ par3 ] by an application to the wright - fisher diffusion with values in @xmath10,1[$ ] conditioned to be killed at @xmath2 .
this diffusion is driven by the sde @xmath5470,1[,\ ] ] and killed when it hits @xmath2 ( @xmath0 is never reached ) . in @xcite , the author proves that the qsd of this process exists and has the density @xmath548 with respect to the lebesgue measure .
define @xmath36 as the law of @xmath549 , where @xmath550 is defined as above .
@xmath36 is the law of the diffusion with values in @xmath10,\pi[$ ] , driven by the sde @xmath5510,\pi[,\ ] ] killed when it hits @xmath2 ( @xmath552 is never reached ) . for all @xmath540,\pi/2[$ ] , define @xmath29 and @xmath33 as in section [ par3 ] . in the following numerical simulation
( see figure [ fig4 ] ) , we set @xmath561 and @xmath544 .
we compute @xmath506 , which is an approximation of @xmath33 , and then of @xmath8 , with the method used in the logistic case ( see part [ par5 ] ) .
i am extremely grateful to my ph.d .
supervisor sylvie mlard for his carefully and essential help on the form and the content of my first paper .
i would like to thank pierre collet for his explanations on some spectral theory tools and krzysztof burdzy for his indication on the proof of the interacting particle process existence .
99 k. burdzy , r. holyst , d. ingerman , p. march ( 1996 ) configurational transition in a fleming - viot - type model and probabilistic interpretation of laplacian eigenfunctions , _ j. phys . a _ * 29
* , 2633 - 2642 .
k. burdzy , r. holyst , p. march ( 2000 ) a fleming - viot representation of the dirichlet laplacian , _ comm .
phys . _ * 214 * , 679 - 703 .
p. cattiaux , p. collet , a. lambert , s. martinez , s. mlard , j. san martin ( 2007 ) quasi - stationarity distributions and diffusion models in population dynamics , _ arxiv : math/0703781v2_. j.a .
cavender ( 1978 ) quasi - stationary distributions of birth and death processes , _ adv .
* 10 * , 570 - 586 .
m. chaleyat - maurel , n. el karoui ( 1978 ) un problme de rflexion et ses applications au temps local et aux quations diffrentielles stochastiques sur @xmath495 .
cas continu , _ astrisque _ * 52 - 53 * , 117 - 144 p. collet , s. martinez , j. san martin ( 1995 ) asymptotic laws for one - dimensional diffusions conditioned to nonabsorption , _ ann .
_ , * 23 - 3 * , 1300 - 1314
darroch and e. seneta ( 1965 ) on quasi - stationary distributions in absorbing discrete - time markov chains , _ j. appl .
* 2 * , 88 - 100 d. down , s.p .
meyn , r.l .
tweedie ( 1995 ) exponential and uniform ergodicity of markov processes , _ ann .
probab . _ * 23 * , 1671 - 1691 .
ethier , t.g .
kurtz ( 1986 ) _ markov processes , characterization and convergence _ , wiley series in probability and mathematical statistics . p.a .
ferrari , nevena maric ( 2006 ) quasi - stationary distributions and fleming - viot processes in countable spaces , _ arxiv : math/0605665 _ , eprint .
ferrari , h. kesten , s. martinez , p. picco , ( 1995 ) existence of quasi - stationary distributions . a renewal dynamical approach , _ ann
. probab . _
* 23 * , 2:511 - 521
. i. grigorescu , m. kang ( 2004 ) hydrodynamic limit for a fleming - viot type system , _ stochastic process .
appl . _ * 110 * , 111 - 143 .
t. huillet ( 2007 ) on wright - fisher diffusion and its relatives , _
theory exp .
_ * p11006*. n. ikeda , s. watanabe ( 1988 ) _ stochastic differential equations and diffusion processes _ , north - holland , amsterdam , 2nd edition .
j. jacod , a. shiryaev ( 1987 ) _ limit theorems for stochastic processes _ , springer , berlin .
a. jakubowski ( 1988 ) tightness criteria for random measures with applications to the principle of conditioning in hilbert spaces , _ probab . math .
_ * 9.1 * , 95 - 114 .
a. lambert ( 2005 ) the branching process with logistic growth , _ ann .
probab . _ * 15 * , 1506 - 1535
. m. lladser , j. san martin ( 2000 ) domain of attraction of the quasi - stationary distributions for the orstein - uhlenbeck process ,
_ j. appl .
prob . _ * 37 * , 511 - 520 .
s. martinez , p. picco , j. san martin ( 1998 ) domain of attraction of quasi - stationary distributions for the brownian motion with drift , _ adv .
prob . _ * 30 * , 385 - 408 .
meyn , r.l .
tweedie ( 1993 ) _ markov chains and stochastic stability _ , springer , berlin .
m. mtivier ( 1982 ) _
semimartingales , a course on stochastic processes _ , gruyter , berlin .
pinsky ( 1985 ) on the convergence of diffusion processes conditioned to remain in a bounded region for large time to limiting positive recurrent diffusion processes , _ ann .
_ * 13 - 2 * , 363 - 378 .
polett ( regularly updated ) quasi - stationary distributions : a bibliography , _ http://www.maths.uq
.edu.au/ pkp / papers / qsds / qsds.pdf_. r. ranga rao ( 1963 ) the law of large numbers for @xmath563$]-valued random variables , _ theor .
_ * 8 * , 70 - 74 .
d. steinsaltz and s.n . evans ( 2004 )
markov mortality models : implications of quasi - stationarity and varying initial distributions , _ theo .
_ * 65 * , 319 - 337 .
d. steinsaltz and s.n .
evans ( 2007 ) quasistationary distributions for one - dimensional diffusions with killing , _ trans .
soc . _ * 359 - 3 * , 1285 - 1324 .
a.m. yaglom ( 1947 ) certain limit theorems of the theory of branching processes ( in russian ) _ dokl .
nauk _ * 56 * , 795 - 798 .
k. yosida ( 1966 ) _ functional analysis , 2nd edition _ springer - verlag . | the long time behavior of an absorbed markov process is well described by the limiting distribution of the process conditioned to not be killed when it is observed .
our aim is to give an approximation s method of this limit , when the process is a @xmath0-dimensional it diffusion whose drift is allowed to explode at the boundary . in a first step ,
we show how to restrict the study to the case of a diffusion with values in a bounded interval and whose drift is bounded . in a second step ,
we show an approximation method of the limiting conditional distribution of such diffusions , based on a fleming - viot type interacting particle system .
we end the paper with two numerical applications : to the logistic feller diffusion and to the wright - fisher diffusion with values in @xmath10,1[$ ] conditioned to be killed at @xmath2 . _
key words : _ quasi - stationary distribution , interacting particle system , empirical process , yaglom limit , diffusion process . _ msc 2000 subject : _ primary 65c50 , 60k35 ; secondary 60j60 |
in the breakthrough experiment reported by k. tanaka et al .
@xcite , the matter radius of the carbon isotope @xmath0c was extracted via a finite - range glauber analysis under an optical - limit approximation of the reaction cross section , for @xmath0c on a liquid hydrogen target , measured around 40 mev / nucleon .
the extracted matter radius presents a huge value of [email protected] fm ( for a viewpoint on @xcite , see also ref .
@xcite . ) , which characterizes this nucleus as the heaviest halo nuclei discovered until now .
for the two - neutron separation energy , @xmath6 , they also quote a value of [email protected] mev .
these experimental results , together with other well - known properties of carbon isotopes @xcite , indicate that @xmath0c is weakly - bound , having a very large two - neutron halo with the @xmath3c as a core , such that the corresponding observables are probably dominated by the tail of the three - body wave function in an ideal @xmath4-wave three - body model , as considered in ref . @xcite .
in addition , within a @xmath7c configuration , we have the @xmath0c as a borromean halo system , considering that a neutron ( @xmath8 ) and @xmath3c is known as an unbound system . in view of its very low - energy properties , within the @xmath9core halo - nuclei systems , the nucleus
@xmath3c has been cited previously @xcite as a good candidate to present three - body efimov states @xcite . considering that this nucleus is more compact than the @xmath0c , with its ground state in a probable @xmath10 configuration , it is also quite natural to suggest the @xmath0c as being still more favorable to have an efimov state @xcite , with its halo predominantly produced by a @xmath11 component .
however , to infer about the possibility of existence of efimov excited states in an ideal @xmath4-wave two - neutron halo nucleus like @xmath0c @xcite is crucial to have a measurement of the virtual state energy of @xmath1c .
the characteristics of @xmath0c , roughly described in the first and second paragraphs , allow us to use a dirac@xmath12 ( zero - range ) interaction , as reviewed in refs .
@xcite , acting on @xmath4-wave to study this problem . in the zero - range limit
three scales emerge for describing the full long - range structure of the @xmath7c wave function : the virtual @xmath13 energy , the @xmath4-wave virtual state energy of the neutron in @xmath1c and the two - neutron separation @xmath6 .
the information on the unbound @xmath14c virtual energy is unknown and @xmath6 has an uncertainty that is about twice its own value . in this study , we calculate a region to guide the experiments to search for the @xmath0c two - neutron separation energy . by first considering that the virtual state energy of @xmath1c is varying from 0 to 100 kev , and that the bound - state energy of @xmath0c is given in an interval from 100 to 1500 kev
, we calculate the mean - square distance of the halo neutron to the center - of - mass ( cm ) of the corresponding three - body system as a function of @xmath6 .
then , by using the extracted one - nucleon mean - distance @xmath15 and its uncertainties as constraints , we are able to estimate a reasonable region for the search of the two - neutron separation energy in @xmath0c , as well as the corresponding region of the virtual state energy of @xmath1c ( directly related to a negative scattering length of the @xmath14c system ) .
the three - body halo wave function allows us to calculate the neutron mean - square distance to the corresponding three - body cm .
this model has already been applied with success to describe halo radii in ref . @xcite and two - neutron correlation functions in ref .
@xcite .
the available quantity that can be used to define limits on the two - neutron halo @xmath0c binding is the extracted matter radius , @xmath17 fm , that was recently reported in ref .
@xcite . the root - mean - square distance , from the cm of the @xmath7c to one of its halo neutrons , can be estimated by using the additional information on the matter radius of the loosely bound @xmath3c , which is given in ref .
@xcite ( @xmath180.05 fm ) . in view of the large difference between the radius of @xmath0c and @xmath3c , we consider it is a reasonable approximation to assume @xmath3c as the core for the present purpose , such that we still can use a three - body approach .
the result of our estimation is given by the following : @xmath19}\approx 15\pm 4\ ; { \rm fm } \ , \label{rn}\end{aligned}\ ] ] where @xmath20 and @xmath21 , with @xmath22 .
this simple approximation shows that @xmath0c is the largest known halo along the neutron dripline . by using this value
, we will be able to define a region where the @xmath1c virtual energy can be found , as well as the corresponding two - neutron separation energy , @xmath6 , in @xmath0c . in the following ,
the faddeev formalism is developed by considering a renormalized zero - range three - body model for a system with a core ( @xmath23 ) , which will be the @xmath24 in the present work , and two - identical particles ( the neutrons ) .
the mass of the core is given by @xmath25 , where @xmath26 defines the mass ratio and @xmath27 is the neutron mass . throughout this article
, we will use units such that @xmath28 . in the renormalization procedure ,
the kernel regularization is done via a subtraction method also considered in @xcite .
after partial wave projection , the @xmath4-wave coupled subtracted integral equations , for two neutrons and a core , can be written in momentum space by a coupled equations for the spectator functions @xmath29 and @xmath30 , as follows : @xmath31 , \label{chi2}\end{aligned}\ ] ] where @xmath32^{-1 } , \label{tau1 } \\
\tau_{nc}(y;\epsilon_3)&\equiv & \frac{1}{\pi}\left(\frac{a+1}{2a}\right)^{3/2 } \left[\sqrt{\epsilon_3+\frac{a+2}{2(a+1 ) } y^2 } + \sqrt{\epsilon_{nc}}\right]^{-1}\ , \label{tau2 } \\
g_1(y , x;\epsilon_3)&\equiv & \log \frac{2a(\epsilon_3+x^2+xy)+y^2(a+1)}{2a(\epsilon_3+x^2-xy)+y^2(a+1 ) } \nonumber \\ & - & \log\frac{2a(1+x^2+xy)+y^2(a+1)}{2a(1+x^2-xy)+y^2(a+1 ) } , \label{g1 } \\
g_2(y , x;\epsilon_3)&\equiv & \log \frac{2(a\epsilon_3 + xy)+(y^2+x^2)(a+1)}{2(a \epsilon_3-xy)+(y^2+x^2)(a+1 ) } \nonumber \\ & - & \log \frac{2(a + xy)+(y^2+x^2)(a+1)}{2(a -xy)+(y^2+x^2)(a+1)}. \label{g2}\end{aligned}\ ] ] in the above , we are using the odd - man - out notation for the spectator functions @xmath33 .
the indexes @xmath8 or @xmath23 in @xmath33 indicates the spectator particle .
the momentum and energy variables are written in terms of a momentum three - body scale @xmath34 , which is used in our subtractive regularization procedure to renormalize the originally singular faddeev equations .
the units considered in eqs .
( [ chi1]-[g2 ] ) are such that all quantities are dimensionless . in view of that , the corresponding dimensionless energies for the three - body system are given by @xmath35 @xmath36 @xmath37 , where @xmath38 kev and @xmath39 are , respectively , the @xmath13 and the @xmath14c virtual - state energies .
the mean - square distance of the neutron to the cm of the three - body system is calculated from the derivative of the fourier transform of the respective matter density with respect to the square of the momentum transfer .
the fourier transform of the one - body densities defines the respective form factor , @xmath40 , as a function of the dimensionless momentum transfer @xmath41 .
thus , for the mean - square distance of the neutron to the cm of @xmath0c , we have @xcite @xmath42 where the form factor is defined as : @xmath43 the above three - body wave function , @xmath44 , in momentum space are given in terms of the spectator functions @xmath33 as : @xmath45,\end{aligned}\ ] ] where @xmath46 is the relative momentum of the pair and @xmath47 is the relative momentum of the spectator particle to the pair .
the calculation of the neutron average distance to the cm of @xmath0c demands as input the @xmath6 , the energies of the virtual @xmath4-wave states of the @xmath13 and @xmath1c systems . the unbound @xmath1c virtual state is poorly known . in our model
we assumed small values of this virtual state between 0 - 100 kev .
the one - neutron mean distance to the cm , @xmath48 , derived from eqs .
( [ ra ] ) and ( [ f3 ] ) and using the wave function ( [ psi ] ) can be written as a general function @xmath49 , dependent on the two - body energies , as : @xmath50 where the plus sign ( minus ) refers to bound ( virtual ) two - body subsystem .
the value of the separation energy is given by @xmath51 .
to convert all results of the calculations to the physical units we have to introduce the physical value of @xmath6 in ( [ rscal ] ) . in this case the value of the parameters @xmath52 and @xmath53 are determined as : @xmath54 from ( [ rscal ] ) and ( [ e2 ] ) , the average distance from the neutron to the cm of the system is given by @xmath55 the limit cycle @xcite is achieved when @xmath52 and @xmath53 tends to zero and it is used to compute the radius of the shallowest @xmath56 bound state .
therefore , in this limit , the dependence on @xmath57 can be dropped out : @xmath58 in practice such limit is achieved fast and the first cycle is enough for the application we are considering ( see ref .
@xcite ) . from experimental data , we have @xmath59 fm , as given by eq .
( [ rn ] ) , and the singlet @xmath13 virtual state energy @xmath38 kev .
therefore , in order to use the model results from ( [ rscal2 ] ) , we have to assume values for the unknown virtual state energy of @xmath1c , to be able to get some information on @xmath6 in @xmath60 . in fig.[c22fig ] , we display our results for the separation energy for different values of the @xmath4-wave neutron virtual state in @xmath61 , ranging from 0 up to 100 kev .
the experimental values of @xmath62 mev @xcite and @xmath63 fm are shown in the figure .
we observe that , for a given @xmath6 , the @xmath15 decreases as the absolute value of the virtual state energy increases .
this can be explained as follows : as the virtual state energy increases , the interaction between the neutron and the core becomes weaker .
therefore , one can fix a given three - body energy by decreasing the size of the system @xcite . by taking into account the value of @xmath64 fm , one obtains @xmath6 below @xmath20.4 mev for a neutron in @xmath1c bound at the threshold .
this result is not far from the central experimental value of 0.42 mev .
we note that a small increase in the virtual state energy up to 20 kev , drops the upper limit of @xmath6 to @xmath20.3 mev .
indeed , the finite - range glauber analysis under an optical - limit approximation of the reaction cross section , for @xmath0c on a liquid hydrogen target , measured around 40 mev / nucleon @xcite , indicates that the observed large enhancement of the cross - section compared to the neighbor carbon isotopes , suggests that values of @xmath6 below 0.4 mev would be possible . the three - body approximation that we have considered for @xmath0c , where the @xmath3c is the core ,
can be justified by comparing the size of @xmath3c with the mean distance of the halo neutrons of @xmath0c and also considering that the halo neutrons in @xmath3c are bound with about 3.5 mev , one order of magnitude greater than @xmath6 in @xmath0c .
thus , the halo neutrons in @xmath0c have a much larger probability to experience the long - range @xmath65 potential derived by efimov than in @xmath3c , as the corresponding wave function tail is extending far beyond the size of @xmath3c .
therefore , the efimov physics should be much more evident in the properties of @xmath0c ground state than in the corresponding properties of @xmath3c . in a microscopic 5-body description , beyond the present model ,
the four neutrons out of @xmath66c , should be in a fully antisymmetric wave function due to the proposed separation of scales . as the @xmath4-wave radial wave functions corresponding to the neutrons in the halo of @xmath3c and in @xmath0c have different sizes , an antisymmetric wave function can be built .
if all spectator neutron interactions are dominated by only @xmath4-waves , as in our model , the pauli exclusion principle would make the halo neutrons in @xmath0c much less bound than in @xmath3c , which indeed seems to be the case .
in essence , with the above remarks , we should emphasize that our three - body model for @xmath0c is not excluding a three - body model for @xmath3c as having a two - neutron halo or an efimov state for @xmath3c very near the scattering threshold @xcite .
one possible correction to our results is due to the interaction range .
range corrections in the calculation of different mean distances were performed by canham and hammer @xcite . by taking the @xmath67li ( borromean @xmath68li system ) for comparison , where @xmath69 300 kev and the neutron average distance to the cm is around 6 fm ,
the correction is a fraction of 1 fm @xcite for a fixed @xmath6 , @xmath70li virtual state energy and @xmath71 scattering length .
therefore , we also expect corrections of the same magnitude , or even smaller , considering that the core is larger but the average distance of the neutron to the cm is more than twice .
we should stress that effects from the detailed core dynamics in our calculation are implicitly carried out by the three - body energy , which in our framework is an external parameter .
summarizing , from the extracted matter radius of @xmath0c , by using a renormalized three - body zero - range model , we estimate the mean - square distance of a halo neutron with respect to the cm of the @xmath0c . from such estimate , we suggest a possible region for an experimental search of both @xmath6 of @xmath0c and the @xmath14c virtual state energy .
the @xmath0c is approximated by a borromean three - body system composed by a point - like core of @xmath3c and two @xmath4-wave halo neutrons .
the validity of our findings relates to the assumption of a large halo compared to the typical range of the nuclear interaction .
we are confident that the guidance provided by this work would help the search for the @xmath0c energy from an experimental analysis .
finally , based on fig . 1 of ref .
@xcite , where it is calculated a region for the appearance of excited efimov states , we would like to mention that a @xmath1c with an energy close to zero can make the @xmath0c as the most promising borromean candidate to present excited efimov states , or a continuum resonance @xcite .
our thanks to the brazilian agencies fapesp and cnpq for partial financial support .
we also would like to thank prof .
yasuyuki suzuki for suggesting the study of @xmath0c .
18 k. tanaka _
et al . _ ,
* 104 * ( 2010 ) 062701 .
k. w. kemper , p. d. cottle , physics * 3 * ( 2010 ) 13 .
d. v. fedorov , a. s. jensen , k. riisager , phys .
* 312 * ( 1993 ) 1 ; phys .
* 73 * ( 1994 ) 2817 .
a. e. a. amorim , t. frederico , l. tomio phys .
c * 56 * ( 1997 ) r2378 .
brown , progr . part
. nucl . phys .
* 47 * ( 2001 ) 517 .
e. nielsen , d.v .
fedorov , a.s .
jensen , e. garrido , phys .
* 347 * ( 2001 ) 373 ; a. s. jensen , k. riisager , d. v. fedorov , e. garrido , rev .
mod . phys .
* 76 * ( 2004 ) 215 ; e. braaten , h .- w .
hammer , phys .
* 428 * ( 2006 ) 259 .
w. horiuchi , y. suzuki , phys .
c * 74 * ( 2006 ) 034311 .
i. mazumdar , a. r. p. rau , v. s. bhasin , phys .
* 97 * ( 2006 ) 062503 .
m. t. yamashita , t. frederico , l. tomio , phys .
* 99 * ( 2007 ) 269201 .
m. t. yamashita , t. frederico , l. tomio , phys .
b * 660 * ( 2008 ) 339 .
v. efimov , phys .
b * 33 * ( 1970 ) 563 ; v. efimov , sov . j. nucl . phys .
12 , 589 ( 1971 ) .
l. tomio , eur .
. j. web of conf . * 3 * , 01013 ( 2010 ) .
m. t. yamashita , t. frederico , l. tomio , nucl .
a * 735 * ( 2004 ) 40 .
m. t. yamashita , t. frederico , l. tomio , phys .
c * 72 * ( 2005 ) 011601(r ) .
et al . _ ,
phys . a * 691 * ( 2001 ) 599 .
s. d. glazek , k. g. wilson , phys .
* 89 * ( 2002 ) 230401 ; r. mohr , r. furnstahl , h. hammer , r. perry , k. wilson , ann . of phys .
* 321 * ( 2005 ) 225 .
m. t. yamashita , t. frederico , a. delfino , l. tomio , phys .
a * 66 * ( 2002 ) 052702 .
f. bringas , m. t. yamashita , t. frederico , phys .
a * 69 * ( 2004 ) 040702(r ) .
d. l. canham and h .- w .
hammer , nucl .
a * 836 * ( 2010 ) 275 . | the recently extracted matter radius of carbon isotope @xmath0c allows us to estimate the mean - square distance of a halo neutron with respect to the center - of - mass of this nucleus . by considering this information , we suggest an energy region for an experimental investigation of the unbound @xmath1c virtual state .
our analysis , in a renormalized zero - ranged three - body model , also indicates that the two - neutron separation energy in @xmath0c is expected to be found below @xmath20.4 mev , where the @xmath0c is approximated by a borromean configuration with a pointlike @xmath3c and two @xmath4-wave halo neutrons .
a virtual - state energy of @xmath1c close to zero , would make the @xmath0c , within borromean nuclei configurations , the most promising candidate to present an excited bound efimov state or a continuum three - body resonance .
pacs 27.30.+t,21.10.gv , 21.45.+v,11.80.jy , 21.10.dr , , and binding energies , faddeev equation , three - body |
molecular electronics describes charge transport processes whereby molecules serve as active elements ( e.g. , rectifiers , switches , sensors ) or passive ones ( resistors or surface passivating agents ) in electronic devices @xcite . since the emergence of this field ,
@xcite it has expanded greatly and now includes several types of device configuration ( e.g. , two - terminal junctions@xcite , three - terminal junctions @xcite and electrochemical devices @xcite ) , where the molecules serve as direct current carriers @xcite or indirectly affect the electrical properties of a junction @xcite . despite the diversity of investigated systems , fundamental questions regarding _ the mechanisms for electrical current passing through molecules between two electrodes and the possibility of gaining predictive power and control over the electrical properties of molecular junctions _ remain mainly unsolved .
detailed discussions on the limitations of existing transport models can be found elsewhere.@xcite in short , even for one of the simplest systems , i.e. that of alkyl chains between au electrodes , there is a large discrepancy between the average tunnel barrier height , extracted from current - voltage ( @xmath1 ) measurements ( @xmath2{ev}$ ] ) , @xcite and the barrier that is expected from the experimentally determined electrode work function and alkyl monolayer ionization potential , i.e. the energy difference between the fermi level and highest occupied molecular orbital ( homo ) as found by ultraviolet photoelectron spectroscopy , ups ( @xmath3{ev}$]).@xcite in spite of this apparent discrepancy , homo - dominated transport is the prevailing concept for such junctions . @xcite in the present study we challenge this concept and find that transport is dominated by the energy difference between the fermi level and the lowest unoccupied molecular orbital ( lumo ) of the molecules .
it would seem that having a semiconductor ( sc ) instead of a metal as one of the electrodes in a molecular junction increases the difficulty in understanding the mechanisms that govern charge transport through the junction .
indeed , with moderately doped scs , the sc depletion layer has a large effect on the overall observed current and if the band bending is very large , it completely overwhelms any molecular effect.@xcite as we shall see , though , the intrinsic asymmetry of a junction with a sc , with respect to charge carrier type , provides unique information on the energy levels involved in transport , _ in ways that are impossible when using only metal electrodes_. to find how charge is transported through a monolayer of saturated alkyl chains we measure and analyze electrical transport across heavy - doped semiconductor - molecular insulator / metal junctions .
our results provide unequivocal evidence that charge transport in these junctions is controlled by the energy difference between the metal fermi level and the lumo of the molecule , @xmath4 , rather than by the energy difference with the homo , @xmath5 .
this finding agrees with complementary spectroscopic measurements of the samples without hg contact , but challenges the prevailing concept of homo - dominated transport . @xcite
highly doped n@xmath0- and p@xmath0- si - c@xmath6h@xmath7/hg junctions ( @xmath8 ) were prepared by alkylation of freshly etched p- or b - doped ( @xmath9{cm^{-3}}$ ] ) si@xmath10 surfaces .
detailed descriptions of the preparation and characterization of both long ( c14-c18 ) and short alkyl chain ( c2 ) samples can be found elsewhere , and are given in the suppl .
( section 1 and section 2 ) .
- ( solid line ) and n@xmath0- ( dashed line ) si - c@xmath11h@xmath12/hg junctions .
the conductance of the n@xmath0-junction is parabolic around @xmath13{v}$ ] , while for the p@xmath0-junction at negative bias on the metal there is a voltage range , where the conductance increases slowly ( shaded area ) , followed by a sharp increase .
the transition bias between the above - mentioned regions is @xmath14{v}$ ] , very close to the si band gap ( @xmath15{ev}$ ] ) . ( a ) ,
( b ) & ( c ) refer to voltage ranges , shown in fig . 2 and discussed in the text , for the p@xmath0 junction only.,title="fig : " ] + in fig . [ normcond ] ,
the normalized conductance of p@xmath0-/n@xmath0-si - c@xmath11h@xmath12/hg junctions is compared on a linear scale . while conductance for the n@xmath0 junction ( dash ) is parabolic around @xmath13{v}$ ] ( i.e. similar to the behaviour expected for metal -insulator -metal ( mim ) tunnel junctions @xcite , for the p@xmath0 junction ( solid ) there is a negative bias range where conductance increases slowly ( shaded ) , followed by a sharp increase at @xmath14{v}$ ] , which closely corresponds to the si band gap ( @xmath16{ev}$ ] ) @xcite .
this behaviour is highly reproducible and independent of the molecular length ( c@xmath17h@xmath18-c@xmath19h@xmath20 ; see ref . for @xmath21 curves of n@xmath0-si data and fig .
[ jvlong ] below for the p@xmath0-si ones ) .
the results of the p@xmath0 junction at negative applied bias can be understood by considering work on metal - inorganic insulator- ( near ) degenerate semiconductor ( mis ) tunnel diodes by esaki , sze , dahlke and co workers,@xcite which forms the basis for the present study .
schematic band diagrams of the relevant model are shown in fig .
[ esaki ] ( top ) , along with an experimentally measured semi log @xmath21 curve of a typical p@xmath0si - c@xmath11h@xmath12/hg junction .
we identify the undulating behaviour at low to moderate negative bias ( region b of fig .
[ esaki ] , which is equivalent to the shaded region in fig .
[ normcond ] ) with a region where negative differential resistance ( ndr ) was predicted originally for the mis system studied by esaki et al .
that effect is smeared out in the @xmath21 curve , because of the effect of interface states , an effect that , together with band bending and image forces , is , for simplicity s sake , neglected in the schematic band diagrams . in the figure , only @xmath22 -lumo ( i.e. , electron tunneling ) dominated transport is considered ( lumo is depicted by thick , black line ) and other cases are discussed below . semiconductor substrates and lumo - dominated transport _ ( top ) _ and semi - log j - v curve of a typical p@xmath0si - c@xmath11h@xmath12/hg junction ( averaged over 12 scans ) _ ( bottom ) _ with the ( a ) , ( b ) and ( c ) labels corresponding to the diagrams above .
three major bias regimes are considered in the different energy schemes ( top ) and separated in the @xmath21 plot by vertical dotted lines _ ( bottom ) _ , corresponding to a ) metal fermi above ( i.e. , closer to vacuum level ) the sc gap ( @xmath23 ) ) ; b ) metal fermi within the sc gap ( @xmath24 ) ; and c ) metal fermi below the sc gap ( @xmath25 ) , where @xmath26 and @xmath27 are the forbidden gap and the energy difference between the sc fermi level and the valence band edge .
top : @xmath28 is the energy difference between the tunneling electrons ( marked by a horizontal arrow ) and the molecular level ( spatially averaged ) , which is taken as the estimated effective tunneling barrier height . the insulator level is assumed to vary across the insulator .
electron tunneling into the vb ( as in b ) is practically identical to holes tunneling in the opposite direction .
surface states , band - bending and image forces are neglected for simplicity . ]
as long as the metal fermi level does not align with the sc forbidden energy gap ( i.e. , regions a , c ) transport is completely equivalent to tunneling in the more common metal - insulator - metal ( mim ) tunnel junction .
@xcite in regions a and c , the bias acts to reduce the averaged tunneling barrier by @xmath29 . with a moderate negative voltage to the metal ( fig .
[ esaki ] ) the metal fermi level moves to energies that are in the forbidden band gap of the sc .
for such a case , neglecting interface states - assisted tunneling ( isat ) , esaki and stiles predicted an effective increase in the tunnel barrier height @xmath30 ( i.e. , @xmath30 ( b ) @xmath31 ( a , c ) ) , and ndr , as has indeed been observed experimentally.@xcite if isat is not neglected , instead of ndr a @xmath32 plot will show undulating behaviour ( as we show in fig .
[ esaki ] , voltage range ( b ) , for molecular junctions ) , which is strongly influenced by the interface state density and distribution in the band gap .
@xcite in our case isat has a clear influence on the @xmath21 data and , therefore , we observe undulating @xmath21 behaviour , instead of pure ndr .
undulating @xmath21 occurs because in region b the highest energy electrons have no matching states to tunnel into , and transport proceeds via deeper electrons , aligned with the sc vb . for those electrons the barrier is now increasing as @xmath33 instead of decreasing with @xmath34 , as in band - to - band tunneling ( ranges a , c ) .
in contrast to p@xmath0 junctions , n@xmath0 junctions are not predicted to show undulating @xmath21 plots for lumo - dominated tunneling because the sc fermi and the tunneling levels are on the same side of the forbidden sc gap , regardless of applied bias ( see fig .
s3.1b of suppl . inf . ) .
simply put , undulating @xmath21 is predicted to occur if tunneling proceeds via carriers of opposite type to the sc majority carriers .
this prediction is confirmed experimentally in fig .
[ normcond ] , which shows a highly symmetrical @xmath35 curve for the n@xmath0-si junction .
if the @xmath5 energy difference did determine the transport barrier ( i.e. , hole , rather than electron tunneling ) , then the special undulating @xmath21 behaviour should be observed in the n@xmath0 junction for positive applied bias on the metal and the p@xmath0 junction should not present such behaviour at all .
the reason for the latter is that for a p@xmath0-sc - based junction and hole tunneling , * with negative bias applied to the metal , holes tunnel from the sc valence band to the metal ; * with positive bias applied to the metal , holes tunnel from the metal to the sc valence band .
neither of these flows involves tunneling of holes into the sc band gap .
thus , the undulating @xmath36 behaviour ( fig . 2 ) and the approximately constant conductance ( fig . [ normcond ] ) for the p@xmath0-si - alkyl / hg junction and lack of it in the n@xmath0 one ( fig . [ normcond ] ) provide direct evidence for @xmath4 ( or electron)- , rather than @xmath5 ( hole)-dominated transport in both doping types . to put this in other words , if transport were homo - dominated , then the tunneling level and the sc fermi level would be below and above the n@xmath0-sc gap ( at positive bias ) , while both the homo and the sc fermi level would be always below the sc gap for p@xmath0 junctions . thus , tunneling via homo predicts undulating @xmath21 for n@xmath0 junctions , which is not observed experimentally ( cf .
fig . in the suppl . inf . for energy schemes that illustrate these alternatives ) . we note that , for an mim junction , this type of asymmetry can not be observed , because it originates in variation in available density of states for tunneling .
indeed , scott et al have used this exact asymmetry and demonstrated that inelastic tunneling is much more pronounced in a metal / molecule - p@xmath0si junction than in a metal - molecule - metal junction @xcite the conclusion that transport is dominated by the @xmath4 energy difference implies that the lumo is closer in energy to the sc @xmath22 for both doping types .
this can be verified by characterizing the energetics of the molecularly modified si , prior to metal contact deposition . to this end
we use _ ultra - violet photoemission spectroscopy _ ( ups ) , _ inverse photoemission spectroscopy _
( ipes ) , _ x - ray photoelectron spectroscopy _ ( xps ) and _ contact potential difference _ ( cpd )
we then study the si band bending at equilibrium after top metal deposition by measuring charge transport across molecular monolayers ( si - c@xmath37h@xmath38/hg ) that are so thin that they do not present an effective tunneling barrier .
finally , we perform a molecular monolayer length - dependent study of sic@xmath6c@xmath7/hg junctions where n=14 , 16 and 18 .
these monolayers are thick enough to make tunneling the dominant current transport process . from this length dependence study
we extract the tunneling barrier height and compare it to the barrier that is expected from the ups / ipes results .
we begin with determining the band bending ( bb ) in si in the p@xmath0- and n@xmath0-si - c@xmath11h@xmath12 systems .
the position of the fermi level is known in our system from frequently repeated ups and ipes measurements of the fermi step on atomically clean metal ( e.g. au ) surfaces .
given the nearly @xmath39{\aa}$ ] thickness of the sam layer , the si valence band maximum ( vbm ) and conduction band minimum ( cbm ) are not visible in ups and ipes , and we must use here the xps measurement of the position of the si 2p core level , which is clearly detectable through the sam layer , to evaluate the vbm position and the bb . indeed , the energy difference between the core level and the vbm , called here ( si2p - vbm ) is fixed , and we can use the position of either one to determine that of the other .
combined ups / xps measurements done in our laboratory on carefully prepared hydrogen - terminated si surfaces give ( si2p - vbm)@xmath40{ev}$ ] , and we use this value here for the determination of the bb .
note that , in spite of being obtained with different experimental systems on differently prepared surfaces , this value compares well with that published in the literature.@xcite the si 2p peaks are found at @xmath41{ev}$ ] and @xmath42{ev}$ ] below the fermi level for the p@xmath0- and n@xmath0-si- c@xmath11h@xmath12 surfaces , respectively .
thus , simple considerations give : @xmath43{ev } - \unit[98.8]{ev } = \unit[0.2]{ev } \\ \mathrm{bb_{n - si } } & = & e_{\mathrm{gap } } - ( e_\mathrm{f } - \mathrm{vbm } ) \\ & = & \unit[1.1]{ev } - ( \unit[99.8]{ev } - \unit[98.9]{ev } ) = \unit[0.2]{ev}\end{aligned}\ ] ] both p-@xmath0 and n@xmath0-si surfaces are therefore seen to be slightly depleted with bb@xmath44{ev}$ ] .
the small bb on both types of substrate is consistent with the known ability of these alkyl monolayers to passivate si.@xcite the cbm in each case is obtained by adding the @xmath45{ev}$ ] si band gap to the position of the vbm . from the position of the photoemission onset of the he(i)(@xmath46{ev}$ ] )
ups spectra , we find that the work functions of p@xmath0- and n@xmath0-si - c@xmath11h@xmath12 are @xmath47{ev}$ ] and @xmath48{ev}$ ] , respectively .
similarly , the cpd measurements performed in the dark also yield a @xmath49{ev}$ ] difference between the two surfaces , in excellent agreement with the photoemission data .
incidentally , cpd also shows these values to be independent of the number of carbons in the alkyl chain ( 2,14,16,18 ) for a given substrate ( i.e. p@xmath0- or n@xmath0-si).the difference between work functions of p@xmath0- and n@xmath0-si - c@xmath11h@xmath12 ( @xmath49{ev}$ ] ) is smaller than that expected from the difference in the si 2p core level positions for the two surfaces ( @xmath50{ev}$ ] ) .
this may indicates that the si - c bond polarization , which contributes to the interface dipole , is affected by the position of the si fermi level .
strictly speaking , the homo and lumo are concepts adapted from a single molecule gas phase model , while here we use them for a monolayer in close contact with a substrate .
therefore , in the remaining of the text , we refer to them as hoso and luso , highest occupied and lowest unoccupied system orbital , @xcite similar to homo * and lumo * in ref . .
[ upsipes ] presents the combined results of he(ii)(@xmath51{ev}$ ] ) ups and ipes measurmenst both for p@xmath0 ( top ) and n@xmath0-si - c@xmath11h@xmath12 .
we define the edges of the monolayer band gap as the crossing between the tangent lines of the spectra . while the edge of the ups spectrum is defined clearly ( both for p@xmath0 and n@xmath0 ) , from the ipes spectra , two different edges can be defined .
the reason is that the ipes incident electrons have a larger penetration depth than the escape depth of the ups electrons . as a result
, the interface states are more pronounced in the ipes spectrum than in the ups spectrum .
these states , which result from hybridization between the alkyl molecular states and the si vb and cb states at the interface are known as induced density of interface states ( idis ) and were calculated to be localized in the first three carbons of the alkyl chain ( with respect to the si surface).@xcite idis is probably the main physical origin for the onset of the intensity in the ipes spectrum ( fig .
[ upsipes ] ) . as a result
, the ipes spectrum has a `` two slope '' behavior , noted as # 1 and # 2 in fig .
[ upsipes ] for both the p@xmath0 and n@xmath0 samples .
we view slope # 1 as expressing the idis and slope # 2 originating from the actual luso .
we note that experimentally determined edges of the gap do not necessarily correspond exactly to the hoso and luso of the monolayer .
the onsets of ups and ipes intensity , which define the gap edges , can be affected by spectral broadening , due to polarization of the molecules in the dense monolayer , @xcite by idis , @xcite and by radiation damage.@xcite the hoso and luso positions can be found by fitting the calculated density of states results , as reported by segev et al . for the si - c@xmath6h@xmath7 system @xcite to the experiment .
in their calculations segev et al . , identified the actual luso to be @xmath52{ev}$ ] above the edge , defined by slope # 2 .
the actual hoso was identified to be @xmath53{ev}$ ] below the ups edge .
the main implication of this calculation for the present study is that , while there is some experimental uncertainty regarding the actual position of the hoso , the luso and band gap width , this uncertainty is unidirectional ( i.e. , the experimental edge to edge gap is the lower limit of the actual gap ) and a similar situation holds for both ups ( hoso ) and ipes ( luso ) . figures [ upsipes]b and c clearly show that for both p@xmath0 and n@xmath0 samples , the fermi level is closer to the onset of the ipes spectrum ( luso ) than to the onset of the ups spectrum ( hoso ) and the estimated uncertainty of the exact position of the hoso and luso is marked by a shaded area .
the apparent difference in the edge to edge gap in fig.[upsipes ] for the molecular monolayers on p@xmath0 and n@xmath0 , is within this uncertainty . for p@xmath0-si - c@xmath11h@xmath12 samples ,
the energy difference between the si @xmath22 and the monolayer luso ( @xmath54{ev}$ ] ) , is smaller than that between @xmath22 and the hoso ( @xmath55{ev}$ ] ) .
similarly , for the n@xmath0-si - c@xmath11h@xmath12 samples , the energy difference between the si @xmath22 and the monolayer luso ( @xmath56{ev}$ ] ) is smaller than that between the si @xmath22 and the hoso ( @xmath57{ev}$ ] ) . a prominent feature in fig . [ upsipes ]
is that the intrinsic band alignment somewhat varies with the doping density of si . in other words ,
the energy difference between the luso ( hoso ) and the conduction ( valence ) band edges of the si is not constant and shifts with the position of the fermi level within the si band gap .
this result is surprising , because it is normally assumed that the electronic molecular properties ( ionization potential and electron affinity ) are independent of the properties of the substrate .
still , this result is highly reproducible experimentally and the same phenomenon was also observed in other hybrid organic - inorganic systems.@xcite possibly this is related to the above - hypothesised doping - dependent si - c bond polarization .
regardless of this result , which is currently under further study , the key experimental finding is that , for both p@xmath0 and n@xmath0 si systems , the fermi level of the system and the carrier transport level in the si are significantly closer to the luso than to the hoso .
we now ask if the spectroscopic results presented in fig . [ upsipes ] can efficiently predict the charge transport characteristics of the alkyl chain monolayer ? to answer this , we compare the effective tunneling barrier height that we can deduce from the @xmath21 measurements with the expected barrier that we find from the ipes spectrum i.e. @xmath58 energy difference . to this end
we need to take under consideration the si band bending in the junction at equilibrium ( zero bias ) , rather than the band bending at the si surface as determined by xps for the free molecularly- modified si surface , because that band bending can change after forming a metal ( hg ) contact on the molecules .
because the molecular monolayer is very thin ( @xmath59{\aa}$ ] ) and supports a large tunneling current , there can be charge transfer between the metal and the si substrate across the molecules , until electronic equilibrium is established ( equalization of the fermi levels of the si and the metal ) .
this charge transfer affects the si band bending , but should not change the alignment between the si bands ( at the surface ) and the monolayer hoso and luso , assuming that the molecules do not become charged .
elsewhere we showed @xcite that , due do the large surface dipole that is introduced by the molecules , depositing hg on a moderately doped n - si - c@xmath6h@xmath7 sample results in a large band bending in the si , up to the point where the si surface is in strong inversion and charge transport in the junction is controlled by minority carrier recombination .
for the heavy - doped si used here , the effect of the band bending on transport is usually much smaller , because the depletion layer is thinner ( @xmath60{nm}$ ] compared to @xmath61{\upmu m}$ ] for moderately doped si ) and allows some charge tunneling through it .
this type of charge transport mechanism , which is intermediate between thermionic emission and pure tunneling , is known as thermionic field emission ( tfe ) .
@xcite measurements of p@xmath0 and n@xmath0 si - c@xmath37h@xmath38/hg junctions ( solid and dashed curves , respectively ) , compared to the j - v behavior of p@xmath0 and n@xmath0si - h / hg junctions ( dotted curve ; p@xmath0 and n@xmath0 curves are identical ) .
the error bars represent the standard deviation of multiple measurements on 8 different junctions .
, title="fig : " ] + to isolate the effect of alkyl molecules on the si band bending from tunneling through the molecules , we performed @xmath21 measurements on a junction with an ultrathin molecular monolayer , c@xmath37h@xmath38 ( fig .
[ jvshort ] ) , and compared the results to those obtained with freshly etched , h - terminated si . because the c@xmath37h@xmath38 monolayer is so thin ( @xmath62{nm}$ ] )
, it presents a negligible charge transport barrier , compared to the depletion layer in the si .
combined with cpd evidence that the surface dipole is un - altered by molecular length , the ultra - short alkyls allow quantifying the effect of the depletion layer on net transport .
the @xmath21 characteristics of the si - h / hg junction were measured as a control ( dotted curve ) and found to be ohmic and identical for p@xmath0 and n@xmath0 junctions .
the results for si - c@xmath37h@xmath38/hg are very different with the p@xmath0 junction exhibiting ohmic behavior with current density even higher than that of the si - h samples .
the n@xmath0 junction yields an exponential increase at both bias polarities , a behavior that is typical of charge transport via tfe .
@xcite the high current density of the p@xmath0 junction is consistent with our assumption that the tunnel barrier presented by the c@xmath37h@xmath38 monolayer is negligible , and that the dipole direction works to slightly decrease the si band bending .
the same dipole , on the n@xmath0-si surface should , therefore , increase the band bending .
indeed , when we fit the n@xmath0 results at forward bias ( positive bias on the metal ) to an analytical expression for tfe developed by padovani and stratton , @xcite we find a barrier height of @xmath63{ev}$ ] .
because the si is heavy - doped and the fermi level is near the cb , this is also roughly the n@xmath0-si band bending value at equilibrium.@xcite thus , for p@xmath0-si junctions there is negligible internal barrier in the si , while such a barrier exists for the n@xmath0-si junction , where it has a large effect on the overall @xmath21 curve . while this does not necessarily mean that the molecular contribution to transport is overwhelmed by band bending in n@xmath0-si , as is the case with moderately doped si junctions,@xcite this band bending decreases the current density significantly .
due to the large internal barrier in the n@xmath0-si , we limit ourselves to the molecular length dependent study of the p@xmath0-si junction , and then only to the data for positive applied bias on the metal , where according to fig .
[ normcond ] , current is mostly due to band - to - band tunneling with the density of states available for tunneling approximately constant with applied bias .
the reason is that , as we noted earlier , for negative applied bias the dominant charge transport mechanism is isat , i.e. , transport is controlled by the recombination rate at the si surface and can not be described with conventional tunneling models.@xcite si - c@xmath6h@xmath7/hg junctions ( @xmath64 and 18 ) .
data for the si - h / hg junctions ( dashed ) are given as reference .
the current density exponentially decreases with the molecular length .
error bars represent the standard deviation of at least 8 different junctions .
, title="fig : " ] + fig .
[ jvlong ] presents the @xmath21 behavior of p@xmath0-si - c@xmath6h@xmath7 /hg junctions ( n=14,16 and 18 ) ( the length dependence of the n@xmath0-si - c@xmath6h@xmath7 /hg junctions were published elsewhere @xcite ) .
it is clear that the qualitative behavior of the j - v curve is independent of chain length .
it is also clearly seen that , as expected for such junctions , the current density decreases exponentially with the number of methylene units in the alkyl chain .
@xcite this type of behaviour is generally described by the landauer relation,@xcite in this model , the conductance of a single channel ( g ) is given as : @xcite @xmath65 where ` @xmath66 ' is the tunnel barrier width , which can be taken as the length of the molecule or the thickness of the monolayer , @xmath67 is the length - decay parameter and @xmath68 is the contact conductance . from a @xmath69 vs. @xmath66 plot at a given @xmath70 ,
we extract both @xmath67 ( slope ) and @xmath68 ( intercept ) .
we start by discussing the meaning of @xmath68 , followed by an interpretation the extracted @xmath67 value .
the extracted @xmath68 value ( at @xmath71{v}$ ] ) of the p@xmath0-si - alkyl / hg junctions is @xmath72 , where @xmath73 is the quantum conductance ( @xmath74{\upmu s } $ ] ) . the gc value of n@xmath0-si - alkyl / hg junctions was @xmath75 , @xcite very low compared to typical au - alkyl mim junctions .
it reflects the large band bending in the si that was extracted from the @xmath21 behaviour of n@xmath0 si - c@xmath37h@xmath38/hg observed in fig .
[ jvshort ] .
the @xmath68 found for the p@xmath0 junction is similar to that reported for an mim junction with one chemi - contact @xcite , a result that can be rationalized by recalling the near flat band conditions in the p@xmath0 -si - c@xmath37h@xmath38/hg junction ( fig .
[ jvshort ] ) . could serve as a standard value to determine the effect of the band bending on the conduction properties of a molecular mis junction , that is unlikely to be generally valid .
the reason is that @xmath68 should be highly contact dependent and will vary between metals and binding groups , ( n.d . ) . ] to further verify the correlation between the spectroscopic evidence ( fig .
[ upsipes ] ) and the transport results , we estimate the tunnel barrier ( @xmath30 ) from the @xmath21 curves . this barrier should , in principle , reflect the energy difference between the fermi level of the electrodes and the energy of the nearest allowed state in the interfacial insulator , as detected by ups and ipes . in addition , we compare the @xmath21 behavior of our p@xmath0-si - alkyl / hg junction to previously reported results on the more common au - s - alkyl - s - au junction . to that effect
, we use the general wkb approximation for tunneling that describes the relation between @xmath67 and @xmath30 as @xcite : @xmath76 where @xmath77 is the decay parameter at equilibrium @xmath78 , @xmath79 is planck s constant , @xmath80 is the elementary charge and @xmath81 is the electron mass .
this is by no means an accurate model that reflects the complexity of molecular junctions . nevertheless , as it is the basis for the simmons model @xcite that is widely used for the extraction of @xmath30 in molecular junctions , it allows direct comparison with previous results .
the difference in dielectric properties between vacuum and insulating materials is commonly accounted for by using an effective mass for the tunneling electron , which is smaller than the free electron mass .
using complex band structure dft calculations , tomfohr and sankey calculated the dispersion relation of the forbidden band gap of an infinite alkyl chain and showed that by using the dispersion relation in vacuum ( eq . [ 2 ] ) with an effective mass of @xmath82 , a lower limit of the barrier is extracted.@xcite thus , we use this value of effective mass to extract an estimated value of @xmath83 , which we can compare to previous studies and to the spectroscopic results presented in fig .
[ upsipes ] .
the average extracted @xmath83 in the low bias range ( @xmath84{v}$ ] ) of the @xmath21 curves in fig .
[ jvlong ] is @xmath85{\aa^{-1}}$ ] . inserting this value into eq .
[ 2 ] , along with a @xmath82 effective mass,@xcite yields a @xmath86{ev}$ ] barrier .
having established that @xmath58 dominates charge transport ( fig . [ normcond ] ) , we can locate the extracted @xmath83 value on the ipes spectrum ( marked by vertical arrow ) in fig .
[ upsipes ] .
this value is in between the ipes edges and , because it is only the lower limit of the barrier , we conclude that we have reasonable agreement with the luso of the alkyl monolayer ( edge # 2 in fig .
[ upsipes ] ) .
a tunneling barrier of @xmath87{ev}$ ] is much smaller than the @xmath88 difference of @xmath89{ev}$ ] ( fig . [ upsipes ] ) , decreasing further the possibility of significant hoso - contribution to transport .
the natural question is how far our observation for luso - dominated tunneling in alkyl - si junctions is applicable to alkyl junctions on any substrate ?
the answer depends on the position of the molecular levels of the alkyl chains relative to the electrode fermi level .
ups / ipes measurements of alkyl thiols on au @xcite yield a hoso - luso gap of @xmath90{ev}$ ] and significantly smaller @xmath58 ( @xmath91{ev}$ ] ) than @xmath88 ( @xmath92{ev}$ ] ) ( the latter is in good agreement with previously reported @xmath88 for alkyl thiols on au ( @xmath3{ev}$ ] ) .
@xcite the hoso - luso gap is similar to the gap measured for alkyl chain monolayers on both si @xcite and gaas @xcite with different binding groups ( si - c , si - o - c and gaas - po@xmath93-c ) , i.e. , _ it is not influenced significantly by binding group and substrate type_. the range of the edge - to - edge gap in all examined samples is @xmath94{ev}$ ] , if the idis onset is ignored .
these values agree with values reported for polyethylene.@xcite thus we conclude that the hoso - luso gap of alkyl monolayers , and their position relative to the vacuum level is rather independent of the contacting electrodes .
therefore , we expect tunneling to be dominated by the @xmath58 energy difference for any alkyl junctions , except maybe for electrodes of extremely high work function ( @xmath95{ev}$ ] ) .
our conclusion is at odds with earlier studies on molecular transport across alkyl chains , which concluded that transport is dominated by the @xmath88 barrier , especially for the extensively studied au-(alkythiol or alkyldithiol)-au systems.@xcite in most cases the evidence for hole tunneling is based on transport across relatively short alkyl chains , up to c12 , and is indirect . an exception is the measurement of thermo - power ( seebeck coefficient ) , where the positive coefficient was taken as direct evidence for hole / hoso - mediated tunneling .
@xcite remarkably , though , the thermo - power decreases with increasing molecular length and , when extrapolated , is predicted _ to change sign _
( i.e. , change from hoso- to luso - dominated , hole to electron tunneling ) for alkyl chains longer than c12 .
this length - dependence was rationalized by invoking metal induced gap states ( migs ) as dominating transport , rather than pure molecular states .
such an explanation is very reasonable , because with the short dithiol molecules used in those experiments ( au - sc@xmath6h@xmath7s - au with @xmath96 ) migs are expected to extend out to several carbon atoms from the contacts . in early cp - afm work on short
mono and dithiol alkyl chains the extrapolated zero molecular length resistance increased as the work function of the contacting metal decreased and this was taken to indicate an increasing tunnel barrier.@xcite similar to the thermo - power , this effect can be due to the migs , rather than to alkyl - derived energy levels . in earlier work with si as substrate ,
the type of dominant charge carrier in the semiconductor was suggested to dictate the type of tunneling through ( longer ) alkyl chains.@xcite in later work on a system , similar to the one studied here , the lack of temperature activation was taken as indicating hoso - mediated /electron tunneling @xcite , an effect that may well be present , but is overwhelmed by the effects that we have shown here . except for the last two reports , we note
that the main difference between ours and other experiments is that si - monolayer / hg junctions ( with a several @xmath97{\upmu m}$ ] diameter contact , rather than a nm - sized one ) make it possible to measure conductance through longer molecules , which allows measuring molecule- , rather than contact - dominated transport by reducing the contribution of migs and idis interface effects to net transport .
to conclude , the analogy to asymmetry in ndr for degenerately - doped mis of different doping types indicates that for both p@xmath0- and n@xmath0-type si - alkyl chain / hg junctions charge transport is controlled by @xmath58 ( rather than by @xmath88 ) .
furthermore , by comparing charge transport through p@xmath0- and n@xmath0-si - c2h5/hg junctions , we showed that there is considerable band bending in the n@xmath0-si junction , which has a large effect on the charge transport ( even though the si is heavy doped ) .
the p@xmath0-si junction , though , is near flat - band in equilibrium , with comparable transport probability to that of mim junctions .
using ups and ipes measurements we elucidated the electronic structure of si - alkyl system and showed that @xmath98 , in line with the electron tunneling qualitative argument . the tunneling barrier height
estimated from the wkb approximation also fits well with the ipes - derived @xmath58 difference .
we thank the us - israel binational science foundation ( dc , ak ) , the israel science foundation , isf , through its centre of excellence programs ( dc , lk ) , the lise meitner minerva center for computational chemistry ( lk ) , the national science foundation ( dmr-1005892 ) ) and the princeton mrsec of the national science foundation ( dmr- 0819860 ) ( ak ) , nanoned , funded by the dutch ministry of economic affairs ( project wsc.6972 ) ( hz ) and the monroe and marjorie burk fund for alternative energy studies ( dc ) for partial support . at the weizmann inst .
this work was made possible in part by the historic generosity of the harold perlman family .
oy thanks the azrieli foundation for an azrieli fellowship .
dc holds the sylvia and rowland schaefer chair in energy research . | we compare the charge transport characteristics of heavy doped p@xmath0- and n@xmath0-si - alkyl chain / hg junctions .
photoelectron spectroscopy ( ups , ipes and xps ) results for the molecule - si band alignment at equilibrium show the fermi level to lumo energy difference to be much smaller than the corresponding fermi level to homo one . this result supports the conclusion we reach , based on negative differential resistance in an analogous semiconductor - inorganic insulator / metal junction , that for both p@xmath0- and n@xmath0-type junctions the energy difference between the fermi level and lumo , i.e. , electron tunneling , controls charge transport
. the fermi level - lumo energy difference , experimentally determined by ipes , agrees with the non - resonant tunneling barrier height deduced from the exponential length - attenuation of the current . |
in the dispersive regime of circuit - qed the resonator frequency depends on the qubit state so that driving the cavity and observing the output of the cavity corresponds to a quantum non - demolition measurement . for outcome 0 " ( 1 ) the mean voltage leaving the cavity represents a time - dependent coherent state , denoted @xmath69 , that is typically small in magnitude .
the phase - space evolution of @xmath69 is determined by the deterministic differential equation @xmath70 where @xmath71 is the measurement drive and @xmath72 is the cavity decay rate . for our experiment ,
the qubit transition frequency is @xmath73 ghz and the readout resonator frequency is @xmath74 ghz .
the qubit anharmonicity is @xmath75 mhz with @xmath0 ( energy relaxation ) time of @xmath76s and coherence time @xmath77s .
the dispersive shift and line width of the readout resonator are measured to be @xmath78 mhz and @xmath79 khz , respectively so that @xmath80 .
the output mode @xmath81 from the resonator can be related to the field inside the resonator by @xmath82 where @xmath83 is the separation between the pointer states , @xmath84 is the mean value of the coherent states and @xmath85 represents the qubit information with @xmath86 and @xmath87 being the probability of the qubit being in state 0 and 1 respectively @xcite .
typically this field is amplified by a linear phase - preserving amplifier to the output mode @xmath88 , @xmath89 where @xmath90 is the power gain and @xmath91 is the extra noise added by the amplifier .
here we have made an assumption that the bandwidth of the amplifier is constant and larger than the bandwidth of the signal being measured . since the commutators must satisfy @xmath92 & = \left[b(t),b^{\dagger}(t^\prime)\right ] = \openone \delta(t - t^\prime),\end{aligned}\ ] ] the noise must satisfy @xmath93 = \openone(1-g)\delta(t - t'),\end{aligned}\ ] ] which , from the generalized uncertainty principle of an arbitrary operator @xmath94 , @xmath95 . using eq .
[ eq : outputmode ] the noise in the output mode @xmath88 is @xmath96is the added noise normalized by the gain @xcite . in our view
this is the best number to quantify an amplifier and for the quantum limit it takes the value 1/2 for a phase preserving amplifier .
other useful quantities are the instantaneous input and output signal - to - noise ratios of the amplifier defined by @xmath97 where @xmath98 is the efficiency of the amplifier and represents how well the input field is mapped to the output field .
another useful quantity is the noise - figure @xmath99 of the amplifier , which is the ratio @xmath100 .
we see that @xmath101 which is 2 for a quantum limited amplifier . in circuit
qed the information about the qubit state is contained in a single quadrature . from eq .
[ eq : field ] this is the quadrature set by @xmath102 and as a result when we subtract the mean value we obtain @xmath103 this overestimates the noise as the information is only in one quadrature . defining the measurement quadrature @xmath104 $ ] it is simple to show that @xmath105 @xmath106 where @xmath107 is the efficiency of measuring information in a single quadrature for a linear phase preserving amplifier . note this is a factor of two less than the efficiency of the amplifier .
a general linear amplifier can be described by the output mode @xmath88 , @xmath108 and from preservation of the commutation relations @xmath109
= \openone(1-m^2+|l|^2)\delta(t - t'),\end{aligned}\ ] ] we have @xmath110 . setting @xmath111 results in @xmath112 . to amplify a single quadrature
@xmath113 $ ] ( phase sensitive amplifier ) we set @xmath114 giving @xmath115 + \mathrm{re } [ h(t)e^{-i\theta}],\\ q(t)&=\frac{1}{\sqrt{g } } \mathrm{im } [ b(t)e^{-i\theta } ] + \mathrm{im } [ h(t)e^{-i\theta}],\end{aligned}\ ] ] where @xmath116 is the power gain . from the generalized uncertainty principle @xmath117 where @xmath118 is the gain normalized added noise .
the instantaneous snr@xmath119 for the quadrature @xmath120 is @xmath121 where @xmath122 .
that is , by using a phase sensitive amplifier tuned to the correct phase the effective snr can be a factor of two better then a phase preserving amplifier . in a typical measurement protocol
the measurement outcome is the integration of the signal @xmath88 from the amplifier with a weighting kernel ( filter ) , @xmath123dt = \sum_j |w_j|\mathrm{re } [ e^{-i\phi_j } c_j ] \delta t,\end{aligned}\ ] ] where @xmath124 is the measurement time and @xmath125 is the kernel . under the assumption that the noise is symmetric a useful measure to quantify the measurement is the separation @xmath126 where 0 and 1 label the two states of the qubit . in ref . @xcite it was shown that the optimal kernel under the additional assumptions of gaussian and diagonal covariance matrices is found by maximizing @xmath94 and is given by @xmath127 where @xmath128 is the separation of the pointer states and @xmath129 $ ] .
the achievable fidelity @xmath28 is the value of the assignment fidelity @xmath31 assuming that the noise in our system is ideal ( symmetric and diagonal gaussian covariance matrices ) . in reality
, the noise does not satisfy these properties so we must fit it to an ideal noise model . to do this
, we implemented the above filtering method on our data and plotted the resulting distributions for the different classes ( @xmath6 and @xmath5 ) of @xmath130 in fig .
[ fig : gaussian_fits ] .
if the noise was ideal we should obtain two gaussian histograms with identical variance .
in reality , we see there are significant non - gaussian statistics and so we fit the histograms to double gaussian distributions of equal variance to obtain the means and standard deviation we would expect in the ideal noise case .
we obtain @xmath131 and @xmath132 , and a standard deviation of @xmath133 , which gives a value for @xmath94 ( defined in eq .
[ eq : sep ] ) of @xmath134 .
we can now compute @xmath28 by combining the ideal noise assumption with the definition of the assignment fidelity , @xmath2 + \mathbb{p}\left[1|0\right]\right)/2.\ ] ] here @xmath3 $ ] ( @xmath4 $ ] ) is the probability of obtaining outcome 1 " ( 0 " ) given the system was prepared in @xmath6 ( @xmath5 ) .
since the noise is ideal @xmath28 takes the form @xmath135 where @xmath94 is defined in eq .
[ eq : sep ] . from this expression
we obtain @xmath136 for our system .
we note that the separation @xmath94 in the time independent limit can be related to the signal - to - noise defined above by snr@xmath137 .
this is straightforward to show by direct substitution of the quantities in eq .
[ eq : sep ] .
@xmath94 is used here as it is a standard measure of separation in ml @xcite .
the quadratic program for the svm @xcite is given by the equation @xmath138 where @xmath139 is the expected outcome ( taken to be -1 or 1 ) .
this has a quadratic objective function with linear inequality constraints .
the soft - margin formalism adds slack variables representing the degree of misclassification to the constraints in eq .
( [ eq : svmprimal ] ) and modifies the objective function to include a mislabelling cost term . the modified quadratic program that includes a soft - margin
is given by the equation @xmath140 in the dual form of this problem , the @xmath141 variables vanish and @xmath142 is a box constraint " that bounds the lagrange multipliers .
we implemented 10-fold cross - validation on the training set and found a misclassification error of 0.0163 which agrees well with @xmath143 .
this implies large @xmath142 value is likely not needed and varying @xmath142 between 0 and 100 gave an optimal value of @xmath31 close to 1 .
k - means clustering @xcite is formulated as the following optimization problem @xmath144 here there are @xmath49 sets @xmath145 with means @xmath146 and so the goal is to partition the data into @xmath49 sets that minimize the within - set distance to the mean @xmath146 .
the k - means algorithm only has guaranteed convergence to a local minimum and finding the global optimum is an np - hard problem @xcite .
therefore initialization of the subclass means in eq .
[ eq : kmeans ] can be important to find meaningful solutions .
typically initialization is a random process that can lead to small variation in the solutions .
this can be circumvented by explicitly defining the initial means to be the average of the output subclass means over many realizations .
this helps ensure the algorithm is reproducible and more stable .
the plot of the heating subclass of @xmath9 found from the k - means algorithm with @xmath51 is given in fig .
[ fig : heating_subclass ] and the plot of the four subclasses of @xmath10 found from the k - means algorithm with @xmath55 is given in fig .
[ fig : four_subclasses ] . | high - fidelity measurements are important for the implementation of quantum information protocols .
current methods for classifying measurement trajectories in superconducting qubit systems produce fidelities systematically lower than those predicted by experimental parameters . here , we place current classification methods within the framework of machine learning ( ml ) algorithms and improve on them by investigating more sophisticated ml approaches .
we find that non - linear algorithms and clustering methods produce significantly higher assignment fidelities that help close the gap to the fidelity achievable under ideal noise conditions . clustering methods group trajectories into natural subsets within the data , which allows for the diagnosis of systematic errors .
we find large clusters in the data associated with @xmath0 processes and show these are the main source of discrepancy between our experimental and achievable fidelities . these error diagnosis techniques help provide a path forward to improve qubit measurements .
the ability to perform accurate measurements is important for maximizing the information one can extract from a physical system .
this is especially true in experimental quantum information processing since quantum systems are highly susceptible to noise effects and error rates of quantum operations and measurements must be small for fault - tolerant quantum computation to be a reality @xcite .
our goal is to provide methods for diagnosing measurement errors and increasing fidelities by using classification and clustering algorithms in machine learning ( ml ) .
while we apply our methods in a superconducting qubit measurement system , we anticipate that the generality of these techniques can be useful in a broader class of systems .
superconducting quantum bits ( qubits ) have become a promising candidate for building a fault - tolerant quantum computer due to their long coherence times @xcite , high - fidelity multi - qubit gate operations @xcite , and the abilitiy to perform single - shot measurements @xcite , in a circuit qed architecture @xcite .
remarkable progress has been made in reducing error - rates of these operations however considerable work is still needed to implement fault - tolerant quantum computation in large networks of qubits . in circuit
quantum electrodynamics ( cqed ) a superconducting anharmonic oscillator , such as a transmon @xcite , is coupled to a resonator , producing a state - dependent shift of the resonator frequency .
this allows for qubit measurements by driving the resonator and recording the output trajectory @xcite in phase ( i - q ) space . in practice
there are significant sources of random noise and systematic effects , such as @xmath0 processes where the system spontaneously decays to its ground state , that can make single - shot trajectories appear complex and difficult to distinguish .
our experimental system is a single qubit ( q4 ) in a planar lattice of four superconducting qubits @xcite .
we show current methods for assigning outcomes to measurement trajectories in this system produce assignment fidelities ( defined below in eq .
( [ eq : assfid ] ) ) that are much lower than the predicted achievable value derived under ideal noise conditions .
we utilize various ml algorithms to obtain deeper insight into the behavior of the trajectories and bring fidelities up closer to our expected values .
we find a total increase in assignment fidelity from 0.9586 using current methods to 0.9821 ( @xmath1 increase ) using non - linear ml classifiers .
the strong performance of non - linear classifiers indicates systematic effects , such as heating and @xmath0 events , could be a significant source of error in our measurements . to verify this , we use ml clustering methods to group the data into naturally occurring subsets .
we find a large cluster consisting of @xmath0 events whose size is consistent with the experimentally measured @xmath0 time .
replacing this cluster with random non-@xmath0 events gives assignment fidelities approaching 0.995 which is much closer to the achievable value of 0.9999 .
going to higher orders we find a much small cluster corresponding to heating of the ground state into the excited state . before moving on ,
let us make a few points about using ml methods for trajectory discrimination and improving measurements .
first , the methods we present here can be useful in a much broader context .
any measurement scheme that produces patterns in a geometric space can potentially benefit from more advanced ml methods . investigating
the applicability to different systems will depend on the details of each situation .
second , these methods are applicable even if we are trying to improve higher fidelity measurements than that of this paper .
the key is that these methods can be tailored according to the types of noise present .
third , ml methods have also been applied to other problems in quantum information such as phase estimation @xcite and asymptotic state estimation @xcite .
the standard metric for characterizing how well a single - qubit measurement assigns outcomes is the assignment fidelity @xmath2 + \mathbb{p}\left[1|0\right]\right)/2.\ ] ] here @xmath3 $ ] ( @xmath4 $ ] ) is the probability of obtaining outcome 0 " ( 1 " ) given the system was prepared in @xmath5 ( @xmath6 ) .
hence @xmath7 $ ] and ideally @xmath8 .
our measurement framework is the dispersive limit of cqed , where observing the resonator output voltage provides a quantum non - demolition qubit measurement . for outcome
0 " ( 1 ) the voltage leaving the cavity represents a coherent state i - q trajectory , and single - shot trajectories are obtained by amplifying the cavity signal .
the main parameters of our system are given in @xcite and complete experimental information is given in ref .
@xcite . our data
consists of 51200 single - shot trajectories ( shots ) , half initially prepared in @xmath6 and the other half in @xmath5 ( denote these classes by @xmath9 and @xmath10 ) .
the trajectories are time - ordered and the first half is used as a training set to predict the second half for classification .
the mean trajectories for each class , denoted @xmath11 and @xmath12 , as well as examples of single - shot trajectories , are given in fig .
[ fig : mean_trajs ] .
we see that each shot can be complicated but there are enough shots to ensure smooth , well - separated means .
the total measurement time @xmath13 is @xmath14s and @xmath15 $ ] is discretized into 163 time - points so trajectories are represented by vectors @xmath16 ( let @xmath17 ) where the first ( last ) 163 entries correspond to the real ( imaginary ) parts of the trajectory .
the current method of classifying trajectories @xcite is to integrate each trajectory with a filter ( kernel , weight ) function @xmath18 .
formally , if @xmath19 is a single trajectory , under the assumption that the covariance matrices @xmath20 and @xmath21 of each class are equal , gaussian , and diagonal , the optimal @xmath18 is equal to the ratio of the difference between the mean trajectories @xmath22 and @xmath23 and the variance @xmath24 .
if @xmath25 is the diagonal covariance matrix then @xmath26.\end{aligned}\ ] ] the assignment fidelity using this method is @xmath27 .
the achievable " assignment fidelity for our experiment , @xmath28 , is the fidelity we would obtain under ideal noise conditions . by
ideal noise " we mean the noise satisfies the assumptions above for the method of @xcite to hold .
the details of this calculation , along with a brief introduction to measurements and amplifiers , is contained in @xcite .
we obtain @xmath29 and so there is a large discrepancy between @xmath30 and @xmath31 that is due to a wide variety of factors such as state - preparation errors and non - gaussian / non - linear effects .
this discrepancy provides the motivation for investigating better methods for classifying trajectories . the idea behind machine learning ( ml ) classification is to obtain a classifying ( discriminating ) surface in @xmath32 under constraints such as the form of the noise . for gaussian noise
the optimal discriminator is a quadratic surface @xcite ( quadratic discriminant analysis - qda ) given by @xmath33x + x^t \left[\sigma_0^{-1}\mu_0-\sigma_1^{-1}\mu_1\right],\end{aligned}\ ] ] and the threshold is @xmath34 where @xmath35 represents the determinant .
if @xmath36 the quadratic term in eq .
( [ eq : qdaeq ] ) disappears and the surface reduces to a hyperplane @xcite ( linear discriminant analysis - lda ) , @xmath37.\end{aligned}\ ] ] comparing with eq .
( [ eq : ldadiag ] ) we see the current method of @xcite is equivalent to lda with the added assumption of diagonal covariance matrices .
qda can achieve a more accurate value of @xmath31 as it allows for @xmath38 and thus a quadratic ( non - linear ) discriminating surface .
we computed @xmath31 using the fitcdiscr " function in matlab for four different methods : ldad , lda , qdad , and qda ( d " represents diagonal covariance matrix and ldad is the method of @xcite ) .
the results are in the second column of table [ table : assfid_discriminant_methods ] .
not surprisingly , we find qdad improves upon ldad and allowing non - diagonal covariance matrices produces higher @xmath31 .
the values in table [ table : assfid_discriminant_methods ] are the sample means from 100 repetitions .
the sample variances @xmath39 are typically on the order of @xmath40 indicating stable and reproducible results .
a value of @xmath31 for qda was not attainable due to singular covariance matrices , which is a result of overfitting the data ( having more variables than required from the correlation time in the trajectories ) . to remedy this problem ,
we first perform dimensionality reduction using principal component analysis ( pca ) @xcite and find @xmath41 of the variance in the data can be accounted for in a subspace of dimension @xmath42 . loosely speaking ,
this implies correlation times of @xmath43 ns . the results with a pca pre - processing step ( using princomp " in matlab ) are in the third column of table [ table : assfid_discriminant_methods ] .
a value of qda can now be computed and as expected it provides the highest @xmath31 out of all cases considered .
c c c c c method & all time - points & pca + [ 0.5ex ] ldad & 0.9586 & 0.9557 + lda & 0.9701 & 0.9586 + qdad & 0.9627 & 0.9648 + qda & & 0.9712 + [ 1ex ] [ table : assfid_discriminant_methods ] these classification methods have assumed gaussian noise .
more robust methods are needed as we expect non - gaussian behavior .
we approach this in two ways .
the first is via the support vector machine ( svm ) @xcite , which makes no assumption on the form of the noise and can be extended to extremely general non - linear discriminating surfaces .
the second is to utilize clustering " methods in ml to naturally goup the data into clusters from which we perform _ multi - class _ classification .
we first describe the svm method .
the linear svm is a quadratic program based on maximizing the minimum margins of the data , where the margin of a data point is its distance to a separating hyperplane @xcite .
the non - linear svm is derived from the dual form of the linear svm by defining a kernel that maps the data to a higher - dimensional space .
the key is that the linear svm in the higher - dimensional space allows for non - linear discrimination in @xmath32 . due to its generality and simplicity , we chose a radial basis ( gaussian ) function kernel .
we implemented the svm using the fitcsvm " function in matlab .
the classification was repeated 100 times and the mean values with the optimal soft - margin parameter are contained in table [ table : assfid_svm_methods ] ( see @xcite for details ) .
the sample variances @xmath39 in @xmath31 are approximately @xmath44 indicating stable results .
the non - linear svm produces the highest assignment fidelity out of all methods considered thus far , indicating non - linear effects are present .
c c c c c method & all time - points & pca + [ 0.5ex ] linear svm & 0.9753 & 0.9571 + non - linear svm & 0.9821 & 0.9739 + [ 1ex ] [ table : assfid_svm_methods ] our second method for implementing a non - linear classifier combines classification and _ clustering _ algorithms .
clustering naturally groups the data into subsets and is unsupervised " since it requires no training data .
we utilize k - means clustering @xcite since it has features that suit our purposes well , however we anticipate similar results can be obtained with other standard clustering algorithms such as heirarchical methods . for an explicit formulation of the k - means clustering problem see @xcite .
we used the matlab
kmeans " function to find @xmath45 clusters in each of @xmath9 and @xmath10 .
we chose @xmath45 to take into account both variance and systematic effects .
the mean trajectories and size of the six subclasses are given in fig .
[ fig : subclass_m4 ] .
we see @xmath9 is split relatively evenly into the subclasses @xmath46 , @xmath47 , @xmath48 that capture variance in the trajectories .
we do not see a subclass of @xmath9 corresponding to heating of the ground state , however we implemented k - means for larger @xmath49 and found a heating subclass of size @xmath50 for @xmath51 ( see fig .
[ fig : heating_subclass ] ) .
@xmath10 has strikingly different properties as subclass @xmath52 is comprised of @xmath0 processes . @xmath53 and @xmath54 are similar in size and capture variance in the trajectories .
the key point is we have found _ explicit shot indices _ for @xmath0 events .
we verified that @xmath52 is comprised of @xmath0 trajectories by performing k - means with @xmath55 .
we found that the @xmath53 and @xmath54 subclasses remain virtually the same while the @xmath0 subclass @xmath52 is now split into two according to variance in these trajectories ( see fig .
[ fig : four_subclasses ] ) . from fig .
[ fig : subclass_m4 ] , @xmath56 of the @xmath5 preparations result in a @xmath0 event , which is consistent with the percentage calculated from system parameters @xcite , @xmath57 . to perform classification
, we lift the @xmath0 subclass @xmath52 to a class @xmath58 of its own , redefine @xmath59 , keep @xmath9 as before , and perform _ multi - class _ classification on @xmath9 , @xmath10 , and @xmath58 .
we implemented four multi - class algorithms in matlab ; multi - class lda , multi - class svm , totalboost " , and
rusboost " .
the latter two are examples of boosting algorithms which assemble an ensemble of weak learners ( classifiers ) in a network to create a final strong learner by iteratively re - weighting data points according to previous results @xcite .
the rusboost method @xcite is particularly useful since it is tailored to the case of one class ( here @xmath58 ) being significantly smaller than the rest .
the results are in table [ table : assfid_multiclass_methods ] .
we again see an increase in assignment fidelities over the discriminant analysis methods of table [ table : assfid_discriminant_methods ] .
not surprisingly , rusboost provides the most significant increase .
we repeated the k - means algorithm 50 times with random initializations and found it to be relatively stable ( sample variance @xmath39 of @xmath60 ) .
we repeated this using fixed initialization of the means and obtained a variance of 0 . out of all methods considered ,
non - linear svm s produce the greatest increase in @xmath31 ( 0.9586 to 0.9821 ) .
we also note all methods are relatively stable with reproducible assignment fidelities ( each method was repeated @xmath61 times ; the sample means of @xmath31 are the table values and the sample variances are @xmath62 ) .
c c c method & all time - points & pca + [ 0.5ex ] multi - lda & 0.9768 & 0.9689 + multi - svm & 0.9784 & 0.9717 + totalboost & 0.9527 & 0.9413 + rusboost & 0.9788 & 0.9723 + [ table : assfid_multiclass_methods ] while we have improved @xmath31 to 0.9821 , we are still far from @xmath63 .
it is possible much of the remaining discrepancy comes from @xmath0 events . to investigate this
we propose the simple error diagnosis test of replacing each @xmath0 event found from the k - means algorithm with a random element from @xmath64 .
this provides a measure of @xmath31 when @xmath0 is negligible .
the means of 100 samples for each method are contained in table [ table : assfid_replace ] ( variances are @xmath65 ) .
non - linear svm produces the highest value of @xmath31 however _ for all methods _
@xmath66 , which is more consistent with @xmath63 .
this confirms @xmath0 events are the significant reason for lower @xmath31 values than expected .
c c c method & all time - points & pca + [ 0.5ex ] ldad & 0.9920 & 0.9909 + lda & 0.9921 & 0.9928 + qdad & 0.9918 & 0.9908 + qda & & 0.9927 + linear svm & 0.9936 & 0.9943 + non - linear svm & 0.9945 & 0.9949 + [ table : assfid_replace ] one attempt to reduce the significance of @xmath0 is to reduce @xmath13 , however this implies the trajectories will spend less time near their steady states and assignment errors due to variance will increase.to observe this , we truncated the trajectories to different @xmath13 and calculated @xmath31 using the non - linear svm . from fig .
[ fig : assfid_vary_time ] we see @xmath67s appears close to optimal .
moreover , a much shorter measurement time of @xmath68s ( not shown in fig .
[ fig : assfid_vary_time ] ) is needed to achieve @xmath31 from lda .
this is a strong message that better classifiers can allow for shorter measurement times .
longer measurement times than the current @xmath14s decrease @xmath31 due to an increase in @xmath0 events .
to conclude , we have utilized ml to understand and improve the readout in a superconducting system .
we find more sophisticated classification algorithms can potentially allow for shorter measurement times and increase assignment fidelities .
non - linear svm s provided the largest increase in assignment fidelity from 0.9586 to 0.9821 ( @xmath1 ) .
clustering helped diagnose the prevalence of systematic effects by finding clusters in the data corresponding to single - shot identification of heating and @xmath0 effects .
we verified @xmath0 events are a significant source of error as the assignment fidelity increases from 0.9821 to 0.9945 when the @xmath0 cluster is replaced with typical trajectories .
this is more consistent with our achievable fidelity and the remaining discrepancy can be due to effects such as heating and state - preparation errors . moving forward
, we expect these methods will help provide insight for improving readout , especially when non - linear and non - gaussian effects are present .
we acknowledge support from aro under contract w911nf-14 - 1 - 0124 and iarpa under contract w911nf-10 - 1 - 0324 .
we acknowledge helpful discussions with oliver dial , stefan filipp , blake johnson , jim rozen , colm ryan , marcus silva , and matthias steffen .
10 p. shor , in _ proceedings of the 37th annual symposium on foundations of computer science ( focs ) _
( ieee press , burlington , vt , 1996 ) .
h. paik et al .
, phys .
rev .
lett . * 107 * , 240501 ( 2011 ) .
j. b. chang et al . , applied physics letters * 103 * , 012602 ( 2013 ) .
r. barends et al . , physical review letters * 111 * , ( 2013 ) .
r. barends et al . ,
nature * 508 * , 500 ( 2014 ) .
f. mallet et al . , nat phys * 5 * , 791 ( 2009 ) .
n. bergeal et al .
, nature * 465 * , 64 ( 2010 ) .
j. e. johnson et al . , phys .
rev .
lett . * 109 * , 050506 ( 2012 ) . d. rist et al .
, phys .
rev .
lett .
* 109 * , 050507 ( 2012 ) .
a. blais et al .
, phys .
rev .
a * 69 * , 062320 ( 2004 ) .
j. koch et al .
, phys .
rev .
a * 76 * , 042319 ( 2007 ) . j. gambetta et al .
, phys .
rev .
a * 77 * , 012112 ( 2008 ) .
a. d. crcoles et al . , arxiv:1410.6419 ( 2014 ) .
a. hentschel and b. c. sanders , phys .
rev .
lett . * 104 * , 063603 ( 2010 ) .
m. gu and w. kotowski , new journal of physics * 12 * , 123032 ( 2010 ) .
see supplemental material for further details .
c. a. ryan et al . , arxiv:1310.6448 ( 2013 ) .
t. m. cover , electronic computers , ieee transactions on * ec-14 * , 326 ( 1965 ) .
r. a. fisher , annals of eugenics * 7 * , 179 ( 1936 ) .
k. pearson , philosophical magazine series 6 * 2 * , 559 ( 1901 ) . c. cortes and v. vapnik , machine learning * 20 * , 273 ( 1995 ) . b. e. boser , i. m. guyon , and v. n. vapnik , in _ proceedings of the fifth annual workshop on computational learning theory _
( acm , new york , ny , usa , 1992 ) , colt 92 , pp . 144152 .
s. lloyd , information theory , ieee transactions on * 28 * , 129 ( 1982 ) .
c. bishop , _ pattern recognition and machine learning _ ( springer , 2007 ) . c. seiffert et al . , systems , man and cybernetics , part a : systems and humans , ieee transactions on * 40 * , 185 ( 2010 ) . |
ring galaxies are dramatic examples of galaxy transformation caused by a remarkably simple interaction .
observations and numerical models ( @xcite ) argue persuasively that these objects are formed by the near central passage of a companion through a spiral along the rotation axis .
the brief additional gravitational force induces epicyclic motions throughout the disk , which act to form radially propagating orbit - crowded rings of gas and stars .
the concentration of the ism into the expanding ring ( at the disk s expense ) is nearly total and can last for @xmath0 myrs .
it is this radical rearrangement of the spiral s ism that is responsible for their interesting star forming properties .
star formation rates ( @xmath1 ) in ring galaxies are typically @xmath2 m@xmath3 yr@xmath4 , i.e. , somewhat enhanced over large spirals but far below the @xmath1s inferred in lirgs ( cf .
@xcite ; @xcite ; @xcite ; @xcite ) .
however , the _ distribution _ of star formation is unique , being completely restricted to the expanding rings while simultaneously extinguished over the interior disk .
both effects are evident in the lindsay - shapley ring ( _ l - s _ , hereafter ) and cartwheel , shown in figure 1 .
a weak nuclear source is responsible for @xmath5 of the star formation in both .
the star forming rings are narrow , with slices showing very sharp radial cutoffs in h@xmath6 emission .
this implies that ob stars remain in the rings for their main sequence lifetimes , which constrains the stellar velocity dispersion of the rings : @xmath7 km s@xmath4 .
the weak line emission that is sometimes found in ring galaxy disks is _ post_-starburst in origin , i.e. , arising from aging hii complexes powered by a - stars .
_ l - s _ and cartwheel have been mapped in hi ( figure 1 ) . in such large systems
@xmath8 of the atomic ism is concentrated in the rings , resulting in high hi surface densities : @xmath9 m@xmath3 pc@xmath10 . at the same time , their interiors are very gas poor , with @xmath11 m@xmath3 pc@xmath10 .
hi line - widths are typically narrow ( @xmath12 km s@xmath4 ) .
kinematic analysis of the hi yields the ring s expansion speed ( @xmath13 ) , and thus an estimate of the ring galaxy s age ( @xmath14 ) . from their measured radii and @xmath13 ( @xmath15 and @xmath16 km s@xmath4 , respectively ) , the rings in cartwheel and _ l - s _
are @xmath17 and @xmath18 myrs old ( @xcite;@xcite ) .
the @xmath1 in ring galaxies correlates with their peak @xmath19 , which explains why young systems ( e.g. , ngc 2793 with @xmath20 myr ) have such low @xmath1 : they are still organizing their ism into a dense ring
. stars form in cold molecular gas , so hi data can only tell part of the story .
ring galaxies are not very luminous in the rotational transitions of @xmath21co , a fact often attributed to reduced metallicities from snow - plowing outer disk gas into the ring ( cf .
@xcite ) . however _
l - s _ s ring possesses @xmath22solar abundances ( @xcite ) , which together with its large angular size , made it an ideal target for the @xmath23 ( @xcite ) .
we observed 16 positions in _ l - s _ in @xmath21co(j=1 - 0 ) and @xmath21co(j=2 - 1 ) transitions : 14 on the ring and one each centered on the nucleus and enclosed disk .
figure 2 shows co detections in 9/14 ring positions , defining two molecular arcs in the ring s north and southwest . the latter coincides with the galaxy s peak @xmath19 and @xmath24 . _ l - s _ s ring is dominated by _
atomic _ rather than molecular gas . for a galactic i@xmath25-n@xmath26
, we find m@xmath26/m@xmath27 .
astonishingly , a typical dwarf galaxy has nearly as much h@xmath28 as _ l - s _ s ring ( @xcite ) .
the molecular gas fraction ( @xmath29 ) varies considerably around the ring , and is lowest in the ring s southwest quadrant ( @xmath30 at p9-p11 ) , where both @xmath19 and @xmath24 peak .
the @xmath21co and hi line profiles in _
l - s _ s ring can be extremely broad ( @xmath31 km s@xmath4 ) , with multiple velocity components or broad tails evident .
it is not clear if this represents out - of - plane gas motions or caustics , though preliminary numerical models suggest the latter ( j. wallin , private communication ) .
the onset of robust star formation in gas disks can be described in terms of local gravitational stability parameters , e.g. , @xmath32 , where @xmath33 is the gas velocity dispersion and @xmath34 is the disk s epicyclic frequency ( @xcite ) .
regions of the disk where @xmath35 are jeans unstable and prone to collapse , leading to the formation of molecular cloud complexes and eventually stars .
the `` bead on a string '' morphology evident in h@xmath6 ( figure 1 ) suggests that the rings are gravitationally unstable .
are they ?
the cartwheel s ring is , even ignoring its ( unknown ) molecular component .
however , @xmath36 essentially everywhere in _ l - s_a ring ( figure 3 ) , due to the large @xmath37 .
this ignores , however , the stellar component s contribution to @xmath38 . using irac 4.5 @xmath39 m data
we find that the stellar mass surface density ( @xmath40 ) everywhere exceeds that of gas ( i.e. , @xmath41 ) .
a gravitational stability parameter combining stars and gas can be written @xmath42 ( @xcite ) , where the ring s radial h@xmath6 profile constrains @xmath43 km s@xmath4 .
figure 3 shows that when the stellar component is included , @xmath44 and _ l - s _ s ring is everywhere jeans unstable .
conversely , the interior disks in both satisfy @xmath45 , i.e. , these regions are stable against the growth of gravitational instabilities and star formation is effectively quenched .
a strong correlation exists between @xmath46 ( @xmath47 ) and the surface density of cold gas in galaxies , i.e. , the `` schmidt law '' , written @xmath48 . in m 51
for example , h@xmath28 and @xmath46 obey @xmath49 ( figure 4 ) .
hi is uncorrelated with @xmath47 , implying that it is a photo - dissociation product and not directly involved in the star formation process .
we show star formation laws derived for the cartwheel and _ l - s _ in the same figure . both are peculiar .
unlike m 51 , atomic gas in the cartwheel correlates with @xmath47 in most of the ring , though with a small @xmath50 .
the exponent becomes negative ( i.e. , _
anti_-correlated ) where @xmath47 peaks .
would this peculiarity disappear if the molecular component were available ?
l - s _ s ring is any guide : atomic gas obeys m 51 s _ molecular _ schmidt law , but its cold molecular ism is _ uncorrelated _ with @xmath47 , which is completely the opposite from m 51 . how can h@xmath28 be so apparently disconnected from star formation ?
star formation efficiency ( @xmath51 ) is the yield of massive stars per unit h@xmath28 mass .
@xcite find nearly constant @xmath51 ( @xmath52 ) ) from s0 to scd ( @xmath53 l@xmath3/m@xmath3 ) .
later types show higher @xmath51 , peaking at @xmath54 l@xmath3/m@xmath3 for irr . detecting molecular gas in _
l - s _ s ring allows the first estimate of @xmath51 in a ring galaxy .
we find @xmath55 l@xmath3/m@xmath3 , i.e. , similar to an irr and an order of magnitude higher than the ( presumably @xmath56sa ) progenitor .
this result depends , of course , on our ability to reliably measure h@xmath28 in the ring using @xmath21co emission .
the rings are gas rich but seemingly h@xmath28 poor .
since hi rapidly converts into h@xmath28 , the low @xmath57 can not signal the consumption of the molecular gas reservoir . nor can metallicity effects by themselves be responsible , at least in _ l - s_. the gas phase pressure ( @xmath58 ) might be a factor , as it directly affects the hi to h@xmath59 conversion rate .
@xcite finds @xmath60 , where @xmath61 is the ambient uv - field . for _ l - s _ s ring we estimated @xmath61 with spitzer and galex images , and @xmath62 , with the ring s stellar surface mass density derived with irac 4.5 @xmath39 m data .
we find very high gas phase pressures , with @xmath63 , leading us to expect @xmath64 everywhere . _
l - s _ s ring should be dominated by molecular gas . why is nt it ?
we used photo - dissociation models in @xcite to estimate average gas volume densities ( @xmath65 ) in _ l - s _ s ring given its @xmath19 and uv - field . in the northern half of the ring , @xmath66 @xmath67 , implying an ism dominated by the _ cold neutral medium _ ( _ cnm _ , @xmath68 k ) , i.e. , the precursor of cold molecular clouds . in the southwest , where @xmath47 and @xmath19 are both much higher , the models give @xmath69 @xmath67 , which taken at face value , points to an ism dominated by the
_ warm neutral medium _
( _ wnm _ , @xmath70 k ) .
how can you form stars out of this ?
we believe the answer lies in fundamental differences in the environments of rings and spiral arms .
consider that molecular clouds spend @xmath71 myrs in the arms of grand design spirals like m 51 , whereas the ism is confined in rings , equally as dense and actively forming stars , for @xmath72 myr .
while molecular cloud growth is enhanced in the high @xmath73 rings , the destructive effects of sne and ob stars are also amplified .
a dominant _ wnm _ might be expected as the molecular clouds become fragmented and `` over - cooked '' by shocks and sustained uv - fields .
co might in this case retreat to the inner - most cloud cores resulting in weak @xmath74 and underestimates of @xmath75 .
this might explain the peculiar schmidt laws and enhanced @xmath51 .
at the same time , higher cloud collision rates might favor the formation of unusually large molecular cloud complexes and more efficient star formation .
more work remains though results from cartwheel and _ l - s _ are intriguing .
the distribution and gravitational stability of a ring galaxy s ism changes dramatically as it evolves .
further , the long confinement in the dense star forming ring is expected result in a fragmented and largely atomic ism , which may explain the peculiar star formation laws , efficiencies , and @xmath57 observed in the two largest and most evolved ring galaxies , cartwheel and _ l - s_. future progress will require sensitive and high resolution assays of the molecular ism in these and other ring galaxies , which will be possible with alma and carma . | compared with ordinary spirals , the ism in ring galaxies experiences markedly different physical conditions and evolution . as a result , ring galaxies provide interesting perspectives on the triggering / quenching of large scale star formation and the destructive effects of massive stars on molecular cloud complexes .
we use high resolution radio , sub - mm , infrared , and optical data to investigate the role of gravitational stability in star formation regulation , factors influencing the ism s molecular fraction , and evidence of peculiar star formation laws and efficiencies in two highly evolved ring galaxies : cartwheel and the lindsay - shapley ring . |
in @xcite , j.w . cannon proved certain geometrical properties of the cayley graph of cocompact discrete hyperbolic groups .
two years later , in 1986 , w. thurston noticed that some of these properties could be reformulated in terms of finite state automata (; this abbreviation will be used for both the singular and plural ) .
in particular , the goedesic paths in the cayley graph that start at the origin form a regular set or , equivalently , they form the language of an . furthermore ,
any pair of such geodesic paths that end at the same or neighbouring vertices lie within a bounded distance of each other .
it can be deduced that such geodesic pairs also form the language of an .
this led thurston to formulate the following general definition .
[ aut ] let @xmath0 be a group with finite generating set @xmath1 , let @xmath2 , and let @xmath3 , where @xmath4 .
then @xmath0 is said to be _ automatic _ ( with respect to @xmath1 ) , if there exist fsa @xmath5 and @xmath6 for each @xmath7 , such that + ( i)@xmath5 has input alphabet @xmath8 , and accepts at least one word in @xmath9 mapping onto each element of @xmath0 .
+ ( ii)each @xmath6 has input alphabet @xmath10 , it accepts only padded pairs , and it accepts the padded pair @xmath11 for @xmath12 if and only if @xmath13 and @xmath14 . here @xmath9 as usual denotes the set of words in @xmath8 . for @xmath15
, @xmath16 denotes the element of @xmath0 onto which @xmath17 maps ; for @xmath18 , we also use @xmath19 to mean that @xmath20 map onto the same element of @xmath0 .
the extra symbol @xmath21 maps onto the identity element of @xmath0 . for @xmath12 , the associated _ padded pair _
@xmath22 is obtained by adjoining symbols @xmath21 to the end of the shorter of @xmath17 and @xmath23 to make them have equal length . the language of the @xmath5 is denoted by @xmath24 . for general properties of finite state automata , the user is referred to any textbook on automata or formal language theory , such as @xcite . in the definition
, @xmath5 is called the _ word - acceptor _ and the @xmath6 the _ multiplier _ automata . the complete collection @xmath25 is known as an _
automatic structure _ for @xmath0 .
note that the multiplier @xmath26 recognises equality in @xmath0 between words in @xmath24 . from a given automatic structure ,
we can always use @xmath26 to construct another one such that @xmath5 accepts a unique word mapping onto each element of @xmath0 ; we simply choose the lexicographically least amongst the shortest words that map onto each element as the ` normal form ' representative of that element .
we shall call such a @xmath5 a word - acceptor with uniqueness .
the best general reference for the theory of automatic groups is the multi - author book @xcite . in particular , it turns out that the automaticity of @xmath0 is independent of the choice of generating set @xmath1 .
this immediately suggests that the definition is a sensible one , because it means that automaticity is an algebraic property of the group , rather than just a geometrical property of its cayley graph .
all finite groups are easily seen to be automatic ; in fact the class of automatic groups is invariant under finite variations , such as sub- and super - groups of finite index .
it is also closed under direct and free products , and includes , for example , all word - hyperbolic groups , braid groups , coxeter groups and artin groups of finite and of ` large ' type .
all automatic groups have finite presentations .
some of the most important and useful applications of this theory only involve an explicit knowledge of a word acceptor with uniqueness , particularly in the frequently occurring case when the accepted words are all geodesics in the cayley graph .
from such a word - acceptor , one can quickly enumerate unique representatives of all words up to a given length .
this can serve as an invaluable time - saving device in certain computer graphics applications , such as drawing tessellations of hyperbolic space on which these groups act freely .
one can also use @xmath5 to compute the growth function for the group ( see @xcite ) .
another important application of automatic structures for groups @xmath0 is their use for the efficient ( quadratic time ) solution of the word problem in @xmath0 .
more precisely , the multiplier automata can be used to reduce an arbitrary word in @xmath9 in quadratic time to the @xmath0equivalent word in @xmath24 . with these applications in mind ,
a collection of programs was written at warwick in the late 1980 s for computing automatic structures .
these programs take a finite presentation of the group @xmath0 as input .
currently , they only work for so - called _ shortlex _ structures , which are those in which @xmath24 consists of the lexicographically least amongst the shortest words that map onto each group element .
( so @xmath5 depends upon the order of @xmath8 as well as on @xmath8 itself . )
many , but not all , of the known classes of automatic groups are known to possess shortlex structures .
the programs are described in some detail in @xcite and @xcite , and in a much more general setting in @xcite .
the latest version is part of a package called and is available by anonymous ` ftp ` from ` ftp.maths.warwick.ac.uk ` in the directory ` people / dfh / kbmag2 ` . from an algorithmic point of view , there is a close connection between automatic groups and rewriting systems for groups , and the programs used make use of the knuth bendix completion process in groups .
however , typically , this process alone would not terminate and in fact automatic groups normally have infinite regular rather than finite complete rewriting systems . when the automatic structure is successfully computed it is , in some sense , enabling this infinite regular system to be used to solve the word problem in a manner that is typically at least as efficient as could be done with a finite rewriting system .
the idea of trying to use infinite regular rewriting systems for this purpose was first proposed by gilman in @xcite . given a word - acceptor automaton for a group , it turns out that the existence and properties of the multiplier automata are equivalent to the so - called ( synchronous ) fellow - traveller property , which was one of the geometrical properties of hyperbolic groups observed originally by j. w. cannon , and is defined as follows . for a word @xmath15
we denote the length of @xmath17 by @xmath27 and , for @xmath28 , @xmath29 ( or more precisely @xmath30 ) denotes the length of the shortest word @xmath15 with @xmath31 . for @xmath32
, @xmath33 denotes the prefix of @xmath17 of length @xmath34 when @xmath35 , and @xmath36 for @xmath37 .
the fellow - traveller property asserts that there exists a constant @xmath38 such that , for all @xmath39 and @xmath40 such that @xmath14 , and all @xmath32 , we have @xmath41 . in other words ,
two travellers proceeding at the same speed along the words @xmath17 and @xmath23 from the base point in the cayley graph of @xmath0 would always remain a bounded distance away from each other .
the fellow - traveller property enables the multiplier automata @xmath6 to be defined in a uniform manner ( see definition 2.3.3 of @xcite ) .
their state set is the set of triples @xmath42 , where @xmath43 are states of @xmath5 , and @xmath28 with @xmath44 .
the start state is @xmath45 , where @xmath46 is the start state of @xmath5 . for @xmath47 ,
there is a transition from @xmath42 to @xmath48 with label @xmath49 if and only if there are transitions @xmath50 and @xmath51 in @xmath5 with labels @xmath52 and @xmath53 , respectively , and if @xmath54 . the state @xmath42 is a success state of @xmath6 if and only if @xmath55 and @xmath56 are success states of @xmath5 , and @xmath57 .
thus the @xmath6 differ only in their accept states .
( we have omitted a technicality from this definition . to deal with the padding symbol , we have to add an extra state to @xmath5 which is reached when @xmath5 is in an accept state and the padding symbol is read . )
it is clear that the @xmath6 behave precisely according to condition ( ii ) of definition [ aut ] .
this method is used to construct the @xmath6 in the programs mentioned above .
note also that it follows from the fellow - traveller property that if @xmath58 is any fixed element of @xmath0 and @xmath13 with @xmath59 , then @xmath17 and @xmath23 fellow - travel with constant at most @xmath60 .
finally , we must mention the weaker concept of an asynchronously automatic group , because it will arise in the next section .
the definition is the same as before , except that the multiplier automata are allowed to read their two input strings at different rates .
more precisely , rather than reading one symbol from each of the two input words at each transition , they read a symbol from one of the two words only , where the choice of which word to read is a function of the state of @xmath6 .
of course , when the end of one of the words is reached , the other word must be selected .
see chapter 7 of @xcite for the formal definition .
again there is a corresponding fellow - traveller property , in which the imaginary travellers are allowed to move at different speeds .
see @xcite or section 7 , part ii of @xcite for details .
the word problem is still solvable for asynchronously automatic groups , but it is unknown whether this can be done in polynomial time .
there are examples known , such as the baumslag solitar groups @xmath61 with @xmath62 , which are asynchronously automatic but not automatic .
there is a more detailed treatment , with references to the literature , of the synchronous and asynchronous fellow - traveller properties in groups in the article @xcite in these proceedings .
let @xmath63 be the language of the word - acceptor in an automatic structure of a group @xmath0 .
a subgroup @xmath64 of @xmath0 is called @xmath65rational if @xmath66 is a regular language ( ie the language of an ) .
such subgroups were studied in @xcite , where it is proved that @xmath65rational is equivalent to @xmath65quasiconvex .
this means that any prefix of a word in @xmath66 lies within a bounded distance of @xmath64 in the cayley graph of @xmath0 .
such subgroups are always finitely generated .
an algorithm for constructing an @xmath67 with language @xmath66 , which takes as input an automatic structure for @xmath0 and a set of generators for an @xmath65rational subgroup @xmath64 of @xmath0 , is described in @xcite .
a practical and efficient version is described in @xcite , and an implementation is available in . the @xmath67 can be used together with the automatic structure to determine whether a given word in @xmath9 lies in @xmath64 ; that is , to solve the generalised word problem for @xmath64 in @xmath0 .
first use the the multiplier automata to reduce the word to one in @xmath65 , and then use @xmath67 to test whether it lies in @xmath64 . given @xmath67 and @xmath68 for two subgroups @xmath64 and @xmath69 of @xmath0 ,
it is easy to intersect their languages to obtain a @xmath70 for their intersection , which can then be used to construct a finite generating set for @xmath71 .
it is possible to generalise the concept of an automatic group from a notion about the elements of the group to one about the cosets of a given subgroup @xmath64 of @xmath0 .
this has been carried out by two doctoral students of the author ( see @xcite and @xcite ) .
the definition is as follows .
let @xmath0 be a group with finite generating set @xmath1 , let @xmath2 , @xmath3 , and let @xmath64 be a subgroup of @xmath0 .
then @xmath0 is said to be _ coset automatic _ with respect to @xmath64 , if there exist fsa @xmath5 , and @xmath6 for each @xmath7 , such that : + ( i)@xmath5 has input alphabet @xmath8 , and accepts at least one word in each right coset of @xmath64 in @xmath0 ; + ( ii)each @xmath6 has input alphabet @xmath10 , it accepts only padded pairs , and it accepts the padded pair @xmath11 for @xmath12 if and only if @xmath13 and @xmath72 . here
@xmath5 is called the _ coset word - acceptor _ and the @xmath6 the _ coset multiplier _ automata . the complete collection @xmath25 is known as an _
automatic coset system _ for the pair @xmath73 .
again the existence of such a system turns out to be independent of the generating set @xmath1 of @xmath0 , and we can , if we wish , always find a new system in which @xmath5 accepts a unique word in each right coset .
it is proved in @xcite that if @xmath65 is the language of the shortlex automatic structure of a word - hyperbolic group @xmath0 ( or even the set of all geodesics in the cayley graph of @xmath0 ) , and if the subgroup @xmath64 is @xmath65quasiconvex , then @xmath0 is coset automatic with respect to @xmath64 . in @xcite the converse
is proved for word - hyperbolic groups , although we shall see from the example in the next section that the converse does not hold in general .
an interesting application to the drawing of limit sets of kleinian groups is described in @xcite . as in the graphical applications of ordinary automatic structures , this involves only the use of @xmath5 to enumerate unique shortest words in each coset .
an algorithm for computing automatic coset systems in the shortlex case was first described in @xcite , and was implemented by him as a standalone program .
it has the disadvantage that it is not usually possible to prove conclusively that the system computed is correct .
a different approach is described in @xcite .
this does enable the output to be proved correct , but it requires an additional hypothesis , to be described below , for it to work at all .
it has the further advantage that it has an optional extension to compute a finite presentation for the subgroup @xmath64 of @xmath0 after the automatic coset system has been found .
this second algorithm , together with the subgroup presentation facility , has been implemented and is available in . the theory , implementation details and performance statistics can also be found in @xcite .
these algorithms provide an alternative method to that described in the previous subsection for solving the generalised word problem for @xmath64 in @xmath0 .
the given word in @xmath15 is reduced ( in quadratic time , using the coset multiplier automata ) to the unique word @xmath74 in the language of the coset word - acceptor for which @xmath75 . then @xmath76 if and only if @xmath74 is the empty word .
the two methods of solving the generalised word problem are to some extent complementary to each other , since there can exist @xmath65quasiconvex subgroups that are not coset automatic and vice versa , although the two concepts are equivalent in word - hyperbolic groups .
the additional hypothesis required for the algorithm developed by hurt is the following generalisation of the fellow - traveller condition .
let @xmath25 be the shortlex automatic coset system for @xmath73 that we are trying to compute . then ,
if @xmath77 for some @xmath40 , there exists @xmath78 such that @xmath79 .
the hypothesis is that there exists a constant @xmath80 such that for all such @xmath81 and @xmath82 , and all @xmath32 , we have @xmath83 . in particular , taking @xmath84 , we get @xmath85 , and so in all such equations , only a finite number of elements @xmath82 occur .
one step in the algorithm is to define the states of the @xmath6 as triples @xmath42 , as in the automatic group case , but now the initial states are @xmath86 , where @xmath46 is the initial state of @xmath5 , and @xmath82 is one of the elements of @xmath64 occurring in the above equations .
so the @xmath6 are in fact constructed initially as non - deterministic automata with multiple initial states , if the hypothesis holds , then we shall say that @xmath0 is _ strongly coset automatic _ with respect to @xmath64 , and call @xmath25 a _ strong automatic coset system _ for @xmath73 .
it is proved in @xcite that word - hyperbolic groups are always strongly coset automatic with respect to their quasiconvex subgroups .
it is easy to construct examples in which the hypothesis does not hold , by choosing @xmath64 to be normal in @xmath0 , in which case @xmath0 coset automatic with respect to @xmath64 is equivalent to @xmath87 automatic , but we do not know of any example in which @xmath88 .
for the application to be described in this section , we need to strengthen the hypothesis defined at the end of the preceding section for strong automatic coset systems .
let @xmath25 be a strong automatic coset system for @xmath73 with respect to the generating set @xmath1 of @xmath0 .
let @xmath89 be a finite set of generators of @xmath64 , and let @xmath90 .
then @xmath89 is said to be efficient with respect to @xmath25 if , for any @xmath13 and any @xmath91 such that @xmath92 , we have either @xmath93 or @xmath94 . we are not currently aware of any particular situations under which an efficient generating set could be shown to exist ; it would be interesting to investigate this question . in specific examples of automatic coset systems that we have calculated with the programs ,
it is often possible to observe directly from the calculation that a particular @xmath89 is efficient .
the concept is useful to us here , because it enables us to prove the following result about hnn extensions , which can then be applied to a specific example .
note that a rather different condition under which an hnn extension of an automatic group is asynchronously automatic has been proved by shapiro in @xcite , and results of a similar nature for amalgamated free products are proved in @xcite .
let @xmath25 be a strong automatic coset system for @xmath73 , let @xmath95 be a finite presentation of @xmath0 , and suppose that @xmath64 has the efficient generating set @xmath89 .
suppose also that @xmath64 is automatic , and let @xmath96 be an automorphism of @xmath64 such that @xmath97 .
let @xmath99 be a right transversal for @xmath64 in @xmath0 .
then by the normal form theorem for hnn extensions ( see , for example , theorem 2.1 ( ii ) , page 182 of @xcite ) , each element of @xmath100 has a unique expression of the form @xmath101 where @xmath78 , @xmath102 , @xmath103 , @xmath104 for @xmath105 and @xmath106 for @xmath107 .
we use this normal form in the natural manner to construct a regular language @xmath108 for @xmath69 on the alphabet @xmath109 where , as before , @xmath2 and @xmath90 .
we are assuming that @xmath64 is automatic , so we can use the language @xmath110 of the word - acceptor from an associated automatic structure with alphabet @xmath111 to obtain a word @xmath112 for the element @xmath78 in the normal form .
for @xmath99 we choose the image in @xmath0 of @xmath24 , and to represent @xmath113 , we choose the unique word @xmath114 with @xmath115 .
this clearly yields a regular language @xmath108 mapping bijectively onto @xmath69 .
we now have to show how to construct the asynchronous multiplier automata @xmath116 for @xmath117 .
since this is fairly routine , we describe the construction in outline only .
suppose that @xmath118 and @xmath119 , and let the hnn normal form of @xmath120 be @xmath121 as above .
if @xmath122 or @xmath123 , then the hnn normal form for @xmath124 in @xmath69 is just @xmath125 , and it is easy to construct @xmath116 .
so suppose @xmath126 .
we shall suppose that @xmath127 and omit the details of the case @xmath128 , which are similar .
there exist words @xmath129 and @xmath130 such that @xmath131 .
let @xmath132 .
then , from the assumptions that the generating set @xmath89 of @xmath64 is efficient and that @xmath97 , it follows that the hnn normal form in @xmath69 for @xmath124 is @xmath133 where there are elements @xmath134 , all having @xmath111length at most @xmath38 , such that @xmath135 , @xmath136 for @xmath137 , @xmath138 for @xmath139 , and @xmath140
. thus we have @xmath141 and @xmath142 , where @xmath143 map onto @xmath144 , and @xmath145 map onto @xmath146 for @xmath137 .
the multiplier @xmath116 proceeds by reading the words @xmath147 and @xmath148 in parallel at the same rate ,
then the @xmath149 together , then @xmath150 and @xmath151 together , and so on . if either of @xmath147 or @xmath148 is longer than the other , then it will wait at the end of the shorter one until the longer word has been read , and similarly for @xmath113 and @xmath152 .
( this explains why @xmath116 needs to be asynchronous .
although @xmath153 and @xmath154 are all bounded , there is no bound on @xmath155 , and so one of the two tapes of the input of @xmath116 may conceivably get indefinitely ahead of the other ; indeed , we have verified that this really can happen in the example below . )
of course , if either of the two words input to @xmath116 is not in @xmath108 , or if they do not both have the same pattern with respect to the occurrences of @xmath156 , then they are rejected .
otherwise , if after @xmath34 transitions , @xmath116 has read @xmath157 symbols from @xmath158 and @xmath159 from @xmath160 , then the element @xmath161 of @xmath69 is remembered as a function of the state of @xmath116 . as in the synchronous case
, it is sufficient to show that @xmath162 is bounded . 1
. @xmath164 and @xmath165 are prefixes of @xmath147 and @xmath166 , where @xmath167 is bounded .
then the the boundedness of @xmath162 from the automaticity of @xmath64 , and the fact that @xmath168 with @xmath169 .
2 . @xmath170 for some @xmath171 and some prefix @xmath172 of @xmath173 , and @xmath174 , where @xmath175 is bounded .
then @xmath176 , and its boundedness follows from the assumptions that @xmath177 and that @xmath25 is a strong automatic coset system for @xmath73 .
3 . @xmath178 for some @xmath171 and some @xmath179 , and @xmath180 , where @xmath181
. then @xmath182 , and its boundedness follows from @xmath183 and the assumption that @xmath97 .
4 . @xmath184 and @xmath185 .
then @xmath186 which is clearly bounded . as an application
, we shall use this theorem together with the results of some machine computations that were done with , to prove that the group defined by the presentation @xmath187 is asynchronously automatic .
this group , which we shall denote by @xmath69 , was originally proposed by mark sapir as a possible building block in his attempts to construct groups with given dehn functions .
however , he later found a different approach to his problem , and so the example is no longer relevant from that viewpoint .
he had hoped that it could be proven automatic , but the methods we have been discussing in this paper only appear to be sufficient to prove it asynchronously automatic .
the computer programs could make no progress with the presentation as given above , but matters improved after manipulating it a little .
eliminating @xmath188 , we get @xmath189 now , putting @xmath190 and @xmath191 , and eliminating @xmath192 and @xmath193 , we get @xmath194 finally , using @xmath195 to simplify the second relation , we get @xmath196 this is now visibly an hnn extension of the group @xmath197 with respect to the subgroup @xmath198 , where @xmath64 is centralised by the new generator @xmath156 .
( in fact @xmath0 is itself an hnn extension with extra generator @xmath155 , but we shall not make use of that fact . ) running the automatic coset system program from on the subgroup @xmath64 of @xmath0 verifies that @xmath0 is strongly coset automatic with respect to @xmath64 . ( the coset word acceptor has 302 states , and the coset multiplers about 1400 states . )
the presentation of @xmath64 computed by the program proves that @xmath64 is free of rank 2 , and so it is certainly automatic .
the programs can also be used to verify that the set @xmath199 is an efficient generating set for @xmath64 .
( briefly , this is done by constructing the multiple initial state multiplier automata for the elements @xmath200 and @xmath201 .
the elements of @xmath64 corresponding to the initial states of these automata can then be inspected from the output , and it turns out that these are just the identity and elements of @xmath90 . ) we can now deduce from the theorem that sapir s group @xmath69 is asynchronously automatic . as a final remark about this example , it turns out ( again using calculations carried out by ) that the subgroup @xmath64 is not @xmath65quasiconvex , where @xmath65 is the language of the word - acceptor of the shortlex automatic structure of @xmath0 . the element @xmath202 of @xmath65 lies in @xmath64 for all @xmath203 , but the coset representative of @xmath204 in the language of the coset word acceptor is @xmath205 . , _ the warwick automatic groups software _ , from : `` geometrical and computational perspectives on infinite groups '' , gilbert baumslag et al ( editors ) , dimacs series in discrete mathematics and theoretical computer science 25 ( 1995 ) 6982 , _ detecting quasiconvexity : algorithmic aspects _ , from : `` geometric and computational perspectives on infinite groups '' , gilbert baumslag et al ( editors ) , dimacs series in discrete mathematics and theoretical computer science 25 ( 1995 ) 9199 | the history , definition and principal properties of automatic groups and their generalisations to subgroups and cosets are reviewed briefly , mainly from a computational perspective .
a result about the asynchronous automaticity of an hnn extension is then proved and applied to an example that was proposed by mark sapir .
the concept of an automatic group was introduced in 1986 by thurston , motivated by some results of jim cannon on hyperbolic groups .
much of the basic theory of this important class of groups was developed by david epstein during the following few years . in the first section of this paper ,
we review briefly the history , definition and properties of automatic groups and their generalisation to subgroups and cosets , mainly from a perspective of carrying out efficient computations within such groups and their subgroups . in the second section ,
we prove a result about the ( asynchronous ) automaticity of an hnn extension , and use it , together with the results of some machine computations , to prove that a particular group , defined by mark sapir , is asynchronously automatic . |
forward simulations of population genetics , track either the genotype of every individual in the population , or the number of individuals that carry a particular genotype .
the former has been implemented in a number of very flexible simulation packages @xcite . in large populations with a moderate number of loci ,
storing the abundance of all possible @xmath4 genotypes is often faster .
simulating such large populations with a small number of loci is for example essential when studying the evolution of drug resistance in viral or bacterial pathogens @xcite .
individual - based population genetic simulations are quite straightforward and usually employ a discrete generation scheme in which processes such as mutation , selection , and migration are applied at every generation to every individual .
individuals are then paired up via a mating scheme and recombinant offspring is produced .
existing toolboxes often emphasize biological realism and allow the user to specify complex life cycles , see e.g. @xcite .
our emphasis here is on efficient simulation of large populations . instead of tracking individuals , we keep track of the distribution @xmath5 of gametes across all possible @xmath4 genotypes , denoted by @xmath6 where @xmath7 .
this genotype distribution changes due to mutation , selection and recombination .
the former two are again straightforward and require at most @xmath8 operations ( each genotype can mutate at any one of the @xmath1 loci ) . in our implementation , selection acts on haploid gametes , precluding dominance effects .
recombination , however , is a computationally expensive operation since it involves pairs of parents ( of which there are @xmath9 ) which can combine their genome in many different ways ( @xmath4 ) . as a consequence
, a naive implementation requires @xmath0 operations to calculate the distribution of recombinant genotypes .
it is intuitive that the complexity of this algorithm can be reduced : given a recombination pattern , only a fraction of the genome is passed on in sexual reproduction and all genotypes that agree on that fraction contribute identically
. we will show below that exploiting this redundancy allows to reduce the number of operations from @xmath0 to @xmath10 .
after selection , mutation , and recombination , the population distribution @xmath5 contains the expected number of individuals of genotype @xmath11 in the next generation . for stochastic population genetics ,
we still need to resample the population in way that mimics the randomness of reproduction .
this is achieved by resampling individuals according to a poisson distribution with mean @xmath12 for each genotype , which will result in a population size of approximately @xmath13 .
the probability of producing a genotype @xmath11 by recombination is @xmath14 where @xmath15 specifies the particular way the parental genomes are combined : @xmath16 if locus @xmath17 is derived from the mother ( resp .
the genotype @xmath18 is summed over ; it represents the parts of the maternal ( @xmath19 ) and paternal ( @xmath20 ) genotypes that are not passed on to the offspring .
we can decompose each parent into successful loci that made it into the offspring and wasted loci , as follows : @xmath21 and @xmath22 , where @xmath23 and a bar over a variable indicate respectively the elementwise and and not operators .
the function @xmath24 assigns a probability to each inheritance pattern , see @xcite for a more detailed explanation . in a facultatively sexual population ,
a fraction @xmath25 of @xmath5 is replaced by @xmath26 , while @xmath27 in an obligate sexual population .
the central ingredient for the efficient recombination is a fast - fourier algorithm for functions on the l - dimensional binary hypercube .
every function on the hypercube can be expressed as @xmath28 where @xmath29 and takes the values @xmath30 . in total
, there are @xmath4 coefficients @xmath31 for every subset of @xmath32 loci out of a total of @xmath1 loci @xcite .
similarly , each coefficient @xmath31 is uniquely specified by @xmath33 these nominally @xmath9 operations can be done in @xmath34 via the fft scheme illustrated in fig .
[ fig : fft ] . operations .
arrow going up indicate addition , going down substraction . for the general @xmath1 dimensional hypercubes @xmath1 cycles are necessary where terms differing at different bits are combined . ] with some algebra ( see online supplement ) , one can show that the generic fourier coefficient of @xmath26 is given by @xmath35 where the sum runs over all partitions of @xmath36 into groups of @xmath37 and @xmath38 denoted by @xmath39 and @xmath40 .
variables such as @xmath41 are the fourier coefficients of the genotype distribution , @xmath42 , and the crossover function @xmath24 is expanded into @xmath43 the latter coefficients can be calculated efficiently by realizing that for @xmath44 , there is exactly one term unequal to zero .
all subsequent terms can be calculated by successive marginalization of unobserved loci . in total
, calculating @xmath45 requires @xmath46 operations .
since there are @xmath47 terms of order @xmath32 , the entire calculation requires @xmath48 operations . in case of single crossover recombination
, the algorithm can be sped up further to @xmath49 .
ffpopsim is implemented in c++ with a python2 wrapper .
documentation , a number of examples , and test routines are provided . as an example
, we discuss here the problem of fitness valley crossing , which has received attention recently in the population genetics literature @xcite .
consider a fitness landscape where the wild - type genotype has ( malthusian ) fitness @xmath50 , while the quadruple mutant has fitness @xmath51 .
all intermediate genotypes have the same slightly deleterious fitness @xmath52 ( @xmath53 relative to wild - type ) . the time required for crossing the valley
can be computed by the following routine :
.... import ffpopsim l = 4 # number of loci n = 1e10 # population size # create population and set rates c = ffpopsim.haploid_lowd(l ) c.set_recombination_rates([0.01 ] * ( l-1 ) ) c.set_mutation_rate(1e-6 ) # start with wildtype : 0b0000 = 0 c.set_genotypes([0b0000],[n ] ) # set positive relative fitness for wildtype # and quadruple mutant : 0b1111 = 15 c.set_fitness_function([0b0000 , 0b1111 ] , [ s1 , s1+s2 ] ) # evolve until the quadruple mutant spreads while c.get_genotype_frequency(0b1111)<0.5 : c.evolve(100 ) print c.generation .... the runtime and memory requirements of @xmath3 still preclude the simulation of more than @xmath54 loci . for this reason , we also include a streamlined individual based simulation package with the same interface that can simulate arbitrarily large number of loci and has an overall runtime and memory requirements @xmath55 in the worst case scenario .
to speed up the simulation in many cases of interest , identical genotypes are grouped into clones .
this part of the library was developed for whole genome simulations of large hiv populations ( @xmath56 , @xmath57 ) and a specific wrapper class for hiv applications is provided . as of now , the library does not support dominance effects which would require a fitness function that depends on pairs of haploid genomes .
such an extension to diploid populations is straightforward .
we would like to thank boris shraiman for many stimulating discussion and pointing us at the fft algorithm .
this work is supported by the erc though stg-260686 .
12 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , * * ( ) , , issn , http://dx.doi.org/10.1016/s0167-739x(02)00171-1 . , and , , * * ( ) , . , , * * ( ) , ,
issn , http://dx.doi.org/10.1109/mcse.2007.55 . , , * * ( ) , . , and , , * * ( ) , . , , _
_ , volume ( ) . , and , , * * ( ) ,
. , and , , * * ( ) , . , , * * , . , , , and , , * * ( ) , . , , and , , * * ( ) ,
. , , and , , * * , .
the toolbox makes extensive use of the gnu scientific library ( http://www.gnu.org/software/gsl/ ) and the boost c++ library ( http://www.boost.org/ ) .
the python wrapper further requires numpy , scipy and matplotlib ( http://www.scipy.org ) .
if all of these are installed and the appropriate path are set , ffpopsim can be compiled using make .
installation instructions are provided in the ` install ` file .
the building process creates files inside the folder ` pkg ` ; c++ headers are created in ` pkg / include ` , the static c++ library in ` pkg / lib ` , and the python module files in ` pkg / python ` .
` ffpopsim ` contains two packages of c++ classes and python wrappers for forward population genetic simulations .
one for large populations and relatively few loci ( @xmath58 ) , another one for longer genomes .
the former is called ` hapoid_lowd ` and tracks the abundance of all possible genotypes .
the other one is called ` haploid_highd ` and tracks only genotypes present in the population .
the latter only allows for limited complextity of fitness functions and crossover patterns .
these two parts of the library have very similar syntax but work quite differently under the hood .
we will therefore describe them separately below .
a complete documentation in html is generated automatically from the source using doxygen and can be found in ` doc / html / index.html ` . since we assume that each locus is in one of two possible states @xmath59 , the genotype space is an @xmath1 dimensional binary hypercube .
the population is a distribution of individuals on that hypercube , and so are the mutant and recombinant genotypes .
also the fitness function @xmath60 , which assigns a number to each genotype @xmath61 , is a function on the hypercube .
for this reason , ` hapoid_lowd ` makes extensive use of a class ` hypercube_lowd ` that stores an @xmath1 dimensional hypercube and implements a number of operations on the hypercube , including a fast - fourier transform ( fft ) .
every function on the hypercube can be expressed as @xmath28 where @xmath62 , i.e. , simply a mapping from @xmath63 to @xmath64 .
there are @xmath65 coefficients @xmath31 for every subset of @xmath32 loci out of @xmath1 loci , so in total @xmath4 coefficients @xcite .
a coefficient @xmath31 is uniquely specified by @xmath66 these nominally @xmath9 operations ( @xmath4 for each coefficient ) can be done in @xmath34 via the fft scheme illustrated in fig .
[ fig : fft ] .
both the forward and reverse transform are implemented in ` hypercube_lowd ` .
arrow going up indicate addition , going down substraction . for the general @xmath1
dimensional hypercubes @xmath1 cycles are necessary where terms differing at different bits are combined . ]
an instance of ` hypercube_lowd ` can be initialized with the function values of the hypercube or with its fourier coefficients .
the population class , ` haploid_lowd ` , holds instances of ` hypercube_lowd ` for the population , the mutant genotypes , the recombinant genotypes and the fitness function . from a practical point of view
, an instance of a low - dimensional population is initialized in three steps , a. the class is instantiated + ....
haploid_lowd::haploid_lowd(int l=1 , int rng_seed=0 ) .... + where l is the number of loci and ` rng_seed ` is a seed for the random number generator ( a random seed by default ) ; b. the initial population structure is set by the functions + .... int haploid_lowd::set_allele_frequencies(double * freq , unsigned long n ) int haploid_lowd::set_genotypes(vector < index_value_pair_t >
gt ) int haploid_lowd::set_wildtype(unsigned long
n ) .... + set the population size and composition .
the first function initializes random genotypes in linkage equilibrium with the specifiec allele frequencies ` freq ` , the second explicitely sets a number of individuals for each genotype using the new type ` index_value_pair ` , and the last one sets a wildtype - only population of size ` n ` ; c. the fitness hypercube is initialized directly by accessing the attribute + .... haploid_lowd::fitness .... + in the population class . in traditional wright - fisher type models , in each generation ,
the expected frequencies of gametes with a particular genotype after mutation , selection , and recombination are calculated and then the population resampled from this gamete distribution .
we will now outline the steps required to update the population .
a more detailed discussion can be found in @xcite .
all the following steps are called by either one of the following functions : * .... int haploid_lowd::evolve(int gen=1 ) ....
+ updates the population looping over a specified number of generations ` gen ` ; * .... int haploid_lowd::evolve_norec(int gen=1 ) .... + is an alternative version that skips the resampling ( deterministic evolution ) ; * .... int haploid_lowd::evolve_deterministic(int gen=1 ) .... + is another alternative that skips the recombination step ( asexual evolution ) . let @xmath5 be the genotype distribution at the beginning of a generation . denoting the mutation rate towards the @xmath67 or @xmath52 state at locus @xmath17 by @xmath68 , the expected @xmath5 after mutation will be @xmath69 where @xmath70 denotes genotype @xmath11 with locus @xmath17 flipped from @xmath52 to @xmath67 or vice versa .
the first term is the loss due to mutation , while the second term is the gain due to mutation from neighbouring genotypes ( in terms of hamming distance ) .
the mutation rates can be specified by the folowing set of overloaded functions : either a single double rate ( same for every position and in both forward and backward sense ) , two double rates ( forward and backward rates ) , a @xmath1 dimensional double array ( site - specific , identical forward and backward rates ) , or a @xmath71 dimensional double array , * .... int haploid_lowd::set_mutation_rate(double rate ) ; .... + takes a single rate and sets it for every position and in both forward and backward sense ; * .... int haploid_lowd::set_mutation_rate(double rate_forward , double rate_backward ) ; .... + takes two rates and sets ` rate_forward ` as the forward rate ( @xmath72 ) and ` rate_backward ` as the backward rate ( @xmath73 ) * .... int haploid_lowd::set_mutation_rate(double * rates ) ; .... + takes the pointer to an array of length l and sets site - specific rates , the same for forward and backward mutations ; * .... int haploid_lowd::set_mutation_rate(double * * rates ) .... + takes a pointer to a pair of ( pointers to ) arrays , each of length l , which contain the site - specific rates for forward ( ` rates[0 ] ` ) and backward ( ` rates[1 ] ` ) mutations .
selection reweighs different the population of different genotypes according to their fitness as follows @xmath74 where @xmath75 is the population average of @xmath76 , which is required to keep the population size constant .
the corresponding function is .... int haploid_lowd::select_gametes ( ) .... for deterministic modeling , one generation would be completed at this point and one would repeat the cycle , starting with mutation again . for stochastic population genetics , we still need to resample the population in a way that mimics the randomness of reproduction . the easiest and most generic way to do this is to resample a population of size @xmath2 using a multinomial distribution with the current @xmath5 as sampling probabilities of different genotypes .
alternatively , one can sample individuals according to a poisson distribution with mean @xmath12 for each genotype , which will result in a population of approximately size @xmath13 . for large populations ,
the two ways of resampling are equivalent and we chose the latter ( much faster ) alternative .
the function .... int haploid_lowd::resample ( ) .... samples the next generation the expected genotype frequencies .
the expected population size used in the resampling is the carrying capacity . the computationally expensive part of the dynamics is recombination , which needs to consider all possible pairs of pairs of parents and all different ways in which their genetic information can be combined . in a facultatively sexual population ,
a fraction @xmath25 of the individuals undergo mating and recombination . in obligate sexual populations , @xmath27 .
the genotype distribution is updated according to the following rule : @xmath77 the distribution @xmath26 of recombinant gametes would naively be computed as follows : @xmath14 where @xmath78 specifies the particular way the parental genomes are combined : @xmath16 if locus @xmath17 is derived from the mother ( resp .
father ) . the genotype @xmath18 is summed over ; it represents the part of the maternal ( @xmath19 ) and paternal ( @xmath20 ) genotypes that is not passed on to the offspring .
we can decompose each parent into successful loci that made it into the offspring and wasted loci , as follows : @xmath79 and @xmath80 , where @xmath23 and a bar over a variable indicate respectively the elementwise and and not operators ( i.e. , @xmath81 ) .
the function @xmath24 assigns a probability to each inheritance pattern . depending on whether the entire population undergoes sexual reproduction or only a fraction @xmath25 of it , the entire population or a fraction @xmath25
is replaced with @xmath26 .
the central ingredient for the efficient computation of @xmath26 is the fourier decomposition introduced above . the generic fourier coefficient of @xmath26 is given by @xmath82 just as @xmath20 and @xmath19
can be expressed as a combination of @xmath11 and @xmath18 , we can invert the relation and express the generic @xmath83 as a function of @xmath20 and @xmath19 , as follows : @xmath84 . using this new basis and exchanging the order of summations ,
we obtain @xmath85 notice that @xmath86 can be pulled out of the two inner sums , because the odds of inheriting a certain locus by the mother / father is independent of what their genetic makeup looks like .
next we expand the product and introduce new labels for compactness , @xmath87 where @xmath37 is the number of loci inherited from the mother among the @xmath32 in @xmath88 .
@xmath37 runs from @xmath52 ( everything happens to be contributed by the father ) to @xmath32 ( everything from the mother ) .
@xmath89 and @xmath90 are all ( unordered ) partitions of @xmath17 into sets of size @xmath37 and @xmath38 , respectively .
now we can group all @xmath91 in the inner sum with @xmath86 , all @xmath92 with @xmath93 , and all @xmath94 with @xmath95 .
the three sums ( over @xmath78 , @xmath96 , and @xmath97 ) are now completely decoupled . moreover
, the two sums over the parental genotypes happen to be the fourier decomposition of @xmath5 .
hence , we have @xmath98 the quantity @xmath99 can be calculated efficiently , for each pair of partitions @xmath100 , by realizing that ( a ) for @xmath44 , there is exactly one term in the sum on the right that is non - zero and ( b ) all lower - order terms can be calculated by successive marginalizations over unobserved loci .
for instance , let us assume that @xmath101 and that the only missing locus is the m - th one .
we can compute @xmath102 there are @xmath65 ways of choosing @xmath32 loci out of @xmath1 , which can be inherited in @xmath103 different ways ( the partitions in @xmath104 and @xmath105 in eq .
( [ eq : reccoeff ] ) ) such that the total number of coefficients is @xmath3 .
note that these coefficients are only calculated when the recombination rates change .
furthermore , this can be done for completely arbitrary recombination patterns , not necessarily only those with independent recombination events at different loci . `
haploid_lowd ` provides a function to calculate @xmath86 from recombination rates between loci assuming a circular or linear chromosome .
the probability of a particular crossover pattern is calculated assuming independent crossovers .
the function .... int haploid_lowd::set_recombination_rate(double * rec_rates ) .... assumes a double array of length @xmath106 for a linear chromosome and of length @xmath1 for a circular chromosome . for a linear ( resp .
circular ) chromosome , the i - th element of the array is the probability of recombining after ( resp . before ) the i - th locus .
furthermore , the mating probability @xmath25 must be specified explicitely via the attribute .... haploid_lowd::outcrossing_rate .... the default is obligate sexual reproduction .
the code offers a simpler alternative for free recombination . in this case , only the global mating probability @xmath25 needs to be entered .
if the user does not set the recombination rates via ` set_recombination_rate ` , free recombination is the default behaviour .
otherwise , this option is controlled by the following boolean attribute ....
haploid_lowd::free_recombination .... note that , in a circular chromosome , there is effectively one more inter - locus segment ( between the last and the first locus ) in which crossovers can occur , and the total number of crossovers has to be even . assuming independent crossovers , the global recombination rate of circular chromosomes is lower than a linear chromosome of the same length by a factor of @xmath107 , where @xmath108 is the recombination rate between the first and last loci .
the recombination process itself is initiated by .... int haploid_lowd::recombine ( ) .... for more than @xmath109 loci , storing then entire genotype space and all possible recombinants becomes prohibitive .
hence we also include a streamlined individual based simulation package that can simulate arbitrarily large number of loci and has an overall runtime and memory requirements @xmath55 in the worst case scenario .
the many - loci package uses the same interface as the few - loci one .
this makes it easy , for example , to first test an evolutionary scenario using many ( all ) loci and to focus on the few crucial ones afterwards . to speed up the program in many cases of interest
, identical genotypes are grouped into clones .
this part of the library was developed for whole genome simulations of large hiv populations ( @xmath56 , @xmath57 ) . a specific wrapper class for hiv applications
is provided . for more than 20 loci ,
it becomes infeasible to store the entire hypercube .
instead , we store individual genotypes as bitsets .
each genotype , together with the number of individuals that carry it , as well as traits and fitness associated with it is stored for as long as it is present in the population .
all of this is aggregated in the structure ` clone_t ` .
the population is a vector of clones .
each generation clones size are updated and added to the new generation , new clones are produced , and empty ones deleted .
fitness functions are again functions on the hypercube .
the latter is implemented as ` hypercube_highd ` . instead of storing all possible fitness values , `
` stores non - zero fourier coefficients .
whenever a new genotype is produced , its fitness is calculated by summing the appropriate coefficients . to implement mutation , a poisson distributed number @xmath110 with mean @xmath111
is drawn for each locus @xmath17 and @xmath110 mutations are introduced at locus @xmath17 into @xmath110 randomly chosen genotypes .
mutations are bit - flip operations in the bitset .
only a global mutation rate is currently supported .
prior to selection , the population average @xmath112 and a growth rate adjustment @xmath113 are computed .
the latter is used to keep the population size close to the carrying capacity @xmath114 .
the size of each clone is then updated with a poisson distributed number with mean @xmath115 , where @xmath116 is the recombination rate .
another poisson distributed number with mean @xmath117 is set aside for recombination later .
the individuals marked for sexual reproduction during the selection step are shuffled and paired .
for each pair , a bitset representing the crossover pattern is produced and two new clones are produced from the two parental genomes .
alternatively , all loci of the two genomes can be reassorted at random .
the c++ library includes python bindings that greatly simplify interactive use and testing .
the wrapping itself is done by swig @xcite .
most notably , the c++ classes ` haploid_lowd ` , ` haploid_highd ` and the hiv - specific subclass are fully exposed to python , including all their public members . the performance speed for evolving a population
is unchanged , since the ` evolve ` function iterates all steps internally for an arbitary number of generations
. the bindings are not completely faithful to the c++ interface , to ensure a more intuitive user experience .
for instance , c++ attribute set / get members are translated into python properties via the builtin ` property ` construct .
furthermore , since direct access to the ` hypercube_lowd ` instances from python is not straightforward , a few utility functions have been written to do common tasks .
the fitness hypercube can be set easily by either one of ....
haploid_lowd.set_fitness_function(genotypes , fitnesses ) haploid_lowd.set_fitness_additive(fitness_main ) .... the former function is used to set specific points on the hypercube : it takes a sequence of genotypes ` genotypes ` ( integers or binary literals using ` 0b ` notation ) and a sequence of fitness values ` fitnesses ` , corresponding to those genotypes .
any missing point on the fitness hypercube will be consider neutral .
the second function creates an additive fitness landscape , in which main effects are specified by the l - dimensional input sequence ` fitness_main ` .
after installation , the ffpopsim library can be used in python as a module , e.g. .... from ffpopsim import haploid_lowd .... the bindings make heavy use of the numpy library and its swig fragments and typemaps @xcite .
we therefore recommend to import numpy before ffpopsim , although this is not strictly necessary .
moreover , the python binsings include a few functions for plotting features of the population , such as genetic diversity .
the python module matplotlib is required for this purpose @xcite .
the hiv - specific part of the code has been expanded further in python to enable quick simulations of viral evolution under commonly studied conditions .
in particular , random genotype - phenotype maps for viral replication capacity and drug resistance can be generated automatically from a few parameters , via the functions ....
hivpopulation.set_resistance_landscape .... the input parameters reflect a number of typical properties of hiv populations , such as the fraction of sites carrying potentially adaptive mutations .
see the inline python documentation for further details on these functions . moreover , since studies of hiv evolution often involve a large number of genotypic data , a function for saving the genotype of random individuals from the current population in a compressed format has been added .
the syntax is the following : .... hivpopulation.write_genotypes_compressed(filename , number_of_individuals ) .... where ` filename ` is the name of the file , in which the data are to be stored , and ` number_of_individuals ` if the size of the random sample .
the data can be accessed again by the standard numpy ` load ` function .
one of the most striking effects of genetic epistasis is the slowdown of evolution when a combination of mutations is beneficial , but intermediate mutants are deleterious compared to wildtype .
such scenario is relevant in applications , for instance for the emergence of bacterial or viral resistance to drugs @xcite .
not surprisingly , recombination plays a central role in this process . on the one hand
, it enhances the production rate of multiple mutants , on the other it depletes the class of complete mutants by back - recombination with deleterious backgrounds .
if the script is run with different recombination rates , the effect of this parameter on the time for valley crossing can be investigated , as shown in fig .
[ fig : valley ] .
the full scripts producing the figures is provides as separate file . during an hiv infection ,
the host immune system targets several viral epitopes simultaneously via a diverse arsenal of cytotoxic t - cells ( ctls ) .
mutations at several loci are thus selected for and start to rise in frequency at the same time but , because of the limited amount of recombination , end up in wasteful competition ( interference ) at frequencies of order one .
the theoretical description of genetic interference is involved and often limited to two - loci models , but ffpopsim makes the simulation of this process straightforward .
the following script evolves a 4-loci population under parallel positive selection and tracks its genotype frequencies : # evolve until fixation of the quadruple mutant , # storing times and genotype frequencies times = [ ] genotype_frequencies = [ ] while pop.get_genotype_frequency(0b1111 ) < 0.99 and pop.generation<1e7 : pop.evolve ( ) times.append(pop.generation ) genotype_frequencies.append(pop.get_genotype_frequencies ( ) ) .... hiv evolution during chronic infection is determined by a number of parallel processes , such as mutation , recombination , and selection imposed by the immune system . in combination , these processes give rise to a complicated dynamics and do nt understand how population features such as population diversity depend on model parameters .
hence simulations are an important source of insight .
ffpopsim offers a specific double c++/python interface to this problem via its class ` hivpopulation ` .
the following script simulates an hiv population for one thousand generations , under a random static fitness landscape , and stores a hundred random genomes from the final population in a compressed file : numpy can be used subsequently to analyze the genome sequences .
alternatively , the internal python functions can be used , e.g. for calculating the fitness distribution directly using ` hivpopulation.get_fitness_histogram ` , as shown in fig .
[ fig : hivfitness ] .
the full scripts producing the figures is provides as separate file . | the analysis of the evolutionary dynamics of a population with many polymorphic loci is challenging since a large number of possible genotypes needs to be tracked . in the absence of analytical solutions ,
forward computer simulations are an important tool in multi - locus population genetics .
the run time of standard algorithms to simulate sexual populations increases as @xmath0 with the number @xmath1 of loci , or with the square of the population size @xmath2 .
we have developed algorithms that allow to simulate large populations with a run - time that scales as @xmath3 .
the algorithm is based on an analog of the fast - fourier transform ( fft ) and allows for arbitrary fitness functions ( i.e. any epistasis ) and genetic maps .
the algorithm is implemented as a collection of c++ classes and a python interface . *
availability : * http://code.google.com / p / ffpopsim/. |
much of what we know about the universe at high redshift arises from observations of the rest - frame ultraviolet emission of distant galaxies that is accessible from the ground at optical and near - infrared wavelengths .
for this reason , significant effort has been devoted over the past several years to improve our understanding of the star formation history and dust attenuation in galaxies drawn from ultraviolet surveys .
recent progress in this area has come mainly from studies of a sample of 57 nearby starburst galaxies , for which a wide range of observations were compiled by @xcite . in particular , for these galaxies there is a remarkably tight correlation between ratio of total far - infrared to ultraviolet luminosities , , and ultraviolet spectral slope , @xmath0 , which is referred to as the ` infrared excess ( irx)-ultraviolet ( uv ) ' relation ( here @xmath0 is defined by a power - law fit of the form @xmath1 to the spectrum at wavelengths @xmath2 ; @xcite ) .
@xcite find that the relation allows reliable estimates of the attenuation by dust at ultraviolet wavelengths , , from either or @xmath0 .
this relation , if universal , represents our best hope to estimate the attenuation by dust and hence the star formation rates of high - redshift galaxies , for which the rest - frame optical and infrared emission can not be observed at present .
the main uncertainty affecting the relation is that it has been established only for starburst galaxies .
recently , @xcite has compiled a small sample of nearby , more quiescent galaxies drawn from ultraviolet observations with various satellites ( here we use the term ` quiescent ' to describe the broad class of star - forming galaxies between active starburst and dormant early - type galaxies ) .
his results suggest that quiescent galaxies deviate from the relation for starburst galaxies , in the sense that quiescent galaxies tend to have ` redder ' ultraviolet spectra at fixed .
this implies that the simple @xcite recipe to estimate from either or @xmath0 may not hold for all galaxy types .
the results of @xcite raise several questions . can star formation history alone account for the distribution of galaxies in the diagram ? or do the galaxies have different dust properties ? @xcite have developed a simple dust model , based on an idealized prescription of the main features of the ( clumpy ) interstellar medium ( ism ) , which allows one to interpret the relation and the emission - line properties of starburst galaxies in terms of star formation rate and dust content .
this model accounts for the different attenuation affecting young and old stars in a galaxy , because of the dispersal of the ( giant molecular ) clouds in which stars form . using this model , @xcite
show that the tight relation for starburst galaxies can be understood most simply as a sequence in the overall dust content .
clearly , to gain insight into the origin of the relation for different types of galaxies , we require further observations and modelling of not only the ultraviolet and infrared luminosities , but also other observable quantities with distinct dependences on dust content and star formation history . in this paper , we compile a new sample of 115 nearby , non - seyfert galaxies spanning a wide range of star formation activities , from starburst to nearly dormant , to explore the physical parameters driving the location of a galaxy in the diagram .
our sample is drawn from ultraviolet observations with various satellites .
we combine these observations with infrared observations and , for a subset of the galaxies , with optical spectroscopy from various sources .
we examine , in particular , the 4000 spectral discontinuity and the emission equivalent width , which are good indicators of star formation activity , and the ratio , which is a good indicator of attenuation by dust . based on these data alone
, it appears that the broadening of the relation for quiescent star - forming galaxies is most likely driven by star formation history .
we combine the simple dust model of @xcite with the population synthesis code of @xcite and show that this model can account for the ultraviolet and far - infrared properties of the galaxies in our sample .
the model allows us to understand why the ratio of present to past - averaged star formation rates is the most likely ` second ' parameter that , in combination with dust content , determines the position of a galaxy in the diagram . in section 2 below , we compile our galaxy sample and investigate the origin of the relation from an observational viewpoint .
in section 3 , we interpret the relation using the model .
we explore the implications of our results for estimates of the ultraviolet attenuation in galaxies in section4 .
our conclusions are summarized in section5 .
in this section , we re - examine the relation from an observational viewpoint .
we combine ultraviolet observations from various satellites with observations from the _ infrared astronomical satellite _ ( _ iras _ ) to compile a new sample of nearby , non - seyfert galaxies spanning a wide range of star formation activities in the diagram .
we explore how the position of a galaxy in this diagram depends on observable properties with distinct dependences on dust content and star formation history .
we compile a new sample of nearby , non - seyfert galaxies spanning a wide range of star formation activities , using available observations with the _ international ultraviolet explorer _ ( _ iue _ ) , the _ orbiting astronomical observatory _ ( _ oao _ ) and the _ astronomical netherlands satellite _ ( _ ans _ ) .
we first extract all ( non - seyfert ) spiral , elliptical , irregular and emission - line galaxies ( classes 80 , 81 , 82 and 88 ) from the _ iue _ newly - extracted spectra ( ines version 3.0 ) archive .
the spectra in this database extend over two wavelength domains , corresponding to the far- ( 11501980 ) and near - ultraviolet ( 18503350 ) spectrographs on - board _ iue_. we reject galaxies with only near - ultraviolet spectra , for which the ultraviolet spectral slope can not be measured reliably .
when both the far- and near - ultraviolet spectra are available for a galaxy , we merge them to form a single ultraviolet spectrum encompassing the range from 1150 to 3350 ( the spectra usually agree to within a few percent in the overlapping region around 19001950 ) .
we remove foreground galactic extinction using the dust maps of @xcite and the galactic extinction curve of @xcite .
furthermore , we co - add the spectra of galaxies observed multiple times to increase the signal - to - noise ratio . for the ultraviolet spectral slope to be measured with reasonable accuracy , a minimum signal - to - noise ratio per pixel of 3 is required ( see appendix ) .
a total of 110 galaxies with both far- and near - ultraviolet spectra and 185 galaxies with only far - ultraviolet spectra satisfy this cut in signal - to - noise ratio .
we require two additional conditions to include galaxies in our final sample : that optical photometry be available to estimate the ultraviolet aperture correction , and that _ iras _ 25 , 60 and 100@xmath3 m observations be available to estimate the total luminosity reradiated by dust .
the above sample of 295 galaxies includes the 57 starburst galaxies with optical angular diameters less than @xmath4 studied by @xcite , for which most of the concentrated ultraviolet emission could be observed within the @xmath5 _ iue _ aperture .
_ iras _ 25 , 60 and 100@xmath3 m observations are available for 50 of the meurer et al
. galaxies , which we include in our final sample . for other galaxies , reasonable ultraviolet aperture corrections may be derived as a function of the effective optical radius and morphological type ( ) . morphological types are known for _ iue _ galaxies , but effective optical radii must be obtained from independent observations .
we thus cross - correlate the above sample of 238 _ iue _ galaxies not included in the @xcite sample with the data release one of the sloan digital sky survey ( sdss dr1 ; @xcite ) . in this way
, we obtain effective @xmath6-band radii for 41 galaxies ( with typical , i.e. median @xmath7 ) , for which we compute ultraviolet aperture corrections using the prescription of .
the corrections vary from galaxy to galaxy , with a mean logarithmic ratio of corrected - to - observed ultraviolet fluxes of @xmath8 .
only 33 of these galaxies have available _ iras _ 25 , 60 and 100@xmath3 m observations and can be included in our final sample .
these typically have morphological types earlier than the starburst galaxies of the @xcite ( sa - sc versus sc - im , although some starburst galaxies are also present ) and similar absolute @xmath9-band magnitudes , @xmath10 ( we adopt a hubble constant @xmath11kms@xmath12mpc@xmath12 ) .
we supplement this sample with ( non - seyfert ) galaxies for which multi - band ultraviolet imaging with _ ans _ and _ oao _ is available from the compilation of . in the spectral range of interest to us ,
the galaxies observed with _
ans _ were imaged at 1550 , 1800 , 2200 and 2500 through a fixed aperture of @xmath13 .
those observed with _
were imaged at 1550 , 1910 and 2460 through a fixed circular aperture of @xmath14 diameter . of the 95 galaxies observed with _
ans _ in the compilation , 34 have available _ iras _ 25 , 60 and 100@xmath3 m observations .
we reject 5 galaxies with optical diameters larger than @xmath15 , for which the ultraviolet emission does not appear to be clearly concentrated within the _ ans _ aperture , and 2 galaxies with larger - aperture imaging available from _ oao _ ( see below ) .
the remaining 27 galaxies have typical morphological types sb - sd and absolute @xmath9-band magnitudes similar to those of the galaxies in the _ iue _ sample , @xmath16 .
of the 36 galaxies with _ oao _ imaging , 11 have available _ iras _ 25 , 60 and 100@xmath3 m observations .
we reject 6 galaxies with optical diameters larger than @xmath17 .
we thus retain only 5 galaxies with _ oao _ imaging in our final sample , two sc galaxies with @xmath18 , two sd galaxies with @xmath19 and an i m galaxy with @xmath20 .
our final sample contains 115 galaxies with both ultraviolet and infrared data , of which 83 were observed with _ iue _ , 27 with _ ans _ and 5 with _ oao_. about half are starburst galaxies ( most belonging to the @xcite sample ) , while the other half are earlier - type , more quiescent galaxies spanning a similar range of absolute @xmath9-band magnitudes .
we compute ultraviolet spectral slopes for all the galaxies in our sample . @xcite
define the ultraviolet spectral slope @xmath0 by a power - law fit of the form @xmath21 ( where @xmath22 is the flux per unit wavelength ) to the spectrum in ten continuum bands in the range @xmath23 .
we can not use this definition here , because it does not apply to galaxies for which only multi - band ultraviolet imaging is available .
instead , we adopt the following definition of the ultraviolet spectral slope for all the galaxies in our sample , @xmath24 where @xmath25 and @xmath26 are the effective wavelengths of the far- ( 13501800 ) and near - ultraviolet ( 18003000 ) filters on board the _ galaxy evolution explorer _ ( _ galex _ , @xcite ) , and and are the mean flux densities ( per unit wavelength ) through these filters . in the appendix , we use a reference sample of 110 _ iue _ spectra to calibrate the conversions between and the ultraviolet spectral slopes estimated from observations with different satellites .
we show that , in particular , ultraviolet spectral slopes estimated from _
ans _ and _ oao _ imaging observations can be converted into with an rms uncertainty of only @xmath27 .
this is similar to the uncertainty introduced in by the lack of near - ultraviolet spectra for some _ iue _ galaxies in our sample .
for all the galaxies in our sample , we compute the ultraviolet flux as @xmath28 , as described in the appendix . to estimate the total far - infrared flux @xmath29 from the observed _ iras _ flux densities @xmath30 , @xmath31 and @xmath32 , we appeal to the recent prescription of @xcite .
this is based on model fits to the spectra of a large sample of nearby , normal ( i.e. non - seyfert ) star - forming galaxies across the whole wavelength range from 3 to 850@xmath3 m .
we use equation ( 5 ) of dale & helou to compute @xmath29 ( @xmath33 in their notation ) from @xmath34 , @xmath35 and @xmath32 . for reference , for the _ iue _ starburst galaxies in our sample , the values of @xmath29 computed in this way are typically 50% larger than those estimated by @xcite from only @xmath35 and @xmath36 using the older prescription of helou et al .
( 1988 ; see also @xcite ) . neglecting possible anisotropies ( section4 ) , we equate the ratio of total far - infrared to ultraviolet fluxes @xmath37 to the corresponding luminosity ratio .
fig.[irx ] shows as a function of for the 115 galaxies in our sample .
different symbols distinguish the 50 starburst galaxies studied by meurer et al .
( 1999 , solid dots ) and the ( typically more quiescent ) galaxies observed with _
iue_+sdss ( stars ) , _ ans _ ( open triangles ) and _ oao _ ( filled triangles ) . to a first , rough approximation , is a function of for the whole sample . as found previously by meurer et al .
( 1995 , 1999 ) , the two quantities are tightly correlated for starburst galaxies , for which the typical scatter at fixed amounts to a factor of less than 2 in .
the main novelty introduced by the inclusion of more quiescent galaxies in our sample is an increase of the scatter to a factor of more than 3 toward lower values at fixed .
lies substantially above the relation for starburst galaxies .
this is not unexpected , since our sample of galaxies drawn from ultraviolet surveys ( with typical infrared luminosities @xmath38 ) does not include ` luminous and ultraluminous infrared galaxies ' , which tend to have larger than starburst galaxies at fixed ( presumably because of the heavy obscuration of the sources of ultraviolet radiation ; @xcite ) . ]
the effect is real , as the observational scatter is much larger than the typical ( i.e. median ) measurement errors ( which are shown in the lower right corner of fig.[irx ] ) .
we note that the _ shape _ of the relation does not appear to be affected by the inclusion of quiescent galaxies in our sample .
what are the main physical parameters driving the relation between ultraviolet spectral slope and ratio of far - infrared to ultraviolet luminosities for star - forming galaxies ?
@xcite show that , for starburst galaxies , the tight relation between and can be understood most simply as a sequence in the overall dust content of the galaxies .
the inclusion of quiesent star - forming galaxies in fig.[irx ] reveals that a ` second parameter ' tends to broaden the relation .
we may begin by identifying this second parameter as the perpendicular distance from the mean relation for starburst galaxies .
following @xcite , we define the mean relation for starburst galaxies by a least - squares fit of the type @xmath39 to the observations of the 50 starburst galaxies in our sample .
the result is shown by the solid line in fig.[irx ] .
we can compute the perpendicular ( i.e. shortest ) distance from each point in the diagram to this line .
we adopt the convention of negative for galaxies with lower than the mean value for starburst galaxies at fixed ( see fig.[irx ] ) .
we expect the second parameter affecting the relation to be connected to star formation history , since the relation becomes broader when quiescent star - forming galaxies are considered in addition to starburst galaxies .
it is thus natural to explore how observed quantities with different dependences on the star formation history correlate with .
we first consider the 4000 spectral discontinuity index , defined as the ratio of the average flux densities in the narrow bands 38503950 and 40004100 ( @xcite ) .
this index depends somewhat on metallicity but correlates more with the ratio of present to past - averaged star formation rates in galaxies , and hence , it is a valuable indicator of the history of star formation ( @xcite ) .
also , the narrowness of the bands used to compute makes it largely insensitive to attenuation by dust .
medium - resolution spectra ( 39 fwhm ) of 64 galaxies in our sample ( 35 starburst and 29 quiescent star - forming ) are available from @xcite , @xcite , @xcite , and @xcite .
fig.[dpobs]a shows measured from these spectra as a function of the perpendicular distance from the mean starburst relation .
the spearman rank correlation coefficient is @xmath40 , indicating that the correlation between and is significant at the @xmath41 level for this sample size .
this result is remarkable , as it is direct observational evidence that star formation history is relevant to the second parameter affecting the relation .
another observable quantity that is sensitive to the ratio of present to past - averaged star formation rates is the emission equivalent width ( @xcite ; @xcite ) .
this is the ratio of the emission - line luminosity produced by young massive stars to the continuum luminosity per unit wavelength produced by older stars at @xmath42 . in fig.[dpobs]b
, we show as a function of for 59 galaxies ( 31 starburst and 28 quiescent star - forming ) in our sample ( we adopt the convention of positive equivalent width for emission ) .
the emission equivalent widths were measured from the same optical spectra as above following the procedure outlined in section2 of @xcite , which allows accurate correction for stellar absorption .
the spearman rank correlation coefficient in this case is @xmath43 , indicating that the correlation between and is significant at the @xmath44 level for the sample size .
we note that dust can reduce through the absorption of ionizing photons . however , the effect is modest for quiescent star - forming galaxies , typically only 2030 per cent , and can not account for the trend seen in fig.[dpobs]b ( @xcite ) .
taken together , figs[dpobs]a and b show that two different tracers of the ratio of present to past - averaged star formation rates , and , correlate with .
this is strong evidence that the ratio of present to past - averaged star formation rates could be the second parameter affecting the relation . in an attempt to further understand the relative influence of star formation history and attenuation by dust on the relation , we plot other properties of the galaxies against in figs[dpobs]c and d : the ratio of the _ iras _ @xmath31 and @xmath32 flux densities and the ratio of the and emission line luminosities .
the 60100ratio , available for all 115 galaxies in our sample , depends on the relative contributions by young and old stars to dust heating in a galaxy ( e.g. , @xcite ) .
hence it is an indirect tracer of the star formation history that also depends on the dust content ( and potentially the spatial distribution of dust relative to stars ) .
the ratio , available for 48 galaxies ( 29 starburst and 19 quiescent star - forming ) from the same optical spectra as above , is strongly sensitive to attenuation by dust but not to the past history of star formation .
the spearman rank correlation coefficients are @xmath45 for the relation between 60100 and and @xmath46 for that between and . for the different numbers of galaxies in each case
, this implies that the correlation between 60100 and is significant at the @xmath44 level , while there appears to be little or no correlation between and .
these findings reinforce our contention that the second parameter affecting the relation is primarily star formation history rather than dust content .
based on the observational evidence above , we conclude that , while dust content is the primary driver of the relation for starburst galaxies , the broadening of the relation for quiescent star - forming galaxies is most likely driven by star formation history , and in particular the ratio of present to past - averaged star formation rates . in the next section , we show that this observational result can be understood in the framework of a simple model for the production of starlight and its transfer through the interstellar medium ( ism ) of galaxies .
in this section , we explore the physical origin of the relation from a more theoretical standpoint . we appeal to a simple model for the production of starlight and its transfer through the interstellar medium in galaxies .
the model is based on a combination of the @xcite population synthesis code and the @xcite prescription for the absorption of starlight by dust .
the charlot & fall model relies on an idealized description of the main features of the ism .
stars are assumed to form in interstellar ` birth clouds ' ( i.e. giant molecular clouds ) .
after @xmath47yr , young stars disrupt their birth clouds and migrate into the ` ambient ism ' .
the ` effective absorption ' curve describing the attenuation of photons emitted in all directions by stars of age @xmath48 in a galaxy is given by the simple formula @xmath49 where @xmath50 is the total effective @xmath51-band optical depth seen by young stars .
the adjustable parameter @xmath3 defines the fraction of the total effective absorption optical depth contributed by the ambient interstellar medium ( @xmath52 on average , with substantial scatter ) .
in addition , a fraction @xmath53 of the effective absorption optical depth @xmath54 of the birth clouds is assumed to arise from the regions ionized by young stars in the inner parts of these clouds .
we note that , in the case of optically thick birth clouds , the ultraviolet radiation escaping from a galaxy is produced by stars older than @xmath55yr , while the total far - infrared luminosity is produced by stars of all ages .
@xcite used their model to interpret the relations between ultraviolet spectral slope , ratio of far - infrared to ultraviolet luminosities and ratio for the starburst galaxies in the @xcite sample ( section 2 ) .
they showed that the finite lifetime of the birth clouds is a key ingredient for resolving the apparent discrepancy between the attenuation of line and continuum radiation in these galaxies .
they also showed that the relation can be understood most simply as a sequence in the overall dust content of the galaxies .
for simplicity , @xcite
limited their study to models with constant star formation rates and young ages , as appropriate to interpret observations of starburst galaxies .
the results of section2 above suggest that star formation history plays an important role in defining the relation for more quiescent galaxies . to investigate the dependence of the relation on star formation history , we assume that the star formation rate varies with time as @xmath56 from @xmath57 to the present galaxy age @xmath58 .
the parameter @xmath59 is the inverse time - scale of star formation . for simplicity ,
all models are assumed for the moment to have fixed solar metallicity .
our main goal is to identify the model parameter that most simply accounts for the offset of normal , quiescent star - forming galaxies from the mean relation for starburst galaxies in fig.[irx ] . to do this
, we must explore the influence of each primary model parameter ( , @xmath59 , and @xmath3 ) on observable quantities available for galaxies in our sample , i.e. the ultraviolet spectral slope , the ratio of far - infrared to ultraviolet luminosities and the 4000 discontinuity .
the influence of and @xmath3 on and can be inferred from the results presented in figure2 of @xcite : these parameters , in particular the effective dust absorption optical depth in the ambient ism , @xmath60 , control the position of a galaxy _ along _ the relation .
the narrow - band spectral index is only weakly sensitive to variations in and @xmath3 ( section2 ) .
we find that the influence of the star formation history on the relation through the parameters and @xmath59 is best described in terms of the ratio of present to past - averaged star formation rates .
this is defined by @xmath61 for the expression of @xmath62 in equation ( [ sfr ] ) , we have @xmath63 for reference , a galaxy with constant star formation rate ( @xmath64 ) has @xmath65 at all ages .
than obtained when using the definition given in equation ( [ bparam ] ) . ]
fig.[irxbpar ] summarizes the main results of our analysis . in fig.[irxbpar]a
, we show as a function of for sequences of models with different ages for four different time - scales of star formation , corresponding to @xmath66 , 0.5 , 1.0 and 2.0gyr@xmath12 , and for @xmath67 and @xmath68 .
the sequences are plotted as solid lines at ages when the ratio of present to past - averaged star formation rates is larger than @xmath69 and as dashed lines at greater ages , down to @xmath70 . from @xmath65 ( age @xmath57 ) to @xmath69 , the distance from the mean starburst relation increases as @xmath71 decreases along the model sequences , models with @xmath72 occupying regions of the diagram populated by more quiescent galaxies .
this behavior arises mainly from an increase in as @xmath71 decreases .
models with @xmath73 are dominated by stellar populations with intrinsically redder spectra and hence larger ultraviolet spectral slopes than starburst galaxies .
we note that , for @xmath74 , all models in fig.[irxbpar]a follow a similar path in the diagram . for @xmath75 , the evolution in the diagram
depends more strongly on the adopted time - scale of star formation . at fixed @xmath71 , is larger in models with short time - scales of star formation ( large @xmath59 ) , which contain a smaller proportion of ultraviolet - bright stars than models with extended star formation . in this @xmath71 range ,
increases at late ages because of the contribution to dust heating by the optical radiation from old stars ( section4 ) .
we note that models with small @xmath71 ` returning ' at late ages to the mean relation for starburst galaxies are considerably fainter than starburst galaxies and have intrinsically red optical spectra .
the similarity for @xmath76 between models with different @xmath59 in the diagram prompts us to plot , in fig.[irxbpar]b , as a function of for sequences of models with different dust content ( i.e. different ) , for two ratios of present to past - averaged star formation rates , @xmath77 and 0.3 . for simplicity , we assume that the fraction of the total effective absorption optical depth contributed by the ambient ism is fixed at the standard value @xmath68 in all models .
the striking result of fig.[irxbpar]b is that the model sequences with constant @xmath71 are remarkably parallel to the mean relation for starburst galaxies , the @xmath69 sequence being offset toward the region of the diagram populated by quiescent star - forming galaxies .
this is consistent with the observational evidence from figs[dpobs]a and b that , while dust content is the primary driver of the relation , the broadening of the relation for quiescent star - forming galaxies is most likely driven by star formation history .
this result does not depend on the specific combination of @xmath59 and @xmath78 used to compute models with @xmath69 in fig.[irxbpar]b .
we find that , for @xmath79 , sequences of models with different dust content are also parallel to the mean relation for starburst galaxies .
however , as expected from fig.[irxbpar]a , the location of the sequences in this @xmath71range depends sensitively on the choices of @xmath59 and @xmath78 ( see also section4 ) .
figs[irxbpar]a and b suggest that models with simple continuous star formation histories can not easily reproduce the observations of galaxies lying at the largest perpendicular distances from the mean starburst relation in our sample , unless very short time - scales of star formation ( @xmath80 ) and very small ratios of present to past - averaged star formation rates ( @xmath81 ) are invoked .
such parameters are not typical of quiescent star - forming galaxies . in fig.[irxbpar]c
, we show as a function of for models with a star formation time - scale parameter @xmath82 , more typical of a spiral galaxy , which undergo an extra burst of star formation of duration @xmath83yr at the age of 6gyr , for @xmath67 and @xmath68 .
we show the time evolution of these models for bursts representing 3 , 10 and 30 per cent of the total mass of stars formed by the continuous model over 13.5gyr .
when the burst starts , each model first joins the mean relation for starburst galaxies . shortly after the burst ends , and hence drop rapidly because of the dispersal of the ( optically thick ) birth clouds of the last stars formed during the burst .
then , and increase again as the ultraviolet light from these stars fades and reddens .
a maximum in is reached about 0.51gyr after the burst ended .
the subsequent evolution toward smaller , as the underlying continuous star formation again starts to dominate the production of the ultraviolet light , is remarkable for two reasons .
first , it is very slow , implying that the models spend several gigayears in the region of the diagram populated by galaxies far away from the mean starburst relation .
second , the @xmath71 parameter is only slightly lower than that of the undisturbed continuous model and hence typical of quiescent star - forming galaxies ( note that this conserves the trend of decreasing @xmath71 with increasing distance from the mean starburst relation ) .
thus , models including minor bursts of star formation on top of continuous star formation histories can account in a natural way for the observations of galaxies at large distances from the mean starburst relation in the diagram .
the simple model described above provides a natural framework for understanding the physical origin of the diagram . as shown in fig.[irxbpar ] , this model can account for the observational evidence from section2 that , while dust content is the primary driver of the relation for starburst galaxies , the broadening of the relation for quiescent star - forming galaxies is most likely driven by star formation history , and in particular the @xmath71 parameter .
these results do not depend sensitively on the population synthesis code used for computing the production of starlight in the model .
we have checked that adopting the pgase population synthesis code (; version 2.0 ) instead of the @xcite code has a negligible influence on the results of fig.[irxbpar ] .
it is worth noting that other effects , such as anisotropic emission from galaxies and spatial distribution of the dust , can potentially affect the positions of galaxies in the diagram .
the inclination angles available for 59 quiescent star - forming galaxies observed with _
iue_+sdss , _ ans _ and _ oao _ in our sample span the full range from 0 to 90 degrees and do not exhibit any specific trend with perpendicular distance from the mean starburst relation .
therefore , we do not expect orientation effects to play a major role in the broadening of the relation for these galaxies .
@xcite show that , for some combinations of optical properties and spatial distributions of the dust in their model , it is possible to produce ratios lower than observed for starburst galaxies at large ultraviolet spectral slopes .
however , the results in their figures 6 and 13 indicate that such models could not account for the low ratios of quiescent star - forming galaxies with small in our sample .
changes in the spatial distribution of the dust also do not provide a natural explanation for the strong drop in equivalent width with increasing distance from the mean starburst relation in fig.[dpobs]b ( see figure 6c of @xcite ) .
we conclude that , although orientation and dust geometry may have minor effects on the positions of galaxies in the diagram , they do not seem responsible for the overall broadening of the relation for quiescent star - forming galaxies in our sample .
this broadening can be most simply understood as a sequence in the star formation histories of the galaxies .
the results presented in the previous sections have important implications for estimates of the ultraviolet attenuation and hence of star formation rates in galaxies . @xcite
derive simple formulae for estimating from either the ultraviolet spectral slope or the ratio of far - infrared to ultraviolet luminosities on the basis of the tight relation for starburst galaxies .
the results of sections2 and 3 above suggest that these formulae can not be straightforwardly applied to more quiescent galaxies .
we can use the model described in section3 to estimate the accuracy to which can be constrained in galaxies of different types . a way to assess
this is to investigate the relations between , and for a library of models encompassing a wide range of physically plausible star formation histories and dust contents .
we thus generate a library of monte carlo realizations of different star formation histories similar to that used by @xcite to interpret the and index strengths of a complete sample of over @xmath84 sdss galaxies .
the main requirement for the model library is to include broad enough ranges of star formation histories and dust contents to ensure a wide coverage of the diagram .
we have checked that the results presented below do not depend sensitively on the detailed parametrization of the models . for simplicity , we follow @xcite and parametrize each star formation history in terms of two components : an underlying continuous model with a star formation law given by equation ( [ sfr ] ) , and random bursts superimposed on this continuous model . over the interval from 0 to 1 gyr@xmath12 .
random bursts occur with equal probability at all times until .
they are parametrized in terms of the ratio between the mass of stars formed in the burst and the total mass of stars formed by the continuous model over the time .
the ratio is taken to be distributed logarithmically from 0.03 to 4.0 . during a burst ,
stars form at a constant rate for a time distributed uniformly in the range @xmath85@xmath86 years .
the burst probability is set so that 50 per cent of the galaxies in the library have experienced a burst in the past 2 gyr ( see @xcite ) . ] for the purpose of the present analysis , we retain only models with ratios of present to past - averaged star formation rates larger than 0.03 . for the attenuation by dust ,
we adopt a broad distribution of with a peak around 1.5 and a broad distribution in @xmath3 with a peak around 0.3 .
this is consistent with the dust properties derived from the ratios and broad - band optical magnitudes of the above sample of @xmath84 sdss galaxies ( @xcite ; @xcite ) .
we distribute our models logarithmically in metallicity from 0.1 to 2 times solar .
our final library consists of 95,000 different models .
fig.[modelib ] shows as a function of for these models . in the upper panel , we have divided the plane into bins and coded each bin with a different colour to reflect the average ratio of present to past - averaged star formation rates of the simulated galaxies that fall into it . as found in section3 above
, sequences of galaxies with different dust contents corresponding to different @xmath71 lie parallel to the mean relation for starburst galaxies . in the lower panel of fig.[modelib ] , we have coded each bin in the plane with a different colour to reflect the average ultraviolet attenuation for the galaxies that fall into it .
it is remarkable that sequences of models with different star formation histories corresponding to different are nearly perpendicular to the @xmath87-axis in this diagram .
this implies that is a much better estimator of than in galaxies with different star formation histories . to quantify the accuracy to which can be estimated from either or
, we have explored functional fits of in terms of these parameters using the models in our library .
we find that the dependence of on is well approximated by two linear relations with a change in slope at some characteristic turn - around point ( corresponding to @xmath88 ) .
this can be represented by an expression of the form @xmath89 + a_2\ , , \label{eq : irxfit}\end{aligned}\ ] ] where @xmath90 and @xmath91 ( the slope of the relation at small @xmath92 ) , @xmath93 ( the slope of the relation at large @xmath92 ) , @xmath94 ( the normalization constant ) , @xmath95 ( the turn - around point ) and @xmath96 ( the smoothness of turn - around ) are fitted parameters .
the accuracy of the estimates obtained from this formula depends on whether or not independent information is available on the ratio of present to past - averaged star formation rates .
if no information is available on @xmath71 , equation ( [ eq : irxfit ] ) with @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 provides estimates of with a typical ( i.e. rms ) relative uncertainty of 20 per cent for all the models in the library .
[ figbeta ] shows , , and and as a function of for this sample .
it can be seen that , and correlate extremely well with . the correlation is worse for , mainly because of the wider wavelength range of the near - ultraviolet filter of _ uit _ , which extends redward to about 3060 ( table [ uvfilters ] ) .
the mean linear relations between the ultraviolet slopes obtained from observations with different satellites and the reference ultraviolet slope are drawn as solid lines in fig .
[ figbeta ] and are indicated at the bottom of each panel along with the 1@xmath102 standard deviation .
the tight correlations in figs .
1(a)1(c ) allow us to estimate reliably from and .
since the relation is significantly worse for , we do not use _ uit _ data in the analysis described in the main text of the paper .
the ultraviolet spectral slope can also be estimated reliably for galaxies with only far - ultraviolet _ iue _ spectra .
the reason for this is that the wavelength range of the far - ultraviolet spectrograph on - board _ iue _ encompasses 9 out of the 10 spectral windows originally used by @xcite to define .
in fact , for the 110 galaxies in the sample considered here , we find that the ultraviolet slope obtained by fitting a power law to the far - ultraviolet spectra in these 9 continuum bands correlates with as tightly as do and .
the mean linear relation is @xmath103 with a standard deviation @xmath104 .
we also adopt a common definition of the ultraviolet flux for galaxies observed with different satellites , @xmath105 . for those galaxies with _ iue _ spectra
, we compute the mean ultraviolet flux density @xmath106 using the response function of the _ galex _ far - ultraviolet filter ( @xmath25 ) .
for those galaxies observed with _
oao _ and _ ans _ , we compute @xmath106 by extrapolating the broadband spectrum from 1550 ( corresponding to the effective wavelength of the far - ultraviolet filters on - board both satellites ) to 1520 , using the ultraviolet slope ( or ) measured for each galaxy . given the narrow wavelength range over which the extrapolation is performed , this procedure does not introduce any significant uncertainties in and hence . | we compile a new sample of 115 nearby , non - seyfert galaxies spanning a wide range of star formation activities , from starburst to nearly dormant , based on ultraviolet observations with various satellites .
we combine these observations with infrared observations to study the relation between ratio of total far - infrared to ultraviolet luminosities and ultraviolet spectral slope ( the ` ' relation ) .
we show that , at fixed ultraviolet spectral slope , quiescent star - forming galaxies in our sample have systematically lower ratio of total far - infrared to ultraviolet luminosities than starburst galaxies .
the strengths of spectral indices sensitive to star formation history , such as the 4000 spectral discontinuity and the emission equivalent width , correlate well with distance from the mean relation for starburst galaxies in the diagram , while there is little or no correlation between the dust - sensitive ratio and this distance .
this is strong observational evidence that the star formation history is relevant to the ` second parameter ' affecting the relation .
we show that these results can be understood in the framework of the simple model of @xcite for the transfer of starlight through the interstellar medium in galaxies .
we confirm that , for starburst galaxies , the tight relation can be understood most simply as a sequence in overall dust content .
in addition , we find that the broadening of the relation for quiescent star - forming galaxies can be understood most simply as a sequence in the ratio of present to past - averaged star formation rates .
we use a library of monte carlo realizations of galaxies with different star formation histories and dust contents to quantify the accuracy to which the ultraviolet attenuation of a galaxy can be estimated from either the ratio of far - infrared to ultraviolet luminosities or the ultraviolet spectral slope .
we provide simple formulae for estimating as a function of either of these observational quantities and show that the accuracy of these estimates can be significantly improved if some constraints are available on the ratio of present to past - averaged star formation rates .
[ firstpage ] dust , extinction galaxies : general ultraviolet : galaxies |
based on the fact that the interaction between the @xmath0 meson and nucleon is attractive haider and liu postulated the existence of the @xmath0-mesic nuclei @xcite , in which the neutral @xmath0 meson might be bound with nucleons via the strong interaction .
the existence of @xmath0-mesic nuclei allows to investigate interaction of the @xmath0 meson and the nucleons inside a nuclear matter .
moreover it would provide information about @xmath5 resonance @xcite and about @xmath0 meson properties in nuclear matter @xcite , as well as about contribution of the flavour singlet component of the quark - gluon wave function of @xmath0 meson @xcite . according to the theoretical considerations
, the formation of the @xmath0-mesic nucleus can only take place if the real part of the @xmath0-nucleus scattering length is negative ( attractive interaction ) , and the magnitude of the real part is larger than the magnitude of the imaginary part @xcite : @xmath6 calculations for hadronic- and photoproduction of the @xmath0 meson gave a wide range of possible values of the s - wave @xmath0n scattering lenght , from @xmath7=(0.27 + 0.22i ) fm up to @xmath7=(1.05 + 0.27i ) fm .
such a high values has not exluded the formation of @xmath0-nucleus bound states for a light nuclei as @xmath8 , t @xcite and even for deuteron @xcite .
those bound states have been searched in several past experiments @xcite , however none of them gave empirical confirmation of their existence .
there are only a promissing experimental signals which might be interpreted as of the @xmath0-mesic nuclei .
the ongoing investigations continue at cosy @xcite , jinr @xcite , j - parc @xcite , mami @xcite and are planned at gsi @xcite .
the measurement of the @xmath1-@xmath0 bound states is carried out with unique precision by means of the wasa detection system , installed at the cosy synchrotron .
signals of the nuclei are searched for via studying the excitation function of specific decay channels of the system , formed in deuteron - deuteron collision @xcite .
the measurement is performed for the beam momenta varying continously around the @xmath0 production threshold .
the beam ramping technique allows to reduce the systematic uncertainities .
the of the bound system should be manifested as a resonance - like structure in the excitation curve of eg .
@xmath4 reaction below the @xmath9 @xmath1-@xmath0 reaction threshold .
the reaction kinematics is schematically presented in fig .
[ fig1 ] .
-@xmath10 production and decay.,width=321,height=170 ] the deuteron - deuteron collision leads to the creation of the @xmath1-@xmath0 bound system .
the @xmath0 meson can be absorbed by one of the nucleons inside helium and may propagate in the nucleus via consecutive excitation of nucleons to the @xmath11 state @xcite until the resonance decays into the pion - nucleon ( n , @xmath12 or p , @xmath13 ) pair outgoing from the nucleus @xcite .
the relative nucleon - pion angle is equal to @xmath14 in the @xmath15 reference frame and it is smeared by about @xmath16 in the center - of - mass frame due to the fermi motion of the nucleons inside the helium nucleus . in june 2008 a search for the @xmath1-@xmath0 bound state was based on the measurement of the excitation function of the @xmath9 @xmath4 reaction near the @xmath0 production threshold . during the experiment
the deuteron beam momentum was varied continuously from 2.185 gev / c to 2.400 gev / c which corresponds the excess energy variation from -51.4 mev to 22 mev .
excitation function was determined after applying cuts on the p and @xmath13 kinetic energy distribution and the @xmath17 opening angle in the cm system @xcite .
the relative normalization of points of the @xmath9 @xmath4 excitation function was based on the quasi - elastic proton - proton scattering . in the excitation function
there is no structure which could be interpreted as a resonance originating from decay of the @xmath0-mesic @xmath1 @xcite . during the second experiment , in november 2010
, two channels of the @xmath0-mesic helium decay were measured : @xmath2-@xmath3 @xmath4 and @xmath2-@xmath3 @xmath18 @xmath19 @xcite .
the measurement was performed with the beam momentum ramping from 2.127 gev / c to 2.422 gev / c , corresponding to the range of the excess energy . until now
the reaction is analysed .
the @xmath20 is identified in the forward detector based on the @xmath21-@xmath22 method .
the neutral pion @xmath12 is reconstructed in the central detector from the invariant mass of two gamma quanta originating from its decay while the neutron four - momentum is calculated using the missing mass technique .
the excitation function for the @xmath9 @xmath18 @xmath19 reaction is studied for the `` signal - rich '' region corresponding to the momenta of the @xmath20 in the cm system below 0.3 gev / c and the `` signal - poor '' region for the @xmath20 cm momenta above 0.3 gev / c .
the contributions from different background reactions is under investigation .
we acknowledge support by the foundation for polish science - mpd program , co - financed by the european union within the european regional development fund , by the polish national science center through grant no .
2011/01/b / st2/00431 and by the ffe grants of the research center juelich . | the @xmath0-mesic nuclei in which the @xmath0 meson is bound with nucleus via strong interaction was postulated already in 1986 , however till now no experiment confirmed empirically its existence .
the discovery of this new kind of an exotic nuclear matter would be very important for better understanding of the @xmath0 meson structure and its interaction with nucleons .
the search for @xmath0-mesic helium is carried out with high statistic and high acceptance with the wasa - at - cosy detection setup in the forschungszentrum jlich .
the search is conducted via the measurement of the excitation function for the chosen decay channels of the @xmath1-@xmath0 system .
untill now two reactions @xmath2-@xmath3 @xmath4 and have been measured with the beam momentum ramped around the @xmath0 production threshold .
this report includes the description of experimental method and status of the analysis . |
the bose - einstein condensation ( bec ) was realized in atomic traps in 1995 by cooling down @xmath0rb atoms in a magnetic trap @xcite . since then
, experiments have been performed on a variety of ultracold alkali atoms ranging from li to cs.@xcite in addition to displaying new phenomena such as the bec - bcs crossover@xcite and the bosonic mott - transition@xcite , these systems provide also clean and flexible realizations for basic theoretical models such as the hubbard model.@xcite alkali metal isotopes with odd ( even ) number of neutrons behave as fermions ( bosons).@xcite although initial experiments were mainly done on bosonic systems , the degenerate fermi systems have also been realized and confined to optical lattices in recent years due to advanced sympathetic cooling techniques .
@xcite because of the pauli exclusion principle , these fermionic atoms can display a variety of interesting phenomena and phases that do not have an analog in bosonic systems.@xcite in the remainder of the paper we shall focus on fermionic systems of three internal degrees of freedom and show how a quantum phase transition appears in this system , which is a simplified version of the color superconductor - baryon phase transition in quantum chromodynamics ( qcd ) .
typical hyperfine couplings are larger than the standard experimental temperatures used to study ultracold gases . in the absence of an external magnetic field , the hyperfine coupling aligns nuclear ( @xmath1 ) and electronic ( @xmath2 ) spin antiparallel to each other , and the hyperfine spin @xmath3 is conserved . in finite magnetic fields , only the hyperfine spin @xmath4 along the external field is a good quantum number .
the hyperfine spin @xmath4 thus provides an internal quantum number that we shall refer to as `` color '' henceforth .
systems with three internal quantum numbers are of special interest , since they are rarely observed in solid state physics .
such a three - component fermionic system may be created , e.g. , by trapping the lowest three hyperfine levels of @xmath5li atoms in all - optical setups in large magnetic fields ( see fig . [
fig : li - hf ] ) . such three - component systems with weak interactions
have been studied first in ref . , where it has been shown that for small attractive interactions , a color superfluid state emerges .
this work has been generalized to incorporate three - body correlations at large attractive interaction strengths in ref . .
at the same time , results for the three - fermion problem in a single parabolic well and a mean field calculation to describe a two - component bec - bcs crossover through a feshbach resonance in @xmath5li appeared.@xcite in this paper , we shall study three - color fermionic systems in an optical lattice .
optical lattices are realized by standing light waves , which create a periodic potential for the trapped atoms.@xcite if the amplitude of the lasers is strong enough , then the atoms are localized to the minima of this potential and at low temperatures can only move by tunneling between the lowest lying states within each such minimum .
( color online ) sketch of the lowest lying hyperfine levels of @xmath5li in external static magnetic fields .
the high - field seeker states have negative tangent , which can not be trapped in magnetic traps .
however , using all - optical setups , one can trap the hyperfine states denoted as @xmath6 , and @xmath7 , and thus creating a three - component quantum degenerate fermionic system .
, width=151 ] the trapped atoms , however , also interact with each other .
the dominant interactions between alkali atoms are short - ranged electrostatic van der waals forces , which can be well approximated by @xmath8 potentials at low energy scales .
if the scattering length of the interacting atoms is smaller than the lattice constant of the optical lattice , then these interactions are restricted to a given lattice site .
correspondingly , the interacting system of atoms in an optical lattice can be accurately described by the following hamiltonian : @xmath9 with @xmath10 the creation operator of a fermionic atom of color @xmath11 at site @xmath12 , and @xmath13 . in the tunneling term , @xmath14 implies restriction to nearest neighbor sites , and the tunneling matrix element is approximately given by @xmath15 , where @xmath16 is the recoil energy , @xmath17 is the wavevector of the lasers , @xmath18 is the mass of the atoms , @xmath19 , and @xmath20 is the depth of the periodic potential.@xcite we neglect the effects of the confining potential in eq . ( [ eq : h ] ) , which would correspond to a site - dependent potential term in the hamiltonian .
the interaction strength @xmath21 between colors @xmath22 and @xmath23 is related to the corresponding @xmath24-wave scattering length , @xmath25 , as @xmath26.@xcite note that fermions with identical colors do not interact with each other . for the sake of simplicity
, we shall first consider the attractive case with @xmath27 .
this case could be realized by loading the @xmath5li atoms into an optical trap in a large magnetic field , where the scattering lengths become large and negative @xmath28 , for all three scattering channels , @xmath29 , @xmath30 , and @xmath31.@xcite introducing the usual gell - mann matrices , @xmath32 ( @xmath33 ) , it is easy to see that global su(3 ) transformations @xmath34 commute with the hamiltonian , which thus also conserves the total number of fermions for each color , @xmath35 .
this conservation of particles is only approximate because in reality the number of the atoms in the trap continuously decreases due to different scattering processes . here
, however , we shall neglect this slow loss of atoms and keep the density @xmath36 of atoms for color @xmath22 as well as the overall filling factor @xmath37 fixed .
( color online ) the ground states for @xmath38 and @xmath39 can be calculated by perturbation theory .
the former is a bcs - state which breaks the su(3 ) symmetry , and the latter is a trionic state with three - particle singlet bound states.,width=226 ] let us first focus on the case of equal densities , @xmath40 . for small attractive @xmath41 , the ground state is a color superfluid:@xcite atoms from two of the colors form the cooper pairs and an @xmath24-wave superfluid , while the third color remains an unpaired fermi liquid .
however , as we discussed in ref . , for large attractive interactions
, this superfluid state becomes unstable , and instead of cooper pairs , it is more likely to form three - atom bound states , the so - called trions " .
these trions are color singlet fermions , and for large @xmath42 they have a hopping amplitude , @xmath43 furthermore , one can easily see that if two trions sit on neighboring lattice sites , then they increase the energy of each other by an amount @xmath44 .
this is because the energy of an individual trion is decreased by quantum fluctuations where one of the atoms virtually hops to one of the neighboring sites .
these quantum fluctuations are reduced if the two trions sit next to each other .
therefore , trions will tend to form a fermi - liquid in any finite dimensions .
this fermi liquid state may be further decorated by charge density wave order at large values of @xmath42 .
also , the fermi liquid scale @xmath45 of the trionic fermi liquid should depend on the value of @xmath46 , and at the transition point , @xmath47 , we expect it to go zero ( see figs .
[ fig : pert ] and [ fig : sketch ] ) . in order to get analytic expressions
, we shall study the ground state in @xmath48 dimensions .
then , to reach a meaningful limit and to get finite kinetic energy , one has to scale the hopping as @xmath49 , with @xmath50 fixed . in this limit , however , trions become immobile . therefore the @xmath51 trionic states are well approximated as @xmath52 where @xmath53 denotes a subset of sites where trions sit .
we can calculate the energy of this state in infinite dimensions : a single trion has an energy @xmath54 , thus the energy of such a state per lattice site is given by @xmath55 , with @xmath56 the total energy of the system and @xmath57 the number of lattice sites .
( color online ) schematic phase diagram of the attractive su(3 ) hubbard model .
the color superfluid phase is stable below a critical temperature @xmath58 , where the appearance of superfluid domains is expected . at large interaction strengths @xmath59
, color singlet trions emerge instead of the cooper pairs and form a fermi liquid at low temperatures @xmath60 .
, width=226 ] clearly , the two ground states obtained by the perturbative expansions have different symmetries : the superfluid state breaks su(3 ) invariance , while the trionic state does not . therefore , there must be a phase transition between them .
note that , relying on symmetries only , this argument is very robust and carries over to _ any _ dimensions . in infinite dimensions
, we find that trions are immobile .
however , this is only an artifact of infinite dimensions and in finite dimensions , a superfluid - fermi liquid phase transition should occur .
one could envision that some other order parameter also emerges and masks the phase transition discussed here .
preliminary results ( not discussed here ) suggest that indeed a charge density state forms at large values of @xmath42 , but except for half filling , which is a special case not discussed here , we do not see any other relevant order parameter that could intrude as a new phase .
one could , in principle , also imagine a phase with simultaneous trionic and fermionic fermi surfaces , similar to the one of ref .
however , in contrast to the calculations of ref .
( which do not take the local constraints on the lattice into account ) , in our scenario , there is no fermi surface at the quantum critical point . from the above argument
, it is unclear whether the phase transition is of first or of second order . to address this question in infinite dimensions and give some more quantitative estimates for the relevant condensation energies and the critical value of @xmath61
, we need an approach that is able to describe the superfluid state and , at the same time , also accounts for three - body correlations . in the present paper ,
we construct a variational gutzwiller wave function that is able to capture both correlations simultaneously .
we then show how averages can be evaluated by constructing an effective action that contains three - body correlations and how calculations can be analytically done in infinite dimensions by using the methodology of dynamical mean field theory.@xcite our method , which uses a single suitably chosen gutzwiller correlator and exploits the cavity method of dynamical mean field theory , is in its present form less flexible than the multiband method reviewed by bnemann _ _ et . al.,__@xcite but it incorporates three - body correlations in a very transparent way .
our analysis shows that , within the gutzwiller approach , the phase transition is of _ second order _ in infinite dimensions .
we therefore expect it to remain of second order in any finite dimension above the lower critical dimension .
nevertheless , as we discuss later , the quantum phase transition may become of slightly first order due to a not perfectly su(3)-symmetrical interaction .
interestingly , our results also suggest that there is a tendency to create an imbalance of the densities @xmath36 in the superfluid phase .
the physical reason is simple : one can gain condensation energy by transferring atoms from the unpaired channel to the paired ones and thereby create _
ferromagnetic order _ as a secondary order parameter . for equal densities
@xmath62 and color conservation , his can only happen if the atoms are segregated and domains are formed , as shown in fig .
[ fig : sketch ] , where we sketched the schematic phase diagram of the attractive su(3 ) model away from half - filling .
this picture , which is first proposed in ref . , has also been confirmed by the ginzburg landau - type effective field theoretical approach of ref . .
so far , we discussed the fully su(3 ) symmetrical case . as we also demonstrate later , the phase transition discussed above is not sensitive to having perfect su(3 ) symmetry . on the other hand , to form the superfluid phase , it is important to have approximately the same fermi momenta for at least two colors .
the rest of the paper is organized as follows . in sec .
[ sec : gutzw_ans ] , we introduce the gutzwiller ansatz for the ground state . in sec .
[ sec : eff_theo ] , we reformulate the gutzwiller expectation values as an effective path integral . in sec .
[ sec : cavity ] , we derive a local action , which can be used to solve the effective action in the @xmath63 dimensional limit . in sec . [
sec : solution ] , we first summarize the results for the su(3 ) symmetric case and then generalize the approach by analyzing the hamiltonian with nonuniform interaction strengths in order to describe a system of @xmath5li atoms . in sec .
[ sec : qcd ] , we present a brief discussion of analogies with qcd . in sec .
[ sec : conclusions ] , we present our conclusions . some of the technical details can be found in the appendices .
to capture the color superfluid - trion transition , we approximate the ground state of the infinite - dimensional system by the following gutzwiller - correlated wave function : @xmath64 \vert { \rm bcs } \rangle \;.
\label{eq : def_gutzwillerstate}\ ] ] here @xmath65 is a gutzwiller variational parameter , which increases ( or decreases ) the amplitude of terms in the `` uncorrelated - state '' which have triple occupancies . in the @xmath66 limit
, @xmath67 becomes a superposition of trionic states similar to eq .
( [ eq : def_trionic_state ] ) .
we choose the uncorrelated ground state as a bcs - like state with colors 1 " and `` 2 '' forming cooper pairs,@xcite @xmath68 the operators @xmath69 in eq .
( [ eq : def_bcsstate ] ) diagonalize the first part of the hamiltonian ( [ eq : h ] ) , and create fermions with momentum @xmath70 , color @xmath22 , and energy @xmath71 , with the vector @xmath72 running over nearest neighbor sites . the coefficients @xmath73 and @xmath74 are the usual bcs coherence factors , with @xmath75 .
the `` chemical potentials '' @xmath76 and @xmath77 appearing in the wave function should be considered merely as parameters that are adjusted to fix the densities @xmath36 at a given value . to perform a variational calculation
, we need to evaluate the gutzwiller expectation value of hamiltonian ( [ eq : h ] ) , @xmath78 at a given filling factor @xmath79 , and then minimize it with respect to @xmath65 , @xmath80 , and eventually the density of the third color , @xmath81 .
we note that our variational wave function smoothly interpolates between the color superfluid @xmath82 and the trionic state @xmath83 .
we can also compare the gutzwiller energy to certain reference states , such as the fermi sea ( @xmath84 , @xmath85 ) , the gutzwiller correlated fermi sea ( @xmath86 , @xmath85 ) , or the uncorrelated bcs state ( @xmath84 , @xmath87 ) , and thereby estimate correlation or condensation energies .
we first derive an effective action that can be used to replace the gutzwiller expectation value of an operator @xmath88 , @xmath89 by a combination of the grassmann path integrals . for this purpose ,
let us first rewrite the denominator of eq .
( [ eq : def_gutzwiller_expval ] ) as @xmath90 \vert { \rm bcs } \rangle .
\label{eq : normalorder_gg}\ ] ] here , the operator in the middle is in a normal ordered form ( i.e. , all @xmath91 appear to the left of the operators @xmath92 ) . since the bcs state is a non - interacting state ( in terms of bogoliubons )
, we can use wick s theorem to evaluate expectation values appearing in eq .
( [ eq : normalorder_gg ] ) . due to the normal
ordering , every @xmath93 and @xmath92 appears in eq .
( [ eq : normalorder_gg ] ) only once .
therefore , we can calculate the quantum mechanical expectation values as the grassmann path integrals , where we replace normal ordered operators by the grassman variables as @xmath94 and @xmath95 and evaluate expectation values with an appropriately chosen action @xmath96 , @xmath97^{-1}_{ij } \psi_j .
\end{aligned}\ ] ] here , we introduced the `` nambu spinors '' , @xmath98 , and the propagator @xmath99 must be chosen so that it satisfies the conditions , @xmath100 with @xmath101 and @xmath102 the normal and anomalous green s functions .
it is easier to express @xmath103 in the fourier space , where it is just a @xmath104 dimensional matrix in the nambu space that can be expressed as @xmath105 we can thus rewrite @xmath106 as @xmath107 \right\rangle_{{\cal s}_0 } \;.
\label{eq : grassman_0}\ ] ] note that in the above procedure , we have not doubled the hilbert space since the integration is over the fields , @xmath108 and @xmath109 , and therefore the @xmath110 and @xmath111 do not represent independent grassman variables .
note also that the green s functions @xmath112 and @xmath113 can easily be determined in terms of the bcs coherence factors .
furthermore , neither the green s functions nor the grassman fields @xmath114 have a time argument .
these grassman fields should thus not be confused with the grassman fields that appear in the path integral expression of the density matrix or the time evolution operator . to make further progress ,
let us define the `` triple occupancy '' as @xmath115 . using the properties of the grassmann variables ,
this quantity can be expressed as a function of the grassmann vectors defined above @xmath116 where @xmath117 , and the third pauli matrix @xmath118 acts in nambu space . clearly , @xmath119 .
therefore , exponentiating the product in eq .
( [ eq : grassman_0 ] ) , we obtain the relation @xmath120 where the full ( interacting ) effective action is given by @xmath121_{ij } \psi_j - ( g^2 - 1 ) \sum_i t_i \;. \label{eq : def_action}\ ] ] interestingly , the gutzwiller variational parameter appears in this effective action as a three - body interaction . for the uncorrelated wave function , @xmath84 , the interaction term vanishes . for trionic correlations one has @xmath122 , i.e. , the effective three - body interaction is attractive , while for @xmath123 the effective interaction is repulsive .
this procedure can be repeated with certain operator s quantum mechanical expectation values after proper normal ordering . for local operators
@xmath124 , the gutzwiller expectation value is @xmath125 \vert { \rm bcs } \rangle}/{\langle g \vert g \rangle } \;,\nonumber \end{aligned}\ ] ] where the gutzwiller normal ordered operator @xmath126 is defined as follows : @xmath127\hat o_i [ 1+(g-1)\hat t_{i}]:\ ] ] and @xmath128 . here
, the numerator can also be converted into expectation values evaluated with @xmath129 : we first use wick s theorem and replace all operators by the grassman variables and the average by @xmath130 .
then , we can insert the missing term @xmath131 $ ] and reexponentiate the product to have an average with the action @xmath129 . with this procedure , the particle density and double occupation on a site can be written as follows : @xmath132 to evaluate the expectation value of the kinetic energy , we also need to evaluate expectation values of the type @xmath133 for @xmath134 .
we therefore define the density supermatrix ( formulas here are valid for @xmath135 ) , @xmath136 with the @xmath137 matrices @xmath138 and @xmath139 defined as @xmath140 and @xmath141 in these expressions , we defined the grassman factors @xmath142 .
clearly , for @xmath84 , the matrix @xmath143 is simply related to the unperturbed green s function , @xmath144^t $ ] .
as we shall now see , although the above expressions look rather complicated and contain formally three - body correlation functions in terms of the effective field theory , we shall still be able to express @xmath145 in terms of the full single particle green s function @xmath146 , defined as @xmath147 and its proper self - energy @xmath148 . at this point
, it is useful to define the generating functional , @xmath149 where the `` current '' @xmath150 is a six - component grassmann vector , @xmath151 , and make use of standard field theoretical methods.@xcite as usual , the components of the green s function can be obtained from @xmath152 by functional derivation , @xmath153 the ( improper ) self energy @xmath154 is defined by cutting off the bare lines from the dressed propagator : @xmath155 while the proper ( one particle irreducible ) self - energy obeys the dyson s equation , @xmath156 it is useful to visualize these self - energies in terms of the feynman diagrams of the effective field theory , as shown in fig .
[ fig : self - energy ] , although we shall sum up these diagrams later by using nondiagrammatic methods .
as mentioned earlier , the gutzwiller correlator generated a three - body interaction in the effective field theory , while anomalous green s functions appear due to the superfluid correlations .
examples of first and second order self - energy diagrams .
continuous lines denote the unperturbed normal green s functions , the staggered lines the anomalous ones , while vertices stand for the three - body interaction , @xmath157 . in infinite dimensions
, we shall sum up these diagrams by using a _
method.,width=226 ] to generate the necessary identities to evaluate the expectation values in eqs .
( [ eq : p_ij ] ) and ( [ eq : q_ij ] ) , we shift the grassmann fields @xmath158 and @xmath159 in the generating functional @xmath152 as @xmath160 where @xmath161 is a real valued parameter . note that this equation also determines how the field @xmath162 is changed , since the components of @xmath162 and @xmath163 are related .
proceeding with the functional derivation after this change of integration variables leads to an expression of the dressed propagator @xmath146 in terms of expectation values of the grassmann fields related to the expectation values in eqs .
( [ eq : p_ij ] ) and ( [ eq : q_ij ] ) ( for details , see appendix [ app : identities ] . ) , and finally comparison of these with the definition of the self - energy gives the following identities ( @xmath164 ) : @xmath165 where @xmath166 .
using these identities , it is possible to express the density supermatrix @xmath145 in terms of the improper self - energy .
this relation is more transparent in the fourier space , where it reads @xmath167 where the superscript @xmath168 refers now to the transposed matrix in color space , and @xmath169 is a @xmath70-independent @xmath104 matrix related to the @xmath170 contributions .
this term does not play a role in the evaluation of the kinetic energy , @xmath171 since we assumed only nearest neighbor hopping , and therefore @xmath172 .
in the previous section , we showed that the gutzwiller expectation values can be expressed by path integrals , which still need to be computed .
these path integrals can not be exactly evaluated in general .
however , an important simplification occurs in the limit of infinite dimensions , @xmath173 .
there the off - diagonal ( @xmath135 ) green s functions decay as @xmath174 , where @xmath175 denotes the minimum number of steps that are needed to reach the lattice point @xmath12 from the other lattice point @xmath176.@xcite therefore , the off - diagonal ( @xmath135 ) components of the proper self energy rapidly go to zero , @xmath177 , and vanish in the infinite - dimensional limit , where the self - energy becomes local , @xmath178 here , we already used the discrete translational symmetry of the lattice : @xmath179 .
as @xmath148 is a central quantity of interacting systems and allows one to compute virtually any other quantity , our main goal shall be to determine @xmath148 .
this step is done similar to the cavity method also used in dynamical mean field theory to extract local properties at a mean field level.@xcite in this approach one integrates out the grassmann variables on all sites except for the origin , @xmath180 where the prime indicates that one should not integrate over the variables @xmath181 and @xmath182 at the origin .
thus , by construction , the local action depends only on the variables @xmath183 , @xmath184 $ ] , and all local expectation values and correlation functions are invariant , @xmath185 importantly , the local green s function is also invariant .
since the hopping is scaled as @xmath186 , it can be shown ( appendix [ app : localaction ] ) that in infinite dimensions , the generated local action takes on a simple functional form , @xmath187 where @xmath188 denotes the `` cavity green s function '' .
the field @xmath189 in eq .
( [ eq : s_l ] ) denotes the grassmann field @xmath190 in the origin , @xmath191 .
the cavity green s function @xmath188 can be determined selfconsistently from the condition that eq .
( [ eq : selfcons000 ] ) holds for any local quantity , including the local dressed propagator of the effective lattice theory , @xmath192 which , by construction , must coincide with the full green s function of the cavity theory , @xmath193 then , on one hand , the local propagator on the lattice can be determined by using dyson s equation and thus can be expressed in terms of @xmath148 , as @xmath194^{-1}\;. \label{eq : dyson_local_latt}\ ] ] on the other hand , we can also compute the local propagator _ exactly _ by evaluating the path integral with eq .
( [ eq : s_l ] ) analytically : using symmetry considerations presented in appendix [ app : diagonal ] , we can show that there is a @xmath195 unitary transformation that diagonalizes @xmath188 : @xmath196 where @xmath197 is a @xmath137 ( real ) diagonal matrix .
we can then evaluate the dressed propagator in the local theory by performing a variable transformation @xmath198 in the intergal , the result being @xmath199 furthermore , dyson s equation also holds for the local theory , and moreover in infinite dimensions , the proper self - energy @xmath148 of the lattice theory is the same as in the self - energy in the local theory .
this follows from the comparison of the skeleton diagrams of the local and lattice theories , and the fact that the full green s functions are the same in both theories .
we can thus write @xmath200 with some algebraic manipulations , eqs .
( [ eq : dyson_local_latt ] ) and ( [ eq : dyson_local ] ) can be recast in the following self - consistency condition , @xmath201^{-1 } \;. \label{eq : selfcons_equations}\ ] ] this is the central equation of the theory . having solved this equation self - consistently for the self - energy matrix @xmath148 , we can express , e.g. , the cavity green s function as @xmath202 and then determine from that the local green s function using eq .
( [ eq : sc ] ) or the improper self - energy from the fourier transform of eq .
( [ eq : sigma_s ] ) .
remember though that the quantities here are @xmath104 matrices and inversions mean matrix inversions . setting all these together and using the relation [ eq .
( [ eq : calp_s ] ) ] between the density matrix @xmath203 and the improper self - energy , the kinetic energy can be finally expressed as @xmath204^{-1 } \nonumber \\ & & - \frac{2}{1+g } [ d^{0 - 1}({{\bf k } } ) - \sigma]^{-1 } \sigma \nonumber \\ & & + \frac{1}{(1+g)^2 } d^{0 - 1}({{\bf k } } ) [ d^{0 - 1}({{\bf k } } ) - \sigma]^{-1 } \sigma \big ] _ { \alpha\alpha } , \nonumber \end{aligned}\ ] ] where the summation runs only over the first three diagonal elements of the @xmath104 matrices .
we can also compute other local expectation values using the same variable transformation [ eq .
( [ eq : u ] ) ] as before .
the total particle density is is given by @xmath205 while the total double occupancy is @xmath206 this is a good point to discuss the similarities and differences with the multiband gutzwiller method of bnemann _ _ et .
al.__@xcite that approach employs a flexible gutzwiller correlator with multiple variational parameters .
some of these parameters are allowed to be fixed so that the proper self - energy of the effective theory vanishes in infinite dimensions .
this leads to important simplifications when calculating the gutzwiller expectation values . in our formalism , on the other hand
, we used a single correlation parameter that was motivated by weak and strong - coupling theories .
nevertheless , this single correlator is able to capture the physics of both limits and it therefore accounts for the most important correlations . including additional ( e.g. , two - body correlators ) should quantitatively modify our results , and change the critical value of @xmath46 , however , we do not expect qualitative changes due to them .
the advantage of our effective field theory approach is that it displays the trionic correlations in a rather transparent way , in terms of three - body interactions .
preliminary results indicate that our cavity functional approach could also be modified to use more flexible correlators , but this is out of the scope of the present paper .
the self - consistency equations are hard to solve in general .
however , we can simplify their solution by observing that the proper self - energy must have certain symmetries .
we therefore need to use only three independent parameters to parametrize the self - energy , @xmath207 with @xmath208 and @xmath118 denoting the second and the third pauli matrices in the nambu space . using this matrix ,
it is possible to solve selfconsistency relations ( [ eq : selfcons_equations ] ) numerically , and we can also calculate the energy , @xmath209 where @xmath57 is the number of lattice sites and we explicitely indicated all implicit variables , with respect to which the energy must be optimized . the kinetic energy of the first two colors is @xmath210 ^ 2 } { ( 1-\sigma_1)(1+f_{{\bf k}}^2)},\ ] ] where @xmath211 is a renormalized occupation number , with @xmath212 . the kinetic energy of the third color can be expressed as @xmath213 ^ 2}{1-\sigma_2 } k_0 \;,\ ] ] where @xmath214 is the noninteracting kinetic energy of the atoms in channel 3 .
finally , the full double occupancy is given by @xmath215 notice that the above momentum sums as well as the summations included in the self - consistency equation contain terms that depend on the momentum only through the single particle energy @xmath216 .
it is therefore possible to turn the momentum sums into energy integrals and convert all these expressions into relatively simple self - consistent integral equations , involving only the density of states @xmath217 , which becomes the gaussian on an infinite - dimensional cubic lattice , @xmath218 for high dimensional hypercubic lattices , the qualitative features of the densities of states are the same as in infinite dimensions . in this regard three dimension
is probably close enough to infinite dimensions , and our results should hold away from half filling .
one- and two - dimensional systems should be investigated by different methods since there spatial fluctuations and van hove singularities will play a much more important role .
nevertheless , even in one dimension , recent bethe ansatz@xcite and density matrix renormalization group@xcite calculations seem to support the results presented here and in ref . .
our numerical procedure is as follows .
first , we fix the filling factor @xmath79 and the interaction strength @xmath46 . for a given set of variational parameters @xmath65 , @xmath80 , and @xmath81
, we use the ansatz ( [ eq : ansatz ] ) and solve the selfconsistency relations [ eq . ( [ eq : selfcons_equations ] ) ] by iteration.@xcite once @xmath148 at hand , we can compute the expectation value of the energy @xmath219 using eqs .
( [ eq : energy])-([eq : df ] ) .
this function can then be minimized numerically to find the optimum values for @xmath65 , @xmath80 , and @xmath81 .
typical results for a given density @xmath220 are shown in fig .
[ fig : landscape ] , where the energy landscape @xmath221 is shown as a function of @xmath65 and @xmath80 for various values of the coupling @xmath46 . for small values of @xmath42 ,
the overall energy minimum is located at some finite value of the correlation parameter @xmath122 and gap @xmath80 .
this energy minimum is shifted to larger and larger values of @xmath65 as we increase @xmath42 .
the optimum value of @xmath80 initially increases with @xmath42 , but at larger values of @xmath42 , it starts to decrease again , the energy minimum is getting shallower and shallower and , eventually , it shifts to @xmath222 and @xmath85 as @xmath46 approaches a critical value @xmath223 .
this behavior of the optimum values of @xmath65 and @xmath80 is similar to the one shown in fig .
[ fig : secondorder ] . within numerical accuracy ,
the gap vanishes continuously at @xmath47 , which is characteristic of a _ second order _
phase transition , i.e. , a quantum critical point .
the correlation parameter @xmath65 precisely diverges at the same value of @xmath46 , implying that for larger values of @xmath42 the optimum ground state is a purely trionic state with no superconducting order .
as we discuss below , this scenario carries over for all filling factors @xmath224 that we investigated , where there is always an interaction strength @xmath225 at which the gutzwiller parameter @xmath65 diverges .
( color online ) variational energy @xmath221 compared to the energy of the trionic state , @xmath226 .
only regions of @xmath227 are shown for different values of the interaction strength and @xmath228 . , width=226 ] in the previous calculation , we fixed all three densities to be the same .
we have done this in the spirit that in an optical lattice , the total number of particles is fixed for each color .
however , one can decrease the energy of the system further in the superconducting phase by letting @xmath81 vary and only fixing the total filling fraction @xmath79 . as shown in fig .
[ fig : n_3_dependence ] , the optimal value of the average occupation of the third color is slightly less than the filling : @xmath229 .
however , as shown in the inset of fig .
[ fig : n_3_dependence ] , @xmath81 approaches @xmath79 as @xmath46 approaches @xmath223 .
this is easy to understand : the reason of the formation of an imbalance is that one can gain condensation energy by transferring particles from color 3 to colors 1 and 2 .
however , the driving force to create this imbalance is the superconducing condensation energy .
this energy does not coincide with @xmath80 , but is rather defined as the difference between the the energy of the state with @xmath230 and the energy of the gutzwiller - correlated state with @xmath231 , @xmath232 @xmath233 this energy is also shown in fig .
[ fig : secondorder ] and it vanishes at the critical point @xmath47 as well .
therefore , the induced charge imbalance must also disappear there , in agreement with our numerical findings .
( color online ) different ground state properties of the su(3 ) attractive hubbard model for @xmath234 . for @xmath235 ,
the gutzwiller parameter @xmath65 diverges , while the superfluid order parameter @xmath80 and the condensation energy @xmath236 vanish .
energy is measured in units of @xmath237.,width=283 ] ( color online ) one can gain energy by transferring particles with color 3 to the superfluid channel : the energy minimum occurs for @xmath229 .
the inset shows the optimal @xmath238 for @xmath239 .
energy is measured in units of @xmath237 .
, width=283 ] in a system with su(3 ) symmetry , this finding implies that , although the total number of particles is fixed for a given color , one can decrease the energy of the system by segregation,@xcite i.e. , by forming superconducting domains with unequal numbers of particles and orienting the order parameters to point in them in different directions , as we already sketched in fig .
[ fig : sketch ] . of course
, this is only possible in a large enough system , where the domain wall energy is compensated by the overall condensation energy gain .
interestingly , the tendency to generate @xmath240 can also be viewed as the appearance of a secondary ferromagnetic order parameter , @xmath241 therefore we conclude that this superconducting state is also a ferromagnet in the su(3 ) language .
so far , we assumed that the scattering lengths are the same in all three scattering channels . in most systems , however , the three scattering lengths are not equal and can vary with an external magnetic field . as we proposed in ref .
, a possible candidate for realizing the trionic state is @xmath5li . in this system
, the magnetic field dependence of the three scattering lengths has been experimentally determined .
@xcite according to the experimental results of ref .
, the interaction strengths can be approximated in the magnetic field region of 60 120 mt as @xmath242\left[1 + \alpha^{\beta\gamma } ( b - b_0^{\beta\gamma})\right ] \ ; , \label{eq : ualphabeta}\ ] ] where the parameters @xmath243 , @xmath80 , and @xmath22 in this equation can be taken from ref .
, and @xmath244 , @xmath245 , and @xmath246 . the interaction amplitude @xmath247
can be changed by tuning the potential depth . for @xmath248 ,
all three interactions become approximately the same , and are negative , @xmath249 . on the other hand ,
if we increase the interaction strength by approaching the feshbach resonance at @xmath250 , then the interaction becomes rather anisotropic in color space .
experimentally , there are thus several ways to drive the system close to the phase transition regime , @xmath251 .
the first possibility is to change the intensity of the laser beams , and thereby tune mostly the hopping amplitude @xmath237 . in this case
, one can apply a large magnetic field in which the interaction is almost su(3 ) symmetrical , although it is somewhat stronger in channel 12 than in the others .
the other possibility is to tune the ratio @xmath252 by changing the magnetic field and approaching a feshbach resonance . in this second case
, the interactions can strongly break su(3 ) symmetry . in both cases
, the anisotropic interaction has an important effect : it locks the superconducting order parameter so that only @xmath253 .
the color superfluid phase thus becomes a more or less standard @xmath254 superfluid with an additional decoupled fermi liquid .
however , as we show , the trionic phase transition can survive in this anisotropic limit .
( color online ) ground state properties of the hamiltonian ( [ eq : h ] ) with the interaction strengths ( [ eq : ualphabeta ] ) for @xmath255 as a function of @xmath256 for @xmath234 filling , @xmath247 .
this describes a cold atomic system where three hyperfine states of @xmath5li are trapped and the magnetic field is tuned .
the inset shows the same quantities with the magnetic field fixed at @xmath257 as a function of @xmath258 .
energy is measured in units of @xmath237.,width=283 ] let us now investigate the superfluid - trion transition , assuming that @xmath259 . in this region , all interactions are attractive , @xmath260 .
perturbation theory tells us that the weak - coupling ground state is the color superfluid state @xcite formed in the channel 12 , and in the strong coupling limit the ground state is a trionic state .
thus , we can use the same type of ansatz as in the su(3 ) case . since the structure of the effective theory
is uniquely determined by the gutzwiller ansatz state , the effective action ,
the ward identities and the self - consistency relations for the local effective action remain the same as before .
the only difference is in the expression of the variational energy that we need to minimize , @xmath261
\;. \nonumber \end{aligned}\ ] ] the previously used numerical procedure can be modified for this case , and we find that the quantum phase transition persists even with the anisotropy in the interaction strengths .
the anisotropy in @xmath21 has an important secondary effect : it induces somewhat different chemical potentials in channels 1 and 2 , i.e. , an su(3 ) `` zeeman field '' . deep in the superfluid phase
, this chemical potential difference is not sufficient to create a charge imbalance , @xmath262 , since breaking cooper pairs requires a finite energy.@xcite in principle , close to @xmath223 , @xmath263 , and therefore , the chemical potential difference could create a first order transition to the trionic state .
this would invalidate our restriction @xmath264 in our calculations .
however , at the critical point trionic correlations also diverge , @xmath66 , and therefore the effects of this zeeman field are suppressed .
in fact , within the gutzwiller approach , the phase transition seems to be continuous even for anisotropic interactions ( see fig .
[ fig : changeu0 ] ) .
in finite dimensions , on the other hand , we can not exclude a first order transition . in fig .
[ fig : changeu0 ] , we show different physical quantities computed for the ground state as a function of @xmath265 at @xmath234 and @xmath266 .
the observed phase transition occurs at @xmath267 , and all parameters behave similarly to as in the su(3 ) symmetrical case .
the phase transition we found is essentially the analogue of the color superconductor - baryon phase transition,@xcite which is believed to occur in qcd . to make this analogy with qcd clearer ,
let us rewrite the partition function of the su(3)-symmetrical hamiltonian in a path integral form , @xmath268 } , \nonumber \\
{ \cal s}&= & \sum_{ij \alpha } \bar c_{i\alpha}(\tau){\cal g}^{-1}_{ij,\alpha}(\tau ) c_{j\alpha}(\tau ) \nonumber \\ & & + \frac u 2 \sum_i \sum_{\alpha \beta } n_{i\alpha}(\tau ) n_{i\beta}(\tau ) , \end{aligned}\ ] ] where @xmath269 denotes the inverse imaginary time green s function for the non - interacting atoms and @xmath270 the hopping between sites @xmath12 and @xmath176 .
we can now decouple the interaction term by using the relation @xmath271 directly following from the su(3 ) identity , @xmath272 and then performing a hubbard stratonovich transformation : @xmath273 , \nonumber\end{aligned}\ ] ] at every lattice and imaginary time point .
here we have suppressed the imaginary time and color indices of the grassmann variables @xmath274 and the `` gluon field '' , @xmath275 , and we introduced the coupling @xmath276 between the gluons and the fermions . using this transformation we can thus express the partition function as @xmath277\right ] , \nonumber \\
{ \cal s}_{gl } & = & \sum_{ij\alpha } \bar c_{i}\ ; { \cal g}^{-1}_{ij } c_{j } + \sum_{i , a}(a_i^a)^2 + i\;\gamma\ ; \sum_{i , a } a^a_i \;(\bar c_{i}\lambda^a c_{i } ) .
\nonumber\end{aligned}\ ] ] in this language , the attractive interaction is mediated by real bosonic fields , `` gluons '' , which are ultimately responsible for the formation of the trionic ( `` baryonic '' ) phase .
note , however , that in our case , gluons do not have a dynamics on their own , rather they are just massive hubbard - stratonovich fields . as a result ,
the mediated interaction is short - ranged , and there is no confinement .
in fact , our phase diagram parallels the famous qcd phase diagram . in qcd , however , one changes the ratio of kinetic energy vs interaction energies through changing the chemical potential , while in our case , the hopping parameter ( i.e. , the mass of the fermions ) is changed .
therefore , large chemical potentials in the qcd correspond in our case to large values of @xmath237 ( i.e. , small @xmath42 ) , while small chemical potentials in qcd correspond to small @xmath237 ( large @xmath42 ) .
another important difference is the order of the phase transition : in our case , the two phases are separated by a quantum critical point , while in qcd , the transition is of first order , due to the long - ranged interactions .
moreover , the superconducting phase we have is somewhat different from the one occurring in qcd , since cold atoms in optical lattices have only a color quantum number , while quarks in qcd have additional flavor degrees of freedom .
the color superfluid emerging in our case has a nontrivial su(3 ) color content and is analogous to the color superconducting phase in two - flavor qcd where only two flavors of light quarks are considered .@xcite in the alternative theoretical scenario of three - flavor qcd , the superconducting state is expected to be color - flavor locked.@xcite
in this paper , we investigated a system of fermionic cold atoms with three internal quantum states and attractive interactions among them .
we have shown by heuristic arguments as well as by a detailed variational calculation in the limit of @xmath278-dimensions that this system shows a quantum phase transition from a correlated superconducting state to a trionic state , where fermions form three - body bound states and the superconducting order disappears .
we also showed that in the case of su(3 ) symmetry , ferromagnetic ordering appears as a secondary order parameter . in a closed system of atoms , where the number of atoms is conserved for every color ,
this implies segregation , i.e. , the formation of domains where the color of the atoms forming the superconducting state changes from domain to domain and the density of the atoms involved in the superconducting order is slightly higher than that of the unpaired ones .
we have also demonstrated that the phase transition is robust against breaking the su(3 ) symmetry of the interaction . this symmetry breaking influences the structure of the superconducting state in that it pins the order parameter to the most strongly interacting channels , but does not influence the existence of the phase transition itself .
we have not studied , however , the case where an unequal number of atoms is loaded for the three hyperfine states .
this situation has been experimentally studied for a two - component @xmath5li system.@xcite there , a segregated superconducting state has been observed in the center of the trap , where the two densities become equal in order to form cooper pairs , while at the edge of the trap , spin densities are different .
a similar scenario is expected for a slight imbalance of the three densities in our case .
it is , however , also possible that for su(3 ) symmetrical interactions , three different superconducting domains of unequal size can form for large enough traps .
formation of a fulde ferrel larkin
ovchinnikov state can not be excluded either , although this state has not been experimentally observed .@xcite here , we should also remark that we solved the gutzwiller problem only in @xmath63 dimensions , where it predicts a phase transition .
although the phase transition survives in finite dimensions too , there we do not know if the variational gutzwiller wave function displays a phase transition or not .
as long as @xmath65 remains finite , one has a step in the momentum distribution at the fermi energy and the system can gain energy by forming a superconductor .
therefore , to have a phase transition , the parameter @xmath65 must diverge at a finite critical value of @xmath46 .
this certainly happens within the local approximation ( gutzwiller approximation ) in finite dimensions too , and one can argue that this should also happen if one could evaluate the finite - dimensional gutzwiller expectation values exactly . unfortunately , restricted monte carlo calculations would be needed to give a definite answer to this question .
finally , let us briefly discuss how the transition from a superfluid to a trionic state could be experimentally detected .
one way to observe the condensate is by detecting vortices . in a perfectly su(3)-symmetrical system
there are no vortices , because any vortex can be twisted away due to the large internal symmetry of the order parameter.@xcite however , if one breaks the su(3 ) symmetry down to u(1 ) then the superconducting phase will contain usual vortices that can be relatively easily observed optically.@xcite in case of @xmath5li , e.g. , this can be achieved by approaching the feshbach resonance at @xmath279 mt from the high - field side , and thereby having a condensate in channel 12 .
then our prediction is that the superfluid state and thus the vortices disappear for large enough @xmath280 .
this ratio can be tuned by either changing the laser beam intensities and thereby the amplitude of the optical lattice or by changing the external magnetic field
. a further consequence of having @xmath281 is that the interaction induces a chemical potential difference for channels 1 and 2 .
although we do not see a sign of it within the infinite - dimensional gutzwiller approach , this may well render the phase transition first order in finite dimensions . pinning the order parameter to a single channel
has a further advantage .
since one knows that the order parameter is simply in channel @xmath282 , one can use the method of ref . to measure @xmath282 by sweeping the magnetic field through the feshbach resonance at @xmath283 and detecting the momentum distribution of the molecules thus formed .
this method would enable one to detect @xmath284 as a function of @xmath280 and verify the existence of a critical value where it goes to zero .
stability of the three - color @xmath5li system may also be an important issue .
high magnetic fields will probably considerably stabilize the three - color condensate , but using mixtures of different fermionic atoms with all scattering lengths being negative may also be a possibility .
in such a composite system , other interesting phenomena could also take place due to the difference in atomic masses .
we would like to thank c. honerkamp for fruitful discussions and exchange of ideas .
we also thank i. bloch , e. demler , d. rischke , p. zoller , and m. zwierlein for useful discussions and comments .
this research has been supported by hungarian otka under grants no .
nf061726 , no .
t046303 , no .
t049571 , and no .
ni70594 , and by the german science foundation ( dfg ) collaborative research center ( sfb trr 49 ) .
g. z. would like to thank the cas , oslo , where part of this work has been completed .
here , we connect certain expectation values with the self energy @xmath154 defined by eq .
( [ eq : def_selfenergy ] ) to express the kinetic energy in terms of the self - energy itself .
first , we shift the integration variables of @xmath152 in eq .
( [ eq : def_gamma ] ) following eq .
( [ eq : def_trf ] ) .
this transformation automatically transforms the conjugate fields by @xmath285 using these , the linear and quadratic terms become @xmath286 transforming the interaction term is more complicated .
first , we transform the on - site density , @xmath287 where we used that @xmath288 . here
, we note that all terms behave like scalars with respect to both the grassmann algebra and the color matrices ; and they commute with each other .
when performing the functional derivation , we will have to expand the exponential in terms of @xmath289 to the order @xmath290 .
thus , when calculating the third power of eq .
( [ eq : transformaction1 ] ) , we can drop certain higher order terms .
the interaction term can then be expanded as follows : @xmath291 now , we can expand the exponential in the definition of the generating functional . introducing the coupling @xmath292 of the effective theory
, we obtain the following long expression @xmath293 where @xmath294 are the transformed grassmann fields , which correspond to @xmath295 .
now one can perform the functional derivation to get the dressed propagator , @xmath296 using that @xmath297 , we can rewrite this in the following form : @xmath298 substituting @xmath299 leads to @xmath300 while setting @xmath301 we obtain another interesting identity , @xmath302 compared these to the definition of the self - energy eq .
( [ eq : def_selfenergy ] ) , one finds the identities eqs .
( [ eq : identities-1 ] ) - ( [ eq : identities-3 ] ) and thus the expression of the kinetic energy in terms of the self - energy .
the local action is defined by @xmath303 where the prime means integration over all grassmann variables except for those defined at the origin .
we shall derive its functional form in @xmath63 dimensions .
it is useful to collect the terms in the action into three groups , @xmath304 where @xmath305_{00 } \psi_0 - ( g^2 - 1 ) t_0\ ] ] contains only terms at the origin , @xmath306_{ij } \psi_j - ( g^2 - 1 ) \sum_{i \neq 0 } t_i\ ] ] is the effective action on a lattice without the origin , and @xmath307_{i0 } \psi_0 + \bar \psi_0 [ d^{0 -1 } ] _ { 0i } \psi_i \right]\ ] ] are the terms which connect the lattice and the cavity .
@xmath308 can be viewed as a generating functional term with the currents @xmath309_{0i } $ ] and their adjungates .
thus , @xmath310 since only connected graphs are generated this way .
let us first restrict ourselves to sites @xmath311 and @xmath312 being only nearest neighbors , and consider terms of order @xmath313 .
if all site indices are different , then summation gives a factor @xmath314 .
since each bare propagator @xmath315 is proportional to @xmath316 , they give a prefactor @xmath317 .
the correlation function is connected , and thus the distance between external sites is @xmath318 , and the scaling of the hopping ensures that the correlation function is @xmath319 thus , these terms give a contribution of the order @xmath320 . if two of the labels @xmath321 and @xmath322 are the same , then the summation gives a factor @xmath323 , however , connected diagrams scale at most @xmath324 , and the final result is again @xmath325 .
this argument can be generalized to any combination of identical labels .
this means that only the quadratic terms @xmath326 survive the @xmath278 limit , which can be added to the quadratic term in @xmath96 , leading to the bare propagator of the local theory @xmath188 .
the self - consistency relations are required to determine its value .
this line of argumentation can also be extended to the case where @xmath311 and @xmath312 run over non - nearest neighbor sites .
one only has to use the property that @xmath327 falls off faster than @xmath328 , with @xmath175 the number of steps needed to reach the lattice site @xmath176 from site @xmath12 .
in this appendix , we show that the matrix @xmath46 that diagonalizes the local propagator @xmath329 has a special symplectic symmetry , which implies that one can evaluate the green s functions by performing a variable transformation with this matrix .
we first observe that the nambu spinor @xmath110 satisfies the relation @xmath330 where @xmath331 and @xmath332 is the first pauli matrix acting in the nambu space .
therefore , we can parametrize @xmath329 in the following way , @xmath333 where @xmath334 is hermitian and @xmath335 is antisymmetric .
thus , the following identity holds @xmath336 this structure of @xmath329 enables us to say something about its eigenvalues .
let us assume that @xmath337 is a right - hand side eigenvector of @xmath329 ( a column vector ) , @xmath338 then , @xmath339 is also an eigenvector , @xmath340 since the eigenvalues are real . as a consequence , we can construct a unitary matrix @xmath341 as @xmath342 which transforms @xmath329 to a diagonal form , @xmath343 with @xmath344 is a real diagonal matrix .
now , we can show or see that @xmath195 is a well - defined transformation for @xmath110 in the sense that the identity @xmath330 also holds for the transformed eigenvectors , @xmath345 , and @xmath346 note that this identity is necessary to carry out the path intergral in terms of the components of the transformed field @xmath295 as new independent variables .
this equality is satisfied if @xmath347 i.e. , if the matrix @xmath348 satisfies the condition @xmath349 equation ( [ eq : well - definition ] ) can readily be verified by using the properties of the eigenvectors @xmath350 relations [ eqs .
( [ eq : diag ] ) and ( [ eq : transposed ] ) ] together with these imply that the green s function of the local green s function can be computed by first transforming to the diagonal basis of @xmath329 and then transforming back the green s function to the original basis by using the transformation @xmath348 .
importantly , the interaction term of the local action is invariant under the transformation @xmath351 .
this follows simply from the facts that the interaction can also be expressed as @xmath352 and that the determinant of @xmath46 is just 1 .
m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
* 92 * , 120403 ( 2004 ) ; m. bartenstein , a. altmeyer , s. riedl , s. jochim , c. chin , j. hecker denschlag , and r. grimm , _ ibid .
_ * 92 * , 120401 ( 2004 ) . using eq . ( [ eq : dyson_local ] ) , we can also express the self - consistency equations [ eq .
( [ eq : selfcons_equations ] ) ] in terms of the components of @xmath188 . for numerical stability reasons , we solved these equivalent equations rather than eq .
( [ eq : selfcons_equations ] ) for @xmath148.then , we determined @xmath148 from @xmath188 using eq .
( [ eq : dyson_local ] ) . | to investigate ultracold fermionic atoms of three internal states ( colors ) in an optical lattice , subject to strong attractive interaction , we study the attractive three - color hubbard model in infinite dimensions by using a variational approach .
we find a quantum phase transition between a weak - coupling superconducting phase and a strong - coupling trionic phase where groups of three atoms are bound to a composite fermion .
we show how the gutzwiller variational theory can be reformulated in terms of an effective field theory with three - body interactions and how this effective field theory can be solved exactly in infinite dimensions by using the methods of dynamical mean field theory . |
in the standard model ( sm ) the top quark is the only fermion with a mass at the electroweak symmetry breaking scale and hence is speculated to be a window on new physics beyond the sm @xcite .
so far the tevatron has measured some properties of the top quark and found good agreements with the sm predictions except for the forward - backward asymmetry @xmath0 , which shows a @xmath6 deviation from the sm expectation @xcite .
although the latest analysis based on the @xmath7 luminosity reduced the deviation to about @xmath8 , it indicated that the forward - backward asymmetry depends on the top pair invariant mass @xmath9 and for @xmath10 gev the deviation is enlarged to @xmath11 @xcite .
so far various new physics schemes have been proposed to explain such a deviation @xcite , among which one attractive way is the @xmath1-channel exchange of a neutral gauge boson @xmath12 which has sizable fcnc coupling for top and up quarks .
these @xmath12-models are especially interesting because , in addition to contributing to @xmath13 , they can also induce the top quark fcnc decays and the like - sign top pair production at the lhc . due to their suppressed rates in the sm @xcite and the rather clean backgrounds @xcite , these fcnc decays and
like - sign top pair production can be a good further test of the fcnc @xmath14 models for explaination of @xmath0 .
on the other hand , the forward - backward asymmetry , the fcnc decays and the like - sign top pair production are correlated with each other and such correlations are model - dependent and thus can help to distinguish different models at the lhc . in this work we concentrate on two such @xmath12-models ,
i.e. the left - right model and the @xmath3 model , to study the correlated effects of top quark forward - backward asymmetry at the tevatron , the fcnc decays @xmath15 ( @xmath5 ) and the like - sign top pair production at the lhc .
this work is organized as follows . in sec .
ii we briefly describe the two models and present the calculation of the observables . in sec .
iii some numerical results are presented . finally , we draw conclusions in sec .
in extensions of the sm with some extra gauge symmetry , a new neutral gauge boson called @xmath12 is often predicted with family universal or non - universal couplings to fermions . for a @xmath16 with family universal couplings
, it contributes to the @xmath17 production only via the @xmath18-channel exchange shown in fig.[fig1](a ) .
such @xmath18-channel contribution does not interfere with the dominant qcd amplitude and thus can not sizably enhance @xmath19 .
however , the situation is quite different for a @xmath16 with non - universal couplings where the fcnc interaction @xmath20 may arise without violating the constraints from flavor physics @xcite . in this case
, @xmath16 can contribute to @xmath17 production via the @xmath1-channel exchange shown in fig.[fig1](b ) , which , as pointed out in @xcite , can interfere with the qcd amplitude to enhance @xmath19 significantly while alter the cross section mildly . since we attempt to explain the deviation of @xmath19 by new physics effects , we in this work consider two models with non - universal @xmath16 couplings , which are called model - i and model - ii , respectively .
model - i extends the sm by a @xmath21 gauge symmetry under which the three generations of the right - handed up - type quarks are charged as @xmath22 @xcite .
therefore the new neutral gauge boson @xmath14 predicted in this model only couples to the right - handed quarks , and can safely escape the constraints from drell - yan measurement at the tevatron and lhc@xcite .
meanwhile , in this framework , the yukawa couplings may be generated by the froggatt - nielsen type mechanism @xcite and the chiral gauge anomalies can be avoided by introducing some extra fermions . the lagrangian relevant to our discussion is @xcite @xmath23 where @xmath24 and @xmath25 are dimensionless parameters , and @xmath26 is the generation index .
recently , it is found that this model can explain the deviation of @xmath19 and meanwhile satisfy other tevatron measurements in the parameter region : @xmath27 , @xmath28 and @xmath29 , which is obtained by the following consideration @xcite : * if @xmath16 is heavier than top quark , it will decay dominantly to @xmath30 or @xmath31 and consequently give excessive like - sign top quark events through the processes @xmath32 , @xmath33 and @xmath34 . *
if @xmath16 is much lighter than top quark , the exotic decay @xmath35 is open . for @xmath36 with @xmath37
, the decay rate will exceed @xmath38 , which can cause a tension between the dileptonic and hadronic channels for the @xmath17 production at the tevatron because such a light @xmath16 will decay into light quarks . *
the presence of a small @xmath39 is necessary to make the model phenomenologically viable
. if @xmath40 , the only decay mode of @xmath16 is @xmath41 , and then both @xmath42 and @xmath43 can give the very similar like - sign top pair signal , which is strongly constrained by the tevatron experiment .
a non - zero @xmath39 can avoid this conflict by allowing @xmath16 to decay into @xmath44 .
on the other hand , @xmath39 can not be too large because it can enhance the rates of both @xmath45 and the loop induced decay @xmath46 , which have been constrained by the measurements at the tevatron .
model - ii is a special left - right symmetric model called the third - generation enhanced left - right model , which is based on the gauge group @xmath47 with gauge couplings @xmath48 , @xmath49 , @xmath50 and @xmath51 respectively@xcite .
the key feature of this model is that the gauge bosons of the @xmath52 group couple only to the third - generation fermions ( the third generation is specially treated ) @xcite .
the gauge interactions relevant to our study are given by @xmath53 where @xmath54 , @xmath55 , @xmath56 is the mixing angle between @xmath57 and @xmath58 , @xmath59 are the unitary matrices which rotate the right - handed quarks @xmath60 and @xmath61 from interaction basis to mass eigenstates and the repeated generation indices @xmath26 and @xmath62 are summed .
similar to model - i , a sizable @xmath63 mixing with other flavor mixings suppressed is allowed by the low energy flavor physics @xcite .
such a sizable @xmath63 mixing can lead to a rather strong @xmath64 interaction with the condition @xmath65 .
similarly , the interaction to leptons are given as follows@xcite : @xmath66 the constraints on model - ii were found @xcite to be : @xmath67 from the requirement of perturbativity , @xmath68 gev for @xmath69 from the global fit of the lep data ( especially @xmath70 ) . among the oblique parameters
, @xmath71 gives the most stringent constraint , which roughly requires @xmath72 at @xmath73 level .
note that the constraints on @xmath74 from the cdf search for new resonant states or from the global fitting of the electroweak precision data are not applicable here since they usually assume a @xmath16 with family universal couplings . from the eqs.(2 ) and ( 3 ) , we can see that the flavor - conserving interactions between the dominantly right - handed @xmath14 and the first two generation fermions are suppressed by small @xmath75 for the chosen parameters in our calculation , and thus make a negligible contribution to the process @xmath76 .
eq.([neucoup ] ) indicates that , unlike model - i , model - ii gives an interaction @xmath77 larger than @xmath78 .
although both models utilize the @xmath78 interaction to explain the deviation of @xmath79 , model - ii will have more side effects than model - i , e.g. , it may also alter sizably @xmath70 , the total and the differential rates of the @xmath17 production .
the consideration of all these effects leads to a favorable region characterized by a very large @xmath80 ( @xmath81 ) and a heavy @xmath16 ( @xmath82 @xcite .
since both models allow for the fcnc @xmath78 interaction , they may predict large top quark fcnc decays @xmath83 ( @xmath5 ) and the like - sign top pair production , as shown in figs.[fig2 ] and [ fig3 ] .
the analytic expressions of the loop amplitudes for the fcnc decays are given in appendix a. three points should be noted here .
first , in model - i the processes @xmath84 and @xmath85 may lead to signals similar to the @xmath86 production , which , however , are suppressed by kinematics or high - order effects .
second , in model - ii the decay @xmath87 can proceed at tree level via the @xmath88 mixing , while in model - i it can only occur at loop level .
third , for the signature @xmath89 at the hera@xcite , in model i , the lepton - phobic @xmath14 will not contribute to this process and can safely avoid the constraints ; for the model ii , due to the large @xmath14 mass , the process @xmath89 is less sensitive to the coupling of @xmath90 .
besides , as mentioned above , the small couplings(@xmath75 ) of @xmath91 will cancel the large flavor - changing coupling(@xmath92 ) of @xmath90 and also not give rise to the significant contribution to process @xmath93 .
the sm parameters used in this calculations are @xcite @xmath94 for new physics parameters , we scan them within the following ranges : @xmath95 it should be noted that the contributions to the @xmath17 cross section mainly come from the flavor - changing t - channel in fig.1(b ) for both models . since the suppressions of small flavor - conserving coupling @xmath25 in model i and @xmath75 in model ii respectively and no interference with the sm qcd process , the s - channel @xmath96 has a negligible effects on the @xmath17 production for both models in our calculations . in additional ,
the measurements of @xmath17 cross section at the lhc still have large uncertainties and may not give a new constraints on our two models@xcite . in our scan
, we require the total cross section of the @xmath17 production and the differential cross section in each bin of @xmath9 to be within the @xmath6 regions of their experimental values at the tevatron@xcite . for model - ii
, we also consider the constraint from the @xmath71 parameter at @xmath97 level @xcite . for the calculation of the hadronic cross sections
, we use the parton distribution function cteq6l @xcite with the renormalization scale @xmath98 and factorization scale @xmath99 setting to be @xmath100 . for each model
we show in fig.4 the correlations between @xmath0 at the tevatron , the branch ratios of the top fcnc decays @xmath101 and the cross section of the @xmath86 production at the lhc with @xmath102tev .
a common feature of the correlations is that all these quantities are proportional to each other .
this is obvious since all the quantities receive contributions from the same @xmath78 interaction . for the forward - backward asymmetry
, one can see that both models can enhance its value significantly to alleviate the deviation , especially , model - i can reduce the deviation to @xmath103 level .
more details about the calculation of @xmath0 were presented in @xcite . about
the fcnc decays , fig.[fig4 ] shows that @xmath104 can have a large branching ratio in model - ii because it can proceed at tree level via the @xmath88 mixing . for each decay model - ii
gives a larger branching ratio than model - i because in model - i the decays are highly suppressed by a factor @xmath105 .
fig.[fig4 ] also indicates that in both models the branching ratios for these fcnc decays are smaller than @xmath106 , which are far below their current experimental bounds ( namely @xmath107 from d0 @xcite , @xmath108 from zeus , @xmath109 at 95% cl . using this limits and eq.(41 ) in arxiv : hep - ph/0003033 in ref[1 ]
, we can obtain the upper limit @xmath110 . ] @xcite and @xmath111 from cdf @xcite ) and also smaller than the maximal values predicted in other new physics models such as low energy supersymmetry @xcite , technicolor model @xcite or little higgs theory @xcite . from the analysis of top fcnc
decays @xcite , one can infer that detecting these decays in the present two models would be quite challenging at the lhc
. about the like - sign top pair production , fig.[fig4 ] shows that for @xmath112 the cross section in both models can be of @xmath113 order , reaching @xmath114 in model - i and @xmath115 in model - ii .
since the signal of such a production is characterized by two isolated like - sign leptons , which is free from the @xmath17 background and the huge qcd @xmath116jets background @xcite , it may be observable at the lhc and will be discussed in the following .
we note that very recently the cdf reported the dependence of @xmath79 on the top pair invariant mass @xmath9 and found a more than @xmath97 discrepancy from the sm prediction for @xmath117 @xcite .
motivated by this , we display the dependence of @xmath79 on @xmath9 in fig.[fig5 ] . in our calculation
, we have included the sm contribution and multiplied the total cross section by a k - factor @xmath118 to include the nlo qcd effect @xcite .
fig.[fig5 ] shows that both models can enhance @xmath0 in large @xmath9 region to ameliorate the discrepancy , though the discrepancy still persists at @xmath6 level .
further , we note that both models can also cause top quark polarization asymmetry in the @xmath119 production at the lhc , which was studied recently in @xcite .
since the like - sign top production can have a large rate and its background is low , we give further study on its observability at the lhc .
first , in fig.[fig6 ] we display some kinematical distributions such as the top quark transverse momentum @xmath120 and its pseudo - rapidity @xmath121 , the total transverse energy @xmath122 of the process and the separation between the two @xmath123-jets @xmath124 .
the new physics parameters are fixed as @xmath125 and @xmath126 for model - i , and @xmath127 and @xmath128 , @xmath129 for model - ii . for the last two distributions in fig.[fig6 ] , we have included the decay chain @xmath130 in our code to simulate the signals of the process . from the upper two and the last frames we can see that the most events are distributed in the region with small transverse momentum or large pseudo - rapidity .
this implies that for a light @xmath16 in model - i the top quarks tend to outgo in parallel with the beam pipe .
the third frame shows that the two @xmath123-jets tend to fly in the opposite direction since they come from the back - to - back top quarks in the @xmath86 rest frame .
the distinct shapes in the second frame are caused by different masses of @xmath2 in the two models ( if the @xmath2 mass is assumed to be in the same region for both models , they give the similar shapes ) .
now we discuss the detection of the @xmath86 production at the lhc with @xmath131 .
we choose @xmath132 ( @xmath133 ) as the signal .
we simulate the energy resolution of the detector effects by assuming a gaussian smearing for the final leptons and jets @xcite @xmath134 where @xmath135 is in gev , and @xmath136 indicates that the energy - dependent and energy - independent terms are added in quadrature .
we take the @xmath123-jet tagging efficiency as @xmath137 .
the main backgrounds are from @xmath138 and @xmath139 , which have been studied in @xcite . in our analysis
we take the same cuts as in @xcite for the signal : @xmath140 where @xmath141 denotes the transverse energy and @xmath142 is the invariant mass of the final states .
note that the production @xmath143 can also mimic our signal @xcite , which , however , has an extra jet and can be suppressed by jet veto . under such cuts
we find the acceptance rate of the signal is @xmath144 for @xmath145 in model - i and @xmath146 for @xmath129 in model - ii .
the acceptance rate is found to increase with @xmath74 , which is @xmath147 for @xmath148 and increased to @xmath149 for @xmath150 . with the backgrounds calculated in @xcite
, we get the @xmath97 sensitivity for the @xmath86 production by requiring @xmath151 for @xmath152 integrated luminosity .
the corresponding results are listed in table i , where one can learn that the @xmath86 production with a rate as large as several tens of @xmath153 can be detected at the lhc .
.@xmath97 observation bound on the rate of the @xmath86 production at the lhc with @xmath131 for @xmath152 integrated luminosity .
[ tab1 ] [ cols="^,^,^,^,^ " , ] finally , in fig.[fig7 ] we project the surviving samples in the scan on the @xmath154 plane with @xmath155 defined as @xmath156 in model - i and @xmath157 in model - ii .
the solid curves in this figure are the @xmath73 sensitivity and above each curve is the observable region .
our results show that about @xmath158 ( @xmath159 ) of the total surviving samples lie above the @xmath73 curve for model - i ( model - ii ) .
combined with fig .
4 , the points lying below the @xmath73 curve in fig.[fig7 ] give a shift of @xmath79 less than @xmath160 .
this fact implies that if the @xmath86 production is not observed at the lhc , then our considered models may not explain the anomaly of @xmath79 .
in this work we studied the correlations between @xmath0 at the tevatron , the fcnc decays @xmath161 and the like - sign top pair production at the lhc in two models with non - universal @xmath16 interactions , i.e. the left - right model and the @xmath3 model .
we also studied the dependence of @xmath0 on the top pair invariant mass @xmath9 .
we found that under the current experimental constraints both models can alleviate the deviation of @xmath0 and , meanwhile , enhance the @xmath86 production to the detectable level of the lhc .
we also found that , since the two models give different predictions for the observables and also their correlations , they may be distinguished by jointly studying these observables . in particular , we emphasize that exploring the @xmath86 production at the lhc will allow for a further test of the models which are used to explain the anomaly of @xmath79 observed at the tevatron
. note added : months after our manuscript finished , the cms collaboration reported their result of searching the like - sign top pair induced by the fcnc @xmath14 at the lhc .
the limit of the cross section @xmath162 pb at 95 % cl@xcite .
we found that in our parameter space , the like - sign top pair cross sections are far bellow the upper limits set by the cms .
this work was supported in part by hastit under grant no .
2009hastit004 , by the national natural science foundation of china ( nnsfc ) under grant nos . 10821504 , 10725526 , 10775039 , 11075045 and by the project of knowledge innovation program ( pkip ) of chinese academy of sciences under grant no .
here we list the amplitudes for the loop induced processes @xmath163 shown in fig.2 .
the notation @xmath164 denotes the amplitude of the diagram with quark @xmath165 ( @xmath166 ) appeared in the loop . for model - i ,
the amplitudes are given by @xmath167p_{r}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(b)}_{g}(q)&=&ag_{s}t^{a}_{\beta\alpha}\bar{u}(p_{u } ) \frac{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}p_{l}}{p_{u}^{2}-m_{t}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(c)}_{g}(q)&=&ag_{s}t^{a}_{\beta\alpha}\bar{u}(p_{u } ) \frac{(2b_{1}^{3}+1)m_{t}^{2}\gamma^{\nu}p_{r}}{p_{t}^{2}-m_{u}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(a)}_{\gamma}(q)&=&\frac{2}{3}ea\bar{u}(p_{u } ) [ -4c^{\nu\rho}\gamma_{\rho}-\gamma^{\nu}+2b_{0}^{1}\gamma^{\nu } -2(\slashed p_{t}-\slashed p_{u})\gamma^{\nu}\gamma^\rho c_{\rho}]p_{r}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(b)}_{\gamma}(q)&=&\frac{2}{3}ea\bar{u}(p_{u } ) \frac{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}p_{l}}{p_{u}^{2}-m_{t}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(c)}_{\gamma}(q)&=&\frac{2}{3}ea\bar{u}(p_{u } ) \frac{(2b_{1}^{3}+1)m_{t}^{2}\gamma^{\nu}p_{r}}{p_{t}^{2}-m_{u}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(a)}_{z}(u)&=&-a\delta_{\alpha\beta}g\bar{u}(p_{u})(\frac{2}{3 } \sin\theta_{w}\tan\theta_{w } ) [ -4c^{\nu\rho}\gamma_{\rho}-\gamma^{\nu}+2b_{0}^{1}\gamma^{\nu}\nonumber \\&&-2(\slashed p_{t}-\slashed p_{u})\gamma^{\nu}\gamma^{\rho } c_{\rho}]p_{r}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{z } ) , \\ \mathcal { m}^{(a)}_{z}(t)&=&-a\delta_{\alpha\beta}g\bar{u}(p_{u } ) \biggl[\frac{m_{t}^{2}}{\cos\theta_{w}}c_{0}\gamma^{\nu}p_{r}+\frac{2}{3 } \sin\theta_{w}\tan\theta_{w}[-4c^{\nu\rho}\gamma_{\rho}\nonumber \\&&-\gamma^{\nu}+2b_{0}^{1}\gamma^{\nu } -2(\slashed p_{t}-\slashed p_{u})\gamma^{\nu}\gamma^{\rho } c_{\rho}]p_{r}\biggl ] u(p_{t})\varepsilon^{\ast}_{\nu}(p_{z } ) , \\
\mathcal { m}^{(b)}_{z}(q)&=&-a\delta_{\alpha\beta}g\bar{u}(p_{u } ) ( \frac{4\sin^{2}\theta_{w}-3}{6\cos\theta_{w } } ) \biggl[\frac{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}p_{l}}{p_{u}^{2}-m_{t}^{2}}\biggl]u(p_{t})\varepsilon^{\ast}_{\nu}(p_{z } ) , \\
\mathcal { m}^{(c)}_{z}(q)&=&-a\delta_{\alpha\beta}g\bar{u}(p_{u } ) ( \frac{2}{3 } \sin\theta_{w}\tan\theta_{w})\frac{(2b_{1}^{3}+1)m_{t}^{2}\gamma^{\nu}p_{r}}{p_{t}^{2}-m_{u}^{2 } } u(p_{t})\varepsilon^{\ast}_{\nu}(p_{z}).\end{aligned}\ ] ] for model - ii , the amplitudes are given by @xmath168g^{u}_{z^{\prime}r}p_{r}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(a)}_{g}(t)&= & bg_{s}t_{\beta\alpha}^{a } \bar{u}(p_{u})\{[-4c^{\nu\rho}\gamma_{\rho}-\gamma^{\nu}+2b_{0}^{1}\gamma^{\nu}- 2(\slashed p_{t}-\slashed p_{u})\gamma^{\nu}\gamma^{\rho } c_{\rho}]g^{t}_{z^{\prime}r}p_{r}\nonumber \\&&+8m_{t}c^{\nu}g^{t}_{z^{\prime}l}p_{l}-4m_{t}(p_{u}-p_{t})^{\nu}c_{0 } g^{t}_{z^{\prime}l}p_{l}\}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(b)}_{g}(u)&=&bg_{s}t_{\beta\alpha}^{a } \bar{u}(p_{u } ) \frac{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}g^{u}_{z^{\prime}r}p_{l}}{p_{u}^{2}-m_{t}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g } ) , \\
\mathcal { m}^{(b)}_{g}(t)&=&bg_{s}t_{\beta\alpha}^{a}\bar{u}(p_{u } ) \{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}g^{t}_{z^{\prime}r}p_{l}+(4b^{2}_{0}-2)m_{t}\slashed p_{u}\gamma^{\nu}g^{t}_{z^{\prime}l}p_{l}\nonumber\\ & & + ( 4b^{2}_{0}-2)m_{t}^{2 } \gamma^{\nu}g^{t}_{z^{\prime}l}p_{r}\ } \frac{u(p_{t})\varepsilon^{\ast}_{\nu}(p_{g})}{p_{u}^{2}-m_{t}^{2}},\end{aligned}\ ] ] @xmath169g^{u}_{z^{\prime}r}p_{r}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(a)}_{\gamma}(t)&= & \frac{2}{3}eb \bar{u}(p_{u})\{[-4c^{\nu\rho}\gamma_{\rho}-\gamma^{\nu}+2b_{0}^{1}\gamma^{\nu}- 2(\slashed p_{t}-\slashed p_{u})\gamma^{\nu}\gamma^{\rho } c_{\rho}]g^{t}_{z^{\prime}r}p_{r}\nonumber \\&&+8m_{t}c^{\nu}g^{t}_{z^{\prime}l}p_{l}-4m_{t}(p_{u}-p_{t})^{\nu}c_{0 } g^{t}_{z^{\prime}l}p_{l}\}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(b)}_{\gamma}(u)&= & \frac{2}{3}eb\bar{u}(p_{u } ) \frac{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}g^{u}_{z^{\prime}r}p_{l}}{p_{u}^{2}-m_{t}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(b)}_{\gamma}(t)&= & \frac{2}{3}eb\bar{u}(p_{u})\{(2b^{2}_{1}+1)m_{t}\slashed p_{u}\gamma^{\nu}g^{t}_{z^{\prime}r}p_{l}+(4b^{2}_{0}-2)m_{t}\slashed p_{u}\gamma^{\nu}g^{t}_{z^{\prime}l}p_{l } \nonumber \\ & & + ( 4b^{2}_{0}-2)m_{t}^{2 } \gamma^{\nu}g^{t}_{z^{\prime}l}p_{r}\ } \frac{u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma})}{p_{u}^{2}-m_{t}^{2 } } , \\
\mathcal { m}^{(c)}_{\gamma}(u)&=&\frac{2}{3}eb\bar{u}(p_{u } ) \frac{(2b_{1}^{3}+1)m_{t}^{2}\gamma^{\nu}g^{u}_{z^{\prime}r}p_{r}}{p_{t}^{2}-m_{u}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma } ) , \\
\mathcal { m}^{(c)}_{\gamma}(t)&=&\frac{2}{3}eb\bar{u}(p_{u } ) \frac{(2b_{1}^{3}+1)m_{t}^{2}\gamma^{\nu}g^{t}_{z^{\prime}r}p_{r}+(4b_{0}^{3}-2)m_{t}\gamma^{\nu}\slashed p_{t}g^{t}_{z^{\prime}l}p_{l } } { p_{t}^{2}-m_{u}^{2}}u(p_{t})\varepsilon^{\ast}_{\nu}(p_{\gamma}).\end{aligned}\ ] ] in above expressions , @xmath170 and @xmath171 denote the momenta of the top and up quark respectively , @xmath172 and @xmath155 are loop functions defined in @xcite and calculated by looptools @xcite , the dependence of the loop functions on momentums and masses is given by @xmath173 and the constants are defined by @xmath174 99 for top quark reviews , see , e.g. , w. bernreuther , 35 , 083001,(2008 ) d. chakraborty , j. konigsberg , d. rainwater , _ ann .
nucl . part .
_ * 53 * , 301 ( 2003 ) ; e. h. simmons , hep - ph/0211335 ; c .- p .
yuan , hep - ph/0203088 ; s. willenbrock , hep - ph/0211067 ; m. beneke , _ et al .
_ , hep - ph/0003033 ; t. han , arxiv:0804.3178 ; for model - independent new physics study , see , e.g. , c. t. hill and s. j. parke , 49 , 4454 ( 1994 ) ; k. whisnant , _ et al .
_ , 56 , 467 ( 1997 ) ; j. m. yang , b .-
young , 56 , 5907 ( 1997 ) ; k. hikasa , _ et al .
_ , 58 , 114003 ( 1998 ) ; j. a. aguilar - saavedra , arxiv:0811.3842 ; r.a .
coimbra , _ et al .
_ , arxiv:0811.1743
. g. stricker _ et al .
_ , cdf note 9724 ( 2009 ) ; t. aaltonen _ et al .
_ [ cdf collaboration ] , phys . rev .
lett . 101 , 202001 ( 2008 ) ; v.
m. abazov _ et al . _
[ d0 collaboration ] , phys .
100 , 142002 ( 2008 ) .
t. aaltonen _ et al . _
[ the cdf collaboration ] , arxiv:1101.0034 [ hep - ex ] .
a. djouadi _ et al .
_ , 82 , 071702 ( 2010 ) ; p. ferrario , g. rodrigo , 80 , 051701 ( 2009 ) ; k. cheung , w. y. keung , t. c. yuan , 682 , 287 ( 2009 ) ; s. jung _ et al .
_ , 81 , 015004 ( 2010 ) ; j. shu , t. tait , k. wang , 81 , 034012 ( 2010 ) ; p. h. frampton , j. shu , k. wang , 683 , 294 ( 2010 ) ; a. arhrib , r. benbrik , c. h. chen , 82 , 034034 ( 2010 ) ; i. dorsner _ et al . _ , 81 , 055009 ( 2010 ) ; d. w. jung _ et al . _ , 691 , 238 ( 2010 ) ; v. ahrens _ et al . _ , jhep * 1009 * , 097 ( 2010 ) ; m. v. martynov , a. d. smirnov , mod . phys
. lett .
a * 25 * , 2637 ( 2010 ) ; r. s. chivukula , e. h. simmons , c .- p .
yuan , 82 , 094009 ( 2010 ) ; q .- h .
cao _ et al .
_ , phys .
rev . * d81 * , 114004 ( 2010 ) ; g. rodrigo and p. ferrario , arxiv:1007.4328 ; v. barger , w .- y .
keung , c .- t .
yu , 81 , 113009 ( 2010 ) ; m. bauer _ et al .
_ , jhep 1011 , 039 ( 2010 ) ; c. zhang , s. willenbrock , arxiv:1008.3869 ; j. a. aguilar - saavedra , 843 , 638 ( 2011 ) ; c. h. chen , g. cvetic , c. s. kim , arxiv:1009.4165 ; k. kumar _ et al .
_ , jhep 1008 , 052 ( 2010 ) ; c. degrande _ et al .
_ , arxiv:1010.6304 [ hep - ph ] ; b. xiao , y .- k .
wang , s. h. zhu , 82 , 034026 ( 2010 ) ; arxiv:1011.0152 ; arxiv:1101.2507 ; g. burdman , l. lima , r. d. matheus , arxiv:1011.6380 ; e. alvarez , l. rold , a. szynkman , arxiv:1011.6557 ; d. w. jung _ et al .
_ , arxiv:1012.0102 ; k. cheung , t. c. yuan , arxiv:1101.1445 ; c. delaunay _ et al .
_ , arxiv:1101.2902 .
d. w. jung , p. ko and j. s. lee , arxiv:1011.5976 ; d. choudhury _ et al .
_ , arxiv:1012.4750 .
s. jung , h. murayama , a. pierce and j. d. wells , 81 , 015004 ( 2010 ) .
et al . _ , 81 , 014016 ( 2010 )
; see for example , g. eilam , j. l. hewett and a. soni , 44 , 1473 ( 1991 ) ; b. mele , s. petrarca and a. soddu , 435 , 401 ( 1998 ) .
t. stelzer , z. sullivan and s. willenbrock , 58 , 094021 ( 1998 ) ; w. s. hou _ et al .
_ , 409 , 344 ( 1997 ) ; f. larios and f. penunuri , j. phys .
g 30 , 895 ( 2004 ) ; j. cao _ et al .
_ , 70 , 114035 ( 2004 ) ; o. cakir _ et al .
_ , eur .
j. c 70 , 295 ( 2010 ) ; o. gedalia , l. mannelli and g. perez , jhep * 1010 * , 046 ( 2010 ) .
yu . p. gouz and s. r. slabospitsky , 457 , 177 ( 1999 ) .
s. k. gupta , arxiv:1011.4960 [ hep - ph ] .
e. l. berger , q. h. cao , c. r. chen , g. shaughnessy and h. zhang , phys .
* 105 * , 181802 ( 2010 ) h. zhang , e. l. berger , q. h. cao , c. r. chen and g. shaughnessy , phys .
b * 696 * , 68 ( 2011 ) t. aaltonen _ et al .
_ [ cdf collaboration ] , phys . rev .
102 , 041801 ( 2009 ) ; m. s. carena , a. daleo , b. a. dobrescu and t. m. p. tait , phys . rev .
d * 70 * , 093009 ( 2004 ) [ arxiv : hep - ph/0408098 ] .
t. aaltonen _ et al . _
[ the cdf collaboration ] , phys .
lett . * 106 * , 121801 ( 2011 ) [ arxiv:1101.4578 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , jhep * 1105 * , 093 ( 2011 ) [ arxiv:1103.0981 [ hep - ex ] ] . c. d. froggatt and h. b. nielsen , nucl .
b * 147 * , 277 ( 1979 ) .
x. g. he and g. valencia , 66 , 013004 ( 2002 ) ; 68 , 033011 ( 2003 ) .
s. chekanov _ et al . _
[ zeus collaboration ] , 559 , 153 ( 2003 ) .
c. amsler _ et al .
_ , particle data group , 667 , 1 ( 2008 ) .
j. pumplin _ et al .
_ , jhep 0602 , 032 ( 2006 )
. m. cacciari _ et al .
_ , jhep 0809 , 127 ( 2008 ) ; s. moch and p. uwer , 78 , 034003 ( 2008 ) ; n. kidonakis and r. vogt , 78 , 074005 ( 2008 ) ; v. m. abazov _ et al . _
[ d0 collaboration ] , 693 , 81 ( 2010 ) .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
lett . 101 , 192002 ( 2008 ) .
see , e.g. , c. s. li , r. j. oakes , j. m. yang , 49 , 293 ( 1994 ) ; g. couture , c. hamzaoui , h. konig , 52 , 1713 ( 1995 ) ; j. l. lopez , d. v. nanopoulos , r. rangarajan , 56 , 3100 ( 1997 ) ; j. m. yang , b. l. young , x. zhang , 58 , 055001 ( 1998 ) ; j. m. yang , c. s. li , 49 , 3412 ( 1994 ) ; j. guasch , j. sola , 562 , 3 ( 1999 ) ; g. eilam _ et al .
_ , 510 , 227 ( 2001 ) ; c. s. li , l. yang , l. jin , 599 , 92 ( 2004 ) ; z. heng _ et al .
_ , 79 , 094029 ( 2009 ) ; j. cao _ et al . _ , 79 , 054003 ( 2009 ) ; 75 , 075021 ( 2007 ) ; 74 , 031701 ( 2006 ) ; 651 , 87 ( 2003 ) .
x. wang _ et al .
_ , 50 , 5781 ( 1994 ) ; g. lu _ et al .
_ , 68 , 015002 ( 2003 ) ; j. cao _ et al .
_ , 76 , 014004 ( 2007 ) ; 67 , 071701 ( 2003 ) ; h. j. zhang , 77 , 057501 ( 2008 ) .
han , l. wang , j. m. yang , arxiv:0903.5491 ; x. wang , _ et al .
_ , 807 , 210 ( 2009 ) ; y. zhang , g. lu , x. wang , arxiv:1011.0552 ; j. j. zhang , _ et . al .
_ , phys .
lett . * 102 * , 072001 ( 2009 ) ; j. drobnak , s. fajfer and j. f. kamenik , phys .
* 104 * , 252001 ( 2010 ) ; phys .
d * 82 * , 073016 ( 2010 ) . | the anomaly of the top quark forward - backward asymmetry @xmath0 observed at the tevatron can be explained by the @xmath1-channel exchange of a neutral gauge boson ( @xmath2 ) which has sizable flavor changing coupling for top and up quarks .
this gauge boson can also induce the top quark flavor - changing neutral - current ( fcnc ) decays and the like - sign top pair production at the lhc . in this work we focus on two models which predict such a @xmath2 , namely the left - right model and the @xmath3 model , to investigate the correlated effects on @xmath0
, the fcnc decays @xmath4 ( @xmath5 ) and the like - sign top pair production at the lhc .
we also pay special attention to the most recently measured @xmath0 in the large top pair invariant mass region .
we find that under the current experimental constraints both models can alleviate the deviation of @xmath0 and , meanwhile , enhance the like - sign top pair production to the detectable level of the lhc .
we also find that the two models give different predictions for the observables and their correlations , and thus they may even be distinguished by jointly studying these top quark observables . |
in recent work @xcite , m. kontsevich has established a remarkable result on the formality of the hochschild complex of a smooth manifold . an important consequence of this result is an explicit construction of an associative @xmath1-product on an arbitrary smooth poisson manifold @xmath7 .
this @xmath1-product is given by a formal power series of bidifferential operators ( depending on a parameter @xmath8 ) .
for a general poisson manifold this series does not converge , except for @xmath9 .
thus the analytic properties of the @xmath1-product are somewhat obscure . however , as we shall show in this paper , if @xmath10 is the dual of a finite - dimensional real lie algebra @xmath0 ( or , in general , if the coefficients of the poisson bracket @xmath11 are linear ) , then the situation is much nicer . in this case , we can set @xmath12 , and kontsevich s @xmath1-product descends to an actual product on the algebra of polynomial functions on @xmath13 , which can be naturally identified with the symmetric algebra @xmath14 . in this paper , we will identify @xmath15 with the convolution algebra of distributions on @xmath0 supported at @xmath16 .
then the @xmath1-product can be viewed as a new associative convolution product of such distributions .
let @xmath17 be the algebra of germs at @xmath16 of differential operators on @xmath0 with analytic coefficients .
by virtue of the pairing between functions and distributions , @xmath18 is naturally a _ right _ @xmath17-module and to emphasize this fact we shall write the action on distributions _ on the right_. our first main result is the following given @xmath19 of order @xmath20 , there is a unique element @xmath21 of order @xmath20 in @xmath17 such that for all @xmath22@xmath23 [ -main - d ] next , in @xcite , kontsevich introduces a certain formal power series @xmath24 \right ) , \label{=s1}%\ ] ] where the constants @xmath25 are expressed as integrals of smooth differential forms over certain manifolds with corners ( compactified configuration spaces , introduced in @xcite ) .
the power series @xmath26 converges to an analytic function @xmath27 in some neighborhood of @xmath16 in @xmath0 .
[ -tau ] for @xmath28 , define the adjoint vector field @xmath29 by the formula @xmath30 let @xmath31 be the subalgebra of invariant distributions , and let @xmath32 be the right ideal of @xmath17 generated by the germs of the differential operators @xmath33 , @xmath28 .
our next result is the following for @xmath34 , the operator @xmath35 belongs to @xmath32 .
[ -main - r ] here @xmath36 and @xmath37 is the corresponding differential operator defined in theorem [ -main - d ] , and for @xmath38 ( convolution of distributions on @xmath0 ) .
we now discuss some applications of these results to invariant distributions on lie groups .
let @xmath6 be a finite - dimensional lie group with lie algebra @xmath0 .
we write @xmath39 for the enveloping algebra of @xmath40 and @xmath41 for the center of the enveloping algebra .
we identify @xmath42 with the convolution algebra of distributions on @xmath6 supported at @xmath43 .
let @xmath44 and @xmath45 be the spaces of germs at @xmath46 and @xmath16 of distributions on @xmath6 and @xmath0 , and let @xmath47 and @xmath48 be the subspaces of @xmath0-invariant germs .
we denote by @xmath49 and @xmath50 the pushforward and pullback of distributions ( germs ) under the exponential map ; and define @xmath51 by @xmath52 where for @xmath53 @xmath54 \right ) \text{.}%\ ] ] here @xmath55 are the bernoulli numbers .
it follows from ( * ? ? ?
8) , that the restriction of @xmath56 to @xmath57 is an _ algebra _
isomorphism from @xmath57 to @xmath58 , i.e. , @xmath59 for @xmath60 , where @xmath61 and @xmath62 denote convolutions in the lie algebra and lie group respectively .
( this fact was first established by duflo in @xcite@xmath63 using kirillov s orbit method ) . from theorems
[ -main - d ] and [ -main - r ] , we can deduce for @xmath34 , the following differential operator @xmath64 lies in @xmath65 : @xmath66 [ -kv - conjecture ] this result was conjectured by m. kashiwara and m. vergne in @xcite , and proved by them for @xmath0 solvable .
it has several important corollaries .
first , since operators from @xmath32 annihilate distributions from @xmath48 , we obtain the following extension of the isomorphism ( [ = duflo - eta ] ) : for @xmath34 and @xmath67 @xmath48 , we have @xmath68 . [ -module - iso ] next , recall that a distribution @xmath69 on @xmath6 is called an eigendistribution if there is a character @xmath70 such that @xmath71 eigendistributions on the lie algebra are defined similarly , and we obtain the following the map @xmath56 takes germs of invariant eigendistributions on the lie algebra to those on the lie group.[-eigendist ] as another consequence of theorem [ -module - iso ] , we obtain every nonzero bi - invariant differential operator on @xmath6 admits a ( local ) fundamental solution.[-fund - sol ] note that the analogous statement for left - invariant differential operators on @xmath6 is known to be false .
in fact , the first example of a not locally solvable partial differential operator ( @xcite ) is a left - invariant vector field on a three - dimensional heisenberg group .
every nonzero bi - invariant differential operator on @xmath6 is locally solvable .
this last result was originally proved by duflo in @xcite ( see also @xcite , @xcite , @xcite for various special cases ) .
an interpretation of the statement of theorem [ -fund - sol ] in terms of kirillov s orbit method can be found in ( * ? ? ? * and 8.2 ) .
most of our results extend to the case of finite - dimensional lie superalgebras without much additional effort .
for simplicity of exposition , we work with ordinary lie algebras and at the end explain the modifications necessary for the super - setting . *
* we would like to thank i. gelfand , f. knop and the other participants of the gelfand seminar at rutgers in which we first learned the main ideas behind the kontsevich formality theorem .
the authors would also like to thank m. kontsevich for sharing his insights with us and p. deligne for many valuable suggestions.[-sec - intro ]
following kontsevich , for @xmath72 , we define @xmath73 here @xmath74 .
@xmath75 is a smooth manifold of dimension @xmath76 , with a free action of the two - dimensional group @xmath77 we consider the quotient space @xmath78 then @xmath79 .
the spaces @xmath75 and @xmath80 have @xmath81 connected components , and we denote by @xmath82 and @xmath83 the component where @xmath84 similarly , for @xmath85 , we define @xmath86 clearly , @xmath87 consider , for example , the space @xmath88 for each point @xmath89 we can choose a unique representative of the form @xmath90 .
thus @xmath91 is homeomorphic to @xmath92 .
similarly , it is easy to see that @xmath93 the spaces @xmath80 have natural compactifications @xmath94 , such that the boundary components in @xmath95 correspond to various degenerations of the configurations of points .
for example , @xmath96,\end{aligned}\ ] ] of particular interest is the space @xmath97 which can be drawn as `` the eye '' .
[ ptbh ] eye.eps the circle @xmath98 represents two points coming close together in the interior of @xmath99 , and two arcs @xmath100 represent the first point ( or the second point ) coming close to the real line . finally , the two corners @xmath101 correspond to both points approaching the real line .
the compactification @xmath94 is described in general in ( * ? ? ?
5 ) , and the boundary components are products of various @xmath102 s and @xmath103s@xmath104 in particular , the boundary components of codimension 1 in @xmath105 are of the form 1 .
@xmath106 ( @xmath107 ) or 2 .
@xmath108 ( @xmath109 or @xmath110 ) .
the first case corresponds to a cluster of @xmath111 points in @xmath99 coming infinitely close to each other in @xmath99 , and the second case corresponds to a cluster of @xmath111 points in @xmath99 and @xmath20 points in @xmath112 coming infinitely close to each other and @xmath112 .
we consider the class of `` admissible '' oriented graphs @xmath113 such that * @xmath114 consists of @xmath115 vertices of the `` first '' type labeled @xmath116and @xmath117 vertices of the `` second '' type labeled \{@xmath118 . * for each vertex @xmath119 of the first type , there are @xmath120 edges in @xmath121 starting from @xmath119 , labeled @xmath122 * @xmath121 contains no edges starting from the vertices of second type , and no loops ( edges of type @xmath123 ) or multiple edges . for each point in @xmath124 , we can draw the graph @xmath125 in the closed upper half - plane @xmath126 as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ vertices @xmath119 are placed at @xmath127 , vertices @xmath128 are placed at @xmath129 @xmath112 , and edges are drawn as lobachevsky geodesics ( semicircles centered on @xmath112 ) connecting the vertices .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for each edge
@xmath130 connecting the vertex at @xmath131 to @xmath132 , we define @xmath133 to be the angle between @xmath134 and the vertical line through @xmath135 .
then @xmath136 define a map @xmath137 from @xmath138to the @xmath111-dimensional torus @xmath139 , where @xmath140 as follows @xmath141 observe that @xmath137 descends to a map from @xmath80 to @xmath139 .
define @xmath142 to be the pullback to @xmath80 of the volume form on @xmath139 .
the form @xmath142 is a smooth @xmath111-form on @xmath80 , which extends continuously to @xmath94 .
we define the _ _ weight _ _ on p.6 of @xcite by the factor of @xmath143 and from the weight @xmath144 on p. 23 by a factor of @xmath145 . ] _
_ @xmath146 by @xmath147 evidently , @xmath148 , unless @xmath149.[ss - weights ] let @xmath150 be a smooth poisson bracket ( bivector field ) on @xmath151 .
kontsevich s main result gives the following construction of an _ associative star - product _ on @xmath152 \right ] $ ] : @xmath153 where @xmath154 , @xmath155 and in general @xmath156 here @xmath157 is defined as a class of admissible graphs with @xmath115 vertices of the first type , two vertices of the second type and @xmath158 @xmath159 .
the weight @xmath146 is as in the previous subsection , and @xmath160 is a bidifferential operator defined as follows : @xmath161 \times\label{=ouch}\\ & \times\left ( \prod_{e\in e_{\gamma},e=(\ast,\overline{1})}\partial _ { i(e)}\right ) f_{1}\times\left ( \prod_{e\in e_{\gamma},e=(\ast,\overline { 2})}\partial_{i(e)}\right ) f_{2}.\nonumber\end{aligned}\ ] ] to obtain this operator , we relabel each vertex @xmath162 @xmath163 by a component @xmath164 of @xmath11 , relabel @xmath165 as @xmath166 and @xmath167 as @xmath168 .
we also label the edge @xmath169 with @xmath170 and the edge @xmath171 with @xmath172 . then an edge leading to
the vertex indicates differentiation of the function labeling the vertex with respect to the index labeling the edge .
below is an example of a graph from @xmath173 with @xmath174@xmath175{ccc}% \overset{1}{\bullet } & \rightarrow & \overset{2}{\bullet}\\ \downarrow & \swarrow & \downarrow\\ \underset{\overline{1}}{\bullet } & & \underset{\overline{2}}{\bullet}% \end{array } \text{\ or ( relabeling)\ \ \ } % \begin{array } [ c]{ccc}% \overset{\gamma^{i_{1}j_{1}}}{\bullet } & \overset{_{j_{1}}}{\rightarrow } & \overset{\gamma^{i_{2}j_{2}}}{\bullet}\\ \,\downarrow^{_{i_{1 } } } & \overset{i_{2}}{\swarrow } & \downarrow^{j_{2}}\\ \underset{f_{1}}{\bullet } & & \underset{f_{2}}{\bullet}% \end{array } .\ ] ] the corresponding bidifferential operator @xmath176 is given by @xmath177 let @xmath178 be a poisson bivector field on @xmath179 as before ; following ( * ? ? ?
8) , we define the map @xmath180 from @xmath151 to @xmath181 \right ] $ ] by @xmath182 here @xmath183 is the set of all admissible graphs with @xmath184 vertices of first type and no vertices of the second type such that @xmath158 @xmath185 and @xmath186 .
the differential operator @xmath187 is defined by labeling the vertex @xmath184 with @xmath188 , the remaining vertices by components of @xmath11 , and then proceeding as in formula ( [ = ouch ] ) .
let @xmath57 be the center of the poisson algebra @xmath189 , i.e. @xmath190 then , as established in ( * ? ? ?
* ch . 8) , @xmath191 for @xmath192 .
let us briefly explain the ideas underlying the above equality .
kontsevich s main result is a construction of an @xmath193-quasiisomorphism @xmath42 between the differential graded lie algebras @xmath194 and @xmath195 consisting of the skew polyvector fields and polydifferential operators on @xmath179 , respectively .
the quasiisomorphism @xmath42 can be regarded as a ( geometric ) map between formal manifolds ( @xmath69-manifolds ) associated to @xmath194 and @xmath195 , respectively .
the skew bivector field @xmath196 can be regarded as a point distribution @xmath197 in the first @xmath69-manifold .
then the tangent spaces at @xmath196 and @xmath198 are differential graded lie algebras @xmath199and @xmath200 , with the same underlying lie algebras as @xmath194 and @xmath195 , but with modified differentials .
the derivative @xmath201 of @xmath42 at @xmath196 induces a graded linear map between @xmath202 and @xmath203 , and @xmath180 is precisely the restriction of @xmath201 to the @xmath16-vector fields ( i.e. , functions ) .
the equality ( [ = cup - on - functions ] ) follows from compatibility of @xmath201 with the natural cup product structures on the ( tangent ) cohomology of @xmath202 and @xmath200 , as established in ( * ? ? ?
* theorem 8.2 ) . the argument proving ( * ? ? ?
* theorem 8.2 ) , which was merely outlined in @xcite , was communicated to us by m. kontsevich ( @xcite ) , and actually proves a slightly stronger result . since we need this stronger version , we shall reproduce kontsevich s argument for the case of lie algebras in section [ -tang - for - lie ] .
in the rest of the paper we restrict ourselves to the situation when @xmath204 @xmath205 is the dual of a finite dimensional lie algebra @xmath0 , and @xmath11 is the standard poisson bracket on @xmath205 . in this case the coefficients of a bivector field @xmath206are linear functions on @xmath205 .
more precisely , let @xmath207 be some basis of @xmath0 viewed as coordinate functions on @xmath205 , and let @xmath208 be the corresponding set of structure constants .
we normalize @xmath207 such that @xmath209 then @xmath210 = \sum_{k=1}^{d}c_{ij}^{k}x_{k}\text{,}%\ ] ] and @xmath211 i.e.
, @xmath212 the linearity of the coefficients of @xmath11 implies that the formula ( [ = star - summation ] ) for the bidifferential operator @xmath213 can be rewritten as @xmath214 where @xmath215 consists of those graphs in @xmath157 for which there is at most one incoming edge at every vertex of the first type .
indeed , if @xmath216 , the corresponding bidifferential operator @xmath217 is automatically @xmath16 .
next , let @xmath166 and @xmath168 be two polynomials on @xmath205 with @xmath218 , @xmath219 .
then we remark that the graphs contributing to the star - product formula for @xmath220 can have no more than @xmath221 vertices of the first type .
indeed , for any @xmath222 the corresponding bidifferential operator @xmath217 contains exactly @xmath223 differentiations . when @xmath224 , @xmath160 is obviously @xmath16 ( because @xmath166 can be differentiated at most @xmath225 times , @xmath168 at most @xmath226 times and each of the coefficients of @xmath11 corresponding to the remaining vertices at most once ) .
hence @xmath227 this sum is finite , and if we set @xmath12 , we obtain a polynomial on @xmath205 of degree @xmath221 .
therefore , kontsevich s @xmath228-product descends to an _ actual product _ on the algebra of polynomial functions on @xmath205 , which can be naturally identified with the symmetric algebra @xmath229 .
let @xmath230 and @xmath231 be the weyl algebras of polynomial coefficient differential operators on @xmath0 and @xmath205 respectively .
both of these algebras are generated by @xmath232 , and there is a unique anti - isomorphism @xmath233 between them , which is the identity on the generators . observe that @xmath233 interchanges multiplication and differentiation operators .
we will now change our point of view and regard @xmath18 as the algebra of distributions with point support ( at @xmath234 ) .
then @xmath18 is naturally a _
right module _ for the algebra @xmath230 via @xmath233 .
as discussed in the introduction , we shall write differential operators acting on distributions , on the right .
this approach leads to the following let @xmath19 @xmath235be a homogeneous distribution of order @xmath20 , and let @xmath236 . then there is a differential operator @xmath237 @xmath230 of order at most @xmath238 with polynomial coefficients of degree at most @xmath115 , such that for any @xmath239 @xmath240 [ -left - to - right ] from the formula ( [ = ouch ] ) for @xmath241 , it is clear that @xmath242 is a differential operator from @xmath243 .
since at most @xmath115 edges of @xmath125 can terminate at the vertex @xmath165 ( corresponding to @xmath244 ) , the order of @xmath245 is at most @xmath246 also , the degree of the coefficients of this differential operator is @xmath247 which is at most @xmath20 . now let @xmath248
. then the statement follows from the discussion above .
* let @xmath249 be the basis of @xmath205 dual to @xmath250 viewed as coordinate functions on @xmath0 .
then , by the lemma above , for @xmath251 homogeneous of order @xmath20 , we can write@xmath252 where @xmath253 and @xmath254 are multi - indices , and , as usual @xmath255 moreover , @xmath256 subtracting second equation from the first , obtain @xmath257 .
we now proceed with the proof of theorem [ -main - d ] .
@xmath258.[-count - graph ] to describe any graph from @xmath215 , it suffices to provide the following data for each vertex @xmath259 @xmath260 : * the source of an incoming edge , if any ( @xmath115 choices , counting the case when there is no incoming edge ) ; * whether the vertex is connected to the vertex @xmath165 , vertex @xmath167 , both or neither ( 4 choices ) .
hence we have at most @xmath261 choices for each of @xmath115 vertices of the first type .
taking into account the possible labelings of the outgoing edges at each vertex , we get @xmath262 now the statement follows from the inequality @xmath263 .
let @xmath236 .
then @xmath264 [ -estimate - weight ] recall that the weight @xmath146 associated to the graph @xmath236 is given by the formula @xmath265 where @xmath142 is the pullback of the volume form from the @xmath223-dimensional torus @xmath266 under @xmath137 .
we consider the preimage of a generic point @xmath267 under @xmath137 .
using the action of @xmath268 , we can fix @xmath269 and @xmath270 , and thus identify @xmath271 with @xmath272 . rewriting the formula ( [ = angle ] ) as @xmath273 we see that each edge gives rise to a pair of quadratic equations involving the coordinates of its endpoints .
thus we obtain a system of @xmath223 quadratic equations in @xmath223 real variables ( the real and imaginary parts of @xmath274 ) .
generically , this system has at most @xmath275 ( complex ) solutions .
therefore , we conclude that @xmath276 where @xmath277 is the volume of @xmath278 @xmath279 @xmath266 . obviously , @xmath280 and the statement follows . [ proof of theorem [ -main - d]]without loss of generality , we may take @xmath281 according to the formulas ( [ = star - pr ] ) , ( [ = star - summation ] ) and lemma [ -left - to - right ] , @xmath282 the expression above can be rewritten as @xmath283 set @xmath284 .
it follows from the discussion after lemma [ -left - to - right ] , that only the graphs @xmath285 can contribute to @xmath286 .
observe that all partial derivatives of @xmath287 are at most 1 in absolute value ( this follows form the inequality ( [ = normalize ] ) ) , and that any derivative of @xmath251 is a monomial with the coefficient not exceeding @xmath288 .
therefore the absolute value of the coefficient @xmath289 in ( [ = c - alpha - beta - gamma ] ) does not exceed @xmath103 .
using lemma [ -estimate - weight ] , we obtain @xmath290 now by lemma [ -count - graph ] , @xmath291 , hence @xmath292 where @xmath293 .
therefore , the series @xmath294 converges absolutely in the polydisk of radius @xmath295 .
hence in this neighborhood the formal series ( [ = bigsum ] ) defines a differential operator of the form @xmath296 , where @xmath297 is the operator of multiplication by some
_ analytic _ function .
the statement follows.[-star - for - lie ]
we start by establishing proposition [ -tau ] . [
proof of proposition [ -tau]]it suffices to show that the constants @xmath298 from @xcite satisfy estimates of the form @xmath299 for some constant @xmath300 .
according to @xcite , we have @xmath301 where `` the wheel '' @xmath302 is the graph from @xmath303 shown in figure [ -fig - wheel ] . [ ptbh ]
wheel.eps arguing as in the proof of lemma [ -estimate - weight ] , we conclude that @xmath304 , hence @xmath305 . by (
8.3.3 ) , if @xmath306 is regarded as a function on @xmath205 , then @xmath307 is given by the application of the infinite differential operator associated to the series @xmath26 from proposition [ -tau ] .
regarding @xmath244 as a distribution on @xmath0 , we conclude that @xmath308 we can now proceed with the proof of theorem [ -main - r ] .
let @xmath125 be a graph in @xmath215 ( @xmath309 ) . for each configuration in @xmath310 ( as opposed to @xmath311 earlier ) , we can draw @xmath125 in @xmath99 and calculate the edge angles .
this gives a map @xmath312 and we denote by @xmath313 the pullback of the volume form on the torus . next recall that @xmath314 ( the eye ) is the compactified configuration space of two points in @xmath99 .
we fix a path @xmath315\rightarrow\overline{c}_{2,0}$ ] , such that @xmath316 [ h ] eye1.eps now for each @xmath317 we have a `` forgetting '' map @xmath318 where we forget the location of first @xmath115 points in the configuration .
we denote by @xmath319 @xmath320 the preimage of the path @xmath321\right ) $ ] under the forgetting map and by @xmath322 its boundary .
the boundary @xmath322 can be written as the following cycle : @xmath323 where @xmath324 are the preimages of the points @xmath325 and @xmath326 respectively and @xmath327 is the sum ( with appropriate signs ) of the components of @xmath328 since @xmath313 is a closed @xmath223-form , by stokes theorem we get @xmath329 [ proof of theorem [ -main - r]]by @xcite , for @xmath330 we have @xmath331 & = ( r\tau)\star(p\tau)\text{,}\\ r\cdot\left [ \sum_{n=0}^{\infty}\frac{1}{n!}\sum_{\gamma\in a_{n}}% \partial_{\gamma}^{p}\left ( \int\limits_{z_{n}^{0}}\omega_{\gamma}^{\prime } \right ) \right ] & = ( r\ast_{\frak{g}}p)\tau\text{.}%\end{aligned}\ ] ] hence for @xmath332 as in the statement of theorem [ -main - r ] , we get @xmath333 \text{.}%\ ] ] we now discuss the structure of @xmath327 .
the relative positions of the vertices @xmath165 and @xmath167 are determined by the point @xmath334 .
the integral @xmath335 is the sum of integrals over the top ( i.e. , @xmath223-dimensional ) components of @xmath327 , which correspond to the following ( degenerate ) configurations : 1 .
two or more points cluster at @xmath165 .
2 . two or more points cluster at @xmath167 .
3 . two or more points cluster somewhere else in @xmath99 .
4 . one or more points cluster on @xmath112 . for components of type 4 @xmath336 ( since the angle for any edge originating from @xmath112 is identically 0 ) . for components of types 13 which involve clusters of _ three or more _ points , @xmath337 by ( * ? ? ?
* lemma 6.6 ) .
now let @xmath338 be a component corresponding to a two - point cluster of type 3 .
the corresponding boundary component of @xmath339 is @xmath340 and we have @xmath341 , where @xmath342 is the preimage of the path @xmath343 in @xmath344 under the forgetting map . as in @xcite ,
@xmath345 @xmath346 decomposes as a product of integrals over @xmath98 and @xmath342 .
the first integral vanishes unless two points in the cluster are connected by an edge . @xmath347 relabeling the vertices and edges as above , we see that the operator @xmath348 is a sum of terms with factors of the form @xmath349c_{ij}^{l}=\frac{1}{4}\left [ x_{k,}\left [ x_{i},x_{j}\right ] \right ] \text{.}%\ ] ] the set of graphs with the above subgraph can be grouped into sets of three corresponding to a cyclic permutation of the three outgoing edges from the cluster .
each of these graphs has the same weight @xmath345 @xmath346 . by the jacobi identity , the sum of the three corresponding operators @xmath348 vanishes . hence @xmath350
now we consider two - point clusters of type 2 .
@xmath351 relabeling the edges and vertices as above , we see that the corresponding operator is a sum of terms with factors of the form @xmath352
\text{,}%\ ] ] where @xmath353 \in\frak{g}$ ] acts as a differential operator .
but this expression is equal to @xmath354 ( where @xmath355 is the adjoint vector field corresponding to @xmath356 ) , which vanishes since @xmath251 is invariant ( @xmath34 ) .
this leaves only two - point clusters of type 1 .
@xmath357 arguing as above , we conclude that @xmath358 is a sum of terms containing factors of the type @xmath359 .
hence @xmath360is an infinite sum of operators from @xmath32 . by theorem [ -main - d ] and proposition [ -tau ]
@xmath35 belongs to @xmath17 .
therefore , @xmath361 , as was to be shown . * remark .
* in the above discussion we did not explicitly take into consideration the possibility of edges originating outside the cluster and terminating at the vertices of the cluster .
however , we can group together graphs which differ only by a permutation of the incident vertices in the cluster , and observe that these graphs have the same weight .
the result now follows from the leibniz formula .
[ -tang - for - lie ]
in this section we prove the kashiwara - vergne conjecture ( theorem [ -kv - conjecture ] ) and discuss its consequences .
we start by observing , that for @xmath362 and @xmath363 in @xmath0 , we have @xmath364 = x^{\prime}\star x^{\prime\prime } -x^{\prime\prime}\star x^{\prime}.\ ] ] then , by the universal property of @xmath365 @xmath366 it follows that there is a unique algebra homomorphism between @xmath42 and @xmath367 .
this morphism ( denoted @xmath368 in kontsevich ) is invertible , and we shall denote its inverse by @xmath369 .
then , as established in @xcite , we have @xmath370 [ proof of theorem [ -kv - conjecture]]in view of theorem [ -main - r ] , it suffices to establish the following identity ( in @xmath17 ) @xmath371 since both sides are germs of analytic differential operators , it is enough to verify that @xmath372 for all @xmath239
. then @xmath373 \\ & = ( r\ast_{\frak{g}}p)q-\exp^{\ast}(\kappa(r\tau)\ast_{g}\kappa(p\tau))\\ & = ( r\ast_{\frak{g}}p)q-\exp^{\ast}(\kappa(r\tau\star p\tau))\\ & = ( r\ast_{\frak{g}}p)q-\exp^{\ast}\left ( \eta\left ( ( r\tau\star p\tau ) \tau^{-1}\right ) \right ) \\ & = ( r\ast_{\frak{g}}p)q-(r\tau\star p\tau)\tau^{-1}q\\ & = \left [ ( r\ast_{\frak{g}}p)\tau-(r\tau\star p\tau)\right ] \tau^{-1}q\\ & = \left ( r\cdot t_{p}\right ) \tau^{-1}q\text{.}%\end{aligned}\ ] ] theorems [ -module - iso ] and [ -eigendist ] follow immediately from the theorem above , and theorem [ -fund - sol ] requires a short argument , which we give below .
[ proof of theorem [ -fund - sol]]a bi - invariant differential operator on @xmath6 is precisely an element of the center @xmath58 of @xmath42 , hence of the form @xmath374 for some @xmath34 .
let @xmath375 and @xmath376 be the delta - distributions supported at @xmath3 and @xmath377 respectively .
then if @xmath378 , by @xcite ( see also @xcite ) , there exists a germ @xmath379 , such that @xmath380 hence by theorem [ -module - iso ] , we get @xmath381 and we see that @xmath382 is a local fundamental solution for @xmath374.[-sec - kv ]
we briefly discuss basic facts about supermanifolds . the study of supermanifolds etc .
was initiated by berezin ( see @xcite , @xcite ) and independently by kostant ( @xcite )
. subsequently a slightly different ( though essentially equivalent ) treatment appeared in the physics literature @xcite .
we will follow the approach of @xcite .
the reader familiar with the physics point of view should have no trouble translating everything into that language .
a supermanifold ( @xmath383-graded manifold ) of dimension @xmath384 is a triple @xmath385 where @xmath386 is a @xmath387-dimensional smooth manifold , @xmath388 is a sheaf of @xmath383-graded commutative algebras and @xmath389 is a sheaf map from @xmath388 to the sheaf @xmath390 of smooth functions on @xmath386 .
moreover , we require that every open subset of @xmath386 can be covered by `` @xmath388-splitting open sets @xmath391 of odd dimension @xmath392 '' .
this means that we can choose ( non - canonical ) subalgebras @xmath393 and @xmath394 of @xmath395 such that @xmath396 @xmath397 and its dual @xmath398 are super ( i.e. , @xmath399-graded ) vector spaces .
a `` pure '' linear functional @xmath400 is called a differentiation at @xmath401 if@xmath402 for any pure @xmath188 and @xmath403 in @xmath397 . here
we write @xmath404 for @xmath405 and @xmath251 for the parity ( @xmath16 or @xmath46 ) .
an arbitrary @xmath406 is called a differentiation if its homogeneous components are differentiations , and we write @xmath407 for the super vector space of all differentiations at @xmath408 .
it is easy to see that @xmath409 _ { i}=d_{i}$ ] , @xmath410 .
moreover , we have a natural surjection @xmath411 @xmath412 ( ordinary tangent space of @xmath386 at @xmath408 ) with @xmath413 _ { 1}$ ] . for each @xmath414 the sheaf @xmath415 of differential operators of order @xmath416 on the supermanifold @xmath417
is defined in @xcite as follows:@xmath418 \in\left ( \operatorname*{diff}\nolimits_{k-1}\mathcal{a}\right ) ( u)\text { for all } f\in\mathcal{a}(u)\right\ } \text{.}%\end{aligned}\ ] ] here the bracket @xmath419 $ ] denotes the graded commutator in @xmath420 . for @xmath421 we can define its _ body _ as the unique differential operator @xmath422 on @xmath386 satisfying the conditions@xmath423 & = \widetilde{\left [ \partial , f\right ] } \text { \ for all } f\in\mathcal{a}(x)\text{.}%\end{aligned}\ ] ] this allows us to introduce a locally convex topology on @xmath397 , as in @xcite .
the space of distributions @xmath424 can then be defined as the topological dual of @xmath425 .
the support of a function @xmath426 is simply the support of @xmath427 , and the support of a distribution @xmath428 @xmath429 is the smallest closed set @xmath430 such that @xmath431 vanishes on all @xmath426 with @xmath432 . following @xcite , we define @xmath433as the subset of those linear functionals on @xmath397 whose kernel contains an ideal of finite codimension .
it is easy to check that @xmath434 is a graded co - commutative coalgebra .
we say that the supermanifold @xmath435 is a lie supergroup ( graded lie group in @xcite ) , if @xmath436 has the structure of a hopf algebra extending the natural coalgebra structure .
this implies that @xmath6 is a lie group and that the super tangent space @xmath437 at the identity @xmath438 has the structure of a lie superalgebra , i.e.,@xmath439 where @xmath440 is the lie algebra of @xmath6 .
moreover , we have a representation @xmath389 of @xmath6 on @xmath441 , whose differential is the adjoint action of @xmath440 on @xmath441 .
we have used the same letter @xmath389 to emphasize the fact that the supergroup can be reconstructed from the representation .
the hopf algebra @xmath436 can be obtained as the smash product of the group algebra @xmath442 and the universal enveloping algebra @xmath366 , and then @xmath443 can be recovered as in @xcite .
the action of a lie supergroup @xmath444 on a supermanifold is defined in @xcite .
each lie supergroup @xmath444 has an `` adjoint action '' on itself and on its lie superalgebra @xmath0 ( regarded as a supermanifold ) .
thus for each @xmath445we obtain derivations of the algebras @xmath446 and @xmath447 , which will both be denoted by @xmath448 .
as before , we regard the ( graded ) symmetric algebra @xmath449 as the convolution algebra of distributions on the super vector space @xmath0 with support at @xmath3 and denote by @xmath57 the subalgebra of @xmath450invariant distributions .
we regard the universal enveloping algebra @xmath39 as the convolution algebra of distributions on the lie supergroup @xmath435 with support at @xmath438 .
the center @xmath58 of @xmath42 can then be regarded as the subalgebra of @xmath451invariant distributions .
the exponential map from @xmath0 to @xmath444 is best described in terms of the corresponding `` pushforward '' map @xmath453 which is the usual exponential from @xmath454 to the group algebra @xmath442 , and the symmetrization map from @xmath18 to @xmath42 . as in the introduction ,
we denote by @xmath17 the algebra of germs at @xmath16 of differential operators on @xmath0 with analytic coefficients , and by @xmath32 the right ideal of @xmath17 generated by the germs of the differential operators
@xmath33 , @xmath28 .
let @xmath0 be a finite - dimensional real lie superalgebra .
then its dual @xmath457 is a poisson supermanifold ( cf .
the poisson bracket @xmath11 defined as the unique biderivation of @xmath458 satisfying@xmath459{ll}% 0 & \text{if } f_{1}\text { or } f_{2}\text { is constant}\\ \frac{1}{2}\left [ f_{1},f_{2}\right ] & \text{if } f_{1},f_{2}\in\frak{g}% \end{array } \right .
\text{.}%\ ] ] for convenience we fix bases @xmath460 of @xmath440 and @xmath461 of @xmath462 then in coordinates we have @xmath463\frac{\partial f_{1}% } { \partial x_{i}}\frac{\partial f_{2}}{\partial x_{j}}\text{.}%\ ] ] for an odd variable @xmath464 the vector field @xmath465 is odd ; nevertheless , @xmath11 is an _ even _ bidifferential operator and preserves the grading on @xmath458 .
following @xcite , we can define the @xmath1-product of @xmath466 by the formula@xmath467 the bidifferential operators @xmath160 for @xmath468 are given by the formula ( [ = ouch ] ) up to a sign which can be calculated in the following way ( illustrated by the example below ) .
we consider the same graph @xmath173 as in section [ ss - weights ] , i.e. @xmath175{ccc}% \overset{1}{\bullet } & \rightarrow & \overset{2}{\bullet}\\ \downarrow & \swarrow & \downarrow\\ \underset{\overline{1}}{\bullet } & & \underset{\overline{2}}{\bullet}% \end{array } \text{\ or ( relabeling)\ \ \ } % \begin{array } [ c]{ccc}% \overset{\gamma^{i_{1}j_{1}}}{\bullet } & \overset{_{j_{1}}}{\rightarrow } & \overset{\gamma^{i_{2}j_{2}}}{\bullet}\\ \,\downarrow^{_{i_{1 } } } & \overset{i_{2}}{\swarrow } & \downarrow^{_{j_{2}}}\\ \underset{f_{1}}{\bullet } & & \underset{f_{2}}{\bullet}% \end{array } .\ ] ] we first form the expression @xmath469 where the bivector fields @xmath470 corresponding to the vertices @xmath46 and @xmath471 are all written on the left ( since @xmath11 is even , the order is irrelevant )
. then we use the koszul rule of signs to move the derivatives to the appropriate positions ( as given by the graph ) , i.e. @xmath472 and @xmath473 should be moved to the left of @xmath166 , @xmath474 to @xmath168 and @xmath475 to @xmath476 .
we obtain the bidifferential operator , as in the formula ( [ = ouch - example ] ) before @xmath477 here the signs of the summands are given by @xmath478 to the power@xmath479 where @xmath480 is the parity of @xmath356 and @xmath481 is the parity of the @xmath482 . note that in the above formula ( [ = super - term ] ) we did not move @xmath472 past @xmath483 , thus preserving the same order as in ( [ = super - mnemonic ] ) .
let us write @xmath485 for the supertrace of an endomorphism of a graded vector space @xcite .
the expression @xmath486 $ ] defines an even element in @xmath447 . with @xmath25 as in ( [ = s1 ] ) , we can verify ( as in section [ -tang - for - lie ] ) that @xmath487 \right)\ ] ] converges to an even superfunction @xmath27 in @xmath447 which is analytic at @xmath3 .
the map @xmath491 is defined ( as before ) by @xmath492 and the restriction of @xmath56 to @xmath57 is an _ algebra _
isomorphism from @xmath57 to @xmath58 ( this extension of duflo s isomorphism to the case of lie superalgebras is due to kontsevich ) .
if a bi - invariant differential operator on a real lie supergroup @xmath6 has a nonzero body , it can be represented as @xmath374 , where @xmath34 and @xmath251 has a nonzero body .
then we can find a germ @xmath379 , such that @xmath380 local solvability of @xmath374 now follows as in the proof of theorem [ -fund - sol ] . | we study kontsevich s deformation quantization for the dual of a finite - dimensional real lie algebra ( or superalgebra ) @xmath0 . in this case
the kontsevich @xmath1-product defines a new convolution on @xmath2 , regarded as the space of distributions supported at @xmath3 . for @xmath4 , we show that the convolution operator @xmath5 is a differential operator with _ analytic _ germ .
we use this fact to prove a conjecture of kashiwara and vergne on invariant distributions on a lie group @xmath6 .
this yields a new proof of duflo s result on local solvability of bi - invariant differential operators on a lie group @xmath6 .
moreover , this new proof extends to lie supergroups . |
the ramsey number @xmath1 of a graph @xmath0 is the smallest integer @xmath6 such that every @xmath7-colouring of the edges of the complete graph @xmath8 on @xmath6 vertices contains a monochromatic copy of @xmath0 .
it has been known @xcite for a long time that the cycles @xmath3 on @xmath4 vertices have ramsey numbers linear in @xmath4 , while the ramsey numbers of complete graphs @xmath9 are exponential .
so from the beginnings of ramsey theory one important question has been hovering in the background : if we keep adding chords to @xmath3 until reaching @xmath9 , at which point will the ramsey number jump up ? very little is known .
when we are close to the complete graph , then , for @xmath10 it holds that @xmath11 , and it is conjectured ( see @xcite ) that this is also true for every @xmath12 . at the other end of the spectrum , that is , when we are close to the cycle , all that is known is the aforementioned result by balister , lehel , and schelp @xcite . using their terms
, we call a graph @xmath0 _ ramsey saturated _ if @xmath13 for every edge @xmath14 in the complement of @xmath0 , and _
ramsey unsaturated _ otherwise . in @xcite it
is proved that the cycle @xmath3 is ramsey unsaturated for every @xmath15 , and it is conjectured that one may add any edge @xmath14 to @xmath3 without increasing the ramsey number , unless @xmath4 is even and adding @xmath14 destroys the bipartiteness of @xmath3 ( conjecture 3 in @xcite ) .
balister , lehel , and schelp also remark ` one would expect to be able to add more than one edge to the cycle without increasing the ramsey number ' .
we show that this is indeed the case , at least asymptotically .
our main result is that one may add a linear number @xmath16 of chords to @xmath3 without changing the ramsey number , as long as the obtained graph @xmath0 has bounded maximum degree @xmath17 , and as long as @xmath0 is bipartite if @xmath4 is even , and almost bipartite if @xmath4 is odd .
( a non - bipartite graph is called _ almost bipartite _ if the removal of one vertex results in a bipartite graph . )
since our proof relies on the regularity method , our result holds only for large graphs , with the constant @xmath18 depending only on @xmath19 .
[ main1 ] for every @xmath20 there are @xmath21 and @xmath22 such that , for every @xmath23 and every collection @xmath24 of chords of @xmath3 with 1 .
@xmath25 and[thm1cond1 ] 2 .
@xmath26,[thm1cond2 ] the following two conditions are equivalent a. @xmath27 , b. either @xmath4 is even and @xmath28 is bipartite , or @xmath4 is odd and @xmath28 is almost bipartite .
observe that theorem [ main1 ] confirms conjecture 3 of @xcite , as an odd cycle plus one chord is almost bipartite .
when @xmath15 is even , @xmath29 ( see @xcite ) and the implication @xmath30 of theorem [ main1 ] can easily be seen by considering the following construction : colour all edges of a maximal cut in @xmath31 blue , and the remaining edges red . by ( a )
, there is a monochromatic copy of @xmath28 .
since the components of the red subgraph of @xmath31 are too small to contain @xmath28 , there is a blue copy of @xmath28 in @xmath31 .
thus @xmath28 is bipartite . for odd @xmath15 , we know that @xmath32 ( see @xcite ) .
take the above colouring of @xmath31 and add one new vertex to @xmath31 all whose incident edges are coloured blue .
then as above , we find a blue copy of @xmath28 .
this copy can not be bipartite since it contains the odd cycle @xmath3 , hence it must be almost bipartite .
the odd case of theorem [ main1 ] follows from a slightly more general result . as a generalization of almost - bipartiteness ,
let us call a graph _
@xmath33-almost bipartite _
( for @xmath34 ) if it contains a set of @xmath33 independent vertices such that the removal of these vertices leaves a bipartite graph , but the removal of any set of less than @xmath33 independent vertices does not .
then a graph is @xmath35-almost bipartite if , and only if , it is almost bipartite .
[ main2 ] for every @xmath36 and @xmath34 there are @xmath21 and @xmath22 such that , for every odd @xmath23 and every collection @xmath24 of chords of @xmath3 with 1 .
@xmath37 and 2 .
@xmath26 , the following two conditions are equivalent a. @xmath38 , b. @xmath28 is @xmath33-almost bipartite . again , the implication @xmath30 of theorem [ main2 ] is straightforward : suppose that @xmath39 and @xmath4 is large ( in particular , @xmath40 ) .
we partition the vertices of @xmath41 into sets @xmath42 , @xmath43 , and @xmath44 such that @xmath45 and @xmath46 and colour all edges within @xmath42 or @xmath43 or @xmath44 red and the rest blue .
clearly , the red subgraph has no large enough component to contain @xmath28 , so we find a blue copy of @xmath28 in @xmath47 .
since the blue subgraph is @xmath33-almost bipartite , we find that @xmath28 is @xmath48-almost bipartite for some @xmath49 . now ,
if @xmath50 , then we can use the implication @xmath51 for @xmath48 to obtain a contradiction .
hence , @xmath28 is @xmath33-almost bipartite .
both our results are , apart from the value of @xmath18 , best possible for every sufficiently large @xmath36 . to see this , we first recall the following result of graham , rdl and ruciski : [ thm : grr ] there exists a constant @xmath52 such that , for every @xmath53 and for every @xmath54 ( except for @xmath55 and @xmath56 ) , there exists a bipartite graph @xmath0 with @xmath4 vertices and maximum degree at most @xmath36 which satisfies @xmath57 .
let @xmath58 be such that @xmath59 , and let @xmath4 be even and larger than @xmath60 .
then theorem [ thm : grr ] assures that there is a bipartite graph @xmath0 on @xmath61 vertices and of maximum degree at most @xmath62 such that @xmath63 .
clearly , we can add @xmath0 into @xmath3 respecting the natural bipartition of @xmath3 .
denote the set of chords by @xmath24 and note that @xmath26 and @xmath64 .
on the other hand , we have @xmath65 hence , this construction shows that the function from condition of theorem [ main1 ] can not be improved to anything above @xmath66 . a similar construction works for large odd @xmath4 .
in this case , let @xmath58 be such that @xmath67 , @xmath68 , and let @xmath0 be a bipartite graph on @xmath69 vertices with @xmath70 and @xmath71 .
let @xmath72 be the ( consecutive ) vertices of @xmath3 .
we add @xmath33 chords @xmath73 , @xmath74 and add @xmath0 into the path @xmath75 , respecting its natural bipartition .
denote the set of thus created chords by @xmath24
. then @xmath28 had maximum degree @xmath36 , it is @xmath33-almost bipartite because it contains @xmath33 vertex - disjoint triangles @xmath76 , @xmath77 and removing vertices @xmath78 from @xmath28 yields a bipartite graph , and @xmath79 . on the other hand
, we have @xmath80 furthermore , theorem [ thm : grr ] also implies that theorems [ main1 ] and [ main2 ] can not be true for non - constant @xmath36 .
suppose that @xmath81 as @xmath82 arbitrarily slowly .
note that we may assume that @xmath83 , because for faster growing @xmath36 , or more precisely , for all @xmath84 , we can add @xmath85 to @xmath3 so that the resulting graph has maximum degree at most @xmath36 and it is bipartite , but its ramsey number is at least @xmath86 for large @xmath4 ( cf . chung and graham @xcite ) .
now , let @xmath87 be such that @xmath88 for every @xmath89 .
let @xmath0 be a bipartite graph on @xmath90 vertices and of maximum degree at most @xmath36 such that @xmath91 .
again , we can add @xmath0 into @xmath3 respecting the natural bipartition of @xmath3 .
denote the set of chords by @xmath24 and note that @xmath92 . on the other hand
, we have @xmath93
we use standard graph - theoretic notation . for a graph @xmath94 ,
let @xmath95 denote the set of its vertices and @xmath96 the set of its edges .
we denote by @xmath97 and @xmath98 the number of vertices and edges of @xmath94 . given a set @xmath99 of vertices and a set @xmath100 of edges , @xmath101 $ ] stands for the subgraph of @xmath94 induced by the vertices of @xmath102 , and @xmath103
is the subgraph of @xmath94 with vertex set @xmath95 and edge set @xmath104
. given two disjoint sets of vertices , @xmath102 and @xmath105 , we write @xmath106 for the set of edges with one endpoint in @xmath102 and the other in @xmath105 , we set @xmath107 , and call @xmath108 $ ] the bipartite subgraph of @xmath94 with bipartition @xmath109 and edge set @xmath106 . the quantity
@xmath110 is the _ density _ of the pair @xmath111 . for a vertex @xmath112 and a set @xmath105 ,
we denote by @xmath113 the set of all vertices adjacent to @xmath112 and set @xmath114 .
next , let us clarify some notation on paths : the _ length _ of a path is the number of edges it contains . an _ odd path _ is a path of odd length , an _ even path _ is a path of even length .
a path is _ trivial _ if it consists of one vertex only .
an _ @xmath115@xmath116 path _ is a path starting at vertex @xmath115 and ending at vertex @xmath116 .
if @xmath115 and @xmath116 are vertices of some path @xmath117 , then we denote by @xmath118 the subpath of @xmath117 starting at @xmath115 and ending at @xmath116 .
if @xmath117 is a path with one endvertex in a set @xmath102 and the other one in a set @xmath105 and does not meet @xmath119 otherwise , then we say that @xmath117 is an _
@xmath102@xmath105 path_. for a set @xmath102 , a non - trivial @xmath102@xmath102 path is also called an _
@xmath102-path_. we finish this subsection by quickly proving two easy facts , which will be needed in section [ preparing ] .
[ bipa ] let @xmath94 be a bipartite graph , and let @xmath120 be such that all @xmath121-paths of @xmath122 are odd .
then @xmath0 has a bipartition @xmath123 of its vertex set so that each @xmath121-path in @xmath122 is a @xmath124@xmath125 path .
take any bipartition of @xmath94 and consider its restriction @xmath123 to the vertices of @xmath0 .
since the endpoints of every odd path can not be in the same partite set , the statement of the fact follows .
[ tripa ] let @xmath126 be a tripartite graphs such that @xmath94 is obtained from @xmath127 by adding at most @xmath128 pairwise internally vertex disjoint @xmath129-paths , each of odd length at least @xmath130 , @xmath131 .
then there is a graph @xmath0 with @xmath132 and @xmath133 which has a tripartition @xmath134 of its vertex set such that each @xmath121-path in @xmath122 is an odd @xmath124@xmath125 path of length at least @xmath135 .
let @xmath127 have colour classes @xmath136 , @xmath137 and @xmath138 .
for each @xmath129-path @xmath139 in @xmath140 that has at least one endvertex ( @xmath112 or @xmath141 ) in @xmath138 , we add both @xmath142 and @xmath116 to @xmath127 .
more precisely , if @xmath143 , then we add @xmath142 to @xmath136 and @xmath116 to @xmath137 , and if only one of @xmath112 , @xmath141 lies in @xmath138 , say @xmath144 , and @xmath145 , @xmath146 , then we add @xmath142 to @xmath147 and @xmath116 to @xmath148 .
call the obtained tripartite graph @xmath0 , with partition classes @xmath124 , @xmath125 and @xmath149 . clearly , the @xmath121-paths in @xmath122 all have odd length @xmath150 , and since @xmath0 is obtained from @xmath127 by adding two vertices from each @xmath129-path , we also have @xmath133 .
we now present some structural results we will use , and which come from @xcite .
we also prove a corollary of one of these theorems , which will be tailor - made for our purposes .
the first two are well - known theorems of erds and gallai .
[ thm : eg ] every graph @xmath94 with at least @xmath151 edges contains a cycle of length at least @xmath152 .
[ thm : eg2 ] let @xmath94 be a graph with @xmath153 and minimum degree at least @xmath154 .
then , for every two vertices @xmath155 and @xmath156 , there exists an @xmath155@xmath156-path containing all vertices of @xmath94 .
the next result is a recent one on embedding bounded degree graphs .
[ genembed ] given a natural number @xmath157 and any @xmath158 , let @xmath159 be an @xmath4-vertex graph in which every vertex has degree at least @xmath160 , and all but at most @xmath161 vertices have degree at least @xmath162 .
let @xmath163 be any @xmath4-vertex graph with @xmath164 .
then @xmath165 .
we also need a well - known result of graham , rdl and ruciski on linear bounds for ramsey numbers of bounded degree graphs .
[ linram ] there exists a constant @xmath166 such that for all @xmath167 and all @xmath54 , every graph @xmath0 with @xmath4 vertices and maximum degree at most @xmath36 satisfies @xmath168 . finally , we will need stability results on monochromatic cycles in graphs whose edges are multicoloured .
( a multi - colouring of the edges of some graphs is an assigment of colours to the edges in which an edge can have more than one colour . )
it will come as no surprise that there are two cases , according to the parity of the cycle .
we start with the odd case : [ jmy - odd ] for @xmath169 , let @xmath170 be integers with @xmath171 and @xmath172 .
suppose that @xmath94 is a graph on @xmath173 vertices with @xmath174 .
then every multi - colouring of the edges of @xmath94 with @xmath7 colours contains either a monochromatic odd cycle @xmath175 for some @xmath176 or there is a vertex @xmath177 , a partition of @xmath178 into two sets @xmath179 and @xmath180 , and an @xmath181 such that a. all edges of @xmath182 \cup g[v_2]$ ] have only colour @xmath183 , b. all edges of @xmath184 $ ] have only colour @xmath185 , and c. @xmath186.[sizev1v2 ] a similar result holds for even cycles .
[ jmy - even ] -3pt for each @xmath187 , there is an @xmath188 such that the following holds for all integers @xmath189 with @xmath190 and @xmath191 .
+ if the edges of a graph @xmath94 on @xmath192 vertices with @xmath193 are multi - coloured with @xmath7-colours , then @xmath94 either contains a monochromatic even cycle @xmath175 for some @xmath194 or there is a partition of @xmath95 into sets @xmath179 , @xmath180 and @xmath195 , and an @xmath181 such that a. all but at most @xmath196 edges of @xmath182 $ ] have only colour @xmath183,[thmevena ] b. all but at most @xmath197 edges of @xmath184 $ ] have only colour @xmath198 , and[thmevenb ] c. @xmath199 , @xmath200 , @xmath201.[thmevenc ] in addition to theorem [ jmy - even ] , we will use its following corollary : [ coro - even ] the sets @xmath179 and @xmath180 from theorem [ jmy - even ] contain sets @xmath102 and @xmath105 , respectively , such that 1 .
the minimum degree in colour @xmath183 in @xmath101 $ ] is at least @xmath202 , the maximum degree in colour @xmath198 in @xmath101 $ ] is at most @xmath203 , 2 . in colour @xmath198 in @xmath108 $ ] , every vertex of @xmath102 has degree at least @xmath204 and every vertex of @xmath105 has degree at least @xmath202 , and 3 .
@xmath205 , @xmath206 .
observe that , firstly , condition ( a ) implies that , in @xmath179 , there are at most @xmath207 vertices that , in colour @xmath198 , have degree at least @xmath208 inside @xmath179 .
secondly , ( b ) implies that @xmath179 contains at most @xmath208 vertices sending at least @xmath209 edges of colour @xmath183 to @xmath180 .
finally , ( b ) implies that , in @xmath180 , there are at most @xmath209 vertices that send at least @xmath208 edges of colour @xmath183 to @xmath179 .
remove these in total at most @xmath210 vertices from @xmath179 and at most @xmath209 vertices from @xmath180 to obtain sets @xmath102 and @xmath105 as required .
in fact , by construction , each vertex of @xmath102 is incident with at most @xmath211 edges of colour @xmath198 that go to other vertices of @xmath102 .
thus ( a ) holds .
similarly we can see that ( b ) holds . for ( c )
, we have to see that @xmath212 and @xmath213 .
this is true as @xmath214 and @xmath215 by construction . using ( c ) ,
we thus obtain ( c ) . in this section
we introduce some very well - known concepts ; the reader familiar with regularity is invited to skip this section .
we start with giving the standard definition of regularity . given @xmath216 and disjoint subsets @xmath109 of the vertex set of a graph @xmath94
, we say that the pair @xmath217 is @xmath218__-regular in _ _
@xmath94 if , for every pair @xmath219 with @xmath220 , @xmath221 , @xmath222 , @xmath223 , we have @xmath224 when there is no danger of confusion , we simply say that the pair @xmath111 is @xmath225-regular .
we now turn to the regularity lemma .
[ regu ] for every @xmath216 and @xmath226 , there are @xmath227 and @xmath228 such that , for each graph @xmath94 on at least @xmath229 vertices , there exists a partition @xmath230 of @xmath95 such that * @xmath231 ; * @xmath232 and @xmath233 ; * for each @xmath183 , @xmath234 , all but at most @xmath235 of the pairs @xmath236 , @xmath237 , are @xmath218-regular .
we shall refer to a partition of @xmath95 satisfying coditions ( i)(iii ) as @xmath218-regular with respect to @xmath94 .
we continue by stating two facts on @xmath225-regular pairs , which are very well - known , and can be found for example in @xcite .
[ fact1 ] let @xmath238 be an @xmath218-regular pair of density @xmath239 in some graph @xmath94 . then all but at most @xmath240 vertices @xmath241 are such that @xmath242
we shall refer to such a vertex @xmath243 as @xmath218-typical in @xmath94 with respect to @xmath180 .
[ fact2 ] let @xmath244 and let @xmath238 be an @xmath218-regular pair of density @xmath245 .
then , for every @xmath246 , @xmath247 such that @xmath248 and @xmath249 , the pair @xmath250 is @xmath251-regular with density at least @xmath252 .
we now state the so - called embedding lemma .
given a graph @xmath253 and a positive integer @xmath254 , let @xmath255 be the graph obtained from @xmath253 by replacing each vertex @xmath156 of @xmath253 with an independent set @xmath256 of size @xmath254 , and each edge @xmath257 of @xmath253 with the complete bipartite graph on @xmath256 and @xmath258 . [ blow - up ] for all @xmath20 and @xmath259 , if @xmath260 is such that @xmath261 , then the following holds for every graph @xmath0 of maximum degree @xmath36 .
let @xmath253 be a graph on vertices @xmath262 , and let @xmath94 be a graph on the union of the sets @xmath263 , where each @xmath264 has the same size @xmath265 , and for each edge @xmath266 the pair @xmath267 is @xmath268-regular and of density at least @xmath269 .
suppose @xmath270 .
if @xmath271 , then also @xmath120 .
in particular , we note that the conditions of lemma [ blow - up ] are satisfied if @xmath272 in this section we prove an important lemma , lemma [ lem : path ] , which shall be used later in the proof of our main theorems .
this lemma is about embedding a family of long paths into a sequence of @xmath225-regular pairs . as a starting point
, we use a lemma from @xcite that shows under which condition we can embed a path into one regular pair .
[ benevides ] for every @xmath273 and for every @xmath274 , there is an @xmath275 with the following property : let @xmath276 and let @xmath277 be an @xmath278-regular pair ( in some graph ) with density at least @xmath279 and with @xmath280 .
then , for every @xmath281 , @xmath282 , and for every two vertices @xmath283 such that @xmath284 , there exists a @xmath243@xmath285-path of length @xmath286 .
we now show how to embed more paths into a longer sequence of regular pairs .
[ lem : path ] let @xmath94 be a graph , let @xmath287 , and let @xmath288 be a collection of pairwise internally vertex disjoint paths of odd lengths .
suppose that @xmath289 is odd and @xmath290 are such that a. @xmath291;[lem : pathe ] b. @xmath292 are pairwise disjoint , except ( possibly ) for @xmath179 and @xmath293;[lem : pathh ] c. for all @xmath294 $ ] , the pair @xmath295 is @xmath225-regular of density at least @xmath296;[lem : pathf ] d. @xmath297;[lem : pathg ] e. for all @xmath298 $ ] , @xmath299;[lem : patha ] f. @xmath300 ; [ lem : pathb ] g. for all @xmath298 $ ] ,
the first vertex of @xmath301 is embedded in a vertex @xmath302 and the last vertex of @xmath301 is embedded in a vertex @xmath303 ; and[lem : pathc ] h. for all @xmath298 $ ] , @xmath304 is @xmath225-typical in @xmath94 with respect to @xmath180 and @xmath305 is @xmath225-typical in @xmath94 with respect to @xmath306.[lem : pathd ] then we can extend @xmath307 to an embedding of all of @xmath308 in @xmath94 such that the only vertices from @xmath309 used are those from @xmath102 .
notice that the sets @xmath179 and @xmath293 can have a non - empty intersection .
consequently , in such a case , it is possible to choose @xmath310 and obtain a cycle of length @xmath311 instead of path of length @xmath311 .
however , in this paper , we will always make sure that @xmath312 . for every @xmath313 $ ] and for all @xmath314 $ ] , we choose an odd number @xmath315 that will indicate how many edges of @xmath301 we plan to embed into the pair @xmath316 .
we choose the numbers @xmath315 so that they satisfy the following three conditions : a. @xmath317 if @xmath318 is odd , and @xmath319 if @xmath318 is even ; [ even1 ] b. @xmath320 for each @xmath314 $ ] ; and [ even2 ] c. @xmath321 for each even @xmath322$].[fitsinthepair ] observe that such a choice is possible because of asumptions and . now , for @xmath323 $ ] and @xmath314 $ ] , set @xmath324 and let @xmath325 denote the @xmath326-th vertex on @xmath301 . in particular , the preimage of @xmath327 is @xmath328 and the preimage of @xmath329 is @xmath330 .
next , successively , for each @xmath314 $ ] , we embed @xmath301 in the following way : for each @xmath331 , we will consider certain sets @xmath332 .
the idea is that , for even @xmath318 , the sets @xmath333 and @xmath334 contain possible images for @xmath335 .
the sets @xmath333 will satisfy the following conditions .
a. @xmath336 , @xmath337 , and @xmath338 for each @xmath339 ; b. @xmath340 contains no vertex of @xmath341 , for any @xmath342 ; c. @xmath343 for each @xmath342 ; d. @xmath344 for @xmath342 ; and e. @xmath345 for each even @xmath346.[gleichgross ] observe that ( g ) , together with fact [ fact2 ] , ensures that , for @xmath347 , @xmath348 first , assuming the existence of the sets @xmath333 , let us embed all the @xmath301 .
we will show later that the sets @xmath332 satisfying ( d)(h ) do exist .
we shall start with the vertices @xmath325 , which will be placed in the following way . for @xmath349
, the vertex @xmath350 is embedded in a vertex @xmath351 of @xmath340 such that @xmath352 by fact [ fact1 ] , all but at most @xmath353 vertices in @xmath340 satisfy . in order to see ,
observe that the sets of vertices satisfying for @xmath318 and for @xmath354 are large enough to apply @xmath251-regularity of @xmath355 , hence there must be an edge @xmath356 between them .
( note that condition holds also for @xmath357 because of ( d ) . )
the last step of our embedding is the embedding of the subpaths @xmath358 for each even @xmath323 $ ] . by
we know that @xmath358 has odd length @xmath359 .
we wish to embed the edges of @xmath358 into the pair @xmath355 . for this
, we plan to use lemma [ benevides ] with @xmath360 and @xmath361 .
the assumptions of lemma [ benevides ] are verified in , ( f ) , ( h ) and , and the only condition being left to check is that @xmath362 . but this follows from @xmath363 this completes the embedding of the paths @xmath301 .
it only remains to show that the sets @xmath332 satisfying ( d)(h ) do exist .
let @xmath364 be given and assume we have already embedded all paths @xmath365 with @xmath366 .
let @xmath367 be the set of all vertices used by our embedding of @xmath368 .
now , for @xmath369 , choose @xmath370 and , for @xmath339 , @xmath371 as large as possible , but such that @xmath345 for each even @xmath346 .
this choice clearly guarantees conditions ( d ) , ( e ) , and ( h ) . to show ( g ) , observe that , for @xmath369 , we have @xmath372 furthermore , for @xmath373 , we know by that @xmath374 thus ( g ) is established . in a similar way we see ( f ) .
for all even @xmath322 $ ] , we have @xmath375 , h\neq i } \frac{q^h_j+1}{2}\\ & \geq \frac{1}{1 - 2\eps^{1/4 } } \frac{q^i_j+1}{2}. \end{aligned}\ ] ] for odd @xmath322 $ ] it suffices to notice that @xmath376 by .
we shall prove at the same time , and with the same parameters @xmath18 and @xmath87 , theorem [ main1 ] for even @xmath23 and theorem [ main2 ] for odd @xmath23 .
recall that we only need to worry about the implication ( b ) @xmath377 ( a ) .
suppose that @xmath378 and , for odd @xmath4 , also a fixed @xmath379 are given .
set @xmath380 where the constant @xmath381 comes from theorem [ linram ] , and set @xmath382 where @xmath166 comes from theorem [ linram ] and @xmath383 is given by theorem [ jmy - even ] for the input @xmath384 for the parameters @xmath218 and @xmath385 the regularity lemma ( theorem [ regu ] ) yields numbers @xmath227 and @xmath228 .
we now set @xmath386 and @xmath387 where @xmath388 is given by lemma [ benevides ] for @xmath389 and @xmath390 .
further , let @xmath24 be a set of chords of @xmath3 satisfying assumptions [ thm1cond1 ] and [ thm1cond2 ] of theorems [ main1 ] and [ main2 ] , and such that @xmath28 is bipartite if @xmath4 is even and @xmath33-almost bipartite if @xmath4 is odd .
our aim is to find a monochromatic copy of @xmath28 in @xmath8 .
the regularity lemma ( theorem [ regu ] ) applied to @xmath392 yields a partition @xmath394 , @xmath395 of @xmath396 that is @xmath225-regular with respect to @xmath392 .
it is well - known that this partition is also @xmath225-regular with respect to @xmath393 . substituting each @xmath264 , @xmath397 , with a vertex @xmath183 and each @xmath225-regular pair @xmath236 with an edge @xmath398
, we obtain the so - called reduced graph @xmath253 with vertex set @xmath399=\{1 , 2,\dots , t\}$ ] such that we define an edge-(multi)colouring of @xmath253 as follows .
colour the edge @xmath266 red ( blue ) if the density of @xmath236 in @xmath392 ( in @xmath393 ) is at least @xmath245 , where @xmath403 notice that every edge of @xmath253 receives a colouring because @xmath404 .
set @xmath406 and , since @xmath25 , notice that @xmath407 let @xmath408 be the subgraph of @xmath28 induced by @xmath24 and all @xmath409-paths in @xmath3 of length at most @xmath141 . then @xmath410 now , @xmath411 is the union @xmath412 of several @xmath409-paths , each of them longer than @xmath141 .
we wish to have them all of the same parity .
for this reason , for each such even path @xmath413 , we remove vertex @xmath285 and edge @xmath414 from @xmath117 , add them to @xmath408 , and call the obtained graph @xmath127 .
observe that we have added at most @xmath415 new vertices .
thus , @xmath416 and @xmath417 if @xmath4 is even , then @xmath28 , and thus also @xmath127 , is bipartite .
by fact [ bipa ] and by , we can arrange the colour classes @xmath124 , @xmath125 of @xmath127 so that @xmath418 is the edge - disjoint union of @xmath124@xmath125-paths of length at least @xmath419 .
set @xmath420 .
if @xmath4 is odd , then , since @xmath421 is @xmath33-almost - bipartite , the graph @xmath127 is @xmath422-colourable .
by fact [ tripa ] , there is a supergraph @xmath0 of @xmath127 contained in @xmath28 , with tripartition @xmath134 of its vertex set such that all @xmath121-paths in @xmath423 are odd @xmath124@xmath125 paths of length at least @xmath424 .
so , in both cases we have obtained a graph @xmath0 with partition classes @xmath124 , @xmath125 and ( if @xmath4 is odd ) @xmath149 such that , by and by , @xmath425 furthermore , for each @xmath121-path @xmath117 of @xmath426 , we have that @xmath427 and , since @xmath428 by , also @xmath429 we wish to apply theorem [ jmy - even ] to @xmath430 with @xmath431 ( as fixed above ) , @xmath432 , and @xmath433 . clearly , by and by the choice of @xmath385
, we have that @xmath434 and @xmath435 thus , as @xmath436 , we obtain that @xmath437 finally , since @xmath438 , we know that @xmath439 1 . [ evencycle ] @xmath253 has a monochromatic even cycle @xmath175 with @xmath440 , or 2 .
[ evenstructure ] there is a colour @xmath181 , and sets @xmath441 so that , for @xmath442 , 1 . the minimum degree in colour @xmath183 in @xmath443 $ ] is at least @xmath444 and the maximum degree in colour @xmath198 in @xmath443 $ ] is at most @xmath445 , 2 . in colour @xmath198 in @xmath446 $ ] , every vertex in @xmath102 has degree at least @xmath447 and every vertex in @xmath105 has degree at least @xmath444 , 3 .
@xmath448 and @xmath449 .
in this case , @xmath253 contains an even monochromatic ( say , red ) cycle @xmath175 with vertices @xmath450 ( in this order ) . let @xmath451 , @xmath452 contain all those vertices that are @xmath225-typical in @xmath392 with respect to @xmath180 or @xmath453 , respectively .
thus , because of fact [ fact1 ] , we have @xmath454 it follows from the definition of @xmath225-regularity that the pair @xmath455 is @xmath456-regular with density at least @xmath457 .
moreover , by condition 2 of theorems [ main1 ] and [ main2 ] , we know that @xmath458 .
thus we verified almost all conditions of lemma [ blow - up ] , which we would like to use with parameters @xmath36 , @xmath459 , @xmath460 , @xmath265 , @xmath461 and graphs @xmath462 , @xmath0 , @xmath463 $ ] in order to embed the bipartite graph @xmath0 into @xmath464 $ ] . by , it is only left to check that @xmath465 by , we know that the edges of @xmath426 form internally vertex - disjoint paths @xmath467 such that for all @xmath314 $ ] the first vertex of @xmath301 is mapped to @xmath302 , while the last vertex of @xmath301 is embedded in @xmath303 . by the choice of @xmath468 and @xmath469 , we know that , for each @xmath314=\{1,\ldots k\}$ ] , @xmath470 let us next establish . for this , it is enough to see that , as @xmath473 , we have @xmath474 & \leq \frac 23\cdot \frac{n}{1-\eps}\\[6pt ] & \overset{\eqref{sizevi}}\leq\frac 23\cdot \frac{|v_1|t}{(1-\eps)^2}\\[6pt ] & \leq\frac 23 \cdot\frac{t}{1 - 3\eps^{1/4 } } ( 1 - 2\epsilon^{1/4})|v_1| \\[6pt ] & \leq\left(\frac23(1+\eps^{1/5})t-2\right)(1 - 2\epsilon^{1/4})|v_1|\\[6pt ] & \overset{\eqref{evencycle}}\leq ( \ell -2)(1 - 2\epsilon^{1/4})|v_1| . \end{aligned}\ ] ] finally , for note that @xmath475 & \geq 4\epsilon^{1/2}(1-\eps)\frac { n_0}t\\[6pt ] & \overset{\eqref{c}}\geq\frac{m_{\rm reg}n_{\ref{benevides}}}t \\[6pt ] & \overset{\eqref{rlarge}}\geq n_{\ref{benevides}}.\end{aligned}\ ] ] we assume the colour @xmath183 is red and colour @xmath198 to be blue .
let @xmath476 denote the union of all vertices that belong to clusters in @xmath102 .
we estimate the number of blue edges in @xmath476 as follows : there are * at most @xmath477 blue edges within clusters in @xmath102 ; * at most @xmath478 blue edges in irregular pairs in @xmath102 ; * at most @xmath479 blue edges in pairs corresponding to blue edges in @xmath102 ; * at most @xmath480 blue edges in the remaining pairs in @xmath102 ( those which are coloured only by red ) . using the facts that @xmath481 , that @xmath482 , and that @xmath483 , we find that @xmath476 contains in total at most @xmath484 blue edges .
since the size of @xmath476 is bounded by @xmath485 we obtain that @xmath476 contains a subset @xmath486 of size at least @xmath487 such that , in @xmath392 , each vertex of @xmath486 has degree at least @xmath488 .
let @xmath489 be the set of all vertices in @xmath490 that send at least @xmath491 red edges to @xmath486 and suppose that @xmath492 .
since @xmath493 we may use lemma [ genembed ] and conclude that the red graph @xmath494 $ ] contains @xmath28 .
so let us from now on assume that @xmath495 .
setting @xmath496 , this means that @xmath497 .
moreover , by definition of @xmath498 , each vertex from @xmath498 sends at least @xmath499 blue edges to @xmath486 .
we embed @xmath28 in @xmath500 $ ] as follows .
say @xmath28 has the bipartition classes @xmath42 and @xmath43 . we know that @xmath501 .
now , embed all of @xmath42 into @xmath498 arbitrarily .
then , we embed the vertices in @xmath43 successively in the following way : say we wish to embed some vertex @xmath502 .
vertex @xmath116 has at most @xmath36 neighbours in @xmath42 .
the images of these neighbours in @xmath498 have at least @xmath503 common blue neigbors in @xmath486 .
since we have used at most @xmath504 vertices of @xmath486 in earlier steps , there is a vertex in @xmath486 we may use to embed @xmath116 .
thus we manage to embed all of @xmath43 as desired in @xmath486 , which finishes the embedding of @xmath28 . in this case , we wish to apply theorem [ jmy - odd ] with parameters @xmath505 , @xmath506 , and @xmath507 to the graph @xmath430 . from and
, we get that @xmath508 , and , clearly , we have @xmath509 . finally , the fact that @xmath510 follows from . 1 .
[ oddcycle ] @xmath253 has a monochromatic odd cycle @xmath175 with @xmath511 , or 2 .
[ oddstructure ] there is a colour @xmath181 , and sets @xmath441 so that 1 .
all edges in @xmath443 $ ] and in @xmath512 $ ] have only colour @xmath183,[d1 ] 2 .
all edges between @xmath102 and @xmath105 have only colour @xmath198 , and[d2 ] 3 .
@xmath513.[d3 ] suppose that @xmath175 is red and contains the vertices @xmath450 ( in this order ) .
first of all , observe that since @xmath281 is odd , the set of all @xmath7-chords of @xmath175 span an odd cycle @xmath514 of the same length in @xmath253 . now
, we split our proof further depending on whether or not @xmath514 contains a red @xmath7-chord .
let @xmath515 be the red @xmath7-chord of @xmath175 .
our plan is to use the lemma [ blow - up ] to embed @xmath0 into the @xmath225-regular triangle @xmath516 such that all vertices of @xmath0 that have neighbours in @xmath517 are mapped to vertices of @xmath293 or @xmath180 that are typical in red with respect to @xmath306 or to @xmath518 , respectively . for this , let @xmath452 and @xmath519 consist of all vertices that are @xmath225-typical in @xmath392 with respect to @xmath453 or to @xmath518 , respectively .
then , by fact [ fact1 ] , we have @xmath520 from the definition of @xmath225-regularity we have that the pairs @xmath521 , @xmath522 $ ] and @xmath523 are @xmath456-regular , and from the even case implies that @xmath524 .
hence lemma [ blow - up ] with parameters @xmath36 , @xmath459 , @xmath460 , @xmath265 , @xmath461 and graphs @xmath525 , @xmath0 , @xmath526 $ ] yields an embedding of the tripartite graph @xmath0 into @xmath94 .
recall that the tripartition classes of @xmath0 were @xmath124 , @xmath125 and @xmath149 .
we map @xmath124 to @xmath179 , @xmath125 to @xmath527 and @xmath149 to @xmath179 . by
, we know that the edges of @xmath426 form internally vertex - disjoint odd paths @xmath467 such that , for all @xmath314 $ ] , the first vertex of @xmath301 is mapped to some @xmath530 , while the last vertex of @xmath301 is embedded to some @xmath303 . by the choice of @xmath531 and @xmath469 , we know that , for each @xmath314 $ ] , we have that @xmath532 let @xmath246 be the set of all vertices in @xmath179 that are typical in red with respect to @xmath293 and @xmath180 , and typical in blue with respect to @xmath453 and @xmath518 .
since @xmath534 by fact [ fact1 ] , we have @xmath535 hence , by theorem [ linram ] , we find a monochromatic copy of @xmath0 inside @xmath468 .
now only have to embed the long connections ( in @xmath392 if our copy of @xmath0 is red , and in @xmath393 if our copy of @xmath0 is blue ) .
this will be done as before , either in @xmath175 or in @xmath536 , with the help of lemma [ lem : path ] .
now the input for lemma [ lem : path ] is @xmath537 and @xmath538 for @xmath323 $ ] and @xmath539 .
* at most @xmath477 blue edges within the clusters in @xmath102 ; * at most @xmath478 blue edges in irregular pairs in @xmath102 ( i.e. non - edges of @xmath443 $ ] , here we use ) ; * at most @xmath480 blue edges in the remaining pairs in @xmath102 ( here we use ( d1 ) ) . using the facts that @xmath481 , that @xmath482 , and that @xmath546 , we find that @xmath476 contains in total at most @xmath547 blue edges . in the same way we obtain that @xmath540 contains at most @xmath548 blue edges .
using the facts that @xmath550 , that @xmath551 , and that @xmath546 , we find that @xmath552 $ ] contains at most @xmath553 red edges .
we remove at most @xmath554 vertices with degree at least @xmath555 in one of @xmath556 $ ] , @xmath557 $ ] , and @xmath558 $ ] , and obtain sets @xmath559 and @xmath560 such that @xmath561 and such that the graphs @xmath556 $ ] , @xmath557 $ ] , and @xmath558 $ ] have maximum degree at most @xmath555 .
set @xmath562 , and let @xmath563 and @xmath564 be the sets of all vertices in @xmath565 that send at least @xmath566 red edges to @xmath486 or to @xmath498 , respectively . if @xmath567 , then we use lemma [ genembed ] ( note that @xmath568 by and by ) and conclude that the red graph @xmath569 $ ] contains @xmath28 .
so let us from now on assume that @xmath570 .
this means that there is a set @xmath571 of @xmath572 vertices which each send at least @xmath573 blue edges to @xmath486 and at least @xmath574 blue edges to @xmath498 .
let @xmath575 and @xmath576 be the set of their common blue neighbours in @xmath486 , or in @xmath498 , respectively .
note that @xmath577 where the last inequality follows from .
we embed @xmath28 as follows : let @xmath44 be the set of @xmath33 independent vertices such that @xmath578 is a bipartite graph with bipartition @xmath42 and @xmath43 . clearly
, the sets @xmath42 and @xmath43 differ in size by at most @xmath33 , and therefore we have that @xmath579 .
embed all of @xmath44 into @xmath571 and all of @xmath42 into @xmath576 arbitrarily ( that @xmath42 fits into @xmath576 in ensured by ) . then embed the vertices in @xmath43 successively , similarly as in the even case . for each vertex
@xmath502 that we wish to embed in @xmath575 , we know that the images of its at most @xmath36 neighbours have at least @xmath580 common neighbours in @xmath575 ( the last inequality follows from and from ) , and thus there is space for embedding @xmath116 properly .
komls , j. and simonovits , m. , _
szemerdi s regularity lemma and its applications in graph theory _ , combinatorics , paul erds is eighty , vol . 2 ( keszthely , 1993 ) , bolyai soc . math . stud .
, vol . 2 , jnos bolyai math .
budapest , 1996 , pp .
295352 .
szemerdi , e. , _ regular partitions of graphs _ ,
problmes combinatoires et thorie des graphes ( colloq .
cnrs , univ .
orsay , orsay , 1976 ) , colloq .
cnrs , vol .
260 , cnrs , paris , 1978 , pp . | we call a graph @xmath0 _ ramsey - unsaturated _ if there is an edge in the complement of @xmath0 such that the ramsey number @xmath1 of @xmath0 does not change upon adding it to @xmath0 .
this notion was introduced by balister , lehel and schelp in @xcite , where it is shown that cycles ( except for @xmath2 ) are ramsey - unsaturated , and conjectured that , moreover , one may add _ any _ chord without changing the ramsey number of the cycle @xmath3 , unless @xmath4 is even and adding the chord creates an odd cycle .
we prove this conjecture for large cycles by showing a stronger statement : if a graph @xmath0 is obtained by adding a linear number of chords to a cycle @xmath3 , then @xmath5 , as long as the maximum degree of @xmath0 is bounded , @xmath0 is either bipartite ( for even @xmath4 ) or almost bipartite ( for odd @xmath4 ) , and @xmath4 is large .
this motivates us to call cycles _ strongly _ ramsey - unsaturated .
our proof uses the regularity method . |
binaries are well known to have a profound impact on the dynamical evolution of globular clusters ( hut et al .
1992 and edmonds et al .
2003 , hereafter paper i ) , and they offer an opportunity to study the results of stellar interactions .
they also offer the crucial advantages of studying binary systems at the same , well - determined distance , age , metallicity and reddening .
accurate distances are particularly important for reducing uncertainties in luminosities , and making reliable cluster - to - cluster comparisons and cluster - to - field comparisons . despite these promising goals ,
observational progress in detecting globular cluster binaries was initially slow , mainly because of spatial resolution and sensitivity limitations .
however , the superb imaging capabilities of _ chandra _ and _ hst _ has now resulted in the detection of large numbers of compact binaries in several globular clusters ( see grindlay et al .
2001a , hereafter ghe01a and references in paper i ) . in particular , using the _ chandra _ data of ghe01a and the _ hst _ data of gilliland et al .
go-8267 ) , a large population of x - ray binaries with optical identifications has been found in 47 tuc , with over 50 binaries reported in paper i. there , the details of the astrometry and photometry for this sample were reported , with most of the systems being identified as either accreting white dwarf - main sequence star binaries ( cataclysmic variables or cvs ) or chromospherically _ active binaries_. this paper reports _ hst _ optical time series and other detailed analysis for the sample of binaries reported in paper i. variability for most of the cvs is detected , confirming the cv identifications given in paper i based on absolute photometry alone . in particular flickering for the brighter optical ids is found , plus periods for several of the higher inclination systems and some long - term variability .
the cv periods and the absolute magnitudes determined in paper i are compared with predictions for roche - lobe filling secondaries and are compared with cvs found in the field .
time series for the active binaries are also presented here , with an emphasis on the variables not discussed by albrow et al .
( 2001 ; hereafter agb01 ) . the radial distribution ( radial offsets from cluster center ) of the cvs and active binaries discovered in paper i are compared with the radial distribution of msps from freire et al .
( 2001 ) and with the total stellar population , to test whether these distributions are consistent with the expectations of mass segregation .
comparisons between the radial distributions of the x - ray detected active binaries and the larger sample of binaries discovered by agb01 are also made .
other important diagnostics examined here are the x - ray luminosities and optical magnitudes .
the flux ratio between the two was shown by richman ( 1996 ) and verbunt et al .
( 1997 ; hereafter vbr97 ) to be inversely proportional to the cv accretion rate ( for non - magnetic systems ) , and here we make detailed comparisons between the 47 tuc cvs and field cvs to help constrain the cv accretion rates in 47 tuc .
the x - ray to optical flux ratio is also useful in searching for additional quiescent low - mass x - ray binaries ( qlmxbs ) besides x5 and x7 ( ghe01a and heinke et al .
we also discuss the x - ray luminosity and absolute magnitude distributions of the 47 tuc cvs and compare to field systems .
the time series will be described in [ sect.tseries ] , followed by an analysis section ( [ sect.anal ] ) , including a study of the radial distribution of the sources and their x - ray to optical flux ratios .
this will be followed by the discussion in [ sect.disc ] , and an interpretation of the results given in both papers i and ii .
paper i discusses the photometric results for the optical ids ( including the cmds ) in detail .
exquisite time series were produced in the @xmath0 and @xmath7 bands for the go-8267 dataset , and are analyzed in this paper .
detailed simulations were carried out by agb01 , giving the recovery rate of simulated variables as a function of period , @xmath0-magnitude and signal amplitude for the pc1 and wf2 chips . for amplitudes of @xmath8(intensity)/intensity
= 0.1 , the recovery rate for pc1 was @xmath6100% from the msto to @xmath0=23.5 for all input periods below 3.2 days , and for amplitudes of 0.01 the recovery rate fell from @xmath670% at @xmath0=20 to @xmath910% at @xmath0=21.5 ( agb01 ) .
the wf2 recovery rates extend to slightly fainter magnitudes ( agb01 ) , but the generally higher crowding and saturation levels in the wf2 images counteract this gain , for variable searches . for the cv candidates found in the go-8267 fov ( paper i )
we have analyzed the high quality time series produced by the planet and binary searches of gilliland et al .
( 2000 ) and agb01 .
time series were produced using difference image analysis combined with aperture and psf fitting photometry ( see gilliland et al . 2000 and agb01 for more details ) .
the size of the region used in the photometry extractions was decreased to as small as one pixel in the cases where plausible x - ray ids were located near bright stars .
least - squares fits to sinusoids and the lomb - scargle periodogram were then applied to search for periodic variations in the time series of the cv candidates .
seven certain periodic variables were confirmed in this way : w1@xmath10 , w8@xmath10 , w15@xmath10 , w21@xmath10 , w34@xmath10
, ako 9 and the marginal i d for w71 . we discuss each of these here in turn .
the i d for w1 is either a 2.89 hr or a 5.78 hr period variable ( ellipsoidal variations ) .
examination of the light curves folded at the 2.89 and 5.78 hr periods , and of the power spectra , do not show unambiguous departures from sinusoidal light curves ( fig .
[ fig.w1 ] ) , although the @xmath0-band phase plot does show some evidence for asymmetry at a little less than the 3-@xmath11 level , arguing for the longer period .
we can use equation ( 2.102 ) from warner ( 1995 ) to place limits on the absolute magnitude of the secondary : @xmath12 where @xmath13 is the absolute magnitude of the secondary and @xmath14 is the orbital period measured in hours ( where 2 hr@xmath15 10 hr ) .
this relationship is appropriate for field cvs of roughly solar metallicity . at fixed mass , stars with lower metallicity
will have brighter secondaries ( stehle , kolb & ritter 1997 ) , and so equation 1 gives a lower limit to the luminosity of secondary stars in 47 tuc ( an upper limit to @xmath1 ) .
for w1@xmath10 ( @xmath16 ) using a 5.78 hr period gives a limit of @xmath17 from equation 1 , while using a 2.89 hr period gives @xmath18 .
the longer period is therefore favored unless the accretion disk completely dominates in the @xmath0-band .
this may be possible if w1 is a non - magnetic nova - like cv , explaining the relatively blue @xmath19 color of w1@xmath10 . however , the @xmath2 value for this system is much more consistent with a magnetic cv or a dn system ( see fig . [ fig.fxfopt ] ) , and a period of 2.89 hrs would be shorter than any of the ux uma systems given in fig .
[ fig.permv]a , and all but one of the dq her systems .
modeling of the accretion disk and secondary in this system may help determine which period is correct .
note that both w8@xmath10 and w15@xmath10 are optically crowded and the absolute photometry in @xmath0 and @xmath7 is relatively poor ( an @xmath7 magnitude could not be self - consistently derived for w15@xmath10 ) .
however , since this project was optimised for time series accuracy , the quality of the time series is much higher than that of the absolute photometry . both w8@xmath10 and
w15@xmath10 are clearly eclipsing systems ( see fig . [ fig.w8w15 ] ) , with unambiguous period determinations of 2.86 hr ( at the upper edge of the period gap for field cvs ) and 4.23 hr respectively .
both of the associated x - ray sources for these cvs are very hard ( ghe01a ) .
detection of these objects as eclipsing strongly supports the argument of ghe01a that self - absorption by the accretion disks are responsible for the hard spectra .
[ sect.permv ] for more details about these two binaries .
the i d for w21 is very near to a strong saturation trail and evidence for a periodic signal was only detected in the single
pixel analysis .
[ fig.w21 ] shows the secure 52 minute variation ( this may also be a 104 minute orbital period ) .
the lack of strong signal at 104 minutes in the power spectrum and the large amplitude of the time series ( [email protected] ) tentatively favor 52 minutes being the orbital period .
poor phase coverage at phases 0.40.75 makes it difficult to distinguish between these two periods , although the phase plot does show suggestive evidence of asymmetry . in the 52 minute period case
the secondary is possibly a degenerate star since 52 minutes is well below the canonical period minimum for field cvs of @xmath21 min , for non - degenerate secondaries .
however , some field cvs with non - degenerate secondaries do have periods about this small ( e.g. a 59 minute orbital period for v485 cen ; augusteijn et al .
1996 ) .
the light curve for w34@xmath10 is discussed in detail in edmonds et al .
( 2002b ) .
although the strong 97.45 minute period is very close to the _ hst _ orbital period ( 96.5 minutes ) , edmonds et al .
( 2002b ) argue that this is not the result of an artifact showing up as a false signal at the _ hst_orbit .
the binary w36@xmath10 ( ako 9 ) is a long period eclipsing cv with a subgiant secondary and a period unambiguously determined to be 1.108 days ( see also edmonds et al .
1996 , ghe01a and knigge et al .
the blue variable and possible counterpart to w71 has a 1.19 hr power spectrum signal with a likely 2.37 hr orbital period , based on the small but significant asymmetry in the phase plot using the longer period ( see fig .
[ fig.w71 ] ) . if this is a cv it would fall within the period gap for field cvs ( @xmath22hr ) .
although this variable may be a magnetic cv , and the source is relatively soft , consistent with an am her system , we favor the explanation ( supported by the astrometry ) , that the _ chandra _ source is not associated with the nearby blue , variable star ( see [ sect.bluevar ] for possible explanations of its nature ) . for the other cv candidates ,
deeper searches for periodic variations were made and several marginal signals were found ( see fig . [ fig.marg-tseries ] ) . for v1 ,
the strongest @xmath0-band power spectrum peak in a reasonable period range corresponds to an orbital period of @xmath67 hr ( for ellipsoidal variations ) . for v2 peaks
are found at 13.67/2 hr ( @xmath0 ) and 6.01/2 hr ( @xmath7 ) and a strong signal is also seen in @xmath0 at 1.44 days and a 1.5 day signal in @xmath7 .
the shortest of these corresponds to a plausible orbital period , while the two longest ones are unlikely to be orbital periods for this apparent ms secondary , unless it is evolved . for v3
, peaks are found in both @xmath0 and @xmath7 corresponding to a possible orbital period of @xmath63.7 hr ( close to the 3.83 hr period present in the x - ray light curve ; ghe01a ) .
the i d for w45 shows a significant peak at 1.87 hr in @xmath0 ( possibly a 3.75 hr orbital period ) .
a weaker signal is found in @xmath7 at the same period and a strong signal at 1.78 hr .
the i d for w2 shows possible periods at 2.2 , 5.9 and 8.2 hr .
these differ from the statistically significant x - ray period of 6.287 hr ( see fig . [ fig.w2 ] for an x - ray phase plot and fig .
[ fig.w2-w120v]a for a @xmath0-band power spectrum and phase plot ) .
the power spectrum is extremely noisy , at least partly because the counterpart lies on a diffraction spike from a nearby giant star .
flickering may be an extra source of noise .
the i d for w120 shows a marginally significant peak in the @xmath0 power spectrum of 1.32 hr , with a false - alarm probability ( fap ) of 1.4@xmath23 ( see fig .
[ fig.w2-w120v]b ) .
the power spectrum is otherwise very clean , with no other peaks having a fap @xmath24 .
the possible counterpart to w10 has a 2.2 or 4.4 hr period and may also be a cv , but crowding prevents us obtaining useful limits on the color .
figure [ fig.rms ] shows plots of time series rms vs @xmath0 and @xmath7 for the pc1 , wf2 and wf4 chips ( the few wf3 variables are shown on wf4 ) . in the cases where @xmath7 values are not available ( w15@xmath10 , w33@xmath10 , w45@xmath10 and w70@xmath10 ) we have assumed that the object is found on the ms in the @xmath0 vs @xmath19 cmd .
as expected , significantly higher rms values than average are found for stars with large amplitude , periodic variations ( e.g. w1@xmath10 , w8@xmath10 and w15@xmath10 ) . among the cv candidates ,
significant non - periodic variations were found for v1 , v2 , v3 and w25@xmath10 .
these objects are clearly visible in the @xmath0 and @xmath7 images , and are not affected by diffraction spikes or saturation trails , so the source of the enhanced noise for these objects is probably flickering from the hot spot or accretion disk .
strong temporal changes in the psfs correlated with telescope breathing caused likely spurious signals to appear in the power spectra near the _ hst _ orbital frequency ( 0.1725 mhz ) for these four large amplitude variables .
the same explanation can not be confidently applied to the possible enhanced noise for w33@xmath10 , w45@xmath10 , w70@xmath10 and w120@xmath10 .
these objects are all so crowded that none of them were detected in the original @xmath0 and @xmath7 band images by agb01 , and the enhanced noise could have some contribution from this crowding .
strong signals near the _ hst _ period are only found for w33@xmath10 in the @xmath7 band .
by contrast , the i d for w44@xmath10 is relatively isolated , though faint , and the higher rms values found here are also suggestive of non - periodic variations .
significant variability was found for several of the cv candidates by comparing the 3rd epoch images of the _ hst _ program go-7503 with those obtained in the first two epochs ( epoch 1 : go-5912 , and epoch 2 : go-6467 ) .
[ fig.7503-var ] shows those cases where f300w variability was discovered ( small amplitude variability was also found for v2 , a known dn , but is not shown here ) .
most of the variations are obvious from fig .
[ fig.7503-var ] with the exception of w44@xmath10 which is 0.36 mag fainter in the 1st epoch than it is in the 2nd ( and 0.30 mag fainter in the 1st than in the 3rd ) , and w122@xmath10 which is 0.87 mag fainter in the 1st epoch than the 2nd and 0.52 mag brighter in the 2nd epoch than the 3rd .
some of these variations may represent dwarf nova ( dn ) outbursts , especially w51@xmath10 and w56@xmath10 .
a detailed analysis of the long - term variability of the cvs will be deferred to a later publication ( where these results will be combined with subsequent _ hst _
observations made with acs ) .
having a sample of cvs and cv candidates with period and @xmath1determinations at a well - known distance allows us to examine the period luminosity relationship expected for cvs .
figure [ fig.permv]a shows a plot of period vs @xmath1 for cvs and cv candidates in 47 tuc along with those of cv1 and cv6 in ngc 6397 ( grindlay et al .
2001b , hereafter ghe01b ; kaluzny & thompson 2002 ) , the dn in m5 ( neill et al .
2002 ) , two cvs in ngc 6752 ( bailyn et al . 1996 ) and the magnetic cv in m67 ( gilliland et al .
the straight line is equation [ eqt.warner ] from warner ( 1995 ) , for solar - metallicity stars .
this will shift to brighter magnitudes ( by about 1 mag ) for metal - poor secondaries , so cvs that are near or above this line are likely to have evolved secondaries
. some of the @xmath0-band light will be from the accretion disk and hot spot ( as well as the secondary ) and this will also push the @xmath0 magnitudes to brighter values ( to the left in [ fig.permv]a ) .
figure [ fig.permv]b shows the plot for the blue variables described earlier . the likely cv w2@xmath10 and the cv candidate v3 both lie close to or above the ms ridge - line in the @xmath0 vs @xmath19 cmd ( see the cmd for pc1 in paper i ) , showing that the secondary dominates the optical light .
as expected these objects lie close to the warner relationship .
the secondary for w2@xmath10might be evolved , but better photometry is needed to make this conclusion secure .
the cvs v1 and w1@xmath10 fall below equation [ eqt.warner ] because the accretion disk and hot spot ( and possibly the wd ) make significant contributions to the light ( the cmds in paper i show that the @xmath19 colors of w1@xmath10 and v1 are well to the blue of the ms ) .
the light curve of the eclipsing cv w15@xmath10 ( fig .
[ fig.w8w15 ] ) implies that the accretion disk and hot spot are responsible for @xmath25% of the @xmath0-band light , and increasing @xmath0 by @xmath61 mag to isolate the ms component brings w15@xmath10 into excellent agreement with equation [ eqt.warner ] . for w8@xmath10 around 40% of
the @xmath0-band light appears to come from a source other than the secondary .
given the similarity of the light contributions from the secondaries , equation [ eqt.warner ] implies that the w8@xmath10secondary should be about 2 mags fainter than the w15@xmath10 secondary , but instead it is almost 1 mag _
brighter_. possible explanations are that : ( 1 ) we are seeing a grazing eclipse with the inclination slightly less than 90 degrees , ( 2 ) a faint ms star is contaminating the light from the cv , or ( 3 ) the normalization for w8@xmath10 is incorrect because of the crowding .
examination of fig .
[ fig.w8w15 ] shows that the ellipsoidal variations in w15@xmath10 are close to the maximum value expected ( see russell 1945 ) , but are around a factor of 2 smaller in w8@xmath10 , consistent with any of these three explanations .
similar ( but less extreme ) issues exist for the two ngc 6752 cvs ( both of these systems lie close to the ms in the @xmath0 vs @xmath19 cmd ) .
the good agreement of the m5 dn with equation [ eqt.warner ] implies that the @xmath0-band light must be dominated by the secondary .
the variable w34@xmath10 has colors that are consistent with those of a cv ( @xmath26 ; @xmath19=1.03 ; @xmath1=9.3 ) and appropriately falls below equation [ eqt.warner ] .
however , as noted by edmonds et al .
( 2002b ) , the relatively large amplitude variation of this object in both @xmath0 and @xmath7 makes an msp another possibility , since the light curve is similar to that of w29@xmath10(47 tuc w ) .
either possibility is consistent with the x - ray properties of this object .
the variable w21@xmath10 ( @xmath27=0.15 ; @xmath19=0.27 ; @xmath1=7.44 ) is also difficult to categorize . assuming that the orbital period is twice the observed period
, the star still falls well below equation [ eqt.warner ] .
could an accretion disk for a cv with a period of 1.73 hr and an expected @xmath1(2)=14 be this bright ? figure 9.8 of warner ( 1995 ) shows that the 13 dne with periods @xmath282 hr mostly have @xmath1(disk)= 89 , with 3 having @xmath1(disk ) @xmath28 7.1 .
however , we would expect a very blue @xmath27 color if the disk is this bright .
alternatively , the binary could be a wd - wd cv or a wd - ns system in quiescence .
finally , ako 9 with its subgiant secondary is well outside the range of applicability of equation [ eqt.warner ] .
figure [ fig.permv-ver ] shows a period vs @xmath1 plot ( similar to fig .
[ fig.permv ] ) for the field cvs with estimated distances in the rosat study by vbr97 .
the orbital periods have been taken from ritter & kolb ( 1998 ) .
the nova - like cvs ( systems that are magnetic or do not show outbursts ) and the dne ( outbursting systems ) have been separated into two plots to show systematic differences between these two general classes of cv .
the labeling of vbr97 for the different cv types is used .
the nova - like cvs are generally further away from equation [ eqt.warner ] than the dne , showing that the relative light contributions from the accretion disk and hot spot ( compared to the secondary ) are larger than for the dne .
the 47 tuc cvs ( and the other cluster cvs ) appear to be more consistent with the dne from vbr97 than the nova - like systems . by calculating the magnitude predicted by equation [ eqt.warner ]
( using the binary period ) and subtracting @xmath1 , we quantify the offsets from equation [ eqt.warner ] and examine the light contribution from sources other than the secondary ( mainly the accretion disk , hot spot and wd ) .
this procedure will overestimate the contribution from the disk and other non - secondary components , since equation [ eqt.warner ] underestimates the contribution from a metal - poor secondary .
figure [ fig.cumul-permv ] shows the cumulative distributions of the offsets from equation [ eqt.warner ] for the 47 tuc and the field cvs .
we have eliminated binaries with periods @xmath2910 hr to remain consistent with the period limitations on equation [ eqt.warner ] .
figure [ fig.cumul-permv]a shows all such systems and fig . [ fig.cumul-permv]b shows only systems with periods above the cv period gap ( the latter figure avoids the shortest period systems likely to have a greater light contamination from the wd ) . for systems above the period gap , the offset distributions for the 47 tuc cvs and the nova - like systems are distinguishable at the 99.55% level using the ks - test .
a subset of the nova - like systems , the dq hers , are distinguishable from the 47 tuc cvs at the 93.8% level ( the nova - like and dq her offsets are distinguishable from each other at the 1.2% level ) .
by contrast , the 47 tuc cv offsets are distinguishable from the dne offsets and the u gem systems at only the 1.4% and 0.9% levels respectively .
this suggests that the 47 tuc cvs have accretion luminosities that are even lower than those found in u gem systems and dne since , as mentioned above , the non - secondary component in the 47 tuc cvs is being underestimated here .
this then suggests that the 47 tuc cvs have lower accretion rates .
more evidence supporting this conclusion is given in
[ sect.fxfopt ] , and discussion is given in
[ sect.cvdisc ] .
as described in paper i , several blue , variable stars discovered by agb01 are not obviously associated with x - ray sources and have colors that are different from most cluster and field cvs . the time series for these blue variables are shown in fig .
[ fig.bluevar-v ] and the period versus @xmath1plot is shown in fig .
[ fig.permv]b .
the variable 1v36 has a sinusoidal , 9.55 hr variation ; 2v08 and 2v30 have 4.8 hr and 28.3 hr variations , and 3v06 , 3v07 and 4v05 have 2.78 hr , 3.79 hr and 3.11 hr variations respectively .
note that the absolute magnitudes of the secondaries ( if ms members ) are likely to be significantly fainter than the observed @xmath1because these stars are all blue in @xmath0 vs @xmath19 .
we have assumed that the orbital period is twice the detected period , as appropriate for most cvs ( roche - lobe filling , ms secondaries with ellipsoidal variations ) with periodic variations detected in the optical .
if this assumption is correct , then 1v36 , 2v08 , 2v30 and 3v06 are unlikely to be filling their roche lobes since they fall above equation [ eqt.warner ] in fig .
[ fig.permv]b , and are therefore unlikely to be cvs ( as already noted for 1v36 by agb01 ) .
if any of the secondaries are evolved then they could fill their roche lobe at the observed period , but such secondaries should be unusually red , and therefore to explain the observed blue @xmath19 colors we probably require an unusually uv - bright , luminous component . even if evolved companions were expected to give unusually bright disks , only 2v08 and 2v30 have @xmath27 colors as blue as most of the other cvs , so the blue variables do not seem to have unusually bright disks .
the blue color for 3v07 implies that only a small fraction of the optical light comes from a ms star , and therefore the secondary for this system is also likely to fall above the warner relationship ( the @xmath1 offset of 3v07 from equation [ eqt.warner ] in fig .
[ fig.permv ] of @xmath61 mag roughly equals its offset from the ms in the @xmath0 vs @xmath19 cmd .
another possibility is that the orbital periods are the same as the observed periods , and that the variations are caused by heating of one side of an ms star . in this case
some of the blue variables ( like 1v36 ) may not have to contain evolved secondaries for roche lobe filling to be taking place .
there are at least two problems with this explanation .
firstly , a uv - bright , relatively luminous component is probably required to cause measurable heating effects , but only 2v08 and 2v30 have @xmath27 colors that are comparable to those of the bluest cvs . secondly , because sinusoidal signals with low amplitudes ( either from heating or ellipsoidal variations ) are generally not detectable even in the brightest cvs because of noise from flickering .
if the blue variables are cvs then we would expect flickering to mask the small sinusoidal signals that are observed .
instead , the rms values for the blue variables are quite low ( fig . [ fig.rms ] ) implying low amplitude flickering , despite having bluer colors and ( presumably ) relatively bright disks , if they are cvs .
we note that these results are unlikely to be some sort of photometric artifact ( such as the effects of bad pixels ) , since several of these objects exist and none of them show unusual psf fits , or excessive levels of crowding .
also , the power spectra show little if any extra power near the _ hst _ orbital frequency .
[ sect.bluevar ] for more discussion about these unusual objects . the phase plots for w12@xmath10 , w92@xmath10 , w137@xmath10 and w182@xmath10 ( eclipsing binaries ) and w41@xmath10 , w47@xmath10 and w163@xmath10 ( contact binaries ) are shown in agb01 .
light curves for most of the by dras and red stragglers from agb01 that have also been identified as active binary candidates are shown in fig .
[ fig.bydra ] , and phase plots for a selection of active binary candidates not found by agb01 are shown in fig . [ fig.other-abs ] .
note that ambiguity exists in the period determination for several of the variables shown in fig .
[ fig.other-abs ] ( especially for w22@xmath10 and w94@xmath10 ) . in these cases
we have assumed that the orbital period is twice the observed period .
clearly , w38@xmath10 and w94@xmath10 are eclipsing binaries , and the other new objects are a likely mixture of by dra and w uma systems ( w66@xmath10 was classified by agb01 as a non - eclipsing contact binary ) .
here , we compare the radial offsets , with respect to the center of 47 tuc ( ` radial distributions ' ) of the different classes of binary . although the go-7503 dataset has uncovered a number of likely cvs and at least 2 active binary candidates , for radial distribution analysis we restrict ourselves to the go-8267 dataset because it contains a much more complete sample of active binary candidates .
figure [ fig.radall ] shows the radial distributions of : ( 1 ) the radio detected msps ( plus w34 ) , ( 2 ) the cvs , ( 3 ) the binaries discovered by agb01 ( with the relatively small fraction of x - ray bright systems and the two confirmed cvs removed ) , ( 4 ) the candidate active binaries or ` xabs ' ( ms or red straggler binaries either from agb01 or this work with apparent x - ray counterparts ) and ( 5 ) the blue variables .
note the similarity between the cv and the msp radial distributions ( ks - test probability = 14.0% ) .
this similarity is not surprising because the single msps should have masses of @xmath30m@xmath31and the msp binaries only @xmath60.020.2m@xmath31 more .
the detected cv candidates should have masses ranging from @xmath60.95m@xmath31 for faint objects like v3 and w120@xmath10 ( @xmath60.4m@xmath31 secondary and @xmath60.55m@xmath31 primary ) up to @xmath61.6 m@xmath31 for ako 9 ( @xmath60.8m@xmath31primary and secondary ) .
the 7 blue variables are slightly less centrally concentrated than the cvs ( ks - test probability = 88.2% ) and the msps ( ks - test probability = 93.8% ) . eliminating 2v08 and 2v30 , the two variables with colors
most like white dwarfs or cvs , causes the above ks - test probabilities to drop to 53% and 55% , suggesting that the blue variables are heavy objects ( 2v08 and 2v30 are relatively distant ) .
however , the remaining sample of 5 is very small for statistical tests .
the radial distribution of the agb01 binaries is slightly less centrally concentrated than those of the cvs and msps ( ks - test probabilities of 77% and 81% respectively ) .
these results are only marginally statistically significant , but are a likely result of lower masses for some of the agb01 binaries , especially the faintest ones .
the active binaries appear to be marginally more centrally concentrated than the agb01 binaries , but the ks - test probability ( 45% ) is not statistically significant . if real , this difference likely results from the generally brighter @xmath0 magnitudes ( and higher masses ) of the active binaries compared to the agb01 binaries ( see fig [ fig.radall]b ) , since the @xmath0 distributions differ at the 90% probability level . also , the @xmath0 magnitudes of the x - ray bright by dras are brighter than the @xmath0 magnitudes of the much larger by dra population from agb01 ( see fig [ fig.radall]d ; ks test probability = 93.5% ) .
the active binary periods extend to higher values than the periods of the cvs and the msps ( fig .
[ fig.radall]c ) , and most strikingly they are significantly shorter than those of the agb01 binaries ( ks probability = 98.9% ) .
we now consider the agb01 binaries and active binaries with @xmath32 and @xmath33 ( this divides the agb01 binaries into two evenly sized groups ) . in the bright subsample
the radial distributions of the agb01 binaries and the active binaries are indistinguishable from each other ( ks - test probability = 16% ) .
the radial distributions of both the agb01 and the bright active binaries are indistinguishable from those of the msps ( ks - test probabilities = 30% and 29% ) and cvs ( ks - test probabilities = 16% and 9% ; see fig .
[ fig.rad-bright-faint]a ) . considering the fainter binaries ( @xmath33 ) , the radial distribution of the faint agb01 binaries and the faint active binaries are indistinguishable from each other ( ks - test probability = 12% ) .
the radial distributions of both the faint agb01 binaries and the faint active binaries ( see figure [ fig.rad-bright-faint]b ) are less centrally concentrated than those of the msps ( ks - test probabilities = 97.0% and 83% ) , with similar results for the cvs ( ks - test probabilities = 93% and 85% ) .
these results are consistent with the faint active binaries being less massive than the bright active binaries .
figure [ fig.rad-bright-faint]c shows the radial distribution of the sources in the go-8267 fov that have not been identified , plotted with the bright and faint active binaries and the general stellar population in three different @xmath0-band ranges .
the unidentified sources are indistinguishable from the bright active binaries , the cvs and the msps , showing that they are consistent with any of these groups , or a combination of all three types .
the faint active binaries have a radial distribution that is almost identical to that of the stars with @xmath34 ( probability = 0.73% ) , and is considerably more concentrated than the stars with @xmath35 ( probability = 84.7% ) .
all of these results are consistent with expectations for mass segregation of binaries and stars in 47 tuc , and with the statement of agb01 that incompleteness as a function of radius and magnitude is expected to be small even for stars as faint as [email protected] .
the most striking result is that the period distribution of the faint active binaries differs from that of the faint agb01 binaries at the 99.99% probability level .
figure [ fig.rad-bright-faint]d shows that while _ all _ of the faint active binaries have periods @xmath280.7 days , only @xmath630% of the agb01 binaries have periods below this value .
we discuss in
[ sect.abdisc ] how these results imply the presence of a stronger relationship between luminosity and period than found in field objects .
a useful , distance independent diagnostic for stars detected in both the optical and x - ray wavelength bands is the ratio of the x - ray flux to the optical flux .
figures [ fig.fxfopt-ver]a and [ fig.fxfopt-ver]b show plots of this ratio vs @xmath1 for nova - like cvs and dne in the field ( vbr97 ) .
the @xmath36 values in the 0.5 - 2.5 kev band from vbr97 were scaled to be consistent with the slightly absorbed , 1 kev thermal bremmstrahlung spectra ( 0.5 - 2.5 kev ) assumed by ghe01a , and used here .
the @xmath37 values were derived from @xmath38 . in cases where the distances to the field cvs of vbr97 were unknown ,
the systems were plotted on the left side of figures [ fig.fxfopt-ver]a and [ fig.fxfopt-ver]b .
the majority of field dne have @xmath2 @xmath39 with the smallest ratios being dominated by z cam systems ( with relatively high accretion rates ) .
the remaining dn classes ( u gem and su uma systems ) mostly have @xmath2 @xmath40 . among the nova - like cvs the ux uma systems have @xmath2values or limits mostly around 0.01 .
these ranges show a clear trend that higher @xmath2 values correspond to lower accretion rates for non - magnetic cvs , as pointed out by vbr97 for this sample of field cvs , and by richman ( 1996 ) .
most magnetic cvs have @xmath2 @xmath41 , although it is unclear if @xmath2 is anti - correlated with accretion rate for these systems . fig .
[ fig.fxfopt]a shows the @xmath2 values for the 47 tuc cv candidates and msps ( plus other blue ids ) and fig .
[ fig.fxfopt]b shows the @xmath2 values for the active binary candidates .
limits are shown where appropriate .
note the clear differences between figures [ fig.fxfopt]a and [ fig.fxfopt]b , with generally much larger @xmath2 values for the cvs and the msps . with the exception of ako 9 , the @xmath2 values for 47 tuc cvs and cv candidates are all @xmath41 , suggesting that the population of 47 tuc cvs , if dominated by non - magnetic systems , are low accretion rate cvs ( u gem and su uma systems , where the latter sub - class are believed to have the lowest accretion rates among field cvs ) . besides v3 , the likely dq her system v1 ( ghe01a ) has the highest @xmath2value .
cumulative distributions of @xmath2 for the 47 tuc cvs , active binaries and msps are compared in fig .
[ fig.cumul-fxfopt]a .
only the objects detected in @xmath0 have been included rather than those with limits , with the exception of 47 tuc t. the clear differences between the cv and the active binary distributions ( and between the msp and active binary distributions ) are confirmed by the median values given in table [ tab.fxfopt-median ] and the large ks - test probabilities shown in table [ tab.fxfopt-ks ] .
the msp distribution only contains 4 systems and the other binary msps detected in the radio fall below our optical detection threshold ( but are mostly detected by _
chandra _ ) , so the true msp distribution will extend to higher @xmath2 values .
the only region of significant overlap between the active binaries and either the cvs or msps is found with 0.1 @xmath9 @xmath2 @xmath9 0.5 .
all of these apparently extreme active binaries are found in the faint group and have radial distributions that are less centrally concentrated than either the cvs , the msps , or the bright active binaries . in fig .
[ fig.cumul-fxfopt]b the 47 tuc @xmath2 distribution for the cvs is compared with those found for field cvs using the data given in vbr97 .
table [ tab.fxfopt-median ] shows the median values for each distribution and table [ tab.fxfopt-ks ] compares the ks - test probabilities .
the trend of increasing @xmath2 for decreasing accretion rate ( richman 1996 & vbr97 ) is clear , with the ux uma ( nova - like ) systems and the su uma ( dne ) systems at opposite ends of the range , and the z cam and u gem systems falling in between .
the 47 tuc @xmath2 values are higher on average than all of the various cv subgroups , including 3 old novae and 3 double - degenerate systems not shown in tables [ tab.fxfopt-median ] and [ tab.fxfopt-ks ] . in a separate calculation we have also included the field systems with x - ray limits ( rather than just detections ) given by vbr97 and this only increases the differences between the @xmath2 values of the 47 tuc cvs and the field cvs . to further investigate the origin of differences in @xmath2 between the 47 tuc cvs and the field cvs , we plot in fig .
[ fig.lxmv]a the cumulative @xmath3 distributions of the 47 tuc cvs and field cvs and a similar plot in fig .
[ fig.lxmv]b for @xmath1 .
limits in @xmath3 ( corresponding to x - ray detection limits , not distance limits ) have been included to increase the sample sizes of the field cvs , since only a subset of the cvs have estimated distances .
the luminosities for most of the field cv classes are indistinguishable from each other , with only the dq her systems having a significantly different ( more luminous ) distribution , as pointed out by vbr97 .
the @xmath3 distribution of the 47 tuc cvs is distinguishable from the am her , dq her , ux uma , z cam , u gem and su uma systems at the 97.6% , 22.4% , 97.3% , 72.2% , 99.3% and 99.8% levels respectively , so that only the dq her systems are obviously a good match to the 47 tuc cvs .
this would argue against the identification of the 47 tuc cvs with low accretion rate cvs like dne , if the accretion rate is proportional to @xmath3 ( for example warner ( 1995 ) suggests @xmath42@xmath3 ) .
however , see the discussion of the @xmath1 distribution below .
note that the majority of field am her and dq her systems from vbr97 were x - ray selected ( as were the 47 tuc cvs ) , unlike other classes of cv , implying that the relatively high x - ray luminosities could be partially selection effects , perhaps weakening the classification of the 47 tuc cvs with magnetic systems .
for example , crowding in x - rays is likely to have prevented the detection of some systems with low x - ray luminosities , and the true cluster distribution will therefore extend to lower luminosities .
robust limits on incompleteness will be presented in future publications .
incompleteness will also apply with the field samples , especially the magnetic systems .
for example , although only 2 out of 10 of the dq her systems in the field sample analysed here involve upper limits , a significant number of faint dq hers ( and other cv classes ) may have been missed in field surveys . the four most luminous dq her systems are at relatively large distances ( @xmath29400 pc ) , suggesting that such bright systems are relatively rare , as pointed out by vbr97 . however , despite this incompleteness at low luminosities , the similarity between the 47 tuc and dq her distributions at high luminosities ( above @xmath43 erg s@xmath5 ) is noteworthy , and v1 , the brightest cv in 47 tuc , has already been shown to be a likely dq her system .
the @xmath1 distribution of the 47 tuc cvs is distinguishable from the am her , dq her , ux uma , z cam , u gem and su uma systems at the 8.6% , 99.97% , 99.99% , 98.80% , 1.03% and 96.2% levels respectively . here , only the am her and the u gem systems are good matches for the 47 tuc cvs .
a comparison of figures [ fig.fxfopt-ver ] and [ fig.fxfopt ] shows that our optical detection limit of @xmath44 ( because of crowding ; see [ sect.summid ] ) is a significant source of incompleteness for faint dne searches , and may prevent the detection of a significant fraction of su uma systems .
clearly , the 47 tuc cvs are significantly fainter in the optical than the nova - like cvs detected in the field , including the dq her systems .
since metal - poor secondaries are expected to be brighter than solar metallicity secondaries , at fixed mass , the difference in accretion luminosity between the 47 tuc cvs and field dq hers will be even more dramatic than suggested by fig .
[ fig.lxmv]b . finally , distances are available for only 2 old novae and none are available for double degenerate systems ( both of these classes of cv may be extremely faint and beyond our detection limits ) .
the faint @xmath1 values for the 47 tuc cvs suggest that they have low accretion rates , consistent with the period/@xmath1 analysis , but conflicting with the interpretation based on their x - ray luminosities .
in fact , an examination of fig .
[ fig.lxmv ] shows that the 47 tuc cvs have combined @xmath3 and @xmath1 distributions that are not consistent with any known class of field cv ( see [ sect.cvdisc ] for discussion of this issue ) . as expected , the optical i d for x5/w58 ( edmonds et al .
2002a ) , a qlmxb , has a relatively high @xmath2 value of 47.8 ( fig .
[ fig.fxfopt ] ; see also heinke et al .
interestingly , the cv candidate v3 has a very similar @xmath2 value .
this ratio is higher than that of all known field and 47 tuc cvs , representing evidence for v3 also being a qlmxb .
the possibility that v3 is a binary containing a neutron star has already been discussed by ghe01a who argue that v3 may be an enshrouded msp . for comparison , source b , the possible qlmxb in ngc 6652
has @xmath2 @xmath69.0 when the @xmath37 value given by heinke , edmonds , & grindlay ( 2001 ) is appropriately converted , while limits for x7 and the qlmxb in ngc 6397 are @xmath29166 ( heinke et al .
2003 ) and @xmath2964 respectively .
however , we caution that the _ chandra _ and go-8267 data were not obtained simultaneously , and since v3 is highly variable both optically ( see paper i ) and in x - rays ( verbunt & hasinger 1998 ; ghe01a ) the high @xmath2 value derived here could be explained by a low optical state during the go-8267 observations , or a high state during the _ chandra _ observations .
the two bright ( @xmath3 @xmath45erg s@xmath5 ) sources and likely qlmxbs in 47 tuc ( x5 and x7 ) and the likely qlmxbs in ngc 6397 and ngc 6440 are all soft sources , using the x - ray color definition of grindlay et al .
( 2002 ) , namely xcolor=2.5 log@xmath46kev@xmath47/[1.5 - 6$]kev@xmath47)$ ] .
therefore , we have searched for evidence that any of the fainter , soft x - ray sources in 47 tuc may be qlmxbs ( besides v3 ) .
the source ako 9 is a very soft source but is more likely a cv than a qlmxb because it has been observed to have at least 2 outbursts without the detection by rosat or rxte of a @xmath48 erg s@xmath5 source in 47 tuc ( associated with an lmxb in outburst ) .
the only other reasonable candidate is w17 , a 103 ct source that is a factor of 3 fainter than the qlmxb in ngc 6397 .
this source is outside the go-7503 fov and unfortunately lies in the middle of a large saturation trail in the go-8267 @xmath0 and @xmath7 images .
however , useful limits can be set from the deep and relatively clean @xmath49 image .
the nearest detectable stars to the x - ray position are at 2.58 , 2.91 and 3.04 @xmath11 , and these have @xmath50 22.4 , 18.3 , & 21.1 .
the fainter two of these are comparable to the cvs w44@xmath10 and w45@xmath10 and the brighter one to a system intermediate between ako 9 and v1 .
we set an upper limit to the detection of a star at the nominal w17 x - ray position ( ie assuming zero error in transforming to _ hst_coordinates ) of @xmath51 .
this is suggestive of a faint optical counterpart like x5 ( edmonds et al .
2002a ) or a faint upper limit as with the qlmxb in ngc 6397 ( ghe01b ) .
other possibilities , besides a qlmxb , are that the source is a soft cv ( e.g. an am her system ) or an active binary .
however , for the latter possibility this source would be a factor of @xmath63 brighter than any of the soft ( and non - flaring ) active binaries .
followup observations with acs should be useful in searching for optical counterparts and determining the nature of this source .
as noted in paper i , 1v36 may be a faint x - ray source ( @xmath3 @xmath52erg s@xmath5 ) , arguing in favor of this object being a cv ( although edmonds et al .
2001 and edmonds et al 2002b both give examples of faint x - ray sources with blue , variable optical counterparts that are msps , not cvs ) .
much deeper _ chandra _ observations are required ( and have been taken ) to confirm this marginal detection of an x - ray source , but there are a number of arguments against the cv identification : ( 1 ) the period of 19.1 hr ( assuming the observed variations are ellipsoidal ) is too long ( see fig . [ fig.permv ] ) .
ellipsoidal variations should be seen in this low noise time series if the system is a cv and therefore roche - lobe filling ( unless the inclination is very low ) ; ( 2 ) the position of this star in the color - color plot is different from most known field and cluster cvs ; ( 3 ) the upper limit on @xmath2 for 1v36 is lower than that of all 47 tuc cvs ( see fig . [ fig.fxfopt ] ) ; ( 4 ) there is no suggestion of the large flickering present in other optically bright cvs such as v1 , v2 and w25@xmath10 .
note that ( 2 ) and ( 4 ) likely rule out the possibility that several ux uma systems with very small @xmath2 values might have slipped beneath our x - ray detection limit .
we therefore believe that the comprehensive data available for 1v36 does not support the claim by knigge et al .
( 2002 ) that 1v36 is a strong cv candidate and look forward to seeing what stis / fuv spectroscopy and variability says about this system .
other explanations for 1v36 fare little better .
the variable appears to be too faint and blue in @xmath19 to be a cluster blue straggler , and the star is much too bright to be a horizontal branch or blue straggler star in the smc .
an explanation for this object remains elusive .
all of the other blue variables are uncrowded in the _ chandra _ image and consistent with non - detection , and their _ hst _
time series show no evidence for flickering . as with 1v36
this argues against them being cvs .
here , we examine possible alternative explanations for these blue variables : ( a ) background variable stars from the smc ( e.g. rr lyraes ) , ( b ) detached wd binaries , ( c ) very low accretion rate cvs , like those discussed in townsley & bildsten ( 2002 ) , and ( d ) exotic collision products
. a combination of these may also apply . considering possibility ( a )
, we plot two smc rr lyraes ( labeled as r1 and r2 ) in the wf3 cmds of paper i. these have periods of 15.2 hr and 8.7 hr and were independently discovered by two of the authors ( edmonds and gilliland ) using a ground - based variability study of a 47 tuc field off the core ( 14.5@xmath53 ; for background see sills et al .
the rr lyrae light curves have a distinctive asymmetrical appearance that is clearly different from that of the blue variables .
also , only 3v07 lies reasonably close to the expected position of the horizontal branch instability strip .
a second possibility is that the variables are pulsating blue stragglers in the smc .
these generally have shorter periods and smaller amplitudes than rr lyraes , for example the range of periods for the main pulsation modes of the 6 known sx phe stars in 47 tuc are 0.72.4 hr and the range of amplitudes is 0.006 - 0.085 mags ( gilliland et al .
so , while the amplitudes overlap , most of the blue variables have periods that are too long . also , only 3v07 appears to lie reasonably close to the expected blue straggler sequence .
consistent with these negative conclusions , we note that the number of blue smc stars above the msto in our field is likely to be very small judging by the detection of only 4 such stars in the study of a wfpc2 47 tuc field by zoccali et al .
proper motion studies with separate f300w epochs may help confirm the blue variables as 47 tuc members .
possibility ( b ) is attractive because fig .
[ fig.permv]b shows that several of the blue variables have periods that are reasonably close to , but longer than , the values given by equation [ eqt.warner ] for roche lobe filling . however , only 2v30 has colors that are obviously consistent with those of a ms star / carbon - oxygen ( co ) wd binary , and the long period for this system of 1.18 or 2.36 days is difficult to understand as either ellipsoidal variations or heating effects .
the colors of 2v08 are very similar to those of a bright wd , so any ms companion is likely to be extremely faint .
the other objects with all three colors available ( 1v36 , 3v07 , 4v05 and the possible i d for w71 ) , have colors , compared to the ms , that are bluer in @xmath19 than in @xmath27 , behavior that is difficult to reconcile with co wd - ms star binaries . however ,
note from the color - color cmd given in paper i that 3v07 lies very close to the he wd companion to the msp 47 tuc u , suggesting that 3v07 may also be a he wd .
this does not obviously have an msp companion , because of the lack of detected x - ray emission , but it could have a wd companion .
case ( c ) appears to be an unlikely explanation because the stars are too bright to be consistent with the townsley & bildsten ( 2002 ) models , and their colors are not obviously consistent with co wds . without having a satisfactory explanation for most of these stars , we turn to possibility ( d ) that at least some of them represent an exotic collision product formed near the center of 47 tuc .
mathieu et al . (
2002 ) point out , in their discussion of red stragglers in m67 , that stellar encounters are common in star clusters and that it would not be surprising to discover products of such encounters ( especially binaries ) that run counter to standard evolutionary theory of single stars .
such expectations apply even more so to 47 tuc .
we have discovered ( or confirmed ) optical counterparts for 22 _ chandra _ sources that are cv candidates as summarized in tables 3 and 4 of paper i. we have included v3 in this list , but , as noted above , it may be a qlmxb . of these 22 cv candidates , definite variability of some type has been seen for all of them except w33@xmath10 , w45@xmath10 , w49@xmath10 , w70@xmath10 , w82@xmath10 and w85@xmath10 . the excellent astrometric match between w45 and w45@xmath10 , the tiny chance of this being a chance coincidence ( 0.012% ) and the relatively bright , hard nature of the x - ray source makes this i d secure .
the other candidates have larger possibilities of being chance coincidences ( 0.8% , 1.0% , 1.4% , 2.3% and 0.2% ) and so without having independent information such as variability or an h@xmath54 excess there remains the possibility that one or even more of these 5 ids is not real . also , w70 , w82 and w85 are relatively soft soures ( fig . [ fig.xcol ] ) and therefore one or two of them may be msps ( see [ sect.alter ] ) . among the marginal candidates we believe that w140@xmath10 and w55@xmath10 have the greatest chances of being cvs , as explained in paper i. the photometric properties of these stars were summarized in paper i. the other properties of these systems
are summarized as : ( 1 ) the @xmath2values are higher , on average , than those of all classes of cv in the field sample of vbr97 , partly because of higher than average x - ray luminosities and partly because of faint optical magnitudes ; ( 2 ) the periods , where available , are generally consistent with expectations for roche lobe filling objects ; ( 3 ) the radial distributions are consistent with those of the cluster msps and the brightest ms - ms binaries , and therefore with wd - ms star binaries ; ( 4 ) three of the cvs are eclipsing ( w8 , w15 and ako 9 ) , several show significant flickering ( v1 , v2 , v3 , & w25@xmath10 ) , two of them have previously been seen in outburst ( v2 and ako 9 ) and several others show large amplitude variations ( e.g. w51@xmath10 , v3 , w56@xmath10 ) that may also be signs of outbursts . the period/@xmath1 data presented in [ sect.permv ] and
the faint optical magnitudes presented in [ sect.fxfopt ] , plus the apparent lack of non - magnetic novalike systems suggest that the 47 tuc cvs are dominated by low accretion rate cvs , i.e. dne , it is unlikely that qlmxbs , with their large @xmath2values , are significantly biasing our results , since based on x - ray spectral information and @xmath2 data , only v3 is a reasonable qlmxb candidate ] .
this result is not necessarily inconsistent with the lack of outbursts seen for 47 tuc cvs , since some dne ( such as su uma systems ) can have quite long recurrence times ( warner 1995 ) .
shara et al .
( 1996 ) mention this possibility for 47 tuc . however , the outbursts seen for v2 and ako 9 , and the possible outbursts shown in paper i ( the go-7503 variability ) do not appear to be as dramatic as those seen in su uma systems , and the @xmath3 distributions of the 47 tuc cvs and field su umas appear to be incompatible . also , all of the 47 tuc cvs except for w21@xmath10are found above the period gap , while nearly all known su uma systems are found below the period gap .
therefore , we believe it is unlikely that a significant fraction of the 47 tuc cvs are su uma systems .
selection effects in the field will almost certainly conspire against the detection of low accretion rate cvs ( warner 1995 ) . for example , nova - like systems are relaitvely bright and blue in the optical .
conversely , su uma systems like wz sge are extremely faint at optical and x - ray wavelengths , and have very long recurrence times for outbursts . also , cvs found in the field have a range of non - uniform selection criteria .
although cvs with very low accretion rates ( like those modeled in townsley & bildsten 2002 ) will be difficult to detect in 47 tuc , the sample of cluster cvs is likely to be relatively complete down to x - ray luminosities of 35@xmath55erg s@xmath5 .
therefore , given ( 1 ) the depth of our observations , ( 2 ) the uniform search methods used , and ( 3 ) that the 47 tuc cvs are effectively all at the same distance , it may be possible that our 47 tuc cvs are a more representative sample of cvs than the field objects of vbr97 , and that they therefore show an expected bias towards lower accretion rate systems .
is the lower metallicity of 47 tuc compared to those found in field ( population i ) cvs likely to result in lower accretion rates ?
stehle , kolb , & ritter ( 1997 ) have studied the long - term evolution of cvs with low metallicity secondaries ( z=@xmath56 ) and have shown that such systems have a smaller period gap and a slightly higher mass transfer rate than cvs where the secondary has a solar composition .
this trend therefore does not explain the low apparent accretion rates for the 47 tuc cvs .
population synthesis modeling by stehle , kolb , & ritter ( 1997 ) shows that most population ii cvs should be old enough to have already evolved beyond the period minimum , and will be extremely faint .
however , this age effect does not explain the low accretion rates for the 47 tuc cvs , since the periods for the cluster cvs , where known , are nearly all above the period gap ( and most of the cvs with unknown periods should also be found above the period gap ) .
a potential problem with the low accretion rate interpretation is that no known class of field cv has both @xmath3 and @xmath1 distributions that are similar to those of the 47 tuc cvs . as noted above
, the x - ray luminosities of the 47 tuc cvs are much higher than those of the u gem systems , and are consistent with those of dq her systems .
so , while the faint optical magnitudes imply that they have relatively low accretion rates , the high x - ray luminosities might imply that they have relatively high accretion rates .
only one of these possibilities , at most , can be correct .
since @xmath1 is clearly a better indicator of accretion rate than @xmath3 for non - magnetic cvs ( vbr97 ) , and @xmath1 is also known to strongly depend on accretion rate for magnetic systems ( e.g. hessman , g " ansicke , & mattei 2000 and yi 1994 ) , we speculate that @xmath3 for the 47 tuc cvs is a poor guide to their accretion rates whether they are magnetic or non - magnetic systems .
we note that two systems with presumably relatively low accretion rates are v2 ( a dn ) and w56@xmath10 , a possible dn .
these are two of the brightest cvs in x - rays , with @xmath3@xmath57erg s@xmath5 and @xmath58erg s@xmath5 respectively , hinting that high x - ray luminosities are compatible with low accretion rates .
we encourage theoretical work on globular cluster cvs to try to explain these results , and caution that the low accretion rate interpretation given here is not secure until the combined x - ray and optical luminosities are understood . given the relatively high x - ray luminosities of the brightest cvs in 47 tuc and the observation that about 25% of field cvs are magnetic ( wickramasinghe & ferrario 2000 ) , is it possible that most of the brighter cvs are dq her systems ?
solid evidence exists for dq her - like behavior in v1 ( ghe01a ) , and the strong resemblance between the spectra of ako 9 and gk per suggests that ako 9 is also a dq her system ( knigge et al .
unfortunately , these two systems represent just a small fraction of the total cv population in 47 tuc , and the period vs @xmath1 and @xmath2data presented in [ sect.permv ] and [ sect.fxfopt ] shows that the 47 tuc cvs are statistically different from dq her systems found in the field ( field dq hers have , on average , significantly brighter optical magnitudes ; fig . [ fig.lxmv]b ) .
this contrasts with the similarity between the @xmath3 distributions of the 47 tuc cvs and the field dq her systems ( see fig .
[ fig.lxmv]a ) .
the @xmath3 and @xmath1 values for the 47 tuc cvs are consistent with the data presented for ngc 6397 by cool et al .
( 1998 ) , edmonds et al . ( 1999 ) and ghe01a .
the @xmath3 values of the ngc 6397 cvs are high ( like those in 47 tuc ) , since 4 out of 9 systems have @xmath3 @xmath59 erg s@xmath5(ghe01b )
. optically the ngc 6397 cvs are faint , despite being expected to have even brighter secondaries than 47 tuc ( because of the low metallicity , [ fe / h]@xmath60 ; harris 1996 ) .
the absolute magnitudes of the 4 relatively bright cvs from cool et al .
( 1998 ) and edmonds et al .
( 1999 ) range between 5.95 and 8.81 .
further information about the ngc 6397 cvs are available because good quality optical spectra have been obtained .
arguments have been made that the cvs in ngc 6397 may be dominated by dq hers ( grindlay et al .
1995 and edmonds et al .
1999 ) , based on the observation that 3 of the 4 ngc 6397 cvs with measured optical spectra have relatively strong he ii 4686 lines , like those found in field dq hers .
figure 8 of edmonds et al .
( 1999 ) shows an analysis of the absolute magnitudes of the accretion disks [ @xmath1(disk ) ] for dq her systems and the ratio between the continuum levels of h@xmath61 and h@xmath54 [ c(h@xmath61)/c(h@xmath54 ) ] for different classes of cv . for field
dq her systems , linear relationships are found between @xmath1(disk ) and the he ii 4686 to h@xmath61 equivalent width ratio ( heii / h@xmath61 ) , and a similar relationship is found between [ c(h@xmath61)/c(h@xmath54 ) ] and heii / h@xmath61 ( brighter and bluer disks , equivalent to higher accretion rates , correlate with larger heii / h@xmath61 values ) .
the 4 ngc 6397 cvs discussed in edmonds et al .
( 1999 ) have @xmath1(disk ) , [ c(h@xmath61)/c(h@xmath54 ) ] and heii / h@xmath61 values that are consistent with the field dq her systems , but intriguingly , the 3 ngc 6397 cvs with moderate heii / h@xmath61 values are all found in the low accretion rate part of the figure , with low @xmath1(disk ) and low c(h@xmath61)/c(h@xmath54 ) values .
therefore , the low accretion rates apparently found for the 47 tuc cvs , combined with the hints of magnetic behavior in a couple of systems , may be consistent with the ngc 6397 results .
optical spectra of the fainter cvs in ngc 6397 and the large population of cvs in 47 tuc are needed to test whether they also have relatively strong he ii 4686 ( useful spectra will be difficult to obtain because the cvs are mostly faint and crowded ) .
searches for rapid wd spin in the optical and x - rays is also needed .
perhaps these observations will identify a new class of low accretion rate , magnetic cvs in globular clusters .
one final possibility worth considering is that somehow the different formation mechanism for globular cluster cvs compared to field cvs is responsible for differences in their respective behavior . to test this hypothesis
we note that cvs in open clusters are much more likely to be formed from primordial binaries than the 47 tuc cvs .
interestingly , the two cvs discovered in the open cluster ngc 6791 by kaluzny et al .
( 1997 ) are both much bluer ( in quiescence ) in the @xmath0 vs @xmath19 cmd than most of the 47 tuc and ngc 6397 cvs ( see edmonds et al .
1999 ) , suggesting brighter accretion disks and higher accretion rates than the 47 tuc and ngc 6397 systems .
indeed , one of the ngc 6791 cvs appears to be either a ux uma or z cam system , with a relatively high accretion rate . only one other cv is known in an open cluster , a faint am her system ( eu cancri ) in m67 ( gilliland et al .
1991 , see fig .
[ fig.permv ] ; belloni , verbunt , & mathieu 1998 ) . clearly ,
any two randomly selected cvs from 47 tuc would not look like the ngc 6791 cvs , but obviously the statistics are very poor .
larger samples of cvs in rich open clusters and low density globular clusters are needed to see if this effect is significant , or just a statistical anomaly .
we have a total of 29 likely active binaries in 47 tuc .
twenty seven of these show statistically significant , mostly periodic , variability , and most of them are found on , or slightly above the ms or subgiant ridge - line except for a handful of red stragglers or red straggler candidates . the total sample of active binaries in 47 tuc above our x - ray detection threshold will inevitably be larger , since a sensitive variability study was only possible with the go-8267 data . a key question is whether the active binaries presented here have similar properties to those of by dras and rs cvns found in the field . here
, we present a brief comparison with the rosat all - sky survey results of dempsey et al .
figure 4 of dempsey et al .
( 1997 ) shows that the x - ray luminosities of rs cvns and by dras are quite different from each other , with median x - ray luminosities of @xmath62erg s@xmath5 for the by dras and @xmath63erg s@xmath5 for the rs cvns and maximum luminosities of @xmath64erg s@xmath5 for the by dras and @xmath65erg s@xmath5 for the rs cvns .
these x - ray luminosities are the quiescent values , as dempsey et al .
( 1997 ) observed flares on all of their targets during their survey and removed data points with enhanced count levels . the red stragglers ( w3@xmath10 , w43@xmath10 and w72@xmath10 ) , the red straggler candidates ( w4@xmath10 , w14@xmath10 and w38@xmath10 ) and the stars near the msto like w9@xmath10 and w75@xmath10 generally have x - ray luminosities ( see ghe01a ) that are in good agreement with the dempsey et al .
( 1997 ) data if the primary stars are subgiants and these systems are rs cvns .
the two red stragglers ( and possible sub - subgiants ) in m67 both have @xmath3= @xmath66erg s@xmath5 ( belloni , verbunt , & mathieu 1998 ) , in good agreement with the x - ray luminosities of w3@xmath10 and w72@xmath10 . also , the relatively bright @xmath0 mags ( and in some cases obviously red colors ) of the x - ray bright by dras ( fig . [ fig.radall]d ) are consistent with the hypothesis that a reasonable number of these objects have at least partially evolved primaries and hence are rs cvns . among the active binaries with optically fainter ids
the source w47 is probably returning from a flare , and x - ray variability in other sources such as w18 , w41 and w94 may also be signs of flares , explaining the relatively high luminosities even though these systems are likely by dras .
the luminosities for the optically faint and apparently non - x - ray variable systems like w26 , w59 and w66 are obviously more extreme .
some of them could have x - ray variability that is intrinsically quite large , but is not detected because of the faintness of the sources .
another possibility is that they are simply outliers in a large population of objects having a distribution that is similar to those found in the field .
agb01 point out that large numbers of by dras may have been missed in their study due to incompleteness , since the binary frequency determined from the by dra sample ( with periods that are mostly a few days long ) is a factor of 17 smaller than that calculated from the shorter period eclipsing and contact binaries .
assuming that the binary frequency is the same among the by dras in the agb01 study as it is among the short period eclipsing and contact binaries , and scaling from the sample of 55 agb01 binaries with @xmath67 this would imply a population of about a thousand by dras in 47 tuc . since figure 4 of dempsey et al .
( 1997 ) shows that @xmath68% of the field by dras have @xmath3 @xmath4erg s@xmath5 , we would therefore have a deficit of x - ray detected by dras with @xmath3 @xmath4erg s@xmath5 , rather than a surplus .
possible reasons for this deficit are that a reasonable fraction of the binaries with periods longer than about 12 days have been destroyed by interactions , or that there is an anti correlation between @xmath3 and period ( see below ) . the plot of x - ray color vs @xmath0 in fig .
[ fig.xcol ] shows that most of the optically faint active binaries are relatively soft sources , consistent with the expected soft x - ray spectra for these objects in quiescence ( dempsey et al .
1993 and dempsey et al . 1997 ) .
the source w47 is a relatively hard source and is variable , and flares are expected to harden the x - ray spectrum ( as noted in paper i w64 may show long - term variability ) .
if the bright active binaries have a substantial fraction of rs cvns and the faint systems are all by dras then there may be an anti - correlation between @xmath3 and @xmath0 ( ie the optically fainter active binaries being associated with fainter x - ray sources ) .
we found no statistically significant evidence for such an anti - correlation , either with or without the x - ray variables ( by contrast the cvs show weak anti - correlations between @xmath3 and @xmath0 , especially when ako 9 is removed from the sample ) . with a much larger number of ( mostly undetected ) by dras than rs cvns it is much more likely that a few very luminous by dras are detected rather than a few very luminous rs cvns , possibly explaining the lack of correlation noted above . in agreement with the trends shown in fig .
[ fig.rad-bright-faint]d , there is a significant anti - correlation between period and @xmath0 mag for the active binaries .
the linear correlations between period and @xmath0 are @xmath69 and @xmath70 when using two different tests ( the ` kendl1 ' and ` pearsn ' subroutines ) from numerical recipes ( press et al .
1992 ) , with chance probabilities of only 0.4% and 1% .
presumably the main cause of this anti - correlation is that the brighter active binaries contain a number of systems with evolved stars and relatively long periods .
however , why have only the short period by dras ( the faint active binaries ) been detected in x - rays ? according to dempsey et al .
( 1997 ) , there is a fairly weak correlation between x - ray flux ( @xmath36 ) as measured at the earth and rotation period ( @xmath71 ) of : @xmath36@xmath72 for by dras ( a stronger correlation exists for rs cvns ) . assuming that @xmath3@xmath73 for the active binaries , then the median period of 0.4 days for the faint active binaries ( the by dras ) and 1.56 days for the faint agb01 binaries means that @xmath3 would be only 17% lower for the agb01 binaries .
this is probably not a large enough difference to explain the very different period distributions , although a steeper slope ( but still remaining within the 1-@xmath11 error limit ) could easily produce a 30 - 40% difference between the x - ray luminosity of the faint active binaries and the faint agb01 binaries .
we also note that the period distribution of the dempsey et al .
( 1997 ) sample is very different from ours , since only 3/29 of the dempsey et al .
( 1997 ) by dras with known periods have periods @xmath9 1 day , but 15/27 of the 47 tuc active binaries with known periods have periods @xmath9 1 day ( our photometrically selected sample is biased towards short periods , explaining part of this effect ) . with this different sample of periods ( with a much stronger bias towards short periods )
it is not surprising that we find a different @xmath36/period relationship . in an important sense
our sample of active binaries in 47 tuc ( at fixed metallicity and age ) is significantly cleaner than the dempsey et al .
( 1997 ) sample containing an inhomogeneous sample of stars with a range of ages and metallicities .
therefore , the examination of trends such as the dependence of @xmath36 on period may be optimally performed with cluster samples ( we defer this study to a future publication following analysis of deeper _ chandra _ observations obtained in late 2002 ) . here , we examine the possibility that some of the active binary candidates could be msps .
such a population would be very interesting for dynamical reasons ( see [ sect.intro ] ) , and it has a direct impact on the number of msps in 47 tuc estimated from the _ chandra _ data . the large range in the estimated number of 47 tuc msps ( 3590 ) given by grindlay et al .
( 2002 ) is dominated by assumptions about the nature of the active binary population ( see also [ sect.revised ] ) .
we begin by noting that the eclipsing binary x - ray sources ( w12@xmath10 , w92@xmath10 , w137@xmath10 , & w182@xmath10 ) are unlikely to be msps ( or cvs ) because two eclipses would not be expected . also , the x - ray sources in the agb01 sample classified as w umas ( w41@xmath10 , w47@xmath10 , & w163@xmath10 ) have periods and colors that obey the rucinski relationship for such systems ( agb01 ) , therefore making the contact binary explanation a more natural one , although here we do not rule out an msp explanation .
the radial distributions of the active binaries presented in
[ sect.rad ] offer powerful constraints on this issue .
there are striking similarities between the radial distributions of the active binaries and the agb01 binaries ( with the active binaries removed ) in both the bright and faint samples ( see [ sect.rad ] and fig . [ fig.rad-bright-faint ] ) .
the similarity in the bright samples suggests that the bright active binaries are not dominated by msps with ` normal ' ms companions , because then mass segregation would mean that these @xmath62.152.25m@xmath31 objects ( 1.4m@xmath31 neutron star + 0.750.85m@xmath31ms star ) should be much more concentrated towards the center of the cluster than the lower mass objects ( the radio detected msps and the cvs ) .
there is a chance that some of the bright active binaries could be msps with low mass ( @xmath74m@xmath31 ) ms star companions that have been heated by x - rays so that they appear optically like msto stars or red stragglers , as may be occurring for 6397-a ( ferraro et al .
2001 ) . in principle the radial distribution of these objects
could then mimic those of msto binaries , however , this would require a coincidence and would not explain an apparent absence of msp binaries with ms star companions having masses close to the turn - off value of @xmath60.85m@xmath31 .
a second possibility is that binary - binary and binary - single star interactions resulting in dynamical kicks may cause the average msp - ms star binary to be further away from the center of the cluster than expected for mass segregation of such objects .
however , this would require another coincidental similarity between radial distributions .
it would also be inconsistent with the difference noted in
[ sect.rad ] between the radial distributions of the bright and the faint active binaries .
this difference probably would not have been seen if msps dominate both the faint and the bright groups because then the percentage difference in mass between the two groups would have been much smaller ( as shown in fig .
[ fig.rad-bright-faint ] the faint active binaries are unlikely to contain a significant number of msps based on their radial distribution ) .
when these radial distribution results are combined with the evidence that ( 1 ) the faint active binaries fall above the ms ridge - line just like the agb01 sample of binaries ( see paper i ) , and ( 2 ) that the x - ray luminosities are consistent with rs cvn and by dra systems ( [ sect.abdisc ] ) , we conclude that the active binaries are indeed dominated by ms - ms binaries .
is this finding consistent with the properties of the known msp population ?
of the 16 msps with timing positions ( freire et al .
2001 ) only half of them are binaries and of these only 5 have companion masses likely to be above the minimum value ( 0.085m@xmath31 ) for being on the ms ( the other 3 have masses of @xmath75m@xmath31 ) .
it is likely that these 5 msps _ all _ have he wd companions based on the radio data ( camilo et al .
2000 ) , e.g. 47 tuc u has an _ hst_-identified
he wd companion ( edmonds et al .
2001 ) while a faint , blue star and possible i d has been found for 47 tuc t and faint limits can be set in the optical for 47 tuc h. only one ms companion was found ( 47 tuc w ; edmonds et al .
2002b ) in the full sample of 20 msps from camilo et al .
( 2000 ) . therefore ,
if the sample of 16 msps is representative of the possible much larger sample of msps ( camilo et al .
2000 ) then optical companions are rare and ms companions even more so . if a large population of msps with ms companions does exist , and we are detecting many of them in x - rays , then the msps would have to be intrinsically radio - faint to avoid any of them being detected by parkes . however , the relatively flat correlation between x - ray and radio luminosity shown by grindlay et al .
( 2002 ) implies that the faintest msps in the radio are less faint as x - ray sources .
given the small number of detections of cluster msps with likely non - degenerate companions ( two ; 47 tuc w and 6397-a ) it is not yet possible to conclude whether such msps are generally relatively bright in x - rays but faint in the radio .
the msp 6397-a would have been easily detected by parkes if it were in 47 tuc , given the radio luminosity of 5 mjy kpc@xmath76 quoted by damico et al .
the radio luminosities of the 47 tuc msps range between 0.81 and 10.9 mjy kpc@xmath76 , with 12 out of 14 of them having radio luminosities less than that of 6397-a .
the msp 47 tuc w , however , is one of the faintest detected msps , and w34 , if it is an msp , is below the detection limit of camilo et al .
the x - ray luminosities for 6397-a and 47 tuc w are slightly higher than the most luminous ` normal ' msp in 47 tuc . without having a deeper sample of msps , and detailed radial velocity information for the active binaries
it is very difficult to rule out the possibility that any _ individual _ , non - eclipsing , optical variable has an msp companion , especially in view of the cases of w29@xmath10 and 6397-a ( and possibly also w34@xmath10 ) .
the former object is very crowded optically and was only discovered using difference image analysis ( edmonds et al .
2002b ) , but was found on close study to have blue colors .
therefore , w29@xmath10 would have been classified as a cv instead of an active binary if not for the remarkable period and phase coincidence with 47 tuc w discovered by edmonds et al .
( 2002b ) , perhaps implying that one or two of the faint cv candidates could be an msp .
none of the active binary candidates show blue colors , and they also show no hint of the large amplitude variability seen in either w29@xmath10 or w34@xmath10 , despite in some cases having periods that are not much longer than those of w29@xmath10(suggesting similar levels of variability caused by irradiation from an msp , if present ) . one potential source of msps masquerading as active binaries are the red stragglers . for example , the companion to the msp 6397-a was previously identified as a by dra by taylor et al .
( 2001 ) on the basis of h@xmath54 emission , a cmd position lying above the ms and optical variability ( this star is clearly a red straggler ) .
however , it was subsequently shown by ferraro et al .
( 2001 ) to have an msp companion based on an accurate radio timing position for the msp . naturally , it is possible for a binary system to be both an active binary and to have an msp companion , since secondaries in short period binaries like 6397-a are undoubtedly tidally locked and rapidly rotating , and hence may have substantial chromospheric activity ( as suggested for 6397-a by orosz & van kerkwijk 2002 ) .
although the 47 tuc red stragglers and 6397-a have similar optical properties , there are some key differences between their x - ray properties . while 6397-a is a reasonably hard source ( as is 47 tuc w ) , the two red stragglers w3 and w72 both have very soft x - ray colors and the two red straggler candidates w4 and w14 also have soft colors ( see fig . [ fig.xcol ] and fig .
[ fig.xcmd ] ) . of the red stragglers , only w43 ( in go-7503 ) has a hard color . generally ,
if most of the active binary candidates are msps and are like 47 tuc w and 6397-a , then they should mostly be hard sources . yet , of the go-8267 active binaries with x - ray color determinations from grindlay et al .
( 2002 ) only 4 ( not including w47@xmath10 ) are hard while 9 ( not including w4 ) are soft .
the leading candidates for msps hidden within our active binary sample include w43 , a faint , moderately hard , variable x - ray source ( like 6397-a ) with an optical i d that falls in a similar region of the optical cmd to 6397-a , as noted above .
no clear optical variability is present , with our limited quality time series .
we note that while w43 is clearly harder than the other red stragglers , this could be because of flaring activity , explaining the variability .
both w23 and w64 are moderately hard sources that were not observed to vary during the _ chandra _ observation of ghe01a , and are brighter than all of the other active binaries ( except w47 ) . however , as explained earlier , w64 may show long - term variability
. the leading candidates for identifications of other 47 tuc u - like msps ( not yet detected in the radio ) , are the weak , soft , non - variable x - ray sources , with faint , blue , optical companions that do not show long - term variability . these are w82 and w85 ( see fig .
[ fig.xcol ] and fig .
[ fig.xcmd ] ) , with plausible optical ids ( lying outside the go-8267 fov ) without obvious long - term variability .
they are beyond the faint limit of the go-7503 data for short - term variability studies .
the i d for w70 shows no evidence for periodic variation in our time series and only marginal evidence for non - periodic variations , but we do not yet have long - term variability information for this object . these sources could be cvs , since am her systems for example are known to be soft x - ray sources ( vbr97 ) .
our high quality h@xmath54 observations with _
hst_/acs may help distinguish between these two possibilities , since he wds should show broad h@xmath54 absorption lines , and cvs strong h@xmath54 emission lines , as observed for the cvs in ngc 6397 ( grindlay et al
. 1995 & cool et al .
1998 ) and ngc 6752 ( bailyn et al .
1996 ) .
finally , we examine the soft , faint , apparently non - variable x - ray sources that have no plausible ( optically variable or photometrically unusual ) counterparts ( see fig . [
fig.xcmd]b ) .
grindlay et al .
( 2002 ) listed the sources in this category , noting in particular the subset of these sources that fall in the go-8267 fov : w4 , w5 , w6 , w24 , w28 , & w98 ( we now add the source w71 to this list ) . as noted by grindlay et al .
( 2002 ) these sources include the leading candidates to be msps . by correcting for the incomplete spatial coverage of the wfpc2 go-8267 fov , and for incompleteness in the detection of radio - detected msps as soft x - ray sources , grindlay et al .
( 2002 ) estimated the presence of at least 19 msps with @xmath3 @xmath77erg s@xmath5 that do not have radio counterparts with timing positions , in addition to the 16 msps with such positions .
this estimate obviously increases if some of the active binaries identified with soft sources are instead msps , but as argued above , few are expected . also , the estimate by grindlay et al .
( 2002 ) implicitly assumes that 2/9 of the msps have x - ray colors @xmath78 , so a small fraction of misidentified ` active binaries ' with harder x - ray colors ( like w9 or w59 ) that are really msps , do not change these statistics .
based on a study of the optical data , is it reasonable that the 7 soft sources just discussed are all msps , rather than cvs or active binaries in some cases ?
first , as already noted , w4 is a candidate red straggler and w24 may be an rs cvn , and both of them have higher x - ray count levels than the brightest detected msp ( except for the hard x - ray source 47 tuc w ) .
also , w98 is located close to several giant stars so it may also be an rs cvn . to estimate how many active binaries , or perhaps cvs , may have missed detection because of crowding , we have examined the stellar count levels present in the go-8267 data around the positions of the 7 soft sources .
figure [ fig.testsky ] shows the mean stellar counts in an annulus ( between 09 and 135 ) around each : ( 1 ) cv , ( 2 ) bright active binary , ( 3 ) faint active binary ( 4 ) msp , ( 5 ) unidentified x - ray source , and ( 6 ) candidate msp ( the 7 sources discussed above ) .
optical positions are used for ( 1 ) , ( 2 ) , ( 3 ) and ( 6 ) , radio positions for ( 4 ) and x - ray positions for ( 5 ) . the median count level for the candidate msps is higher than any of these other groups , and is 2.0 times larger than the median count level for the msps ( the two distributions differ at the 98.8% level using the ks - test ) .
this suggests that some of the 7 soft sources are active binaries ( or cvs ) that have been missed because of crowding .
we tentatively classify w4 , w24 and w98 as active binaries , based on their promixity to bright stars , and classify the 4 remaining sources as msps .
this implies a population of 13 extra msps for a total of 29 msps with @xmath3 @xmath79 erg s@xmath5 . incorporating the optical identifications reported here and in paper i ( plus radio / msp identifications from ghe01a and grindlay et al .
2002 ) and just considering the go-8267 fov alone ( where our high quality time series coverage is spatially complete ) a significant fraction of the 78 sources are active binaries ( 34.6% ) compared to 19.2% for cvs and 9.0% for msps , with the source w46/x7 being a qlmxb ( 1.3% ) .
these sources , where x - ray colors are available , are labeled in the x - ray cmd shown in fig .
[ fig.xcmd ] . the remaining 27 ( 34.6% )
sources in the go-8267 fov have uncertain identifications , or have no plausible i d . of these sources
, we use the x - ray luminosity and color , or the plausible match - up with bright , blue or marginally variable stars to tentatively identify w10 , w16 , w17 , w20 , w32 , w35 , w37 and w140 as cvs , and w4 , w24 , w37 , w71 , w93 , w141 and w168 as active binaries leaving 13 faint sources that we identify as possible msps ( see table [ tab.undet8267 ] ) .
this leaves us with final fractions of 42.3% active binaries , 29.5% cvs and 26.9% msps ( plus the single qlmxb ) , although some of the 13 weak x - ray sources could be cvs or active binaries , lowering the msp number .
the cv fraction is very similar to that calculated by ghe01a , but the active binary fraction is significantly higher than the 15% estimated by ghe01a .
correspondingly , the msp estimate , and the one given in [ sect.revised ] , are smaller than the value of @xmath650 msps originally presented for the 108 sources in ghe01a ( that estimate was based on incomplete analysis of the optical data ) . based on the period/@xmath1 analysis and the @xmath1 distribution presented here ,
the 47 tuc cvs appear to have lower accretion rates than field cvs systems found above the period gap
. theoretical work on cv evolution predicts that the fainter , shorter period cvs expected to exist in 47 tuc ( but not observed because of crowding ) should have even lower accretion rates than the bright ones that we do observe ( see di stefano & rappaport 1994 and townsley & bildsten 2002 ) . the suggestive evidence for low accretion rates and dq her type behavior in cvs in ngc 6397 , combined with : ( 1 ) the evidence for magnetic behavior in a couple of the 47 tuc cvs ( v1 and ako 9 ) and ( 2 ) the high ( dq her - like ) x - ray luminosities of the 47 tuc cvs implies that magnetic activity may play an important role in globular cluster cvs .
however , more work is needed to understand the unusual @xmath1 and @xmath3distributions for the cvs in both 47 tuc and ngc 6397 .
perhaps the most unexpected result , when compared with other clusters , is that the active binaries outnumber the cvs . in ngc 6397 ,
the detected cvs outnumber the active binaries by a factor of @xmath623 ( based on deeper _ chandra _ data than obtained for 47 tuc ) , but significant depletion of the ms binary population in ngc 6397 is likely to have occurred during and after core collapse . in ngc 6752 ,
the detected cvs outnumber the active binaries by a factor of @xmath6410 , but the ngc 6752 _ chandra _ data is not as deep as the 47 tuc data , and several unidentified soft sources may still be identified with active binaries , given optical data of higher quality . with this large population of detected active binaries in 47 tuc , many of them soft sources , we find support for the lower range given by grindlay et al .
( 2002 ) for the number of msps in 47 tuc with @xmath3 @xmath80erg s@xmath5(@xmath63040 ) .
it is possible that a few of the active binaries may really be msps , and we continue to search for new objects that are similar to 47 tuc u or 47 tuc w , but , given the spatial , photometric , x - ray spectral and timing information available for the observed active binaries , a large population of msps masquerading as active binaries is unlikely .
the lower estimated number of msps in 47 tuc means that the ratio between the number of msps and the number of cvs is smaller , and closer to the corresponding value found in ngc 6397 by ghe01b .
a comparison of figures [ fig.fxfopt-ver ] and [ fig.fxfopt]a shows that we are sensitive to detection of most classes of cv except for the faintest dne ( su umas ) , and possibly also double degenerate systems ( where the statistics are poor ) .
there are good prospects for finding at least some of the fainter cv population not yet detected here by using the much deeper x - ray limits set by the 300 ks acis - s observation .
although the gains will be limited near cluster center because of x - ray crowding , the faintest cvs should have low - mass companions of @xmath81m@xmath31 , and with wd masses of @xmath60.55m@xmath31 the total system masses should be less than or equal to the msto mass . therefore , many of these objects should be found in relatively uncrowded regions more than 20 - 30@xmath82 away from the cluster center .
unfortunately , optical identifications will be difficult or impossible for many of these objects because of crowding and the high background levels present in the _ hst _
images , a penalty for studying objects in this massive , concentrated globular cluster .
the 300 ks followup _ chandra _ observations will provide much better quality spectral , timing and positional information , especially for the faintest sources discussed here , and these data should be very useful for distinguishing between msps and active binaries .
many new , faint sources will also likely be discovered , dominated by msps and active binaries .
new radio observations are obviously needed to search for the @xmath6200 undetected msps believed to exist in 47 tuc ( camilo et al . 2000 ) .
we acknowledge comments on the cvs from brian warner , and a number of extremely helpful comments from the referee , christian knigge . this work was supported in part by stsci grants go-8267.01 - 97a ( pde and rlg ) and hst - ar-09199.01-a ( pde ) .
albrow , m. d. , gilliland , r. l. , brown , t. m. , edmonds , p. d. , guhathakurta , p. , & sarajedini , a. 2001 , , 559 , 1060 ( agb01 ) augusteijn , t. , van der hooft , f. , de jong , j. a. , & van paradijs , j. 1996 , , 311 , 889 bailyn , c. d. , rubenstein , e. p. , slavin , s. d. , cohn , h. , lugger , p. , cool , a. m. , & grindlay , j. e. 1996 , , 473 , l31 belloni , t. , verbunt , f. , & mathieu , r. d. 1998 , , 339 , 431 camilo , f. , lorimer , d. r. , freire , p. , lyne , a. g. , & manchester , r. n. 2000 , , 535 , 975 cool , a. m. , grindlay , j. e. , cohn , h. n. , lugger , p. m. , & bailyn , c. d. 1998 , , 508 , l75 window into astrophysics , 277 damico , n. , possenti , a. , manchester , r. n. , sarkissian , j. , lyne , a. g. , & camilo , f. 2001 , , 561 , l89 dempsey , r. c. , linsky , j. l. , schmitt , j. h. m. m. , & fleming , t. a. 1993 , , 413 , 333 dempsey , r. c. , linsky , j. l. , fleming , t. a. , & schmitt , j. h. m. m. 1997 , , 478 , 358 di stefano , r. & rappaport , s. 1994 , , 423 , 274 edmonds , p. d. , gilliland , r. l. , guhathakurta , p. , petro , l. d. , saha , a. , & shara , m. m. 1996 , , 468 , 241 edmonds , p. d. , grindlay , j. e. , cool , a. , cohn , h. , lugger , p. , & bailyn , c. 1999 , , 516 , 250 edmonds , p. d. , gilliland , r. l. , heinke , c. o. , grindlay , j. e. , & camilo , f. 2001 , , 557 , l57 edmonds , p. d. , heinke , c. o. , grindlay , j. e. , & gilliland , r. l. 2002a , , 564 , l17 edmonds , p. d. , gilliland , r. l. , camilo , f. , heinke , c. o. , & grindlay , j. e. 2002b , , 579 , 741 edmonds , p. d. , gilliland , r. l. , heinke , c. o. , & grindlay , j. e. 2003 , , submitted ( paper i ) ferraro , f. r. , possenti , a. , damico , n. , & sabbi , e. 2001 , , 561 , l93 freire , p. c. , camilo , f. , lorimer , d. r. , lyne , a. g. , manchester , r. n. , & damico , n. 2001 , , 326 , 901 gilliland , r. l. et al .
1991 , , 101 , 541 gilliland , r. l. , bono , g. , edmonds , p. d. , caputo , f. , cassisi , s. , petro , l. d. , saha , a. , & shara , m. m. 1998 , , 507 , 818 gilliland , r. l. et al . 2000 , , 545 , l47 grindlay , j. e. , cool , a. m. , callanan , p. j. , bailyn , c. d. , cohn , h. n. , & lugger , p. m. 1995 , , 455 , l47 grindlay , j. e. , heinke , c. o. , edmonds , p. d. & murray , s. 2001a , science , 292 , 2290 ( ghe01a ) grindlay , j. e. , heinke , c. o. , edmonds , p. d. , murray , s. s. , & cool , a. m. 2001b , , 563 , l53 ( ghe01b ) grindlay , j. e. , camilo , f. , heinke , c. o. , edmonds , p. d. , cohn , h. , & lugger , p. 2002
, , 581 , 470 harris , w. e. 1996 , , 112 , 1487 heinke , c. o. , edmonds , p. d. , & grindlay , j. e. 2001 , , 562 , 363 heinke , c. o. , grindlay , j. e. , lloyd , d. & edmonds , p. d. 2003 , , accepted hut , p. et al .
1992 , , 104 , 981 kaluzny , j. , stanek , k. z. , garnavich , p. m. , & challis , p. 1997
, , 491 , 153 kaluzny , j. & thompson , i. b. 2002 , astro - ph/0210626 knigge , c. , zurek , d. r. , shara , m. m. , long , k. s. , & gilliland , r. l. 2001 , asp conf .
ser . : stellar collisions , mergers , and their consequences knigge , c. , zurek , d. r. , shara , m. m. , & long , k. s. 2002 , , 579 , 752 mathieu , r. d. , van den berg , m. , torres , g. , latham , d. , verbunt , f. , & stassun , k. 2002 , astro - ph/0209568 neill , j. d. , shara , m. m. , caulet , a. , & buckley , d. a. h. 2002 , , 123 , 3298 orosz , j. a. & van kerkwijk , m. h. 2003 , , 397 , 237 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, cambridge : university press , 2nd ed .
richman , h. r. 1996 , , 462 , 404 ritter , h. & kolb , u. 1998 , , 129 , 83 russell , h. n. 1945 , , 102 , 1 shara , m. m. , bergeron , l. e. , gilliland , r. l. , saha , a. , & petro , l. 1996 , , 471 , 804 sills , a. , bailyn , c. d. , edmonds , p. d. , & gilliland , r. l. 2000 , , 535 , 298 stehle , r. , kolb , u. , & ritter , h. 1997 , , 320 , 136 taylor , j. m. , grindlay , j. e. , edmonds , p. d. , & cool , a. m. 2001 , , 553 , l169 townsley , d. m. & bildsten , l. 2002 , , 565 , l35 verbunt , f. , bunk , w. h. , ritter , h. , & pfeffermann , e. 1997 , , 327 , 602 ( vbr97 ) verbunt , f. & hasinger , g. 1998 , , 336 , 895 warner , b. 1995 , cambridge astrophysics series , cambridge , new york : cambridge university press wickramasinghe , d. t. & ferrario , l. 2000 , , 112 , 873 yi , i. 1994 , , 422 , 289 zoccali , m. et al .
2001 , , 553 , 733 u gems & 77.39 & & & & & & & + dq hers & 86.13 & 13.40 & & & & & & + am hers & 97.06 & 42.68 & 0.82 & & & & & + z cams & 99.84 & 99.13 & 87.52 & 92.22 & & & & + ux umas&100.00 & 99.99 & 99.26 & 99.65 & 98.13 & & & + cvs & 97.72 & 98.20 & 93.36 & 98.44 & 99.93 & 100.00 & & + msps & 41.28 & 1.13 & 1.43 & 4.57 & 90.46 & 99.11 & 97.97 & + xabs&100.00 & 100.00 & 99.73 & 99.93 & 99.74 & 8.17 & 100.00 & 99.01 + 20 & 1 & a & cv ? + 24 & 1 & b & ab ? + 28 & 1 & a & msp ? + 31 & 1 & b & msp ?
+ 32 & 1 & d & cv ?
+ 35 & 1 & c & cv ?
+ 37 & 1 & b & cv ?
+ 39 & 1 & b & 47 tuc o + 46 & 1 & b & qlmxb + 96 & 1 & a & msp ? + 97 & 1 & c & msp ?
+ 98 & 1 & d & ab ? + 141 & 1 & c & ab ?
+ 4 & 2 & d & ab ? + 5 & 2 & d & msp ?
+ 10 & 2 & d & cv ?
+ 13 & 2 & c & 47 tuc n + 16 & 2 & d & cv ?
+ 17 & 2 & d & cv ?
+ 19 & 2 & c & 47 tuc g , i + 67 & 2 & c & 47 tuc d + 74 & 2 & c & 47 tuc h + 91 & 2 & a & msp ? + 95 & 2 & b & msp ?
+ 142 & 2 & d & msp ?
+ 168 & 2 & d & ab ? + 115 & 3 & c & msp ? + 6 & 4 & d & msp ? + 7 & 4 & d & 47 tuc e + 71 & 4 & c & msp ? + 93 & 4 & c & ab ? + 99 & 4 & a & msp ? + 140 & 4 & a & cv ? | we report time series and variability information for the optical identifications of x - ray sources in 47 tucanae reported in paper i ( at least 22 cvs and 29 active binaries ) .
the radial distribution of the cvs is indistinguishable from that of the msps detected by freire et al .
( 2001 ) . a study of the eight cvs with secure orbital periods ( two obtained from the _ chandra _ study of grindlay et al .
2001a ) shows that the 47 tuc cvs have fainter accretion disks , in the @xmath0 band , than field cvs with similar periods .
these faint disks and the faint absolute magnitudes ( @xmath1 ) of the 47 tuc cvs suggests they have low accretion rates .
one possible explanation is that the 47 tuc objects may be a more representative sample of cvs , down to our detection threshold , than the cvs found in the field ( where many low accretion rate systems are believed to be undiscovered ) , showing the advantages of deep globular cluster observations .
the median @xmath2 value for the 47 tuc cvs is higher than that of all known classes of field cv , partly because of the faint @xmath1values and partly because of the relatively high x - ray luminosities ( @xmath3 ) .
the latter are only seen in dq her systems in the field , but the 47 tuc cvs are much fainter optically than most field dq hers .
previous work by edmonds et al .
( 1999 ) has shown that the 4 brightest cvs in ngc 6397 have optical spectra and broadband colors that are consistent with dq hers having lower than average accretion rates .
some combination of magnetic behavior and low accretion rates may be able to explain our observations , but the results at present are ambiguous , since no class of field cv has distributions of both @xmath3 and @xmath1 that are consistent with those of the 47 tuc cvs .
the radial distribution of the x - ray detected active binaries is indistinguishable from that of the much larger sample of optical variables ( eclipsing and contact binaries and by dra variables ) detected in previous wfpc2 studies by albrow et al ( 2001 ) .
the x - ray properties of these objects ( luminosity , hardness ratios and variability ) are consistent with those of active binaries found in field studies , and the @xmath2 distribution is significantly different from those of the cvs and the millisecond pulsars ( msps ) that are detected ( or possibly detected ) in the optical . despite these results , we examine the possibility that a few of the active binaries are msps with main sequence companions resulting from double exchanges in the crowded core of 47 tuc .
no solid evidence is found for a significant population of such objects , and therefore using the methods of grindlay et al .
( 2002 ) we estimate that the number of msps in 47 tuc with luminosities @xmath4erg s@xmath5 is @xmath63040 , near the previous lower limit .
we present the results of a new , deeper search for faint low - mass x - ray binaries ( lmxbs ) in quiescence .
one reasonable and one marginal candidate for optical identification of a quiescent lmxb was found ( one is already known ) .
finally , it is shown that the periods of the blue variables showing little or no evidence for x - ray emission are too long for roche lobe filling ( if the variations are ellipsoidal ) .
these blue variables also show no evidence for the large flickering levels seen in comparably bright cvs . at present
we have no satisfactory explanation for these objects , although some may be detached wd - main sequence star binaries . |
a universal weight function of a quantum affine algebra is a family of functions with values in its borel subalgebra satisfying certain coalgebraic properties .
it can be used either to construct solutions of @xmath3-difference knizhnik - zamolodchikov equations @xcite or to construct the off - shell bethe vectors , thus generalizing the nested bethe ansatz procedure @xcite .
a general construction of a weight function has been suggested in @xcite .
it uses the existence of borel subalgebras of two different types in a quantum affine algebra .
one type is related to the realization of @xmath4 as a quantized kac - moody algebra , and the other comes from the `` current '' realization of @xmath4 proposed by drinfeld in @xcite .
it was proved in @xcite that the weight function of the quantum affine algebra can be represented as the projection of a product of so called drinfeld currents on the intersection of borel subalgebras of @xmath4 of different types . in the paper
@xcite the method of the algebraic bethe ansatz was developed for the case of the quantum twisted affine algebra @xmath0 .
in particular there was suggested an inductive procedure for the construction of off - shell bethe vectors for finite dimensional representations of @xmath0 .
the main result of the present paper consists in the applying of the approach of @xcite to obtain an explicit formula for the universal weight function of the quantum twisted affine algebra @xmath0 . as a corollary
we derive an integral presentation for the factors of the universal @xmath5-martix of @xmath0 , connecting the usual and the drinfeld comultiplications , as well as for the universal @xmath5-matrix itself . an analogous formula for the universal @xmath5-martix of @xmath6
is presented in @xcite .
the paper is organized as follows . in section @xmath7 , we describe two realizations of the quantum twisted affine algebra @xmath0 . in section @xmath8 , we adapt the general result of @xcite and @xcite to the quantum twisted affine algebra @xmath0 . in section @xmath9 , we introduce the so called `` composite currents '' and derive their analytical properties , which are crucial for our investigations .
section @xmath10 contains an exposition of the main result . here
combinatorial formulae for the universal weight function and other related objects are presented .
most of the proofs are collected in section @xmath11 .
finally , section @xmath12 contains examples of the universal weight function .
quantum affine algebra @xmath0 is an associative algebra generated by elements @xmath13 subject to certain commutation relations .
the relations are given as formal power series identities for the following generating functions ( currents ) : @xmath14 as follows : ( z - q^2w)(qz+w)e(z)e(w ) & = & ( q^2z - w)(z+qw)e(w)e(z ) , [ ee ] + ( q^2z - w)(z+qw)f(z)f(w ) & = & ( z - q^2w)(qz+w)f(w)f(z ) , [ ff ] + k^+(z)e(w)k^+(z)^-1 & = & ( w / z)e(w ) , [ kpe ] + k^+(z)f(w)k^+(z)^-1 & = & ( w / z)^-1f(w ) , [ kpf ] + k^-(z)e(w)k^-(z)^-1 & = & ( z / w)^-1e(w ) , [ kme ] + k^-(z)f(w)k^-(z)^-1 & = & ( z / w)f(w ) , [ kmf ] + k^(z)k^(w ) & = & k^(w)k^(z ) , [ kk ] + k^-(z)k^+(w ) & = & k^+(w)k^-(z ) , [ kmkp ] + e(z)f(w)-f(w)e(z ) & = & 1q - q^-1 .
[ ef ] where [ a ] ( x)= , and @xmath15 is a formal laurent series , given by [ delta ] ( z / w ) = _ n(z / w)^n . generating functions @xmath16 satisfy the cubic serre relations ( see @xcite ) : [ serre ] _
z_1,z_2,z_3 e(z_1)e(z_2)e(z_3 ) & = & 0 , + _
z_1,z_2,z_3 f(z_1)f(z_2)f(z_3 ) & = & 0 , + _
z_1,z_2,z_3 e(z_1)e(z_2)e(z_3 ) & = & 0 , + _
z_1,z_2,z_3 f(z_1)f(z_2)f(z_3 ) & = & 0 . define the _ principal _ grading on the algebra @xmath0 by the relations [ grading ] e_n= 3n+1,f_n=3n-1,a_n=3n , k_0 ^ 1=0 .
the assignment defines @xmath0 as a graded associative algebra .
the hopf algebra structure on @xmath0 can be defined as follows : ^(d)(e(z ) ) & = & e(z)1+k^-(z)e(z ) , [ dde ] + ^(d)(f(z ) ) & = & 1f(z)+f(z)k^+(z ) , [ ddf ] + ^(d)(k^(z ) ) & = & k^(z)k^(z ) , [ ddk ] + a^(d)(e(z ) ) & = & -(k^-(z))^-1e(z ) , [ ade ] + a^(d)(f(z ) ) & = & -f(z)(k^+(z))^-1 , [ adf ] + a^(d)(k^(z ) ) & = & ( k^(z))^-1 , [ adk ] + ^(d)(e(z ) ) & = & 0 , [ cde ] + ^(d)(f(z ) ) & = & 0 , [ cdf ] + ^(d)(k^(z ) ) & = & 1 .
[ cdk ] where @xmath17 , @xmath18 and @xmath19 are the comultiplication , the counit and the antipode maps respectively .
we will call @xmath17 a _ drinfeld comultiplication_. the map @xmath17 defines the structure of topological bialgebra on @xmath20 which means that for any @xmath21 this map sends an element @xmath22 of the graded component @xmath23 $ ] to the sum of spaces @xmath24\otimes { \mathcal a}[-m(n)+k]$ ] , where @xmath25 is an integer such that for each @xmath26 the corresponding summand on the right hand side of @xmath27 is finite
. see ( * ? ? ?
* section 2 ) for details .
another realization of @xmath0 can be obtained using chevalley generators .
quantum affine algebra @xmath0 is an associative algebra generated by elements @xmath28 , @xmath29 , @xmath30 , @xmath31 , satisfying the following relations : k_e_k^-1 _
= q^1e _ , & & k_e_(-2)k^-1 _
= q^2e_(-2 ) , + k_-2e_k^-1_-2 = q^2e _ , & & k_-2e_k^-1_-2 = q^4e _ , + k_^2 k_-2 = 1 , & & = 0 , + = , & & = , + ( _ qe_)^5 e_(-2 ) = 0 , & & ( _ qe_(-2))^2 e _ = 0 , where ( _ qe_)(x ) & = & e_x - k_^1xk_^1e _ , + ( _ qe_(-2))(x ) & = & e_(-2)x - k_-2 ^ 1xk_-2 ^ 1e_(-2 ) . the principal grading is given by relations e_=1,e_(-2)=1
, k_^1=k_-2 ^ 1=0 . the hopf algebra structure associated with this realization can be defined by ( e_)=e_1+k_e _ , & & ( e_-2)=e_-21+k_-2e_-2 , + ( e_-)=1e_-+e_-k^-1 _ , & & ( e_-(-2))=1e_-(-2)+e_-(-2)k^-1_-2 , + ( k^1_)=k^1_k^1 _ , & & ( k^1_-2)=k^1_-2k^1_-2 , + ( e_)=0 , & & ( e_(-2))=0 , + ( k^1_)=1 , & & ( k^1_-2)=1 , + a(e_)=-k^-1_e _ , & & a(e_-2)=-k^-1_-2e_-2 , + a(e_-)=-e_-k _ , & & a(e_-(-2))=-e_-(-2)k_-2 , + a(k^1 _ ) = k^1 _ , & & a(k^1_-2 ) = k^1_-2 , where @xmath32 , @xmath33 and @xmath34 are the comultiplication , the counit and the antipode maps respectively .
we will call @xmath32 `` standard '' comultiplication .
the isomorphism between two descriptions of the quantum affine algebra @xmath0 can be established by the following maps : [ isom ] rclrcl k_-2 & & k_0 ^ -2 , & k _ & & k_0 , + e_-2 & & a(qf_1f_0-f_0f_1)k_0 ^ -2 , & e _ & & e_0 , + e_-(-2 ) & & ak_0 ^ 2(q^-1e_0e_-1-e_-1e_0 ) , & e_- & & f_0 , where @xmath35 . for generic @xmath3 ,
the mapping above is an isomorphism of associative algebras .
it preserves the counit , but does not respect the comultiplication maps @xmath32 and @xmath17 .
the relation between the two comultiplications is described in section .
in this section , we adapt the theory of orthogonal decompositions and twists of drinfeld double developed in @xcite to the @xmath0 case .
recall that in any quantum affine algebra there exist borel subalgebras of two types .
borel subalgebras of the first type come from the drinfeld ( `` current '' ) realization of @xmath0 .
let @xmath36 denote the subalgebra of @xmath0 , generated by the elements @xmath37 , and let @xmath38 denote the subalgebra of @xmath0 , generated by the elements @xmath39 .
they are hopf subalgebras of @xmath0 with respect to comultiplication @xmath17 .
the `` current '' borel subalgebra @xmath36 contains the subalgebra @xmath40 generated by elements @xmath41 .
the `` current '' borel subalgebra @xmath38 contains the subalgebra @xmath42 , generated by elements @xmath43 .
borel subalgebras of the second type are obtained via the chevalley realization .
let @xmath44 and @xmath45 denote a pair of subalgebras of @xmath0 generated by elements e_,e_-2,k^1 _ e_-,e_-(-2),k^1 _ respectively . in terms of the drinfeld realization ,
these subalgebras are generated by elements k_0 ^ 1,e_0,qf_1f_0-f_0f_1 k_0 ^ 1,f_0,q^-1e_0e_-1-e_-1e_0 , respectively .
the algebras @xmath46 are hopf subalgebras of @xmath0 with respect to comultiplication @xmath32 and , moreover , coideals with respect to the drinfeld comultiplication .
more precisely : [ prop3.1 ] for any element @xmath47 we have an equality [ coid ] ^(d)(x)=x1+a_ib_i b_iu_q(_+),(b_i)=0 .
in particular , subalgebra @xmath44 is a left coideal of @xmath0 with respect to coproduct @xmath17 , i.e. @xmath48 .
analogously , @xmath45 is a right coideal of @xmath0 with respect to the same coproduct .
the proof of proposition is given in section .
let @xmath49 , @xmath50 , @xmath51 and @xmath52 denote the following intersections of borel subalgebras : u_f^-=u_fu_q(_- ) , & u_f^+=u_fu_q(_+ ) , + u_e^+=u_eu_q(_+ ) , & u_e^-=u_eu_q(_- ) . here
the upper sign indicates the borel subalgebra @xmath46 containing the given algebra , and the lower letter indicates the `` current '' borel subalgebra @xmath36 or @xmath38 that it is intersected with .
these letters are capitals if the subalgebra contains imaginary root generators @xmath53 and are lower case otherwise .
the intersections have coideal properties with respect to both comultiplications . in particular , proposition and its analog for borel subalgebra @xmath45
imply the inclusions ^(d)(u_f^+ ) u_fu_f^+ , & ^(d)(u_f^- ) u_f^-u_f , + ^(d)(u_e^- ) u_e^-u_e , & ^(d)(u_e^+ ) u_eu^+_e .
let @xmath54 be a bialgebra with multiplication map @xmath55 , comultiplication map @xmath56 , unit @xmath57 and counit @xmath33 .
we say that its subalgebras @xmath58 and @xmath59 determine an orthogonal decomposition of @xmath54 ( see @xcite ) , if : * the algebra @xmath54 admits a decomposition @xmath60 , such that the multiplication map @xmath61 establishes an isomorphism of linear spaces .
* @xmath58 is the left coideal of @xmath54 , @xmath59 is the right coideal of @xmath54 : ( _ 1)_1 , ( _ 2)_2 .
there is a pair of biorthogonal decompositions of `` current '' borel subalgebras , equipped with comultiplications @xmath17 and its opposite , namely @xmath62 and @xmath63 .
the condition @xmath64 is a corollary of the theory of cartan - weyl bases ( see @xcite ) ; the condition @xmath65 follows from proposition .
let @xmath66 and @xmath67 denote the projection operators corresponding to the second decomposition , so that for any @xmath68 and for any @xmath69 [ pdef ] p^+(f_-f_+)=(f_-)f_+ , p^-(f_-f_+)=f_-(f_+ ) .
projection operators related to the first decomposition are denoted by @xmath70 and @xmath71 , in such a way that for any @xmath72 and for any @xmath73 [ pdefa ] p^*+(e_+e_-)=e_+(e_- ) , p^*-(e_+e_-)=(e_+)e_- .
algebra @xmath0 admits a natural completion @xmath74 , which can be described as the minimal extension of @xmath0 , that acts on every left highest weight and on every right lowest weight @xmath0-module with respect to the standard borel subalgebras(see @xcite ) . in a highest weight representation of @xmath0 , any matrix coefficient of an arbitrary product of currents
@xmath75 is a laurent polynomial in @xmath76 over the ring of formal power series in variables @xmath77 .
it converges to a rational function in the domain @xmath78 .
the latter observation and the commutation relation , which dictate the way of taking an analytic continuation from the domain above , allow us to regard products of currents as operator - valued meromorphic functions with values in @xmath74 .
the projection operators @xmath79 , defined in the previous subsection , extend to the completion @xmath80 .
the projection operators @xmath81 extend to another completion @xmath82 , where @xmath83 denotes the minimal extension of @xmath0 that acts in every left lowest weight and in every right highest weight @xmath0-module .
let @xmath84 and @xmath85 denote the following formal series of elements of the tensor product @xmath86 : [ pairing ] & = & = 1+(q - q^-1)e(z)f(z ) + + & & + e(z_1)e(z_2)f(z_1)f(z_2)+ , + & = & ( _ n>0 a_-n a_n ) .
by @xmath87 we mean a formal integral that is equal to the coefficient of @xmath88 in the formal series @xmath89 .
we further set [ rfactors ] _
+ = ( p^*-p^+ ) ( ) , _
-=(p^*+p^- ) ( ) , = _
+ ^21q^hh^21_- , where @xmath90 . [ proposition0 ]
@xmath64 the tensor @xmath91 is a cocycle for @xmath17 such that for any @xmath92 @xmath93 @xmath65 the tensor @xmath94 is the universal @xmath5-matrix for @xmath0 with opposite comultiplication @xmath95 : @xmath96 for any element @xmath97 , let @xmath98 denote the same element presented as a sum @xmath99 , where @xmath100 , @xmath101 , @xmath102 , @xmath103 . in the work @xcite
, it was proved that the element @xmath104 is a well defined series in elements of the tensor product @xmath105 , which represents the tensor of the hopf and @xmath106 for any @xmath107 and @xmath108 . ]
pairing @xmath109 , that is @xmath110 for any @xmath111 and @xmath112 .. however , the arguments therein are general and work for any quantum affine algebra . ] due to proposition and to the theory of cartan - weyl bases for @xmath0 developed in @xcite ,- commutators with @xmath113-commutators in the construction of the cartan - weyl basis . ]
the pair @xmath114 forms an orthogonal decomposition of the bialgebra @xmath38 .
analogously , the dual pair @xmath115 forms an orthogonal decomposition of the bialgebra @xmath36 .
this implies the decomposition @xmath116 , where @xmath117 is the tensor of the pairing of @xmath51 and @xmath50 , @xmath118 is the tensor of the pairing of @xmath52 and @xmath49 .
they are given by the relations : @xmath119 the last equalities in both lines of are implied by the following observation : the application of projection operators to both tensor components automatically performs the normal ordering of the tensor @xmath84 .
the factorization of the tensor @xmath85 follows from the definitions of the projections . now , part ( i ) of the proposition
becomes a direct generalization of proposition 3.8 of @xcite .
one can check that the conditions ( h1)(h6 ) of section 2.3 of @xcite are satisfied for the algebras @xmath120 , @xmath121 , @xmath122 , @xmath123 due to the theory of cartan - weyl bases , proposition [ proposition0 ] and the relations . under the same settings part ( ii )
is the direct generalization of the results of ( * ? ? ?
* section 2.3 ) .
note that all the results of this and of previous sections could be easily modified for the central extended algebra @xmath0 with an added grading element .
see e.g. @xcite .
let @xmath124 be a representation of the algebra @xmath0 and @xmath125 be a vector in @xmath124 .
we call @xmath125 a highest weight vector with respect to the current borel subalgebra @xmath38 if e(z)v & = & 0 , + k^(z)v & = & ( z)v , where @xmath126 is a meromorphic function , decomposed in series in @xmath88 for @xmath127 and in series in @xmath128 for @xmath129 .
representation @xmath124 is called a representation with highest weight vector @xmath130 with respect to @xmath38 if it is generated by @xmath125 over @xmath0 .
suppose that for every finite ordered set @xmath131 , an element @xmath132 is chosen .
it is a laurent polynomial in @xmath133 over the ring @xmath134 satisfying the following conditions : 1 . for any representation @xmath124 that is a highest weight representation with respect to @xmath38 with highest weight vector @xmath125 the function w_v(z_i_1, ,z_i_n)=w(z_i_1,
,z_i_n)v converges in the domain @xmath135 to a meromorphic @xmath124-valued function ; 2 .
@xmath136 if @xmath137 ; 3 .
let @xmath138 be a tensor product of the highest weight representations with highest weight vectors @xmath139 and @xmath140 and the highest weight series @xmath141 and @xmath142 ; then for an ordered set @xmath143 we have @xmath144 a collection @xmath132 for all @xmath145 is called the universal weight function . using proposition [ proposition0 ] ,
we get the following direct generalization of ( * ? ? ?
* theorem 3 ) : [ proposition000 ] the collection @xmath146 is a weight function for @xmath0 : w(z_i_1, ,z_i_n ) = p^+ .
our main goal from now on is to calculate the latter projection .
in the following sections we use the abbreviated notation @xmath147 for the projection @xmath66 . we introduce a pair of new currents @xmath148 and @xmath149 the first one plays a crucial role in the following discussion since the main result is formulated in terms of the projection @xmath150 thus @xmath148 and @xmath151 are a pair of laurent series with coefficients in the completed algebra @xmath152 given by [ s ] s(z ) & = & _ w =- q^-1 zf(z)f(w)w , + t(z ) & = & _ w = q^2 zf(z)f(w)w .
we call generating functions @xmath148 , @xmath151 the _ composite currents_. we treat them as meromorphic operator - valued functions in the category of left highest weight @xmath0-modules . in @xmath152 the composite currents @xmath148 and @xmath151 can be represented as the following products : [ s ] s(z ) & = & f(-q^-1z)f(z ) , + [ t ] t(z ) & = & f(q^2z)f(z ) . @xmath153 the second equality is proved in a similar way
. we use the following vanishing conditions on products of drinfeld currents : _ diagonal conditions _
( i ) follow from relation and _ serre conditions _
( ii ) follow from analytic serre relations ( see @xcite ) : * the product @xmath154 has a simple zero on the hyperplane @xmath155 ; * the product @xmath156 has a simple zero on the following lines : @xmath157 , @xmath158 and @xmath159 .
we collect data derived from the vanishing conditions in the following table : [ cols="^,^,^,^ " , ] in this case the first four zeros are derived from diagonal conditions and the rest from serre relations . thus the product @xmath160 has simple zeros at the points @xmath161 and @xmath162 and simple poles at the points @xmath163 and @xmath164 .
once again we obtain the following equality of holomorphic functions : [ ss ] s(z)s(w ) = s(w)s(z ) .
in this section we introduce some notation that is frequently used below .
the following rational functions are formal power series , converging in the domain @xmath165 .
@xmath166 let @xmath167 and @xmath168 denote the following combinations of projections : [ fc ] ( z_1, ,z_n-1;z_n)=p(f(z_n))-_k(z_1,
,z_n-1;z_n)p(f(z_k ) ) , [ sc ] ( z_1, ,z_n-1;z_n)=p(s(z_n ) ) & - _ k(z_1,
,z_n-1;z_n)p(s(z_k ) ) - + & - _ k(z_1,
,z_n-1;z_n)p(s(-qz_k ) ) .
let @xmath169 and @xmath170 be ordered subsets of an ordered set @xmath171 .
we call an ordered pair of subsets @xmath172 a _
@xmath66-admissible pair _ of cardinality @xmath173 if the following hold : * @xmath174 ; * @xmath175 ; * @xmath176 , for @xmath177 . for each @xmath66-admissible pair @xmath172 of cardinality @xmath173 , and for any @xmath178
, we introduce a formal power series @xmath179 that is the decomposition in the domain @xmath78 of the rational function [ tau - fin ] _ i , j^k = -_i_k , where @xmath180 are skipped in the row @xmath181 of variables of the function @xmath182 above . in other words ,
each of @xmath180 occur in the variables of function @xmath182 , but only once .
now , for each @xmath66-admissible pair @xmath172 of cardinality @xmath173 and for any @xmath183 such that @xmath184 , we introduce a generating function @xmath185 .
namely , for such a @xmath26 there exists a unique @xmath186 for which @xmath187 , with @xmath188 , @xmath189 .
then we set [ fc - fin ] _
i , j^k(z_1,
,z_n)= , where again @xmath190 are skipped in the row @xmath191 . finally , we define a q - commutator [ qcomm ] _
q^1 = ab - q^1ba and a pair of ordered products @xmath192 [ th1 ] projections of currents @xmath89 and @xmath148 can be written as follows : p(f(z ) ) & = & _ n>0f_nz^-n , + p(s(z ) ) & = & -1q+q^-2 .
[ th - main ] the projection @xmath193 can be expressed by the following explicit formula : @xmath194 where @xmath168 , @xmath179 , @xmath195 are defined by , , respectively . the second sum in the formula
is taken over @xmath66-admissible pairs @xmath172 of cardinality @xmath173 , and in the case @xmath196 we have only one summand @xmath197 we illustrate theorem by giving examples in section . in this section we obtain some formulae for projections @xmath198 .
the proofs almost literally reproduce the ones of theorem , and we skip them . let us introduce a composite current @xmath199 as a matter of fact @xmath200 , but @xmath201 is more convenient for calculating projection @xmath67 .
the following theorem holds : projections @xmath202 and @xmath203 can be written as follows : p^-(f(z ) ) & = & f_nz^-n , + p^-(s(z ) ) & = & 11+q^3 .
now we introduce some more notation : @xmath204 define the following combinations of projections : [ fc- ] ( z_1;z_2, ,z_n)= p^-(f(z_1))-_k(z_1;z_2, ,z_n)p^-(f(z_k ) ) , [ sc- ] ( z_1;z_2, ,z_n)= p^-(s(z_1 ) ) & - _ k(z_1;z_2,
,z_n)p^-(s(z_k ) ) - + & - _ k(z_1;z_2,
,z_n)p^-(s(-q^-1z_k ) ) , now , let @xmath169 and @xmath170 be ordered subsets of an ordered set @xmath171 .
we call an ordered pair of subsets @xmath172 a _
@xmath67-admissible pair _ of cardinality @xmath173 if the following hold : * @xmath174 ; * @xmath205 ; * @xmath206 , for @xmath177 . for each @xmath67-admissible pair @xmath172 of cardinality @xmath173 , and for any @xmath178
, we introduce a formal power series @xmath207 that is the decomposition in the domain @xmath78 of the rational function [ tau - fin- ] _ i , j^k = _ j_k , where @xmath208 are skipped in the row @xmath209 of variables of the function @xmath210 above . in other words
each of @xmath208 occur in the variables of function @xmath210 , but only once .
now , for each @xmath67-admissible pair @xmath172 of cardinality @xmath173 and for any @xmath183 such that @xmath211 , we introduce a generating function @xmath212 .
namely , for such a @xmath26 there exists a unique @xmath186 for which @xmath213 , with @xmath214 , @xmath215 .
then we set [ fc - fin- ] _ i , j^k(z_1,
,z_n)= , where again @xmath216 are skipped in the row @xmath217
. then we have the following theorem : [ theor- ] the projection @xmath218 can be expressed by the following explicit formula : @xmath219 where @xmath220 , @xmath207 , @xmath221 are defined by , , respectively . the second sum in the formula
is taken over @xmath67-admissible pairs @xmath172 , and in the case @xmath196 we have only one summand @xmath222 to obtain the formulae for projections @xmath70 and @xmath71 we introduce an involution @xmath223 of the algebra @xmath0 such that ( e_n ) = f_-n , ( f_n ) = e_-n , ( a_n ) = a_-n , ( k_0 ) = k_0 ^ -1 .
one can see that [ pstar ] p^= p^*. using this identity we can compute @xmath224 as @xmath225 .
an integral presentation for the factors of the universal @xmath226-matrix of @xmath0 can be obtained by applying formulae , and to relations .
the statement is multiplicative with respect to @xmath22 ; hence it is sufficient to prove it for generators @xmath227 , @xmath228 , @xmath229 of the subalgebra @xmath44 . for elements @xmath227 and @xmath229
the property follows directly from the isomorphism between the realizations and the formulae of the comultiplication @xmath17 . thus we only need to check the property for @xmath228 .
let @xmath230 be a grading on the commutative algebra , generated by the elements @xmath231 and @xmath232 , such that @xmath233 and @xmath234 .
using drinfeld comultiplication formulae we obtain @xmath235 where @xmath236 $ ] are polynomials in variables @xmath237 of degree @xmath145 over the quotient field @xmath238 . according to
it only remains to check the statement of the lemma for @xmath239 the equality ^(d)(qf_1f_0-f_0f_1 ) = 1(qf_1f_0-f_0f_1 ) + f_-i+1(qj_nf_0-f_0j_n ) holds modulo @xmath240 .
formulae imply that @xmath241 to prove the same for the second summand we use equality @xmath242 which one can obtain from relation .
the first part of the theorem immediately follows from the definition of the projection @xmath66 .
the proof of the second part is a little bit more complicated .
we start with an equality @xmath243 using the cauchy residue theorem we rewrite the latter residue as the following sum : @xmath244 it remains to calculate the residues at @xmath245 and @xmath246 . @xmath247
since @xmath154 is well defined only in the domain @xmath248 we evaluate the residue of its analytic continuation at @xmath246 : _
w= = _ w= , which after putting @xmath249 equals @xmath250 previous relations imply @xmath251 where @xmath252 is defined by .
recalling projection properties , @xmath253 to finish the proof we rewrite relation in coordinates : [ ffcomp ] q^2f_n+2f_m - qf_mf_n+2+q^2f_m+2f_n - qf_nf_m+2 = ( 1-q^3)(f_m+1f_n+1+f_n+1f_m+1 ) .
using relation , the projection properties and the borel subalgebras description , given in the section , we obtain that @xmath254 for @xmath255 the relation above is obvious , for @xmath256 , @xmath257 it follows from the formulae , and finally for @xmath258 , @xmath259 equality provides us with the desired relation
. therefore we obtain p(s(z))=-1q+q^-2 .
assume @xmath260 .
we introduce the formal power series @xmath261 , which is the decomposition in the domain @xmath78 of the rational function [ tau ] _ k , p(z_1,
,z_n-1;z_n ) = -_k(z_1, ,z_n-1;z_n ) .
now , we use iterative application of : [ string ] the projection @xmath262 can be written as @xmath263 where @xmath167 is defined by .
we reduce evaluation of @xmath264 to the problem of calculating @xmath265 or @xmath266 . applying lemma to @xmath266
once more we reduce the case to the evaluation of @xmath265 .
the latter is achieved by using : [ strings ] the projection @xmath267 admits a decomposition p = p(z_1, ,z_n-1;z_n ) , where @xmath168 is defined by .
one can see that on every step of the iterative application of lemma , we either factor projection @xmath262 into the product @xmath268 @xmath269 or replace product @xmath270 with the current @xmath271 .
thus , sets @xmath143 and @xmath272 in the formula indicate that on the @xmath273-th step , we replaced @xmath274 with @xmath275 , on the @xmath276-th step we replaced @xmath277 with @xmath278 etc . for any current @xmath279 let @xmath280 denote currents @xmath281 one can see , that @xmath282 let us introduce notation that is used in the proof below .
let @xmath283 be a square matrix of order @xmath284 , and @xmath285 , @xmath286 and @xmath287 be @xmath288-dimensional vectors as follows : @xmath289 [ rv ] _
c = 11-c^-1z_n / z_1, ,11-c^-1z_n / z_n-1 , [ rm ] _
c = 11-c^-1z_1/z_1 & & & & 11-c^-1z_1/z_n-1 + & & & & + 11-c^-1z_n-1/z_1 & & & & 11-c^-1z_n-1/z_n-1 , with @xmath290 .
next , we introduce the following rational functions of @xmath22 : @xmath291 let also [ res - abg ] _
c(w / z ) = _ u = cz , _ c(w / z ) = _ u = cz ,
_ c(w / z ) = _ u = cz , for @xmath290 .
one can see that @xmath292 , @xmath293 , @xmath294 , for some constants @xmath295 , @xmath296 , @xmath297 .
now , by the definition of projection @xmath147 we have @xmath298 . using projection properties we obtain @xmath299 to calculate the second summand on the right hand side we need to move current @xmath300 in front of the product @xmath301 .
this can be done by inductive application of : [ cp ] the following commutation relations hold : * let @xmath302 and @xmath303 be as in and ; then @xmath304 * let @xmath305 and @xmath306 be as in and ; then @xmath307 informally , the proposition above states that on pushing @xmath300 to the left in the product @xmath308 some rational functions of @xmath309 arise , which after decomposition into partial fractions appear to have only simple poles on the hyperplanes @xmath310 and @xmath311 . thus the following formula holds : @xmath312 where @xmath313 and @xmath314 are @xmath288-dimensional vectors with coordinates @xmath315 and @xmath287 is given by .
notice that @xmath316 and @xmath317 do not depend on @xmath309 .
since we have @xmath284 vanishing conditions , @xmath318 has simple zeros on the hyperplanes @xmath319 for @xmath320 ; we can compose a system of linear equations @xmath321 with @xmath285 and @xmath322 defined by , . expressing @xmath313 from the system above and substituting it into equation
, we get @xmath323 [ vm ] we have the following equalities : * let @xmath167 be as in ; then p(f(z_n ) ) - _
q^2_q^2 ^ -1^= ( z_1, ,z_n-1;z_n ) ; * let @xmath324 be as in ; then ^= _ i(z_1, ,z_n-1;z_n)y_i .
applying proposition we arrive at the equation @xmath325 now let us notice that @xmath326 for @xmath327 , since the product @xmath328 does not have poles on the hyperplanes @xmath311 for @xmath327 . taking into account that the action of projection @xmath329 is well defined on every highest weight @xmath0-module for any arbitrary @xmath128 and @xmath330 , we obtain the following commutation relations : @xmath331 with @xmath302 , @xmath305 , @xmath332 defined by , and respectively .
notice , that in the relations above one should take an analytic continuation of the left hand or the right hand side when necessary .
using relation , we rewrite @xmath333 as follows : @xmath334 where @xmath335 , and @xmath302 is defined by .
finally , using relation and mutual commutativity of the projection @xmath147 and of the operator of taking a residue , we obtain @xmath336 hence the statement of the lemma holds .
we start with : [ decomp ] current @xmath148 admits the following expansion : [ sdec ] s(z)=p^+(s(z))+^+-s^-(z ) .
hence the projection @xmath267 can be evaluated via pushing @xmath337 and @xmath338 in front of the product @xmath339 .
we use proposition and : [ cp2 ] the following commutation relation holds : @xmath340 where @xmath332 and @xmath341 are given by and
. then we arrive at the following equation : [ sss ] p = pp(s(z_n ) ) - _
-q^-1 ^ - _ -q^3^ , where @xmath313 , @xmath314 and @xmath342 are @xmath288-dimensional vectors with coordinates @xmath343 and @xmath344 is defined by .
notice that @xmath345 for @xmath346 since the product @xmath339 does not have poles on the hyperplanes @xmath311 , while @xmath316 and @xmath347 are functions independent of @xmath309 .
regarding @xmath348 vanishing conditions , namely @xmath349 on the hyperplanes @xmath350 and @xmath351 , we can compose a system of @xmath348 linear equations with @xmath348 variables @xmath316 and @xmath347 for @xmath346 .
let @xmath352 where @xmath283 , @xmath287 and @xmath286 are defined by , and respectively .
thus we can present our system of equations as @xmath353 solving the system and substituting @xmath313 and @xmath342 into the equation we get @xmath354 now we finish the proof by applying : [ vm2 ] we have the following equality : p(s(z_n ) ) - ^-1^=(z_1,
,z_n-1;z_n ) , where @xmath168 is given by . since the proofs are almost the same we present only the proof of proposition part @xmath355 .
let us recall relation : ( q^2z - u)(z+qu)f(z)f(u)=(z - q^2u)(qz+u)f(u)f(z ) . since product @xmath356 has simple poles on the planes @xmath357 and @xmath358 , we derive the following equality of two holomorphic functions f(z)f(u ) + 11-q^-2 u / zt(z ) +
11+q u / zs(z ) = + = ( z / u)f(u)f(z ) - 11-q^2 z / ut(z ) + 11+q^-1 z / us(z ) . as a matter of fact ,
the coefficients of @xmath151 and @xmath148 on the left hand side and the right hand side are the same rational functions , but they are represented as power series in @xmath359 and @xmath360 respectively for the reasons mentioned in section . hence we get @xmath361 under the action of @xmath362 the previous equality turns into @xmath363 since the integral above can be rewritten as @xmath364 using relations , , we derive the statement of the proposition . to prove part @xmath355 of the proposition we only need to find out that @xmath365 is an @xmath288-dimensional vector @xmath366 with @xmath367 for @xmath368 .
regarding @xmath369 as a rational function over @xmath309 , we observe that it has simple poles only on the planes @xmath370 for @xmath346 .
we also know that @xmath369 is a rational function of degree @xmath371 in @xmath309 .
let us notice that @xmath344 is equal to the @xmath372-th row of matrix @xmath322 if @xmath350 .
therefore @xmath369 equals @xmath57 on the hyperplane @xmath373 and equals @xmath374 on the hyperplanes @xmath350 for @xmath375 .
these conditions totally determine @xmath369 .
now , replacing @xmath3 by @xmath376 we derive statement @xmath355 of the proposition . to prove the statement of part @xmath377 we need to show that [ la1 ]
_ k(z_n;z_1,
,z_n-1)=11+qz_n / z_k-_q^2_q^2 ^ -1_-q^-1^(k ) , where @xmath378 stands for the @xmath26-th column of matrix @xmath379 . regarded as a rational function of @xmath309
, @xmath324 has simple poles only on the hyperplanes @xmath310 for @xmath346 and @xmath380 .
since vector @xmath344 is equal to the @xmath372-th row of the matrix @xmath322 for @xmath350 , @xmath381 has simple zeros on the hyperplanes @xmath382 .
considering the fact that @xmath381 is a rational function of degree @xmath371 in @xmath383 we can present it in the following form : _
,z_n-1;z_n ) = 1qz_n+z_i c(z_1,
equality implies that _ z_n - q^-1z_k((1+qz_n / z_k)_k(z_1, ,z_n-1;z_n))=1 .
we finish the proof by obtaining @xmath384 from the latter condition .
the proof of proposition is an easy modification of the proof of proposition part @xmath355 .
the statement of the proposition follows from equality , projection properties and the statement _ b ) _ of theorem .
in this section we work out particular cases of the formula for @xmath385 .
recall that all rational functions are considered as formal power series converging in the region @xmath165 .
first of all let us recall that @xmath386 @xmath387 now , for @xmath388 we have @xmath389 since @xmath390 and @xmath391 , we get @xmath392 here @xmath393 for @xmath394 , formula turns into @xmath395 here again ( z_i ) = p(s(z_i ) ) , and @xmath391 .
therefore we obtain @xmath396 one can also see that @xmath397 and @xmath398 the following expressions are obtained from corresponding rational functions by expansion in power series in the domain @xmath399 : @xmath400 finally , we obtain formula in the case @xmath401 : @xmath402 where the sum is taken over all permutations @xmath403 of an ordered set @xmath404 such that @xmath405 and @xmath406 . in the latter example
@xmath407 for all @xmath408 , and @xmath409 where @xmath381 and @xmath302 are given by and respectively .
the authors thank stanislav pakuliak for fruitful discussions .
the first author was supported by interdisciplinary rfbr grant 09 - 01 - 12185-ofi - m , joint grant 09 - 02 - 90493-ukr - f , and grant for support of scientific schools 3036.2008.2-nsh .
the second author was supported by rfbr grant 08 - 01 - 00667 , joint cnrs - rfbr grant 09 - 01 - 93106-ncnil , and by federal agency for science and innovations of russian federation under contract 14.740.11.0081 . | in this article , we give an explicit formula for the universal weight function of the quantum twisted affine algebra @xmath0 .
the calculations use the technique of projecting products of drinfeld currents onto the intersection of borel subalgebras of different types . * weight function for the quantum + affine algebra @xmath0 * * s. khoroshkin@xmath1 , a. shapiro@xmath2 * + |
the recent experimental realization of synthetic spin - orbit ( so ) coupling in ultracold quantum gases @xcite is considered as an important breakthrough , as it provides new possibilities for ultracold quantum gases to be used as quantum simulation platforms , and paves a new route towards exploring novel states of matter and quantum phenomena @xcite .
it has been found that the so coupling can not only stabilize various topological defects , such as half - quantum vortex , skyrmion , composite soliton and chiral domain wall , contributing to the design and exploration of new functional materials @xcite , but also lead to entirely new quantum phases , such as magnetized phase and stripe phase @xcite , providing support for the study of novel quantum dynamical phase transitions @xcite and exotic supersolid phases @xcite .
all the intriguing features mentioned above are based on the characteristics that the so coupling ( either of the nist @xcite , rashba @xcite or weyl @xcite types ) makes the internal states coupled to their momenta via the su(2 ) pauli matrices .
however , if the ( pseudo)spin degree of freedom involves more than two states , the su(2 ) spin matrices can not describe completely all the couplings among the internal states .
for example , a direct transition between the states @xmath0 and @xmath1 is missing in a three - component system @xcite . from this sense , an su(3 ) so coupling with the spin operator spanned by the gell - mann matrices is more effective in describing the internal couplings among three - component atoms @xcite .
the su(3 ) so coupled system has no analogue in ordinary condensed matter systems , hence may lead to new quantum phases and topological defects . in this article
, we show that a new type of topological defects , double - quantum spin vortices , can exist in the ground states of su(3 ) so coupled bose - einstein condensates ( becs ) .
it is found that the su(3 ) so coupling leads to two distinct ground - state phases , a magnetized phase or a lattice phase , depending on the spin - exchange interaction being ferromagnetic or antiferromagnetic . in the magnetized phase , the su(3 ) so coupling leads to a ground state with three - fold degeneracy , in stark contrast to the su(2 ) case where the degeneracy is two , thus may offer new insights into quantum dynamical phase transitions @xcite . in the lattice phase , the su(3 )
so coupling breaks the ordinary phase requirement @xmath2 for ordinary spinor becs , where @xmath3 is the winding number of the @xmath4-th spin component @xcite , and induces three types of exotic vortices with cores filled by different magnetizations .
the interlaced arrangement of these vortices leads to the spontaneous formation of multiply quantized spin vortices with winding number @xmath5 .
this new type of topological defects can be observed in experiments using magnetization - sensitive phase - contrast imaging technique .
we consider the @xmath6 spinor becs with su(3 ) so coupling . using the mean - field approximation ,
the hamiltonian can be written in the gross - pitaevskii form as @xmath7,\label{hamiltonian}\end{aligned}\ ] ] where the order parameter @xmath8^\top$ ] is normalized with the total particle number @xmath9 .
the particle density is @xmath10 , and the spin density vector @xmath11 is defined by @xmath12 with @xmath13 being the vector of the spin-@xmath14 matrices given in the irreducible representation @xcite .
the so coupling term is chosen as @xmath15 , where @xmath16 is the spin - orbit coupling strength , @xmath17 represents 2d momentum , and @xmath18 is expressed in terms of @xmath19 and @xmath20 , with @xmath21 being the gell - mann matrices , i.e. , the generators of the su(3 ) group @xcite . note that the su(3 ) so coupling term in the hamiltonian involves all the pairwise couplings between the three states
this is distinct from the previously discussed su(2 ) so coupling in spinor becs , where the states @xmath22 and @xmath23 are coupled indirectly @xcite . the parameters @xmath24 and @xmath25 describe the strengths of density - density and spin - exchange interactions , respectively .
the hamiltonian with su(3 ) so coupling can be realized using a similar method of raman dressing as in the su(2 ) case @xcite . as shown in fig .
[ fig1](a ) , three laser beams with different polarizations and frequencies , intersecting at an angle of @xmath26 , are used for the raman coupling .
each of the three raman lasers dresses one hyperfine spin state from the @xmath6 manifold ( @xmath27 , @xmath28 and @xmath29 ) to the excited state @xmath30 [ see fig . [ fig1](b ) ] . when the standard rotating wave approximation is used and the excited state is adiabatically eliminated due to far detuning
, one can obtain the effective hamiltonian in eq .
( [ hamiltonian ] ) , as discussed in appendix [ app : deri ] . by diagonalizing the kinetic energy and so coupling terms
, we can obtain the single - particle energy spectrum , which can provide useful information about the ground state of bose condensates .
for the su(2 ) case , it is known that the single - particle spectrum with the nist type so coupling acquires either a single or two minima , depending on the strength of the raman coupling @xcite , while for the case of rashba type there exist an infinite number of minima locating on a continuous ring in momentum space @xcite . for the su(3 ) so coupling discussed here , we find that there are in general three discrete minima residing on the vertices of an equilateral triangle [ see figs . [ fig1](c)-[fig1](d ) ] .
this unique property of the energy band implies the possibility of a three - fold degenerate many - body magnetized state @xcite or a topologically nontrivial lattice state , depending on the choices among the three minima made by the many - body interactions .
next , we discuss the phase diagram of the many - body ground states . for the case of su(2 ) so coupling , it is shown that two many - body ground states , magnetized state and stripe state , can be stabilized in a homogeneous system @xcite .
although the rashba so coupling provides infinite degenerate minima in the single - particle spectrum , a many - body ground state condensed in one or two points in momentum space is always energetically favorable due to the presence of spin - exchange interaction @xcite . as a result , a lattice state with the condensates occupying three or more momentum points for su(2 ) so coupling is unstable , unless a strong harmonic trap is introduced @xcite . for the present case of su(3 ) so coupling , we first analytically calculate the possible ground states using a variational approach with a trial wave function @xmath31 , where [ groundstate ] @xmath32 correspond to the many - body states with all particles condensing on one of the three minima of the single - particle spectrum , and @xmath33 are expansion coefficients .
substituting eqs .
( [ groundstate1])-([groundstate3 ] ) into the interaction energy functional @xmath34 one obtains @xmath35 where @xmath36 is the mean particle density . by minimizing the interaction energy with respect to the variation of @xmath37
, one finds that the spin - exchange interaction plays an important role in determining the phase diagram .
when @xmath38 , it favors @xmath39 , indicating that the ground state is a triangular lattice phase with an equally weighted superposition of the three single - particle minima . on the other hand , as @xmath40 , the system prefers a state with either @xmath41 , @xmath42 , or @xmath43 , @xmath44 , or @xmath45 , @xmath46 , indicating that the ground state occupies one single minimum in the momentum space , and corresponds to a three - fold degenerate magnetized phase .
note that the variational wave function eqs .
( [ groundstate1])-([groundstate3 ] ) is a good starting point as the so coupling is strong enough to dominate the chemical potential . for the case with weak so coupling , one must rely on numerical simulations to determine the many - body ground state . in such a situation , we find a stripe phase with two minima in momentum space occupied for @xmath47 , which will be discussed latter .
the many - body ground states can be numerically obtained by minimizing the energy functional associated with the hamiltonian eq .
( [ hamiltonian ] ) via the imaginary time evolution method .
it is found that the numerical results are consistent with the analytical analysis discussed above for rather weak interaction with @xmath48 .
figure [ fig2 ] illustrates the two possible ground states of spinor becs with su(3 ) so coupling . when @xmath49 , the three components are immiscible and arranged as an interlaced triangular lattice with the spatial translational symmetry spontaneously broken [ see figs .
[ fig2](a)-[fig2](d ) ] .
this lattice is topologically nontrivial and embedded by vortices and antivortices as shown in fig .
[ fig2](b ) . from this result
, we conclude that a lattice phase can be stabilized in a uniform su(3 ) so coupled bec , which is in clear contrast to the su(2 ) case where a strong harmonic trap is required @xcite .
more details on the structure of vortices as well as their unique spin configurations will be investigated later . on the other hand , as @xmath50 ,
the three components are miscible , and the system forms a magnetized phase with the spatial transitional symmetry preserved but the time - reversal symmetry broken [ see figs . [ fig2](e)-[fig2](f ) ] .
this magnetized phase occupies one of the three minima of the single - particle spectrum by spontaneous symmetry breaking , hence is three - fold degenerate instead of doubly degenerate in the su(2 ) case @xcite . between the numerical simulation and the variational calculation
are shown by solid ( lattice state ) and dashed ( stripe state ) lines .
( b)-(d ) the ground - state density , phase and momentum distributions of the stripe phase with the parameters @xmath51 and @xmath52.,scaledwidth=40.0% ] for strong antiferromagnetic spin interaction with @xmath47 , however , a stripe phase is identified with two of three minima occupied in the momentum space .
we take the states with two or three minima occupied in the momentum space as trial wave functions , and perform imaginary time evolution to find their respective optimized ground state energy . a typical set of results
are summarized in figure [ fig3](a ) , showing the energy comparison with different values of interatomic interactions .
obviously , one finds that the stripe phase will has lower energy than the lattice phase when the interatomic interaction exceeds a critical value .
due to the finite momentum in vertical direction of the stripe [ see fig . [ fig3](d ) ] , both the spatial translational and time - reversal symmetries are broken [ see figs .
[ fig3](b)-[fig3](c ) ] .
this is distinct from the stripe phase induced by su(2 ) so coupling , where the time - reversal symmetry is preserved @xcite .
the vortex configuration of spinor becs depends on the phase relation between the three components .
we next discuss the influence of so coupling on the phase requirement of the vortex configuration .
we first assume that the spinor order parameter of a vortex in the polar coordinate @xmath53 can be described as @xmath54 where @xmath55 and @xmath56 . in the absence of so coupling ,
the phase - dependent terms in the hamiltonian are @xmath57 where the first term results from the kinetic energy and the second from the spin - exchange interaction . substituting eq .
( [ spinor order parameter ] ) into ( [ phase - dependent terms ] ) , one obtains @xmath58d\theta . \label{eint}\end{aligned}\ ] ] it is easy to read from eq .
( [ ekin ] ) that the system favors small winding numbers energetically .
moreover , from eq .
( [ eint ] ) the energy minimization requires the winding number and phase satisfy the following relations [ phase requirement1 ] @xmath59 where @xmath60 is odd for @xmath49 and even for @xmath50 .
the phase requirement of eq .
( [ winding number1 ] ) indicates that the following types of winding combination , such as @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 and @xmath66 are allowed in a spinor bec , where the symbol
@xmath67 " denotes the absence of the @xmath68 component . for the case of su(2 ) so coupling , we take the rashba type as an example , and write the hamiltonian as @xmath69 where @xmath70^\top$ ] . substituting eq .
( [ spinor order parameter ] ) into ( [ rashba hamiltonian ] ) , one can obtain @xmath71 } \notag \\ & -(\phi_{1}r\partial_{r}\phi_{0}+\mathrm{w}_{0}\phi_{1}\phi_{0 } ) e^{-i[(\mathrm{w}_{1}-\mathrm{w}_{0}+1)\theta+(\alpha_{1}-\alpha_{0}-\frac{\pi}{2 } ) ] } \notag \\ & + ( \phi_{0}r\partial_{r}\phi_{-1}+\mathrm{w}_{-1}\phi_{0}\phi_{-1 } ) e^{i[(\mathrm{w}_{-1}-\mathrm{w}_{0}-1)\theta+(\alpha_{-1}-\alpha_{0}-\frac{\pi}{2 } ) ] } \notag \\ & -(\phi_{-1}r\partial_{r}\phi_{0}-\mathrm{w}_{0}\phi_{-1}\phi_{0 } ) e^{-i[(\mathrm{w}_{-1}-\mathrm{w}_{0}-1)\theta+(\alpha_{-1}-\alpha_{0}-\frac{\pi}{2})]}\big].\end{aligned}\ ] ] in order to minimize the so coupling energy , it is preferred that [ phase requirement2 ] @xmath72 then the so coupling energy is rewritten as @xmath73dr\cos{m\pi } \notag \\ & + 2\pi\!\!\int\![\phi_{0}r\partial_{r}\phi_{-1}-\phi_{-1}r\partial_{r}\phi_{0 } + ( \mathrm{w}_{-1}+\mathrm{w}_{0})\phi_{0}\phi_{-1}]dr\cos{n\pi},\label{esoc - su2}\end{aligned}\ ] ] where @xmath74 and @xmath60 are odd or even , which can be determined by minimizing the energy expressed in eq .
( [ esoc - su2 ] ) .
it is found that the su(2 ) so coupling does not violate the ordinary requirement on the winding combination in eq .
( [ winding number1 ] ) , but introduces further requirements in eqs .
( [ winding number2a])-([winding number2b ] ) . as a result , while @xmath75 , @xmath76 and @xmath77 are still allowed , some winding combinations such as @xmath64 , @xmath61 , @xmath62 , @xmath78 , @xmath79 and @xmath80 are forbidden .
obviously , one can see that the so coupling break the chiral symmetry , thus may lead to chiral spin textures . for the case of su(3 ) so coupling , the effective hamiltonian can be written as @xmath81 substituting eq .
( [ spinor order parameter ] ) into ( [ su3 soc hamiltonian ] ) , we get @xmath82 } \notag \\ & -(\phi_{1}r\partial_{r}\phi_{0}+\mathrm{w}_{0}\phi_{1}\phi_{0 } ) e^{-i[(\mathrm{w}_{1}-\mathrm{w}_{0}+1)\theta+(\alpha_{1}-\alpha_{0}-\frac{\pi}{2 } ) ] } \notag \\ & + ( \phi_{0}r\partial_{r}\phi_{-1}+\mathrm{w}_{-1}\phi_{0}\phi_{-1 } ) e^{i[(\mathrm{w}_{-1}-\mathrm{w}_{0}-1)\theta+(\alpha_{-1}-\alpha_{0}-\frac{\pi}{2 } ) ] } \notag \\ & -(\phi_{-1}r\partial_{r}\phi_{0}-\mathrm{w}_{0}\phi_{-1}\phi_{0 } ) e^{-i[(\mathrm{w}_{-1}-\mathrm{w}_{0}-1)\theta+(\alpha_{-1}-\alpha_{0}-\frac{\pi}{2 } ) ] } \notag \\ & + ( \phi_{-1}r\partial_{r}\phi_{1}+\mathrm{w}_{1}\phi_{-1}\phi_{1 } ) e^{i[(\mathrm{w}_{1}-\mathrm{w}_{-1}-1)\theta+(\alpha_{1}-\alpha_{-1}-\frac{\pi}{2 } ) ] } \notag \\ & -(\phi_{1}r\partial_{r}\phi_{-1}-\mathrm{w}_{-1}\phi_{1}\phi_{-1 } ) e^{-i[(\mathrm{w}_{1}-\mathrm{w}_{-1}-1)\theta+(\alpha_{1}-\alpha_{-1}-\frac{\pi}{2})]}\big].\end{aligned}\ ] ] by minimizing the so coupling energy , one obtains the following relations [ phase requirement3 ] @xmath83 then the so coupling energy can be rewritten as @xmath84dr\cos{m\pi } \notag \\ & + \!2\pi\!\!\int\![\phi_{0}r\partial_{r}\phi_{-1}-\phi_{-1}r\partial_{r}\phi_{0 } + ( \mathrm{w}_{-1}+\mathrm{w}_{0})\phi_{0}\phi_{-1}]dr\cos{n\pi } \notag \\ & + \!2\pi\!\!\int\![\phi_{-1}r\partial_{r}\phi_{1}-\phi_{1}r\partial_{r}\phi_{-1 } + ( \mathrm{w}_{1}+\mathrm{w}_{-1})\phi_{-1}\phi_{1}]dr\cos{l\pi},\label{esoc - su3}\end{aligned}\ ] ] where @xmath74 , @xmath60 and @xmath85 are odd or even , which can be determined from eq .
( [ esoc - su3 ] ) . however , the three winding requirements eqs .
( [ winding number3a])-([winding number3c ] ) can not be satisfied simultaneously .
thus the su(3 ) so coupling may choose two out of the three winding requirements for the following three cases : case : [ hanhan2 ] @xmath86 case : [ hanhan2 ] @xmath87 case : [ hanhan2 ] @xmath88 for case , the winding combination @xmath75 is allowed , while @xmath78 is not allowed , indicating the chiral symmetry is broken . for case and case , one can find that the su(3 ) so coupling breaks the ordinary requirement on the winding combination in eq .
( [ winding number1 ] ) , thus new winding combinations , such as @xmath89 and @xmath90 , are possible .
the vortex configurations of spinor becs can be classified according to the combination of winding numbers and the magnetization of vortex core @xcite .
for example , a mermin - ho vortex has winding combination @xmath91 with a ferromagnetic core , where the plus and minus signs represent different chirality of the vortices @xcite , and the expression of @xmath92 indicates that the components of @xmath93 , @xmath94 and @xmath95 in the wave function acquire winding numbers of @xmath96 , @xmath97 and @xmath98 , respectively . using this notation , a polar - core vortex has winding combination @xmath99 with an antiferromagnetic core , and a half - quantum vortex has winding combination @xmath100 with a ferromagnetic core , where the symbol
@xmath67 " denotes the absence of the @xmath68 component . in the lattice phase induced by the su(3 )
so coupling with antiferromagnetic spin interaction , there exists three types of vortices : one is a polar - core vortex with winding combination @xmath75 , and the other two are ferromagnetic - core vortices with winding combinations @xmath101 and @xmath102 [ see fig . [
fig4](a ) ] .
however , the vortex configurations with opposite chirality of each type , such as @xmath78 , @xmath103 and @xmath104 , are not allowed , because the chiral symmetry is intrincically broken in su(3 ) so coupled systems , as discussed in sec . .
surprisingly , one finds that the two types of ferromagnetic - core vortices @xmath90 and @xmath102 violate the conventional phase requirement @xmath2 for ordinary spinor becs @xcite . this can be understood by noting that the relative phase among different wave function components are no longer uniquely determined by the spin - exchange interaction but also affected by the su(3 ) so coupling , as qualitatively explained in sec . .
thus , the interlaced arrangement of the three types of vortices forms a new class of vortex lattice which has no analogue in systems without so coupling .
the configurations of the three types of vortices induced by the su(3 ) so coupling with antiferromagnetic interaction are essentially different from those usually observed in ferromagnetic spinor becs , as can be illustrated by the spherical - harmonic representation @xcite . from figs .
[ fig4](b)-[fig4](d ) one can find that for the polar - core vortex , the antiferromagnetic order parameter varies continuously everywhere , while for the ferromagnetic - core vortex , the magnetic order parameter acquires a singularity at the vortex core .
in contrast , in the ordinary ferromagnetic spinor becs , the ferromagnetic order parameter varies continuously everywhere for the ferromagnetic - core vortex , but has a singularity at the core for the polar - core vortex @xcite .
spin vortex is a complex topological defect resulting from symmetry breaking , and is characterized by zero net mass current and quantized spin current around an unmagnetized core @xcite .
it is not only different from the magnetic vortex found in magnetic thin films @xcite , but also from the 2d skyrmion @xcite due to the existence of singularity in the spin textures @xcite .
single - quantum spin vortex with the spin current showing one quantum of circulation has been experimentally observed in ferromagnetic spinor becs @xcite .
multi - quantum spin vortices with @xmath85 ( @xmath105 ) quanta circulating spin current , however , are considered to be topologically unstable and have not been discovered yet @xcite .
a particularly important finding of our present work is that the polar - core vortex in the lattice phase has a spin current with two quanta of circulation around the unmagnetized core , hence can be identified as a double - quantum spin vortex .
figure [ fig5 ] presents the transverse magnetization @xmath106 , longitudinal magnetization @xmath107 , and amplitude of the total magnetization @xmath108 in the lattice phase , which are experimentally observable by magnetization - sensitive phase - contrast imaging technique @xcite . from these results
, one can find two distinct types of topological defects , double - quantum spin vortex ( dsv ) and half skyrmion ( hs ) @xcite , which correspond to the polar - core vortex with winding combinations @xmath109 and the ferromagnetic - core vortex with winding combinations @xmath101 or @xmath102 , respectively .
in particular , for the double - quantum spin vortex , the core is unmagnetized and the orientation of the magnetization along a closed path surrounding the core acquires a rotation of @xmath110 .
this finding indicates that a regular lattice of multi - quantum spin vortices can emerge spontaneously in antiferromagnetic spinor becs with su(3 ) so coupling . by exploring the effect of a small but finite temperature , we confirm that the double - quantum spin vortices are robust against thermal fluctuations and hence are observable in experiments , as discussed in appendix [ app : stab ] .
the emergence of spin current with two quanta of circulation can be analytically understood by expanding the wave function obtained by the variational methods around the center of a double - quantum spin vortex .
we suppose that the wave function of the lattice phase is written as @xmath111 then one can expand @xmath112 around the center of a vortex with winding number @xmath75 , e.g. , at the location of @xmath113 . substituting @xmath114 and @xmath115 into @xmath112 and expanding with respect to the infinitesimal @xmath116
, we obtain @xmath117 notice that the second - order terms with @xmath118 have no essential influence on the phases , thus the winding number for each component can still be represented as @xmath109 [ see figs .
[ fig6](a)-[fig6](c ) ] .
however , since the first - order terms are canceled out when calculating the transverse magnetization @xmath119 $ ] , the second - order terms play a dominant role , leading to the emergence of spin current with two quanta of circulation around an unmagnetized core @xmath120 as illustrated in fig . [
fig6](d ) . ) , displaying the winding combination @xmath109 .
( d ) direction of the transverse magnetization , indicating the emergence of spin current with two quanta of circulation.,scaledwidth=40.0% ]
to summarize , we have mapped out the ground - state phase diagram of su(3 ) spin - orbit coupled bose - einstein condensates .
several novel phases are discovered including a three - fold degenerate magnetized phase , a vortex lattice phase , as well as a stripe phase with time - reversal symmetry broken .
we also investigate the influence of su(3 ) spin - orbit coupling on the phase requirement of the vortex configuration , and demonstrate that the su(3 ) spin - orbit coupling breaks the ordinary phase rule of spinor bose - einstein condensates , and allows the spontaneous emergence of stable double - quantum spin vortices . as a new member in the family of topological defects , double quantum spin vortex has never been discovered in any other systems .
our work deepen the understanding of spin - orbit phenomena , and will attract extensive interest of scientists in the cold atom community .
this work was supported by nkrdp under grants nos .
2016yfa0301500 , 2012cb821305 ; nsfc under grants nos . 61227902 , 61378017 , 11274009 , 11434011 , 11434015 , 11447178 , 11522436 , 11547126 ; nkbrsfc under grant no .
2012cb821305 ; sklqoqod under grant no .
kf201403 ; sprpcas under grant nos .
xdb01020300 , xdb21030300 ; jsps kakenhi grant no .
26400414 and mext kakenhi grant no .
. w. h. and x .- f .
z. contributed equally to this work .
we consider spinor bose - einstein condensates ( becs ) illuminated by three raman laser beams , which couple two of the three hyperfine spin components respectively , as illustrated in figs .
1(a)-1(b ) of the main text .
the internal dynamics of a single particle under this scheme can be described by the hamiltonian @xmath121,\label{single - particle hamiltonian}\end{aligned}\ ] ] where @xmath122 is the momentum of the particles , and @xmath123 and @xmath124 are the energies of the ground and excited states , respectively . in the atom - light coupling term , @xmath125 and @xmath126 are the wave vectors and frequencies of the three raman lasers with @xmath127 the corresponding rabi frequencies , and @xmath128 is the matrix element of the dipole transition .
one can see that this hamiltonian is similar to that used in the scheme for creating 2d spin - orbit ( so ) coupling in ultracold fermi gases @xcite , thus can be readily realized in bose gases .
taking the standard rotating wave approximation to get rid of the time dependence of the hamiltonian , and adiabatically eliminating the excited states for far detuning , the hamiltonian can be rewritten as @xmath129 where @xmath130 , @xmath131 and @xmath132 are the two - photon detunings , and the real parameters @xmath133 describe the raman coupling strength between hyperfine ground states @xmath134 and @xmath135 , which can be expressed as @xcite @xmath136 here , @xmath137 is the intensity of each raman laser , and @xmath138 denotes the one - photon detuning .
other parameters @xmath139 , @xmath140 and @xmath141 in eq . (
[ raman coupling strength ] ) are the speed of light , permittivity of vacuum and elementary charge , respectively . in eq .
( [ raman coupling strength ] ) , @xmath142 is an index labeling the components of @xmath143 in the spherical basis , and @xmath144 describes the middle excited hyperfine spin state in the raman process .
for simplicity , we assume @xmath145 , which can always be satisfied by adjusting the system parameters , such as the laser intensity . introducing a unitary transformation @xmath146 and a time - dependent unitary transformation @xmath147 , the effective hamiltonian becomes @xmath148 where the laser vectors @xmath149 , @xmath150 and @xmath151 are defined with @xmath152 .
the spin - dependent uniform potential induced by the raman detuning @xmath153 and raman coupling strength @xmath154 can be eliminated by applying a zeeman field , leading to @xmath155 where @xmath156 and @xmath157 with @xmath158 and @xmath159 denoting the linear and quadratic zeeman energy respectively . by tuning the detuning , the zeeman energy and the raman coupling strength , one can reach the regime @xmath160 and @xmath161 which satisfying @xmath162 .
then we have @xmath163 which is the single - particle hamiltonian with su(3 ) so coupling considered in the main text .
in order to verify the stability of the phases discovered in this manuscript , we have explored the effects of a small but finite temperature , and concluded that the double - quantum spin vortex states are robust against the thermal fluctuations . in particular , we considered a random fluctuation @xmath164 in the real - time evolution of the gross - pitaevskii equation , which causes an energy fluctuation about @xmath165 with @xmath166 the ground - state energy .
an estimation shows that this level of fluctuation corresponds to the energy scale @xmath167 with @xmath168 nk , which is higher enough for a usual system of bose - einstein condensates in realistic experiments .
according to numerical simulations , we find that the structure of the double - quantum spin vortex state is stable under this level of fluctuation in tens of millisenconds [ see fig .
[ fig7 ] ] , suggesting that this phase is indeed observable in experiments .
99 y. j. lin , k. jimnez - garca , and i. b. spielman , nature * 471 * , 83 ( 2011 ) .
p. wang , z. q. yu , z. fu , j. miao , l. huang , s. chai , h. zhai , and j. zhang , phys . rev . lett . * 109 * , 095301 ( 2012 ) .
l. w. cheuk , a. t. sommer , z. hadzibabic , t. yefsah , w. s. bakr , and m. w. zwierlein , phys .
lett . * 109 * , 095302 ( 2012 ) .
j. y. zhang , s. c. ji , z. chen , l. zhang , z. d. du , b. yan , g. s. pan , b. zhao , y. j. deng , h. zhai , s. chen , and j. w. pan , phys .
rev . lett . * 109 * , 115301 ( 2012 ) . c. qu , c. hamner , m. gong , c. zhang , and p. engels , phys .
a * 88 * , 021604(r ) ( 2013 ) .
z. fu , l. huang , z. meng , p. wang , l. zhang , s. zhang , h. zhai , p. zhang , and j. zhang , nat . phys .
* 10 * , 110 ( 2014 ) . s. c. ji , j. y. zhang , l. zhang , z. d. du , w. zheng , y. j. deng , h. zhai , s. chen , and j. w. pan , nat
. phys . * 10 * , 314 ( 2014 ) .
a. j. olson , s .-
wang , r. j. niffenegger , c .- h .
li , c. h. greene , and y. p. chen , phys .
a * 90 * , 013616 ( 2014 ) .
l. huang , z. meng , p. wang , p. peng , s .- l . zhang , l. chen , d. li , q. zhou , and j. zhang , nat . phys .
* 12 * , 540 ( 2016 ) .
z. wu , l. zhang , w. sun , x .-
xu , b .- z .
wang , s .- c .
ji , y. deng , s. chen , x .-
liu , and j .- w .
pan , arxiv:1511.08170 ( 2015 ) .
a. m. dudarev , r. b. diener , i. carusotto , and q. niu , phys .
lett . * 92 * , 153005 ( 2004 ) . v. galitski and i. b. spielman , nature * 494 * , 49 ( 2013 ) .
j. dalibard , f. gerbier , g. juzelinas , and p. hberg , rev .
* 83 * , 1523 ( 2011 ) .
n. goldman , g. juzelinas , p. hberg , and i. b. spielman , rep .
. phys . * 77 * , 126401 ( 2014 ) .
h. zhai , int . j. mod
b * 26 * , 1230001 ( 2012 ) .
h. zhai , rep . prog . phys .
* 78 * , 026001 ( 2015 ) .
w. yi , w. zhang , and x. cui , sci .
china phys .
mech . astron .
58,014201 ( 2015 ) .
j. zhang , h. hu , x. j. liu , and h. pu , annu . rev .
* 2 * , 81 ( 2014 ) . c. wu , mod .
b * 23 * , 1 ( 2009 ) . c. wu , i. mondragon - shem , and x. f. zhou , chin .
28 , 097102 ( 2011 ) .
h. hu , b. ramachandhran , h. pu , and x. j. liu , phys .
lett . * 108 * , 010402 ( 2012 ) .
h. sakaguchi , b. li , and b. a. malomed , phys .
e * 89 * , 032920 ( 2014 )
. w. han , g. juzelinas , w. zhang , and w. m. liu , phys .
rev . a * 91 * , 013607 ( 2015 ) . c. wang , c. gao ,
c. m. jian , and h. zhai , phys .
lett . * 105 * , 160403 ( 2010 ) . t. l. ho and s. zhang , phys .
lett . * 107 * , 150403 ( 2011 ) .
y. li , l. p. pitaevskii , and s. stringari , phys .
lett . * 108 * , 225301 ( 2012 ) . t. f. j. poon and x. j. liu , phys .
a * 93 * , 063420 ( 2016 ) .
a. polkovnikov , k. sengupta , a. silva , and m. vengalattore , rev .
mod . phys . * 83 * , 863 ( 2011 ) .
y. li , g. i. martone , l. p. pitaevskii , and s. stringari , phys .
. lett . * 110 * , 235302 ( 2013 ) . g. i. martone , y. li , l. p. pitaevskii , and s. stringari , phys .
a * 86 * , 063621 ( 2012 ) .
k. sun , c. qu , y. xu , y. zhang , and c. zhang , phys .
a * 93 * , 023615 ( 2016 ) .
b. m. anderson , g. juzelinas , v. m. galitski , and i. b. spielman , phys .
lett . * 108 * , 235301 ( 2012 ) .
t. gra , r. w. chhajlany , c. a. muschik , and m. lewenstein , phys .
b * 90 * , 195127 ( 2014 ) .
r. barnett , g. r. boyd , and v. galitski , phys .
lett . * 109 * , 235308 ( 2012 ) . y. kawaguchi and m. ueda , phys . rep .
* 520 * , 253 ( 2012 ) .
t. isoshima , k. machida , and t. ohmi , j. phys .
. jpn . * 70 * , 1604 ( 2001 ) .
t. mizushima , n. kobayashi , and k. machida , phys .
a * 70 * , 043613 ( 2004 ) .
d. m. stamper - kurn and m. ueda , rev .
85 * , 1191 ( 2013 ) .
d. s. wang , y. r. shi , k. w. chow , z. x. yu , and x. g. li , eur .
j. d 67 , 242 ( 2013 ) .
d. s. wang , y. q. ma , and x. g. li , commun .
nonlinear sci .
19 , 3556 ( 2014 ) .
g. b. arfken , h. j. weber , and f. e. harris , _ mathematical methods for physicists _
( 7th ed . ) ( 2000 ) .
z. f. xu , y. kawaguchi , l. you , and m. ueda , phys . rev .
a * 86 * , 033628 ( 2012 ) .
z. lan and p. ohberg , phys .
rev . a * 89 * , 023630 ( 2014 ) .
y. zhang , l. mao , and c. zhang , phys .
lett . * 108 * , 035302 ( 2012 ) .
t. d. stanescu , b. anderson , and v. galitski , phys .
a * 78 * , 023616 ( 2008 ) .
s. sinha , r. nath , and l. santos , phys .
lett . * 107 * , 270401 ( 2011 ) .
h. saito , y. kawaguchi , and m. ueda , phys .
* 96 * , 065302 ( 2006 ) .
h. saito , y. kawaguchi , and m. ueda , phys .
a * 75 * , 013621 ( 2007 ) .
y. kawaguchi , h. saito , k. kudo , and m. ueda , phys .
a * 82 * , 043627 ( 2010 ) .
j. lovegrove , m. o. borgh , and j. ruostekoski , phys .
a * 86 * , 013613 ( 2012 ) .
k . liu , y .- c .
tsai , w. m. liu , and s .- c .
gou , phys .
a * 86 * , 023601 ( 2012 ) .
a. hubert and r. schafer , _ magnetic domains _ ( springer , berlin , 1998 ) .
t. shinjo , t. okuno , r. hassdorf , k. shigeto , and t. ono , science * 289 * , 930 ( 2000 ) .
a. wachowiak , j. wiebe , m. bode , o. pietzsch , m. morgenstern , and r. wiesendanger , science * 298 * , 577 ( 2002 ) .
l. s. leslie , a. hansen , k. c. wright , b. m. deutsch , and n. p. bigelow , phys .
* 103 * , 250401 ( 2009 ) .
jae - yoon choi , woo jin kwon , and yong - il shin , phys .
lett . * 108 * , 035301 ( 2012 ) .
s. yi and h. pu , phys .
* 97 * , 020401 ( 2006 ) .
l. e. sadler , j. m. higbie , s. r. leslie , m. vengalattore , and d. m. stamper - kurn , nature * 443 * , 312 ( 2006 ) .
other experimental techniques , such as selective absorption and time - of - flight measurement , can also be used to observe the double - quantum spin vortex states .
we suggest the magnetization - sensitive phase - contrast imaging method just because it has advantage in observing the phase information of the spin - vortex structure .
g. e. brown and m. rho , _ the multifaceted skyrmion _ ( world scientific , singapore , 2010 ) . m. demarco and h. pu , phys .
rev . a * 91 * , 033630 ( 2015 ) .
t. a. savard , s. r. granade , k. m. ohara , m. e. gehm , and j. e. thomas , phys . rev .
a * 60 * , 4788 ( 1999 ) . | we show that double - quantum spin vortices , which are characterized by doubly quantized circulating spin currents and unmagnetized filled cores , can exist in the ground states of su(3 ) spin - orbit coupled bose gases .
it is found that the su(3 ) spin - orbit coupling and spin - exchange interaction play important roles in determining the ground - state phase diagram . in the case of effective ferromagnetic spin interaction , the su(3 )
spin - orbit coupling induces a three - fold degeneracy to the magnetized ground state , while in the antiferromagnetic spin interaction case , the su(3 ) spin - orbit coupling breaks the ordinary phase rule of spinor bose gases , and allows the spontaneous emergence of double - quantum spin vortices .
this exotic topological defect is in stark contrast to the singly quantized spin vortices observed in existing experiments , and can be readily observed by the current magnetization - sensitive phase - contrast imaging technique . |
kleiner @xcite described all partially ordered sets ( _ posets _ ) with finite number of nonequivalent indecomposable representations .
we extend his description to unitary representations of posets .
the notion of poset representations was introduced by nazarova and roiter @xcite ( see also @xcite ) .
a _ matrix representation _ of a finite poset @xmath0 over a field @xmath16 is a block matrix @xmath17 $ ] over @xmath16 .
two representations @xmath17 $ ] and @xmath18 $ ] are _ equivalent _ if @xmath19 can be reduced to @xmath20 by elementary row transformations , elementary column transformations within @xmath21 , and additions of linear combinations of columns of @xmath21 to columns of @xmath22 if @xmath5 .
the _ direct sum _ of @xmath19 and @xmath20 is the representation @xmath23.\ ] ] a representation is called _ indecomposable _ if it is not equivalent to a direct sum of two representations .
it is sufficient to classify only indecomposable representations since each representation is equivalent to a direct sum of indecomposable representations , uniquely determined up to isomorphism of summands .
kleiner @xcite ( see also ( * ? ? ?
* theorem 5.1 ) and ( * ? ? ? * theorem 10.1 ) ) proved that a poset @xmath8 has only a finite number of nonequivalent indecomposable representations if and only if it does not contain a full poset whose hasse diagram is one of the form @xmath24 } \xymatrix @c=0.3 cm @r=0.5 cm { & & & & & & & & & & & & & & & \bullet \ar@{-}[d ] & & & & \\ & & & & & & & & & & & & & & & \bullet \ar@{-}[d ] & & & & \bullet \ar@{-}[d ] \\ & & & & & & & & & & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d ] & & & & \bullet \ar@{-}[d ] & & & & \bullet \ar@{-}[d ] \\ & & & & & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d ] & & & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d ] & & & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d ] & & \bullet \ar@{-}[d ] \ar@{-}[rd ] & \bullet \ar@{-}[d ] & \bullet \ar@{-}[d]\\ \bullet & \bullet & \bullet & \bullet & , \resizebox{0.3 cm } { ! } { } & \bullet & \bullet & \bullet & , \resizebox{0.3 cm } { ! } { } & \bullet & \bullet & \bullet & , \resizebox{0.3 cm } { ! } { } & \bullet & \bullet & \bullet & , \resizebox{0.3 cm } { ! } { } & \bullet & \bullet & \bullet&. } \end{split}\ ] ] an equivalent definition of poset representations can be given in terms of subspaces .
a _ subspace representation _ of @xmath0 is a tuple @xmath25 , in which @xmath2 is a vector space over @xmath16 and @xmath26 are its subspaces such that @xmath27 if @xmath5 ( i.e. , each representation is a homomorphism from @xmath8 to the poset of all subspaces of @xmath2 ) .
two subspace representations @xmath25 and @xmath28 are _ equivalent _ if there exists a linear bijection @xmath29 such that @xmath30 for all @xmath31 . for each subspace representation @xmath25
, one can construct a matrix representation @xmath32 $ ] in such a way that ( i ) for each @xmath31 the columns of all @xmath22 with @xmath33 generate the subspace @xmath3 ; and ( ii ) two subspace representations are equivalent if and only if the corresponding matrix representations are equivalent ; see ( * ? ? ?
* chapter 3 ) . from now on , all representations that we consider are over the field @xmath34 of complex numbers . by a _ unitary representation of dimension _
@xmath35 , we mean a subspace representation @xmath36 in which @xmath10 is a unitary space of dimension @xmath35 .
two unitary representations @xmath36 and @xmath25 of a poset @xmath8 are _ unitarily equivalent _ if there exists a unitary bijection @xmath37 such that @xmath38 for all @xmath31 .
orthogonal sum _ of unitary representations @xmath39 and @xmath40 is the unitary representation @xmath41 in which @xmath42 denotes the orthogonal sum of @xmath10 and @xmath2 .
a unitary representation is called _ orthogonaly indecomposable _ if it is not equivalent to an orthogonal sum of two unitary representations .
note that the problem of classifying unitary representations is hopeless even for the poset @xmath43 since by ( * ? ? ?
* theorem 4 ) it contains the problem of classifying an operator on a unitary space , and hence it contains the problem of classifying any system of operators on unitary spaces @xcite .
the classification becomes possible for a broader class of posets if we impose additional conditions on unitary representations .
we denote the orthogonal projection onto a subspace @xmath44 by @xmath45 and the set of positive real numbers by @xmath46 .
we say that a unitary representation @xmath36 is a _ representation of weight _ @xmath47 ( or _ @xmath7-representation _ ) if @xmath48 such relations appear in many areas of mathematics , see for example @xcite and references therein . our goal is to prove that kleiner s theorem holds for @xmath7-representations too : [ maintheoremfin ] the following conditions are equivalent for each finite poset @xmath8 with @xmath49 elements : 1 . for each @xmath50 , @xmath8 has only a finite number of indecomposable unitarily nonequivalent @xmath7-representations ; 2 . for each @xmath50 and @xmath51
, @xmath8 has only a finite number of indecomposable unitarily nonequivalent @xmath7-representations of dimension @xmath35 ; 3 .
@xmath8 does not contain a full poset whose hasse diagram is one of the form .
in what follows we suppose that the elements of a poset @xmath8 are numbered from @xmath52 to @xmath53 . a poset is called _ primitive _ and is denoted by @xmath54 if it is the disjoint ( cardinal ) sum of linearly ordered sets of orders @xmath55 . the diagrams and corresponding posets are called _
critical_. the poset which corresponds to the last diagram in the list is denoted by @xmath56 . to simplify the notation we denote a subspace representation @xmath1 of @xmath8 by @xmath57 .
the similar notation will be used for unitary representations and weights . a subspace representation @xmath58 is called _ schurian _ if all its endomorphisms are trivial ; that is , the ring @xmath59 is isomorphic to @xmath16 .
any schurian representation is indecomposable .
any unitary representation @xmath60 can be viewed as a subspace representation ; the forgetful map is denoted by @xmath61 .
if @xmath39 is an indecomposable @xmath7-representation , then @xmath62 is schurian ( see ( * ? ? ?
* theorem 1 ) ) .
[ lemdiag ] let @xmath63 , @xmath64 be orthogonal projections such that @xmath65 for @xmath66 with positive real @xmath67 .
let there exist a diagonal matrix @xmath68 with positive components such that @xmath69 for all @xmath31
. then @xmath70 and @xmath71 for all @xmath31 .
write @xmath72 $ ] , @xmath73 $ ] , @xmath74 $ ] , where @xmath75 . without losing generality
, we may assume that @xmath76 .
since @xmath77 , we have @xmath78 for @xmath79^t \in \mathbb c^{nm},\\ y&:=[\sqrt{\chi_1}q^{(1)}_{1,1},\ldots,\sqrt{\chi_1}q^{(1)}_{1,n},\ldots,\sqrt{\chi_m}q^{(m)}_{1,1},\ldots,\sqrt{\chi_m}q^{(m)}_{1,n}]^t \in \mathbb c^{nm } , \end{aligned}\ ] ] we have @xmath80 note that @xmath81 similarly , @xmath82 . by , we have @xmath83 , hence @xmath84 . by the cauchy - schwartz inequality ,
@xmath85 comparing this inequality with ( [ firstin ] ) , we have @xmath70 and @xmath71 for all @xmath31 .
[ unitvslin ] two @xmath7-representations @xmath86 and @xmath87 are unitarily equivalent if and only if the corresponding subspace representations @xmath62 and @xmath88 are equivalent .
if @xmath39 is unitarily equivalent to @xmath89 , then @xmath62 is equivalent to @xmath90 .
let us prove the converse statement .
@xmath62 is equivalent to @xmath88 if and only if there exists an invertible @xmath91 such that @xmath92 let @xmath93 be the polar decomposition of @xmath94 , where @xmath95 and @xmath96 are unitary maps and @xmath97 is a positively defined diagonal operator .
then @xmath98 hence @xmath99 for all @xmath31 .
since @xmath39 and @xmath89 are @xmath7-representations , @xmath100 lemma [ lemdiag ] ensures that @xmath101 for all @xmath31 .
therefore , @xmath102 for all @xmath31 , and so @xmath39 is unitarily equivalent to @xmath89 . by similar argumentation
, one can show that @xmath7-representation @xmath39 is orthogonally indecomposable if and only @xmath62 is indecomposable
. the connection between usual and orthoscalar representations of quivers was established ( * ? ? ?
* theorem 1 ) in the same way as in theorem [ unitvslin ] .
a representation @xmath58 of weight @xmath103 is called @xmath7-_stable _ if @xmath104 and @xmath105 for any proper subspace @xmath106 .
[ stab1 ] if @xmath107 is an indecomposable @xmath7-representation , then @xmath62 is @xmath7-stable . equating the traces of both sides in
, we obtain @xmath108 if @xmath109 is any proper subspace of @xmath10 , then @xmath110 . equating the traces of both sides in the last equality
, we get @xmath111 by ( * ? ? ?
* theorem 2 ) , @xmath112 for each two subspaces @xmath113 and @xmath114 , and so @xmath115 it remains to prove that the last inequality is strict .
indeed , assume that @xmath116 for all @xmath31
. then each @xmath117 commutes with @xmath45 .
hence the subspace @xmath109 is invariant with respect to the projections @xmath117 and the representation @xmath39 is decomposable .
this contradicts the assumption .
the converse statement to lemma [ stab1 ] also holds : if a representation @xmath58 is @xmath7-stable , then one can choose a scalar product in @xmath2 in such a way that @xmath40 becomes a @xmath7-representation ; see ( * ? ? ?
* theorem 3.5 ) . using results from @xcite
, one can prove the following theorem .
[ stability - un ] an indecomposable unitary representation @xmath39 is a @xmath7-representation if and only if the corresponding subspace representation @xmath62 is @xmath7-stable .
the implication @xmath118 is trivial .
assume that the hasse diagram of @xmath8 does not contain any of critical diagrams .
if @xmath8 has an infinite number of indecomposable unitarily nonequivalent @xmath7-representations for some weight @xmath7 , then by theorem [ unitvslin ] it has an infinite number of nonequivalent indecomposable subspace representations . by kleiner s theorem , @xmath8 contains a critical diagram ; a contradiction .
we say that a poset @xmath8 is _ unitary representation - infinite _ if there exist @xmath121 and @xmath122 such that @xmath8 has an infinite number of indecomposable unitarily nonequivalent @xmath123-representations of dimension @xmath35 .
our aim is to prove that critical posets are unitary representation - infinite .
one can show that critical primitive posets are unitary representation - infinite using @xcite .
namely , there exists a correspondence between the @xmath7-representations of a given poset @xmath8 and the representations of a certain @xmath124-algebra @xmath125 associated with a star - shaped graph @xmath126 , which is determined by the hasse diagram of @xmath8 , and the parameter @xmath127 is determined by the weight @xmath7 . if @xmath126 is an extended dynkin graph ( which corresponds to some primitive critical @xmath8 ) , then one can choose the parameter @xmath127 such that @xmath125 has an infinite number of unitarily nonequivalent irreducible representations .
the complete description of such representations was given in @xcite ( see also remark [ remfinale ] ) .
but we use another method that handles both primitive and non - primitive cases . denote by @xmath128 ( or @xmath129 if no confusion can arise ) the @xmath130-dimensional vector in which the @xmath31-th coordinate is @xmath52 and the others are @xmath131 .
denote by @xmath132 the vector @xmath133 and by @xmath134 the vector space spanned by @xmath135 . * if @xmath138 , then @xmath139 consists of the space @xmath140 and its subspaces @xmath141 * if @xmath142 , then @xmath139 consists of the space @xmath143 and its subspaces @xmath144&\langle e_1 \rangle\ar@{->}[u ] & \langle e_3 \rangle\ar@{->}[u]}\ ] ] * if @xmath145 , then @xmath139 consists of the space @xmath146 and its subspaces @xmath147&\langle e_2,e_3 \rangle\ar@{->}[u]\\ \langle e_{123},e_{24 } \rangle&\langle e_4 \rangle\ar@{->}[u]&\langle e_3 \rangle \ar@{->}[u]}\ ] ] * if @xmath148 , then @xmath139 consists of the space @xmath149 and its subspaces @xmath150\\ & & \langle e_2,e_3,e_4 \rangle\ar@{->}[u]\\ & \langle e_1,e_2,e_5,e_6 \rangle&\langle e_3,e_4 \rangle\ar@{->}[u]\\ \langle e_{123},e_{245},e_{16 } \rangle&\langle e_5,e_6 \rangle\ar@{->}[u]&\langle e_4 \rangle\ar@{->}[u]\\}\ ] ] * if @xmath151 , then @xmath139 consists of the space @xmath152 and its subspaces @xmath153\\ \langle e_{235},e_{134},e_5,e_3+\lambda e_4 \rangle&\langle e_1,e_2,e_5 \rangle&\langle e_3,e_4 \rangle\ar@{->}[u]\\ \langle e_{235},e_{134 } \rangle\ar@{->}[u]&\langle e_5 \rangle\ar@{->}[ul]\ar@{->}[u]&\langle e_4 \rangle\ar@{->}[u]}\ ] ] denote by @xmath154 the only subspace from @xmath155 that depends on the parameter @xmath156 and denote by @xmath157 the element from @xmath158 that corresponds to @xmath154 . deleting @xmath154 from @xmath155
, we obtain the subspace representation @xmath159 of primitive poset @xmath160 .
let @xmath158 be a critical poset .
the poset @xmath164 is primitive and does not contain any of the critical posets , its subspace representation @xmath163 is schurian . by (
* proposition 3.1 ) , there exists a weight which we denote by @xmath165 , such that @xmath163 is @xmath165-stable .
write @xmath166 the subspace representation @xmath167 is @xmath165-stable , hence @xmath168 .
let @xmath169 be such that @xmath170 .
write @xmath171 and @xmath172 by proposition [ nonequiv ] and proposition [ stabcrit ] , any critical poset @xmath8 has an infinite number of nonequivalent @xmath123-stable subspace representations . by theorem [ stability - un ]
, @xmath8 has an infinite number of indecomposable unitarily nonequivalent @xmath123-representations .
suppose that @xmath8 contains a critical poset @xmath177 . by proposition [ crinf ] ,
there exists a weight @xmath178 such that @xmath177 has an infinite number of indecomposable unitarily nonequivalent @xmath179-representations of dimension @xmath35 .
define the following subset of @xmath8 : @xmath180 for each @xmath178-representation @xmath181 , define the unitary representation @xmath182 of @xmath8 as follows : @xmath183 it is easy to check that @xmath89 is @xmath184-representation , in which @xmath185 is defined by @xmath186 hence @xmath8 is unitary representation - infinite . [ remfinale ] define the following weights : @xmath187 each weight @xmath123 obtained from the minimal imaginary root of the quadratic form related to a critical poset @xmath8 .
we checked ( describing all possible subdimension vectors ) that the representations @xmath155 are @xmath123-stable for any @xmath188 .
hence they give rise to an infinite family of nonequivalent @xmath123-representations .
for primitive @xmath8 one can obtain the precise description of projections for such representations using the results from @xcite .
the description in the case @xmath189 is unknown .
* acknowledgments .
* we would like to thank s.a .
kruglyak , v.l .
ostrovskii , and l.b .
turowska for useful discussions and advises and also v.v .
sergeichuk for numerous helpful remarks .
the second author was partially supported by dfg grant schm1009/4 - 1 and fapesp grant 2010/15781 - 0 .
nazarova , a.v .
roiter , representations of partially ordered sets , zap .
otdel . mat .
inst . steklov .
( lomi ) 28 , 1972 , 531 ( in russian ) ; english version : j. soviet math .
5 ) ( 1975 ) 585606 .
ostrovskyi , yu.s .
samoilenko , on spectral theorems for families of linearly connected selfadjoint operators with prescribed spectra associated with extended dynkin graphs , 58 ( 11 ) ( 2006 ) 17681785 . | a subspace representation of a poset @xmath0 is given by a system @xmath1 consisting of a vector space @xmath2 and its subspaces @xmath3 such that @xmath4 if @xmath5 .
for each real - valued vector @xmath6 with positive components , we define a unitary @xmath7-representation of @xmath8 as a system @xmath9 that consists of a unitary space @xmath10 and its subspaces @xmath11 such that @xmath12 if @xmath13 and satisfies @xmath14 , in which @xmath15 is the orthogonal projection onto @xmath11 .
we prove that @xmath8 has a finite number of unitarily nonequivalent indecomposable @xmath7-representations for each weight @xmath7 if and only if @xmath8 has a finite number of nonequivalent indecomposable subspace representations ; that is , if and only if @xmath8 contains any of kleiner s critical posets
. representations of partially ordered sets , representation - finite type , kleiner s theorem 15a63,15a21,16g20 . |
in the past two decades , improved understanding of dimension has been stimulated by studies of nonlinear dynamical systems and their strange attractors .
canonical dimensions used to characterize strange attractors have been defined in terms of zero - scale limits .
these limit - based dimensions are inferred by varying a partition scale and extrapolating results to an invariant limit . in this paper
we consider _ scale - local _ rnyi dimensions of a strange attractor of the hnon map as explicit functions of scale .
we demonstrate that approximations to the zero - scale limit by extrapolation represent scale averages of scale - local dimension .
we also contrast dimension defined as a topological invariant with dimension based on numerical analysis of arbitrary distributions and the problem of estimation based on computationally - accessible scale intervals , for which extrapolations based on _ a priori _ considerations are not relevant .
in this paper we review limit - based and scale - local dimension definitions , identify limit - based dimensions as scale averages of scale - local dimensions , survey previous limit - based dimension analyses of a hnon attractor , present a scale - local dimension analysis of this attractor , and examine the general structure of scale - local and running - average dimension distributions and their monotonicity with rnyi index @xmath0 .
we conclude by examining the status of limit - based dimensions as topological invariants and their relation to scale - local dimension , information and dimension transport .
dimensionality is a fundamental property of distributions characterizing their correlation structure .
modern dimension theory is based on the work of carathodory and hausdorf @xcite .
dimension is there defined in terms of asymptotic limits of set measures depending on partition scale .
limit - based dimensions are invariant properties of distributions also invariant under certain transformations .
hausdorff - besicovitch dimension @xcite is based on the set measure @xmath1 where @xmath2 is a partition of the embedding space , @xmath3 is the _ support _ of set @xmath4 and @xmath5 is the size of the @xmath6 partition element .
the characteristic scale @xmath7 of a hausdorff partition is an
_ upper limit _ on partition - element size .
if the point measure or dimension function is assumed to have a power - law form @xmath8 with @xmath9 arbitrary , then hausdorff - besicovitch dimension @xmath10 is the value of @xmath9 for which @xmath11 remains finite , @xmath12 .
box - counting dimension is based on partition elements ( boxes ) which have a common size ( in contrast to hausdorff partitions ) .
the number @xmath13 of boxes of size @xmath7 required to cover a set @xmath4 is @xmath14 for self - similar sets @xmath13 is expected to follow a power law on scale : @xmath15 , where @xmath16 is the fractal or box - counting dimension , in which case @xmath17 with @xmath18 the boundary size of @xmath4 . by analogy
one can define a box - counting set measure @xcite @xmath19 the limit @xmath20 should remain finite for @xmath9 = @xmath16 . box - counting dimension can be evaluated with the computationally convenient eq .
( [ boxdim ] ) , whereas hausdorff set measure , an _ infinite _ sum over partition elements , can not be factored as in the second line of eq .
( [ hausbox ] ) .
hausdorff and box - counting dimensions were initially applied to strange attractors with the goal of better understanding the nonlinear processes which define such distributions .
the dimension concept was later expanded to information , correlation and higher rnyi dimensions @xcite , reflecting multipoint ( @xmath0-point ) correlations of measure distributions .
these dimensions are also conventionally defined in terms of zero - scale limits .
box - counting and hausdorff dimensions correspond to @xmath21 .
information dimension @xmath22 is defined @xmath23 application of lhpital s rule in the limit @xmath24 .
correlation dimension @xmath25 @xcite measures two - point ( @xmath26 ) correlations as defined by the correlation integral c_2(e , e ) & = & 1 n^2(e ) _
i , j=1^n ( e-|x_i - x_j| ) + & & _ i=1^m(e , e ) n_i^[2](e , e)/n^[2](e ) , with @xmath27 as @xmath28 in the case of a fractal . in general , @xmath29 , @xmath30 and @xmath31 , with the rnyi dimensions
@xcite defined by @xmath32 and @xmath33 , \end{aligned}\ ] ] where we approximate the rank-@xmath0 normalized correlation integral by @xmath34 } / n^{[q]}$ ] @xcite , @xmath35 } = n ! / ( n - q)!$ ] , @xmath36 is the normalized contents ( point count or measure total ) of the @xmath6 partition element , @xmath37}/n^q \rangle \approx \langle n_i / n \rangle^q$ ] , @xmath18 is the boundary size and @xmath38 is the total point count or measure in @xmath4 .
for the analysis of arbitrary distributions over scale intervals limited by computational or experimental accessibility we must understand dimension in a larger context without assumptions of recursiveness and extrapolation to asymptotic limits ( generalizing beyond distributions defined by recursive mappings or iterated part removal ) .
within bounded scale intervals distinctions are required which are not relevant for limit - based dimensions . scale - local dimension , in contrast to limit - based dimension , is an explicit function of scale relating _ relative change of a measure to relative change of scale _ at arbitrary scale and space points @xmath39 represented compactly by @xmath40 .
the dimension distribution on scale depends on scale resolution @xmath41 .
the convenient differential notation should not imply an asymptotic resolution limit ( @xmath41 will be suppressed notationally but must be specified for numerical analysis ) .
in contrast to limit - based dimensions no _ a priori _ assumption is made about scale dependence or asymptotic values .
dimension @xmath42 relates relative change in a correlation integral to relative change in scale @xmath43 @xmath44 represents the _ projection _ of a @xmath45-dimensional space onto a @xmath46-dimensional subspace or hyperplane perpendicular to its main diagonal , the domain of the correlation integral @xmath47 .
@xmath48 is the corresponding rank - q dimension of set @xmath4 in the @xmath49-dimensional primary space defined by @xcite @xmath50 scale - local dimensions are expressed in terms of rnyi entropies eq .
( [ renent ] ) as @xmath51
entropy @xmath52 depends on _ scale intervals _ : @xmath53 is determined by the number of bins of size @xmath7 within a distribution boundary of size @xmath18 .
rnyi entropies are expressed as definite integrals of scale - local dimensions by inverting eq .
( [ renyidim ] ) @xmath54 limit - based rnyi dimensions in the form [ dlim ] d_q^(e ) & & _ e 0 + & = & _ e 0 + & & _ e 0 .
are thereby interpreted as _ mean values _ of scale - local dimensions averaged ( in principle ) over a ` semi - infinite ' scale interval bounded above by the boundary size of the distribution and below ( in practice ) by finite computing power . if distribution @xmath4 is described by a single dimension value over the averaging interval the distribution on dimension is then a delta function ( basis for idealized arguments ) and the mean obviously takes this value .
more generally , limit - based dimension is the mean of a finite - width distribution on dimension
. the width and skewness of the distribution may also be of interest for characterizing attractors
. the distribution may even be multimodal ( for more complex scale dependencies ) or have no central tendency . the mean value defined as a zero - scale limit in eq . ( [ dlim ] )
can be generalized to an explicit function of scale interval : @xmath55 .
scale interval @xmath56 $ ] then ranges from the limit - based @xmath57 $ ] to the scale - local @xmath58 $ ] and may be chosen to minimize boundary - scale and void - bin biases in a scale - local dimension distribution .
variation of @xmath55 with decreasing lower bound @xmath59 may give a misleading impression of convergence to @xmath60 .
integration over an increasing scale interval attenuates scale - local variation in proportion to the averaging interval , suggesting convergence to a zero - scale dimension value ( if it exists ) which is not necessarily the limit of the mean .
for illustration we consider limit - based and scale - local dimensions of a well - known strange attractor of the hnon map @xcite .
@xmath61 the attractor we analyze corresponds to parameter values @xmath62 and @xmath63 , whose properties have been extensively described in the literature , and falls in the bounded region @xmath64 $ ] .
orbits are either trapped within this region or unbounded @xcite .
this attractor is useful because it demonstrates central tendency ( on dimension ) in the @xmath48 distributions on scale , and the @xmath65 variations of the distributions are significant : @xmath66% of mean values as indicated in fig .
[ henplotsa1 ] .
a finite - length mapping generates a sample of the attractor , not the attractor itself ( the parent distribution for the sample ) . for each mapping
we choose an initial random seed @xmath67 and discard iterations until we sample points on the attractor itself .
increased sample number ( larger point density ) reduces the small - scale limit of the dimension distribution .
multiple sampling with randomly - chosen seeds indicates which dimension characteristics are independent of initial conditions .
the limit - based dimension @xmath68 is defined in principle by eqs . ( [ boxdim],[classdim ] ) .
[ classdim ] d^ _ e 0 .
the boundary size @xmath18 is often not specified , assumed to be unity or used as a free fit parameter . in the limit
, the choice of @xmath18 would not matter . for practical analysis on a bounded scale interval
the choice of @xmath18 may be a critical bias source for dimension estimates .
value @xmath69 @xcite was determined by a linear least - squares fit to @xmath70 _ vs. _ @xmath71 over a bounded scale interval , with @xmath18 a free fit parameter .
@xmath72 minimization determines an effective dimension definition ( for fit - optimized @xmath18 ) [ avdim ] d^ & & = , a ratio of weighted scale averages favoring smaller scales .
since each @xmath73 is effectively a scale average the @xmath72 fit results in an _ average of averages_. for arbitrary dimension distributions this difference in effective definition could be quite significant .
local averages over six scale intervals were determined from corrected box counts @xmath13 at seven scale points with @xmath74 $ ] @xcite [ null ] d_i & & = i .
these six values are compared ( solid points ) to other averaging schemes and measurements in fig .
[ dimsurv ] and to the full scale - local dimension distributions in fig .
[ henplotsa1 ] , where we compare @xmath75 to @xmath76 obtained from a scale - local distribution .
value @xmath77 was obtained in @xcite by extrapolating the @xmath75 from eq .
( [ null ] ) to a ` zero - scale ' limiting value , driving the estimate somewhat high compared to a uniformly - weighted average over the entire @xmath78 $ ] scale interval .
convergence is apparently expected with decreasing scale to a constant value @xmath79 , confusing @xmath80 with @xmath81 . ) for @xmath82 based on corrected box counts from @xcite .
solid dots are from eq .
( [ null ] ) and open triangles are a running average from eq .
( [ double ] ) , again using the same box counts .
the solid horizontal line at 1.28 represents the extrapolation from @xcite , the dashed line at 1.26 represents a box - counting result and lyapunov upper limit from @xcite , width=384 ] in fig .
[ dimsurv ] and elsewhere the attractor boundary size assumed for the plot axis definition is @xmath83 .
scale values for the solid points ( @xmath75 from @xcite ) were rescaled accordingly for this figure . using the @xmath84 from @xcite we also plot a variant of eq .
( [ classdim ] ) [ single ] d_i & & i , where @xmath85 is @xmath86 for the upper open circles and @xmath87 for the lower open circles .
these curves suggest the bias possible from different choices of boundary scale in this definition and the slow convergence in the mean noted in @xcite . with the same @xmath84 we also plot ( open triangles ) the running average .
[ double ] d_i & & = i .
the solid horizontal line at 1.28 represents the extrapolation from @xcite , the dashed line at 1.26 represents the box - counting result and lyapunov upper limit from @xcite .
we contend that _ scale - local _ rnyi dimensions defined over an arbitrary bounded scale interval provide the most fundamental and accessible representation of the dimensionality and correlation structure of an arbitrary distribution , that limit - based dimensions , as asymptotic limits ( by extrapolation ) of averages , are secondary . for a finite sample with bin contents @xmath88 @xmath89
where @xmath90 , and the brackets indicate bin dithering to further reduce bias , especially important near the boundary scale @xcite . while @xmath13 ( and therefore @xmath91 ) can in principle be obtained by correcting void - bin bias as noted above , we do not make a correction in this analysis , relying instead on large sample number @xmath38 to minimize bias within a specified scale interval .
attractor point - set data are compactly summarized by applying an `` atomic '' or microscale binning at some small scale well below the lower limit of the intended working scale interval .
the microbin contents are stored as a zero - suppressed list of bin occupancies . to calculate the relevant topological measures the microbinning is itself binned at increasing scale ( with scale points defined by integer multiples of the microbin scale ) .
this approach maintains analysis generality while minimizing computation time .
the number of scale points available for analysis ( scale resolution ) is limited by the microbin scale .
this scale can always be chosen small enough to allow for any requisite scale resolution , at the expense of greater computation time . in this analysis
the microbin scale is chosen to be more than a decade below the lower bound of the working scale interval .
this choice insures that integer granularity does not limit the desired scale resolution .
a very high resolution study over 10 decades of scale is reported in @xcite .
entropy values are calculated for approximately 20 scale points in each decade of scale , implying a scale resolution @xmath92 , and dimension values are calculated between pairs of neighboring entropy points as ratios of finite differences .
details of scale - local entropy calculation , including bin dithering to minimize boundary - scale bias , are given in @xcite . in fig .
[ henondim ] we present scale - local dimensions @xmath93 for @xmath94 and @xmath95 ( solid and dashed respectively in right panel ) for an analysis with scale resolution @xmath96 .
distributions in the left panel rise from zero above the boundary scale ( the whole attractor appears as a single point ) to values near 1.2 at scales well below the boundary .
distributions return to zero at small scale , reflecting resolution of individual sample points .
the right panel with expanded vertical scale indicates that typical variation of either distribution is larger in amplitude than the difference between their means .
horizontal lines at 1.28 ( solid ) and 1.26 ( dashed ) reflect estimates of limit - based dimensions @xmath97 respectively in @xcite and two sample numbers : n = 10k ( solid ) and 300k ( dashed ) .
right panel
n = 300k distributions from left panel for @xmath94 ( solid curve ) and @xmath98 ( dotted curve ) . expanded vertical scale allows comparison with limit - based values ( horizontal lines ) .
[ henondim],width=384 ] fig .
[ henplotsa1 ] shows higher - resolution distributions of scale - local dimensions @xmath99 ( solid ) and @xmath100 ( dashed ) for 10 m mapping interactions on scale interval @xmath101 $ ] also used to form running - average distributions in fig .
[ henplotsa2 ] . within this interval
the distributions closely approximate those of the parent .
extensions beyond this optimum interval ( gray curves ) are subject to significant bias ( boundary and void - bin )
. dotted curves at smaller scale ( for 1 m iterations ) illustrate void - bin bias .
also plotted are the local - average @xmath75 values from @xcite ( solid dots ) shown in fig .
[ dimsurv ] .
averages @xmath102 of the scale - local distribution over nearly matching scale intervals ( open boxes ) are in good agreement .
( solid ) and @xmath103 ( dashed ) over four scale decades for 10 m map iterations .
dotted curves are for 1 m map iterations showing void - bin bias .
solid dots are local scale averages @xmath75 from @xcite .
open boxes are comparable averages of scale - local dimensions showing good agreement .
horizontal lines correspond to limit - based estimates of @xmath104 @xcite ( solid ) and of @xmath105 and @xmath106 = 1.26 @xcite ( dashed).,width=384 ] the distributions are highly structured ( structure in fig .
[ henplotsa1 ] limited by scale resolution @xmath107 ) and very reproducible ( typical variation with different mapping seeds is within two line widths ) .
scale - local dimension is clearly not point - wise monotonically ordered with @xmath0 ( discussed further in sec .
[ mono ] ) .
there is no trend within this scale interval for decreasing structure at smaller scale .
it is suggested in another high - resolution study over 11 decades @xcite that the dimension distribution may itself be self - similar on scale _ resolution _ and does not obviously converge to a limiting value at smaller scale . for @xmath108 for distributions in fig .
[ henplotsa1 ] with @xmath109 .
open triangles show data from @xcite also as a running average according to eq .
( [ double ] ) .
lower curves show boundary bias due to starting point @xmath110 .
uppermost curves show results from modified limit - based definition eq .
( [ single ] ) and @xmath111 .
open triangles show data from @xcite as a running average according to eq .
( [ double]).,width=384 ] in fig .
[ henplotsa2 ] are shown running averages of two scale - local distributions of the form @xmath112 eqs .
( [ dlim],[double ] ) based on @xmath113 distributions in fig . [ henplotsa1 ] .
middle curves ( solid : @xmath21 , dashed : @xmath103 ) correspond to @xmath114 chosen to minimize bias near the boundary scale .
these running averages continued over a semi - infinite scale interval would terminate in limit - based dimension values .
the low - scale limiting values in this plot furnish the best _ estimates _ of limit - based dimensions given finite computing resources .
lower curves correspond to @xmath115 and are severely biased .
horizontal lines correspond to limit - based estimates of @xmath104 @xcite ( solid ) and of @xmath105 and @xmath106 = 1.26 @xcite ( dashed ) .
open triangles are data from @xcite treated as a running average according to eq .
( [ double ] ) .
uppermost two curves illustrate eq .
( [ single ] ) for @xmath116 showing a different aspect of boundary bias for this dimension definition .
similar results obtain for eq .
( [ avdim ] ) ( there is a small vertical shift due to the different scale weighting ) . for @xmath117 $ ] ( @xmath0 increasing from top ( solid ) to bottom ( dashed ) curve ) .
right panel running scale averages @xmath112 of the distributions in the left panel , showing monotonic @xmath0-ordering for large averaging intervals.,width=384 ] the left panel in fig .
[ qplotgen ] shows scale - local @xmath118 for @xmath117 $ ] ( @xmath0 increasing from top to bottom curve ) .
we observe large excursions in dimension values ( up to 30% over the scale interval ) .
these results compare well with @xmath119 in fig . 2 of @xcite . increasing @xmath0 acts as a contrast adjustment , enhancing sensitivity to large density variations on the attractor .
there is a suggestion of periodicity on scale for larger @xmath0 values .
the right panel shows running averages @xmath120 , also for @xmath117 $ ] . for averages over a sufficiently large scale interval the running averages are consistent with monotonicity on @xmath0 , as discussed in sec .
values at the small - scale endpoints are numerical estimators for limit - based dimensions .
we observe in figs .
[ henondim ] , [ henplotsa1 ] that scale - local @xmath121 and @xmath122 are not monotonic with respect to @xmath0 , @xmath0-ordering alternates with varying scale .
monotonic decrease of rnyi _ entropies _ with @xmath0 has been established for certain conditions .
following @xcite [ qderiv ] s_q(e ) / q = - _ i z_qi ( z_qi / p_i ) , with @xmath123 .
since the sum on the right ( kullback information ) is non - negative definite the @xmath0-derivative is never positive .
limit - based dimensions @xmath124 then inherit this monotonicity .
however , a similar @xmath0-ordering of scale - local dimensions is not generally valid .
the @xmath0-derivatives for scale - local dimension are [ dimq ] d_q(e ) / q = \ { _ i z_qi ( z_qi / p_i ) } .
there is no _ a priori _ restriction on the sign of the scale derivative on the right ( a form of _ dimension transport _ see sec .
[ topinv ] and @xcite ) .
dimension transport , the scale derivative of scale - local information , can have positive or negative values .
however , dimension transport averaged over a scale interval descending from the boundary scale _ must _ be negative - definite , since the integral over that scale interval is the information : @xmath125 @xcite .
if the ` information ' in the bracket of eq . ( [ dimq ] ) is positive definite ( a possible condition on transformations ) , scale - averaged dimension over the corresponding scale interval is monotone - decreasing with @xmath0 : @xmath126 .
the running averages in the right panel of fig .
[ qplotgen ] are indeed monotonic with @xmath0 over most of the observed scale interval , whereas the scale - local distributions ( left panel ) are significantly non - monotonic .
the invariance of dimension under a transformation depends on the amount of information generated by the transformation .
we show that invariance of limit - based dimensions under transformation holds if the amount of information generated by the transformation is bounded over a semi - infinite scale interval .
this condition can be compared to the definition of homeomorphism as a ` continuous ' or ` smooth ' bijective map .
we have established that limit - based dimensions are averages of scale - local dimensions over semi - infinite scale intervals .
transformations alter scale - local dimension distributions in the form of _ dimension transport _ , the scale derivative of information @xcite .
some transformations produce transport over a finite scale interval , others over semi - infinite scale intervals .
an example of the latter is the hnon map itself , which maps a 2d space to a @xmath127d attractor .
such maps , and recursive transformations like peano and baker foldings , have no lower - scale boundary : limit - based dimensions are not protected as invariants .
the change in limit - based rnyi dimension with dimension transport resulting from a transformation can be written [ diminv ] d_q^ & = & _ e 0 \ { _ e^l d_q(e ) d e / _ e^l d e } + & = & _ e 0 , where _ e.g. _
dimension transport for a 2d embedding space is a signed number bounded by @xmath128 $ ] .
if a transformation generates dimension transport over a _ finite _ scale interval its integral in eq . ( [ diminv ] ) , which is information @xmath129 , should be zero in the asymptotic limit , and limit - based dimensions remain invariant .
the scale dependence of dimension transport reflects the scale structure of transformations and thus provides a basis for classifying them .
an analogy can be drawn between the logarithmic system of entropy , information , dimension and dimension transport which describes correlation over arbitrary scale intervals and the linear system of correlation integral , cumulant , autocorrelation and autocorrelation difference suitable for correlation structure restricted to small scale intervals .
difference quantities which compare two distributions ( possibly related by a transformation ) can be defined in the two systems by c_q(e ) & = & ^e_0 a_q(e ) d e + c_q(e ) & = & ( 1-q ) _
e^l d_q(e ) d e , where @xmath130 is an autocorrelation difference or _
net _ autocorrelation , @xmath131 is dimension transport , @xmath132 can be related to cumulants and @xmath133 to information .
autocorrelation and scale - local dimension play analogous roles as local or differential correlation measures on scale , in linear and logarithmic systems respectively .
entropies and correlation integrals are the corresponding integral quantities .
changes in correlation result in _ transport _ of autocorrelation or dimension on scale as conserved quantities .
choice of linear or logarithmic correlation measures depends on the scale structure of distributions .
the linear system ( with a long history of development ) is better suited for linear periodic systems ; the logarithmic system for complex , nonlinear aperiodic systems .
topology seeks static properties of transformation - invariant sets amenable to rigorous treatment .
science seeks to describe the evolution of transient complex systems viewed imperfectly .
science thus requires a broader descriptive system , a generalized , self - consistent treatment containing topological invariance and limit - based measures as a special case .
studies of nonlinear dynamics have provided a fruitful interface between mathematical idealization and real - world arbitrariness . after a period of rapid development
one could pursue a path toward greater rigor which emphasizes a subset of interesting dynamics and methods and favors limit - based measures .
alternatively , one could strive to encompass arbitrarily scale - dependent systems in a generalized scale - local theory of dimension , with limit - based dimension as a special case . in this paper
we compare scale - local and limit - based dimensions through analysis of a well - known distribution a strange attractor of the hnon map .
we find that limit - based dimension is an asymptotic scale average of scale - local dimension .
the former is not @xmath134 , a limiting scale - local value at zero scale , but rather @xmath135 , an average over a semi - infinite scale interval .
the general quantity is @xmath136 , an average over arbitrary scale interval with limiting cases @xmath137 ( limit - based dimension ) and @xmath138 ( scale - local dimension at scale @xmath71 and resolution @xmath41 ) . running average @xmath139 descending from a boundary scale _ estimates _ limit - based dimension in experimental or computational contexts .
scale - local rnyi dimensions of the hnon map attractor analyzed here are _ highly scale dependent _ , with variations on scale up to 30% of mean values , no indication of convergence to a fixed limit and no consistent monotonicity on @xmath0 index .
running averages based on scale - local distributions do exhibit monotonicity on @xmath0 index for significant averaging intervals . running scale averages
provide the best means for estimating limit - based dimension values from numerical or experimental data , but as statistical estimators these averages are subject to several significant sources of bias , as demonstrated in this and other analyses .
we re - express dimension invariance in terms of scale averages , information and dimension transport .
limit - based dimension defined on a semi - infinite scale interval is invariant under transformations corresponding to _ finite information _ , or equivalently dimension transport restricted to a finite scale interval ( recursive transformations for example do not satisfy this condition ) . by the same argument ,
dimension _ estimators _ derived from finite scale intervals are not generally invariant under transformations , in fact may be used to study the scale - dependent structure of transformations .
dimension transport provides an alternative classification basis .
7 pesin y b 1997 _ dimension theory in dynamical systems _ ( the university of chicago press ) .
hausdorff f 1919 math .
ann . * 79 * 157 .
czy j 1994 _ paradoxes of measures and dimensions originating in felix hausdorff s ideas _ ( world scientific , singapore ) .
falconer k 1990 _ fractal geometry _
( wiley , chichester ) .
rudolph o 1994 _ thermodynamic and multifractal formalism and the bowen - series map _ , desy preprint ( issn 0418 - 9833 ) 94 - 122 .
baker g l and gollub j p 1990 _ chaotic dynamics _ ( cambridge , cambridge university ) .
grassberger p 1983 _ phys .
lett . _ * 50 * 346 - 349 rnyi a 1960 _ mta iii .
* 10 * 251 - 282 .
reid j g and trainor t a math - ph/0304010 , submitted to _
hnon m 1976 _ commun .
phys . _ * 50 * 69 .
hunt b r 1996 _ nonlinearity _ * 9 * 845 - 852 .
grassberger p 1983 _ phys .
lett . _ a * 97 * 227 - 230 .
russell d a , hansen j d and ott e 1980 _ phys .
* 45 * 1175 .
grassberger p 1983 _ phys .
lett . _ a * 97 * 224 - 226 . , edited by shlesinger m f , zaslovsky g m and frisch u 1995 ( springer - verlag , berlin ) .
feller w 1971 _ an introduction to probability theory and its applications vol ii _
( wiley , ny ) .
sprott j c 2003 _ chaos and time - series analysis _ ( oxford university press ) .
grassberger p 1988 _ phys .
lett . _ a * 128 * 369 - 373 .
trainor t a 1998 _ scale - local topological measures _ preprint , university of washington . | we compare limit - based and scale - local dimensions of complex distributions , particularly for a strange attractor of the hnon map .
scale - local dimensions as distributions on scale are seen to exhibit a wealth of detail .
limit - based dimensions are shown to be averages of scale - local dimensions , in principle over a semi - infinite scale interval .
we identify some critical questions of definition for practical dimension analysis of arbitrary distributions on bounded scale intervals .
.2 in keywords : scale , dimension , information , correlation integral , hnon map |
the dust cycle is currently considered as the key process controlling the variability of the martian climate at interseasonal and interannual time scales , as well as the weather variability at much shorter time scales .
the atmospheric thermal and dynamical structures , as well as the transport of aerosols and chemical species , are all strongly dependent on the dust spatio - temporal distribution , particle sizes , and optical properties . since the first scientific observations of a planet - encircling dust storm by ground - based telescopes in the late 50 s
, dust has been one of the main objectives of many of the spacecraft missions to mars over more than 40 years .
recent and ongoing missions , such as mars global surveyor ( mgs ) , mars odyssey ( ody ) , mars express , mars exploration rover ( mer ) , mars reconnaissance orbiter ( mro ) , and mars science laboratory , have included spectrometers , radiometers or imagers to measure radiances at wavelengths sensitive to dust .
atmospheric dust can be qualitatively observed by spacecraft cameras , such as the mars orbiter camera ( moc ) aboard mgs , and the mars color imager ( marci ) aboard mro ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for dust climatologies based on camera observations , as well as the mars daily weather websites ) .
one of the key physical parameters used to quantify the presence of mineral dust in the atmosphere is the column ( or total ) optical depth ( or optical thickness ) .
it is a measure of the fraction of radiation at a specific wavelength that would be removed from the vertical component of a beam during its path through the atmosphere by absorption and scattering due to airborne dust .
it can also be defined as the integral over an atmospheric column of the profile of the extinction opacity ( or extinction coefficient ) , where extinction accounts for absorption and scattering .
the column optical depth is the product of dust retrievals when the radiances are obtained by nadir - viewing instruments .
profiles of extinction opacity can be derived from radiances measured by limb - viewing instruments , providing important information on the vertical extension of the dust .
other important properties related to dust , such as the size distribution and the optical parameters , are more difficult to retrieve from direct measurements of radiances @xcite .
the knowledge of the spatio - temporal distribution of dust is of primary importance to produce quantitative estimates of dust mass mixing ratios , and calculate the atmospheric heating rates due to absorption and scattering of solar and infrared ( ir ) radiation by airborne particles .
these calculations are the basis for describing the thermal forcing in mars atmospheric models , and producing accurate predictions of the atmospheric state .
the choice of scenarios ( i.e. spatio - temporal distributions ) of dust optical depth has a significant impact on martian global climate model ( gcm ) simulations .
model studies have often been carried out with analytical specifications of dust distributions , both in the horizontal and in the vertical @xcite .
more recently , modeling groups have been carrying out simulations with more realistic horizontal dust distributions , tied to tes observations @xcite .
we propose in this paper to create a well - documented set of dust scenarios to be used in model experiments .
although we focus our attention on the column - integrated dust distribution , it should be noted that the spatial variation of the vertical distribution of dust also plays a significant role in thermal response ( see e.g. * ? ? ?
there are a number of approaches to represent this vertical structure , which is beyond the scope of this paper . to date
, there exist several datasets of retrieved column dust optical depth ( cdod ) for mars , spanning more than 20 martian years ( my ) since the mariner era .
these datasets are highly heterogeneous , as they have been created using data from different instruments having different geometric views , spatial and temporal coverage , as well as different observing wavelengths .
nonetheless , we show in this paper that at least some of these datasets can be appropriately used to quantitatively reconstruct the recent dust climatology on mars , and characterize the variability over many seasonal cycles .
this paper seeks to produce a continuous , multiannual climatology of cdod from early march 1999 ( solar longitude @xmath0 in my 24 ) to the end of july 2013 ( @xmath1 in my 31 ) . during this period of time , the thermal emission spectrometer ( tes , *
* ) aboard mgs , the thermal emission imaging system ( themis , * ? ? ?
* ) on ody , and the mars climate sounder ( mcs , * ? ? ?
* ) on mro provided global coverage of radiance observations at ir wavelengths , from which @xcite obtained direct retrievals of cdod or estimates of this quantity from the integrated extinction profiles . while images from orbiting spacecraft can provide information over large areas on the planet at any given time , observations of ir radiances from orbiting instruments have a very discrete coverage in longitude and local time due to the choice of orbit geometry .
mgs and mro , for instance , have sun - synchronous , nearly 2-hour polar orbits , which provide good latitude coverage but only sample about a dozen longitudes , usually at close to two fixed local times of day except when crossing the poles ( dust can often only be retrieved in the daytime , restricting further to one local time ) . because dust storms on mars have a wide range of spatial and temporal scales @xcite , these discrete observations can affect the space - time representation of dust storm activity .
extrapolating the data collected along orbit tracks to a broader range of local time and longitude introduces even more biases .
synoptic maps produced using simple average binning may alter the representation of rapidly evolving dust distribution .
we have developed a gridding methodology that is specifically adapted to heterogeneous observations , and to the discrete longitudinal / temporal coverage typical of spacecraft data acquisition .
the objective is to produce daily , regularly gridded maps of absorption cdod at 9.3 @xmath2 for several consecutive martian years ( dust opacity in absorption is less dependent on the particle size than in extinction ) . to achieve the goal ,
we have adopted a two - step procedure .
the first step consists in calculating iteratively averages and standard deviations of observations on a regularly spaced but likely incomplete spatio - temporal grid , after having binned the data using time windows of different size , and applied appropriate weighting functions in space , time , and retrieval uncertainty at each iteration .
we have established acceptance criteria for the gridded values , based on the number of reliable observations in each bin .
we have used this first product to statistically study the dust variability over almost eight complete martian years .
the second step consists in producing regular , daily maps of cdod by spatially interpolating and/or extrapolating the irregularly gridded maps , using a kriging method ( see e.g. * ? ? ?
* for a general introduction on the technique ) .
this multiannual series of synoptic maps of cdod is used for the dust scenarios in the gcm simulations that produce the current version of the mars climate database ( mcd version 5.1 , * ? ? ?
we provide open access to both the irregularly gridded and regularly kriged datasets , to foster scientific analyses and applications of the long - term martian dust climatology .
the most up - to - date version of these products ( currently v1.5 ) can be downloaded in the form of netcdf files from the mcd project website , hosted by the laboratoire de mtorologie dynamique ( lmd ) .
we also provide atlases of these maps as supplementary material of this publication .
the outline of this paper is the following .
section [ instruments ] describes the instruments and data we have used to produce the dust climatology .
we provide details of the time coverage , data quality control , processing , and uncertainties for each instrument . in section [ gridding ] we introduce the iterative weighted binning methodology we have adopted to create irregularly gridded ( but regularly spaced ) cdod maps from the spacecraft observations .
section [ validation ] is devoted to discuss the internal validation of the gridded maps , as well as the validation with independent observations .
we report in section [ analysis ] the statistical analysis of the dust climatology , in relation to the intraseasonal and interannual variabilities . in section [ kriging ]
we discuss the assumptions we have made and the kriging technique we have applied , with the purpose of producing regular and complete maps to be used as dust scenarios .
a summary is outlined in section [ summary ] , including a discussion on future developments .
mars global surveyor started its science mapping phase in march 1999 ( @xmath0 , my 24 ) , after a lengthy period of aerobraking . only about six months of global observations
are available in my 23 during the aerobraking , when the spacecraft orbit was still very elliptical ( 6 hour orbit instead of the nominal 2 hour mapping phase orbit ) .
mgs stopped working properly in november 2006 , and the mission was officially ended in january 2007 .
mars odyssey started its mapping phase in february 2002 ( @xmath3 , my 25 ) and is still working at the time of writing .
it is the longest running spacecraft orbiting an extra - terrestrial planet to date .
mars reconnaissance orbiter started its mapping phase in november 2006 ( @xmath4 , my 28 ) and , although it encountered a few problems that kept it in safe mode for an extended period at the end of 2009 , is currently operating nominally .
the mars exploration rovers `` spirit '' and `` opportunity '' started their missions on mars respectively on 4th january 2004 and 25th january 2004 ( @xmath5 and @xmath6 , my 26 ) .
spirit ceased communications with earth in march 2010 ( @xmath7 , my 30 ) , whereas opportunity is still active on the surface of the planet .
for building the dust climatology described in this paper , we have used cdod retrievals and estimates obtained from the following instruments observing at ir wavelengths : * tes aboard mgs , from @xmath8 in my 24 to @xmath9 in my 27 .
after this date , the number and quality of tes observations rapidly decreased ( august 2004 ) .
cdod retrievals @xcite are in absorption ( scattering is not modeled ) from nadir observations at wavelengths centered around 1075 @xmath10 ( 9.3 @xmath2 ) .
local times are narrowly centered around 2 pm at most latitudes , except when the orbit crosses high latitudes .
* themis aboard ody , from @xmath11 in my 26 ( we did not use observations taken at the end of my 25 ) to @xmath1 in my 31 .
the instrument is still working at the time of writing .
cdod retrievals @xcite are in absorption from nadir observations at wavelengths centered around 1075 @xmath10 ( 9.3 @xmath2 ) as for tes .
local times are between 2.30pm and 6 pm at most equatorial and mid - latitudes ( except for latitudes polewards of @xmath12 ) . *
mcs aboard mro , from @xmath13 in my 28 to @xmath1 in my 31 .
mcs was switched on in september 2006 before the official beginning of mro primary science phase , but had a long period in my 28 ( between 9th february and 14th june 2007 ) during which a mechanical problem preventing the use of its elevation actuator forced the team to keep it in limb - staring mode . the event precluded any nadir or off - nadir observation during this period , with the effect of limiting the vertical extension of the retrieved profiles ( including those of extinction opacity ) in the lower part of the atmosphere . since 9th october 2007 , off - nadir measurements with surface incidence angles between about @xmath14 and @xmath15 have resumed with nearly every limb sequence , but for retrievals of aerosol extinction opacities only limb views are currently used .
@xcite obtained profiles of dust extinction opacity from limb observations at wavelengths centered around 463 @xmath10 ( 21.6 @xmath2 ) as standard mcs product . in this paper
we have used estimates of cdod from the limb observations , as detailed in section [ qualitymcs ] .
local times are centered around 3 am and 3 pm at most latitudes , except when the orbit crosses the polar region . since 13th september 2010 ( @xmath16 , my 30 ) mcs has also been able to observe cross - track , thus providing information in a range of local times at selected positions during mro orbits @xcite .
we include these cross - track observations in our gridding , when available . for the validation of the gridded maps with independent observations , we have used retrievals of near - ir and ir cdod from the following instruments : * pancam cameras @xcite aboard mer - a `` spirit '' and mer - b `` opportunity '' , respectively from @xmath5 and @xmath6 in my 26 .
spirit pancam stopped providing measurements after @xmath7 in my 30 .
@xcite have retrieved cdod from upward - looking nadir observations at wavelengths centred around 880 nm ( near ir ) and 440 nm ( visible blue ) .
in this work we use only retrievals at 880 nm .
there are no significant differences between the values provided at the two wavelengths . * mini - tes aboard mer `` spirit '' and `` opportunity '' @xcite , starting from the same times as pancam cameras until @xmath17 ( spirit ) and @xmath18 ( opportunity ) in my 28 .
after @xmath18 the detectors were covered by dust from the my 28 planet - encircling dust storm and became unreliable .
@xcite have retrieved cdod from upward - looking nadir observations at ir wavelengths centered around 1075 @xmath10 ( 9.3 @xmath2 ) .
* crism @xcite aboard mro , from @xmath19 in my 28 until @xmath1 in my 30 .
cdod retrievals have been obtained from nadir observations at wavelengths centred around 900 nm @xcite .
although there are crism observations available after @xmath1 in my 30 , and crism is still working at the time of writing , we are not using these observations in the present work because an issue of truncated epfs in my 31 requires further investigation .
figure [ fig_timeline ] shows a summary of the time periods for which observations from the above mentioned instruments are available , together with the time limits within which we have used them in this paper .
figure [ fig_timeline ] in addition to cdod data , we have also used visible wide - angle pictures from mgs / moc @xcite , and visible pictures from mro / marci @xcite .
the main purpose is to compare the evolution of selected dust storms in the reconstructed cdod maps to the daily evolution that can be appreciated in camera images of the martian surface . before using the cdod datasets retrieved from the different instruments , we have firstly checked the values against quality control criteria to eliminate unreliable data . for all datasets ,
negative values are only allowed if the sum of the value and the associated uncertainty ( see section [ uncertainty ] ) is non - negative .
although a single negative measurement ( in this specific case a ` measurement ' corresponds to a ` retrieval ' ) can be considered unphysical , the weight of negative measurements must be taken into account in statistical calculations to prevent biases towards statistically non - zero values .
it is anticipated that both the gridded values described in section [ gridding ] and the interpolated ones described in section [ kriging ] are only accepted if they are positive , otherwise they are replaced by a minimum positive value of 0.02 . for specific datasets ,
we have adopted the following quality control criteria .
we show in figures [ fig_retrievalnumberday ] and [ fig_retrievalnumbernight ] the number of cdod retrievals used in the present work . for tes
, we have retained cdod observations flagged as `` good quality '' by the tes team , which passed additional quality control criteria .
specifically , we require that the surface temperature is greater than 220 k @xcite , the difference between surface temperature and maximum atmospheric temperature is greater than 5 k , the radiance fit residual is lower than 20 , the opacity of the carbon dioxide hot bands is between -0.01 and 0.05 , and the water ice opacity is greater than -0.05 ( the negative value threshold is constrained by the uncertainty ) .
the surface temperature threshold insures a good signal - to - noise ratio , whereas the threshold on the temperature difference between surface and atmosphere prevents unreliable retrievals when the temperature contrast is too small .
the latter case is particularly valid during high dust loading conditions , as showed by results from the geophysical fluid dynamics laboratory gcm ( gfdl - gcm ) and comparison with moc images during the 2001 planet - encircling dust storm @xcite . with the choice of these quality control criteria , retrievals
are practically limited to daytime conditions over ice - free surfaces , and may be reduced during intense dust storms ( see figs [ fig_retrievalnumberday ] and [ fig_retrievalnumbernight ] ) .
themis cdod retrievals include several `` framelets '' in the same `` image '' , i.e. several values within the same stripe of observations .
only the last framelet of each image is properly calibrated , but in this study we have considered all available framelets to increase the number of themis observations , especially in my 27 when we use no other observations .
we associate a slightly larger uncertainty to non - calibrated framelets , as explained in section [ uncertainty ] . for quality control purposes ,
we have retained cdod values when the rms residual from the aerosol opacity fit is lower than 0.4 , and the surface temperature is greater than 210 k. mcs observes the atmosphere in limb and off - nadir modes , which has the advantage of allowing the retrieval of vertical profiles of dust opacity , but has the disadvantage of not being able to observe the first few kilometres above the ground ( depending on the angle of observation ) . in order to estimate the column optical depth for aerosol dust
, the mcs team has integrated the full profile of dust extinction opacity produced by the retrieval algorithm ( v4.0 is the version used in this work ) after a successful retrieval . the profile is extended upward and downward under the assumption of well mixed dust , based on the last valid value .
the standard retrieved dust profile is often truncated when regions have measured radiances that are not fit by the forward radiative model within defined thresholds in dust , water ice or temperature . in the upper levels ,
the minimum extinction coefficient threshold for dust is @xmath20 km@xmath21 . in the lower levels ,
the profiles can get saturated by high opacity values , exceeding the threshold of @xmath22 km@xmath21 @xcite .
column optical depth estimates , on the other hand , use non - truncated profiles to make sure all available and reasonable information is taken into account .
the mcs dust profiles can not be retrieved down to the surface , and the dust in the un - retrieved part of the profile can account for a significant fraction of the total dust column .
estimates of cdod from mcs observations , therefore , are likely to introduce errors attributable to either the extrapolation to the surface under the well mixed assumption or the use of dust opacity values at altitudes where the fit to observed radiances is not within the standard threshold
. this problem may be particularly acute in light of evidence of elevated dust layers @xcite .
mcs cdod estimates can be fairly inaccurate if the lowest retrieved level of dust opacity is above @xmath23 km altitude , depending on the time of the year and the dust / water ice conditions .
we have therefore retained cdods that correspond to dust extinction profiles with valid values at least at or below 25 km altitude .
dayside retrievals are particularly affected by low altitude aereosols . to prevent large extrapolation of dust extinction coefficient to the surface during daytime observations ,
we only accept estimated daytime cdods when the corresponding extinction profile has valid values below 8 km altitude .
furthermore , we have rejected cdod estimates when the temperature profile dropped below the condensation temperature of carbon dioxide at some pressure levels , because co@xmath24 ice opacity can affect retrievals of dust opacity at those levels ( carbon dioxide ice opacity is not currently taken into account in mcs retrievals ) .
it is worth mentioning that this selection might introduce a systematic bias , especially during the northern spring and summer seasons in the tropics .
tes provides daytime cdods while mcs estimates of cdod in this season are largely based on nighttime viewing , because daytime clouds in the aphelion tropical belt limit the success of retrievals .
figure [ fig_retrievalnumberday ] figure [ fig_retrievalnumbernight ] the cdods from tes / themis and mcs are retrieved at different ir wavelengths , in absorption for tes / themis and in extinction for mcs , which prevents the direct comparison .
we account for this problem by converting the estimated mcs extinction cdods at 21.6 @xmath2 into equivalent absorption cdods at 9.3 @xmath2 .
we carry out the conversion by multiplying the mcs values by a factor 2.7 , which takes into account the effective dust particle radius ( 1.06 @xmath2 ) and variance ( 0.3 ) used in the retrievals ( version 4.0 , * ? ? ?
* ; * ? ? ?
it is worth mentioning that the dust size distribution is maintained fixed throughtout all seasons and locations .
we use cdod from pancam cameras aboard mer - a and mer - b rovers at the near - ir wavelength of 880 nm .
crism cdod retrievals are obtained at the near - ir wavelength of 900 nm .
when using pancam and crism for validating the gridded maps in section [ validation ] , we have to account for the differences between the absorption - only gridded values at 9.3 @xmath2 and the full extinction near - ir retrievals .
this is done by converting the gridded values into equivalent visible values .
the process firstly converts values from absorption - only at 9.3 @xmath2 to full - extinction values by multiplying by a factor 1.3 @xcite , then converts the latter to mean visible values by multiplying by a factor 2.0 , consistent with 1.5- to 2.0-@xmath2 dust particles @xcite .
overall , the factor we use to convert from absorption - only at 9.3 @xmath2 to equivalent full - extinction visible is 2.6 .
this factor , though , is affected by large errors , deriving from both the conversion to extinction @xcite , and the infrared - to - visible conversion @xcite , see also section [ uncertainty ] .
this is the reason why we have decided to provide our gridded products at the original ir wavelengths , despite the advantage of producing equivalent visible cdod maps for global and mesoscale atmospheric models , whose radiation schemes usually compute dust heating rates based on mean visible opacities using assumed ir / visible ratios . in order to calculate spatial averages
, we require that cdods are normalized to a reference pressure level , hence eliminating the effect of topographic inhomogeneities .
the retrieved values are divided by the surface pressure at the appropriate location and time of year , extracted from the pres0 tool included in the mcd v4.3 @xcite , and multiplied by 610 pa , which is our choice for the reference pressure level .
the cdods used in the present study , therefore , are equivalent to integrate or extrapolate the optical depth down to 610 pa at all locations on the planet , under the assumption of well - mixed dust .
see section [ uncertainty ] for a discussion of uncertainties associated to surface pressure . we have estimated the uncertainty of each single cdod value that we have used in this work .
the total uncertainty of an observation is provided as the propagation of single uncertainties detailed in this section , for each instrument .
the error propagation is carried out using relative errors , which highly simplify the formulas .
the squared relative propagated error reduces to the sum of squared relative errors on each independent variable when operations among variables include only multiplications and divisions , as in our case .
an uncertainty that is common to all observations is associated to the value of the surface pressure extracted from the mcd v4.3 , which is used to extrapolate cdods to a specific pressure level .
after the corrections for topography and total mass of the atmosphere , which are carried out within the pres0 routine of the mcd , the largest sources of surface pressure variability on mars are the weather systems , mainly in the form of dust storms and high latitude winter baroclinic waves . in order to account for this variability
, we have extracted the day - to - day rms of surface pressure from the mcd ( see the mcd detailed design document associated to the version of the database described in * ? ? ?
* ) to build a @xmath25 solar longitude @xmath26 latitude array .
we provide an estimate of the surface pressure uncertainty associated to the specific location and season of an observation by interpolating this array .
retrievals of cdod from tes and themis observations come with an associated nominal uncertainty , estimated by taking into account random errors in the instrument and calibration , as well as possible systematic errors in the retrieval algorithms .
@xcite gives an estimate for the total uncertainty in tes absorption - only ir dust optical depth for a single retrieval of 0.05 or 10% of the column optical depth , whichever is larger .
@xcite provides this value for themis absorption - only ir dust optical depth for the calibrated framelet : 0.04 or 10% of the column optical depth , whichever is larger . for themis observations
, @xcite also states that `` uncertainties are likely somewhat higher ( perhaps 20% or even higher ) during the most intense dust storms because large corrections to the temperature profile must be made for those observations '' .
it is likely that tes retrievals during high dust loading are also affected by larger uncertainties , due to the reduced surface - atmosphere temperature contrast and uncertainty in dust size distribution . in order to account for this ,
for tes we use the nominal uncertainty up to ir optical depth 1.0 , then we increase the uncertainty to 20% up to ir optical depth 2.0 , and to 30% for ir optical depths greater than 2.0 . for themis
, we use the nominal uncertainty up to ir optical depth 0.5 , we increase the uncertainty to 20% up to ir optical depth 2.0 , and to 30% for ir optical depths greater than 2.0 . as for the non - calibrated themis framelets , we currently increase their uncertainties by 20% ( e.g. 0.06 instead of 0.05 when the cdod value is 0.5 ) as explained in section [ quality ] ,
the procedure adopted to estimate cdod from mcs observations is likely to introduce errors , particularly when the truncated dust profiles do not have valid values close to the surface . estimating these errors is particularly difficult , as crucial information may be missing in the atmospheric layers where the dust loading is significant .
our estimate of the uncertainty on mcs cdods , therefore , can only be empirical . in section [ summary ] , we mention future developments in relation to mcs cdod retrievals . in this work ,
we assign an uncertainty on the basis of the altitude of the lowest valid level in the dust profile used for the vertical integration .
the uncertainty is linearly increasing from 5% ( nominal value ) to 60% as a function of this altitude , from the surface to the highest accepted level of 25 km .
although this uncertainty estimate is arbitrary , it provides a way to distinguish mcs values that are likely to be biased , and to relatively evaluate this bias .
this feature is used during the gridding procedure , as detailed in section [ gridding ] .
very small values of cdod estimated from dust extinction profiles that do not have valid values below 4 km are considered spurious and substituted with the minimum value of 0.01 , with 10% uncertainty ( i.e. 0.001 ) .
finally , we consider that the 2.7 factor used to convert mcs extinction cdods at 21.6 @xmath2 to equivalent absorption cdods at 9.3 @xmath2 is affected by 10% uncertainty ( i.e. 0.3 ) , due to the uncertainty on the particle sizes .
we take it into account in the propagation of uncertainties for mcs .
we have averaged cdod retrievals from pancam and mini - tes observations in each available sol for both mer - spirit and mer - opportunity .
we have also interpolated in time to fill the gaps in the sol - averaged time series for both mers . for sol - averaged pancam and mini - tes cdods , we use the largest value between the standard deviation and the average of single measurement errors as a measure of uncertainty on the daily mean .
this choice takes into account cdod variations with local time as well as single measurement errors .
as explained in section [ dproc ] , we have decided to work at the original tes and themis absorption ir wavelength of 9.3 @xmath2 , except when comparing with pancam and crism retrievals .
the conversion factors from absorption to extinction and from ir to visible depend on the aerosol refractive indexes , which ultimately depend on the aerosol size , shape and composition , and are likely to vary with seasons and locations .
both factors are therefore affected by large uncertainties .
@xcite shows in fig .
4a a graph of the relationship between effective absorption and full extinction optical depth for dust at the equator at northern autumn equinox in my 24 ( it is stated that the curve is typical of other times and locations ) . for ir dust
optical depths lower than 0.5 , the relationship is fairly linear ( @xmath27 ) .
the departure from the linear relationship at higher dust optical depth can be estimated as @xmath28 .
this estimate could be used to introduce an uncertainty on the absorption - to - extinction factor , which increases with the optical depth . as for the infrared - to - visible factor
, uncertainties have been estimated in a few measurement campaigns , although limited in time and location .
@xcite reports a value of @xmath29 from mer - pancam observations , @xcite reports in table 2 dust optical depth values from mer - pancam and mini - tes , which can be averaged to provide a value of @xmath30 . in section [ validation ] , we have simply used a single factor 2.6 ( @xmath31 ) throughout the time series to convert the gridded values to equivalent visible values , without explicitly taking the uncertainty on this factor into account .
the process of creating uniformly - spaced data ( i.e. a regular grid ) from irregularly - spaced ( scattered ) data is generally known as ` gridding ' .
there exist several techniques to solve this problem , depending on the applications .
a basic technique commonly adopted when dealing with orbital spacecraft observations is ` binning ' using space - time box averages .
the original data values which fall in a given multi - dimensional interval ( a ` bin ' ) are replaced by a value representative of that interval , often the average , but the median can also be used to filter outliers .
if the bins are uniformly distributed , the problem of gridding the data is solved . for our purposes ,
the simple box - average binning is not suitable because 1 ) we desire to achieve the highest possible spatio - temporal resolution for the cdod maps , compatibly with having a reasonable number of observations to grid , and 2 ) simple box averages introduce temporal and spatial biases if long time intervals and/or large spatial boxes are used ( several orbits at different times are inevitably averaged together ) .
reconstructing the dust climatology while preserving the desired variability of dust storms at short time scale and small spatial scale , therefore , requires a more sophisticated data gridding . in order to achieve this objective ,
we have developed an efficient ` iterative weighted binning ' methodology , described in the following section .
the application of our gridding procedure is equivalent to a moving weighted average characterized by the use of successive spatio - temporal windows .
the basic principle of the methodology is that , for a given grid point at a given time , all observations within a defined time window and spatial range are averaged using weights that depend on 1 ) the lag between the time of each observation and the required time at the grid point , 2 ) the spatial distance between each observation location and the location of the grid point , and 3 ) the uncertainty ( standard deviation ) of each observation . in this work ,
we use the mars universal time ( mut ) of the prime meridian as absolute time variable to which we refer both the time of observations and the time of grid points on a synoptic map .
the weighted average value @xmath32 at a given time @xmath33 and location @xmath34 for a generic observed variable @xmath35 , which is a function of time and spatial coordinates , is given by : @xmath36 where @xmath37 represents the index of the @xmath38 observations @xmath39 that are included in defined time window and space range ( a multi - dimensional bin built around the time @xmath33 and the location @xmath40 ) .
@xmath41 is the time weighting function , which determines the contribution of the @xmath42 observation according to the time difference @xmath43 ( positive for observations in the future , negative for observations in the past ) .
@xmath44 is the distance weighting function , which determines the contribution of the @xmath42 observation according to its distance @xmath45 from the location @xmath40 .
the distance weight is also a function of the time lag @xmath46 so that observations at different times in the past or in the future with respect to @xmath33 have different distance weights .
@xmath47 is the uncertainty weight , which determines the contribution of the @xmath42 observation according to its relative uncertainty ( @xmath48 is the observation standard deviation ) .
the particular choice for the form of the weighting functions @xmath44 , @xmath41 , and @xmath47 depends on the specific application . for our purpose of creating synoptic maps of cdod using orbiting satellite observations , it is convenient to use weighting functions similar to those implemented in the analysis correction data assimilation scheme of the uk meteorological office @xcite and in the derived mars analysis correction data assimilation scheme @xcite .
in fact , eq . [ eq1 ] applied to @xmath49 grid points of a regular grid can be considered as the weighted average equivalent of the model grid point increment eq . 3.18 in the analysis correction scheme of @xcite , although this analogy has no formal basis .
[ eq1 ] applies to observations in specifically defined time window and space range .
our methodology includes the iterative use of less and less restrictive time windows and space ranges , as we explain in the next section where we describe the application of iwb to the problem of gridding dust opacity observations . if we apply eq .
[ eq1 ] to each grid point @xmath49 of a pre - defined space - time grid for the cdod variable @xmath50 , the weighted average value @xmath51 at each grid point can be written as : @xmath52 where @xmath53 are the cdod observations that are included in a specified time window and distance range built around the time and location of the grid point @xmath49 .
additionally , one can define the weighted root mean squared deviation ( or weighted standard deviation ) associated to the weighted average value @xmath54 as : @xmath55 we choose @xmath44 to be a second - order autoregressive correlation function of the distance @xmath56 from observation @xmath37 to grid point @xmath49 and of the correlation scale @xmath57 ( see also eq . 3.19 in * ?
* ) : @xmath58 where @xmath59 is a function of the time difference between observation time and grid point time .
@xmath57 is chosen to be a linearly increasing function of this difference , symmetric with respect to @xmath60 , with a minimum value @xmath61 at @xmath60 and a maximum value @xmath62 at the extrema of the chosen time window ( tw ) .
the equation for @xmath57 ( defined over the range @xmath63 ) is therefore provided by : @xmath64 with this choice of @xmath57 , observations with times closer to the time of the requested grid point have weights @xmath65 that decrease faster with distance ( see fig . [ fig_weights ] ) . in other words ,
smaller weights are assigned to distant observations when their time is closer to the current time .
a further parameter , @xmath66 , sets the distance range on the spherical surface within which the contributions of the observations to the average are considered .
we use the haversine formula to calculate the distance between two locations on the spherical planetary surface , which is numerically better conditioned for small distances @xcite .
@xmath41 is chosen to be a decreasing quadratic function of the difference @xmath67 between observation time and grid point time @xcite , symmetric with respect to @xmath60 .
the equation for @xmath41 ( defined over the range @xmath63 ) is therefore : @xmath68 where @xmath69 is the minimum value at the extrema of the chosen time window ( see fig . [ fig_weights ] for an example with @xmath70 ) .
finally , @xmath47 is chosen as a second - order autoregressive correlation function of the relative uncertainty of observation , namely : @xmath71 where @xmath72 is the uncertainty ( standard deviation ) of each observation , and @xmath73 is a scaling factor used to obtain the desired width at half maximum ( whm ) for the @xmath47 function .
we choose @xmath74 to have whm=0.2
( i.e. @xmath75 when the relative uncertainty of an observation is @xmath76 ) .
see fig .
[ fig_weights ] for a plot of the @xmath47 function we use .
figure [ fig_weights ] the following is a summary of our gridding procedure with iwb . 1 .
we have tested the sensitivity of the method to different spatial and temporal resolutions and the results are reported in section [ bingridex ] . on the basis of the sensitivity tests , we have chosen the resolutions of the longitude - latitude grids reported in table [ tab1 ] , according to different datasets .
the choice of the time resolution is one sol .
each synoptic gridded map is centred around local time noon at the prime meridian , i.e. a sol at @xmath77 longitude is defined between 00:00 and 24:00 mut .
for my 24 and 25 we have used only tes cdods . for my 26 and 27
we have used both tes and themis observations until @xmath78 in my 27 , then only themis observations . for my 28 ,
we have used only themis until @xmath79 , then both mcs and themis , as well as in most of my 29 , 30 , and 31 , apart from @xmath80 , my 29 , to @xmath81 , my 30 , when only themis observations are available .
3 . for each set of observations ( tes , tes+themis , themis , mcs+themis )
we have defined the parameters required by eq .
[ eq2 ] , which are summarized in table [ tab1 ] .
the iwb procedure then requires that , for each sol and for each spatial grid point , we define a time window and distance range within which we use observations to calculate the weighted average and its associated weighted standard deviation .
we repeat the previous step iteratively , using different time windows , from the smallest to the largest ( the values we use are reported in table [ tab1 ] ) , and calculating the weights accordingly . at each iteration
, more valid grid points are added to the synoptic map , because more observations are considered , but we do not overwrite grid points flagged as valid in previous iterations .
the criterion to accept a value of weighted average at a particular grid point is that there must be at least a minimum number of observations @xmath82 within a distance @xmath83 from the grid point , having relative uncertainties lower than a threshold value @xmath84 ( see values in table [ tab1 ] ) .
if this is not the case , a missing value is assigned to the grid point at that iteration .
the final result of the successive application of the weighted binning equation consists in daily synoptic maps of cdod on the pre - defined regular grid , with missing values in places where the gridding criterion was not satisfied at any iteration .
7 . for the specific years my 27 to 31
, we apply an additional step in the procedure by substituting some of the missing values with valid gridded values obtained with only themis observations using the parameters listed in table [ tab1 ] .
this step is quite effective in reducing the gaps induced by the poor mcs coverage of major dust events and of the tropical regions affected by the aphelion cloud belt .
it provides full weight to themis observations with less restrictive gridding parameters , allowing for valid grid points which would otherwise not satisfy the acceptance criterion of the normal procedure .
table [ tab1 ] in order to illustrate the procedure of gridding when the observation coverage is good both in space and time ( at least for most latitudes ) , we use an example from tes observations in the northern winter of my 24 .
[ fig_tes_time ] highlights one of the main reasons why the application of weighted binning in space and time is essential when gridding at high spatio - temporal resolution . in this figure , tes retrievals of cdod
are shown within a time window of 7 sols ( about @xmath85 solar longitude at this season ) centered around noon at @xmath77 longitude in sol 449 , @xmath86 , my 24 . the area shown in the figure
is limited to @xmath87 longitude and @xmath88 latitude south of the equator .
the colours indicate the time lag between the time of an observation and the center time .
it is clear that adjacent orbits have quite different time lags .
a simple box average would certainly introduce biases in the creation of a synoptic map of cdod for this sol . figure [ fig_tes_time ] the result of the iwb with increasing time windows is shown in fig .
[ fig_tes_tw ] .
we separate each iteration ( tw=1 , 3 , 5 , and 7 sols ) in the panel series ( a ) and ( b ) , respectively showing the retrieved non - uniform values of tes cdod , and the uniformly gridded values .
clearly , gridded values of cdod with larger tws are smoothed with respect to smaller tws , even if time weighting is applied in both cases .
the result of the successive application of weighted binning is then shown in panel ( c ) , where the valid grid points in each iteration add to the valid values of the previous one , without overwriting .
eventually , we obtain a fairly complete , regular map of cdod where the observations closer to the mut time of the synoptic map provide most of the unbiased information .
this shows that an iterative weighted binning produces more valid and less biased values than the application of a single weighting with a fixed tw .
figure [ fig_tes_tw ] the optimal spatial resolution for the gridded maps can be considered as the highest one which includes a reasonable number of observations in each spatio - temporal bin and , at the same time , produces results not dissimilar to those obtained at a lower resolution .
ideally , the gridding method should not show much sensitivity with respect to the choice of reasonable spatial and temporal resolutions , particularly in the standard deviation field . in order to verify the quality of the choice we have made for the datasets reported in table [ tab1 ] ,
we have carried out a sensitivity test using four different cases , as illustrated in fig.[fig_my24_restests ] for a typical map with high dust optical depth contrasts ( same sol as in fig .
[ fig_tes_tw ] ) . both averaged values and standard deviations ( which in this case represent the variability within the group of averaged observations ) show little variation throughout the four cases .
this result illustrates ( at least for tes ) that our gridding procedure is not particularly sensitive to the choice of the spatial resolution within a limited range .
this allows us to choose for tes in my 24 , 25 and 26 fairly high longitude and latitude resolutions , namely @xmath89 .
the same latitude resolution can not be used for mcs , given the fact that the path of the limb observations usually spans several degrees in latitude , and there are fewer observations per degree than tes .
finally , the sparse distribution of themis observations must be taken into account in the choice of the spatial resolution , when only themis data are available . in order to maintain a consistent resolution throughout my 28 , 29 , 30 and 31 we use @xmath90 in longitude and latitude .
figure [ fig_my24_restests ] the combination of fig .
[ fig_tes_tw ] , [ fig_my24flush_evol ] , and [ moc_tes_my24 ] in section [ camval ] provides the basis for our choice of the 1-sol time resolution .
ideally , we want the highest time resolution to follow the development of regional dust storms , using as many independent observations as possible .
given the number of tes and mcs orbits per sol ( respectively 12 and 13 ) , there are not enough independent observations to provide a sufficient spatial resolution below the 1-sol time resolution .
the top panels of the ( a ) and ( b ) series in fig .
[ fig_tes_tw ] show that the main features of the regional storm occurring in my 24 at @xmath86 can be captured with a 1-sol tw gridding .
[ fig_my24flush_evol ] illustrates that , for the same period , we can characterize the daily evolution of the storm , and fig .
[ moc_tes_my24 ] demonstrates that such daily evolution can be followed in moc synoptic visible images , thus validating our gridded maps in this context .
tes has , nonetheless , the advantage of a good spatial coverage even in condition of high aerosol loading , which is often not the case for mcs . fig . [ fig_my29flush_evol ] shows the daily evolution of another regional storm occurring in my 29 at @xmath91 , which clearly highlights the large number of missing grid points .
as previously mentioned in section [ qualitymcs ] , mcs cdod estimates are few in numbers during episodes of high dust loading , and have large associated uncertainties because the retrieved dust extinction profile is missing in the dusty part of the atmosphere .
the criterion we use for the acceptance of grid points , based on a minimum number of observations within a threshold distance having relative uncertainty below a defined threshold , is particularly strict for mcs and themis observations .
the thresholds we have used in this work ( see table [ tab1 ] ) might be relaxed from my 28 onwards , especially the relative uncertainty threshold .
other ways to increase the number of valid grid points are discussed in sections [ camval ] and [ summary ] , and include the possibilities to estimate the cdods from observations of brightness temperatures , using a gcm , and to retrieve proper cdods from nadir and off - nadir mcs observations . figure [ fig_my24flush_evol ] figure [ fig_my29flush_evol ]
we have carried out a statistical validation of the incomplete maps of cdod based on two approaches : 1 . an internal validation comparing tes and mcs cdods to gridded cdods , interpolated in the locations and at the times of the observations ; 2 . an external validation using independent observations .
these include the values of visible cdod retrieved from crism observations , the time series of visible cdod retrieved from pancam cameras aboard spirit and opportunity , the time series of ir cdod retrieved from mini - tes aboard the two mers , and some moc and marci images during the evolution of regional storms . for the statistical internal validation
, we have interpolated the gridded maps ( linearly in time and bilinearly in space ) at the location and time of each observation , if a complete set of neighbours was available ( i.e. four adjacent spatial grid points at two consecutive sols ) .
we have interpolated values for both the gridded average field and the standard deviation one .
the first test we have carried out is a simple correlation test .
for all examined years , there is clearly a correlation between the observed values and the reconstructed ones , despite the uncertainties introduced by the linear interpolation .
the linear correlation is very good in my 25 , 28 , 29 , 30 and 31 , as the high values of the pearson correlation coefficient suggest ( @xmath92 ) , and the data points indeed accumulate around a straight line with slope close to 1 . for the other years
, the pearson correlation coefficient provides lower values ( @xmath93 ) , but still observed values and reconstructed ones appear visually correlated ( not shown here ) . for each available interpolated value
, we have then calculated the standardized mean difference ( smd ) between this value and the observed value , which is weighted using the combined standard deviation . under the approximation
that the observed value and the interpolated value are independent ( i.e. the covariance is neglected ) , the variable we calculate is : @xmath94 which is equivalent to expressing the difference between the two values in terms of their combined standard deviation . strictly speaking ,
the interpolated value is not independent of the observed value because the latter has been used to calculate the former , so the value of @xmath95 might be underestimated . nonetheless , this variable is useful for estimating whether clear biases are present in our gridded maps , or whether the differences are within statistical limits .
we display in figs .
[ fig_histo_grid ] the histograms of the smd for the tes ( my 24 , 25 , 26 ) and mcs datasets ( my 28 , 29 , 30 , 31 ) . the difference in the number of values from year to year reflects the number of available retrievals , as shown in figs .
[ fig_retrievalnumberday ] and [ fig_retrievalnumbernight ] .
all histograms show that most of the values have @xmath96 ( the difference is much lower than the combined standard deviation ) with @xmath97 , and peak values are very close to zero .
these general characteristics provide a sound internal validation from the statistical point of view .
other important factors to be considered are the shape of the histograms ( i.e. skeweness and kurtosis ) , which can highlight possible biases .
all distributions are leptokurtic ( i.e. more sharply peaked than a gaussian distribution ) , which indicates differences with respect to the combined standard deviation generally smaller than what expected by a random process .
mcs histograms are very symmetric ( small skeweness ) therefore no particular bias is evident .
tes histograms , on the other hand , are clearly right - skewed , which suggests that , in the gridded maps of my 24 , 25 and 26 , there are more largely overestimated cdods than largely underestimated ones .
the asymmetry of the smd distributions for the tes years is consistent with lower values of the correlation coefficient for my 24 and 26 .
the origin of such bias is mainly related to overestimated gridded cdods in correspondence to small observed ones .
figure [ fig_histo_grid ] a third diagnostic we use to internally validate our gridded maps is the relative standard deviation ( std ) of the grid points .
this is expressed by the ratio between the weighted std , calculated with eq .
[ eq2bis ] ( a measure of the variability of cdod at a grid point ) , and the weighted average calculated with eq .
[ fig_percrelerr ] shows that my 24 , 29 , 30 , and 31 have distributions peaked at values of relative std lower than 10% .
distributions in the other years peak at relative std lower than 20% , with my 26 and 27 being the worst .
all years have very right - skewed distributions , but the values of relative std of the grid points do not exceed much the values of relative uncertainty of the single retrievals , which can be larger than 30% , particularly in the mcs years .
[ fig_percrelerr ] , therefore , suggests reasonable values for the relative std of the gridded maps , and provides an indirect validation of the goodness of the chosen spatio - temporal resolution .
if the resolution was too coarse , in fact , one would expect large variability at most grid points , which is not the case here .
figure [ fig_percrelerr ] we used independent crism retrievals of visible cdod in my 28 , 29 , and 30 to compare with equivalent visible gridded values interpolated in the positions and at times of crism observations . as explained in section [ dproc
] , we have multiplied the gridded cdod by a factor 2.6 to estimate equivalent visible extinction cdods , and compare to crism values , retrieved at 900 nm wavelength . as in section
[ intval ] , we have calculated the pearson correlation coefficients ( @xmath98 in my 28 , @xmath99 in my 29,and @xmath100 in my 30 ) , and produced histograms of the smd ( fig . [ fig_histo_grid_crism ] ) for each available year .
the comparison with crism data produces less correlated values and more biased histograms than the internal validation . in particular ,
the smd distributions are clearly left - skewed , which suggests a majority of large underestimated values of cdod in the gridded maps with respect to those observed by crism .
it must be remembered , though , that uncertainties in the absorption - to - extinction and ir - to - visible factors would reflect in the skeweness of the smd distributions .
nevertheless , only factors that combine to provide much larger values than the chose 2.6 value could make the distributions unskewed ( not shown here ) .
it is therefore very difficult to draw conclusions from the validation with crism observations in the visible .
figure [ fig_histo_grid_crism ] we have also compared the gridded values , interpolated in the location of mer spirit and opportunity , to the cdod measured by pancam in the visible and mini - tes in the ir on the two rovers .
the time series of the comparison , for each sol when there are available observations , are shown in fig .
[ fig_mertimeseries ] .
we have chosen to display also my 24 , 25 , and 26 , even if mer observations are not available in those years , in order to increase the statistics on the interannual variability at meridiani planum and gusev crater .
as for the comparison with crism , we have multiplied the values extracted from the gridded maps and mini - tes values by a factor 2.6 , to convert them to a mean equivalent visible wavelength .
the comparison in gusev crater is quite satisfactory throughtout the time series , with mini - tes , pancam and the gridded maps generally agreeing within the uncertainties ( represented by the grey envelope ) , except for a couple of mini - tes peaks not seen by pancam as large . even in the non - dusty seasons ,
the time series show consistency in all years , with values of cdod between 0.2 and 0.3 on average .
the comparison in meridiani is , on the contrary , satisfactory during the dusty season but problematic during the non - dusty one , when it highlights a strong difference between the cdod measured from ground and the one gridded from satellite observations .
consistently in every year , the gridded maps provide values of cdod that are about a half of those observed by pancam and mini - tes during the period @xmath101 $ ] .
this bias seems to be present in all three satellite datasets considered in this work , even when looking at single cdod retrievals . @xcite
have recently provided a possible explanation for this bias . according to them ,
water ice clouds and hazes contributed to the observed opacity at the opportunity site in the summer season .
clouds were seen over the range @xmath102 $ ] , with peak activity near @xmath103 and @xmath104 .
when looking at the sun through the atmospheric column , pancam in meridiani is likely to add water ice optical depth to the dust optical depth , thus explaining the bias .
@xcite confirms that ice clouds and hazes were not a significant part of the opacity at the spirit site , which does not show biases with respect to observations from satellites .
satellite measurements and gridded maps , therefore , should be considered more reliable in meridiani at this time of the year . at other times , in fact , the differences between pancam cdod and gridded cdod are well within the uncertainties , even during the my 28 planet - encircling dust storm when only themis observations are available for the gridding .
when comparing the time series in gusev and meridiani , we can clearly observe and confirm from previous studies ( e.g. * ? ? ?
* ) that 1 ) the two planet - encircling dust storms ( my 25 and 28 ) had similar growth , peak values , and decay ; 2 ) there are many regional storms that affected both meridiani and gusev ( on the opposite side of mars ) , reaching comparable peak values ; 3 ) there is a tendency to have three distinct peaks of cdod every second half of the year , with an early peak around @xmath105 , a major peak around @xmath106 , and a late peak around @xmath107 .
the presence of three seasonal peaks of cdod is not only confined to the equatorial latitude band , as it is discussed in section [ analysis ] and shown in fig .
[ fig_zonalmean_grid ] .
figure [ fig_mertimeseries ] we have used moc and marci processed daily global maps @xcite to qualitatively compare in visible images the evolution of the two selected regional storms shown in section [ bingridex ] .
[ moc_tes_my24 ] shows the evolution of the flushing storm occurring in my 24 at @xmath108 , and fig .
[ marci_mcs_my29_3 ] shows the evolution of the regional storm occurring in my 29 at @xmath91 .
although the lack of strong contrast between the dust and the background in both moc and marci images makes it difficult to clearly visualize the presence of dust , the overlaid satellite observations help to recognize where a storm is ongoing . in my 24 , the contours of gridded cdod correlate remarkably well in position and shape with the presence of the dust haze in the moc images at several sols during the evolution of the regional storm .
this validation of the gridded maps during a particularly dynamic and fast - evolving event encourages their use for statistical analysis of the evolution of dust storms , particularly in my 24 , 25 and 26 .
figure [ moc_tes_my24 ] given the limited coverage of valid grid points in the maps of the my 29 regional dust storm ( see fig .
[ fig_my29flush_evol ] ) , it is not easy in this case to validate its evolution with the marci images .
nevertheless , it can be observed that the position of the dust haze in the marci images corresponds to the high values of cdod in the gridded maps at corresponding sols .
the overlaid satellite observations , in the case of the marci images , are surface temperature anomalies , which show a remarkable correlation with the dust haze .
@xcite devised a technique to estimate the cdod from satellite observations of surface temperature .
observed surface temperatures are actually top - of - atmosphere brightness temperatures in spectral regions where the atmosphere is relatively transparent .
the technique uses a gcm with dust transport capability ( the gfdl mgcm ) to find the best value of dust optical depth at any given time and location that allows to match simulated brightness temperatures to observed ones .
they applied this technique to estimate cdods during the my 25 planet - encircling dust storm using tes observations , and in principle the same technique can be applied to mcs observations of brightness temperature .
given the fact that brightness temperature observations are available even when proper cdod retrievals fail , the gridding procedure would certainly benefit from the inclusions of cdod data estimated as described above .
figure [ marci_mcs_my29_3 ]
the daily time series of cdod spanning eight martian years allows for analysis of intraseasonal and interannual variability .
we have shown that the evolution of individual regional dust storms can be followed when the gridded maps are fairly complete .
even when the number of valid grid points in the daily maps is low , the cdod variability can still be analysed statistically , provided data are averaged or filtered to prevent contamination from spurious values occurring at high frequencies ( e.g. `` on / off '' , isolated , large optical depths ) .
we provide an atlas of daily , irregularly gridded maps for each available martian year as supplementary material of this paper .
the reader can therefore explore the entire time series that forms the current version 1.5 of the cdod climatology obtained with iwb , before downloading the corresponding netcdf dataset from the lmd website ( at the address reported in the footnote of section [ intro ] ) .
the atlas is organised as 669 pages ( one for each sol ) with eight maps on each page corresponding to the available martian years .
see also [ calendar ] for a description of the sol - based martian calendar devised in the present work .
the structure of the atlas helps the analysis of interannual variability .
it is interesting , for instance , to look simultaneously at the cdod around sol 370 ( @xmath109 ) in my 24 , 25 , and 26 , a few sols before the beginning of the my 25 planet - encircling dust storm .
it appears clearly that the cdod distribution was very similar in the three years observed by tes at this season , until sol 372 ( @xmath110 ) , when slightly higher cdod values appeared in the hellas basin in my 25 , prior to expansion out of this crater towards the northern rim and in hesperia planum .
another feature that clearly stands out when looking at cdod at high southern latitudes at this season is that tes observed fairly high values near the polar cap edge , where dust is likely to be lifted by southern baroclinic waves , active at this time of the year , whereas no sign of such high values is present in the other years .
very few themis observations are available in my 27 and 28 , so it is difficult to judge these two years , but it is evident that mcs observations are biased towards very low cdod values at high southern latitudes . this bias of mcs observations can be observed even more clearly in the zonal means of gridded cdods , plotted as a function of solar longitude and latitude in each martian year ( fig .
[ fig_zonalmean_grid ] ) .
there is an evident dichotomy between the tes years , where the high northern and southern latitudes in northern spring and summer present increased optical depth values , and the mcs years , where values decrease towards high latitudes .
panel ( a ) of fig .
[ fig_stat_latallyears ] offers a striking summary of this dichotomy around @xmath111 .
the reason for this important bias is most likely the fact that dust is aloft only in the lower portion of the atmosphere ( within the first kilometers ) at this season and at these latitudes .
mcs limb observations are not able to scan through these low levels , therefore dust is missed by the radiometer , which detects only very small values of opacity above the low dust layer .
future retrievals of cdod from mcs in - planet observations , or estimates from top - of - atmosphere brightness temperature fit , might eventually correct this bias .
biases apart , the latitudinal , seasonal , and interannual variability of cdod can be fully appreciated in fig .
[ fig_zonalmean_grid ] , which summarizes eight years of dust climatology on mars .
this figure clearly highlights four distinctive phases in the distribution of dust during the second half of each year without a global - scale storm , and confirms what other studies have found , by using the longest available record of observations .
dust starts to increase in the atmosphere around northern autumn equinox in the southern hemisphere , but the largest increase usually occurs between @xmath112 and @xmath113 , when baroclinic activity at high northern latitudes favours cross - equatorial flushing storms ( although not all regional storms occurring at this time originated in the northern plains ) .
a third phase in the dust distribution is characterized by large lifting of dust occurring in the southern polar region between @xmath114 and @xmath115 , after the @xmath116 ice has mostly sublimated away . at other latitudes , instead
, there is a clear decrease of atmospheric dust in every martian year after @xmath117 ( with the exception of my 28 , characterized by the late planet - encircling storm ) .
this pause in large dust storms coincides with the decrease in the amplitude of low - altitude northern baroclinic waves ( the so - called solsticial pause , see e.g. * ? ? ?
when the solsticial pause is over , and the baroclinic wave activity at low altitude reinforces again , the probability of late flushing storms increases . every year , therefore , a fourth phase in the dust distribution starts around or after @xmath118 , producing a late peak of cdod .
figure [ fig_zonalmean_grid ] the fairly repeatable pattern of cdod from year to year contrasts with the highly unpredictable occurrence of global - scale dust storms , both in terms of frequency and season .
if global - scale storms did not occur on mars , and we did not take into account single regional storms , the `` typical '' annual distribution of cdod would look like the one in fig .
[ fig_dodtypyear ] , where we have averaged all years together , excluding the largest cdod value for each grid point . in this figure ,
the four distinct phases described above appear perfectly well defined .
obviously , also the mcs bias around the edges of the polar caps stands out in this figure .
figure [ fig_dodtypyear ] figs .
[ fig_stat_eq_allyears ] and [ fig_stat_latallyears ] are useful to summarize similarities and differences among the years . at equatorial latitudes ,
there is little interannual variability in northern spring / summer , and low cdod .
the gradient of cdod increases moving towards southern tropical latitudes and northern high latitudes in tes years , whereas in mcs years the bias mentioned above appears . in the second half of the year ,
the background dust level shows again little variability in the equatorial / tropical regions during the solsticial pause , except for the years with a global - scale storm .
large cdod gradients are again observed moving towards high latitudes in tes years , but not in mcs years . outside the solsticial pause ,
cross - equatorial flushing storms are likely to produce peaks of cdod in the equatorial region , which affect all longitudes ( as we have also seen in the comparison between meridiani and gusev ) . figure [ fig_stat_eq_allyears ] figure [ fig_stat_latallyears ] as an example of how our gridded products can be used to explore the multiannual statistics of dust loading in particular regions , we consider the case of esa s exomars 2016 entry , descending and landing demostrator module ( edm `` schiaparelli '' ) .
the landing is planned to take place in meridiani planum during the time window @xmath119 in my 33 , which is well inside the dust storm season .
although this will provide a unique opportunity to characterize a dust - loaded atmosphere during the entry , descending , and landing procedure , it will also pose strict contraints on the engineering parameters of the landing , and increase the associated risks .
it is therefore very important to produce an accurate statistical prediction of the expected range of dust loading at the time and location of the landing , based on historical records . if one plots the time series of cdod in all years in meridiani , as in fig .
[ fig_stat_eq_allyears ] , but limited to the period @xmath120 , the tail of the planetary - encircling dust storm stands out with respect to all other years , which show moderate dust loading ( not shown here ) .
it is worth noting that this season is characterized by the solsticial pause of the baroclinic wave activity , with associated lack of flushing storms , which in general have trajectories with potential to affect meridiani . by calculating the cumulative histogram for the landing window @xmath121 ( not shown here )
, we can make the statistical prediction , based on past observations , that exomars 2016 `` schiaparelli '' lander is likely to encounter a moderate dust loading with cdod of @xmath122 ( ir absorption referred to 610 pa ) .
this prediction , though , does not exclude the possibility of high dust loading induced by an equinoctial planet - encircling dust storm , which , as seen above , can be quite unpredictable even just a few sols beforehand .
with the application of the iwb procedure , the maps of cdod that we obtain are incomplete and present higher spatial resolutions in my 24 , 25 , 26 than in the other years . for many practical applications
it is desirable to have complete , regularly gridded maps spanning several years with the same resolution .
one of such application is to prescribe realistic aerosol dust distributions for global - scale or meso - scale climate model simulations .
the production of the mcd statistics using the lmd gcm , for instance , is an obvious example @xcite .
other possible applications include retrieving surface or atmospheric variables ( e.g. surface albedo or atmospheric water vapor ) using observations from which the aerosol dust component needs to be subtracted . for these reasons , in this section
we discuss our method to derive multiannual , regularly gridded ` dust scenarios ' from the irregularly gridded maps we have described so far .
the process of producing complete gridded maps at a given resolution from maps that have missing data and different resolutions is , generally speaking , a problem of interpolation and/or extrapolation .
there exist several techniques , more or less ` optimal ' , to solve this problem , as it is the case for the gridding problem .
kriging ( which is synonymous with ` optimal prediction ' ) is a technique that belongs to the family of linear least squares estimation algorithms .
it is a method of interpolation that predicts unknown values from data provided at known locations . unlike other common interpolation methods , such as polynomial , spline , and nearest - neighbor
, kriging does not require an exact fit at each tabulated data point .
another important difference of kriging and other linear estimation methods is that kriging aims to minimize the error variance of the predicted values .
kriging applies a weighting to each of the tabulated data points based on spatial variance and trends among the points .
weights are computed by combining calculations of the spatial structure and dependence of the data , and building a statistical model of their spatial correlation ( called ` semivariogram ' model ) .
alternatively , empirical semivariograms are often approximated by theoretical model functions , the most common of which are the spherical and the exponential semivariograms ( we use the latter in this work ) .
the reader can refer to @xcite for a general overview of the method . for
the application of kriging to data gridding , @xcite provide an example for the case of temperature ground stations on the earth , and @xcite evaluate kriging among other methods of spatial interpolation . given the spatial characteristics of the incomplete maps of cdod that we have obtained after gridding the observations with the iwb procedure , kriging is the interpolation method that is likely to provide the best results , producing smooth spatial variations even in cases when many missing values are present . before applying the kriging to the maps
, we have replaced some of the missing values using the following methods ( in order of application ) . 1 . in tes , themis and mcs datasets there are gaps in data coverage that sometime extend for few sols , e.g. during mars solar conjunctions . in those cases
, gridded maps are missing as well , if the gaps in data coverage are longer than 7 sols ( the maximum tw we use ) .
in addition , one or two daily maps before the first missing map and after the last missing map might have very few valid grid points in longitude . when long data gaps occur , we have increased the tw up to 25 sols , in order to accumulate enough observations to bypass the data coverage gap , and we have applied the gridding procedure again with more successive iterations .
this method produces a smooth time interpolation by combining the moving average and the time weighting . in the absence of alternative information , or of a dynamical model of dust transport , this is a reasonable way to interpolate maps during periods of missing data .
issues of data coverage are both temporal and spatial .
tes and themis datasets have few valid retrievals of cdod at high latitudes during the winter seasons , and the mcs dataset does not include estimates when the dust loading is too high ( e.g. during dust storms ) , the water ice opacity is large ( e.g. during the aphelion cloud tropical belt season ) , or the temperature at some height is below the @xmath116 condensation temperature ( e.g. during the polar nights , when @xmath116 clouds might form ) . in order to fill most of the spatial gaps in our gridded maps with climatological values , for each sol of the martian year we have used the average of all eight years , excluding from the average the largest value of cdod to eliminate single dust storms . for this climatological year ( see fig .
[ fig_dodtypyear ] for a plot of its zonal mean ) , spatial inhomogeneities are smoothed out , and the optical depths that result are generally underestimated , particularly in the second half of the year . before using these 669 climatological maps to fill out gaps in the original gridded maps , therefore , we have to re - normalize the values of cdod using anchoring values from independent observations .
for each sol from my 27 to 31 we can use the observations of optical depth from pancam ` spirit ' and pancam ` opportunity ' as anchoring value .
we actually use the minimum between the sol - averaged observations from spirit and opportunity , after having interpolated the time series to fill any gap .
the use of the minimum value avoids re - normalizing the climatological maps with values characteristic of specific dust storms located at either spirit s or opportunity s site .
furthermore , it is consistent with satellite observations in northern summer ( see section [ validation ] ) , so that we avoid biases in the re - normalization . for a given sol
, we calculate the average of the climatological cdod map @xmath123 in a latitude band [ @xmath124,@xmath77 ] ( we recall that the locations of spirit and opportunity are , respectively , close to @xmath126 and @xmath127 latitude ) , and we re - normalize the values of the map using the following factor @xmath128 , weighted in latitude @xmath129 : + [ eq6 ] = r_+ ( 1- ) & if @xmath130 + r_+ ( 1 + ) & if @xmath131 + where @xmath132 is the ratio between the minimum mer dust optical depth @xmath133 ( converted to equivalent ir absorption by dividing by 2.6 ) and the calculated average @xmath123 .
we replace a missing grid point in a gridded map from my 27 to 31 with a value obtained as described above , if the distance between the missing grid point and the closest valid grid point is greater than a certain threshold distance , set equal to 1000 km .
the threshold avoids the introduction of possible artificial discontinuities between valid gridded values and filled values in the maps .
no valid grid points exist at very high latitudes during the polar winters in tes and themis years ( my 24 to 27 ) . in order to constrain the kriging interpolation at high latitudes and at the poles , we set to 0.1 the value of the missing grid points located @xmath134 poleward of the northermost or southermost valid latitude , including the value at the poles ( when missing ) .
this value , although arbitrary , is suggested as being reasonable by numerical simulations of the dust cycle in gcms , and may be revised in future updates of the dust scenarios , as new observations become available . after replacing some missing values in the incomplete gridded maps ,
the data fields are ready to be interpolated by krigging .
we have used the idl built - in routine griddata for kriging interpolation on the sphere with an exponential semivariogram model .
we have chosen two different grids : a low resolution grid ( @xmath135 in longitude - latitude ) , and a high resolution grid ( @xmath136 in longitude - latitude ) . at the end of the interpolation procedure , carried out for each available sol from my 24 to my 31 , we obtain complete , regularly gridded maps of ir absorption cdod at the reference pressure of 610 pa . fig . [ fig_krigmap ] shows an example based on the same sol as in fig.[fig_tes_tw ] , kriged at high resolution .
figure [ fig_krigmap ] the series of daily maps is divided into annual dust scenarios , with 669 sols each , instead of using the martian calendar described in [ calendar ] .
the reason for this choice is that atmospheric models that use dust scenarios are often based on an integer number of sols in the year . for years with 668 sols in our sol - based martian calendar ,
we add the first sol of the following year .
the 669th sol of my 24 , 27 , 29 , therefore is the same as the first sol of , respectively , my 25 , 28 , and 30 .
since there are no maps available for the first part of my 24 ( before sol 225 ) , we use the first 224 sols of my 25 to fill the gap , therefore creating a `` hybrid '' my 24 year .
the junction at sol 225 is smoothed using a running average for the three sols before and the three sols after .
it is worth mentioning that if one wants to use a scenario for a specific year in a cyclic model simulation , the first few sols and the last few sols should be averaged ( e.g. with a running average ) to avoid the discontinuity .
similarly to what we do for the irregularly gridded maps , we provide an atlas of the daily dust scenario maps as supplementary material of this paper .
the structure of the atlas is the same : 669 pages ( one per sol ) , each one having eight maps for the respective years .
the dust scenarios ( currently v1.5 ) can be downloaded from the same url in netcdf format .
[ fig_zonalmean_krig ] shows zonal means built from the dust scenario maps as a function of solar longitude and latitude , as in fig .
[ fig_zonalmean_grid ] for the irregularly gridded maps .
figure [ fig_zonalmean_krig ]
in this paper we have described the procedures to obtain daily gridded maps of column dust optical depth for eight martian years , from @xmath137 in my 24 to the end of my 31 .
we have used retrieved and estimated cdods from three different instruments ( tes , themis , and mcs ) aboard different polar orbiting satellites ( mgs , ody , and mro ) .
data have been collected for about 14 earth years of spacecraft missions , from 1999 to 2013 .
observed cdods have been firstly gridded using the iwb methodology , which produces daily maps on a regularly distributed grid , but with missing values at some grid points .
subsequently , incomplete maps have been interpolated onto a regular , complete grid , using the kriging method after the application of a series of procedures to replace most of the missing values with climatological values .
the two datasets of ir absorption cdod at the reference pressure of 610 pa ( irregularly gridded maps and regularly gridded ones ) , have been separated in martian years , and made publicly available in the netcdf format on the mcd project website hosted by the lmd ( url : ` http://www-mars.lmd.jussieu.fr/ ` ) .
two atlases of the current version 1.5 of the maps are also available as supplementary material of this paper . key achievements of this work are the production and analysis of a continuous , multiannual climatology of dust based on several heterogeneous datasets , as well as the assessement of uncertainties ( both in the cdod retrievals and in the resulting gridded maps ) , and biases .
eight martian years of temporal and spatial ( 2d ) dust distribution have been presented in term of interannual and intraseasonal variability , down to the daily evolution of single dust storms .
we have confirmed that , from the statistical point of view , the years without global - scale storms are characterized by four phases in the solar longitude - latitude dust distribution , clearly highlighted in fig .
[ fig_dodtypyear ] .
this figure represents a ` typical ' climatological year for the dust optical depth , on top of which the annual optical depth is locally increased by the evolution of single dust storms , particularly in the second half of the year .
future developments of this work include the update of the climatology with the addition of new observations , not only for martian years beyond my 31 , but also for those already described in this paper .
the mcs team expects to carry out retrievals of cdod ( as additional standard product ) using nadir and off - nadir observations , which would greatly improve the quality of the dataset and reduce the biases identified in this work
. the use of estimated cdods from fitting top - of - atmosphere brightness temperatures with a gcm can be another important source of information to be used in the gridding procedure , in order to increase the number of valid grid points and reduce the need for the kriging interpolation in some areas .
reliable retrievals of cdod from the planetary fourier spectrometer ( pfs ) observations aboard mars express would be invaluable to improve the quality of the maps in my 27 and my 28 , when only sparse themis observations are available .
the iwb procedure would be particularly adapted to grid pfs observations , which are fairly heterogeneous in local time .
the use of data assimilation in a dynamical model of the dust cycle is the obvious improvement of the simple spatial kriging technique described in this work to produce complete dust scenarios , with ensemble methods providing the capability to estimate uncertainties . despite possible use of data assimilation techniques as ` optimal interpolators ' of dust observations ( or other types of aerosols and variables ,
e.g. @xcite ) , gridding retrievals from satellite orbits with iwb procedure can retain its key advantages of 1 ) being easy to implement , 2 ) not requiring a model , 3 ) providing statistical estimates of uncertainties based on those of the retrievals , and 4 ) giving immediate access to statistical analysis of the spatio - temporal evolution of events .
furthermore , this gridding technique could be applied to fields beyond the study of martian dust climatology , where data assimilation techniques might not be accessible or might be more difficult to implement .
lm is indebted to many people whose direct or indirect help made it possible to write this paper .
a particular mention is due to aymeric spiga , who always believed this paper would be finished one day , and provided time and competence during long discussion sessions .
lm is grateful to mathieu vincendon , tanguy bertrand , frank daerden for feedback on earlier versions of the dust scenario maps , and to helen wang for initially guiding us through the marci images .
a martian year has 668.5921 mean solar days ( sols ) or , with good approximation , 668.6 sols .
the solar longitude resets to zero by definition at the beginning of each new year ( spring equinox ) .
the scientific community primarily have been using a martian calendar based on the combination of martian year and solar longitude , since data are usually averaged over a few solar longitude degrees .
this combination is not satisfactory in our work , because our series of maps uses the sol as time variable .
if we want to devise a martian calendar based on sols , we need to keep track of the fraction of a sol left at the end of each martian year , similarly to what is done for the earth calendar .
many sol - based calendars have been proposed for mars , with different solutions for the structure of months and weeks , and for the way they deal with the leap years .
we mention here only the example of the darian calendar ( first described in a paper published by * ? ? ?
* ) , which has been proposed to serve the needs of possible future human settlers on mars . for scientific purposes , and in the interest of simple time - keeping , we have devised for the present work our own sol - based martian calendar
since there are approximately 668.6 sols in a martian year , every 5 years the number of elapsed sols is very close to the integer 3343 .
it is convenient , therefore , to define cycles of 5 martian years , with a number of sols respectively of 669 , 668 , 669 , 668 , and 669 ( a total of 3343 sols ) . by doing this ,
the end of each cycle is ( without approximation ) only 0.04 sol shorter than an integer number of sols , which accumulates to 1/4 of a sol after 30 martian years ( a reasonable amount for scientific purposes ) .
furthermore , assuming the first sol of a year starts when it is midnight at the prime meridian ( 00:00 mut ) , if there is a time bias @xmath138 between the first sol and @xmath11 at the beginning of a 5-year cycle , there is again the same bias at the beginning of the following cycle , if we approximate the martian year with 668.6 sols .
thanks to this property , the time biases between our sol - based calendar and a solar longitude - based calendar follow a repeatable pattern , and they are never too large for a reasonable number of martian years , if @xmath138 is properly set at the beginning of my 1 .
@xcite proposed an arbitrary convention to numerate the martian years starting from 11th april 1955 ( my 1 ) .
this date corresponds to the martian northern spring equinox preceding the planet - encircling dust storm of 1956 .
this storm marks the beginning of the new period of scientific observation and exploration of the planet mars . using this convention ,
all scientific data can be easily compared using a calendar based on positive martian years . since the introduction of this convention
, most of the scientific papers on mars adopted the numeration introduced by clancy et al .
, according to which @xmath11 in my 1 occurred at 08:31 utc on 11th april 1955 . for a solar longitude - based calendar , every new martian year since this date is marked by the occurrence of the northern hemisphere spring equinox . for the sol - based calendar introduced in the present paper ,
we maintain the numeration of the martian year introduced by @xcite , but we have to start our my 1 at a slightly different time , because at @xmath11 in my 1 the mut time ( local mean solar time ) was 13:26 rather than 00:00 .
the first sol of my 1 in our calendar starts on 11th april 1955 at 19:22 utc , when it was @xmath139 and 00:00 mut on mars , i.e. @xmath140 sol . by the beginning of my 26 ,
this bias has reduced to @xmath141 , therefore the sol - based calendar we use is less than 6 hours late with respect to the calendar based on solar longitude at the spring equinox of my 26 , and less than 5 hours late at the beginning of my 31 ( i.e the beginning of the following 5-year cycle ) .
this bias is satisfactory for the climatology from my 24 to my 31 , and we make the approximation that @xmath141 sol ( @xmath142 solar longitude ) for the three cycles we deal with in this paper . by doing so
, we can easily subtract a constant offset at the new year s sol in my 21 , 26 and 31 , to find the solar longitude of the corresponding spring equinox , with an error of about one hour in my 21 and 31 with respect to my 26 . in table
[ tab2 ] , we summarize the number of sols for each martian year and the solar longitude corresponding to each new year s sol . table [ tab2 ]
00 bell iii , j. f. , and 24 co - authors , mars exploration rover athena panoramic camera ( pancam ) investigation . _ j. geophys .
res . _ * 108 * , 8063 + , doi:10.1029/2003je002070 , 2003 bell iii , j. f. , and 20 co - authors , mars reconnaissance orbiter mars color imager ( marci ) : instrument description , calibration , and performance . _
j. geophys .
res . _ * 114 * , e08s92 , doi:10.1029/2008je003315 , 2009 cantor , b. a. , p. b. james , m. caplinger , m. j. wolff , martian dust storms : 1999 mars orbiter camera observations
_ j. geophys .
res . _ * 106 * , e10 , 23653???23687 , 2001 .
cantor , b. a. , moc observations of the 2001 mars planet - encircling dust storm .
_ icarus _ * 186 * , 60 - 96 , 2007 cantor , b. a. , observations of the martian atmosphere by mro - marci , an overview of 1 mars year .
_ mars atmosphere : modeling and observations workshop _ ,
williamsburg , virginia ( usa ) , abstract 9075 , http://www.lpi.usra.edu/meetings/modeling2008/modeling2008.authorindex.html , 2008 christensen , p. r. , and other 19 co - authors , miniature thermal emission spectrometer for the mars exploration rovers .
_ j. geophys .
( planets ) _ * 108 * , 8064-+ , 2003 christensen , p. r. , et al .
, mars global surveyor thermal emission spectrometer experiment : investigation description and surface science results . _
j. geophys .
_ * 106 * , 23,82323,871 , 2001 .
christensen , p. r. , et al .
, the thermal emission imaging system ( themis ) for the mars 2001 odyssey mission , _ space sci .
_ , 110 , 85130 , 2004 . clancy r. t. , b. j. sandor , m. j. wolff , p. r. christensen , m. d. smith , j. c. pearl , b. j. conrath , and r. j. wilson , an intercomparison of ground - based millimiter , mgs tes , and viking atmospheric temperature measurements : seasonal and interannual variability of temperatures and dust loading in the global mars atmosphere .
_ j. geophys .
res . _ * 105 * , 95539571 , 2000 .
clancy r. t. , m. j. wolff , and p. r. christensen , mars aerosol studies with the mgs tes emission phase function observations : optical depths , particle sizes , and ice cloud types versus latitude and solar longitude . _
j. geophys .
_ * * 108**e9 , doi:10.1029/2003je002058 , 2003 fenton , l. k. , j. c. pearl , and t. z. martin , mapping mariner 9 dust opacities . _ icarus _ * 130 * , 115 - 124 , 1997 .
forget , f. , f. hourdin , r. fournier , c. hourdin , o. talagrand , m. collins , s. r. lewis , p. l. read , j .-
hout , improved general circulation models of the martian atmosphere from the surface to above 80 km . _ j. geophys .
res . _ * 104 * , 24155???24176 , 1999 .
gangale , t. , martian standard time .
_ j. of the british interplanetary soc .
_ , * 39 * , no . 6 , 282 - 288 , 1986 .
guzewich , s.d .
, e. r. talaat , a. d. toigo , d. w. waugh , t. h. mcconnochie , high - altitude dust layers on mars : observations with the thermal emission spectrometer . _ j. geophys .
_ , * 118 * , issue 6 , 1177???1194 , 2013a .
guzewich , s.d . , a. d. toigo , m.i .
richardson , c.e .
newman , e. r. talaat , d. w. waugh , t. h. mcconnochie , the impact of a realistic vertical dust distribution on the simulation of the martian general circulation .
_ j. geophys .
_ , * 118 * , issue 5 , 980???993 , 2013b .
haylock , m.r . , n. hofstra , a. m. g. klein tank , e.j .
klok , p.d .
jones , and m. new , a european daily high - resolution gridded data set of surface temperature and precipitation for 1950 - 2006 .
_ j. geophys .
res . _ * 113 * , d20119 , doi:10.1029/2008jd010201 , 2008 .
hofstra , n. , m.r .
haylock , m. new , p.d .
jones , and c. frei , comparison of six methods for the interpolation of daily european climate data . _ j. geophys .
res . _ * 113 * , d21 , doi:10.1029/2008jd010100 , 2008 .
journel a.g . , and c. g. huijbregts , mining geostatistics .
600 pp , academic london , 1978 .
kavulich , jr . , m. j. , i. szunyogh , g. gyarmati , and r. j. wilson , local dynamics of baroclinic waves in the martian atmosphere .
_ j. atmos .
_ , * 70(11 ) * , doi:10.1175/jas - d-12 - 0262.1 , 2013 .
kleinbhl , a. , j. t. schofield , d. m. kass , w. a. abdou , c. r. backus , b. sen , j. h. shirley , w. g. lawson , m. i. richardson , f. w. taylor , n. a. teanby , and d. j. mccleese , mars climate sounder limb profile retrieval of atmospheric temperature , pressure , and dust and water ice opacity . _
j. geophys .
_ * 114 * , e10006 , doi:10.1029/2009je003358 , 2009 kleinbhl , a. , j. t. schofield , w. a. abdou , p. g. j. irwin , r. j. de kok , a single - scattering approximation for infrared radiative transfer in limb geometry in the martian atmosphere .
_ j. quant .
transfer _ , * 112 * , 1568 - 1580 , 2011 .
kleinbhl , a. , r. j. wilson , d. kass , j. t. schofield , d. j. mccleese .
the semidiurnal tide in the middle atmosphere of mars .
_ geophys .
_ , * 40 * , issue 10 , 1952???1959 , 2013 .
kuroda , t. , a. s. medvedev , p. hartogh , m. takahashi , semiannual oscillations in the atmosphere of mars .
_ geophys .
lett . _ * 35 * , issue 23 , 2008 .
heavens , n. g , m. i. richardson , a. kleinb??hl , d. m. kass , d. j. mccleese , w. abdou , j. l. benson , j. t. schofield , j. h. shirley , p. m. wolkenberg , the vertical distribution of dust in the martian atmosphere during northern spring and summer : high - altitude tropical dust maximum at northern summer solstice .
_ j. geophys .
_ , * 116 * , e01007 , 10.1029/2010je003692 , 2011 . lemmon , m.t .
wolff , m.d .
smith , r.t .
clancy , d. banfield , g.a .
landis , a. ghosh , p.h .
smith , n. spanovich , b. whitney , p. whelley , r. greeley , s. thompson , j.f .
bell iii , s.w .
squyres , atmospheric imaging results from the mars exploration rovers : spirit and opportunity .
_ science _ * 306 * , 1753 - 1756 , 2004 .
lemmon , m.t .
, m. j. wolff , j. f. bell iii , m. d. smith , b. cantor , and p. h. smith .
dust aerosol , clouds , and the atmospheric optical depth record over 5 mars years of the mars exploration rover mission _ icarus _ , submitted , 2014 lewis , s. r. , p. l. read , b. j. conrath , j. c. pearl and m. d. smith , assimilation of thermal emission spectrometer atmospheric data during the mars global surveyor aerobraking period .
_ icarus _ * 192 * , 327 - 347 , 2007 .
lorenc , a. c. , r. s. bell , and b. macpherson , the meteorological office analysis correction data assimilation scheme .
_ q. j. r. meteorol .
_ * 117 * , 58 - 89 , 1991 .
malin , m. c. , and 9 co - authors , an overview of the 1985 - 2006 mars orbiter camera science investigation .
_ mars _ , * 5 * , 1 - 60 , doi:10.1555/mars.2010.0001 , 2010 martnez - alvarado , o. ; montabone , l. ; lewis , s. r. ; moroz , i. m. and read , p. l. , transient teleconnection event at the onset of a planet - encircling dust storm on mars .
_ annales geophysicae _ * 27 * , 36633676 , 2009 .
mccleese , d. j. ; schofield , j. t. ; taylor , f. w. ; calcutt , s. b. ; foote , m. c. ; kass , d. m. ; leovy , c. b. ; paige , d. a. ; read , p. l. ; zurek , r. w. , mars climate sounder : an investigation of thermal and water vapor structure , dust and condensate distributions in the atmosphere , and energy balance of the polar regions . _
j. of geophys .
_ , * 112 * , e5 , doi:10.1029/2006je002790 , 2007 millour , e. , and 11 co - authors , an improved mars climate database .
proceedings of the fourth international workshop on the mars atmosphere : modelling and observation , paris , france , 8 - 11 february , 2011 .
millour , e. , and 18 co - authors , a new mars climate database v5.1 , proceedings of the fifth international workshop on the mars atmosphere : modelling and observation , oxford , uk , 13 - 16 january , 2014 .
montmessin , f. , f. forget , p. rannou , m. cabane , r. m. haberle , origin and role of water ice clouds in the martian water cycle as inferred from a general circulation model .
_ j. geophys .
res . _ * 109 * , issue e10 , 2004 .
mulholland , d.p .
lewis , p. l. read , the solsticial pause on mars : modelling and investigation of causes , in preparation , 2014 murchie , .s , and 40 co - authors , compact reconnaissance imaging spectrometer for mars ( crism ) on mars reconnaissance orbiter ( mro ) _ j. geophys .
( planets ) _ * 112 * , e11 , 5-+ , 2007 .
sinnott , r.w , virtues of the haversine . _ sky and telescope _ * 68 * , no .
2 , 159+ , 1984 .
smith m. d. , interannual variability in tes atmospheric observations of mars during 1999 - 2003 .
_ icarus _ * 167 * , 148165 , 2004 .
smith , m. , b. , themis observations of mars aerosol optical depth from 2002 - 2008 . _ icarus _ doi:10.1016/j.icarus.2009.03.027 , 2009 .
smith , m. d. , conrath , b. j. , pearl , j. c. , and christensen , p. r. , thermal emission spectrometer observations of martian planet - encircling dust storm 2001a _ icarus _ * 157 * , 259 - 263 , 2002 .
smith , m. d. , bandfield , j. l. , christensen , p. r. , richardson , m. i. thermal emission imaging system ( themis ) infrared observations of atmospheric dust and water ice cloud optical depth _ j. geophys .
_ * 108 * , 5115 + , 2003 .
smith , m. d. , wolff , m. j. , spanovich , n. , ghosh , a. , bandfield , j. l. , christensen , p. r. , landis , g. , a. , squyres , s. w. , one martian year of atmospheric observations using mer mini - tes . _ j. geophys .
_ * 111 * , e10 , doi:10.1029/2006je002770 , 2006 .
steele , l. j. ; lewis , s. r. ; patel , m. r. ; montmessin , f. ; forget , f. ; smith , m. d. , the seasonal cycle of water vapour on mars from assimilation of thermal emission spectrometer data .
_ icarus _ * 237 * , 97115 , 2014 .
vincendon , m. , langevin , y. , poulet , f. , pommerol , a. , wolff , m. , bibring , j .-
p . , gondet , b. , jouglet , d. , yearly and seasonal variations of low albedo surfaces on mars in the omega / mex dataset : constraints on aerosols properties and dust deposits , _ icarus _ * 200 * , 395405 , 2009 .
wang , h. , m. i. richardson , r. j. wilson , a. p. ingersoll , and r. w. zurek , cyclones , tides , and the origin of major dust storms on mars . _ geophys .
_ , * 30 * , 9 , 1488 , doi:10.1029/2002gl016828 , 2003 .
wang , h. , r. w. zurek , and m. i. richardson , the relationship between frontal dust storms and transient eddy activity in the northern hemisphere of mars as observed by mars global surveyor . _ j. geophys .
_ , * 110 * , e7 , e07005 , doi : 10.1029/2005je002423 , 2005 .
wang , h. , and m. i. richardson , m.i .
, the origin , evolution , and trajectory of large dust storms on mars during mars years 24 - 30 ( 1999 - 2011 ) . _ icarus _ , available online at http://dx.doi.org/10.1016/j.icarus.2013.10.033 , 2013 .
wilson , r. j. , haberle , r. m. , noble , j. , bridger , a. f. c. , schaeffer , j. , barnes , j. r. , cantor , b. a. , simulation of the 2001 planet - encircling dust storm with the nasa / noaa mars general circulation model .
proceedings of the third international workshop on the mars atmosphere : modelling and observation , williamsburg , va , usa , no 1447 , p.9023 , 10 - 13 november , 2008 .
wilson , r. j. , j. noble , s. j. greybush , the derivation of atmospheric opacity from surface temperature observations .
proceedings of the fourth international workshop on the mars atmosphere : modelling and observation , paris , france , pp.116 - 119 , 8 - 11 february , 2011 .
wolff , r. t .. , and p. r. clancy , constraints on the size of martian aerosols from thermal emission spectrometer observations . _ j. geophys .
* 108 * , 5097+ , 2003 .
wolff , r. t .. , and 12 co - authors , constraints on dust aerosols from the mars exploration rovers using mgs overflights and mini - tes .
_ j. geophys .
_ * 111 * , e12s17 , doi:10.1029/2006je002786 , 2006 .
wolff , r. t. , smith , m .
d. , clancy , r. t. , arvidson , r. , kahre , m. , seelos , f. , murchie , s. , savijarvi , h. , wavelegth dependence of dust aerosol single scattering albedo as observed by the compact reconnaissance imaging spectrometer .
_ j. geophys .
_ * 114 * , e00d04 , doi:10.1029/2009je003350 , 2009 .
[ cols="^,^,^,^,^,^,^,^,^ " , ]
* figure [ fig_timeline ] * a summary of the time coverage of all instruments used in this work , and the time limits within which observations are available .
lighter colors in the bars indicate the periods when observations from that particular instrument are available but either there are no cdod retrievals available or we do not currently use them .
* figure [ fig_retrievalnumberday ] * the number of dayside retrievals of column dust optical depth in absorption at 9.3 @xmath2 ( local times between 06:00 and 18:00 ) passing the quality control procedure described in the text .
the number of retrievals is summed in 1 sol @xmath143 @xmath144 latitude bins , and plotted for each year as a function of sol and latitude .
tes observed at higher spectral resolution ( 5 cm@xmath21 ) between sol @xmath146 in my 25 and sol @xmath147 in my 26 , therefore there are considerably less observations in this period , because the high spectral resolution data take twice as long to acquire as the low spectral resolution .
note that themis retrievals are summed together with tes retrievals and mcs retrievals , when available .
* figure [ fig_weights ] * plots of the weighting functions used in eq .
panel ( a ) shows the distance weight @xmath44 as a function of @xmath148 ( see eq .
[ eq3 ] ) for two different values of correlation scale @xmath57 , corresponding to @xmath149 km and @xmath150 km .
the cut - off distance @xmath66 is drawn here at 800 km as an example .
the @xmath57 function is plotted in panel ( b ) , as a function of @xmath151 , defined over the range @xmath152 for tw=7 sol . in panel ( c )
we plot the time weight function @xmath41 ( see eq . [ eq3tris ] ) for the same tw=7 sol .
the value of the @xmath41 function at the extrema of the time window interval is fixed to 0.05 in this work .
finally , panel ( d ) shows the plot of the uncertainty weight @xmath47 ( see eq .
[ eq4 ] ) as a function of the relative uncertainty @xmath153 , where @xmath154 is the standard deviation of a single column dust optical depth retrieval @xmath50 . * figure [ fig_tes_time ] * the difference between the time of each tes observation and noon at @xmath77 longitude in sol 449 , @xmath86 , my 24 ( the ` current time ' ) .
the observations are shown within a time window of 7 sols centered around the current time .
the picture is a zoomed view centered around @xmath155 longitude and @xmath156 latitude .
the black square shows a @xmath157 longitude - latitude box whereas the black circle indicates the 800 km distance range ( @xmath66 ) we use in the weighted binning for the tes dataset when the time window is 7 sols ( see table [ tab1 ] ) .
we apply the time and spatial weighting within this circle , as detailed in section [ bingrid ] .
* figure [ fig_tes_tw ] * ( a ) the panels from top to bottom show tes observations accumulated within time windows of 1 , 3 , 5 and 7 sol , centred around noon at @xmath77 longitude in sol 449 .
( b ) the panels from top to bottom show the corresponding partial results of the application of the weighted binning for each specific time window .
finally , in panel ( c ) we show the result of the iterative application of all 4 tw , where the valid grid points in each iteration are not overwritten .
there are 60x60 grid points in the gridded maps , separated by @xmath157 in longitude and latitude .
missing values are assigned to bins where the acceptance criterion is not satisfied ( see table [ tab1 ] ) .
* figure [ fig_my24_restests ] * tests on the sensitivity of the gridding procedure on the spatial resolution .
we use the same sol as in fig .
[ fig_tes_tw ] .
panel ( a ) and ( f ) show the case with @xmath135 longitude - latitude resolution , respectively in the average and in the standard deviation fields ; panel ( b ) and ( g ) are for the case with @xmath157 resolution ; panel ( c ) and ( h ) are for the case with @xmath157 resolution but longer range for the space weighting ( @xmath158 km and @xmath159 km ) ; panel ( d ) and ( i ) are for the case with @xmath160 resolution .
* figure [ fig_my24flush_evol ] * evolution of the my 24 , @xmath108 flushing storm .
each panel shows gridded column dust optical depth ( in absorption at 9.3 @xmath2 ) at the reference pressure level of 610 pa . from top
left to bottom right , maps are provided from sol 440 to sol 451 ( i.e. from @xmath161 to @xmath162 ) . in all panels , mut is noon and solar longitudes are provided accordingly .
see also [ calendar ] for the description of the sol - based martian calendar we use in this paper .
* figure [ fig_my29flush_evol ] * evolution of the my 29 , @xmath91 regional storm . as in fig .
[ fig_my24flush_evol ] , each panel shows gridded column dust optical depth ( in absorption at 9.3 @xmath2 ) at reference pressure level of 610 pa .
from top left to bottom right , maps are provided from sol 457 to sol 468 ( i.e. from @xmath163 to @xmath164 ) . * figure [ fig_histo_grid ] * histograms of smd values ( eq . [ eq5 ] ) for tes and mcs datasets in different martian years : ( a ) tes , my 24 ; ( b ) tes , my 25 ; ( c ) tes , my 26 ; ( d ) mcs , my 28 ; ( e ) mcs , my 29 ; ( f ) mcs , my 30 ; ( g ) mcs , my 31 .
* figure [ fig_percrelerr ] * plots of relative standard deviation curves for different martian years .
the relative standard deviation is defined as the ratio between the weighted standard deviation and the weighted average calculated , respectively , with eq .
[ eq2 ] and eq .
[ eq2bis ] , for each valid grid point .
* figure [ fig_mertimeseries ] * plot of column dust optical depth time series at ( a ) spirit site in gusev crater , and
( b ) opportunity site in meridiani planum .
the red curves are the time series of values interpolated from the daily gridded maps obtained with iwb procedure ( only values where 4 neighbor grid points are valid ) .
the green curves are sol - averaged values from mer mini - tes , whereas the blue curves are sol - averaged values from mer pancam .
corresponding uncertainty values are drawn as gray envelopes .
values interpolated from the gridded maps and mini - tes values are multiplied by 2.6 to convert them from absorption at 9.3 @xmath2 to equivalent visible , to compare to pancam values at 800 nm .
* figure [ moc_tes_my24 ] * tracks of tes column dust optical depth ( at 9.3 @xmath2 in absorption ) overlaid on moc mars daily global map images for 6 sols at 2 sol intervals in my 24 .
contours show the analyzed optical depth from the gridded maps derived from tes observations as described in the text .
( a ) sol 440 ( @xmath165 ) , ( b ) sol 442 ( @xmath166 ) , ( c ) sol 444 ( @xmath167 ) , ( d ) sol 446 ( @xmath168 ) , ( e ) sol 448 ( @xmath169 ) , and ( f ) sol 450 ( @xmath86 ) . in all maps , mut is noon , and solar longitude values are provided accordingly .
see [ calendar ] for the description of the sol - based martian calendar we use in this paper .
moc images are provided by bruce cantor .
* figure [ marci_mcs_my29_3 ] * tracks of surface temperature anomaly overlaid on marci mars daily global map images for ( a ) sol 460 ( @xmath170 ) and ( b ) sol 468 ( @xmath171 ) .
see [ calendar ] for the description of the sol - based martian calendar we use in this paper . brightness temperature anomaly is the apparent decrease in afternoon surface temperature ( 32 micron brightness temperature ) , relative to a clear atmosphere , that can be attributed to the influence of dust .
marci images are provided by helen wang . * figure [ fig_zonalmean_grid ] * zonal means of 9.3 @xmath2 absorption column dust optical depth maps ( referred to 610 pa ) as a function of solar longitude and latitude for all eight available martian years .
data are extracted from the irregularly gridded maps obtained with the application of the iwr procedure .
* figure [ fig_dodtypyear ] * plot of the zonal mean of the 9.3 @xmath2 absorption column dust optical depth in the climatological year , as a function of solar longitude and latitude .
the climatological year is obtained by averaging the irregularly gridded maps for all eight years , excluding the largest value of cdod for each grid point .
* figure [ fig_krigmap ] * map of 9.3 @xmath2 absorption column dust optical depth corresponding to fig .
[ fig_tes_tw ] ( sol 449 , my 24 ) after the application of kriging at high resolution .
the map is regular and complete , on a @xmath136 longitude - latitude grid . * figure [ fig_zonalmean_krig ] * zonal means of 9.3 @xmath2 absorption column dust optical depth maps ( referred to 610 pa ) as a function of solar longitude and latitude for all eight available martian years , calculated using the complete maps after the application of the kriging procedure . | we have produced a multiannual climatology of airborne dust from martian year 24 to 31 using multiple datasets of retrieved or estimated column optical depths .
the datasets are based on observations of the martian atmosphere from april 1999 to july 2013 made by different orbiting instruments : the thermal emission spectrometer ( tes ) aboard mars global surveyor , the thermal emission imaging system ( themis ) aboard mars odyssey , and the mars climate sounder ( mcs ) aboard mars reconnaissance orbiter ( mro ) .
the procedure we have adopted consists of gridding the available retrievals of column dust optical depth ( cdod ) from tes and themis nadir observations , as well as the estimates of this quantity from mcs limb observations . our gridding method calculates averages on a regularly spaced , but possibly incomplete , spatio - temporal grid , using an iterative procedure weighted in space , time , and retrieval uncertainty . in order to evaluate strengths and weaknesses of the resulting gridded maps , we associate values of weighted standard deviation with every grid point average , and compare with independent observations of cdod by pancam cameras and mini - tes spectrometers aboard the mars exploration rovers ( `` spirit '' and `` opportunity '' ) , as well as the compact reconnaissance imaging spectrometer for mars aboard mro .
we have statistically analyzed the irregularly gridded maps to provide an overview of the dust climatology on mars over eight years , specifically in relation to its interseasonal and interannual variability .
finally , we have produced multiannual , regular daily maps of cdod by spatially interpolating the irregularly gridded maps using a kriging method .
these synoptic maps are used as dust scenarios in the mars climate database version 5 , and are useful in many modelling applications in addition to forming a basis for instrument intercomparisons .
the derived dust maps for the eight available martian years ( currently version 1.5 ) are publicly available and distributed with open access .
= 1 mars , atmosphere , martian dust , dust climatology , martian dust storms |
the behavior of high temperature superconductors ( hts ) is characterized by a strong dependence on doping .
for example , changing the sr concentration in la@xmath7sr@xmath8cuo@xmath9 , or the o concentration in yba@xmath10cu@xmath11o@xmath12 tunes these materials from antiferromagnetic insulators , to spin glasses , and finally to superconductors .
the short coherence length and relatively low superfluid density in the underdoped region emphasize the existence of phase fluctuations and enhance the competition between superconductivity and other long range order phases @xcite . in such systems , where different ground states emerge as the particle density is tuned , it is often important to explore the possibility inhomogeneous mixtures of phases ; for example , antiferromagnetic and superconducting , at commensurate and incommensurate filling .
indeed , there has already been considerable study of whether the doped holes are uniformly distributed or whether they accumulate preferentially in two - dimensional phase separated regions@xcite or in one - dimensional stripes@xcite .
there have been many direct studies of doping dependent transitions in correlated electron hamiltonians like the repulsive hubbard model , which provide microscopic models of hts@xcite .
these are numerically challenging since they attempt to understand the origin of hts at the same time as addressing additional subtle issues like inhomogeneous phases .
a more simple , if somewhat less fundamental , way in which related questions can be addressed is to consider models , like the @xmath13 hubbard hamiltonian , which have an attractive interaction ` built in ' .
this separates out the problem of how superconducting pairs form in disordered , short coherence length systems in which the particle density is fixed but the interactions are diluted , from the complexities of the origins of the attractive interaction .
the key underlying question is how superconductivity spreads through a system as the local attraction between charge carriers is strengthened .
indeed , such centers of attraction played important role in the several models of hts@xcite .
another possible experimental realization is the percolative , granular - like transition to superconductivity via hydrogen doping which has been reported in eu@xmath14ce@xmath15nbsr@xmath16cu@xmath16o@xmath17@xcite , a ceramic with the superconducting critical temperature @xmath18= 29 k. in this paper , we shall adopt this approach and study a simple model which allows us to address the issue of superconducting transitions tuned by changing the concentration of attraction centers .
specifically , we study a two - dimensional negative @xmath0 hubbard model which has an on - site attraction on some fraction of its lattice sites and ask what fraction of sites is needed to be attractive for the system to superconduct . recently , this issue has been addressed by litak and gyrffy @xcite who examined the problem of superconducting percolation by using hartree - fock - gorkov decoupling as an approximation for interacting electrons and coherent potential approximation ( cpa ) to treat randomly distributed centers .
they have found the critical concentration of negative centers @xmath19 , where @xmath20 denotes a band filling in the limit of large @xmath21 interaction . for a small concentration of centers @xmath22
the system becomes normal because every site , where @xmath2 interaction is present , is doubly occupied .
the effect of dilution of the interactions in the attractive @xmath0 hubbard model at half - filling , @xmath23 can also be inferred from the quantum monte carlo work of ulmke _
et al._@xcite , who investigated the suppression of antiferromagnetism by @xmath24 impurities in the repulsive hubbard model ( @xmath25 ) at @xmath23 .
the critical concentration @xmath26 of impurities which destroy the antiferromagnetism was estimated as @xmath27 .
since a particle - hole transformation maps the attractive and repulsive @xmath0 hubbard models onto one another , at half - filling , this work immediately identifies the critical concentration @xmath6 for suppression of superconductivity in the attractive hubbard model , @xmath28 .
further analytic work on the disordered attractive hubbard model , for example based on numerical solution of the bogliubov - de gennes equations , is contained in @xcite
. however , there have to date been no quantum monte carlo ( qmc ) studies for the doped case where the density is not precisely half - filled and particle - hole symmetry is broken . in this paper
we examine the problem of percolating superconductivity using qmc in the negative @xmath0 hamiltonian changing the concentration of negative centers @xmath3 as well as band filling @xmath20@xcite .
the use of qmc enables us to go beyond the limitations of mean field approximations and investigate the effect of charge and pairing potential fluctuations on an equal footing .
the paper is organized as follows .
after an introduction in sec . 1 we present the microscopic model and formulate the problem ( sec . 2 ) .
section 3 defines the qmc monte carlo method used in simulations and shows the key numerical results : the prediction of superconductivity percolation and the critical concentration @xmath6 .
the paper ends with conclusions .
we consider the random negative @xmath0 hubbard model @xcite defined by the following hamiltonian : @xmath29 here @xmath5 and @xmath30 label sites on a two - dimensional square lattice , @xmath31 is the hopping integral connecting only nearest neighbour @xmath5 and @xmath30 s , @xmath32 is the electronic chemical potential , @xmath33 , @xmath34 create and annihilate , respectively , electrons at the single site @xmath5 with spin @xmath35 , and the coupling constant , @xmath36 we are interested in determining , away from particle - hole symmetry , the critical concentration @xmath6 for the @xmath0 centers such that for @xmath22 the configurationally averaged , superconducting long range order parameter vanishes . at the same time , we would like to obtain some insight into the mechanism of the formation of long range order as one moves to @xmath37 .
such questions can be addressed by computing expectation values of operators like the hamiltonian , the kinetic energy and double occupation , and pair - pair correlation functions .
such expectation values take the general form , @xmath38 and can be evaluated by a standard procedure @xcite which involves writing a path integral expression for the imaginary time evolution operator @xmath39 and performing a stochastic integration over a set of auxiliary fields introduced to treat the electron - electron interaction exactly .
this will be described in greater detail in the following section . typically , we can study system of a few hundred electrons , far larger than the competing exact diagonalization approach , but still small enough to require careful treatment of finite size effects , as we shall discuss below
. we will focus especially on the equal time pair correlation function , which is defined by @xmath40 here @xmath41 creates a pair of electrons at lattice site @xmath5 .
the long range order in @xmath42 can be probed by calculating its structure factor , @xmath43 a finite positive value @xmath44 , in the limit of large lattice size @xmath45 , signals superconductivity , otherwise the system is normal . for our disordered system
[ e2 ] , fig .
1 ) we have to do a disorder average over many realizations of random configurations of the attractive @xmath0 centers .
in order to describe the quantum monte carlo method @xcite in more detail , let us first rewrite the hubbard hamiltonian in a way which puts the interaction term in a particle - hole symmetric form . @xmath46 the trace over the fermion degrees of freedom can not be performed analytically due to the quadratic interaction of term @xmath0 . to reduce the problem to a quadratic hamiltonian we discretize the imaginary time @xmath47 ( eq . [ e3 ] ) and employ the trotter approximation @xcite to decompose the full imaginary time evolution .
the partition function can be expressed as , @xmath48^l \approx \left [ { \rm e}^{-\delta \tau \hat k } { \rm e}^{-\delta \tau \hat p } \right]^l~ , \label{e7}\ ] ] where @xmath49 includes the quadratic part of @xmath50 ( eq . [ e6 ] ) @xmath51 while @xmath52 the on - site interaction part @xmath53
then we apply the discrete hubbard stratanovich transformation to decouple the attractive interaction @xmath54 @xcite @xmath55 where @xmath56 the replacement of the quartic interaction term between up and down spin electron densities by an expression in which the densities couple to a hubbard - stratonovich field makes all terms in the exponential of eq . 7 quadratic in the fermion operators .
the trace can then be performed analytically , leaving a product of determinants which depend on the hubbard - stratonovich field . because the up and down spin species couple to the field with the same sign , these two determinants are identical , and
hence their product is always positive .
so , unlike the case of the repulsive model where @xmath57 couples to the field , there is no sign problem in the attractive model , even in the presence of disorder in the interaction .
using the above procedure we have simulated the hamiltonian ( eq . [ e6 ] ) and calculated the pair structure factor @xmath58 ( eq . [ e5 ] ) as a function of lattice size @xmath45 .
spin - wave theory suggests that in the ordered phase the finite size correction for @xmath58 should be proportional to the inverse linear system size @xmath59@xcite .
figure 2(a ) presents the results of finite size scaling of @xmath58 for various concentrations of @xmath0 sites @xmath3 .
the calculations were done for @xmath60 and @xmath61 , and the average number of electrons per site was chosen to be @xmath23 .
one can easily see that there exists a critical concentration @xmath62 . to see how @xmath6 depends on @xmath20 in fig .
2(b ) the concentration of @xmath0 centers is kept fixed at @xmath63 , and we have plotted @xmath58 for various system fillings ( @xmath64 and @xmath65 ) .
here @xmath23 corresponds to the case at the percolation threshold ( @xmath66 ) .
it is clear that for a larger average charge @xmath67 the system is not superconducting .
a different situation emerges for @xmath68 where @xmath69 appears to be slightly above 0 , that is , the system is superconducting .
clearly , disorder in the interaction @xmath0 has broken the particle hole symmetry .
in fact , this is connected with the modification of the bare chemical potential at @xmath0 sites , in the quadratic part of hamiltonian @xmath49 ( eq . [ e6],[e8 ] ) @xmath70 in some previous work on the repulsive model@xcite , disorder was inserted in a way which preserved particle - hole symmetry by using a separate shift in chemical potential depending on the interaction strength on each site .
this enabled simulations at lower temperature since it eliminated the sign problem , but also was somewhat less physically motivated .
interestingly , the structure factor associated with the pairing correlation function @xmath71 for @xmath72 is smaller than the one for @xmath73 .
this is due to the existence of a competing charge density wave ( cdw ) phase at @xmath23 .
a similar effect was found for the system with disordered site energies and the same attraction @xmath0 on every lattice site @xcite .
thus , in addition to phase fluctuation of the pairing potential@xcite , it is clear that the distribution of charges @xmath74 on the lattice ( charge fluctuations ) is also important for the destruction of superconductivity . indeed
, previous work suggests that percolation of the occupied sites is a crucial phenomenon in this process@xcite . examining this effect further
, one can look for the change of the pairing factor correlation function @xmath58 and simultaneously the average occupation of @xmath0 sites ( @xmath75 ) ( where @xmath76 ) .
figures 3a - c show these quantities as a function of average site occupation @xmath20 for a finite system @xmath77 for @xmath73 ( a ) , @xmath78 ( b ) and c=0.375 ( c ) , respectively .
note that @xmath58 does not display particle - hole symmetry .
the dotted lines in fig .
3(a - c ) represent the occupation of attractive sites for a simplified hamiltonian , where @xmath49 reads @xmath79 it is easy to see that in case of the full hamiltonian ( eqs .
[ e2],[e6 ] ) the system displays charge fluctuation ( @xmath80 ) and comparing @xmath75 curves to the dotted lines one can see that charge fluctuations are stronger for smaller @xmath3 .
in our case , for large enough interaction @xmath0 , the small concentration of @xmath0 centers leads to double occupation in almost all attractive sites .
this makes it harder for electrons to hop to another @xmath81 site .
nevertheless the system can be metallic because the other electrons which occupy the @xmath24 site can easily hop to the other @xmath24 sites .
this result is consistent with the results obtained by litak and gyrffy @xcite .
the small dip visible in the pairing structure factor @xmath58 around @xmath23 ( in fig .
3a ) can be associated with the creation of a cdw order .
this effect is not present in figs .
3b - c as cdw expected to disappear faster than superconductivity @xcite .
we have examined the problem of percolating superconductivity in the context of a random @xmath0 hubbard model .
we have studied the case where @xmath1 is @xmath2 and 0 with probability @xmath3 and @xmath4 respectively by the quantum monte carlo method . for the half filled system @xmath23 we obtain the critical concentration @xmath82 . as we expected
the transition is due to phase and charge fluctuations and depends strongly on the density @xmath20 . in that sense
our results agree qualitatively with those obtained by litak and gyrffy@xcite .
moreover , the present approach is exact as opposed to earlier work@xcite employing the mean field approximation .
we find that the onset of superconductivity in this diluted , attractive hubbard model is linked to percolation of the order parameter and charge fluctuations .
gl acknowledges a partial support by the polish state committee for scientific research ( kbn ) , project no
. 5 p03b 00221 .
the work of rts was supported by nsf dmr9985978 .
tp snd rrds are grateful to brazilian agencies faperj and cnpq for financial support . | we study the effect of random dilution of attraction centers on electron pairing in a generalization of the negative @xmath0 hubbard model where the interaction @xmath1 , defined on the two dimensional lattice , can be @xmath2 and 0 with probability @xmath3 and @xmath4 depending on the lattice site @xmath5 . using the determinant quantum monte carlo approach
, we find the critical concentration @xmath6 of negative centers which leads to superconductivity , and show how the evolution of the local density on the two types of sites lends insight into the formation of global phase coherence . , , ,
disorder , doping , fluctuations |
the bogoliubov de gennes equations and approximations to these equations , for example the eilenberger equations have in the last 10 years been used to great sucsess to examine the properties of normal superconducting hybrid structures .
most of this work , in recent times , has centered upon solving these equations for the electrical mesoscopic tranport properties of such hybrid structures for a recent review of calculations of electrical conductance properties of normal superconducting hybrid structures see lambert and ramondi @xcite .
virtually all of this work has been centred around the properties of systems which contain conventional superconducting regions , ie .
the superconducting gap has been isotropic .
also all but a few @xcite have been non - self - consistent calculations , this however has not inhibited the sucsess of such calculations . in more recent times , with the discovery of high - temperature superconductivity again the bogoluibov de gennes equations have been used to examine the properties of hybrid structures , where now the superconducting gap is taken to be non - isotropic , usually @xmath1wave in symmetry , but mainly so far most of the theoretical analysis of _ simple _ experiments @xcite have relied upon macroscopic symmetry arguments and not the microscopic details of the actual interfaces .
however , the microscopic physics at the interface can be an important factor in understanding the experimental results .
for example , if mixing of different order parameter symmetries occurs at the interface ( because the interface breaks the bulk tetragonal or orthorhombic symmetry ) the extent of such mixing can only be determined from microscopic calculations .
similarly , suppression of the order parameter ( either @xmath1wave or @xmath0wave ) near an @xmath4 interface can lead to significant local density of states within the bulk energy gap , and this can complicate the analysis of single electron tunnelling spectra@xcite . in the past few years there have been a number of microscopic calculations of surfaces and interfaces in superconducting systems with a @xmath1wave order parameter .
most of the theoretical results have been obtained using tunnelling theory , or andreev s approximation@xcite in which the tunnelling barrier and the order parameter are not found self - consistently . for
tunnel junctions these approximations may be adequate , but we show below that self - consistency has significant effects for interfaces with direct contact between the constituents .
self - consistent properties of interfaces have previously been computed using the eilenberger equations @xcite , which are an approximation to the bogoliubov de gennes equation .
these self - consistent solutions to the eilenberger equations have shown some interesting effects .
also there have now in the last couple of years been a few self - consistent calculations of the bogoliubov de gennes equations , with non - local interactions , looking at the effect of impurity scattering @xcite and superconducting interfaces @xcite . in this paper
we tackle a _
simple _ problem , solving the bogoliubov de gennes equations self - consistently , in the presence of an @xmath0wave and a @xmath1wave order parameter with the aim of asking , how important is self - consistency ?
we present a method of how to calculate self - consistent properties of superconducting interfaces by directly solving the bogoliubov de gennes equations @xcite and that certainly in the case of a @xmath1wave interface it is crucial to perform such a calculation to obtain the correct form for the low energy density of states in the region of an interface .
the method of performing our self - consistent calculations will be described in section ii . here
we introduce the bogoliubov de gennes equation with a general interaction , @xmath5 , and demonstrate how this general hamiltonian can be solved self - consistently using the recursion method @xcite .
we then present results for bulk @xmath0wave and @xmath1wave systems ; looking at both the local particle density of states and in the case of @xmath1wave we look at the critical temperature as a function of band filling , we see that a @xmath1wave solution , for the interaction we have chosen , is not the only possible solution , for our hamiltonian . then in section iii we present results for both @xmath2 and @xmath3 systems .
we examine both the form of the superconducting order parameter in the region of the interface , for different strengths of insulating barrier , and how the local particle density of states varies in the region of the interface .
what we see is the sub - gap local particle density of states , on the superconducting side of the interface , is not in the case of @xmath0wave superconductivity effected by changing the strength of the insultaing barrier , this is in marked contrast to what we see in the @xmath1wave case .
finally in the conclusions we comment on what can be said in the light of our calculations .
our starting point is the non - local hubbard model , which is described by the following hamiltonian @xmath6 where the creation and annihilation operators @xmath7 and @xmath8 , respectivily , create and annihilate electrons with spin @xmath9 in the orbital centred at the lattice point labelled by @xmath10 , @xmath11 is the amplitude for hopping from site @xmath12 to site @xmath10 , @xmath13 is the chemical potential and @xmath5 is the interaction energy of two electrons with opposite spin on sites @xmath10 and @xmath12 . to obtain the , generalized non - local , bogoliubov de gennes equation from equation ( [ eq : hub ] ) we first make a mean field approximation for the _ pairing field _ @xmath14 and assuming that the fluctuations about this mean are small we then perform the bogoliubov cononical - transformation , enabling us to obtain @xmath15 with @xmath16 where @xmath17 and @xmath18 are the particle and hole amplitudes , on site @xmath10 , associated with an eigenenergy @xmath19 and where @xmath20 is the , possibly non - local , pairing potential or gap function , which does not have to be real . in the fully self - consistent bogoliubov de gennes equation
the normal state hamiltonian @xmath21 is given by @xmath22 where @xmath13 is the chemical potential , @xmath23 is the normal on site energy of site @xmath10 and @xmath11 , as mentioned above is the hopping integral between site @xmath10 and site @xmath12 .
the interaction terms @xmath24 and @xmath25 are the hartree - fock potentials , which are defined by @xmath26 similarly the pairing potentials are defined as @xmath27 where the anomalous density is @xmath28 in equations ( [ eq:3 ] ) the sum only considers terms @xmath19 up to the condenstate chemical potential ( @xmath13 ) as compared to equation ( [ eq:5 ] ) where the sum is over all terms .
a solution to the above system of equations will be fully self - consistent provided that both the normal ( @xmath29 ) and anomalous ( @xmath20 ) potentials are determined consistently with the corresponding densities @xmath30 and @xmath31 via equations ( [ eq:3 ] ) and ( [ eq:5 ] ) .
figure 1 illustrates the geometry corresponding to this system of equations , on a tight binding lattice .
the tight - binding lattice has nearest neighbour hopping interactions ( @xmath11 ) , as well as a coupling between particle and hole space , via a superconducting order parameter ( @xmath20 ) . if the interactions are purely on - site ( @xmath32 ) attractions then the pairing potential will be purely local ( @xmath33 ) , corresponding to the dashed line in figure 1
. on the other hand when the interaction is non - local ( @xmath5 , @xmath34 ) the pairing potential @xmath20 will also be non - local , as illustrated , for nearest nieghbour interactions , by the solid lines in figure 1 .
if we now wish to solve the bogoliubov de gennes equations self - consistnetly , we have found that the most efficient method of doing this is the recursion method @xcite .
this method allows us to calculate the electronic green s functions @xmath35 where the indices @xmath10 and @xmath12 denote sites , while @xmath36 and @xmath37 represent the particle or hole degree of freedom on each site .
we denote particle degrees of freedom by @xmath38 and hole degrees of freedom by @xmath39 .
for example @xmath40 represents the greens function between the particle degree of freedom on site @xmath10 and the hole degree of freedom on site @xmath12 . to compute the green s functions ( [ eq:6 ] )
we can closely follow the method described by litak , miller and gyrffy@xcite , they solved the bogoliubov de gennes equation self - consistently , using the recursion method , for a local interaction . using this method we can transform the hamiltonian to a block tridiagonal form @xmath41 where @xmath42 and @xmath43 are @xmath44 matrices
given this form for @xmath45 and expressing the green s function as @xmath46 the greens functions above can be evaluated as a matrix continued fraction so that @xmath47 where @xmath48 within equations ( [ eq:7 ] ) and ( [ eq:9 ] ) we have a formally exact representation of the green s functions .
however in general both the tridiagonal representation of the hamiltonian , and the matrix continued fraction ( [ eq:7 ] ) will be infinite .
in practice one can only calculate a finite number of terms in the continued fraction exactly . in the terminology of the recursion method it is necessary to _ terminate _ the continued fraction @xcite . if we were to calculate up to and including @xmath42 and @xmath49 and then simply set subsequent coefficients to zero then the green s function would have @xmath50 poles along the real axis .
the density of states would then correspond to a set of @xmath50 delta functions .
integrated quantities such as the the densities @xmath30 and @xmath31 could depend strongly on @xmath51 , especially since only a few of the @xmath50 delta functions would be within the relevant energy range within the bcs cut off , @xmath52 . in order to obtain accurate results it would be necessary to calculate a large number of exact levels , which would be expensive in terms of both computer time and memory . as a more efficient alternative we choose to terminate the continued fraction using the extrapolation method , as used previously by litak , miller and gyrffy@xcite .
we calculate the values for @xmath42 and @xmath49 exactly up to the first @xmath53 coefficients using the recursion method .
then , noting the fact that the elements of the matrices @xmath42 and @xmath49 vary in a predictable manner @xcite , we extrapolate the elements of the matrices for a further @xmath54 iterations , where @xmath54 is usually very much greater than @xmath53 .
this enables us to compute the various densities of states , and the charge densities @xmath30 and @xmath31 accurately with relatively little computer time and memory . in terms of the green s functions @xmath55
the pairing and normal hartree - fock potentials @xmath20 and @xmath56 can be expressed as @xmath57 and @xmath58 where @xmath59 is a small positive number . to obtain the above equations we have used the fact that @xmath60 are the eigenvectors of equation ( [ eq:1 ] ) with eigenvalues @xmath19 and @xmath61 .
it should be noted that up to this point we have been completely general , so we are not just restricted to calculating the greens functions on the same site , @xmath62 or just between nearest neighbours , in essence we could use this formalism to calculate a greens function between any two sites .
this enables us to also calculate such quantities as @xmath63 and @xmath64 , which can then in turn be used to calculate normal reflection and transmission coefficients , thus enabaling in principle for transport properties or our hamiltonian to be calculated . using the above methods to calculate @xmath20 and @xmath29 we need to achieve a fully self - consistent solution .
firstly we make use of any symmetries in the system in order to minimise the number of calculations which are necessary .
for example on an infinite square lattice with no variation in any of the potentials only one independent site needs to be calculated since this site can be mapped onto all of the other sites .
secondly , once we have decided which sites need to be calculated self - consistently @xmath20 and @xmath30 can be calculated for those sites , remembering that on a square lattice each site will have four nearest neighbours .
this implies that in general we will have to calculate nine different green s functions in order to calculate @xmath20 and @xmath30 .
this can be seen by considering site @xmath10 in figure 1 and noting that we need to calculate the greens functions shown in table i , depending on whether the interaction is purely local , purely non - local , or both local and non - local . having calculated the appropriate green s functions new values for @xmath20 and @xmath30 can be calculated , which we will denote as @xmath65 and @xmath66 .
inserting these into the hamiltonian and repeating the calculation of then green s functions leads to a new set @xmath67 and @xmath68 and so on .
we repeat this iteration for all @xmath10 and @xmath12 until @xmath69 and @xmath70 since @xmath20 and @xmath30 can be complex we need to also check for convergence in their associated phases . up to this point
we have said nothing about the interaction itself , just that it exists . in conventional bcs theory
@xcite we can impose a cut - off for our interaction , since we know that in this case the interaction is phonon mediated , thus in equations ( 14 ) and ( 16 ) we can have a cut - off for our integrals .
however we also wish to consider the properties of high - temperature superconductors , where the mechanism for the interaction is a contriversial area , so we state at this point that we are not interested in the mechanism for high - temperature superconductivity , but we do know , through experiment @xcite , that the symmetry of the superconducting order parameter in high - temperature materials is of a @xmath1wave nature , therefore , generally we wish to consider the propeties of @xmath1wave superconductors . the simplest model to consider is one where we have a nearest neighbour attractive interaction , @xmath5 . for such a situation , where the interaction is constant we can calculate the critical temperature for a given interaction strength , as a function of band filling , see figure 2 .
it is found that there are two possible solutions for the symmetry of the superconducting order parameter . in this figure
we can see there are two critical temperatures , one for extended @xmath0wave solutions , near the edges of the band , and a @xmath1wave solution near the centre of the band .
it is this @xmath1wave solution which we will be considering , however it is important to note , and will be seen later , that in non - uniform systems although we may be at a band filling where only the @xmath1wave solution exists this does not nessesitate that the extended @xmath0wave components to the superconducting order parameter will thence be zero . before proceeding
it is instructive to examine the bulk local particle density of states ( @xmath71 ) for the systems we are concerned with . in figure 3 we have plotted the local particle density of states , @xmath72 for three different scenarios , no interaction ( thin dashed line ) , a local interaction ( thin solid line ) and a , nearest nieghbour , non - local interaction ( thick solid line ) .
what we see from this is exactly what one would expect , i.e. for no interaction , we obtain the density of states for a regular , nearest nieghbour tight binding lattice , for a local interaction we obtain a bcs type gap centred around the fermi energy , and finally for a non - local interaction we see a typical @xmath1wave gap , ie .
the local particle density of states goes to zero linearly at the fermi energy .
having said that there is no way , at this point to descern that we actually have a @xmath1wave solution , but we define a @xmath1wave order parameter in the following manner @xmath73 where @xmath74 , @xmath75 , @xmath76 and @xmath77 are the nearest nieghbours to @xmath10 , moving in a clockwise or anticlockwise direction , this quantity is then the @xmath1wave component to the order parameter at @xmath10 , similarly we can also write @xmath78 for the extended @xmath0wave component to the order parameter . for the results obtained in figure 3 @xmath79 and @xmath80 .
now we can proceed to perform self - consistent calculations for interfaces .
our _ model _ is as follows , we consider a system which does not vary in the @xmath81 coordinate and has a step like function in the interaction strength in the @xmath82 coordinate ; ie . for @xmath83
the interaction is finite ; whereas for @xmath84 our interaction ( be it local or nearest nieghbour non - local ) is zero .
the interface itself is modelled by a line of sites at @xmath85 , where we change the on - site energy of these sites in accordance with the strength of the barrier at the interface we wish to model .
we then proceed to calculate the bogoliubov de gennes equations in the self - consistent manner prescribed above .
first we look at the results obtained for a local interaction ( @xmath86 ) .
we see in figure 4 the nomalized magnitude of the superconducting order parameter ( upper graph ) and @xmath87 ( lower graph ) , where in reference to the prevoius equations we have made a transformation from sites @xmath10 to position @xmath82 . in this figure
we see how these two quantities vary in the region of the interface ( @xmath85 ) , for different strengths of the barrier at the interface , @xmath88 ( thin solid line ) , @xmath89 ( thin dashed line ) , @xmath90 ( thick solid line ) and @xmath91 ( thick dashed line ) .
the main point to deduce from this is that the barrier has very little effect upon the actual order parameter or the local density .
this is in comparison to what we see for an equivalent calculation for a non - local interaction , the results of which are shown in figure 5 , where we have plotted the nomalised @xmath1wave contribution to the superconducting order parameter ( upper graph ) and the extended @xmath0wave contribution to the superconducting order parameter ( lower graph ) , for the same interface configurations as used to obtain figure 4 .
what we see here is that although @xmath92 is not seroiusly modified in the presence of a barrier , @xmath93 is .
we can see that as we increase the strength of the barrier @xmath93 in the region of the interface is reduced .
we now wish to focus our attention on the local particle density of states , near the interface .
how this quantity changes will be important in deducing whether it is crucial to perform self - consistent calculations to obtain quantities which are reliant upon the local particle density of states near the interface .
as an example we are going to focus on the sub - gap density of states and how this changes for different barrier strengths , and whether it is then valid to use non - self - consistent calculations in calculating such quantities as the sub - gap conductance ? with this in mind
we have in figure 6 plotted the low energy local particle density of states for a local interaction ( @xmath86 ) , at @xmath94 .
we have plotted this quantity for different strengths of barrier , @xmath88 ( thin dashed line ) , @xmath89 ( thin solid line ) , @xmath90 ( thick dashed line ) , @xmath91 ( thick solid line ) and @xmath95 ( thin dashed - dotted line ) . again , as in figure 4 there is little change in the quantity we are interested in .
in fact from this plot we can say that a quantity such as the sub - gap , @xmath2 conductance depends very little on changes in the density of states and is totally governed by the tunnelling conductance of the barrier .
however again taking the @xmath1wave case we can see an altogther different story .
in figure 7 we have plotted equivalent local particle density of states as in figure 6 , but for a @xmath1wave interface .
what we see is that the low energy density of states in the region of the barrier is modified significantly , when we change the strength of the barrier , and hence we can no longer state that the conductance of an @xmath3 interface will be mainly governed by the tunnelling conductance of the barrier .
this implies that in such calculations , when comparing , for example the conductances of different @xmath3 interfaces , the details of the barrier effect the sub - gap density of states on the superconducting side of the barier and hence self - consistency has an important role to play . finally to highlight the point that the low energy local particle density of states in the region of an interface is modified significantly , by the strength of the barrier at the interface ,
when the supercunducting region has a @xmath1wave symmetry , we have in figure 8 plotted the local particle density of states at @xmath96 ( a ) , @xmath94 ( b ) , @xmath97 ( c ) and @xmath98 ( d ) .
for two different strengths of barrier @xmath88 ( thin solid line ) and @xmath99 ( thick solid line ) . from this figure
we can see the _
dramatic _ change in the sub - gap local density of states . particularly in figure 8(a ) ( nearest the interface ) we see that in the region of @xmath100 , when thier is no interface barrier , there is always a significant density of states at the interface , as compared to @xmath101 when we have a finite barrier .
in this paper we have developed the recursion method for calculating the bogoliubov de gennes equations self - consistently with both local and non - local interactions .
then as a _ simple _ problem we examined the self - consistent local particle density of states in the region of both @xmath2 and @xmath3 interfaces , in particular we have focused on how the local particle density of states is modified as we change the strength of the barrier .
what we see is that in the case of the @xmath2 interface the low energy particle density of states , on the superconducting side of the interface , does not change dramatically as we change the strength of the insulating barrier .
this is in stark contrast to the comparitively dramatic change we see when we perform the same calculation for a @xmath3 interface .
these results imply that in the case of @xmath1wave interfaces the details of the the interface are important and hence to get these details correct we should perform a self - consistent calculation .
if we do a non - self - consistent calculation for the case of the @xmath1wave interface , i.e. just let the order parameter vary in a step like manner at the interface , then we see that the low energy density of states on the superconducting side of the barrier is unaltered as we change the strength of the insulating barrier , at the interface .
this discrepancy , between self - consistent and non - self - consistent calculations , arises from the fact that we have a mixing of symmetries in the region of the self - consistent @xmath1wave interface and when a self - consistent calculation is performed we see that this mixing changes as the interface is changed , this effect will not be seen in a non - self - consistent calculation .
these results may well explain why non - self - consistent conductance calculations for hybrid @xmath0wave structures have over the past years been so sucsessful , i.e the problem can simply be seen as calculating the tunnel conductance of the barrier , but also raises a warning flag that such a bold approach for @xmath1wave structures may not be as effective , since our calculations show that the low energy density of states changes as we change the properties of the interface , we believe that the only way to obtain the correct low energy density of states behavoir is to perform a self - consistent calculation .
this work was supported by the epsrc under grant number gr / l22454 and by the tmr network dynamics of nanostructures .
we would like to thank b.l gyrffy for useful discussions .
lambert and r. raimondi , j. phys . : condens .
matter * 10 * , 901 ( 1998 ) . c. bruder , phys .
b * 41 * , 4017 ( 1990 ) .
j. hara , m. ashida and k. nagai , phys .
b * 47 * , 11263 ( 1993 ) .
barash , a.v .
galaktionov and a.d .
zaikin , phys .
b * 52 * , 665 ( 1995 ) .
t xiang and j.m .
wheatley , phys .
b * 51 * , 11721 ( 1995 ) . p. miller and b.l .
gyrffy , j. phys . : condens .
matter * 7 * , 5579 ( 1995 ) .
canizares and f. sols , j. phys . : condens .
matter * 7 * , l317 ( 1995 ) .
canizares and f. sols , phys .
b * 55 * , 531 ( 1997 ) .
a.m. martin and c.j .
lambert , phys .
b * 51 * , 117999 ( 1995 ) .
a.m. martin and c.j .
lambert , j. phys . : condens .
matter * 8 * , l731 ( 1996 ) .
l. chang , s. chaudhuri and p.f .
bagwell , phys .
rev b * 54 * , 9399 ( 1996 ) .
r. a. riedel , l. chang and p.f .
bagwell , phys .
b * 54 * ( 1996 ) .
wollman , d.j .
van harlingen , w.c .
lee , d.m .
ginsberg and a.j .
leggett , phys .
lett . * 71 * , 2134 ( 1993 ) .
a.g . sun , d.a .
gajewski , m.b .
maple and r.c dynes , phys .
. lett . * 72 * , 2267 ( 1994 ) .
et al . _ ,
73 * , 593 ( 1994 ) .
k. kitizawa , science * 271 * , 313 ( 1996 ) .
hu , phys .
72 * , 1526 ( 1994 ) .
miller and c.s ting , phys .
b * 53 * , 3604 ( 1996 ) .
y. tanaka and s. kashiwaya , phys .
lett . * 74 * , 3451 ( 1995 ) .
zhu , z.d .
wang , and h.x .
tang , phys .
b * 54 * , 7354 ( 1996 ) .
y. tanaka , phys .
lett . * 72 * , 3871 ( 1994 ) .
s. kashiwaya , y. tanaka , m. koyanagi , h. takashima and k. kajimura , phys .
b * 51 * 1350 ( 1994 ) . h.x .
tang , z.d .
wang and j.x .
zhu , phys .
b * 54 * , 12509 ( 1996 ) .
m. suzuki , m. taira and x zheng , czechoslovak j. phys . * 46 s3 * , 1351 ( 1996 ) . y. tanaka and s. kashiwaya , phys .
b * 56 * , 892 ( 1997 ) .
l. alff , h. takashima , s. kashiwaya , n. terada , h. ihara , y tanaka , m. koyanagi and k. kajimura , phys .
b * 55 * r14757 ( 1997 ) .
y.s barash , a.v .
galaktionov and a.d .
zaikin , phys .
b * 52 * , 665 ( 1995 ) . m. matsumoto and h. shiba , j. phys .
* 64 * , 3384 ( 1995 ) .
m. matsumoto and h. shiba , j. phys .
. jpn . * 64 * , 4867 ( 1995 ) . m. matsumoto and h. shiba , j. phys .
65 * , 2194 ( 1996 ) .
barash , a.a .
svidzinsky and h. burkhardt , phys .
b * 55 * , 15282 ( 1997 ) . y. morita , m. kohmoto and k. maki , phys
* 78 * , 4841 ( 1997 ) .
y. onishi , y. ohashi , y shingaki and k. miyake , j. phys .
. jpn . * 65 * , 675 ( 1995 ) .
y. tanuma , y. tanaka , m. yamashiro and s. kashiwaya , physica c * 282 - 287 * , 1857 ( 1997 ) .
a.m. martin and j.f .
annett , phys .
b * 57 * , 8709 ( 1998 ) .
r. haydock , in : solid state physics , vol .
* 35 * , eds .
ehrenreich , f. seitz and d. turnbull ( academic press , new york , 1980 ) .
g. litak , p. miller and b.l .
gyrffy , physica c * 251 * , 263 ( 1995 ) .
a. magnus in : solid state sciences * 58 * eds .
pettifor and d.l .
f. ducastelle , p. turchi and g. treglia in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
nex in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
g. allan in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
a. trias , m. kiwi and m. weissmann , phys . rev .
b * 28 * , 1859 ( 1983 ) .
j. bardeen , l.n .
cooper and j.r.schieffer , phys . rev . *
108 * , 1175 ( 1957 ) .
annett and n.d .
goldenfeld , j. low temp . phys . * 89 * , 197 ( 1992 ) .
.this table shows which greens functions need to be calculated for systems with interactions which are local , @xmath32 , non - local , @xmath102 , or both .
the site labels correspond to the notation of fig . 1 . [ cols="^,^,^,^,^,^",options="header " , ] | we develop a method to solve the bogoliubov de gennes equation for superconductors self - consistently , using the recursion method .
the method allows the pairing interaction to be either local or non - local corresponding to @xmath0 and @xmath1wave superconductivity , respectively . using this method we examine the properties of various @xmath2 and @xmath3 interfaces .
in particular we self - consistently calculate the spatially varying density of states and the superconducting order parameter .
we see that changing the strength of the insulating barrier , at the interface , does not , in the case of an @xmath0wave superconductor , dramatically , change the low energy local density of states , in the superconducting region near the interface .
this is in stark contrast to what we see in the case of a @xmath3 interface where the local particle density of states is changed dramatically .
hence we deduce that in calculating such properties as the conductance of @xmath2 and @xmath3 structures it is far more important to carry out a self - consistent calculations in the @xmath1wave case . |
the full dynamical content of the ads / cft correspondence @xcite ( for a review @xcite ) can be summarized by an equivalence between two path integrals @xcite , @xcite : @xmath0=\displaystyle{\int_{\phi\sim\phi_0 } } { \cal d } \phi~\exp(-s[\phi])=\big<\exp\int_{m_d } d^dx \phi_0 { \cal o}\big>_{cft } \label{adscft}\ ] ] here @xmath1 represents the fields propagating on anti - de sitter space ( ads ) and @xmath2 are the dual gauge invariant operators in the conformal field theory ( cft ) .
@xmath3 $ ] is the partition function on ( euclidian ) ads@xmath4 , computed with the boundary condition that the field @xmath1 approaches @xmath5 at the boundary of ads .
the right - hand side of ( [ adscft ] ) is the generating functional of the correlation functions of the dual cft living on the boundary @xmath6 of ads@xmath4 . in the limit where classical field theory is a good description of the physics on ads ( corresponding to the large @xmath7 limit and large t hooft parameter on the cft side ) , @xmath8=\exp(-s_{cl}[\phi_0])\ ] ] where @xmath9 $ ] is the on - shell value of the classical action for fields that satisfy the condition @xmath10 at the boundary of ads . in order to compute @xmath9 $ ]
typically one first reconstructs , using the classical equations of motion , the on - shell value of @xmath1 with the required boundary behavior and then substitute in the action .
this can be done iteratively using bulk - to - boundary and bulk - to - bulk propagators .
the goal of this paper will be to show how to compute @xmath9 $ ] using the method of brane running . in principle , no explicit knowledge of bulk - to - boundary an bulk - to - bulk propagators is required .
this approach was originally proposed in @xcite and extended in @xcite to gauge fields and gravity , to compute effective field theories of a brane - localized observer in the randall - sundrum scenario with two branes ( rs1 ) @xcite . as in that paper
, we will consider the compactification of ads on an orbifold @xmath11 with fixed points at @xmath12 ( the ultraviolet brane ) and at @xmath13 ( the infrared brane ) , where @xmath14 is the radial coordinate of ads . in the context of the full ads
/ cft correspondence , this geometry should be simply regarded as a regularization : the @xmath15 brane modifies ads near the boundary and the @xmath16 brane cuts the space in the direction of the horizon . in the rs1 scenario instead , two physical branes placed at the fixed points end the space . in @xcite the authors were interested in computing effective theories of an observer localized on the @xmath16 brane , while here we will use the brane running method to compute the on - shell action as a function of the fields at @xmath12 .
the basic idea is that the dependence of the on - shell value of the action on the fields at @xmath12 can be obtained integrating out bulk degrees of freedom . specifically , we consider theories with different locations of the @xmath16 brane .
effective brane actions are generated which guarantee that @xmath9 $ ] does not change . by moving the @xmath16 brane up to @xmath17 , by construction @xmath9 $ ]
equals the effective @xmath16 brane action . using ( [ adscft ] )
, the couplings of the effective @xmath16 brane action can be identified with the correlation functions of the cft .
the couplings on the effective brane `` run '' in a way that resembles an ordinary renormalization group ( rg ) flow . in particular , in the limit where the physical @xmath16 brane is at infinity , corresponding to the rs2 scenario @xcite , all the brane couplings flow to a non - local fixed point .
notice that the running of the brane couplings should not be confused with the rg flow of the observables in the dual field theory ( see for example @xcite ) , since we are keeping the cutoff , represented by the location of @xmath15 brane , fixed .
this formalism appears particularly appropriate for computations in the rs1 scenario . in the holographic interpretation ,
this model is dual to a cft with a cutoff @xmath18 , where conformal invariance is also broken in the infrared by the @xmath16 brane at the energy scale @xmath19 .
since conformal invariance is broken , the couplings on the effective brane are now not at their fixed points but , at energies @xmath20 , we can consider this as a small deviation from conformality .
brane actions can be included with ease , they simply modify the initial condition for the flow at @xmath13 .
the paper is organized as follows .
in section [ 2point ] we present the general techniques in the case of the @xmath21point function of a scalar field and reproduce well known results . in section
[ interactions ] we extend the procedure to @xmath22 and @xmath23point functions and consider the general structure of higher - point functions . in section [ rs ]
we apply the brane running method to holographic computations in the rs1 model .
section [ conclusions ] contains the conclusions .
a more detailed analysis of the @xmath21point function is left to the appendix .
we start considering a bulk scalar field propagating in an rs1-type geometry .
we will work in euclidean ads@xmath4 with the metric in poincar coordinates : @xmath24 where @xmath25 is the radius of ads@xmath4 .
the radial dimension @xmath14 is compactified on an orbifold @xmath11 with fixed points located at @xmath12 ( the ultraviolet brane ) and at @xmath13 ( the infrared brane ) .
one can think of this configuration as a regularization of the full ads , which can be recovered in the limit @xmath26 and @xmath27 .
the euclidean action of a free scalar of mass @xmath28 is : @xmath29 where : @xmath30 @xmath31 and @xmath32 are the induced metrics on the branes .
we will consider @xmath33 and @xmath34 to be derivative couplings .
these are the most general brane actions quadratic in the fields and compatible with @xmath35dimensional poincar invariance .
henceforth we take @xmath36 .
we systematically use the fourier transform of the field @xmath1 in the directions parallel to the brane : @xmath37 in momentum space , the equation of motion is given by : @xmath38\phi(p , z)=0 \label{massivescalar}\ ] ] the brane actions imply boundary conditions on the fields at @xmath12 and at @xmath13 . assuming that field @xmath1 is even on the orbifold we have : @xmath39 since we are interested in computing @xmath9 $ ] , we consistently impose dirichlet boundary conditions on the @xmath15 brane : @xmath40 in the ads / cft correspondence , the on - shell value of the action on ads as a function of the boundary values of the fields is the generating functional of the correlation functions of the gauge invariant operators of the cft . in particular ,
a bulk scalar field of mass @xmath28 is dual to a scalar operator of the cft with conformal dimension @xmath41 .
this case can be obtained by adding appropriate brane actions on the @xmath15 brane but we will not pursue this possibility here . ] one popular way to compute the on - shell action is to consider a dirichlet boundary problem at @xmath17 @xcite and use solutions of the wave - equation ( for the two @xmath21point function , @xmath9 $ ] depends only on the asymptotic behavior of the solutions near the boundary ; explicit wave - functions are necessary to compute higher - point functions ) .
we will now see how the dependence of the on - shell action on the boundary values of the fields can be obtained by flowing in the space of theories with different sizes of the extra - dimension . for applications to the ads / cft correspondence , we are only interested in the effective action seen by an observer living on the @xmath15 brane .
we consider a new theory with an effective @xmath16 brane located at @xmath42 .
requiring that the original theory and the theory with the effective @xmath16 brane at @xmath43 have precisely the same solutions in the overlapping region leaves any observable on the @xmath15 brane unchanged .
technically this program is accomplished including brane actions on the effective @xmath16 brane which provide the same boundary conditions at @xmath43 as the one implied by the original brane action at @xmath13 .
this construction is equivalent to requiring that @xmath9 $ ] is invariant . since the whole information about the cft
is contained in @xmath9 $ ] , the ads theory with the effective @xmath16 brane at @xmath43 describes the same cft . at the quadratic level
the necessary effective brane action is : @xmath44 a few remarks are in order .
the coupling @xmath45 has an explicit @xmath43 dependence and an implicit one from the momentum ( this follows from @xmath35dimensional poincar invariance ) .
if the original brane action at @xmath13 is local , i.e. it has an expansion polynomial in @xmath46 , the flow respects such a property .
however , when the physical @xmath16 brane is removed to infinity , as in the full ads / cft or in the rs2 model , non - local terms will also be generated .
this is easily understood from a field - theoretic point of view .
when ads extends to infinity we have a continuum of @xmath35dimensional particles which generates branch cuts in the effective action seen by an observer located on the @xmath15 brane .
these non - local terms are crucial since they reproduce the correlation functions of the cft at non - coincident points .
therefore we consider brane actions which contain local and non - local terms : @xmath47 @xmath48 has a polynomial expansion in momentum : @xmath49 where @xmath50 is the induced metric at @xmath51 .
the precise form of @xmath52 depends on the value of the bulk mass .
the flow of the brane couplings , as a function of the radial coordinate @xmath43 , can be cast in a form that resembles an ordinary rg flow . in the theory with the brane action ( [ braneaction ] ) , the boundary condition at @xmath43 is : @xmath53 in order to compute the differential flow of @xmath54 , we take the logarithmic derivative with respect to @xmath43 of ( [ bca2 ] ) .
since by hypothesis the solutions remain unchanged in the overlapping region , we use the bulk equation of motion to eliminate the second derivatives of the fields . requiring that the resulting equation is satisfied for any value of @xmath1
, we obtain the flow equation of @xmath54 : @xmath55 this is the main equation of this section .
it contains the same information about the cft as does the original action .
one can integrate this equation with the boundary condition : @xmath56 to obtain the couplings of the effective brane at @xmath43 .
solving ( [ flowequation ] ) allows one to compute the on - shell value of the action in terms of the values of the fields at @xmath12 .
explicitly , `` running '' the effective brane all the way down to the @xmath15 brane , @xmath9 $ ] is simply given by the sum of the @xmath15 and the effective brane actions , because the bulk contribution goes to zero .
assuming for simplicity that @xmath57 , we have from ( [ braneaction ] ) : @xmath58=s_b(z_{uv})= \frac { \pi } { z_{uv}^{d } } \int d^dp \delta(p+p ' ) \phi_0(p ' ) \lambda_2(z_{uv } ) \phi_0(p ) \label{scl2}\ ] ] the flow equation ( [ flowequation ] ) has a fixed point in the sense @xmath59 ( that is , the coupling depends on @xmath43 only through the dependence on momentum ) : @xmath60 where @xmath61 is the modified bessel function of order @xmath62 . in the limit where the @xmath16 brane is at infinity ,
the physics on the @xmath15 brane does not depend on the @xmath16 brane . in this case
( see appendix ) the coupling @xmath45 is a function of @xmath63 only : it is at its fixed point .
in other words , choosing the brane coupling at the fixed point ( [ fixedpoint ] ) is equivalent , from the point of view of an observer living on the @xmath15 brane , to removing the @xmath16 brane to infinity . in the appendix
we show that the flow equation ( [ flowequation ] ) has in general a line of fixed points , but these correspond to solutions of the wave - equation which are not regular at infinity , so they are rejected .
one can check that , in the limit @xmath64 and zero brane actions , @xmath45 approaches the fixed point ( [ fixedpoint ] ) .
according to the ads / cft prescription ( [ adscft ] ) , @xmath65 is , up to a normalization , the @xmath21point function of a scalar operator of the cft of dimension @xmath66 ( in the large @xmath7 limit and strong t hooft parameter ) . in fact , with @xmath54 at the fixed point ( [ fixedpoint ] ) , the brane action evaluated at @xmath12 coincides up to a factor of two ( coming from working in the orbifold covering space ) with the standard ads / cft on - shell action computed , for example , in @xcite .
the expansion of ( [ fixedpoint ] ) for integer @xmath67 is : @xmath68 where the coefficient @xmath69 depends on @xmath67 only . for non - integer @xmath67 one
finds a similar expansion with first non - analytic term @xmath70 .
notice that , as promised , the brane action necessary to reproduce the infinite space limit is non - local . in ads
/ cft computations one then proceeds taking the limit @xmath71 in ( [ scl2 ] ) .
the divergent local terms in @xmath9 $ ] need to be renormalized away @xcite .
the non - local term @xmath72 reproduces the @xmath21point function at non - coincident points of a scalar operator with conformal dimension @xmath66 .
higher - point functions can be computed in a very similar way to the @xmath21point functions , by integrating out bulk degrees of freedom . as an example we will consider the case of cubic interactions .
the bulk action in momentum space contains : @xmath73 where @xmath74 are scalar fields of mass @xmath75 and @xmath76 is symmetric in the indexes .
the bulk equation of motion reads : @xmath77 \phi_i(p , z)-3 \sigma^{ijk } \int d^dq \phi_j(p - q , z)\phi_k(q , z)=0\ ] ] in this case the effective brane action also contains cubic and higher order terms : @xmath78 the boundary condition at @xmath43 becomes : @xmath79 where @xmath80 is the quadratic brane coupling of the field @xmath81 . to compute the flow equation for the coupling @xmath82
, we proceed by taking the logarithmic derivative of ( [ bca3 ] ) , as in the case of the @xmath21point function .
we then use the bulk equation of motion and require that the equation is satisfied for any value of the fields . with a cubic brane action this can be done up to order @xmath83 in the fields . of course
, this is because a cubic interaction generates higher - point functions as well .
the important point is that , at the classical level , lower order point functions do not mix with higher orders ones , so the flow of the coupling @xmath84 depends only on the couplings @xmath85 with @xmath86 .
the equation for @xmath82 is : @xmath87 \label{threepoint}\ ] ] while this equation can be difficult to solve in general , it can be solved perturbatively expanding in powers of momenta , as explained in the appendix for the case of the @xmath21point function . in the infinite ads case , each brane coupling should be set at the fixed point , so they do not depend explicitly on @xmath43 .
one can check that the fixed point in this case is given by : @xmath88 notice that , since the @xmath89function for @xmath82 is linear in @xmath82 , the fixed point is completely determined by @xmath90 . the fixed point ( [ fp3 ] )
is just the @xmath22point function in the limit @xmath27 , computed with bulk - to - boundary propagators @xcite .
the method of brane running is completely general and can be extended to arbitrary order point functions .
the computation of @xmath23 and higher - point functions is complicated in the ads / cft correspondence because it involves bulk - to - bulk ads propagators . in our case ,
the structure of the differential equations for the higher - point functions is similar to the case of the @xmath22point function and can be readily understood in a diagrammatic way . in order to compute the @xmath23point function
we need to include a quartic brane action : @xmath91 following the same steps as for the @xmath22point function , one obtains : @xmath92+\nonumber \\ & & + \frac 9 8 [ \lambda_3^{imk}(-p , p - q , q)\lambda_3^{mjl}(q - p , p - q - k , k)+\lambda_3^{ijm}(-p , p - q - k , q+k ) \lambda_3^{mlk}(-q - k , q , k)]\nonumber \\ \label{4fp}\end{aligned}\ ] ] in the previous equation , the momentum dependence of the coupling is @xmath93 .
the fixed point of ( [ 4fp ] ) is , up to normalization , the amplitude for the @xmath23point function in the ads / cft which can be expressed in terms of bulk - to - bulk and bulk - to - boundary propagators .
we now analyze the general structure of the @xmath89functions for the couplings .
we consider in particular the @xmath23point function but the same considerations hold for higher - point functions .
schematically , the diagrams contributing to the @xmath23point function when brane actions are included are represented in the figure below .
= 1.5 in = 1.5 in = 1.35 in to move the brane from @xmath43 to @xmath94 , we need to provide counter - terms for the degrees of freedom which have been integrated out . as a consequence ,
the infinitesimal variation of @xmath95 contains a contribution proportional to @xmath96 , a contribution proportional to @xmath97 as well as a contribution linear in @xmath95
. a quartic bulk interaction would act as a source in the @xmath89function of @xmath95 .
the dependence on the momenta can be worked out as in ordinary feynman diagrams .
in this section we apply the brane running method to the randall - sundrum scenario @xcite .
we take @xmath98 . in the full ads
/ cft correspondence , a theory on ads is equivalent to a pure conformal field theory living on the boundary . in the previous analysis the @xmath15 brane and the @xmath16 brane should be understood as regulators of the theory so the physical quantities are extracted in the limit @xmath71 and @xmath99 . as we have argued , the second limit ( corresponding to the geometry of the rs2 model ) implies that all the couplings on the effective brane are at their fixed points . in the rs1 scenario , five dimensional ads is truncated by two physical branes parallel to the boundary , placed at the fixed points of an orbifold . since small values of the coordinate @xmath14 correspond to the ultraviolet of the cft , the presence of the @xmath15 brane represents a cutoff applied to the cft .
on the other hand , the @xmath16 brane modifies the space only at large @xmath14 so naively it corresponds , in the holographic picture , to breaking ( spontaneously ) conformal invariance in the infrared , at a scale @xmath19 .
it should be clear from the present paper that this conclusion may not hold when brane actions are included : theories with different size of the extra - dimension can have the same cft dual . since for phenomenological applications one takes @xmath100 , the theory should appear approximately conformal for a large range of energies , namely @xmath101 . to any order in the fields ,
the dependence of the action on the boundary values of the fields can be computed integrating the flow equations presented in the previous sections .
this can be done perturbatively in @xmath46 as shown in the appendix for the @xmath21point function . in particular
it is straightforward , even for higher - point functions , to compute the zero momentum dependence of the coupling @xmath102 .
the leading dependence on the external momentum is also not hard to compute .
since the @xmath16 brane breaks conformal invariance in the infrared , the brane couplings are not at a fixed point , but acquire an explicit dependence on the scale @xmath43 . in the case of the @xmath21point function , equation ( [ flowequation ] ) can be solved explicitly , the solution being : @xmath103 the integration constant @xmath104 is determined by the boundary condition at the @xmath16 brane , namely @xmath105 .
this allows the inclusion in a simple and systematic way of arbitrary brane actions .
for example , for a massless scalar and @xmath106 , corresponding to ordinary neumann boundary condition , the result is : @xmath107 for @xmath108 this is just the analytic continuation to euclidean ads of the inverse of the brane - to - brane propagator of a massless scalar computed in @xcite . in general
, @xmath65 is the inverse of the brane - to - brane propagator , so the zeros in @xmath46 of @xmath65 correspond to the masses of the kaluza - klein ( kk ) modes . for @xmath109 one can expand @xmath45 in powers of @xmath46 .
the expansion is now completely polynomial in momentum .
technically this happens because if the boundary condition is local , this property will be preserved by the flow . from the cft point of view , this fact is simply the statement that below the scale @xmath19 , conformal symmetry is completely broken , and in particular , a mass gap is generated .
on the ads side we have a discrete set of @xmath110 particles whose mass scale is set by @xmath19 . , i.e. tachyons satisfying the lower bound on the masses of scalars in ads@xmath4 , the kk spectrum also contains a single tachyon whose mass squared is @xmath111 @xcite . ] this discrete set of particles gives rise to isolated poles in the propagator but no branch cuts .
therefore , at energies below the masses of the kk modes , this system should be describable in terms of a local action . at energies above @xmath19
, conformal invariance is approximately restored . for @xmath112 ,
i.e. energies below the cutoff of the cft , we can expand @xmath65 around @xmath113 . in the case of the massless scalar ( [ 2pirs ] )
we have : @xmath114 in the limit @xmath27 we recover the expansion of the fixed point .
this should be expected since in this limit all the couplings flow to their fixed points . using the asymptotic expansion of the bessel functions , it is easy to see that the deviation from @xmath90 becomes exponentially small at energies greater that @xmath19 . since for large momenta the presence of the @xmath16 brane represents a small deviation from conformality , all the brane couplings are close to the fixed point . in this regime
it is useful to simplify the flow equation expanding to linear order around @xmath90 : @xmath115 where @xmath116 . to leading order , for @xmath117 ,
the solution of the previous equation is simply @xmath118 , independently on @xmath67 . as we move the effective brane to smaller @xmath43 , @xmath119 becomes exponentially suppressed .
similar manipulations can be done for higher - point functions .
localized fields can be naturally included in this formalism .
this case is important for phenomenological applications .
for example , in the original proposal by randall and sundrum , in order to solve the hierarchy problem , the standard model fields were localized on the @xmath16 brane . for fields localized on the @xmath15 brane
, it is obvious that their action should be added to the effective action . from the cft point of view ,
fields living on the @xmath15 brane are interpreted as fields external to the cft and coupled to gauge invariant operators of the cft so it is natural to add the action of the cft and that of the external fields . in the case of fields localized on the @xmath16 brane , or in the bulk , the analysis is more intricate because these fields couple only indirectly to the @xmath15 brane . as an example , following @xcite , we consider a free scalar field coupled to an external source @xmath120 localized at @xmath121 . in @xmath122 dimensions ,
the ads action contains : @xmath123 consistency requires the brane action : @xmath124 the running of @xmath125 is given by : @xmath126 while the @xmath89function of @xmath54 remains unchanged .
notice that only in this particular case a coupling depends on higher order ones.function of the coupling @xmath84 will depend on the couplings up to @xmath127 . ] for a localized source in the case where the @xmath16 brane is at infinity , @xmath54 will still be at its fixed point ( [ fixedpoint ] ) .
it is not hard to check that the solution of ( [ onepoint ] ) is : @xmath128 a tadpole for the field @xmath1 is generated in the effective field theory of an observer on the @xmath15 brane . on the cft side
, this corresponds to an expectation value for the dual operator .
in this paper we have shown how to compute the value of the on - shell ads action of a scalar field by flowing in the space of theories with a different size for the extra - dimension . by integrating out bulk degrees of freedom , we generate effective brane actions which reproduce exactly the same physics in the restricted region . to any order in the fields ,
the dependence of the on - shell action on the boundary values of the fields at @xmath12 is obtained by solving the flow equations for the couplings on the effective brane . in principle
, this procedure does not require the knowledge of the bulk - to - boundary and bulk - to - bulk propagators .
the values of the couplings are related in a simple way to the correlation functions of the cft . in the case of infinite ads or rs2 ,
the brane couplings are at their fixed points , corresponding to an unbroken cft .
this formalism is particularly appropriate for holographic computations in the rs1 model . here
the couplings on the effective brane are not at the fixed points but acquire an explicit dependence on the radial coordinate . at energies much above the scale set by the @xmath16 brane , corresponding to the scale where conformal invariance is broken , the couplings are very close to the fixed points so we can simplify the flow equations .
gauge fields and fluctuations of the metric can be included in a very similar way .
the relevant steps are presented in @xcite , where the brane running is used to compute the effective field theory of an observer living on the @xmath16 brane .
the extension to fermionic fields is non - trivial because the wave - equation is a first order differential equation in this case .
more in general , the method of brane running can be used in geometries which contain a poincar invariant subspace .
the structure of the @xmath89functions for the couplings would be similar to the one in the present paper . in this case
though , the brane couplings , even in the infinite space limit , will not be at a fixed point .
i would like to thank kaustubh agashe and jon bagger for reading the manuscript .
this work was supported in part by the u.s
. national science foundation , grant nsf - phy-9970781 and nsf - phy-0099468 .
in this appendix we give a self - contained derivation of the flow equation for the coupling @xmath54 and carefully study the solutions .
the boundary condition at @xmath43 determined by the brane action is : @xmath129 the bulk equation of motion is : @xmath130 we are interested in the differential flow of the coupling @xmath54 . by assumption
the solution of the equations of motion do not depend on the location of the brane .
this physical condition determines a ( classical ) flow for @xmath54 . taking the logarithmic derivative of ( [ bcapp ] ) with respect to @xmath43 we obtain : @xmath131 using the bulk equation of motion to eliminate the second derivative and requiring that this new equation is satisfied for any value of the field @xmath132 , we find the flow equation : @xmath133 the solutions of this equation are functions of @xmath63 .
the boundary condition at @xmath13 introduces a dependence on @xmath134 . a quick way to find
the solutions of ( [ flowapp ] ) is to use explicit wave - functions and substitute in ( [ bcapp ] ) .
the solution of the bulk equation of motion is : @xmath135 where @xmath136 and @xmath61 are bessel s functions of order @xmath62 . plugging in ( [ bcapp ] ) and using properties of the bessel s functions we find : @xmath137 the coefficients @xmath138 and @xmath139 have a dependence on @xmath140 determined by the @xmath16 brane . particularly important for the applications to the ads / cft
are the solutions of this equation which depend on @xmath141 only .
we call these solutions fixed points because the couplings in the expansion of @xmath54 are independent on @xmath43 in this case . from ( [ fpapp ] ) we see that the flow equation has a line of fixed points depending on the ratio @xmath142 ( with @xmath142 independent of @xmath46 ) . setting @xmath54 at a fixed point is physically equivalent , for an observer living on the @xmath15 brane , to removing the @xmath16 brane to infinity
. regularity at infinity selects the wave - function @xmath61 so the relevant fixed point corresponds to @xmath143 . for illustrative purposes we now show an alternative method to compute the fixed points of ( [ flowapp ] ) .
the same technique could be used to solve the equations for the higher order couplings or to study the @xmath21point functions in backgrounds which are only asymptotically ads .
equation ( [ flowapp ] ) can be solved iteratively expanding in powers of @xmath46 .
notice that for the fixed points the momentum expansion is the same as the expansion around @xmath144 , that is a near boundary analysis . for simplicity
, we will consider the case of @xmath67 integer which is the most common in the string theory literature .
we start with the following ansatz for the solution : @xmath145 substituting in the flow equation we obtain a set of coupled differential equations for the coefficients @xmath146 : @xmath147 for the fixed points all the derivatives are equal to zero so we have a set of algebraic equations .
the solution of the first equation in ( [ flowexpapp ] ) is : @xmath148 the solution with the @xmath149 sign corresponds to the fixed point with @xmath136 only .
this solution has an expansion purely polynomial in momentum . in position space
this generates only contact interactions . with the choice of the @xmath150 sign we can solve ( [ flowexpapp ] ) iteratively only up to the order @xmath151
this equation is inconsistent . in order to find a solution we need to include a term @xmath152 in the expansion of @xmath54 .
now the equation @xmath153 determines the coefficient @xmath69 while @xmath154 remains unconstrained .
the fact that @xmath154 is undetermined is of course related to possibility of choosing @xmath142 in ( [ fpapp ] ) .
s. de haro , s. n. solodukhin and k. skenderis , commun . math .
* 217 * , 595 ( 2001 ) [ arxiv : hep - th/0002230 ] .
m. bianchi , d. z. freedman and k. skenderis , jhep * 0108 * , 041 ( 2001 ) [ arxiv : hep - th/0105276 ] .
m. bianchi , d. z. freedman and k. skenderis , nucl .
b * 631 * , 159 ( 2002 ) [ arxiv : hep - th/0112119 ] .
h. verlinde , nucl .
b 580 , 264 ( 2000 ) [ arxiv : hep - th/9906182 ] ; j. maldacena , unpublished remarks ; e. witten , itp santa barbara conference ` new dimensions in field theory and string theory ' , http://www.itp.ucsb.edu/online/susy_c99/discussion ; s. gubser , phys . rev .
d 63 , 084017 ( 2001 ) [ arxiv : hep - th/9912001 ] ; e. verlinde and h. verlinde , jhep 0005 , 034 ( 2000 ) [ arxiv : hep - th/9912018 ] ; n. arkani - hamed , m. porrati and l. randall , jhep 0108 , 017 ( 2001 ) [ arxiv : hep - th/0012148 ] ; r. rattazzi and a. zaffaroni , jhep 0104 , 021 ( 2001 ) [ arxiv : hep - th/0012248 ] . | we extend the formalism introduced in the paper hep - th/0209050 to compute correlation functions in the ads / cft correspondence .
we show how the on - shell action of a scalar field in a compactification of ads space can be obtained by flowing in the space of ( non - local ) theories with different sizes of the extra - dimension .
our method is relevant in particular for holographic computations in the randall - sundrum scenario with one or two branes and it allows the inclusion of brane - localized actions in a systematic way .
the method can also be generalized to other backgrounds and does not rely on explicit knowledge of the solutions of the wave - equations . |
in the recent paper @xcite , the following refinement of the grss inequality was proved .
[ t1.1]let @xmath0 be an inner product space over @xmath1 @xmath2 and @xmath3 @xmath4 if @xmath5 are real or complex numbers and @xmath6 are vectors in @xmath7 such that either@xmath8or , equivalently , @xmath9hold , then we have the following refinement of the grss inequality@xmath10 ^{\frac{1}{2}}\left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{\frac{1}{2 } } \notag \\ & \leq \frac{1}{4}\left\vert \phi -\phi \right\vert \left\vert \gamma -\gamma \right\vert .
\notag\end{aligned}\]]the constant @xmath11 is best possible in both inequalities .
note that the inequality between the first and last term in ( [ 1.3 ] ) was firstly established in @xcite .
a generalisation of the above result for finite families of orthornormal vectors has been pointed out in @xcite .
[ t1.2]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath15 @xmath16 @xmath17 @xmath18 and @xmath19 if@xmath20or , equivalently,@xmath21hold , then we have the inequalities@xmath22 ^{\frac{1}{2}}\left [ \func{re}% \left\langle \sum_{i\in f}\gamma _ { i}e_{i}-y , y-\sum_{i\in f}\gamma _ { i}e_{i}\right\rangle \right ] ^{\frac{1}{2 } } \notag \\ &
\leq & \frac{1}{4}\left ( \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}\cdot \left ( \sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}. \notag\end{aligned}\]]the constant @xmath11 is best possible .
we note that the inequality between the first term and the last term for real inner products under the assumption ( [ 1.4 ] ) , has been proved by n. ujevi in @xcite .
the following corollary provides a counterpart for the well known bessel s inequality in real or complex inner product spaces ( see also @xcite ) .
[ c1.3]with the above assumptions for @xmath23 @xmath24 and @xmath25 one has the inequalities@xmath26with @xmath11 as the best possible constant in both inequalities .
it is the main aim of this paper to point out a different counterpart for the bessel and grss inequalities stated above .
some related results are also outlined .
the following lemma holds .
[ l2.1]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath15 @xmath27 @xmath18 , @xmath28 and @xmath29 if@xmath30then we have the inequality@xmath31 consider@xmath32and@xmath33 \\ & = \sum_{i\in f}\left\vert \lambda _ { i}\right\vert ^{2}+\sum_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}-\sum_{i\in f}% \overline{\lambda _ { i}}\left\langle x , e_{i}\right\rangle -\sum_{i\in f}\lambda _ { i}\overline{\left\langle x , e_{i}\right\rangle } .\end{aligned}\]]if we subtract @xmath34 from @xmath35 we deduce an equality that is interesting in its own right@xmath36from which we easily deduce ( [ 2.2 ] ) .
the following counterpart of bessel s inequality holds .
[ t2.2]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath15 @xmath37 @xmath38 @xmath39 real or complex numbers . for a @xmath40 if either 1 .
@xmath41 + or , equivalently , 2 .
@xmath42 holds , then the following counterpart of bessel s inequality@xmath43is valid .
the constant @xmath11 is best possible in both inequalities .
firstly , we observe that for @xmath44 the following are equivalent@xmath45and@xmath46now , for @xmath47 @xmath48 we have@xmath49giving , for @xmath50 the desired equivalence .
now , if we apply lemma [ l2.1 ] for @xmath51 and @xmath52we deduce the first inequality in ( [ 2.4 ] ) .
let us prove that @xmath11 is best possible in the second inequality in ( [ 2.4 ] ) .
assume that there is a @xmath53 such that@xmath54provided that @xmath37 @xmath55 and @xmath14 satisfy ( i ) and ( ii ) .
we choose @xmath56 @xmath57 @xmath58 @xmath59 @xmath60 @xmath61 to get from ( [ 2.7 ] ) that@xmath62provided@xmath63from ( [ 2.8 ] ) we get@xmath64provided ( [ 2.9 ] ) holds .
if we choose @xmath65 @xmath66 then ( [ 2.9 ] ) is fulfilled and by ( [ 2.10 ] ) we get @xmath67 giving @xmath68 [ r2.3]if @xmath56 @xmath69 @xmath70 and for @xmath71 and @xmath72 one has either@xmath73or , equivalently,@xmath74then@xmath75the constant @xmath11 is best possible in both inequalities .
[ r2.4]it is important to compare the bounds provided by corollary c1.3 and theorem [ t2.2 ] . for this purpose ,
consider@xmath76and@xmath77where @xmath7 is a real inner product , @xmath3 @xmath78 @xmath79 @xmath80 with @xmath81or equivalently , @xmath82 if we choose @xmath83 @xmath84 then we have@xmath85provided @xmath86 choose @xmath87 with @xmath88 then we get@xmath89which shows that @xmath90 if @xmath91 and @xmath92 if @xmath93 we may state the following proposition . [ p2.5]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath15 @xmath37 @xmath94 @xmath95 .
if @xmath72 either satisfies ( i ) , or , equivalently , ( ii ) of theorem [ t2.2 ] , then the upper bounds@xmath96for the bessel s difference @xmath97 can not be compared in general .
the following result holds .
[ t3.1]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath98 , @xmath99 @xmath100 and @xmath19 if either@xmath101or , equivalently,@xmath102hold , then we have the inequalities@xmath103the constant @xmath11 is best possible .
using schwartz s inequality in the inner product space @xmath104 one has@xmath105and since a simple calculation shows that @xmath106and @xmath107for any @xmath108 then by ( [ 3.4 ] ) and by the counterpart of bessel s inequality in theorem [ t2.2 ] , we have@xmath109 \notag \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \times \left [ \frac{1}{4}\sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}-\sum_{i\in f}\left\vert \frac{\gamma _ { i}+\gamma _ { i}}{2}-\left\langle y , e_{i}\right\rangle \right\vert ^{2}\right ] \notag \\ & : = k .
\notag\end{aligned}\]]using aczl s inequality for real numbers , i.e. , we recall that@xmath110provided that @xmath111 @xmath112 we may state that@xmath113 ^{2}.\end{gathered}\]]using ( [ 3.5 ] ) and ( [ 3.7 ] ) we conclude that@xmath114 ^{2}.\end{gathered}\]]taking the square root in ( [ 3.8 ] ) and taking into account that the quantity in the last square brackets is nonnegative ( see for example ( 2.4 ) ) , we deduce the second inequality in ( [ 3.3 ] ) . the fact that @xmath11 is the best possible constant follows by theorem [ t2.2 ] and we omit the details .
the following corollary may be stated .
[ c3.2]let @xmath3 @xmath78 @xmath115 and @xmath116 such that either@xmath117or , equivalently,@xmath118then we have the following refinement of grss inequality@xmath119the constant @xmath11 is best possible in both inequalities .
the following companion of the grss inequality also holds .
[ t4.1]let @xmath12 be a family of orthornormal vectors in @xmath13 @xmath14 a finite part of @xmath98 and @xmath120 , @xmath116 and @xmath121 such that either@xmath122or , equivalently,@xmath123holds .
then we have the inequality@xmath124 \label{4.3 } \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \sum_{i\in f}\left\vert \phi
_ { i}-\phi _ { i}\right\vert ^{2 } \notag \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\frac{1}{4}\frac{1}{\lambda \left ( 1-\lambda \right ) } \sum_{i\in f}\left\vert \frac{\phi _ { i}+\phi _ { i}}{2}-\left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2 } \notag \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \sum_{i\in f}\left\vert \phi
_ { i}-\phi _ { i}\right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath125 is the best possible constant in ( [ 4.3 ] ) in the sense that it can not be replaced by a smaller constant . we know that for any@xmath126 one has@xmath127then for any @xmath128 and @xmath129 one has@xmath130since@xmath131for any @xmath108 then , by ( [ 4.4 ] )
, we get@xmath124 \label{4.5 } \\ & = \func{re}\left [ \left\langle x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i},y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right\rangle \right ] \notag \\ & \leq \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda \left ( x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i}\right ) + \left ( 1-\lambda \right ) \left ( y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right ) \right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda x+\left ( 1-\lambda \right ) y-\sum_{i\in f}\left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle e_{i}\right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left [ \left\vert \lambda x+\left ( 1-\lambda \right ) y\right\vert ^{2}-\sum_{i\in f}\left\vert \left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2}\right ] .
\notag\end{aligned}\]]if we apply the counterpart of bessel s inequality in theorem [ t2.2 ] for @xmath132 we may state that@xmath133now , by making use of ( [ 4.5 ] ) and ( [ 4.6 ] ) , we deduce ( [ 4.3 ] ) . the fact that @xmath125 is the best possible constant in ( [ 4.3 ] )
follows by the fact that if in ( [ 4.1 ] ) we choose @xmath134 then it becomes ( i ) of theorem [ t2.2 ] , implying for @xmath135 ( [ 2.4 ] ) , for which , we have shown that @xmath11 was the best constant .
[ r1]in practical applications we may use only the inequality between the first and the last term in ( [ 4.3 ] ) . [ r2]if in theorem [ t4.1 ] , we choose @xmath136 then we get@xmath124 \label{4.7 } \\ & \leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}-\sum_{i\in f}\left\vert \frac{\phi
_ { i}+\phi _ { i}}{2}-\left\langle \frac{x+y}{2},e_{i}\right\rangle \right\vert ^{2 } \notag \\ &
\leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2 } , \notag\end{aligned}\]]provided@xmath137or , equivalently,@xmath138 we note that ( [ 4.7 ] ) is a refinement of the corresponding result in ssd3 .
[ c4.2]with the assumptions of theorem [ t4.1 ] and if@xmath139 or , equivalently@xmath140then we have the inequality@xmath141 \right\vert \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}. \label{4.10}\]]the constant @xmath125 is best possible in ( [ 4.10 ] ) .
[ r4.3]if @xmath7 is a real inner product space and @xmath142 with the property@xmath143or , equivalently,@xmath144 ^{\frac{1}{2 } } , \label{4.12}\]]then we have the grss type inequality@xmath145
let @xmath146 be a measure space consisting of a set @xmath147 a @xmath148algebra of parts @xmath149 and a countably additive and positive measure @xmath150 on @xmath149 with values in @xmath151 let @xmath152 be a @xmath153measurable function on @xmath154 denote by @xmath155 the hilbert space of all real or complex valued functions defined on @xmath156 and @xmath157integrable on @xmath147 i.e.,@xmath158 consider the family @xmath159 of functions in @xmath160 with the properties that@xmath161where @xmath162 is @xmath163 if @xmath164 and @xmath165 if @xmath166 @xmath167 is an orthornormal family in @xmath168 [ p5.1]let @xmath159 be an orthornormal family of functions in @xmath169 @xmath14 a finite subset of @xmath15 @xmath170 @xmath171 and @xmath172 so that either@xmath173 d\mu \left ( s\right ) \geq 0 \label{5.3}\]]or , equivalently,@xmath174then we have the inequality@xmath175the constant @xmath11 is best possible in both inequalities .
[ p5.2]let @xmath159 and @xmath14 be as in proposition [ p5.1 ] . if @xmath177 @xmath95 and @xmath178 so that either@xmath179 d\mu \left ( s\right ) & \geq 0 , \label{5.6 } \\
\int_{\omega } \rho \left ( s\right ) \func{re}\left [ \left ( \sum_{i\in f}\gamma _ { i}f_{i}\left ( s\right ) -g\left
( s\right ) \right ) \left ( \overline{% g}\left ( s\right ) -\sum_{i\in f}\overline{\gamma _ { i}}\text { } \overline{f_{i}% } \left ( s\right ) \right ) \right ] d\mu \left ( s\right ) & \geq 0 , \notag\end{aligned}\]]or ,
equivalently,@xmath180then we have the inequalities@xmath181@xmath182the constant @xmath11 is the best possible .
[ c5.4]let @xmath159 be an orthornormal family of functions in the real hilbert space @xmath183 @xmath14 a finite part of @xmath15 @xmath184 @xmath171 and @xmath185 so that@xmath186then we have the inequalities@xmath187 ^{2 } \label{5.10 } \\
& \leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}-\sum_{i\in f}% \left [ \frac{m_{i}+m_{i}}{2}-\int_{\omega } \rho \left ( s\right ) f\left ( s\right ) f_{i}\left ( s\right ) d\mu \left ( s\right ) \right ] ^{2 } \notag \\ & \leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}. \notag\end{aligned}\]]the constant @xmath11 is best possible .
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , rgmia res .
, 6(2003 ) , supplement , article 10 . `
[ on line http://rgmia.vu.edu.au/v6(e).html ] ` | a new counterpart of bessel s inequality for orthornormal families in real or complex inner product spaces is obtained . applications for some grss type results
are also provided . |
the quasi two - dimensional electron system ( 2des ) in high mobility gaas / algaas heterostructures has served to identify a new type of zero - resistance state in the 2des , one which occurs at high filling factors or low magnetic fields under microwave photo - excitation .
@xcite such zero - resistance states have been fascinating in part because they could help to point out the necessary conditions for obtaining vanishing resistance in the 2des in a magnetic field .
the associated oscillatory effect is also interesting from the application perspective since it could lead to new approaches for a frequency and power sensitive radiation detector in the microwave and terahertz bands , for a large number of applications.@xcite numerous experimental @xciteand theoretical@xcite works about the radiation - induced magnetoresistance oscillations ( rimo ) and associated zero - resistance states have been published over the past decade . at the present
, it is understood that rimos are `` @xmath4-cycle phase - shifted '' in @xmath5 , and the oscillatory minima occur in @xmath1 vs. @xmath6 plots broadly about @xmath7b_f$ ] , where @xmath8 , @xmath9 is the microwave frequency , @xmath10 is the effective electron mass in gaas and @xmath11 .
the theoretically proposed physical mechanisms for rimos include the displacement model@xcite , the nonparabolicity model@xcite , the inelastic model@xcite and the radiation - driven electron - orbit model@xcite .
these different theories have suggested dissimilar behavior of some physical properties such as , for example the polarization - angle - dependence and the power - dependence of the rimos . as a result , an interesting issue concerns the sensitivity of the rimos to the polarization angle of linearly polarized microwaves .
early linear microwave polarization sensitivity work carried out on l - shaped samples@xcite showed that the the frequency and the phase of rimos are insensitive to the linear polarization angle of the microwaves .
later work indicated that rimos were insensitive to circularly and linearly polarized microwaves in square - shaped samples in a quasioptical measurement@xcite .
more recently , a polarization - angle - dependence in the amplitude of rimos has been demonstrated@xcite .
the results were roughly consistent with the predictions of the displacement model , the nonparabolicity model , and the radiation - driven electron - orbit model for @xmath12 , where @xmath13 is damping factor and @xmath14.@xcite finally , ramanayaka _ et al .
_ showed that the @xmath1 varied sinusoidally vs. @xmath0 at low microwave power , following the empirical relation @xmath15 , where @xmath0 is microwave polarization angle , @xmath16 is phase shift , and the plus and minus signs corresponded to the oscillatory maxima and minima , respectively.@xcite the results suggested both a @xmath9-dependence and a @xmath6-dependence in @xmath2 , although the phase shifts did not appear to be systematically responsive to any experimental parameters .
these studies indicated that the observed phase shifts required further experimental investigation .
thus , we extract the frequency dependence of the phase shift in the @xmath1 vs. @xmath0 results by applying an analysis that averages over other smaller contributions , when those contributions are smaller than estimates of the experimental uncertainty .
the results suggest a non - vanishing frequency dependence in the phase shift , i.e. , @xmath17 , over the frequency interval @xmath18 ghz .
the experimental setup for the polarization - orientation measurements is shown in fig .
linearly polarized microwaves are generated by an antenna inside a rotatable microwave launcher , and they are transmitted via a cylindrical waveguide to the sample .
the samples consist of 400-@xmath19-wide hall bar with alloyed gold - germanium contacts fabricated from gaas / algaas heterojunctions with a 2des with carrier density @xmath20 and mobility @xmath21 .
the samples are immersed in pumped liquid helium to maintain a temperature of 1.5 k during the measurements .
standard four - terminal lock - in techniques are utilized to measure the diagonal resistance @xmath1 .
finally , the polarization angle @xmath0 is defined as the angle between the microwave electric field @xmath22 and the hall bar axis .
thus , in experiment , the gradual increase of @xmath0 from @xmath23 to @xmath24 is achieved by rotating the microwave launcher . at the outset to the experiment
, a power detector is connected to a power meter and placed at the end of cylindrical waveguide , see fig .
1(c ) inset .
this power detector is sensitive to the radiation field along its preferred axis .
the orientation of this detector is fixed parallel to the antenna , setting @xmath25 .
then , the antenna is turned from @xmath23 to @xmath24 at @xmath26 increment for a number of different frequencies , @xmath9 , from @xmath27 ghz to @xmath28 ghz , to characterize the linear polarization angle of the microwaves at the bottom of the waveguide sample holder .
1(c ) shows the normalized detected power as a function of @xmath0 at @xmath29 ghz .
as expected for linearly polarized microwaves , the detected power varies sinusoidally with @xmath0 , and this sinusoidal curve can be described by @xmath30 , where @xmath31 is the detected power , @xmath32 is the dark response without microwaves , and @xmath33 is the amplitude of cosine square function . the cosine square function has been used here because the microwave power is proportional to the square of the microwave electric field .
such data fits to this sinusoidal function were used to extract a phase shift , @xmath16 , in the absence of a sample to characterize the polarization angle in the measurement setup .
1(c ) demonstrates good fit between detector response and the sinusoidal function , as the fit indicates that @xmath34 at @xmath35 ghz .
this same method was used to extract @xmath16 at a number of frequencies .
these fit - extracted @xmath16 of the bare sample holder plus detector have been plotted as a function of @xmath9 in fig .
1(d ) , which indicates that @xmath36 for @xmath18ghz . here
, one expects that @xmath37 if the microwave antenna is well aligned with the microwave detector when the @xmath38 condition is defined at each @xmath9 .
thus , the observed spread in these @xmath2 , which is @xmath39 , is attributed to experimental issues such as misalignment and readout errors .
is defined as the angle between the microwave electric field @xmath22 and the hall bar axis .
( c ) the sinusoidal normalized detected power as a function of @xmath0 , with a power detector in place of the sample , can be fit by a cosine square function in order to extract @xmath16 .
( d ) a plot of @xmath16 vs @xmath9 , without the sample , implies that the measurement uncertainty in @xmath16 is below @xmath40.,width=283 ] with ( photo - excited ) and without ( dark ) microwave illumination for @xmath41 t , at @xmath42 and @xmath38 . here
, panel ( a ) shows the @xmath1 measured from the left ( l ) pair of contacts while panel ( b ) shows the @xmath1 measured from the right ( r ) pair of contacts on the hall bar .
panels ( c)-(h ) exhibit oscillatory photo - excited and dark @xmath1 vs. the polarization angle @xmath0 , with extracted @xmath16 obtained from the fit curve at ( c ) @xmath43 ( d ) @xmath44 ( e ) @xmath45 ( f ) @xmath46 ( g ) @xmath47 ( h ) @xmath48 respectively .
panels ( c)-(h ) show that the fit extracted @xmath16 at a given extrema of the mimos have similar values for contact pairs on opposite sides ( ( l ) and ( r ) ) of the device with the phase shift difference being smaller than estimated measurement uncertainty ( @xmath49).,width=283 ] in the next experimental phase , @xmath6-field sweeps of @xmath1 with microwave ( photo - excited @xmath1 ) and without microwave excitation ( dark @xmath1 ) were carried out from @xmath50 t to @xmath51 t at a number of frequencies
2(a ) , and 2(b ) both show photo - excited and dark @xmath1 vs @xmath6 with @xmath52 ghz , @xmath53 at @xmath25 for @xmath54 t . here , fig .
2(a ) ( labeled with the superscript sign of l ) represents measurements from the contact pair on the left side of the hall bar , while fig .
2 ( b ) ( labeled with the superscript sign of r ) represents measurements from the right side of the hall bar . both fig .
2(a ) and 2(b ) exhibit perceptible rimos below @xmath55 t. thus , in fig .
2(a ) and ( b ) , the predominant oscillatory extrema of rimos have been labeled as @xmath56 , @xmath57 , and @xmath58 , to indicate the first peak , the first valley , and the second peak , respectively , of rimos for @xmath59 .
2(c)-(h ) display dark and photo - excited @xmath1 as a function of @xmath0 at fixed @xmath6 corresponding to @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , and @xmath65 .
note that the photo - excited @xmath1 varies sinusoidally with @xmath0 , and the dark @xmath1 maintains a constant value vs. @xmath0 . with ( photo - excited ) and without ( dark ) microwave illumination for @xmath41 t , at @xmath42 and @xmath38 .
here , panel ( a ) shows the @xmath1 measured from the left ( l ) pair of contacts while panel ( b ) shows the @xmath1 measured from the right ( r ) pair of contacts on the hall bar .
panels ( c)-(h ) show oscillatory photo - excited and dark @xmath1 from the left - sided pair and the right - sided pair of contacts on the hall bar , along with the extracted @xmath16 at ( c ) @xmath66 ( d ) @xmath67 ( e ) @xmath68 ( f ) @xmath69 ( g ) @xmath70 ( h ) @xmath71 , respectively .
panels ( c)-(h ) show that the fit extracted @xmath16 at a given extrema of the mimos have similar values for contact pairs on opposite sides ( ( l ) and ( r ) ) of the device with the phase shift difference being smaller than estimated measurement uncertainty ( @xmath49).,width=283 ] , i.e. , @xmath72 vs @xmath9 for various extrema .
the figure shows that @xmath73 for all extrema at a given @xmath9 are close to each other .
however , there is a large variation in @xmath73 with @xmath9.,width=283 ] the phase shifts @xmath16 in the @xmath1 vs. @xmath0 data are extracted by fitting to @xmath74 , with `` + '' sign for peaks and `` - '' sign for valleys .
associated fit curves are also shown in fig .
2(c)-(h ) .
to compare these @xmath16 , let us begin by focusing upon fig .
2(c ) ( @xmath43 ) and fig
. 2(d ) ( @xmath44 ) .
again , both of @xmath43 and @xmath44 are measured at the @xmath56 magnetic field , but from contact pairs on opposite sides of the hall bar .
the results show @xmath75 for @xmath43 and @xmath76 for @xmath44 , where the small standard errors are due to the excellent fits to the cosine square function for oscillatory @xmath1 .
the @xmath77 between @xmath43 and @xmath44 , is less than the measurement uncertainty of @xmath78 .
similarly , if we compare fig .
2(e ) ( @xmath45 ) and ( f ) ( @xmath46 ) or ( g ) ( @xmath47 ) and ( h ) ( @xmath48 ) , the @xmath79 for @xmath57 , and @xmath80 for @xmath58 .
these phase shift differences @xmath81 are again smaller than the measurement uncertainty .
next , a comparison of the photo - excited and the dark @xmath1 vs @xmath6 is shown in fig .
3(a ) and ( b ) for the same experimental parameters over the range @xmath82 t. similar to fig . 2 ,
@xmath1 measured on the left contact pair on the hall bar is plotted in fig .
3(a ) ( with superscript l ) , and that measured via right contact pair is shown in fig .
3(b ) ( with superscript r ) .
three prominent extrema , which are now labeled as @xmath83 , @xmath84 and @xmath85 , represent the first peak , the first valley , and the second peak , respectively , for @xmath86 .
3(c)-(h ) also exhibit traces of oscillatory photo - excited and dark @xmath1 vs @xmath0 at the @xmath87 , @xmath88 , @xmath89 , @xmath90 , @xmath91 , and @xmath92 magnetic field values indicated in fig .
3(a ) and ( b ) .
the @xmath16 extracted by fitting the data to @xmath74 are also displayed in fig .
3 ( c)-(h ) . as for the data of fig . 2
, we compare the @xmath16 of opposite contact pairs at the extrema of rimos in the range of @xmath86 . at @xmath66 , @xmath93 , and at @xmath67 , @xmath94 , which yields @xmath95 , still much smaller than the measurement uncertainty .
for @xmath84 and @xmath85 ( fig .
3(e)-(h ) ) , the @xmath96 and @xmath97 , respectively , which are also within the estimated measurement uncertainty of @xmath40 .
similar results to those shown fig . 2 and fig .
3 for @xmath52 ghz were observed at other microwave frequencies .
that is , each @xmath98 at these @xmath9 were smaller than the measurement uncertainty . as a consequence
, we make the assumption that the @xmath16 values obtained from the two opposite contact pairs at given @xmath6 in this sample are practically indistinguishable .
thus , we average the @xmath99 and @xmath100 obtained from the opposite contact pairs of the hall bar to reduce the measurement error in the extracted @xmath16 , and assume that this average @xmath16 is more representative of the corresponding sample area ( see fig .
table 1 provides a summary of the @xmath101 obtained after averaging over opposite contact pairs . here ,
@xmath102 and @xmath103 are the @xmath2 s for positive and negative magnetic fields , respectively .
[ cols="^,>,>,>,>,>,>",options="header " , ] for the sake of extracting the frequency dependence in @xmath2 , we evaluate also @xmath73 , which is the average of @xmath104 and @xmath105 .
for example , @xmath106 , etc .
the @xmath107 at the various extrema are shown in fig . 4 as a function of @xmath9 .
4 shows that the differences between any two of @xmath107 are less than @xmath78 at each @xmath9 .
further , the points appeared clustered at each @xmath9 , and the point - clusters show a systematic variation with @xmath9 .
these results indicate a frequency dependence in the phase shift @xmath2 even when a worst case scenario is applied for averaging over other smaller contributions to the phase shift .
from the results presented above , we suggest the following : ( a ) at low @xmath31 , opposing contact pairs on a hall bar present the same value for a given extremum , a given magnetic field direction , and a given @xmath9 , for @xmath16 that is extracted from the fit of @xmath15 , within measurement uncertainty .
( b ) at a given @xmath9 , the @xmath73 for all the considered extrema exhibit similar values , see fig .
4 . however ,
these values show a systematic dependence upon @xmath9 .
note that point ( a ) is expected since the two edges of the hall bar are parallel to each other and their orientation with respect to the microwave polarization is the same . since the gaas / algaas heterojunctions represent an extraordinarily clean system with mean - free paths approaching the mm- or sample size - scale , it is difficult to develop a scenario where the two edges of the homogeneous specimen would not exhibit the same linear polarization response . in this context
it might be interesting to introduce non - parallel edges to see if that feature in the specimen introduces a phase shift in the response observed on the two edges .
this will be a topic of future experiments .
point ( b ) suggests that the magnitude of @xmath6 does not produce profoundly distinguishable differences on representative @xmath16 at the extrema of rimos in the specimen .
the difference in this aspect between present and previous experimental work could be attributed to differences in sample quality and defect configuration within the specimen . from the theoretical perspective , lei _ _ et al.__@xcite have simulated sinusoidal responses of @xmath1 as a function of polarization angle @xmath0 using the balance - equation formulation of their photon - assisted magnetotransport model .
the results have indicated that the phase shift in the @xmath1 vs. @xmath0 response is dependent upon @xmath6 or extremum , and @xmath9 .
further , they suggested that @xmath108 , @xmath109 , etc . in an isotropic system , which is not observed here .
however , it should be noted that real samples possibly include additional complexity , such as asymmetry , that was not considered in their theory.@xcite finally , although the cause of observed frequency - dependent @xmath16 is not yet fully understood experimentally , and theory has not considered this possibility in our context , we draw a comparison with faraday rotation in quantum hall systems . in faraday rotation , the polarization - plane of the transmitted linearly polarized radiation in the presence of magnetic field becomes a rotated by an angle , @xmath110 , which is called the faraday angle .
generally , faraday rotation , @xmath110 , is a function of radiation frequency per the drude - lorentz model@xcite .
the phase shifts @xmath2 reported here appear , however , in a @xmath111 response , the magnetoresistance , observed under @xmath112-excitation .
one might qualitatively understand the observed @xmath2 by suggesting that the 2des rotates the polarization of the @xmath112 excitation , and the @xmath111 response then follows polarization of the rotated @xmath112 excitation .
in such a situation , @xmath2 could be a manifestation of @xmath113 .
further theory is needed , however , to provide more understanding of this possibility .
the phase shifts observed in the polarization - angle - dependence of microwave - induced magnetoresistance oscillations have been processed using an analysis which emphasized averaging over the smaller contributions , when the smaller contributions were smaller than estimates of the experimental uncertainty .
the analysis was carried out in order to extract a possible microwave frequency contribution to the phase shift , @xmath3 , observed in the @xmath1 vs. @xmath0 response of the microwave radiation - induced magnetoresistance oscillations .
the analysis demonstrates a non - vanishing frequency dependent contribution to the phase shift over the frequency interval @xmath18 ghz .
magnetotransport studies and t.ye at georgia state university were supported by the us department of energy , office of basic energy sciences , material sciences and engineering division under de - sc001762 .
liu was supported by the aro under w911nf-10 - 1 - 0450 .
0 r. g. mani , j. h. smet , k. von klitzing , v. narayanamurti , w. b. johnson , and v. umansky , nature ( london ) * 420 * , 646 ( 2002 ) . m. a. zudov , r. r. du , l. n. pfeiffer , and k. w. west , phys . rev
. lett . * 90 * , 046807 ( 2003 ) .
p. h. siegel , ieee trans .
microwave theory and techniques , * 50 * , 910 - 928 ( 2002 ) .
b. b. hu and m. c. nuss , optics lett .
20 , 1716 - 1718 ( 1995 ) .
r. g. mani , appl .
phys . lett . *
92 * , 102107 ( 2008 ) .
r. g. mani , v. narayanamurti , k. von klitzing , j. h. smet , w. b. johnson and v umansky , phys .
b * 69 * , 161306(r ) ( 2004 ) ; phys . rev .
b * 70 * , 155310 ( 2004 ) .
r. g. mani , j. h. smet , k. von klitzing , v. narayanamurti , w. b. johnson and v. umansky , phys .
* 92 * , 146801 ( 2004 ) ; phys .
b * 69 * , 193304 ( 2004 ) .
s. a. studenikin , m. potemski , p. t. coleridge , a. s. sachrajda and z. r. wasilewski , solid state commun .
* 129 * , 341 - 345 ( 2004 ) .
r. g. mani , physica e ( amsterdam ) * 22 * , 1 - 6 ( 2004 ) ; physica e amsterdam * 25 * , 189 - 197 ( 2004 ) ; int .
b * 18 * , 3473 - 3480 ( 2004 ) ; appl .
lett . * 85 * , 4962 - 4964 ( 2004 ) ; phys .
b * 72 * , 075327 ( 2005 ) ; z. phys .
b 92 , 335 ( 1993 ) ; j. phys . soc .
65 , 1751 ( 1996 ) ; appl .
* 91 * , 132103 ( 2007 ) ; solid state comm . *
144 * , 409 - 412 ( 2007 ) ; physica e ( amsterdam ) * 40 * , 1178 - 1181 ( 2008 ) .
a. e. kovalev , s. a. zvyagin , c. r. bowers , j. l. reno and j. a. simmons , solid state commun .
* 130 * , 379 - 381 ( 2004 ) .
r. r. du , m. a. zudov , c. l. yang , l. n. pfeiffer and k. w. west , physica e * 22 * , 7 - 12 ( 2004 ) .
j. h. smet , b. gorshunov , c. jiang , l. pfeiffer , k. west , v. umansky , m. dressel , r. meisels , f. kuchar and k. von klitzing , phys .
* 95 * , 116804 ( 2005 ) .
a. wirthmann , b. d. mccombe , d. heitmann , s. holland , k. friedland and c .- m .
hu , phys .
b * 76 * , 195315 ( 2007 )
. s. a. studenikin , a. s. sachrajda , j. a. gupta , z. r. wasilewski , o. m. fedorych , m. byszewski , d. k. maude , m. potemski , m. hilke , k. w. west and l. n. pfeiffer , phys .
b * 76 * , 165321 ( 2007 ) .
o. e. raichev , phys .
b * 78 * , 125304 ( 2008 ) .
s. wiedmann , g. m. gusev , o. e. raichev , t. e. lamas , a. k. bakarov and j. c. portal , phys .
b * 78 * , 121301(r ) ( 2008 ) .
r. g. mani , w. b. johnson , v umansky , v. narayanamurti and k. ploog , phys .
b * 79 * , 205320 ( 2009 ) .
tung , c. l. yang , d. smirnov , l. n. pfeiffer , k. w. west , r. r. du and y .- j .
wang , solid state commun . * 149 * , 1531 - 1534 ( 2009 ) .
r. g. mani , c. gerl , s. schmult , w. wegscheider and v. umansky , phys .
b * 81 * , 125320 ( 2010 ) . a. n. ramanayaka and r. g. mani , phys .
b * 82 * , 165327 ( 2010 ) .
s. wiedmann , g. m. gusev , o. e. raichev , a. k. bakarov and j. c. portal , phys .
b * 81 * , 085311 ( 2010 ) .
o. m. fedorych , m. potemski , s. a. studenikin , j. a. gupta , z. r. wasilewski and i. a. dmitriev , phys .
b * 81 * , 201302(r ) ( 2010 ) .
s. wiedmann , g. m. gusev , o. e. raichev , a. k. bakarov and j. c. portal , phys .
lett . * 105 * , 026804 ( 2010 ) .
a. n. ramanayaka , r. g. mani and w. wegscheider , phys .
b * 83 * , 165303 ( 2011 ) .
r. g. mani , a. n. ramanayaka and w. wegscheider , phys .
b * 84 * , 085308 ( 2011 )
. a. n. ramanayaka , r. g. mani , j. iarrea and w. wegscheider , phys .
b * 85 * , 205315 ( 2012 ) .
a. bogan , a. t. hatke , s. a. studenikin , a. sachrajda , m. a. zudov , l. n. pfeiffer and k. w. west , phys .
b * 86 * , 235305 ( 2012 ) .
r. g. mani , j. hankinson , c. berger and w. a. de heer , nat .
* 3 * , 996 ( 2012 ) .
s. dietrich , s. byrnes , s. vitkalov , a. v. goran and a. a. bykov , j. appl .
* 113 * , 053709 ( 2013 ) .
r. g. mani , a. n. ramanayaka , t. ye , m. s. heimbeck , h. o. everitt and w. wegscheider , phys .
b * 87 * , 245308 ( 2013 ) .
r. g. mani , a. kriisa and w. wegscheider , sci .
rep . * 3 * , 2747 ( 2013 ) .
r. g. mani and a. kriisa , sci .
rep . * 3 * , 3478 ( 2013 ) .
t. ye , r. g. mani and w. wegscheider , appl .
. lett . * 102 * , 242113 ( 2013 ) ; ibid * 103 * , 192106 ( 2013 ) ; ibid . * 105 * , 191609 ( 2014 ) .
a. t. hatke , m. a. zudov , j. d. watson , m. j. manfra , l. n. pfeiffer and k. w. west , phys .
b * 87 * , 161307(r ) ( 2013 ) .
d. konstantinov , y. monarkha and k. kono , phys .
* 111 * , 266802 ( 2013 ) .
t. ye , h .- c .
liu , w. wegscheider and r. g. mani , phys .
b * 89 * , 155307 ( 2014 ) .
a. c. durst , s. sachdev , n. read and s. m. girvin , phys .
* 91 * , 086803 ( 2003 )
. v. ryzhii and r. suris , j. phys . :
matter * 15 * , 6855 - 6869 ( 2003 ) .
x. l. lei and s. y. liu , phys .
lett . * 91 * , 226805 ( 2003 ) .
a. a. koulakov and m. e. raikh , phys .
b * 68 * , 115324 ( 2003 ) . a. v. andreev and i. l. aleiner and a. j. millis , phys .
lett . * 91 * , 056803 ( 2003 ) .
v. ryzhii , phys .
b * 68 * , 193402 ( 2003 ) .
x. l. lei , j. phys . : condens .
matter * 16 * , 4045 - 4060 ( 2004 ) . k. park , phys .
b * 69 * , 201301(r ) ( 2004 ) . p. h. rivera and p. a. schulz , phys .
b * 70 * , 075314 ( 2004 )
. s. a. mikhailov , phys .
b * 70 * , 165311 ( 2004 ) .
a. auerbach , i. finkler , b. i. halperin and amir yacoby , phys .
rev . lett . * 94 * , 196801 ( 2005 ) .
j. iarrea and g. platero , phys .
* 94 * , 016806 ( 2005 ) .
x. l. lei and s. y. liu , phys .
b * 72 * , 075345 ( 2005 ) .
i. a. dmitriev , m. g. vavilov , i. l. aleiner , a. d. mirlin and d. g. polyakov , phys .
b * 71 * , 115316 ( 2005 ) .
j. iarrea and g. platero , appl .
. lett . * 89 * , 052109 ( 2006 ) .
j. dietel , phys .
b * 73 * , 125350 ( 2006 ) .
j. iarrea and g. platero , phys .
b * 76 * , 073311 ( 2007 ) .
a. d. chepelianskii , a. s. pikovsky and d. l. shepelyansky , eur .
j. b * 60 * , 225 - 229 ( 2007 ) .
j. iarrea and g. platero , phys .
b * 78 * , 193310 ( 2008 ) . j. iarrea and g. platero , appl .
. lett . * 93 * , 062104 ( 2008 ) .
s. wang and t .- k .
ng , phys .
b * 77 * , 165324 ( 2008 ) .
j. iarrea and g. platero , appl .
. lett . * 95 * , 162106 ( 2009 ) .
i. g. finkler and b. i. halperin , phys .
b * 79 * , 085315 ( 2009 ) .
p. h. rivera , a. l. c. pereira and p. a. schulz , phys .
b * 79 * , 205406 ( 2009 ) . x. l. lei and s. y. liu , appl . phys
. lett . * 94 * , 232107 ( 2009 ) .
j. iarrea and g. platero , nanotechnology * 21 * , 315401 ( 2010 ) .
jess iarrea , r. g. mani and w. wegscheider , phys .
b * 82 * , 205321 ( 2010 ) .
j. iarrea and g. platero , phys .
b * 84 * , 075313 ( 2011 ) .
j. iarrea , appl .
. lett . * 99 * , 232115 ( 2011 ) .
s. a. mikhailov , phys .
b * 83 * , 155303 ( 2011 ) .
a. d. chepelianskii , j. laidet , i. farrer , h. e. beere , d. a. ritchie and h. bouchiat , phys .
b * 86 * , 205108 ( 2012 ) . j. iarrea , appl .
phys . lett . *
100 * , 242103 ( 2012 ) .
x. l. lei and s. y. liu , phys .
b * 86 * , 205303 ( 2012 ) .
j. iarrea , phys .
a * 377 * , 2642 ( 2013 ) .
j. iarrea , j. appl . phys .
* 113 * , 183717 ( 2013 ) .
a. kunold and m. torres , physica b * 425 * , 78 - 82 ( 2013 ) .
o. v. zhirov , a. d. chepelianskii and d. l. shepelyansky , phys . rev .
b * 88 * , 035410 ( 2013 ) .
j. iarrea , europhys . lett .
* 106 * , 47005 ( 2014 ) .
x. l. lei and s. y. liu , j. appl .
* 115 * , 233711 ( 2014 ) .
iris crassee , julien levallois , andrew l. walter , markus ostler , aaron bostwick , eli rotenberg , thomas seyller , dirk van der marel and alexey b. kuzmenko , nat . phys .
* 7 * , 48 - 51 ( 2011 ) .
a. m. shuvaev , g. v. astakhov , a. pimenov , c. brne , h. buhmann and l. w. molenkamp , phys .
* 106 * , 107404 ( 2011 ) .
r. shimano , g. yumoto , j. y. yoo , r. matsunaga , s. tanabe , h. hibino , t. morimoto and h. aoki , nat . commun .
* 4 * , 1841 ( 2013 ) . | linear polarization angle , @xmath0 , dependent measurements of the microwave radiation - induced oscillatory magnetoresistance , @xmath1 , in high mobility gaas / algaas 2d electron devices have shown a @xmath0 dependence in the oscillatory amplitude along with magnetic field , frequency , and extrema - dependent phase shifts , @xmath2 . here , we suggest a microwave frequency dependence of @xmath3 using an analysis that averages over other smaller contributions , when those contributions are smaller than estimates of the experimental uncertainty . |
measurements of solar and atmospheric neutrinos have provided strong evidence for non - zero neutrino masses and mixing @xcite .
atmospheric results have been further strengthened by the k2k collaboration s accelerator based results @xcite .
the solar neutrino results have been confirmed by the kamland collaboration in a reactor based experiment that has shown that the large mixing angle ( lma ) solution is most likely the correct one @xcite .
interpretation of the experimental results is based on oscillations of one neutrino flavor state , @xmath8 or @xmath9 , into the others , and described quantum mechanically in terms of neutrino mass eigenstates , @xmath10 and @xmath11 .
the mass squared differences involved in the transitions are measured to be approximately @xmath12 for the solar neutrinos and @xmath13 for the atmospheric neutrinos , with large mixing strengths , @xmath14 and @xmath15 in both cases .
these existing data on neutrino oscillations@xcite and the prospect of searching for cp violation make clear that the next generation of oscillation experiments must be significantly more ambitious than before . in particular
, the source of neutrinos needs to be accelerator based so that both the neutrino flavor content and the energy spectrum of the initial state can be selected .
several approaches have been explored in the literature .
these involve either a narrow band `` off - axis '' beam produced with a conventional magnetic focusing system @xcite or a neutrino factory based on a muon storage ring @xcite . in this paper , we show that the currently favored parameters open the possibility for an accelerator based very long baseline ( bvlb ) experiment that can explore both solar and atmospheric oscillation parameters in a single experiment , complete the measurement of the mixing parameters , and search for new physics .
there are four measurements of primary interest that can be addressed with the experimental arrangement described in this paper . using the parameter convention described in @xcite , they are : 1 .
\(i ) definitive observation of oscillatory behavior ( multiple oscillations ) in the @xmath16 disappearance mode and precise determination of the oscillation parameters @xmath4 and @xmath17 , with statistical errors @xmath18 .
\(ii ) detection of the oscillation @xmath19 in the appearance mode and measurement of the parameter @xmath20
. this will involve matter enhancement and also allow definitive measurement of the sign of @xmath4 , i.e. , which neutrino @xmath11 or @xmath21 is heavier .
\(iii ) measurement of the cp violation parameter , @xmath0 , in the lepton sector .
\(iv ) measurement of @xmath6 in the appearance channel of @xmath7 .
we describe how these measurements could be carried out with good precision in a single accelerator based experiment . for precise and definitive measurement of oscillations
we must observe multiple oscillation nodes in the energy spectrum of reconstructed muon and electron neutrino charged current events .
the multiple node signature is also necessary in order to distinguish between oscillations and other more exotic explanations such as neutrino decay @xcite or extra dimensions @xcite for the muon neutrino deficit in atmospheric neutrinos . since the cross - section , fermi motion , and nuclear effects limit the statistics and the energy resolution ( for reconstructed neutrino energy ) of low energy charged current events , we use neutrinos with energies greater than few hundred mev and clean events with a single visible muon or electron that are dominated by quasielastic scattering , @xmath22 , where @xmath23 denotes a lepton .
figure [ nodes ] shows that the distance needed to observe at least 3 nodes in the reconstructed neutrino energy spectrum is greater than 2000 km for @xmath24 , the currently favored value from super - kamiokande atmospheric data @xcite
. a baseline of longer than 2000 km coupled with a wide band beam with high flux from 0.5 to 6 gev will provide a nodal pattern in the @xmath25 disappearance channel and good sensitivity over a broad range of @xmath4 . .
the distances from fnal to soudan and from bnl to homestake are shown by the vertical lines.,scaledwidth=48.0% ] ) and solid ( @xmath26 ) curves are for _ no _ and the lower dot - dashed ( @xmath26 ) curves is for _ ro_. the parameters used for the figure are @xmath27 , @xmath28 , and @xmath29 and @xmath30 , @xmath31 . , scaledwidth=48.0% ]
the low energy wide band beam and the very long baseline also have a number of important advantages for the appearance channel of @xmath32 .
these advantages can be summarized using figure [ app1 ] which shows the probability of @xmath7 oscillation as a function of neutrino energy for the distance of 2540 km , the distance between bnl and the homestake mine in south dakota .
the oscillation parameters that we have assumed are indicated in the figure and the caption .
we define the natural mass ordering ( _ no _ ) of neutrinos to be @xmath33 , and the other two possibilities ( @xmath34 and @xmath35 ) are collectively called reversed ordering ( _ ro _ ) . both _
ro _ possibilities have very similar physical consequences , but @xmath35 is strongly disfavored by the explanation of the solar neutrino deficit using matter enhanced neutrino oscillation ( the lma solution ) . since neutrinos from an accelerator beam must pass through the earth to arrive at a detector located 2540 km away , the probability in figure [ app1 ] includes the effects of matter , which enhance ( suppress ) the probability above 3.0 gev for _ no _ ( _ ro _ ) @xcite .
therefore the appearance probability above 3.0 gev is sensitive to both the mass ordering and the parameter @xmath36 .
the probability in the region 1.0 to 3.0 gev is less sensitive to matter but much more sensitive to the cp phase @xmath0 as shown in figure [ app1]@xcite .
the increase in the probability below 1.5 gev is due to the presence of terms involving the solar oscillation parameters , @xmath37 and @xmath38 .
therefore , the spectrum of electron neutrino events measured with a wide band beam over 2540 km with sufficiently low background has the potential to determine @xmath39 , @xmath0 , and the mass ordering of neutrinos as well as @xmath37 because these parameters affect different regions of the energy spectrum . in the following ,
we examine how well the parameters can be determined and the implications for the detector performance and background .
the high energy proton accelerator , to be used for making the neutrino beam , must be intense ( @xmath40 1 mw in power ) to provide sufficient neutrino - induced event rates in a massive detector very distant from it .
such a long baseline experimental arrangement @xcite may be realized with a neutrino beam from the upgraded 28 gev proton beam from the alternating gradient synchrotron at the brookhaven national laboratory ( bnl ) and a water cherenkov detector with 0.5 megaton of fiducial mass at either the homestake mine in south dakota or the waste isolation pilot plant ( wipp ) in new mexico , at distances @xmath402540 km and @xmath402900 km from bnl , respectively .
a version of the detector is described at length in @xcite and is a candidate for location in the homestake mine , where it would occupy five independent cavities , each about 100 kton in fiducial volume .
another version of the detector in a single volume is described in @xcite . for much of the discussion below
we have used the distance of 2540 km to homestake for our calculations . although the statistics obtained at 2900 km to wipp will be somewhat smaller , many of the physics effects will be larger , and therefore the resolution and the physics reach of the bvlb experiment are approximately independent of the choice between homestake or wipp .
the accelerator upgrade as well as the issues regarding the production target and horn system are described in @xcite .
we briefly describe it here for completeness .
currently , the bnl - ags can accelerate @xmath41 protons upto 28 gev approximately every 2 sec .
this corresponds to average beam power of about 0.16 mw .
this average power could be upgraded by increasing the repetition rate of the ags synchrotron to 2.5 hz while keeping the number of protons per pulse approximately the same .
currently a 200 mev room temperature linac in combination with a small synchrotron , called the booster , injects protons into the ags at 1.2 gev .
the process of collecting sufficient number of protons from the booster into the ags takes about 0.6 sec .
therefore , for 2.5 hz operation the booster must be replaced by a new injector .
a new super - conducting linac to replace the booster could serve the role of a new injector ; the remaining modifications to the ags are well understood and they involve power supplies , the rf system , and other rate dependent systems to make the accelerator ramp up and down at 2.5 hz .
we expect the final upgraded accelerator configuration to yield @xmath42 protons every 400 ms at 28 gev .
our studies show that this intensity can be further upgraded to 2 mw by incremental improvements to the linac and the ags repetition rate .
in addition to the 1 mw accelerator upgrade , the pion production target , the horn focusing system , and the decay tunnel , which must be aimed at an angle of 11.26 degrees into the earth ( towards homestake ) need to be built .
our current studies indicate that a carbon - carbon composite target inserted into a conventional aluminum horn cooled with forced helium and water will function in the 1 mw beam for a sufficient length of time .
our current plan is to build a hill instead of an underground tunnel to accommodate the 200 m long pion decay tunnel .
we consider both the rapid cycling accelerator upgrade which maintains a relatively low intensity per pulse and the 1 mw capable target technically less risky than other alternatives . for comparison , at jparc @xcite
the first ( preliminary ) stage of 0.75 mw is to be achieved at @xmath40 1/3 hz with @xmath43 protons per pulse at 50 gev . at fermilab
a 2 mw upgrade to the 120 gev main injector calls for either an 8 gev super - conducting linac or a new rapid cycling proton synchrotron as injectors for the main injector@xcite . as explained in the following ,
the combination of the very long baseline and a 500 kt detector allows the bvlb experiment to use a lower average power at 1 mw compared to the proposed jparc or numi(off axis ) @xcite projects and achieve wider physics goals .
this reduces the technical difficulties ( liquid mercury or moving targets , radiation damage , shielding of personnel , etc . )
associated with using beam powers in excess of 1 mw .
we have performed detailed simulations of a wide band horn focused neutrino beam using 1 mw of 28 gev protons on a graphite target .
the neutrino spectrum obtained in these simulations and used for the results in this paper is shown in figure [ nuspec ] normalized for pot ( protons on target ) .
we calculate that with such a beam ( total intensity of @xmath44 at distance of 1 km from the target ) we will obtain about 60,000 charged current and 20,000 neutral current events in a total exposure lasting @xmath45 seconds in a 0.5 megaton water cherenkov detector located at the homestake mine 2540 km away from bnl . for the results reported below
, we use the particle spectra obtained by making cuts to select single muon or electron quasielastic events .
events with multiple particles ( about 4 times higher in statistics ) could be used to further enhance the statistical significance of the effects .
a conventional horn focused beam can be run in either the neutrino or the anti - neutrino mode . in this paper
we show that most of the physics program in the case of 3-generation mixing can be carried out by taking data in the neutrino mode alone .
it is well known that experiments that must take data with anti - neutrinos , which have lower cross section , must run for long times to accumulate sufficient statistics @xcite .
this problem is largely alleviated in our method .
we will discuss the anti - neutrino beam in a separate paper ; here we wish to focus on the physics reach of the neutrino beam alone @xcite .
protons on target ) . ,
sec of data - taking .
the top histogram is without oscillations ; the middle histogram with error bars is with oscillations .
both histograms include the dominant single pion charged current background .
the bottom histogram shows this background contribution to the oscillating spectrum .
this plot is for @xmath46 .
the error bars correspond to the statistical error expected in the bin . at low energies the fermi movement , which is included in the simulation , will dominate the resolution .
, scaledwidth=48.0% ] an advantage of the very long baseline is that the multiple node pattern is detectable over the entire allowed range of @xmath4 .
the sensitivity to @xmath47}}{l\rm [ km ] } ~ev^2 $ ] extends to the low value of about @xmath48 , significantly below the allowed range from super kamiokande . at
energies lower than @xmath49 gev , the @xmath16 energy resolution will be dominated by fermi motion and nuclear effects , as illustrated in figure [ nodes ] .
the contribution to the resolution from water cherenkov track reconstruction depends in first approximation on the photo - multiplier tube coverage . with
coverage greater than 10% , a reconstruction energy resolution of better than @xmath50 should be achieved @xcite .
the simulated spectrum of the expected @xmath16 disappearance signal including backgrounds and resolution is shown in figure [ wcnodesa ] for @xmath51 as a function of reconstructed neutrino energy .
the background , which will be primarily charged current events , will also oscillate and broaden the dips in the nodal pattern . from this spectrum
we estimate that the determination of @xmath52 will have a statistical uncertainty of approximately @xmath53% at @xmath54 and @xmath55 . the experiment can determine @xmath56 at 90% confidence level .
within the parameter region of interest there will be little correlation in the determination of @xmath4 and @xmath5 .
the precision of the experiment is compared in figure [ cntr ] with the precision expected from minos @xcite and the result from super - kamiokande @xcite .
since the statistics and the size of the expected signal ( distortion of the spectrum ) are both large in the disappearance measurement , we expect that the final error on the parameters will be dominated by the systematic error .
a great advantage of the very long baseline and multiple oscillation pattern in the spectrum is that the effect of systematic errors on flux normalization , background subtraction , and spectrum distortion due to nuclear effects or detector calibration is small .
the error on the overall detector energy scale is expected to be the dominant systematic error @xcite .
the large event rate in this experiment will allow us to measure @xmath4 precisely in a short period of time ; this measurement will be important to predict the shape of the appearance signal which we now discuss .
the importance of matter effects on long - baseline neutrino oscillation experiments has been recognized for many years @xcite . for our study , we have included the effect of matter with a full numerical calculation taking into account a realistic matter profile in the earth , following @xcite . for our present discussion ,
it is useful to exhibit an approximate analytic formula for the oscillation of @xmath7 for 3-generation mixing obtained with the simplifying assumption of constant matter density @xcite .
assuming a constant matter density , the oscillation of @xmath19 in the earth for 3-generation mixing is described approximately by equation [ qe1 ] . in this equation @xmath58 , @xmath59 , @xmath60 , @xmath61 .
@xmath62 is the density of electrons in the earth .
recall that @xmath63 .
also notice that @xmath64 is sensitive only to the sign of @xmath65 .
@xmath66 for anti - neutrinos , the second term in equation [ qe1 ] has the opposite sign .
it is proportional to the following cp violating quantity .
@xmath67 equation [ qe1 ] is an expansion in powers of @xmath68 .
the approximation becomes inaccurate for @xmath69 as well as @xmath70 .
for the actual results shown in figure [ app1 ] we have used the exact numerical calculation , accurate to all orders .
nevertheless , the approximate formula is useful for understanding important features of the appearance probability : 1 ) the first 3 terms in the equation control the enhancement ( for _ no _ ) or suppression ( for _ ro _ ) of the oscillation probability above 3 gev ; 2 ) the second and third terms control the sensitivity to cp in the intermediate 1 to 3 gev range ; and 3 ) the last term controls the sensitivity to @xmath37 at low energies . and @xmath5 for the 2540 baseline experiment ; assuming 1 mw , 0.5 mt , and @xmath71 sec of exposure .
we have included a 5% bin - to - bin systematic uncertainty in the energy calibration as well as a 5% systematic uncertainty in the normalization .
we have not included a systematic uncertainty on the global energy scale ; this should be added in quadrature to the expected resolution on @xmath4 .
the expected resolution from the minos experiment at fermilab ( dashed ) and the allowed region from super kamiokande ( dotted ) is also indicated .
, scaledwidth=48.0% ] while the @xmath16 disappearance result will be affected mainly by systematic errors , the @xmath72 appearance result will be affected by the backgrounds .
the @xmath73 signal will consist of clean , single electron events ( single showering rings in a water cherenkov detector ) that result mostly from the quasielastic reaction @xmath74 .
the main backgrounds will be from the electron neutrino contamination in the beam and reactions that have a @xmath75 in the final state .
the @xmath75 background will depend on how well the detector can distinguish events with single electron induced and two photon induced electromagnetic ( e.m . )
backgrounds due to @xmath76 conversion , @xmath77 charged current reactions in the detector , and subsequent @xmath78 decays are small because of the low energy of the beam and the consequent low cross section for @xmath79 production @xcite .
because of the rapid fall in the bvlb neutrino spectrum beyond 4 gev ( figure [ nuspec ] ) , the largest contribution to the @xmath75 background will come from neutral current events with a single @xmath75 in the final state .
it is well known that resonant single pion production in neutrino reactions has a rapidly falling cross section as a function of momentum transfer , @xmath80 , @xcite up to the kinematically allowed value .
this characteristic alone suppresses this background by more than 2 orders of magnitude for @xmath75 ( or shower ) energies above 2 gev @xcite .
therefore a modest @xmath75 background suppression ( by a factor of @xmath3 below 2 gev and @xmath402 above 2 gev ) is sufficient to reduce the @xmath75 background to manageable level over the entire spectrum .
the electron neutrino contamination in the beam from decays of muons and kaons , is well understood to be 0.7% of the muon neutrino flux @xcite , with a similar spectrum ( figure [ nuspec ] ) .
the experimentally observed electron neutrino energy spectrum will therefore have three components : the rapidly falling shape of the @xmath75 background , the spectral shape of the @xmath81 beam contamination slightly modified by oscillation , and the oscillatory shape of the appearance signal .
the shape of the appearance spectrum will be well known because of the precise knowledge of @xmath4 from the disappearance measurement as well as the improved knowledge of @xmath82 from kamland @xcite .
these distinguishing spectra will allow experimental detection of @xmath7 with good confidence .
figure [ app2 ] shows a simulation of the expected spectrum of reconstructed electron neutrino energy after @xmath71 sec of running .
the parameters assumed are listed in the figure as well as the caption .
figure [ app2 ] further illustrates the previously described three regions of the appearance spectrum : the high energy region ( @xmath83 gev ) with a matter enhanced ( for _ no _ ) appearance has the main contribution to the background from @xmath81 contamination in the beam ; the intermediate region ( @xmath84 gev ) with high sensitivity to the cp phase , but little dependance on mass ordering , has approximately equal contribution from both background sources ( @xmath81 and @xmath75 ) ; and the low energy region ( @xmath85 gev ) , where the effects of the cp phase and @xmath86 dominate , has the main background from the @xmath75 events .
matter enhancement of the oscillations has been postulated for a long time without experimental confirmation @xcite .
detection of such an effect by observing a matter enhanced peak around 3 gev will be important .
however , in the case of _ ro _ mass ordering this enhanced peak will be missing , but the effect ( depending on @xmath0 ) on the rest of the electron neutrino spectrum will be small .
the very long baseline combined with the low energy spectrum make it posible to observe @xmath7 conversion even if @xmath87 because of the contribution from @xmath37 if the solar neutrino large mixing angle solution ( lma ) holds . in figure [ app3 ]
we show the expected appearance spectrum for @xmath88 ( lma - i ) with 90 excess events and @xmath89 ( lma - ii ) with 530 excess events above background .
the lma - i and lma - ii solutions with their respective confidence levels are discussed in @xcite .
this measurement will be sensitive to the magnitude and knowledge of the background because there will be no oscillating signature to distinguish the signal .
we estimate that the statistical and systematic errors on this measurement will allow us to determine the combination @xmath90 to a precision of about 10% at the lma best fit point .
this is competitive with the expected measurement from kamland .
however , it will be done in the explicit ( @xmath7 ) appearance mode , which is qualitatively different from the measurements made so far in the sno or the kamland experiments .
therefore , such a measurement will have important implications on the search for new physics such as sterile neutrinos . in figure [ app4 ]
we show the 90 and 99.73% ( 3 sigma ) c.l .
sensitivity of the bvlb experiment in the variables @xmath20 versus @xmath0 .
the actual limit obtained in the case of a lack of signal will depend on various ambiguities . here
we show the 99.73% c.l .
lines for _
no _ and _ ro _ , on the right hand of which the experiment will observe an electron appearance signal due to the presence of terms involving @xmath1 with greater than 3 sigma significance and thus determine the corresponding mass ordering .
currently from atmospheric data there is a sign uncertainty on @xmath91 ; this introduces an additional ambiguity onto figure [ app4 ] of @xmath92 . for this plot
we have assumed that the other parameters are well known either through other experiment or by the disappearance measurement : @xmath93 , @xmath94 .
the values of @xmath95 and @xmath96 are set to 0.86 , 1.0 , respectively .
the sensitivity to @xmath36 is approximately constant in the @xmath4 range allowed by super kamiokande @xcite .
the value of @xmath37 affects the modulation of the @xmath36 sensitivity with respect to @xmath0 .
if there is no excess electron appearance signal , other than the expected signal due to @xmath37 alone , then it would indicate a switch to anti - neutrino running to determine if the _ ro _ hypothesis with parameters on the left hand side of the dashed line in figure [ app4 ] applies @xcite . including background contamination .
the points with error bars are for @xmath97 ; the error bars indicate the expected statistical error on each bin .
the histogram directly below the error bars is for @xmath98 and the third histogram is for @xmath99 .
the hatched histogram shows the total background .
the @xmath81 beam background is also shown .
the plot is for @xmath100 .
we have assumed @xmath101 and @xmath102 .
the values of @xmath103 and @xmath17 are set to 0.86 , 1.0 , respectively . running conditions as in figure [ wcnodesa ] .
, title="fig:",scaledwidth=48.0% ] + .
important parameters in this figure are @xmath102 ( lma - i ) or @xmath104 ( lma - ii ) and @xmath105 .
all other parameters and the running condition as in figure [ wcnodesa ] .
, title="fig:",scaledwidth=48.0% ] +
to get a qualitative understanding for the measurement of cp violating parameters we compare the size of terms involving @xmath0 with the first term in equation [ qe1 ] .
this ratio is seen to be approximately proportional to @xmath106 . as shown in figure [ app1 ] , the fractional contribution from the cp violating terms increases for lower energies at a given distance .
the energy dependence of the cp effect and the matter effect tend to be opposite and therefore can be distinguished from each other from the energy distribution using neutrino data alone . on the other hand ,
the statistics for a given size detector at a given energy are poorer by one over the square of the distance , but the term linear in @xmath107 grows linearly in distance @xcite .
the statistical sensitivity ( approximately proportional to the square root of the event rate ) to the effects of cp violation , therefore , is independent of distance because the loss of event rate and the increase of the cp effect approximately cancel each other in the statistical merit if backgrounds remain the same .
therefore , the two important advantages of the bvlb approach are that the cp effect can be detected without running in the anti - neutrino mode and the sensitivity to systematic errors on the background and the normalization is considerably reduced because the fractional size of the cp effect is large .
for example , in figure [ app1 ] , the cp effect at @xmath108 in the first oscillation peak is @xmath109 while the effect in the second oscillation peak is more than 50% .
therefore , it is unnecessary to know the background and the normalization to better than 10% to obtain a significant measurement of @xmath0 at the second oscillation maximum .
there could be a contribution to the systematic error from the theoretical calculation of the probability shown in figure [ app1 ] .
because of the very long baseline , this probability depends on the earth s density profile which is known to about 5% .
random density fluctations on that order will lead to a relative systematic uncertainty in the @xmath81 appearance probability of about 1% @xcite , which is not significant for the bvlb method , but could be significant in the case of an experiment that performs the cp measurement at the first oscillation maximum with a shorter baseline . versus @xmath0 for the natural ( _ no _ ) and reversed ordering of parameters ( _ ro _ ) .
the solid lines are for _ no _ , the left line for 90% and right line for 99.73% c.l .
the dashed and dotted lines are for 90 and 99.73% c.l . for _
ro_. on the right hand side of the 99.73% c.l .
lines the experiment will observe an electron appearance signal with greater than 3 sigma significance . , title="fig:",scaledwidth=48.0% ] + an experiment with a wide band beam is needed to exploit the energy dependence of the cp effect to extract the value of @xmath0 and @xmath110 using only the neutrino data .
the energy dependence also reduces the correlation between the two parameters .
figure [ app5 ] shows the expected resolution on @xmath111 versus @xmath0 at a particular case , @xmath29 and @xmath112 with all other parameters fixed as indicated . the 1@xmath113 error on @xmath0 from this measurement as a function of @xmath0
is shown in figure [ app6 ] .
it is clear that the precision of the measurement will be limited if @xmath114 because of the background .
however , for larger values the resolution on @xmath0 is approximately independent of @xmath20 .
this result can be understood by examining equation [ qe1 ] in which the first term grows as @xmath20 and the cp violating term grows as @xmath115 .
therefore the statistical sensitivity to the cp violating term , which is the ratio of the second term to the square - root of the first term , is indenpendent of @xmath1 as long as the background does not dominate @xcite . at @xmath116
the event rate reaches maximum and does not change rapidly with respect to @xmath0 , therefore the resolution on @xmath0 becomes poor . because of the multiple node spectrum the appearance probability is sensitive to both the @xmath107 and @xmath117 terms in equation [ qe1 ]
this eliminates the @xmath118 ambiguity normally present in a narrow band single node experiment @xcite .
the sign uncertainty @xmath119 introduces a @xmath120 ambiguity .
if @xmath37 is known poorly , there will be a contribution to the @xmath0 resolution due to the correlation between @xmath0 and @xmath37 .
we expect to know @xmath37 from kamland with precision of @xmath121 @xcite ; this does not introduce a dominant source of uncertainty on @xmath0 in our experimental method . if a four parameter , @xmath37 , @xmath103 , @xmath36 , and @xmath0 , fit to the data is performed , the correlations between the parameters dilute the sensitivity to each parameter
. however , the ambiguities and correlations do not reduce the ability of the experiment to establish that 3-generation neutrino mixing contains a non - zero complex phase and hence a cp violating term .
this is seen if we consider the resolution on the quantity @xmath122 which is cp violating and by definition does not exhibit any of the above correlations . the measurement of this quantity will simply depend on the statistics in the low and medium energy components of the spectrum . versus @xmath0 for statistical and systematic errors .
the test point used here is @xmath123 and @xmath124 .
@xmath125 , and @xmath126 .
the values of @xmath103 and @xmath17 are set to 0.86 , 1.0 , respectively . ,
scaledwidth=48.0% ] in degrees for several values of @xmath36 .
all other parameters are held fixed assuming they will be known before this measurement .
, title="fig:",scaledwidth=48.0% ] +
the alternatives to the bvlb plan fall into two categories : experiments with conventional horn focused beams over baselines of several hundred km and with beams based on neutrinos from muon decays in a high intensity muon storage ring . for a recent review of muon storage ring based neutrino factory physics as well as comparisons to conventional neutrinos beams
see @xcite .
recently , for the conventional beam an `` off - axis '' beam is preferred because the energy spectrum is narrower and can be tuned , within a limited range from 1 to 2 gev by choosing the off - axis angle @xcite .
a narrow energy spectrum could be better suited for identifying @xmath7 conversion because the neutral current backgrounds from interactions of high energy neutrinos are lowered .
the jparc to super kamiokande project ( baseline of 295 km ) has adapted this strategy and a new initiative at fermilab based on the numi beamline ( baseline of 732 km ) is also considering it @xcite . both of these projects focus on the observation of a non - zero value for @xmath1 . since it is not possible to eliminate the @xmath400.5% @xmath81 contamination in a conventional @xmath127 beam ( even if it is `` off - axis '' )
the sensitivity of all projects based on a conventional beam , including the bvlb plan , is limited to @xmath128 ( ignoring the effect of @xmath0 ) .
sensitivity below this value is background limited and requires very good understanding of the systematic errors associated with the purity of the beam .
the bvlb plan does , however , extend this sensitivity to values of @xmath129 down to 0.001 @xmath130 .
the `` off - axis '' projects must be upgraded to several megawatts of beam power and detectors of several hundred kt ( the second phase of the jparc project calls for 4 mw beam and a 1 megaton detector ) to perform the cp measurement .
the narrow band nature of the beam requires running in the anti - neutrino mode to observe cp violation .
the bvlb plan uses a wide band beam over a much longer distance to achieve the same sensitivity to @xmath20 over a wider range of @xmath4 .
it also has the ability to see a multiple node structure in both the disappearance and appearance channel ; thus making the observation of @xmath7 experimentally more robust against unanticipated backgrounds .
in addition , the unique aspect of the wide band beam over the very long baseline is the ability to measure @xmath0 with few ambiguities using neutrino data alone . the muon storage ring based `` neutrino factory '' solves the background problem faced with conventional beams .
a @xmath131 and a @xmath81 is produced in each @xmath132 decay .
an appearance search looks for the conversion of @xmath81 into @xmath127 by looking for events with negatively charged @xmath133 particles in the presence of @xmath131 charged current events with positively charged @xmath132 particles in the final state .
such a physics signature should potentially have very low background .
therefore , a neutrino factory based @xmath134 search is estimated to be sensitive to @xmath20 perhaps as small as @xmath135 .
such a facility is very well tuned for measurement of @xmath20 .
nevertheless , because of the relatively narrow energy spectrum and higher energy of the neutrino factory , it lacks the ability to observe multiple nodes in the observed spectrum .
for the neutrino factory the effects of cp violation are observed by comparing @xmath136 and @xmath137 in two different runs of @xmath133 and @xmath133 decays in the presence of very large matter effects .
the narrow spectrum also limits the ability of the neutrino factory to resolve the ambiguities in the @xmath0 determination regardless of the value of @xmath36 .
this has been extensively discussed in the literature @xcite and one of the proposed solutions is to construct several detectors with different baselines to uniquely determine @xmath0 versus @xmath36 .
although the neutrino factory can extend the reach for @xmath36 to very small values because the backgrounds to the appearance search are small , the resolution on @xmath0 disregarding the ambiguities inherent in the neutrino factory approach is comparable to the bvlb resolution .
we therefore conclude , that if @xmath138 the bvlb experiment is either equal to or exceeds the physics reach for neutrino mixing parameters of the neutrino factory .
other suggestions in the literature include @xmath81 or @xmath139 beams from either decays of radioactive nuclei accelerated to high energies ( beta - beam ) @xcite or @xmath139 s from nuclear power generation reactors such as the beam used for kamland @xcite .
both of these concepts aim to create very pure , but low energy beams . for cp sensitivity
both @xmath81 and @xmath139 must be produced in separate runs in the beta - beam concept .
the low background should enable the beta - beam concept to have good sensitivity to @xmath1 , if sufficient flux can be obtained .
the reactor beams are below @xmath127 charged current reaction threshold and therefore must rely on disappearance to search for a non - zero value of @xmath1 . a disappearance based search for @xmath1
will most likely be limited by systematic errors .
further comments on these techniques must await more detailed studies .
we have simulated and analyzed a feasible , very long baseline neutrino oscillation experiment consisting of a low energy , wide band neutrino beam produced by 1 mw of 28 gev protons incident on a carbon target with magnetic horn focusing of pions and kaons , and a 500 kt detector at a distance of @xmath140 2500 km from the neutrino source .
the bvlb neutrino beam with a total intensity of about @xmath141 at a distance of 1 km from the target could be provided by an upgrade to the bnl - ags @xcite .
the single bvlb experiment could produce measurements of all parameters in the neutrino mixing matrix through observation of the disappearance channel , @xmath142 and the appearance channel , @xmath19 .
the experiment is also sensitive to the mass ordering of neutrinos using the observation of the matter effect in the appearance channel through the currently unknown parameter @xmath36 .
nevertheless , the experiment is intended primarily to measure the strength of cp invariance violation in the neutrino sector and will provide a measurement of the cp phase , @xmath0 or alternatively the cp violating quantity , @xmath143 , if the one currently unknown neutrino oscillation mixing parameter @xmath2 , a value about 15 times lower than the present experimental upper limit .
we point out that for a given resolution on @xmath0 the number of neutrino events needed which determines the detector size and beam intensity , is approximately independent of the baseline as well as the value of @xmath144 as long as the electron signal is not background dominated .
therefore , the concept of very long baseline ( @xmath145 ) is attractive because it provides access to much richer physics phenomena as well as reduces the need to understand systematic errors on the flux normalization and background determination .
a shorter baseline ( @xmath146 ) will obviously limit the reach of the experiment for @xmath147 in the appearance mode as well as reduce the resolution on the cp parameter . on the other hand
, a much longer baseline ( @xmath148 ) will result in a matter effect that is large enough to dominate the spectrum and make the extraction of the cp effect more difficult .
the larger distance will also make it necessary to make the neutrino beam directed at a higher angle into the earth , a technical challenge that may not be necessary given the current values of neutrino parameters .
lastly , we have also shown that most of this rich physics program including the search for cp effects can be carried out by neutrino running alone .
once cp effects are established with neutrino data , anti - neutrino data could be obtained for more precision on the parameters or search for new physics .
the shape of the disappearance signal over multiple oscillations from neutrino running alone ( as well as in combination with anti - neutrino running ) can be used to contrain effects of new physics from : sterile neutrino mixing , extra dimensions , exotic interactions with matter , etc .
it has not escaped our notice that the large detector size ( @xmath149500 kt ) , mandated by the above described neutrino program , naturally lends itself to the important physics of proton decay , supernova detection , etc .
although some experimental requirements such as shielding from cosmic rays , detector shape , or photo - sensor coverage , may differ , it is clear that they can be resolved so that this large detector can also be employed to address these frontier physics issues .
we thank all contributors to the bnl neutrino working group , as well as the bnl - ags technical staff and engineers whom we consulted during this work .
we especially thank thomas roser , nick tsoupas , alessandro ruggiero , deepak raparia , nick simos , and hans ludewig for their contribution to the accelerator and beam design .
we also thank thomas kirk who encouraged us throughout this work .
this work was supported by doe grant de - ac02 - 98ch10886 and nsf grant phy02 - 18438 .
y. fukuda et al . , phys .
lett . * 81 * , 1562 ( 1998 ) ; s. fukuda et al .
lett . * 86 * 5656 , 2001 ; e.w .
beier , phys . lett . *
b283 * , 446 ( 1992 ) ; t. kajita and y. totsuka , rev .
phys . * 73 * , 85 ( 2001 ) . the jhf - kamioka neutrino project , y. itow et al .
, arxiv : hep - ex/0106019 , june 2001 .
letter of intent to build an off - axis detector to study numu to nue oscillations with the numi neutrino beam , d. ayres _
et al . _ , hep - ex/0210005 .
barger , s. geer , r. raja , k. whisnant , phys .
d63 : 113011 ( 2001 ) ; v. barger et al .
, hep - ph/0103052 ; p. huber , m. lindner , w. winter , nucl . phys .
b645 , 3 ( 2002 ) ; v. barger , d. marfatia , k. whisnant , phys .
d65 : 073023 ( 2002 ) . c. albright et al . ,
physics at a neutrino factory ; hep - ex/0008064 ; n. holtkamp et al .
, feasibility study of a neutrino source based on a muon storage ring , fermilab - pub-00 - 108-e * june 2000 ) ; s. ozaki , r. palmer , m. zisman , and j. gallardo , eds .
feability study ii of a muon - based neutrino source , bnl-52623 ( 2001 ) and references therein to other neutrino factory studies .
accelerator technical design report for high - intensity proton accelerator facility project , japan atomic energy research institute and high energy accelerator research organization , 2002 .
http://j-parc.jp/ if the value of @xmath36 turns out to be too small or the true mass ordering is _ ro _ ,
then a switch to antineutrino data - taking would be indicated .
should a moderate amount of antineutrino data show a peak in the positron spectrum , then we can conclude that the _ ro _ hierarchy ( @xmath150 ) holds true . if the neutrino data shows that the _ no _ hierarchy is correct and that the cp parameter is non - zero , then we could switch to antineutrino data to explicitly show cp non - conservation and measure the parameters more precisely .
the event rate for an anti - neutrino beam running for the same running time of @xmath71 sec is about 19,000 charged current and 6000 neutral current events . | we analyze the prospects of a feasible , brookhaven national laboratory based , very long baseline ( bvlb ) neutrino oscillation experiment consisting of a conventional horn produced low energy wide band beam and a detector of 500 kt fiducial mass with modest requirements on event recognition and resolution .
such an experiment is intended primarily to determine cp violating effects in the neutrino sector for 3-generation mixing .
we analyze the sensitivity of such an experiment .
we conclude that this experiment will allow determination of the cp phase @xmath0 and the currently unknown mixing parameter @xmath1 , if @xmath2 , a value @xmath3 times lower than the present experimental upper limit .
in addition to @xmath1 and @xmath0 , the experiment has great potential for precise measurements of most other parameters in the neutrino mixing matrix including @xmath4 , @xmath5 , @xmath6 , and the mass ordering of neutrinos through the observation of the matter effect in the @xmath7 appearance channel .
= 11.6pt |
in the current star formation paradigm , low - mass stars grow in mass over time through accretion of matter from a circumstellar disc ( e.g. lynden - bell & pringle 1974 ; appenzeller & mundt 1989 ; bertout 1989 ) .
this accretion process is believed to occur via the magnetic field lines that connect the star to its disc and that act as funnels for the gas ( e.g. knigl 1991 ; shu et al .
the strong excess emission observed in most of these pre - main sequence ( pms ) stars is believed to originate through gravitational energy , released by infalling matter , that ionises and excites the gas .
thus , the excess luminosity of these objects ( the classical t tauri stars ) can be used to measure the mass accretion rate .
a reliable measurement of the rate of mass accretion onto pms stars is of paramount importance for understanding the evolution of both the stars and their discs ( see e.g. review by calvet et al .
2000 ) . of particular interest
is determining how the mass accretion rate changes with time as a star approaches its main sequence , whether it depends on the final mass of the forming star and whether it is affected by the chemical composition and density of its parent molecular cloud or by the proximity of hot , massive stars .
ground - based spectroscopic studies of nearby young star - forming regions ( e.g. in taurus , auriga , ophiuchus , orion ) show that the mass accretion rate appears to decrease steadily with time , from @xmath16m@xmath13yr@xmath14 at ages of @xmath17myr to @xmath18m@xmath13yr@xmath14 at @xmath19myr ( muzerolle et al .
2000 ; sicilia - aguilar et al .
2005 ; 2006 ) . while at face value this is in line with the expected evolution of viscous discs ( hartmann et al .
1998 ) , the scatter on the data is very large , exceeding 2 dex at any given age .
this is at least in part explained by the recent finding that the mass accretion rate seems also to depend quite steeply on the mass of the forming star .
muzerolle et al .
( 2003 ; 2005 ) , natta et al .
( 2004 ; 2006 ) , white & hillenbrand ( 2004 ) and sicilia - aguilar et al . ( 2006 ) consistently report that the accretion rate @xmath6 decreases with the stellar mass @xmath20 as @xmath21 , albeit with large uncertainties on the actual value of the power - law index . on the other hand ,
clarke & pringle ( 2006 ) warn that such a steep decline might be spurious and strongly driven by detection / selection thresholds .
while observational uncertainties and potential systematic errors may well contribute to the present uncertainty , the true limitation in this field of research currently comes from the paucity of available measurements .
indeed , all the results so far obtained are based on the mass accretion rates of a small number of stars ( just over one hundred ) , all located in nearby galactic star forming regions , covering a very limited range of ages and no appreciable range of metallicity ( all clouds mentioned above having essentially solar metallicity ; e.g. padgett 1996 ) .
the origin of this limitation is to be found in the technique used to measure pms mass accretion rates .
regardless as to whether the latter are derived from the analysis of veiling in the photosperic absorption lines ( usually at ultraviolet and optical wavelengths ) or through a detailed study of the profile and intensity of emission lines ( usually h@xmath3 , pa@xmath22 or br@xmath23 ) , these thechniques typically require medium to high resolution spectroscopy for each individual object .
even with modern multi - object spectrographs at the largest ground - based telescopes , this approach is not very efficient and can not reach beyond nearby star forming regions or the orion cluster , as crowding and sensitivity issues rule out but the brightest objects in more distant regions like the centre of the milky way or the magellanic clouds .
while in principle the pms mass accretion rate can be estimated from the @xmath15-band excess if the spectral type is known ( see e.g. gullbring et al .
1998 ; romaniello et al .
2004 ) , this method is potentially subject to large uncertainties in the extinction and in the determination of the correct spectral type and effective temperature when spectroscopy is not available . in order to overcome these limitations and
significantly extend the sample of stars with a measured mass accretion rate , we have developed and successfully tested a new method to reliably measure @xmath24 that does not require spectroscopy .
the method , described in the following sections , combines broad - band @xmath0 and @xmath1 photometry with narrow - band @xmath2 imaging and allows us to identify all stars with excess h@xmath3 emission and to derive from it the accretion luminosity @xmath25 and hence @xmath6 for hundreds of objects simultaneously .
the paper is organised as follows : in section2 we address the identification of pms stars via their colour excess and in section3 we show how to reliably convert the measured excess into h@xmath3 luminosity and equivalent width .
section4 shows how the accretion luminosity and mass accretion rate can be obtained from the h@xmath3 luminosity and section5 presents an application of the method to the field of sn1987a , in the large magellanic cloud ( lmc ) .
a thorough discussion of all systematic uncertainties involved in the method is provided in section6 , while a general discussion and conclusions follow in section7 .
the appendix provides additional figures and tabular material .
one of the characteristic signatures of the accretion process on to pms stars is the presence of excess emission in h@xmath3 ( see calvet et al .
although surveys to search for stars with h@xmath3 excess emission over large sky areas have been traditionally carried out via slitless spectroscopy ( e.g. herbig 1957 ; kohoutek & wehmeyer 1999 ) , extensive studies in which narrow - band h@xmath3 imaging is combined with broad - band photometry also exist ( e.g. parker et al . 2005 ; drew et al .
2005 ) . recently , romaniello et al .
( 1998 ) , panagia et al .
( 2000 ) and romaniello et al .
( 2006 ) used hst / wfpc2 photometry in h@xmath3 ( f656n ) and the r band ( f675w ) to provide the first direct detection of extra - galactic pms stars with strong h@xmath3 emission in the lmc .
their approach is based on the widely adopted use of the r band magnitude as an indicator of the level of the photospheric continuum near the h@xmath3 line .
therefore , stars with strong h@xmath3 emission will have a large @xmath26 colour . from the analysis of the specific f656n and f675w filter response curves of the wfpc2 instrument , panagia et al .
( 2000 ) conclude that an equivalent width ( @xmath27 ) of the h@xmath3 emission line in excess of 8 will result in a colour excess @xmath28 .
this way of identifying pms stars is more accurate and reliable than the simple classification based on the position of the objects in the hertzsprung russell diagram , i.e. stars placed well above the main sequence ( e.g. gilmozzi et al .
1994 ; hunter et al . 1995 ; nota et al .
2006 ; gouliermis et al .
the reason is twofold : _
( i ) _ contamination by older field stars and the effects of differential extinction may lead to an overestimate of the actual number of candidate young , low - mass pms stars , i.e. those `` above '' the ms ; _ ( ii ) _ older pms stars , already close to their ms , can not be properly identified with this method , thereby resulting in an underestimate of the number of stars that are still forming .
when pms stars are identified by means of their @xmath26 colour excess , both problems disappear . similarly , this approach is more practical , efficient and economical than slitless spectroscopy for studying pms stars in dense star forming regions in the milky way and nearby galaxies , which are today accessible thanks to high - resolution observations , in particular those made with the hst
. we will show in section3 that the level of the photospheric continuum near h@xmath3 can be directly derived from the observed @xmath0 and @xmath1 magnitudes .
this means that pms stars with an h@xmath3 excess can be identified without actually measuring the canonical @xmath26 index . on the other hand ,
since @xmath26 has traditionally been used in pms stars studies , we first show here how one can derive analytically the @xmath29 magnitude of an object from the measured @xmath0 and i magnitudes .
in this way , we intend to make it possible for those who do not dispose of r - band photometry for their fields to derive it in an accurate way by using measurements in the neighbouring v and i bands .
the latter are more commonly used than the r band in photometric studies of resolved stellar populations and are particularly popular among hst observers .
the absence of remarkable spectral features between 5000 and 9000 in the stellar continuum of dwarfs and giants with effective temperatures in the range 4000k @xmath30 10000k implies that their @xmath31 and @xmath32 colours depend primarily on the effective temperature of the stars and less on their metallicity . in effect , @xmath32 remains a useful colour index for temperature determinations ( von braun et al .
1998 ; bessell , castelli & plez 1998 ) .
the colour colour sequences in the @xmath33 passbands published by johnson ( 1966 ) revealed a very tight correlation between the mean @xmath31 and @xmath32 colours of local dwarfs and giants ( see figure[fig1 ] ) .
the correlation is remarkably linear in the range @xmath34 ( i.e. 4000k @xmath35 10000k ) , where @xmath36 ( dashed line in figure[fig1 ] ) , and it remains linear for @xmath37 ( i.e. 3000k @xmath38 4000k ) , albeit with a shallower slope ( @xmath39 ; dot - dashed line ) .
it is , therefore , possible and meaningful to attempt to derive the value of the @xmath29 magnitude from the knowledge of @xmath0 and @xmath1 , particularly because no extrapolation beyond the available wavelength range is involved .
the simple inversion of the relationship above gives @xmath40 for the system of johnson .
is not the same as the much more commonly used cousin s @xmath1 .
therefore , the colour relationships given here can not be used for the cousin system .
proper relationships for the most common magnitude systems are provided in the appendix . ]
interestingly , this relationship is very similar to the one that we would obtain if we assumed that the proportion with which the @xmath0 and @xmath1 magnitudes contribute to the @xmath29 magnitude is simply dictated by the ratio of the effective wavelengths of the filters .
the effective wavelengths of the v , r and i bands in johnson s system are , respectively , @xmath41 , @xmath42 and @xmath43 , which imply that the v band should contribute for @xmath44 and the i band for @xmath45 to the r - band magnitude .
these values are remarkably close to those observed . the apparent disarming simplicity of this conclusion , however , should not lead one to think that this is the case for any photometric system .
in particular , for spectral bands much wider than johnson s , such as some of the hst / wfpc2 and hst / acs filters , the matter becomes more complicated because the true effective wavelength can depend significantly on the spectral properties of the source ( i.e. its temperature ) . in these cases ,
the relationship should be derived via direct observations in the specific bands or through synthetic photometry , when the properties of the photometric system are well caracterised . due to the large number of inter - related observing modes offered by the scientific instruments on board the hst , synthetic photometry has long been established as the most practical and efficient approach to their calibration ( see koornneef et al .
1986 ; horne 1988 ; sirianni et al .
for this reason , reference tables exist that accurately describe the main optical components of the telescope and instruments as well as the response and quantum efficiency of the detectors as a function of wavelength .
therefore , thanks also to the availability of standardised software packages ( e.g. synphot ; see laidler et al . 2008 ) , it is relatively easy to derive synthetic magnitudes from observed or theoretical stellar spectra for various hst instrumental configurations .
we have used both observed and theoretical stellar spectra , taken respectively from the bruzual - persson - gunn - stryker ( bpgs ) spectrophotometry atlas ( see gunn & stryker 1983 ) and from the stellar atmosphere models of bessell et al .
( 1998 ) , to compute colour relationships between @xmath29-like magnitudes and @xmath32-like colours for various hst instruments and ground photometric systems .
details on how the colour transformations were derived and a table of coefficients ( table[tab3 ] ) can be found in the appendix , while a brief comparison with hst data is given in the following section .
to verify the validity of the predicted colour relationships , we have used wfpc2 photometry of a field around sn1987a studied by romaniello ( 1998 ) , panagia et al .
( 2000 ) and romaniello et al .
the observations cover a field of @xmath8arcmin@xmath9 and were collected over three epochs , namely 1994 september , 1996 february and 1997 july .
the advantage of this catalogue is that the magnitude of each star in it is individually corrected for the effects of interstellar extinction by using the information provided simultaneously by all broad - band filters ( see romaniello et al .
2002 for more details ) .
while it would be possible to include the effects of reddening in the calculations of the colour relationships of table[tab3 ] , it is more appropriate in this case to correct the observed magnitudes , since considerable differential extinction is likely to be present in this field ( see panagia et al . 2000 ; de marchi & panagia , in preparation ) . from the catalogue of romaniello et al . (
2002 ) , we have selected all those stars whose mean error @xmath46 in the three bands f555w , f675w and f814w does not exceed @xmath47mag , where @xmath48 and @xmath49 , @xmath50 and @xmath51 are the photometric uncertainties in each individual band after correction for interstellar extinction .
a total of @xmath52 stars satisfy the condition set by equation[eq1 ] , out of @xmath53 objects in the complete catalogue .
the colour colour diagram for the stars selected in this way is shown in figure[fig2 ] .
the dashed line in that figure corresponds to the colour relationship derived for the bands in question from the models of bessell et al .
( 1998 ; see the appendix ) and is practically indistinguishable from the best linear regression fit to the data , which has a slope of @xmath54 .
although these models were selected to match the metallicity of the lmc , having used the relationship derived with the bpgs atlas or that for @xmath55=-1 $ ] would have still resulted in a very good fit , within the observational uncertainties .
this result confirms that , at least over the range explored here , metallicity does not play a dominant role in the colour relationships .
we provide a further proof of the reliability and accuracy of the interpolated r - band magnitudes in figure[fig3 ] , where we compare the identification of pms stars based on the measured and interpolated @xmath26 colour . to distinguish the
observed from interpolated r - band magnitude , we indicate the latter with @xmath56 , as explained below . ,
@xmath29 , @xmath1 and @xmath2 to indicate the respective wfpc2 bands .
when differences between photometric systems are important ( e.g. in table[tab3 ] ) , all bands will be indicated with their specific name . ] in this case , from the photometry of romaniello et al .
( 2002 ) already corrected for reddening , we have further restricted the selection to objects with @xmath57 , comprising @xmath58 stars .
diamonds in figure[fig3 ] represent the average @xmath26 colour of the population and , as such , define the reference with respect to which one should look for excess emission .
the dashed lines represent the theoretical colour relationship obtained using the bessell et al .
( 1998 ) model atmospheres mentioned above for normal main - sequence stars and is in rather good agreement with the average observed colours in the range @xmath59 ( filled diamonds ) .
the rms deviation between the model and the data over this range amounts to @xmath60mag and is dominated by the systematic departure in the range @xmath61 .
the latter could be due to a number of reasons , including the inability of the models to accurately describe the properties of the h@xmath3 line since they are meant to be used for broad - band photometry .
the departure of the models from the data at @xmath62 stems from both small number statistics and the fact that the bulk of the population there comes from red giants , which have lower surface gravity and possibly lower metallicity than those assumed in the models . in the following
, we will concentrate our analysis to the range @xmath63 marked by the filled diamonds .
figure[fig3]a is based on the observed data and we indicate as large dots all stars whose observed @xmath26 colour exceeds the local average by at least four times the individual photometric uncertainty in the colour , i.e. the one specific to each star . consequently , these objects must be regarded as having a bona - fide h@xmath3 excess above the @xmath10 level .
there are 199 such objects in panel ( a ) . in panel
( b ) we have replaced the observed @xmath29 magnitude with its interpolated value @xmath56 .
all but 16 of the 199 stars with @xmath10 excess in panel ( a ) have a @xmath10 excess also in panel ( b ) and are indicated as large dots .
the remaining 16 objects , marked as triangles , fail this requirement as their colour is closer to the local average colour than four times their specific combined photometric uncertainty .
in panels ( c ) and ( d ) we study the uncertainty in the other direction , i.e. from the interpolated to the observed plane .
panel ( c ) simulates the case in which , no @xmath29-band data being available , one uses the interpolated @xmath56 value from table[tab3 ] to identify stars with h@xmath3 excess emission .
stars indicated as large dots in panel ( c ) are those with interpolated @xmath64 colour exceeding the local average by at least four times the individual photometric uncertainty in the colour , i.e. those with h@xmath3 excess above the @xmath10 level .
a total of 194 stars meet this condition .
all but 11 of them would have met the @xmath10 excess condition if we had used the actually observed @xmath29-band magnitude .
they are indicated as tlarge dots in panel ( d ) , while the 11 objects failing the condition are marked as triangles . from figure[fig3 ]
we can draw some quantitative conclusions on the statistical significance of the detection of the colour excess in h@xmath3 .
first of all , it does not surprise that some objects detected as bona - fide h@xmath3 excess stars in panel ( a ) are rejected in panel ( b ) , since the typical photometric uncertainty on @xmath56 is a factor of @xmath65 higher than that on @xmath29 , as one must combine the uncertainties on @xmath0 and @xmath1 that concur to the estimate of @xmath56 ( although the major contribution to the uncertainty on the @xmath26 colour of a star comes from the h@xmath3 band ) .
for the same reason , it is to be expected that panel ( c ) will yield less objects with bona - fide h@xmath3 excess than panel ( a ) .
considering the number of detections and mismatches of figure[fig3 ] , we estimate that using the interpolated @xmath56 in place of @xmath29 a fraction of @xmath66 of the objects with h@xmath3 excess at the @xmath10 level would be missing and that for about @xmath67 of the detections the assigned statistical significance would be higher than indicated by direct @xmath29 measurements .
we conclude that our interpolation scheme is quite reliable in that it can provide the correct identification of pms stars in no less 94% of the cases .
while helpful to accurately identify pms stars , the @xmath26 colour excess alone does not immediately provide an absolute measure of the h@xmath3 luminosity @xmath4 nor of the equivalent width @xmath68 , which are necessary to arrive at the mass accretion rate . in order to measure @xmath4 , one needs a solid estimate of the stellar spectrum in the h@xmath3 band ( i.e. without the contribution of the emission ) .
similarly , to measure @xmath68 , a solid estimate of the level of the stellar continuum inside the h@xmath3 band is needed .
since the r band is over an order of magnitude wider than the h@xmath3 filter , it can not provide this information accurately , but only gives a rough estimate of the continuum level . in this section
we present a simple method that allows us to measure @xmath4 for the objects with an h@xmath3 excess and to derive the level of their spectral continuum inside the h@xmath3 band , determining in this way also @xmath69 .
the method is based on the simple consideration that , at any time , the largest majority of stars of a given effective temperature @xmath70 in a stellar population will have no excess h@xmath3 emission .
therefore , for stars of that effective temperature , the _ average _ value of a colour index involving h@xmath3 ( for instance @xmath71 , and thus not just @xmath26 ) effectively defines a spectral reference template with respect to which the h@xmath3 colour excess should be sought .
note that this is not only true for populations comprising both young and old stars , such as those typical of extragalactic stellar fields , but also for very young populations since pms objects show large variations in their h@xmath3 emission over hours or days ( e.g. fernandez et al .
1995 ; smith et al . 1999 ; alencar et al .
2001 ) , with only about one third of them at any given time being active h@xmath3 emitters above a @xmath72 threshold ( panagia et al .
2000 ) . to better clarify the working of our method , we display in figure[fig4 ] the @xmath71 colour as a function of @xmath32 for a set of stars taken from the catalogue of romaniello et al .
this time , we have selected a total of @xmath73 stars , namely all objects with @xmath74mag , where the mean uncertainty @xmath46 is defined as in equation[eq1 ] , but for the v , i and h@xmath3 bands instead of v , i and r. therefore , no r - band information is used in this case and , in fact , no r - band data is needed . it should be noted that @xmath46 is dominated by the uncertainty on the h@xmath3 magnitude , while the median value of the uncertainty in the other two bands is @xmath75 and @xmath76 , respectively .
the thick dashed line in figure[fig4 ] represents the average @xmath71 colour obtained as the running median with a box - car size of 100 points .
the thin solid line shows the colours in these filters for the model atmospheres of bessell et al .
( 1998 ) mentioned above ( note again the @xmath77mag discrepancy between models and observations around @xmath78 , most likely due to the coarse spectral sampling of the h@xmath3 line in the models ) . at the extremes of the distribution ( @xmath79 and @xmath62 ) the density of observed points decreases considerably . to determine the reference template there
, we have reduced the box - car size to 10 points and averaged the running median obtained in this way with the theoretical models ( thin solid line ) . the thick dot - dashed line in figure[fig4 ]
shows the best linear fit to the resulting average ( which is extrapolated for @xmath80 ) .
a total of 189 objects , indicated by large dots in figure[fig4 ] , have a @xmath71 index exceeding that of the reference template at the same @xmath32 colour by more than four times the uncertainty on their @xmath71 values .
these are the objects with a @xmath71 excess at the @xmath10 level and , as we shall see later , most of them are bona - fide pms stars . since the contribution of the h@xmath3 line to the @xmath0 magnitude is completely negligible , the magnitude @xmath81 corresponding to the excess emission is simply : @xmath82 where the superscript _ obs _ refers to the observations and _ ref _ to the reference template .
once @xmath81 is determined in this way , the h@xmath3 emission line luminosity @xmath4 can be immediately obtained from the photometric zero point and absolute sensitivity of the instrumental set - up and from the distance to the sources .
we have assumed a distance to the lmc and , more specifically , to sn1987a of @xmath83kpc ( panagia et al .
1991 , later updated in panagia 1999 ) , whereas the photometric properties of the instrument were taken from the wfpc2 instrument handbook ( heyer & biretta 2004 ; namely inverse sensitivity photflam@xmath84 and photometric zero - point zp(vegamag)@xmath85 ) .
we derive a median value of the luminosity of the 189 objects with h@xmath3 excess of @xmath86erg / s or @xmath87l@xmath88 .
a histogram of the individual @xmath4 values is shown in figure[fig5 ] ( dashed line ) .
it should be noted that , because of the width of the h@xmath3 filter , @xmath81 includes a small contribution due to [ nii ] emission features at @xmath89 and @xmath90 .
generally , their intensities do not exceed @xmath91% and @xmath92% , respectively , of the h@xmath3 line intensity , and in many instances the [ nii ] doublet lines are much weaker , i.e. @xmath93 of the h@xmath3 line intensity ( e.g. edwards et al .
1987 ; hartigan , edwards & ghandour 1995 ) .
these intensities should be even lower for pms stars in the lmc because of the lower metallicity relative to the milky way .
however , to take a conservative approach , we have simply adopted an average value that is half of the maximum for galactic t - tauri stars and we give as uncertainty half of the measured range .
we will therefore adopt average values of @xmath94% and @xmath95% relative to the h@xmath3 intensity , for the 6584 and 6548 [ nii ] lines , respectively ( note that the uncertainties after the @xmath96 sign here are meant to show the entire span of the range , not just the @xmath97 value ) . considering the width of the h@xmath3 filter of the wfpc2 ,
only the 6548 [ nii ] line contaminates the h@xmath3 measurements , whereas for measurements made with the h@xmath3 filter on board the acs , both lines will be included ( albeit only in part for the 6548 line ) .
although the resulting corrections on the h@xmath3 intensity are quite small ( namely @xmath98 for the wfpc2 f656n filter and @xmath99 for the acs f658n filter , on average ) , they are systematic and it is therefore good practice to apply them in all cases . typically , the combined total uncertainty on our @xmath4 measurements is @xmath100 and is completely dominated by the statistical uncertainty on the h@xmath3 photometry . the systematic uncertainty on the distance to the field of sn1987a accounts for 5% ( see panagia 1999 ) , whereas the typical uncertainty on the absolute sensitivity of the instrumental setup is of order 3% . if the stars defining the reference template had no h@xmath3 absorption features ( we have established above that they do not have h@xmath3 in emission ) , their @xmath71 index would correspond to that of the pure continuum and could then be used to derive the equivalent width of the h@xmath3 emission line , @xmath68 . while this approximation might be valid for stars with a conspicuous h@xmath3 emission , it is clearly not applicable in general .
it is , however , possible in all cases to derive the level of the h@xmath3 continuum by using properly validated models atmospheres . as discussed above , the models of bessell et al .
( 1998 ) reproduce quite reliably the observed broad - band colours ( see e.g. figure[fig2 ] ) .
this means that , even though the models might lack the resolution necessary to realistically reproduce spectral lines ( hence the small discrepancy between the dashed line and the squares in figure[fig3 ] ) , the level of the continuum can be trusted .
one can , therefore , fit the continuum in those models in a range containing the h@xmath3 line ( e.g. @xmath101 ) and fold the resulting `` h@xmath3-less '' model spectra through the instrumental set - up ( e.g. with synphot ) , so as to derive the magnitude of the sole continuum in the h@xmath3 band ( @xmath102 ) as a function of spectral type or effective temperature .
we present in the appendix the relationships between @xmath0 , @xmath1 and @xmath102 for several hst instruments and filters .
the difference between the observed @xmath2 magnitude and @xmath102 provides a direct measure of @xmath68 , as we explain below .
we recall that the equivalent width of a line is defined as : @xmath103 where @xmath104 is the profile of the line , or the spectrum of the source divided by the intensity of the continuum , and the integration is extended over the spectral region corresponding to the line .
if the line profile is very narrow when compared to the width of the filter , i.e. if the line falls completely within the filter bandpass as is typical of emission lines , then @xmath69 is simply given by the relationship : @xmath105 \label{eq4}\ ] ] where @xmath106 is the rectangular width of the filter ( similar in definition to the equivalent width of a line ) , which depends on the characteristics of the filter .
values of @xmath106 for the h@xmath3 bands of past and current hst instruments are given in the appendix ( table[tab4 ] ) .
as in the case of @xmath4 , the statistical uncertainty on the equivalent width is dominated by the uncertainty on the @xmath2 magnitude , typically of order 15% in our selection . moreover ,
the value of @xmath69 obtained in this way is also subject to some systematic uncertainties produced by the model atmospheres used to interpolate the level of the continuum , such as e.g. metallicity , surface gravity or spectral resolution , and should therefore be taken with caution . on the other hand ,
we show in the appendix that the equivalent widths ( in absorption ) obtained by applying this method to the high - resolution model spectra of munari et al .
( 2005 ) are in excellent agreement with those obtained via standard spectro - photometry ( see figure[fig18 ] ) .
similarly , the values of @xmath69 that we obtain in the field of sn1987a are fully consistent with those of young pms stars . in figure[fig6 ] we show the value of @xmath69 as a function of @xmath32 for the @xmath73 selected stars .
as mentioned earlier , there are 189 objects with a @xmath81 excess above the @xmath10 level and they are indicated as thick symbols ( circles and triangles ) in figure[fig6 ] . since the equivalent width shown in figure[fig6 ] is that of the pure emission component , the spectra of stars with small @xmath27 , say @xmath107 , actually show an h@xmath3 in absorption .
for this reason , we conservatively ignore stars with measured @xmath108 , since this value is about the largest absorption equivalent width expected for normal stars ( see figure[fig18 ] in the appendix ) .
a total of 164 objects satisfy this condition , but , in light of our conservative threshold , this must be considered as a lower - limit to the number of objects with genuine h@xmath3 excess .
values of @xmath69 for the sample range from 10 to 650 , with a median of 19 .
these values are in excellent agreement with those obtained by panagia et al .
( 2000 ) from the @xmath26 colour excess of the stars in this field and represent a refinement of those measurements , since these are based on a more accurate determination of the level of the continuum .
these equivalent widths are typical of pms stars ( e.g. muzerolle , hartmann & calvet 1998 ) and about two orders of magnitude smaller than those expected for pure nebular emission ( see discussion in section6 ) .
while objects bluer than @xmath109 are probably ae / be stars , there are 133 redder stars with significant h@xmath3 excess ( marked as circles in figure[fig6 ] ) that are most likely bona - fide low - mass pms stars of various ages , as we will show in the following sections .
a histogram of their h@xmath3 luminosities is shown in figure[fig5 ] ( solid line ) .
it is generally accepted ( see e.g. hartmann et al . 1998 ) that the energy @xmath5 released by the magnetospheric accretion process goes towards ionising and heating the circumstellar gas . in this hypothesis ,
the bolometric accretion luminosity @xmath5 can be determined from a measurement of the _ reradiated energy _ and , in particular , via the measured h@xmath3 luminosity that is produced in the process . in order to determine the exact relationship between @xmath4 and @xmath5 , we have used literature values of @xmath5 and @xmath4 measurements as recently summarized by dahm ( 2008 ) for pms stars in the taurus
auriga association .
as pointed out by dahm himself , there may be a serious problem in determining an empirical relationship between @xmath5 and @xmath110 measurements because the two quantities were determined from non - simultaneous observations .
the intensity of the h@xmath3 line from pms sequence stars is known to vary by about 20% on a time scale of a few hours and by as much as a factor of 2 3 in a few days ( e.g. fernandez et al .
1995 ; smith et al . 1999 ; alencar et al .
2001 ) . for this reason ,
a relationship between @xmath4 and @xmath111 is necessarily rather uncertain , although a very clear trend can be seen in the @xmath112 plot shown by dahm ( 2008 ) .
his logarithmic best fit would provide a slope of @xmath113 for such a relationship . on the other hand , theoretical models ( e.g. muzerolle
, calvet & hartmann 1998 ) would predict logarithmic slopes of about unity for low accretion rates , hence for faint h@xmath3 luminosities , and shallower slopes for higher luminosities . in the absence of compelling evidence in favour or against a steep slope
, we will adopt a logarithmic slope of unity for the empirical @xmath112 relationship , i.e. a constant ratio @xmath114 , and determine the proportionality constant from an elementary fit of the data summarised by dahm ( 2008 ) . on this basis
we obtain : @xmath115 as anticipated , the resulting uncertainty on @xmath5 is rather large , as high as a factor of 3 , but this is the best one can obtain with the available data sets . in view of the convenience and the feasibility of determining @xmath5 from @xmath4 for pms outside our own galaxy , it would be of paramount importance to obtain a proper calibration of the @xmath112 relationshipby making simultaneous spectral observations of pms stars over a suitably wide wavelength baseline , i.e. covering at least the range 30008000 , and possibly extending it to the near ir with measurements of paschen and brackett lines .
high quality data of this type are already available for the orion nebula cluster , collected by dedicated projects both from ground based observatories and with the hst ( i.e. the legacy program on orion ; p.i .
massimo robberto ) and their analysis is in progress to provide a more accurate calibration of the @xmath112 relationship ( da rio et al . , in preparation ; robberto et al . , in preparation ; panagia et al . , in preparation ) .
the median value of the accretion luminosity thus obtained for our sample of 133 pms stars is @xmath116l@xmath88 , in good agreement with the value of @xmath117l@xmath88 derived from the u - band excess measured by romaniello et al .
( 2004 ) for stars in the same field , especially considering that their selection of stars with excess emission is less stringent than our @xmath10 level ( we discuss this point in section7 ) .
once @xmath5 is known , the mass accretion rate follows from the free - fall equation that links the luminosity released in the impact of the accretion flow with the rate of mass accretion @xmath6 according to the relationship : @xmath118 where @xmath119 is the gravitational constant , @xmath120 and @xmath121 the mass and photospheric radius of the star and @xmath122 the inner radius of the accretion disc .
the value of @xmath122 is rather uncertain and depends on how exactly the accretion disc is coupled with the magnetic field of the star .
following gullbring et al .
( 1998 ) , we adopt @xmath123 for all pms objects .
as regards @xmath121 , we derive it from the luminosity and effective temperature of the stars , which in turn come from the observed colour @xmath124 and magnitude ( @xmath0 ) , properly corrected for interstellar extinction as provided by romaniello ( 1998 ) , panagia et al .
( 2000 ) and romaniello et al . (
the adopted distance to sn1987a is @xmath125kpc ( panagia et al .
1991 ; panagia 1999 ) , as mentioned above .
the stellar mass @xmath120 was estimated by comparing the location of a star in the hertzsprung
russell ( h r ) diagram with the pms evolutionary tracks . as for the latter
, we adopted those of deglinnocenti et al .
( 2008 ; see also tognelli , prada moroni & deglinnocenti , in preparation ) , for metallicity @xmath126 or about one third @xmath127 , as appropriate for the lmc ( e.g. hill , andrievsky , & spite 1995 ; geha et al .
these new pms tracks were specifically computed with an updated version of the franec evolutionary code ( see chieffi & straniero 1989 and deglinnocenti et al .
2008 for details and cignoni et al .
2009 for an application ) .
however , in order to assess how differences in the evolutionary models affect our results , in section6.1 we also consider pms tracks from other authors , namely those of dantona & mazzitelli ( 1997 ) and siess , dufour & forestini ( 2000 ) .
combining equations[eq5 ] and [ eq6 ] , we can link the mass accretion rate @xmath6 , in units of m@xmath13yr@xmath14 , to @xmath4 : @xmath128 the distribution of mass accretion rates found in this way for the 133 pms stars in our sample is shown in figure[fig7 ] .
the median value of @xmath129m@xmath13yr@xmath14 is in remarkably good agreement with the median rate of @xmath130m@xmath13yr@xmath14 found by romaniello et al .
( 2004 ) from the u - band excess .
the uncertainty on @xmath6 is dominated by the approximate knowledge of the ratio of @xmath5 and @xmath4 in equation[eq5 ] . as explained above
, empirical measurements and theoretical models suggest a value of @xmath131 but with an uncertainty of a factor of three .
since the ratio , however , is the same for all stars , the comparison between different objects is not hampered by this uncertainty , as long as the statistical errors are small .
as for the other quantities in equation7 , we discuss here the sources of statistical uncertainty , while systematic errors are addressed separately in section6 . with our selection criteria ,
the typical uncertainty on @xmath4 is 15% and is dominated by random errors .
the uncertainty on @xmath121 is typically 7% , including a 5% systematic uncertainty on the distance modulus . as for the mass @xmath120 , since it is determined by comparing the location of a star in the h
r diagram with evolutionary tracks , both systematic and statistical uncertainties are important .
the uncertainty on the temperature is mostly statistical and of order 3% , while that on the luminosity is @xmath132 , comprising both random errors ( 1% uncertainty on the bolometric correction and 3% on the photometry ) and systematic effects ( 5% uncertainty on the distance modulus ) .
when we interpolate through the pms evolutionary tracks to estimate the mass , the uncertainties on @xmath133 and @xmath134 imply an error of @xmath135 on @xmath120 .
however , an even larger source of systematic uncertainty on @xmath120 comes from the evolutionary tracks , as we explain in section6 . in summary ,
the combined statistical uncertainty on @xmath6 is 17% .
having identified a population of bona - fide pms stars , with well defined accretion luminosities and mass accretion rates , it is instructive to place them in the h r diagram .
we do so in figure[fig8 ] where bona - fide pms stars are marked as large dots , with a size proportional to their @xmath6 value , according to the legend . also shown in the figure are the theoretical isochrones from the franec models of deglinnocenti et al .
( 2008 ) for @xmath126 and ages as indicated ( in units of myr , from the stellar birth line ; see palla & stahler 1993 ) .
the dashed line , corresponding to a 40 myr isochrone , defines in practice the zero - age main sequence ( ms ) .
it is noteworthy that the majority of pms objects in figure[fig8 ] are rather close to the ms and would have easily been missed if no information on their h@xmath3 excess had been available .
indeed , it is customary to identify pms stars in a colour magnitude diagram by searching for objects located above and to the right of the ms , since this is where one would expect to find very young objects .
this method was first used by gilmozzi et al .
( 1994 ) to identify a population of pms stars outside the milky way , namely in the lmc cluster ngc1850 .
more recent applications of this method include e.g. those of sirianni et al .
( 2000 ) , nota et al .
( 2006 ) and gouliermis et al .
( 2006 ) . unfortunately ,
this method of identification is not very reliable , since the existence of an age spread and the presence in the same field of a considerably older population , as well as unaccounted patchy absorption , all tend to fill up the parameter space between the ms and the birth line , thereby thwarting any attempts to identify pms stars on the basis of their effective temperature and luminosity alone . in the specific case of the sn1987a field studied here , panagia et al .
( 2000 ) have shown that the age spread is remarkable , with several generations of young stars with ages between 1 and 150myr superposed on a much older field population ( @xmath136gyr ) . in such circumstances , as figure[fig8 ] shows , broad - band photometry alone would have not identified as such most of our bona - fide pms stars . in accordance with their proximity to the ms ,
the majority of bona - fide pms stars turn out to be relatively old .
the ages of individual objects were determined by interpolating between the isochrones in the h
r diagram .
they range from 1 to @xmath137myr , with a median age of @xmath11myr , in very good agreement with the estimated age of both sn1987a and its nearby star 2 of @xmath138myr ( see scuderi et al . 1996 and references therein ) .
panagia et al .
( 2000 ) find that about 35% of the stars younger than 100myr in this field have a typical age of 12myr , again consistent with the estimated age of the bona - fide pms stars .
the relatively old age of our pms stars might appear at odds with the rather large mass accretion rates that we obtain for that population , namely @xmath139m@xmath13yr@xmath14 . according to current models of viscous disc evolution ( see hartmann et al .
1998 ; muzerolle et al .
2000 ) , the mass accretion rate of a @xmath140myr old pms population should be of order @xmath141m@xmath13yr@xmath14 . those models , shown in figure[fig9 ] as a solid line , adequately reproduce the trend of decreasing @xmath6 with stellar age observed for very low - mass galactic t - tauri stars ( see e.g. calvet et al . 2004 ) , but under - predict our measurements ( indicated by diamonds ) by over an order of magnitude . on the other hand , recent observations of pms stars in the cluster trumpler37 by sicilia
aguilar et al .
( 2006 ) , from whom the data - points in figure[fig9 ] are taken , indicate that the value of @xmath6 for @xmath142myr old g - type pms stars ( large circles ) is considerably larger , indeed of order @xmath16m@xmath13yr , in excellent agreement with our results . with a median mass of @xmath91m@xmath13and a median @xmath32 colour of @xmath143 ,
our bona - fide pms stars are fully consistent with the g spectral type and suggest that stars of higher mass have a higher @xmath6 value at all ages . as mentioned in the introduction , muzerolle et al .
( 2003 ; 2005 ) , natta et al .
( 2004 ; 2006 ) and calvet et al .
( 2004 ) suggest a dependence of the type @xmath21 from the analysis of a sample of low - mass and intermediate - mass t - tauri stars , albeit with a large uncertainty on the index ( see also clarke & pringle 2006 ) .
a similar trend is proposed by sicilia
aguilar et al . ( 2006 ) , yet with even lower statistical significance due to the limited mass range covered by their observations .
the range of masses that we cover here is also quite limited and does not allow us to address this issue in detail
. we will , however , do so in a forthcoming paper ( de marchi et al .
2008 , in preparation ) , where we compare the accretion rates in this field with those of a much younger and an order of magnitude more numerous population of pms stars in the small magellanic cloud . while in the sn1987a field the detection limits imposed by the wfpc2 photometry
do not allow us to reach stars with mass below @xmath144m@xmath88 , in that paper , based on acs observations , we will study stars down to @xmath145m@xmath88 .
the main sources of systematic uncertainty on the derived mass accretion rates are ( _ i _ ) discrepancies in the isochrones and evolutionary tracks , ( _ ii _ ) reddening , ( _ iii _ ) h@xmath3 emission generated by sources other than the accretion process , and ( _ iv _ ) the contribution of the nebular continuum to the colours of the stars .
we discuss all these effects in this section .
as explained in sections4 and 5 , the mass and age of pms stars are determined by interpolation of their location in the h
r diagram , using model evolutionary tracks and isochrones as references .
apart for possible inaccuracies in the models input physics and errors in the interpolation , the largest source of systematic uncertainty on the derived mass and age comes from the use of models that might not properly describe the stellar population under study ( e.g. because of the wrong metallicity ) and from differences between models of various authors .
for instance , if we had used siess et al.s ( 2000 ) tracks for metallicity 1/2 solar in place of those for 1/3 solar , the masses of our pms objects would be systematically higher by about 20% .
for metallicity 1/2 solar , using the tracks of siess et al .
( 2000 ) or those of dantona & mazzitelli ( 1997 ) results in masses that agree to within a few percent . on the other hand ,
when it comes to the determination of the age , using dantona & mazzitelli s ( 1997 ) isochrones one would derive ages twice as young for stars colder than @xmath146k , while for hotter stars ages would only be 20% lower . in summary ,
systematic uncertainties of order 20% are to be expected for the mass and possibly higher for the age determination .
dust extinction systematically displaces stars in the h
r diagram towards lower luminosities and effective temperatures .
this is shown graphically in figure[fig10 ] , for six model stars , three with @xmath147k ( large circles ) and three with @xmath148k ( large squares ) .
the effects of a reddening increase from @xmath149 to @xmath150 on the location of the stars in the h
r diagram is shown by the smaller symbols , corresponding to increments of @xmath151 .
we have adopted here the extinction law as measured specifically in the field of sn1987a by scuderi et al .
( 1996 ) . since the reddening vector in the h
r plane is almost parallel to the lines of constant radius ( as shown by the dot - dashed line for @xmath152r@xmath88 ) , extinction affects only marginally the estimate of the stellar radius
. however , at low temperatures the reddening vector crosses the evolutionary tracks ( solid lines ) , thus leading to an underestimate of the stellar mass , if no extinction correction is applied . in turn , this will result in a systematically higher value of the @xmath153 ratio and , therefore , in an overestimate of the mass accretion rate derived via equation[eq6 ] . in order to better gauge the uncertainty on the value of @xmath153 caused by unaccounted extinction
, we have determined the mass and radius of each one of the artificially reddened model stars shown in figure[fig10 ] , in exactly the same way in which we did that for bona - fide pms stars , and compared those with the known input values .
the result is shown graphically in figure[fig11 ] , where , the relation between the derived and input @xmath153 ratios is shown as a function of colour excess @xmath154 .
regardless of the input parameters ( @xmath134 , @xmath133 or @xmath153 ) , it appears that over the range explored here the derived @xmath153 ratio increases slowly with colour excess ( see dashed line that best fits the median values indicated by the crosses ) . in particular , underestimating the @xmath154 colour excess by @xmath145mag would lead to a 30% overestimate of @xmath153 ( and hence of @xmath24 ) .
this appears unlikely , considering that the best estimate of the total extinction including the milky way component towards star 2 , one of the two bright close companions to sn1987a , is @xmath155 ( scuderi et al .
1996 ) , with a @xmath97 dispersion over the field of @xmath156 ( romaniello 1998 ; panagia et al .
according to zaritsky et al . (
2004 ) , the average reddening towards cool ( 5500k@xmath1576500k ) stars in the lmc is @xmath158 , thus implying @xmath159 .
therefore , for an average lmc star forming region , omitting the extinction correction would result in a 10% overestimate of the mass accretion rate . this error , albeit systematic , is smaller than most other systematic uncertainties and comparable to the measurement errors .
nonetheless , we would like to stress here that the effects of patchy absorption can be more severe in regions of high extinction .
therefore , whenever possible , one should apply extinction corrections to each individual stars , as it was done for the photometric catalogue of romaniello et al .
( 2002 ) that we use here .
the underlying assumption in the determination of the mass accretion rate of pms stars is that the gravitational energy @xmath5 released in the accretion process goes into heating of the gas at the boundary layer ( see e.g. hartmann et al .
1998 ) . in this case
, the luminosity of the h@xmath3 emission line can provide a measure of the accretion energy , because it acts as a natural `` detector '' of the luminosity released in the accretion process .
thus , the issue to address is whether the measured @xmath4 is produced only and entirely as a result of the energy released in the accretion process .
obviously , depending on the geometry and orientation of the circumstellar disc and on the specifics of the accretion process , some of the ionising energy could escape without effectively ionising the local gas .
this would result in an underestimate of the true mass accretion rate for some stars , but this is true of any method that uses h emission lines as indicators of the accretion process and , therefore , it does not affect the comparison of our results with those of others ( see figure[fig9 ] ) .
one possibility that we can readily exclude is that the h@xmath3 emission that we detect is due to chromospheric activity .
the total chromospheric emission of a solar - type star on the subgiant branch corresponds to a luminosity of about @xmath160l@xmath161 ( ulmschneider 1979 ; pasquini et al . 2000 ) , i.e. two orders of magnitude less than what we measure . on the other hand
, we must also consider the possibility that h@xmath3 emission might occur in discrete h knots along the line of sight , in places unrelated with the pms object .
furthermore , hot nearby massive stars may ionise the h gas in which pms stars are immersed , eventually producing h@xmath3 emission when the gas recombines .
both these effects would result in a overestimate of @xmath6 , but we show below that the probability of this happening is very small . as for intervening h@xmath3 emission along the line of sight , it would have to arise in knots of ionised h , e.g. very compact hii regions , contained within our photometric apertures . however ,
if such knots exist , they should not project only against stars and , therefore , narrow - band images should reveal h@xmath3 emission where broad - band imaging shows no or very low signal ( see section 6.3 ) .
since this is not observed , statistical arguments rule out this possibility as a significant source of the observed h@xmath3 excess emission . while it is possible that some of the h@xmath3 emission that we detect is due to diffuse nebular emission in the hii region not powered by the accretion process , if the emission is extended and uniform over an area of at least @xmath162 radius around the star its contribution cancels out with the rest of the background when we perform aperture photometry ( using an aperture of @xmath163 radius for the object and an annulus of @xmath164 for the background ) .
the subtraction would not work if the emission were not uniform , as for example in the case of a filament that projects over the star but that does not cover completely the background annulus .
however , we have visually inspected all objects with excess h@xmath3 emission and excluded the few cases where a spurious filament of this type could be seen ( see romaniello 1998 ; panagia et al .
2000 ) .
another case in which the emission does not cancel out is when the gas ionised by an external source is associated to the star and completely contained within our photometric aperture .
we recall here that the typical size of a circumstellar disc is of order 100 au ( e.g. hartmann et al . 1998 ) , or about @xmath165 pixel at the distance of sn1987a , and that the median @xmath4 that we measure for our pms stars is @xmath166ergs@xmath14 , or about @xmath87l@xmath88 ( see figure[fig5 ] ) .
we can then calculate how close a pms stars should be to a hot , uv - bright star for the latter to flood the pms circumstellar disc with enough ionising radiation to cause at least 10% of the h@xmath3 luminosity that we measure from that source .
let @xmath167au be the external radius of the circumstellar disc and @xmath168 the distance to an ionising star that illuminates the disc , assumed to have zero inclination ( i.e. face on ) , so that @xmath169 is the solid angle subtended by the disc .
let @xmath170 be the recombination coefficient to the first excited level in an optically thick h gas .
if @xmath0 is the volume of the gas and @xmath171 and @xmath172 are the electron and proton densities per unit volume , the global ionisation equilibrium equation ( case b ; see baker & menzel 1938 ) is then : @xmath173 where @xmath174 is the rate of ionising photons produced by the external star . denoting with @xmath175 the h@xmath3 emission coefficient for the same gas
, the luminosity of the h@xmath3 recombination line can be written as : @xmath176 which allows us to express the luminosity of the h@xmath3 emission caused by the external star as a function of its ionising photon rate @xmath174 : @xmath177 the minimum distance beyond which the contribution of the ionising star to the total h@xmath3 emission becomes negligible ( i.e. 10% or less ) is given by : @xmath178 table[tab1 ] gives the value of @xmath179 for a typical h@xmath3 luminosity of @xmath87l@xmath88 , as a function of the spectral type of the ionising star .
here we have used the compilation of panagia ( 1973 ) for the appropriate stellar parameters .
the table also provides the minimum distance @xmath180 in pc and in arcsec ( at the distance of sn1987a ) for a circumstellar disc of radius @xmath167au .
we recall here that 1 arcsec corresponds to 10 wfpc2 pixels .
the brightest star in the field studied here is an object with @xmath181 and @xmath182 , with an implied mass of @xmath183m@xmath13and an age less than 1myr ( see panagia et al .
the corresponding value of @xmath174 , from panagia ( 1973 ) , is @xmath184s@xmath14 , thus implying @xmath185pc or 38 wfpc2 pixel for a gas temperature of 10000k .
a careful inspection of the images shows that none of our candidate bona - fide pms stars falls closer to this star than @xmath180 .
the same applies to the second and third brightest stars in the field , respectively with @xmath186 , @xmath187 and @xmath188 , @xmath189 , as well as to all fainter objects .
lcccccc o4 & 50000 & 6.11 & 49.93 & 3.42e-08 & 1.31 & 5.26 + o6 & 42000 & 5.40 & 49.08 & 2.43e-07 & 0.49 & 1.97 + o7 & 36500 & 4.81 & 48.62 & 6.93e-07 & 0.29 & 1.17 + o9 & 34500 & 4.66 & 48.08 & 2.43e-06 & 0.16 & 0.62 + b0 & 30900 & 4.40 & 47.36 & 1.27e-05 & 0.07 & 0.27 + b1 & 22600 & 3.72 & 45.29 & 1.50e-03 & 0.01 & 0.03 + [ tab1 ] it is convenient to calculate , for each pms object , the total h@xmath3 luminosity @xmath190 caused by all neighbouring ionising stars , expressed as : @xmath191 where @xmath192 is the ionising photon rate of star @xmath193 and @xmath194 is its distance from the pms object being considered .
besides the three brightest stars mentioned above , we have included as potential sources of the detected h@xmath3 emission an additional 12 stars in the field with @xmath195 and @xmath196 , as classified by romaniello et al .
we note that these estimates of @xmath190 are firm upper limits because we are assuming that _
( i ) _ all disks are _ face - on _ relative to the hot stars , and that _ ( ii ) _ the projected distances on the sky are indeed the actual distances from the hot stars .
the values of @xmath197 obtained in this way for the 133 bona - fide pms stars span the range @xmath198l@xmath88 , or typically three orders of magnitude less than the median pms h@xmath3 luminosity , as shown by the histogram in figure[fig12 ] .
we can , therefore , conclude that in this specific field the contribution of hot ionising stars to the measured excess h@xmath3 emission is negligible .
we must stress , however , that the effect might be much more pronounced in younger star forming regions , such as the orion nebula or 30 dor .
finally , consideration must be given to the role of the nebular continuum , i.e. radiation caused by bound
free and free free transitions in the gas and unrelated to the stellar photosphere .
if present , nebular continuum will add to the intrinsic continuum of the star , thereby affecting both the observed total level and the slope .
this could ultimately alter the measured broad - band colours of the source , thereby thwarting our attempts to infer the level of the continuum in the h@xmath3 band from the observed @xmath0 and @xmath1 magnitudes .
it is important to estimate the magnitude of this effect , as our method relies on the intrinsic @xmath32 colour of a star to derive its @xmath81 ( see equation[eq2 ] ) .
fortunately , the contribution of the nebular continuum appears to be insignificant . to prove this ,
we have assumed a fully ionised gas of pure h , considering only bound free and free free transitions and ignoring the contribution to the continuum from two - photon emission ( spitzer & greenstein 1951 ) .
using osterbrock s ( 1989 , chapter 4 ) tabulations , we find h@xmath3 line intensity and h@xmath3 continuum fluxes of the nebular gas as shown in table[tab2 ] , for gas electron temperatures in the range 5000k 20000k .
the purely nebular @xmath69 ranges from @xmath199 to @xmath200 , or two orders of magnitude higher than what we measure for pms objects .
we can , therefore , safely conclude that the nebular contribution to the continuum is negligible ( less than 1% ) . furthermore , we find that for gas temperatures in the range 5000k 10000k the @xmath32 colour of the nebular continuum varies from @xmath201 to @xmath202 , spanning a range typical of g k type stars .
thus the effects of the nebular continuum on the @xmath32 colour of pms stars remains insignificant even for the objects with the highest @xmath68 in our sample .
rrrrr 5000k & @xmath203 & @xmath204 & @xmath205 & 9378 + 10000k & @xmath206 & @xmath207 & @xmath208 & 6763 + 20000k & @xmath209 & @xmath210 & @xmath211 & 4764 + [ tab2 ]
we have developed and successfully tested a new self - consistent method to reliably identify all pms objects actively undergoing mass accretion in a resolved stellar population , regardless of their age , without requiring spectroscopy .
the method combines broad - band @xmath0 and @xmath1 photometry with narrow - band @xmath2 imaging to : ( 1 ) identify all stars with excess h@xmath3 emission using as a reference template of the photospheric luminosity in the h@xmath3 band the average @xmath71 colour of stars with small photometric errors ; ( 2 ) convert the excess h@xmath3 magnitude into h@xmath3 luminosity @xmath4 via the absolute photometric calibration of the h@xmath3 band ; ( 3 ) convert , if desired , the excess h@xmath3 magnitude into the h@xmath3 emission equivalent width , using model spectra to fit the level of the continuum in the h@xmath3 band ; ( 4 ) derive the accretion luminosity @xmath5 from @xmath4 using a relationship based on recent literature values ; ( 5 ) obtain the mass accretion rate @xmath212 from the free - fall equation that links it to @xmath5 via the stellar parameters ( mass and radius ) , that we derive from the h
r diagram using pre - main sequence tracks for the appropriate metallicity .
since no spectroscopy of individual objects is needed , observations requiring just a few hours of exposure time can suffice to reveal hundreds of pms stars simultaneously , determine their h@xmath3 luminosity with an uncertainty of less than 15% and derive their mass accretion rates with the same accuracy as allowed by conventional spectral line analysis . as regards the h@xmath3 luminosity ,
the virtue of this new method is that it derives the reference luminosity of the stellar photosphere inside the specific h@xmath3 band simply from the average @xmath71 colour of stars with the same @xmath32 index ( see figure[fig4 ] ) .
this is possible because , at any given time , the majority of the stars in a stellar field have no excess h@xmath3 emission . obviously , the average luminosity of the stellar photosphere in the h@xmath3 band reflects more or less pronounced h@xmath3 absorption features , depending on the spectral type of the objects , and precisely for this reason it defines the reference luminosity with respect to which any excess emission should be sought and measured .
the uncertainty associated with the derived @xmath4 is dominated by statistical errors in the photometry and does not exceed 15% for our @xmath10 selection ( see section3.1 ) .
note that this approach is also applicable to populations comprising mostly very young objects , such as the orion nebula , since excess h@xmath3 emission is found in only about one third of the pms stars at any given time and , therefore , the majority of the very young objects will not have a conspicuous h@xmath3 emission , thus correctly defining the reference template ( see section3.1 ) . in spite of the good accuracy on @xmath4 ,
deriving the mass accretion rate requires the use of a rather uncertain relationship linking @xmath4 to the accretion luminosity ( see section4 and also dahm 2008 ) .
the uncertainty on the relationship can have a systematic effect of up to a factor of 3 on the value of @xmath24 .
this is the unfortunate consequence of our incomplete understanding of the physics of the accretion process , plagued by the lack of direct measurements of the bolometric accretion luminosity @xmath5 , as well as by the paucity of simultaneous measurements of the hot continuum excess emission and of the emission line fluxes in h@xmath3 or other lines ( e.g. pa@xmath22 and br@xmath23 ) .
although annoying , the systematic uncertainty of up to a factor of 3 on @xmath212 is similar to that plaguing age determinations based on the comparison with theoretical isochrones , which can reach a factor of 2 ( see section6.1 ) . on the other hand ,
these systematics affect in the same way all determinations of @xmath6 that make use of emission lines diagnostics , including those based on a detailed study of the profile and intensity of the h@xmath3 line ( e.g. muzerolle et al .
1998b ) , since their calibration relies on the same measurements .
for this reason , the relative values of @xmath6 that we obtain are trustworthy and the comparison of our results with those obtained with other methods remains meaningful . as discussed in sections3 and 4 , as an application of our method , we have studied the field of sn1987a , where we identify 133 bona - fide pms stars , based on their large h@xmath3 excess .
the median value of the mass accretion rate that we find for these stars , namely @xmath213m@xmath13yr@xmath14 , is in very good agreement with the median value of @xmath214m@xmath13yr@xmath14 found by romaniello et al .
( 2004 ) from the analysis of the distribution of the @xmath215 colour excess in the same field . while we have identified 133 bona - fide pms stars with an h@xmath3 excess above the @xmath10 level , romaniello et al .
conclude that 765 stars with an excess @xmath15-band luminosity larger than @xmath216l@xmath88 could be bona - fide pms candidates , albeit at the @xmath217 level .
the approach followed by romaniello et al .
( 2004 ) is statistical in nature and , while providing a solid estimate of the average mass accretion rate , it does not allow one to reliably identify individual pms stars
. therefore , romaniello et al . turned to the h@xmath3 excess emission ( namely @xmath26 ) as a more direct indicator of ongoing accretion and found a strong correlation between the @xmath15-band continuum excess and @xmath69 for stars with @xmath218 and @xmath219 .
not surprisingly , the more stringent threshold on the equivalent width adopted in the present study , i.e. @xmath220 , results in a smaller number of identified of bona - fide pms stars , but also in a much smaller uncertainty .
indeed , the method that we present here is more direct than the one based on the @xmath15-band excess , since it does not require the use of model atmospheres to predict the level of the intrinsic photospheric continuum in the @xmath15 band from the observed optical colours , and as such it results in a more reliable determination of the accretion luminosity . our method , therefore , makes it possible for the first time to carry out a quantitative and systematic study of the properties of a large number of pms stars , including those in dense and/or distant star forming regions .
this allows us to tackle some still open questions as to the evolution of both the stars and their discs .
for instance , in the field of sn1987a , we have studied how the mass accretion rate changes with time as our bona - fide pms stars approach their main sequence .
we find that @xmath6 decreases more slowly with time than what is observed for low - mass t - tauri stars in the galaxy ( see section5 ) and predicted by models of viscous disc evolution ( hartmann et al .
however , our results are in excellent agreement with the ( so far limited ) measurements of the mass accretion rate of g type stars in the milky way ( e.g. sicilia aguilar et al .
2006 ) , thus lending credit to the hypothesis that @xmath6 might depend on the stellar mass .
moreover , our method allows us to investigate whether and how the accretion process is affected by the chemical composition and density of the molecular clouds or by the proximity of hot , massive ionising stars .
we will address the effects of metallicity in a forthcoming paper ( de marchi et al .
, in preparation ) devoted to a comparison of star formation in the small and large magellanic clouds . as for the effect of hot massive stars ,
we have studied the distribution of the 133 bona - fide pms objects with respect to the baricentre of the 15 hottest stars in the field , described in section6.3 .
we find a clear anti - correlation , shown in figure[fig13 ] , between the frequency of the identified pms stars and their distance from the ionising objects ( marked with blue star symbols in figure[fig14 ] ) .
this effect is not seen for non - pms objects of similar brightness , whose distribution is uniform over the whole field , confirming that the trend is not due to problems in detecting faint objects near the brightest stars .
the trend remains the same whether the baricentre of the hot stars is computed using as weight their bolometric luminosity or their ionising flux .
this anti - correlation is quite remarkable in several ways , as we discuss below . in figure[fig14 ]
we show a schematic map of the distribution of all 133 pms objects with respect to the 15 ionising stars .
the baricentre of the ionising flux is shown by the large square and concentric annuli are drawn from it , in increments of 35 . for reference , a red ellipse marks the position of sn1987a .
symbols of different colours indicate pms stars with different h@xmath3 luminosity , from red for @xmath221l@xmath88 to white for @xmath222l@xmath88 , with green diamonds corresponding to intermediate values .
an inspection to figure[fig14 ] immediately reveals two important facts : ( _ i _ ) there are very few pms stars near the baricentre of the massive objects and ( _ ii _ ) their h@xmath3 luminosity is systematically lower than that of pms stars farther away .
panagia et al .
( 2000 ) had already pointed out that there is no overdensity of young pms objects around hot massive stars in this field .
this shows that low - mass and massive stars do not follow each other , suggesting that their formation mechanisms are separate and distinct .
but what figure[fig14 ] additionally shows is that there are very few pms stars with strong h@xmath3 luminosity ( red squares ) around the young massive stars , with only 20% of them within a 70arcsec radius of the baricentre .
conversely , stars with low h@xmath3 luminosity ( white squares ) are distributed rather uniformly across the field . with a median age of @xmath11myr ,
most pms objects are much older than the @xmath223myr old hot stars and must belong to a previous generation .
thus , no spatial relationship ( whether a correlation or anti - correlation ) should be expected between the distribution of the two types of objects . on the other hand , while we can not exclude that originally fewer pms stars formed in the region now populated by young massive stars , there are indications that the local paucity of pms objects is only apparent and caused by the presence of the hot young stars . the fact that pms objects near the youngest hot stars are both less numerous and fainter in h@xmath3 emission suggests that their circumstellar discs have been considerably eroded and made less efficient by enhanced photo - evaporation .
we have shown earlier ( see section6.3 and figure[fig12 ] ) that the instantaneous contribution to the h@xmath3 luminosity from the ionising radiation of nearby massive stars is negligible .
the results shown in figure[fig14 ] confirms that this is the case .
if it were not and the observed h@xmath3 luminosity were mostly due to fluorescence of the circumstellar discs under the effect of the ionising radiation of massive stars , the radial distribution should be opposite to the one observed , with brighter h@xmath3 luminosities for pms stars closer to the baricentre of the youngest hot stars .
thus , if the apparent paucity of pms objects in their vicinity is the result of photo - evaporation , the latter must be mostly due to the effects of non - ionising far ultraviolet ( fuv ) radiation .
when integrated over the lifetime of massive stars ( @xmath224myr ) , their fuv radiation can be quite significant for the disruption of nearby circumstellar discs .
in particular , this also means that there might be many more objects still in their pms phase , close to the hot stars , but we have no way to identify them as such , since the photo - evaporation of their discs has stopped the accretion process .
recent theoretical studies of the erosion of circumstellar discs via photo - evaporation due to external fuv radiation ( e.g. adams et al .
2004 ) or to the pms star itself ( e.g. alexander , clarke & pringle 2006a ; 2006b ) suggest timescales of order 810myr for the complete dissolution of the discs .
although this is the right order of magnitude for the timescale of the process that we see , there are still some discrepancies . for instance , it is clear that if discs were to completely dissolve under the effect of their own central star within 510myr , as suggested by alexander et al .
( 2006b ) , the majority of our pms objects , with a median age of @xmath11myr , should not show any sign of accretion .
similarly , with a typical age of 2myr , the hottest stars in our field would have not had time to deposit enough uv photons on to the pms discs to seriously erode them , if the process is to take of order 10myr ( adams et al .
2004 ) . while recently clarke ( 2007 ) suggested that the effect of nearby massive young stars may be stronger , leading to disc exhaustion timescales of order 2myr
, her calculations apply to the immediate vicinity ( @xmath225pc ) of very massive stars , such as the o6 star @xmath226 in orion .
the density of massive stars in the field of sn1987a is much lower than that present in the orion nebula and the distances to our pms objects typically an order of magnitude larger than those considered by clarke ( 2007 ) . nevertheless ,
while it is clear that the theory of disc photo - evaporation does not yet capture all the details , we believe that it can greatly benefit from the observations presented here .
in particular , it will be necessary to explain which mechanisms allow isolated circumstellar discs to feed their central pms stars for well over 20myr , at least in a low - metallicity , low - density environment such as the field of the lmc .
at the same time , the models should justify the quick ( @xmath227myr ) demise of the pms discs within a ( projected ) radius of @xmath19pc of a handful of late o stars . extending our investigation ,
as we plan to do , to more and denser regions , such as ngc1850 and 30dor in the large magellanic cloud and ngc346 and ngc602 in the small magellanic cloud , will offer increased statistics and a much wider variety of star formation environments .
we are very thankful to pier giorgio prada moroni and scilla deglinnocenti for sharing with us their pre - main sequence evolutionary tracks ahead of publication and to massimo robberto for useful discussions .
gdm is grateful to stsci and eso for the hospitality , via their science visitor programmes , during the preparation of this paper .
the research of np was supported in part by nasa grant hst - go-11547.06-a .
we are also indebted to an anonymous referee for precious comments and suggestions that have helped us to improve the presentation of our work .
lclll wfpc2 & @xmath228 & @xmath229 & @xmath230 & @xmath231 + & @xmath228 & @xmath232 & @xmath232 & @xmath233 + & @xmath234 & & @xmath235 & @xmath236 + & @xmath234 & & @xmath237 & @xmath238 + acs/ & @xmath239 & @xmath240 & @xmath241 & @xmath242 + wfc & @xmath239 & @xmath243 & @xmath244 & @xmath243 + & @xmath245 & & @xmath246 & @xmath247 + & @xmath245 & & @xmath248 & @xmath249 + acs/ & @xmath239 & @xmath250 & @xmath251 & @xmath252 + hrc & @xmath239 & @xmath253 & @xmath254 & @xmath255 + & @xmath245 & & @xmath256 & @xmath257 + & @xmath245 & & @xmath258 & @xmath259 + wfc3/ & @xmath239 & @xmath260 & @xmath261 & @xmath262 + uvis & @xmath239 & @xmath263 & @xmath264 & @xmath265 + & @xmath234 & & @xmath266 & @xmath267 + & @xmath234 & & @xmath268 & @xmath269 + johnson- & @xmath29 & @xmath270 & @xmath271 & @xmath272 + cousin & & & & [ tab3 ] over the years , a large number of observations of stellar populations have been collected with the hst / wfpc2 and hst / acs through the @xmath0-like filters f555w and f606w and the @xmath1-like filter f814w .
we have therefore computed colour relationships between the magnitude in the @xmath29-like band f675w ( @xmath228 ) and the @xmath32-like colours @xmath273 and @xmath274 . to this aim ,
we have used both observed and theoretical stellar spectra , taken respectively from the bruzual - persson - gunn - stryker ( bpgs ) spectrophotometry atlas ( see gunn & stryker 1983 ) and from the stellar atmosphere models of bessell et al .
( 1998 ) .
the bpgs atlas contains nearby bright dwarfs and giants of near - solar metallicity , whereas from the library of bessell et al .
( 1998 ) we have selected specifically those models with metallicity index @xmath55=-0.5 $ ] , as appropriate for the lmc ( dufour 1984 ) , surface gravity @xmath275 and effective temperatures in the range 3500 k @xmath276 40000k . using the standard hst synthetic photometry package synphot ( laidler et al . 2008 ) , we have folded these spectra through the instrumental response of the wfpc2 , the acs / wfc , the acs / hrc and the wfc3/uvis cameras , in order to obtain the expected magnitudes in the bands listed above . for comparison with ground - based photometry ,
we have also calculated the expected magnitudes through the johnsn
cousin @xmath277 and @xmath1 bands .
we show in figure[fig15 ] , for the wfpc2 filter system , the predicted @xmath278 colour as a function of @xmath273 for all the stars in the bpgs atlas .
the distribution is remarkably linear and the dashed line represents the best fit in the colour range @xmath279 ( filled symbols ) , corresponding to the relationship @xmath280 .
the colour colour diagram in the same bands based on the models of bessell et al .
( 1998 ) is shown in figure[fig16 ] and the relationship is practically indistinguishable , namely @xmath281 . in this case , the dashed line shows the least - square fit to stars in the range 4000k @xmath282 10000k , marked by the filled symbols .
as already noted in section2 , it appears that neither the luminosity class nor the metallicity has a noticeable impact on the colour relationship in these bands .
this is also the case when we consider the @xmath274 colour , for which the relationship with @xmath228 is given in table[tab3 ] . in the same table
we also list the results for the acs / wfc , acs / hrc and wfc3/uvis cameras , as well as those for the standard johnson cousin system normally used from the ground ( note that the @xmath29 and @xmath1 bands in this system do not coincide with the @xmath29 and @xmath1 bands of johnson s original system , as mentioned in section2 ) .
as expected , the coefficients in the colour relationship vary considerably with the band , and differences exist also between the same band on various hst cameras ( or , at least , bands with the same name ) . as briefly discussed in section3 , the magnitude corresponding to the level of the continuum in the h@xmath3 band can be derived by means of properly validated model atmospheres , such as those of bessell et al .
( 1998 ) that we have shown to reproduce quite accurately the observed broad - band colours ( see figure[fig2 ] ) .
this means that , even though the models might lack the spectral resolution necessary to realistically reproduce spectral lines ( hence the small discrepancy between the dashed lines and the squares in figure[fig2 ] ) , the level of the continuum can be trusted .
we have , therefore , fitted the continuum in the bessell et al .
( 1998 ) spectra over the range @xmath283 and have replaced the actual spectrum in that range with the interpolated continuum .
finally , we have used synphot to fold the model spectra , modified in this way , through the instrumental set - up and derive in this way the continuum h@xmath3 magnitudes ( @xmath102 ) for the various h@xmath3 filters on - board the hst , as a function of the spectral type or @xmath133 of the star . with this information
it is now possible to build colour colour diagrams like those of figure[fig15 ] and [ fig16 ] , and look for a colour relationship between @xmath102 and @xmath32 . in figure[fig17 ] we show the specific case of the wfpc2 .
the @xmath284 colour varies very smoothly and very slowly with the effective temperature and is a remarkably linear function of the @xmath273 colour over the whole range , and particularly for stars with 4000k @xmath285k ( filled circles ) with which we are mostly concerned in this work .
similar relationships exist for other bands and photometric systems and they are given in table[tab3 ] .
we note that the constant @xmath286 in the colour relationship accounts for the conspicuous h@xmath3 absorption line in the spectrum of vega , which defines the standard for the vegamag magnitude system . as indicated in section3 , the difference between the observed @xmath2 magnitude and @xmath102 can be converted to the equivalent width @xmath287 by means of equation[eq4 ] . to prove this , we have applied equation[eq4 ] to the model atmospheres of munari et al .
( 2005 ) , which provide spectra with a resolving power of @xmath288 and a 1 sampling , and compared the results to literature measurements of h@xmath3 equivalent widths ( in absorption ) .
ducati ( 1981 ) compiled an extensive homogeneous catalogue of h@xmath3 line measurements , from photographic and photoelectric techniques , for about 2300 stars .
we have converted ducati s index @xmath3 to @xmath289h@xmath290 following the calibration provided by strauss & ducati ( 1981 ) and cester et al .
the values of @xmath291h@xmath290 obtained in this way are shown in figure[fig18 ] , where crosses correspond to dwarfs and boxes to giants .
ducati s ( 1981 ) stars are classified according to their spectral type , which we have converted to @xmath32 following bessell ( 1990 ) .
as for the application of equation[eq4 ] to the models of munari et al .
( 2005 ) , this was done for model spectra with @xmath292 , @xmath293 and solar metallicity and for the wfpc2 photometric system , namely f555w for v , f814w for i and f656n for h@xmath3 .
it should be noted that , for stars with broad h@xmath3 lines ( around spectral type a ) , the equivalent width determined in this way is systematically smaller than the true one , since part of the line wings fall outside of the h@xmath3 filter .
however , the measurements of ducati ( 1981 ) are subject to exactly the same effect , since they were derived in a similar fashion and with a h@xmath3 filter of comparable effective width .
the value of @xmath289h@xmath290 obtained in this way is shown in figure[fig18 ] ( solid line ) , as a function of @xmath32 .
the agreement with the observations is very satisfactory .
lccc wfpc2 & f656n & @xmath294 & @xmath295 + acs / wfc & f658n & @xmath296 & @xmath297 + acs / hrc & f658n & @xmath296 & @xmath298 + wfc3 & f656n & @xmath299 & @xmath300 + & f658n & @xmath301 & @xmath302 [ tab4 ] sirianni , m. , jee , m. , benitez , n. , blakeslee , j. , martel , a. , meurer , g. , clampin , m. , de marchi , g. , ford , h. c. , gilliland , r. , hartig , g. , illingworth , g. , mack , j. , mccann , w. 2005 , pasp , 117 , 1049 | we have developed and successfully tested a new self - consistent method to reliably identify pre - main sequence ( pms ) objects actively undergoing mass accretion in a resolved stellar population , regardless of their age .
the method does not require spectroscopy and combines broad - band @xmath0 and @xmath1 photometry with narrow - band @xmath2 imaging to : ( 1 ) identify all stars with excess h@xmath3 emission ; ( 2 ) convert the excess h@xmath3 magnitude into h@xmath3 luminosity @xmath4 ; ( 3 ) estimate the h@xmath3 emission equivalent width ; ( 4 ) derive the accretion luminosity @xmath5 from @xmath4 ; and finally ( 5 ) obtain the mass accretion rate @xmath6 from @xmath5 and the stellar parameters ( mass and radius ) . by selecting stars with an accuracy of 15% or better in the h@xmath3 photometry , the statistical uncertainty on the derived @xmath6 is typically @xmath7 and is dictated by the precision of the h@xmath3 photometry .
systematic uncertainties , of up to a factor of 3 on the value of @xmath6 , are caused by our incomplete understanding of the physics of the accretion process and affect all determinations of the mass accretion rate , including those based on a spectroscopic h@xmath3 line analysis . as an application of our method
, we study the accretion process in a field of @xmath8 arcmin@xmath9 around sn1987a , using existing hubble space telescope photometry . we identify as bona - fide pms stars a total of 133 objects with a h@xmath3 excess above the @xmath10 level and a median age of @xmath11myr .
their median mass accretion rate of @xmath12m@xmath13yr@xmath14 is in excellent agreement with previous determinations based on the @xmath15-band excess of the stars in the same field , as well as with the value measured for g - type pms stars in the milky way .
the accretion luminosity of these pms objects shows a strong dependence on their distance from a group of hot massive stars in the field and suggests that the ultraviolet radiation of the latter is rapidly eroding the circumstellar discs around pms stars . |
recently , a theorem announced by schrder and simpson in @xcite and presented with a proof in @xcite has attracted some attention . before stating this result we need some preparations .
let @xmath1 be an arbitrary topological space .
the emphasis is on non - hausdorff spaces like directed complete partially ordered sets (= dcpos ) with the scott topology ( see @xcite ) and quotients of countably based spaces ( qcb - spaces , see @xcite ) .
but , of course , compact and locally compact hausdorff spaces are classical examples .
we consider functions defined on @xmath1 with values in @xmath2 , the set @xmath3 of nonnegative real numbers extended by @xmath4 as top element .
such functions @xmath5 are _ lower semicontinuous _ if , for every @xmath6 , the set @xmath7 is open in @xmath1 .
we denote by @xmath8 the set of all lower semicontinuous functions @xmath9 with the pointwise defined order .
the pointwise supremum of any family of lower semicontinuous functions is lower semicontinuous so that @xmath8 is a complete lattice .
the pointwise defined sum @xmath10 of two lower semicontinuous functions @xmath11 as well as the scalar multiple @xmath12 for @xmath13 are also lower semicontinuous .
the maps @xmath14 which are linear and scott - continuous , that is , those satisfying @xmath15{$\scriptstyle\uparrow$}}}}}_i f_i ) = { \mathop{\bigvee{}^{^{\,\makebox[0pt]{$\scriptstyle\uparrow$}}}}}_i\mu(f_i)\ ] ] for all @xmath16 , @xmath6 and for every directed family @xmath17 in @xmath8 , form the _ valuation powerdomain _ @xmath18 .
on @xmath19 one also has a pointwise defined order , addition and multiplication by nonnegative real numbers .
we endow @xmath19 with the weak@xmath0upper topology , that is , the coarsest topology such that the evaluation maps @xmath20 remain lower semicontinuous for all @xmath21 .
the schrder - simpson theorem says : [ ss ] for every linear functional @xmath22 which is lower semicontinuous with respect to the weak@xmath0upper topology , there is an @xmath23 such that @xmath24 for all @xmath25 .
this theorem sounds exactly like a simple general fact from classical functional analysis ( see , e.g. , ( * ? ? ?
* chapter iv , 1.2 ) ) : _ let @xmath26 be a topological vector space over the field of real numbers and @xmath27 the dual vector space of all continuous linear functionals @xmath28 , then every linear functional @xmath29 defined on the topological dual @xmath27 which is continuous with respect to the weak@xmath0 topology is the evaluation at some @xmath30 , that is , @xmath31 for every @xmath32 .
_ here the weak@xmath0 topology is the coarsest topology on @xmath27 such that all the evaluation maps @xmath33 , @xmath30 , remain continuous .
( here @xmath34 is meant to carry the usual hausdorff topology ) .
the idea behind the quite simple proof of this classical result is to reduce the question to the finite dimensional case , where it is trivial .
the schrder - simpson theorem is a kind of asymmetric version of this classical fact in a special situation .
its original proof was technical and involved .
a new proof in a conceptual framework was published in @xcite . a short proof by j. goubault - larrecq @xcite has appeared just recently .
let us remark that the schrder - simpson theorem has a predecessor in domain theory : if @xmath1 is a continuous dcpo ( with the scott topology ) , then @xmath19 is a continuous dcpo , too , and every scott - continuous linear functional on @xmath19 is the evaluation at some @xmath23 ( see c. jones @xcite and kirch @xcite ) . for a continuous dcpo @xmath1 , the scott topology on @xmath19 agrees with the weak@xmath0upper topology as kirch @xcite has shown , a fact that is now longer true for more general spaces .
thus , the schrder - simpson theorem is a far reaching generalization .
the proofs of the schrder - simpson theorem due to keimel and goubault - larrecq are based on two lemmas that hold for every lower semicontinuous linear functional @xmath29 from the valuation powerdomain @xmath35 with the weak@xmath0upper topology into @xmath2 , where @xmath1 is any topological space : ( lemma 2 in @xcite , theorem 5.3 in @xcite ) [ lem:2 ] there is a family @xmath17 of functions in @xmath8 such that @xmath36 for every @xmath25 .
( lemma 1 in @xcite , lemma 5.5 in @xcite ) [ lem:1 ] the set of all @xmath23 such that @xmath37 for all @xmath25 is directed .
the schrder - simpson theorem [ ss ] can now be deduced as follows : given a lower semicontinuous linear @xmath38 , lemma [ lem:2 ] gives us a family of @xmath17 of functions in @xmath8 such that @xmath36 for all @xmath25 . by lemma [ lem:1 ]
the family @xmath17 can be chosen to be directed and we let @xmath39 . since the valuations @xmath40 are scott - continuous on @xmath8 , they preserve directed suprema and we obtain that @xmath41 for all @xmath40 .
in fact , it suffices to find a subcone @xmath42 of @xmath8 and to prove the two lemmas above for subfamilies of @xmath42 , for example for the subcone of step functions .
it is desirable to prove an analogue of the schrder - simpson theorem for situations different from the valuation powerdomain . for the moment
i do not know how to generalize lemma [ lem:1 ] . for lemma [ lem:2 ] ,
a general conceptual argument had been developed in ( * ? ? ?
* section 5.1 ) by reducing the problem to a situation close to the classical vector space situation ; but this argument needed some quite heavy background from the theory of quasi - uniform locally convex cones .
goubault - larrecq found a simple direct argument for the proof of lemma [ lem:2 ] for the special situation of the valuation powerdomain . in this note
we present a direct approach to lemma [ lem:2 ] based on goubault - larrecq s idea but in an appropriate generality .
it uses the classical idea to reduce the problem to the finite dimensional case .
let us recall the motivation behind all of this .
the schrder - simpson theorem concerns the probabilistic powerdomain over arbitrary topological spaces used in semantics for modelling probabilistic phenomena in programming .
goubault - larrecq @xcite recently has applied the result in his work on modelling probabilistic and ordinary nondeterminism simultaneously .
the schrder - simpson theorem can not be proved with methods from classical functional analysis , since the real - valued lower semicontinuous functions on a topological space @xmath1 do not form a vector space . for a lower semicontinuous function @xmath43 ,
its negative @xmath44 is not necessarily lower semicontinuous .
similarly , we are in an asymmetric situation for the valuations , those being scott - continuous linear functionals .
we therefore work with cones , a weakening of the notion of a vector space by restricting scalar multiplication to nonnegative reals and not requiring the existence of additive inverses .
this offers the advantage of including infinity . in this section
we provide some basic concepts .
the connection to our goal will appear only later .
we denote by @xmath3 the set of nonnegative real numbers with the usual linear order , addition and multiplication . the letters @xmath45 will always denote nonnegative reals . further @xmath46 denotes the nonnegative reals extended by @xmath4 as a top element .
we also extend addition and multiplication from @xmath3 to @xmath2 by defining @xmath47 for all @xmath48 and @xmath49 [ cone]a _ cone _ is a commutative monoid @xmath42 carrying a scalar multiplication by nonnegative real numbers satisfying the same axioms as for vector spaces ; that is , @xmath42 is endowed with an addition @xmath50 and a neutral element @xmath51 satisfying : @xmath52 and with a scalar multiplication @xmath53 satisfying:@xmath54 for all @xmath55 and all @xmath56 . an _ ordered cone _ is a cone @xmath42 endowed with a partial order @xmath57 , such that addition and multiplication by fixed scalars @xmath58 are order preserving , that is , for all @xmath59 and all @xmath60 : @xmath61 unlike for vector spaces , addition in cones need not satisfy the cancellation property , and cones need not be embeddable in vector spaces . for example
@xmath62 is an ordered cone that is not embeddable in a vector space .
the same holds for finite powers @xmath63 and for infinite powers @xmath64 with the pointwise defined order @xmath65 iff @xmath66 for all indices @xmath67 .
maps @xmath29 from a cone @xmath42 into @xmath2 are simply referred to as _
functionals_. thus , functionals are allowed to have the value @xmath4 . on any set of functionals we consider the pointwise order @xmath68 if @xmath69 for all @xmath70 .
let @xmath42 be a cone .
a functional for homogeneous functionals ; note that we omit this requirement .
] @xmath71 is called + [ cols="<,>,^ , < , < " , ] + we say that @xmath29 is _ sublinear _ ( _ superlinear _ , _ linear _ , respectively ) , if it is homogeneous and subadditive ( superadditive , additive , respectively ) . as in real vector spaces , we have a notion of convexity : [ def : convex]a subset @xmath72 of a cone @xmath42 is _ convex _ if , for all @xmath73 , the convex combinations @xmath74 belong to @xmath72 for all real number @xmath75 with @xmath76 .
convex sets in cones may look different than those in vector spaces because of the presence of infinite elements .
indeed , in @xmath2 not only intervals @xmath77 are convex but also the sets @xmath78 . to every functional @xmath79
we assign the sets @xmath80 which are complements one of the other .
these sets reflect properties of the functionals : [ rem : mink ] for homogeneous functionals @xmath29 and @xmath81 on a cone @xmath42 we have : \(a ) @xmath82 if and only if @xmath83 if and only if @xmath84 . indeed , @xmath85 iff , for all @xmath86 , @xmath87 . for @xmath86 , @xmath87 iff @xmath88 iff @xmath89 iff @xmath90 iff @xmath91 iff @xmath92 .
\(b ) if @xmath29 is sublinear then @xmath93 is convex . indeed , if @xmath29 is sublinear and @xmath94 are elements in @xmath93 , then @xmath95 , whence @xmath96
\(c ) if @xmath81 is superlinear then @xmath97 is convex .
the following technical result is the key to our main result : [ lem : sep ] let @xmath98 and @xmath72 be a convex subset in @xmath99 disjoint from @xmath26
. then there are nonnegative real numbers @xmath100 such that @xmath101 and @xmath102 let @xmath26 and @xmath72 be as in the statement of the lemma .
the convex hull of @xmath103^n$ ] does not meet @xmath26 . indeed , if @xmath104 , then @xmath105 for some coordinate @xmath106 of @xmath107 ; as @xmath108 for every coordinate of @xmath109^n$ ] , the @xmath67-th coordinate of every convex combination of @xmath107 and @xmath110 is @xmath111 .
as @xmath26 is an upper set , the lower set @xmath112 of all elements @xmath113 that are below some element in the convex hull of @xmath114^n$ ] does not meet @xmath26 , and @xmath112 is convex , too .
we now consider the convex subset @xmath115 and the convex open subset @xmath116 of @xmath117 which are disjoint . by a standard separation theorem ( see the remark below ) , there is a linear functional @xmath118 such that @xmath119 for all @xmath120 and all @xmath121 . as for every linear functional on @xmath117 , there are real numbers @xmath100 such that @xmath122 for every @xmath123 .
let us show that @xmath124 for every @xmath67 and @xmath125 . since @xmath126
, we have @xmath127 . since @xmath128 for every @xmath129
, we have @xmath130 for all @xmath131 , whence @xmath132 .
we finally show that @xmath133 .
since the vector @xmath134 belongs to @xmath135 for every @xmath129 , we have @xmath136 , that is , @xmath137 for every @xmath129 , which implies @xmath133 .
similarly , @xmath124 for all @xmath67 .
the restriction of @xmath5 to the positive cone @xmath138 has a unique scott - continuous extension to @xmath99 also given by @xmath139 for all @xmath140 which has the desired separation property for @xmath72 and @xmath26 . in the proof of the preceding lemma [ lem : sep ]
we have used a standard separation theorem which can be proved for arbitrary topological vector spaces @xmath141 over the reals ( see , e.g. , ( * ? ? ? * theorem 3.4(a ) ) ) : _ for every open convex subset @xmath135 and every convex subset @xmath142 disjoint from @xmath135 there are a continuous linear functional @xmath143 and a real number @xmath75 such that @xmath144 for every @xmath120 and every @xmath145 .
since @xmath146 in the situation in the proof above , one may choose @xmath147 by replacing @xmath5 by @xmath148 .
_ the proof of this standard separation theorem uses the axiom of choice .
we only need this separation theorem for finite dimensional real vector spaces . for this special case ,
there exist constructive proofs .
one of the referees for this paper has worked out such a constructive proof .
one may also use a separation theorem for convex subsets in finite dimensional vector spaces ( see aliprantis and border [ 1 , theorem 7.35 , p. 279 ] ) often attributed to minkowski . in @xcite ,
k. c. border discusses the same result under the name of finite dimensional separating hyperplane theorem 11 , and border s proof looks constructive , as far as i can see .
_ thus , our lemma [ lem : sep ] has a constructive proof , not using the axiom of choice or any weak form thereof , and the same holds for the following proposition .
_ i am indebted to one of the referees for a detailed discussion on the issue of a constructive proof .
moreover , the proof of lemma [ lem : sep ] has been simplified by a suggestion of this referee , namely to replace @xmath72 by the convex hull of @xmath72 with the cube @xmath149^n$ ] .
the following key result uses the reduction to the finite dimensional case as indicated in the introduction : [ key ] consider a cone @xmath150 together with a sublinear functional @xmath151 and finitely many linear functionals @xmath152 , @xmath153 such that @xmath154
. then there are nonnegative real numbers @xmath100 such that @xmath125 and @xmath155 .
let @xmath156 be the set of all @xmath157 such that @xmath158 . since @xmath29 is supposed to be sublinear
, @xmath159 is convex .
further let @xmath160 be the set of all @xmath157 such that @xmath161 .
consider the map @xmath162 defined by @xmath163 .
this map @xmath164 is linear , since all the @xmath165 are linear .
for @xmath166 we have @xmath167 for every @xmath67 , that is @xmath168 .
since @xmath164 is linear , the image @xmath169 of @xmath159 is convex . for @xmath170
, we have @xmath171 , whence @xmath172 .
thus , @xmath169 is a convex subset of @xmath99 disjoint from @xmath26 . by the separation lemma [ lem : sep ]
there are nonnegative real numbers @xmath100 such that @xmath173 and @xmath174 for all @xmath175 and all @xmath176 .
we denote by @xmath177 the linear map @xmath178 .
then , on one hand , @xmath179 is a linear functional on @xmath150 .
if @xmath180 , then @xmath166 , hence @xmath181 which implies @xmath182 , whence @xmath183 by remark [ rem : mink](a ) . on the other hand , if @xmath184 , then @xmath170 , whence @xmath185 ; thus @xmath186 again by remark [ rem : mink](a )
noticing that , @xmath187 , that is , @xmath188 , we have the desired result . in the above lemma ,
the maps @xmath165 are supposed to be linear ; so the linear combination @xmath189 is linear , too . in particular
, we have interpolated a linear functional between @xmath29 and @xmath190 . under the hypotheses of the preceding lemma
, there are general separation theorems that yield directly the existence of a linear functional @xmath81 between the sublinear functional @xmath29 and the superlinear functional @xmath190 .
our lemma above gives us more information as it tells us that @xmath81 may be chosen to be a convex combination of the @xmath165 and this property will be crucial in the following .
one may prove the preceding lemma also under the weaker hypothesis that the maps @xmath165 are only superlinear and not linear .
but we have no use for this generalization .
every partially ordered set @xmath191 can be endowed with a simple topology , the _ upper topology _
@xmath192 : a subbasis for the closed sets is given by the principal ideals @xmath193 , @xmath194 .
if not specified otherwise , we will use the upper topology on @xmath3 and on @xmath2 .
the proper nonempty open subsets are simply the open upper intervals @xmath195r,+\infty[=\{x\mid x > r\}$ ] on @xmath3 and the open upper intervals @xmath195r,+\infty]$ ] on @xmath2 , where @xmath6 .
this asymmetric upper topology on @xmath2 is appropriate to talk about lower semicontinuity . in agreement with classical analysis , for an arbitrary topological space @xmath1 , we call _ lower semicontinuous _ those functions @xmath196 which are continuous with respect to the upper topology on @xmath2
. extended addition and multiplication are lower semicontinuous as maps @xmath197 also at infinity . for topological spaces @xmath198 ,
a function @xmath199 will be said to be _ separately continuous _ if , for each fixed @xmath200 , the function @xmath201 and , for each fixed @xmath202 , the function @xmath203 is continuous .
we say that @xmath5 is _ jointly continuous _ if it is continuous for the product topology on @xmath204 .
a _ semitopological cone _ is cone equipped with a topology satisfying the t@xmath205 separation axiom such that addition @xmath50 and scalar multiplication @xmath206 are separately continuous .
if these maps are jointly continuous , we have a _ topological cone_. [ ex : rn ] for a nonnegative integer @xmath207 , clearly @xmath99 is a cone ordered by the coordinatewise order . the upper topology on @xmath99 is the product of the upper topology on @xmath2 .
addition and extended scalar multiplication are continuous for the upper topology on @xmath99 so that @xmath99 is a topological cone .
the same holds for infinite powers @xmath208 .
any topological t@xmath205 space @xmath1 comes with an intrinsic partial order , the _ specialization preorder _ defined by @xmath65 if the element @xmath107 is contained in the closure of the singleton @xmath209 or , equivalently , if every open set containing @xmath107 also contains @xmath110 . in this paper
, an order on a topological space will always be the specialization order .
open sets are upper (= saturated ) sets and closed sets are lower sets with respect to the specialization preorder .
the saturation @xmath210 of a subset @xmath159 can also be characterized to be the intersection of the open sets containing @xmath159 . in hausdorff spaces ,
the specialization order is trivial . on @xmath2 ,
the specialization order for the upper topology agrees with the usual linear order .
continuous maps between topological t@xmath205 spaces preserve the respective specialization preorders . for more details , see e.g. @xcite , section 0 - 5 .
as we use the upper topology on @xmath3 , a semitopological cone @xmath42 can not satisfy the hausdorff separation property : since continuous maps preserve the respective specialization preorders , the continuity of scalar multiplication implies firstly that , for every @xmath211 , the map @xmath212 is order preserving , that is , the rays in the cone are linearly ordered , and secondly that the cone is pointed ; indeed , @xmath213 for all elements @xmath214 of the cone .
in particular , the specialization order on @xmath42 is nontrivial and the topology is far from being hausdorff except for the trivial case @xmath215 . as continuous maps
preserve the specialization order , we conclude that _ every semitopological cone is an ordered cone with bottom element @xmath51_. let @xmath150 be a semitopological cone
. there is a close relation between classes of open subsets of @xmath150 and classes of lower semicontinuous functionals on @xmath150 .
an arbitrary functional @xmath216 is lower semicontinuous if and only if , for every @xmath13 , the set @xmath217 is open . for a homogeneous
functional this simplifies : a homogeneous functional is lower semicontinuous if and only if the set @xmath218 is open . since @xmath219 , the
open set @xmath220 does not contain @xmath51 .
conversely , for any open subset @xmath221 containing @xmath51 , the _ upper minkowski functional _
@xmath222 is lower semicontinuous and homogeneous ; it is understood that @xmath223 , if @xmath224 for all @xmath225 .
for any nonempty finite family @xmath226 of homogeneous lower semicontinuous functions the pointwise infimum @xmath227 is again homogeneous and lower semicontinuous .
the same holds for the pointwise supremum @xmath228 of an arbitrary family @xmath229 of homogeneous lower semicontinuous functionals .
we have the following properties sharpening remark [ rem : mink ] ( see ( * ? ? ?
* propositions 7.4 and 7.5 ) and @xcite ) : [ lem : hom ] 1 .
the assignment @xmath230 establishes an order isomorphism between the collection @xmath231 of all lower semicontinuous homogeneous functionals @xmath232 ordered pointwise and the collection @xmath233 of all proper open subsets @xmath160 of @xmath150 ordered by inclusion ; the inverse map is given by assigning to every @xmath234 its minkowski functional @xmath235 .
for homogeneous lower semicontinuous functionals we have:@xmath236 2 .
a homogeneous functional @xmath29 is superlinear if and only if the corresponding open set @xmath220 is convex .
3 . a homogeneous functional @xmath29 is sublinear if and only if the complement @xmath237 of the corresponding open set is convex .
we fix a _ dual pair of cones _ which , by definition , consists of two cones @xmath42 and @xmath150 together with a bilinear map @xmath238 .
bilinearity means that for each @xmath70 , the map @xmath239 is linear and similarly for fixed @xmath240 .
we will always suppose this bilinear map to be non - singular , that is , for any two different elements @xmath241 there is an @xmath70 such that @xmath242 .
as before , we endow @xmath2 with the _ upper _ topology .
we consider the coarsest topology @xmath243 on @xmath150 rendering lower semicontinuous the maps @xmath244 for all @xmath70 .
we call it the _ weak upper topology _ on @xmath150
. a subbasis of this topology is given by the sets @xmath245 addition and scalar multiplication are jointly continuous for the weak upper topology @xmath243 so that @xmath150 with the weak upper topology is a topological cone .
alternatively , @xmath150 may be mapped into the product space @xmath246 via the map @xmath247 .
the weak upper topology @xmath243 is the topology induced by the upper topology on the product space @xmath246 . as we suppose the bilinear form to be non - singular
, the weak upper topology @xmath243 satisfies the t@xmath205 separation axiom .
the specialization order is given by @xmath248 if @xmath249 for all @xmath70 , that is , if @xmath250 for all @xmath70 .
we may define a weak upper topology @xmath251 for any subset @xmath252 by taking the coarsest topology rendering continuous the maps @xmath253 for all @xmath254 .
this weak upper topology agrees with the weak upper topology @xmath255 , where @xmath256 denotes the subcone of @xmath42 generated by @xmath252 .
this follows from the fact that finite linear combinations of lower semicontinuous functionals are also lower semicontinuous . as the weak upper topology is generated by the subbasic open sets @xmath257 , @xmath70 , the finite intersections @xmath258 where @xmath259 is a finite subset of @xmath42 form a basis of the weak upper topology .
since these basic open sets are convex , this topology is locally convex in the sense that every point has a neighborhood basis of open convex sets .
every @xmath243-open set is the union of a family of basic open sets @xmath260 . translating this statement to the homogeneous functionals according to lemma [ lem : hom ]
we obtain : if @xmath150 carries the weak upper topology @xmath243 , the homogeneous lower semicontinuous functionals @xmath261 are those obtained from the functionals @xmath253 , @xmath70 , as pointwise sups of finite pointwise infs .
we now come to our general form of lemma [ lem:2 ] : [ main ] let @xmath262 be a pair of cones together with a non - singular bilinear map @xmath263 .
we equip @xmath150 with the weak upper topology @xmath243 .
then every lower semicontinuous sublinear functional @xmath264 is the pointwise supremum of a family of point evaluations @xmath253 ; more explicitly : for every @xmath240 , @xmath265 by the preceding lemma every homogeneous lower semicontinuous functional @xmath81 on @xmath150 is the pointwise sup of finite pointwise meets @xmath266 where the @xmath106 range over finite families of elements in @xmath42 . if @xmath81 is sublinear , proposition [ key ] tells us that there is a convex combination @xmath267 , such that @xmath268 . since @xmath269
, we have found an element @xmath270 in @xmath42 with @xmath271 .
thus @xmath81 is the pointwise sup of functionals @xmath253 .
concerning the previous proof , note that the functional @xmath272 is superlinear and lower semicontinuous .
thus the sandwich theorem @xcite allows to find a continuous linear functional @xmath29 with @xmath273 .
but that is not what we want .
we want to show that we can choose @xmath274 for some @xmath275 . for this
we use proposition [ key ] which was proved through a reduction of the problem to the finite dimensional case . in the following
we rephrase our theorem for some special situations .
one specializes the cones @xmath42 and @xmath150 above by a dual pair of ordered cones @xmath42 and @xmath150 with an order preserving nonsingular bilinear map @xmath276 .
note that the given order on @xmath150 may be stronger than the specialization preorder induced by the weak upper topology @xmath243 .
both orders agree if and only if the bilinear form is _ order non - singular _ in the sense that for @xmath277 in @xmath42 there is a @xmath240 such that @xmath278 , and similarly for the other argument . under these hypotheses ,
theorem [ main ] specializes to : for every sublinear functional @xmath279 which is lower semicontinuous with respect to the weak upper topology @xmath243 we have @xmath280 for every @xmath240 .
we now consider a pair @xmath262 of semitopological cones together with a separately continuous non - singular bilinear form @xmath276 .
the weak upper topology @xmath243 is coarser than the given topology on @xmath150 . applying theorem [ main ] under these hypotheses we obtain : for every sublinear functional @xmath279 which is lower semicontinuous with respect to the weak upper topology @xmath243 we have @xmath281 for every @xmath240 .
since the given topology on @xmath150 may be strictly coarser than the weak upper topology @xmath243 , the claim of the corollary need not be true for the sublinear functionals @xmath279 which are lower semicontinuous with respect to the original topology on @xmath150
. recall that a poset @xmath150 is _ directed complete _
( a _ dcpo _ , for short ) , if every directed family @xmath282 of elements in @xmath150 has a least upper bound , denoted by @xmath283{$\scriptstyle\uparrow$}}}}}_i x_i$ ] .
a map @xmath284 between dcpos is called scott - continuous if it preserves least upper bounds of directed sets , that is , if it is order preserving and if @xmath285{$\scriptstyle\uparrow$}}}}}_i x_i)={\mathop{\bigvee{}^{^{\,\makebox[0pt]{$\scriptstyle\uparrow$}}}}}_if(x_i)$ ] for every directed family of elements in @xmath42 .
these are precisely the functions that are continuous for the respective scott topologies , where a subset @xmath159 of a dcpo is called scott - closed if @xmath159 is a lower set and if @xmath283{$\scriptstyle\uparrow$}}}}}_i x_i\in a$ ] for every directed family of @xmath286 .
we refer to @xcite for background on dcpos . a _
d - cone _ is a cone equipped with directed complete partial order in such a way that scalar multiplication @xmath287 and addition @xmath50 are scott - continuous .
with respect to the scott topology a d - cone is a semitopological cone . @xmath2 with its usual order is a d - cone ; its scott topology agrees with the upper topology .
thus the scott - continuous functionals on a d - cone are the lower semicontinuous ones .
we can specialize our results to dual pairs of d - cones @xmath262 with a scott - continuous non - singular bilinear form @xmath288 .
but here we want to consider also subcones @xmath289 which are _ d - dense _ in @xmath42 , that is , the only d - subcone of @xmath42 that contains @xmath290 is @xmath42 itself .
we have : [ subbase ] the weak upper topology @xmath291 agrees with the weak upper topology @xmath243 . in order to verify our claim , we first notice : if @xmath282 is a directed family in @xmath42 and @xmath292{$\scriptstyle\uparrow$}}}}}_ix_i$ ] , then @xmath293 . indeed , @xmath294 if and only if @xmath295{$\scriptstyle\uparrow$}}}}}_i x_i , y\rangle = { \mathop{\bigvee{}^{^{\,\makebox[0pt]{$\scriptstyle\uparrow$}}}}}_i\langle x_i , y\rangle$ ] ( by the scott - continuity of the bilinear form ) if and only if @xmath296 for some @xmath67 if and only if @xmath297 for some @xmath67 .
this shows that adding suprema of directed sets to the set @xmath290 does not refine the weak upper topology . by (
* corollary 6.7 ) this suffices to justify our claim . under the hypotheses of this subsection ,
theorem [ main ] now specializes to the following : let @xmath262 be a dual pair of d - cones and @xmath290 a d - dense subcone of @xmath42 .
we endow @xmath150 with the weak upper topology @xmath291 and consider a scott - continuous sublinear functional @xmath279 .
then @xmath298 for every @xmath240 .
the standard situation for applying the previous setting is to start with a cone @xmath42 , to form the _ algebraic dual _
@xmath299 of all linear functionals @xmath300 which is again a cone under pointwise addition and scalar multiplication and to take the bilinear map @xmath301 for @xmath302 which is non - singular for obvious reasons . for ordered cones @xmath42 it is natural to restrict to the _ order dual _ of all order preserving linear functionals .
if we start with a semitopological cone @xmath42 , we consider the _ topological dual _
@xmath303 of all lower semicontinuous linear functionals on @xmath42 which is a subcone of the algebraic dual @xmath304 .
we consider the weak upper topology @xmath305 on @xmath303 which is also called the _
weak@xmath0upper topology _ in agreement with the terminology in functional analysis .
it is the coarsest topology for which the evaluation maps @xmath306 are lower semicontinuous for all @xmath70 .
the bilinear map @xmath307 is separately lower semicontinuous .
we ask : [ qu : bij ] characterize those semitopological cones @xmath42 with the property that , for every linear functional @xmath308 that is lower semicontinuous for the weak@xmath0upper topology , there is a unique element @xmath70 such that @xmath309 for every @xmath310 , that is , @xmath311 for a uniquely determined @xmath70
. since the evaluation maps are linear , @xmath312 yields a map from @xmath42 to the double dual @xmath313 which is linear .
the question formulated above is equivalent to the question whether the map @xmath314 is an isomorphism of cones .
a simple example for cones having this property are the finite dimensional cones @xmath99 endowed with the upper topology ( which agrees with the product topology with respect to the upper topology on @xmath2 ) .
this follows from the fact that the dual cone of all lower semicontinuous linear functionals on @xmath99 is isomorphic to @xmath99 and its weak@xmath0upper topology agrees with the upper topology . for the proof
we just need the following : [ ex : rn ] for every lower semicontinuous linear functional @xmath315 there are @xmath316 such that @xmath317 .
given @xmath318 , let @xmath319 where @xmath320 is the standard basis vector all entries of which are zero , except the i - th entry which is @xmath321 .
by linearity , @xmath322 for all @xmath323 . by lower semicontinuity
this formula extends to all of @xmath99 .
the schrder - simpson theorem yields a whole class of examples for which the answer to question [ qu : bij ] is affirmative . indeed , for any topological space @xmath1 , the cone @xmath324 of lower semicontinuous functions @xmath325 is a d - cone and its dual is the valuation powerdomain @xmath326 .
of course , question [ qu : bij ] can be split .
the injectivity of the map @xmath312 is equivalent to the question whether the lower semicontinuous linear functionals on @xmath42 separate the points of @xmath42 .
indeed , for elements @xmath327 in @xmath42 , there is a @xmath328 such that @xmath329 if , and only if , @xmath330 , that is , @xmath331 .
this criterion for injectivity is equivalent to the property that , whenever @xmath332 , there is a convex open set @xmath333 containing @xmath107 but not @xmath110 .
this is a consequence of the separation theorem ( @xcite ) .
this property is in particular guaranteed by local convexity : a semitopological cone is _ locally convex _ if every point has a neighborhood base of open convex sets .
thus , injectivity is not the problem ; surjectivity is the issue .
we are mainly interested in the case of a d - cone @xmath42 with its scott topology .
we then form the dual cone @xmath334 of all scott - continuous linear maps @xmath300 which is a d - cone , too , for the pointwise defined order .
the bilinear map @xmath301 is scott - continuous .
again , the topology @xmath305 on @xmath303 is called the weak@xmath0upper topology . by lemma [ subbase ] above
, we can replace @xmath42 by any d - dense subcone @xmath290 since then @xmath335 .
theorem [ main ] now specializes to the following general version of lemma [ lem:2 ] in the introduction : let @xmath42 be a d - cone and @xmath303 its dual cone of scott - continuous linear functionals @xmath300 endowed with its weak@xmath0-upper topology .
let @xmath290 be a d - dense subcone of @xmath42 in the sense that there is no proper d - subcone of @xmath42 containing @xmath290 .
then for every lower semicontinuous sublinear functional @xmath336 we have @xmath337 for every @xmath328 , that is , @xmath338 .
what we are lacking is a general version of lemma [ lem:1 ] .
so we are left with the question : characterize the d - cones @xmath42 for which the conclusion of the previous corollary can be strengthened as follows : * for every lower semicontinuous linear functional @xmath339 there is a directed family @xmath106 of elements in @xmath42 such that @xmath340{$\scriptstyle\uparrow$}}}}}\!\!\!_i\ \mu(x_i)$ ] for all @xmath328 .
if this is the case , then we may form @xmath292{$\scriptstyle\uparrow$}}}}}\!\!\!_i\ x_i$ ] and we obtain @xmath340{$\scriptstyle\uparrow$}}}}}\!\!\!_i\ \mu(x_i)=\mu({\mathop{\bigvee{}^{^{\,\makebox[0pt]{$\scriptstyle\uparrow$}}}}}\!\!\!_i\ x_i)=\mu(x)$ ] , since all the @xmath328 are scott - continuous .
the following example may be instructive . even in the finite dimensional case not every linear functional on a cone with values in @xmath3 may be lower semicontinuous for a weak upper topology .
take the discrete two elements space @xmath341 and the sierpinski space @xmath342 , the two element set @xmath343 with @xmath344 as the only proper nonempty open subset , let @xmath345 and @xmath346 .
the standard inner product is a bilinear map @xmath347 .
one might think that this situation gives a counterexample to theorem [ main ] .
indeed , the second projection @xmath348 is a linear functional , but there is no @xmath349 such that @xmath350 .
( the only point @xmath107 satisfying this equation is @xmath351 which does not belong to @xmath42 . ) also , it is not possible to obtain @xmath352 as the pointwise sup of functionals of the type @xmath353 with @xmath70 .
but there is not contradiction to theorem [ main ] .
the point is , that @xmath352 is not lower semicontinuous with respect to the weak upper topology @xmath243 .
indeed , the specialization order @xmath354 associated with the topology @xmath243 has to be preserved by functionals that are lower semicontinuous with respect to the weak upper topology @xmath243 . in this case
the specialization order on @xmath150 is given by @xmath355 the projection @xmath352 does not preserve @xmath354 , since @xmath356 but @xmath357 .
_ let @xmath42 and @xmath150 be subcones of @xmath117 which are closed and have interior points for the usual euclidean topology .
suppose in addition that , for the standard inner product , @xmath358 for all @xmath359 .
we claim : @xmath42 is the dual @xmath360 of @xmath150 in the sense that for every linear functional @xmath361 , which is lower semicontinuous for the weak upper topology @xmath243 , there is a unique @xmath70 such that @xmath274 for some @xmath70 . _ for the proof
consider such a linear functional @xmath29 .
it may be extended in a unique way to a linear functional on @xmath117 .
thus there is a unique element @xmath362 such that @xmath363 . by proposition [ main ] ,
@xmath364 for all @xmath240 , where @xmath1 is the set of all @xmath365 such that @xmath366 , that is @xmath367 , that is @xmath368 for all @xmath240 .
this amounts to say that @xmath369 , that is @xmath370 for the order @xmath371 defined by the cone @xmath360 .
this shows that the set @xmath1 is compact .
clearly , @xmath1 is convex . if @xmath372
, then @xmath373 is compact and convex but does not contain @xmath51 .
take any @xmath374 in the interior of @xmath150 .
then @xmath375 is a strictly positive functional on @xmath360 , that is , @xmath376 for all @xmath377 in @xmath360 .
hence , @xmath378 for all @xmath379 .
since @xmath1 is compact , it follows that @xmath380 , a contradiction . +
similar results as above hold for d - subcones of @xmath99 . but note that @xmath42 is not the dual of @xmath150 , in general , if we use the usual euclidean topology on @xmath150 .
i. battenfeld , m. schrder and a. simpson , a convenient category of domains . in : computation ,
meaning and logic , articles dedicated to gordon plotkin l. cardelli , m. fiore and g. winskel ( eds ) , electronic notes in computer science 172 ( 2007 ) , pages 6999 . k. c. border , separating hyperplane theorems .
notes , california institute of technology .
+ www.hss.caltech.edu/~kcb/notes/separatinghyperplane.pdf .
g. gierz , k. h. hofmann , k. keimel , j. d. lawson , m. mislove , and d. s. scott , continuous lattices and domains . _ encyclopedia of mathematics and its applications , vol .
93 _ , cambridge university press , 2003 , xxxvi+591 pages .
m. schrder and a. simpson , probabilistic observations and valuations .
talk at new interactions between analysis , topology and computation , birmingham ( 2009 ) . | in functional analysis it is well known that every linear functional defined on the dual of a locally convex vector space which is continuous for the weak@xmath0 topology is the evaluation at a uniquely determined point of the given vector space .
m. schrder and a. simpson have obtained a similar result for lower semicontinuous linear functionals on the cone of all scott - continuous valuations on a topological space endowed with the weak@xmath0upper topology , an asymmetric version of the weak@xmath0 topology .
this result has given rise to several proofs , originally by schrder and simpson themselves and , more recently , by the author of these notes and by j. goubault - larrecq .
the proofs developed from very technical arguments to more and more conceptual ones .
the present note continues on this line , presenting a conceptual approach inspired by classical functional analysis which may prove useful in other situations . |
in standard non - relativistic quantum mechanics time enters as a parameter , external to the quantum system being investigated , and as such is not a dynamical variable associated in any way with the system s dynamics . however , there are many cases , such as experiments measuring the time of arrival of particles at a detector , the decay time of an unstable quantum system etc . , which call for a more dynamical point of view with respect to time .
one approach to this problem consists of the construction of time operators through the use of covariant positive operator valued measures ( povms ) @xcite .
since the time operators built via the use of povms are in general maximally symmetric , non - self - adjoint operators , the construction of these objects bypasses an old theorem of pauli @xcite , stating that there does not exist a self - adjoint time operator @xmath3 canonically conjugate to a hamiltonian @xmath4 whose spectrum is semibounded , such that the pair @xmath3 and @xmath4 forms together an imprimitivity system @xcite .
however , in comparison to standard self - adjoint quantum observables these operators present some difficulties , e.g. , in the relation between the algebra they generate and their spectral representations . the dynamical role of time within the framework of standard quantum mechanics
has recently been considered in @xcite through the construction of a family of self - adjoint lyapunov operators . throughout this text
we define a lyapunov operator as a self - adjoint operator whose expectation value is monotonically decreasing in time .
more precisely , let @xmath5 be a hilbert space corresponding to a given quantum mechanical system , let @xmath4 be a self - adjoint hamiltonian generating its evolution , and let @xmath6 be the state of the system at time @xmath7 . define the trajectory @xmath8 , corresponding to an initial state @xmath9 , as the set of states @xmath10 then the definition of a lyapunov operator is as follows @xcite : + + * definition 1 : * _ _ let @xmath11 be a bounded self - adjoint operator on @xmath5 .
let @xmath8 be a trajectory corresponding to an arbitrarily chosen initial state @xmath12 .
denote by @xmath13 the collection of all expectation values of @xmath11 for normalized states in @xmath8 .
then @xmath11 is a forward lyapunov operator if the mapping @xmath14 defined by @xmath15 is monotonically decreasing in time.__@xmath16 + + * remark 1 : * if in the definition above we require that @xmath17 be monotonically increasing instead of monotonically decreasing we also obtain a valid definition of a lyapunov operator .
the requirement that @xmath17 is monotonically decreasing is made purely for the sake of convenience .
it follows from this definition that a lyapunov operator allows for the temporal ordering of states in any trajectory @xmath8 according to the ordering of the expectation values in @xmath18 , thereby introducing temporal ordering into the hilbert space @xmath5 of any problem for which such an operator can be constructed .
the construction of lyapunov operators constitutes a somewhat conservative approach to the problem of time in quantum mechanics . in general , lyapunov operators indicate the direction of flow of time in a system but do not serve as time operators and are not intended for answering questions of direct time measurements .
rather , the construction of lyapunov operators , and in particular the family of lyapunov operators introduced in @xcite , was carried out with a different goal in mind .
the temporal ordering introduced into the hilbert space is a fundamental property of a quantum mechanical problem admitting lyapunov operators .
a question then arises as to the possible implications this property may have on the description of the dynamics of the system , i.e. is there a way to ` inject ' the direction of time ordering introduced by the lyapunov operator into the description of the dynamics of a quantum system in such a way that certain processes in the evolution of the system obtain simple descriptions amenable to thorough analyses ?
the study of such consequences of the existence of a lyapunov operator has been addressed to some extent in @xcite . in the present paper
we continue this line of investigation , and demonstrate the beginning of an affirmative answer to the question raised above .
specifically , we show that the existence of the family of lyapunov operators introduced in @xcite leads to a certain description of the scattering process , which we shall term _ the transition decomposition _ , that is particularly useful for treating the evolution of scattering resonances .
in this section we state some key results from @xcite , which will provide the basis for the all that follows . the following theorem is an adaptation of results stated in @xcite : + + * theorem 1 : * [ thm : forward_lypunov_op]_let @xmath5 be the hilbert space representing some quantum mechanical system , let @xmath4 be its hamiltonian , and let the spectrum of @xmath4 be @xmath19 , absolutely continuous , and uniformly degenerate . then for any choice of generalized eigenbasis @xmath20 of @xmath4 , @xmath21 being the degeneracy index , the operator @xmath22 where the summation over @xmath21 may also stand for integration , is a lyapunov operator in the sense of definition 1 , and , moreover , for any state @xmath23 @xmath24 _ @xmath16 + * remark 2 : * in this paper we shall consider only hamiltonians for which the spectrum is @xmath25 , absolutely continuous , and uniformly degenerate . the results below apply also to the more general case of hamiltonians for which the absolutely continuous spectrum is @xmath25 and is uniformly degenerate . in such cases the constructions below hold in the subspace @xmath26 .
+ + * remark 3 : * note that each complete set of generalized eigentates of @xmath4 has associated with it a different lypaunov operator , and as such theorem 1 represents a prescription for the construction of a family of self - adjoint lyapunov operators . in particular , this still holds true for two sets that differ only by an energy dependent phase .
when discussing general properties of the set of lypaunov operators defined above , we shall often refer to @xmath11 without specifying the underlying complete set of generalized eigenstates .
+ + * remark 4 : * the above family of self - adjoint lyapunov operators has recently been generalized in @xcite .
+ + * remark 5 : * throughout this paper we employ natural units , i.e. @xmath27 . + the following corollary to theorem 1 is crucial for what follows
. + + * corollary 1 : * _ let @xmath28 .
then for any state @xmath23 @xmath29 _
in this section we present a decomposition of the wave - function into two components one that is asymptotic to the state of the system in the far past and vanishes in the far future , and one that is asymptotic to the state of the system in the far future and vanishes in the far past and a corresponding decomposition for operators when working in the heisenberg picture .
the transition over time between the two asymptotic components suggests the possibility that this decomposition yields an appropriate description of transient phenomena in the evolution of a quantum system .
this expectation is realized when we apply the decomposition to scattering problems , where it will be seen to afford a highly useful description of the evolution of scattering resonances .
this task is taken up in the next section .
+ let @xmath30 , @xmath9 , @xmath31 . at any moment in time @xmath7
express @xmath32 as the sum of two components as follows : @xmath33 where @xmath34 the limits below are then an immediate consequence of theorem 1 and corollary 1:@xmath35 eqs . (
[ eq : transition representation_schrodinger ] ) , ( [ eq : backward_component_limits ] ) , and ( [ eq : forward_component_limits ] ) imply that @xmath36 can be decomposed into a sum of two components , @xmath37 and @xmath38 , such that @xmath39 vanishes in the future time asymptote and is asymptotic to @xmath36 in the past time asymptote , and @xmath38 vanishes in the past time asymptote and is asymptotic to @xmath36 in the future time asymptote .
we refer to @xmath39 as the _ backward asymptotic component _ and to @xmath38 as the _ forward asymptotic component _ of @xmath36 .
since the decomposition here is of the evolving state @xmath36 , we call the decomposition in eq .
( [ eq : transition representation_schrodinger ] ) the _ transition decomposition in the schrdinger picture . _ in this decomposition the evolution of @xmath40 is represented as a transition from its backward asymptotic component to its forward asymptotic component .
+ the transition decomposition in the schrdinger picture , eq .
( [ eq : transition representation_schrodinger ] ) , gives rise to a corresponding decomposition in the heisenberg picture .
let @xmath41 be a self - adjoint operator representing some physical observable and let @xmath42 be its heisenberg evolution . consider the expectation value of @xmath43 for some arbitrary state @xmath9 and apply the decomposition in eq .
( [ eq : transition representation_schrodinger ] ) .
we have @xmath44 define now the following decomposition of @xmath41 , which will be seen below to correspond to the three different terms on the right - hand side of eq .
( [ eq : schrodinger_to_heisenberg_trans_decomp ] ) , @xmath45 with @xmath46 indeed , if @xmath47 , @xmath48 , and @xmath49 , respectively , are the heisenberg evolutions of @xmath50 , @xmath51 , and @xmath52 , so that @xmath53 then @xmath54 @xmath55 @xmath56 using the limits of @xmath39 and @xmath38 , eqs .
( [ eq : backward_component_limits ] ) and ( [ eq : forward_component_limits ] ) , we obtain that the components of @xmath43 satisfy the following limits @xmath57 eq . ( [ eq : trans_rep_heisenberg_t ] ) provides the transition decomposition in the heisenberg picture .
the observable @xmath43 decomposes into a sum of three components , @xmath47 , @xmath48 , and @xmath49 , such that @xmath47 vanishes in the future time asymptote and is asymptotic to @xmath43 in the past time asymptote , @xmath49 vanishes in the past time asymptote and is asymptotic to @xmath43 in the future time asymptote , and @xmath48 is transient and vanishes both in the past and in the future time asymptotes .
the evolution of @xmath43 is represented as a transition from @xmath47 in the backward asymptote to @xmath49 in the forward asymptote .
accordingly , @xmath47 will be termed the _
backward asymptotic transition observable _ , @xmath49 will be termed the _ forward asymptotic transition observable _ , and @xmath48 will be termed the _
in this section we apply the transition decomposition to a quantum mechanical scattering problem . in particular , we find that this decomposition is especially useful for the description of scattering resonances .
specifically , we shall apply the transition decomposition to the evolution of approximate resonance states defined in the context of the formalism of _ the semigroup decomposition of resonance evolution _ @xcite .
this formalism , developed in recent years out of efforts to adapt the lax - phillips scattering theory @xcite to the description of the evolution of scattering resonances in quantum mechanics @xcite , is a mathematical framework serving as a basis for a time - dependent theory of resonances in quantum mechanical scattering problems @xcite .
we give here a short presentation of the simplest form of this framework that is sufficient for our purposes .
+ recall that theorem 1 defines a large class of lypaunov operators in the sense that for any complete set of generalized eigenstates of the hamiltonian there corresponds a different lyapunov operator .
it follows that each choice of lyapunov operator has associated with it its own transition decomposition .
we would like to apply the transition decomposition to scattering problems .
for such problems there are at least two distinguished energy representations for the hamiltonian , i.e. the incoming and outgoing energy representations , and we must choose which of the two to employ .
the incoming and outgoing energy representations are defined in terms of the incoming and outgoing solutions of the lipmann - schwinger equation .
let @xmath58 and @xmath59 be the incoming and outgoing solutions of the lipmann - schwinger equation satisfying @xmath60 , where @xmath21 stands for the degeneracy indices of the energy spectrum .
( the degeneracy is assumed to be uniform over the energy spectrum . )
the _ outgoing lyapunov operator @xmath61 _ is defined as@xmath62 and gives rise to an _ outgoing transition decomposition in the schrdinger or heisenberg pictures . _ defining @xmath63 , the outgoing transition decomposition in the schrdinger picture is given by @xmath64 while for an observable @xmath41 defined on @xmath5 , in the outgoing transition decomposition in the heisenberg picture , we have @xmath65 with @xmath66 1 .
let @xmath5 be a hilbert space corresponding to a given quantum mechanical scattering problem .
a self - adjoint ` free ' unperturbed hamiltonian @xmath67 and a self - adjoint perturbed hamiltonian @xmath4 are defined on @xmath5 and form a complete scattering system , i.e. , we assume that the mller wave operators @xmath68 exist and are complete .
we assume that the ( absolutely ) continuous spectrum of @xmath4 is uniformly degenerate . to simplify matters
we assume that this degeneracy is one .
the @xmath69-matrix in the the energy representation , denoted by @xmath70 , is the boundary value of a function @xmath71 analytic in some strip above the positive real axis and having an analytic continuation across the cut on the positive real energy axis into some simply - connected region @xmath72 below the real axis in which it has a single , simple resonance pole at the point @xmath73 with @xmath74 .
it is shown in @xcite that there exists a dense set @xmath75 and a well defined state @xmath76 such that for any states @xmath77 and @xmath78 the above assumptions lead to a decomposition , induced by the pole of the @xmath69-matrix at @xmath73 , of matrix elements of the evolution @xmath79 generated by @xmath4 , having the form @xmath80 the second term of the decomposition on the right - hand side of eq .
( [ eq : semigroup_decomp ] ) exhibits the typical decay behavior of a resonance .
this term is the _ semigroup term _ , or _
resonance term_. the term @xmath81 in eq .
( [ eq : semigroup_decomp ] ) is the so - called _ background _ term .
the state @xmath82 appearing in the second term in eq .
( [ eq : semigroup_decomp ] ) is called the _ approximate resonance state_. note that if the state @xmath83 on the left - hand side of eq .
( [ eq : semigroup_decomp ] ) is chosen to be orthogonal to @xmath84 then the resonance term in that equation vanishes .
the reference to @xmath84 as an approximate resonance state stems from the fact that it can be shown that there is no choice of @xmath85 and @xmath83 in the matrix element @xmath86 for which the background term @xmath81 disappears @xcite .
indeed , the schrdinger evolution of a closed system does not allow for an exponential decay law for the survival probability @xcite and deviations , such as the zeno effect for short times are inevitable @xcite .
an explicit expression for the approximate resonance state @xmath84 is given by @xcite @xmath87 this expression is obtained under the assumption that there is only a single resonance pole of the @xmath69-matrix below the positive real axis in the region @xmath72 . in the case
that @xmath72 contains multiple resonance poles of the @xmath69-matrix eq .
( [ eq : psi_approx ] ) provides only a _ zeroth order approximate resonance state _ @xcite . throughout the rest of the paper we only consider approximate resonance states given by eq .
( [ eq : psi_approx ] ) .
this restriction pertains also to the example worked out below for which the @xmath69-matrix possesses multiple resonance poles .
hence , all states calculated there are of zeroth order .
the restriction to zeroth order approximate resonance states is made for the sake of simplicity and clarity of exposition and it should be emphasized that there is no a priori difficulty in working with higher order approximate resonance states .
we shall use this decomposition to represent the evolution of a particular resonance in a simple one - dimensional scattering problem .
the model we consider is the scattering along the half - line @xmath25 off a square barrier potential .
thus , we consider a free hamiltonian @xmath90 acting on @xmath91 ( where @xmath67 is taken to be the self - adjoint extension in @xmath91 of @xmath92 from its original domain of definition @xmath93 ) and a full hamiltonian @xmath94 with @xmath95 a multiplicative operator @xmath96 such that @xmath97 where @xmath98 and @xmath99 . in this case
there are no bound state solutions of the eigenvalue problem for @xmath4 and the ( absolutely ) continuous spectrum of @xmath4 is @xmath19 . in order to find the scattering states and calculate the @xmath69-matrix one solves the eigenvalue problem for the continuous spectrum generalized eigenfunctions @xmath100 of @xmath4 @xmath101 imposing appropriate boundary conditions we find that @xmath102 where @xmath103 and @xmath104 for @xmath105 or @xmath106 for @xmath107 .
the coefficients in eq .
( [ eq : e_general_eigenfunc ] ) are given by @xcite @xmath108 where @xmath109 is to be determined by normalization conditions .
given the full set of solutions @xmath110 for the continuous energy spectrum one can find the sets @xmath111 of solutions of the lippmann - schwinger equation corresponding to incoming and outgoing asymptotic conditions .
we have @xmath112 where @xmath113 and @xmath114 are , respectively , the incoming and outgoing lippmann - schwinger solutions .
the normalization conditions for the lippmann - schwinger states in eq .
( [ eq : lippmann_schwinger_sol ] ) give us @xmath115 . in the energy representation the @xmath69-matrix
is given by @xmath116 the above expression for the @xmath69-matrix leads to the calculation of the scattering resonances of the problem . for a resonance point @xmath117 in the lower half - plane below the positive real axis we set @xmath118 with @xmath119 the resonance energy and @xmath120 the resonance width . for barrier parameters @xmath121 , @xmath122 , and @xmath123 the three lowest energy resonance poles are given by @xmath124 , @xmath125 and @xmath126 .
we shall focus on the third resonance pole @xmath127 . utilizing the outgoing lippmann - schwinger eigenfunctions @xmath114 , given by eqs .
( [ eq : e_general_eigenfunc]-[eq : lippmann_schwinger_sol ] ) , the spatial wave function of the approximate resonance state @xmath128 , and its energy density , can be calculated numerically .
the probability density @xmath129 is shown in fig . 1 ( see @xcite ) , while the energy density @xmath130 is shown in fig .
now apply the outgoing transition decomposition of eq .
( [ eq : psi_approx_trans_rep ] ) to the evolution of the approximate resonance state @xmath131 . according to eq .
( [ eq : psi_approx_trans_rep ] ) the spatial probability distribution of @xmath132 decomposes into three components@xmath133 the right - hand side of eq .
( [ eq : trans_rep_prob_density ] ) is the outgoing transition decomposition of the position probability density of @xmath134 . the first term on the right hand side of eq .
( [ eq : trans_rep_prob_density ] ) is the backward asymptotic component , the second term is the transient component , and the third term is the forward asymptotic component .
4 presents the results of the application of the outgoing transition decomposition to the evolution of the approximate resonance state @xmath135 .
each row represents a ` snapshot ' corresponding to a particular time @xmath7 .
the left graph in each row shows @xmath136 at time @xmath7 .
the middle graph shows the contribution of its backward asymptotic component @xmath137 at time @xmath7 , and the right graph in each row shows the sum of transient and forward asymptotic components @xmath138 .
it is clearly seen from this sequence of snapshots that the formation phase of the resonance , starting at the negative time asymptote @xmath139 and following through the scattering process up to @xmath140 , is captured by the backward asymptotic component ( the middle column of graphs ) , while through this whole time interval the contribution of the transient and forward asymptotic components is small . in the decay phase of the resonance , commencing at @xmath141 , the middle column of graphs essentially shows the spatial probability density profile of a resonance state , multiplied by an exponentially decaying factor @xmath142 , which gives the decay of the resonance over time .
as the resonance state decays the probability is transferred to the forward asymptotic term , a process captured in the right most column of graphs , and is eventually carried to spatial infinity as the scattering process evolves further towards the forward time asymptote @xmath143 .
observe also that the graphs on the right column in fig .
4 obtain also negative values .
this is due to the contribution of the transient term which is not necessarily positive .
further understanding of the behavior of the probability density @xmath144 may be gained by integrating it over @xmath145 . in this way
we get @xmath146 which is just the expectation value of the lyapunov operator @xmath61 in the state @xmath147 . a plot of this behavior is presented in fig .
the decay of the expectation value of @xmath61 matches the resonance decay factor @xmath142 mentioned above .
y. strauss , and j. silman and s. machnes , acknowledge support from the isf ( grants .
1169/06 and 784/06 , respectively )
. j. silman also acknowledges the support of the inter - university attraction poles programme ( belgian science policy ) under project iap - p6/10 ( photonics@be ) and of the fnrs . | * we show that the existence of the family of self - adjoint lyapunov operators introduced in [ j. math .
phys . * 51 * , 022104 ( 2010 ) ] allows for the decomposition of the state of a quantum mechanical system into two parts : a past time asymptote , which is asymptotic to the state of the system at @xmath0 and vanishes at @xmath1 , and a future time asymptote , which is asymptotic to the state of the system at @xmath1 and vanishes at @xmath2 .
we demonstrate the usefulness of this decomposition for the description of resonance phenomena by considering the resonance scattering of a particle off a square barrier potential .
we show that the past time asymptote captures the behavior of the resonance .
in particular , it exhibits the expected exponential decay law and spatial probability distribution . * |
galactic black hole binaries ( bhbs ) are ideal objects to study the physics of accretion flow onto a black hole .
bhbs take several different states with distinct spectral and timing properties ( for a review see , e.g. , * ? ? ?
generally , when the mass accretion rate ( and hence the luminosity ) is low , a bhb exhibits the low / hard state ( lhs ) , where the spectrum is dominated a power law with a photon index of @xmath0 and a high energy cut - off at @xmath1 kev . at a higher mass accretion rate ,
the spectrum is dominated by optically thick , thermal emission from the accretion disk , which is well represented by the multi - color disk ( mcd ) model @xcite .
this state is called the high / soft state ( hss ) .
when the mass accretion rate becomes even larger and the luminosity is close to the eddington limit , a bhb takes the so - called very high state ( vhs ) , where the spectra are characterized by strong comptonization .
although the vhs is a key state to understand the accretion disk physics under very high mass accretion rates , it has been rather poorly investigated because it is quite rare that a bhb takes this state .
the evolution of the state of a bhb during an outburst is complex , and not solely determined by mass accretion rate . in typical outbursts , a bhb shows hysteresis with a q - shaped curve on a hardness - intensity diagram ( e.g. , * ? ? ?
sources start from the lhs , then make a transition into the hss , sometimes via the vhs .
the source remains in the hss while the luminosity declines , and then turns back to the lhs , at a luminosity which can be a factor of 2 - 20 times smaller than the initial lhs - hss transition .
this hysteresis behavior probably originates from the accretion rate changing faster than the viscous timescale in the geometrically thin disk @xcite .
however , bhbs in outbursts do not always exhibit similar evolutional tracks on this diagram ( e.g. , @xcite ) , implying that the physical processes governing the spectral evolution in bhbs may be even more complicated .
an important issue in understanding black - hole accretion flow in each state is whether the standard disk extends down to the innermost stable circular orbit ( isco ) around the black hole or is truncated at a larger radius . in the mcd model , the disk bolometric luminosity @xmath2
is related to the disk temperature @xmath3 and innermost disk radius @xmath4 as @xmath5 , where @xmath6 is stefan - boltzmann constant .
the standard disk is believed to extend to the isco in the hss , since the observed @xmath4 value of a bhb is found to be constant over a wide range of @xmath2 ( e.g. , * ? ? ?
by contrast , many studies suggest that the accretion disk in the lhs is truncated before reaching the isco ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and the inner part is probably replaced by radiative inefficient accretion flow ( riaf ) ( e.g. , * ? ? ? * ) .
the geometry of the innermost part of the accretion flow in the vhs is less clear , however .
this is because in this state a hot corona might be strongly compton - scattering photons emitted from the standard disk @xcite , making it difficult to directly measure the intrinsic disk parameters .
there are some reports that the innermost disk radius in the vhs is somewhat larger than that in the hss @xcite , indicating that the disk may be truncated before the isco .
it is crucial to establish this picture by using high - quality broad band x - ray spectra of bhbs in the vhs .
4u 163047 is identified as a galactic bhb because of the similarity of the spectral and temporal behaviors to other bhbs , although the mass of the compact object has not been measured yet .
this object exhibits quasi - periodic outbursts with 600700 day cycles .
this period is very rapid as a bhb , probably indicating a high time - averaged mass accretion rate .
the x - ray spectra of 4u 163047 are heavily absorbed ( @xmath7 @xmath8 ; * ? ? ?
* ) , suggesting that the object is located near the galactic center , although the accurate source distance @xmath9 is unknown .
the nature of the companion has not been firmly identified yet ; @xcite infer that it may be an early type star by assuming that the whole @xmath10-band magnitude in outburst comes from the companion .
outbursts of 4u 163047 are observed with many missions since 1980s , including
@xcite , _ einstein / ginga / asca _ @xcite , _ beppo - sax _
@xcite , _ rxte _
( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , _ integral _
@xcite , _ suzaku _
@xcite , _ xmm - newton _
@xcite , and _ nustar
_ @xcite . by fitting the rxte spectra taken between 2002 and 2005 with the mcd plus power - law model , @xcite
suggest that the accretion disk exhibited a state transition between the standard disk and the slim disk at high mass accretion rates .
similar results are also reported by @xcite using the rxte data .
most of these studies , however , apply simple , phenomenological models to the spectra that are limited in the energy resolution or band coverage . to understand
the physical conditions of the accretion flow in the vhs , detailed analysis based on physically self - consistent models would be required .
4u 163047 is an important target to study the disk wind and jet formation mechanism . @xcite
discover absorption lines from highly ionized iron ions in the _ suzaku _ spectra of 4u 163047 in the hss , indicating the presence of disk wind , at least , in this state .
recently , @xcite also report an absorption - line feature around 7 kev when the source was in a very similar state to that in the first observation of @xcite .
these facts suggest a relatively high inclination angle of the system , @xmath11 , from analogy with other bhbs with disk wind signatures @xcite .
this is also supported by @xcite from the rather hard color of the disc in the hss , which can be explained by relativistic effects on the disc spectrum at high inclination . @xcite
discover doppler shifted emission lines of relativistic baryonic jets from an _ xmm - newton _ observation performed on 2012 september 28 . in this paper , we analyze the _ suzaku _ data of 4u 163047 taken in the vhs during the 2012 outburst .
the observations and data reduction are described in section [ obs ] .
we present the analysis results of the vhs spectra in section [ analysis ] .
the method to correct for the dust scattering effects is explained there .
the results of re - analysis of the hss spectra taken in 2006 february are presented in section [ analysis2 ] .
we report the results from infrared observations performed one day before the _ suzaku _ one at the infra - red survey facility ( irsf ) in section [ infrared ] .
discussions are given in section [ discuss ] , focusing on the inner disk truncation , jet lifetime , and disk wind . throughout the paper ,
we adopt the distance of @xmath12 kpc and inclination angle of @xmath13 as the reference values .
we assume the solar element abundances given by @xcite .
all errors attached to spectral parameters are those at 90% confidence limits for a single parameter of interest .
we performed a too ( target of opportunity ) observation of 4u 163047 with _ suzaku _ from 2012 october 2 00:24:59 ( ut ) to 20:38:24 .
figure [ maxi_bat ] shows its long - term light curves in the 310 kev and 1550 kev bands obtained with the gas slit camera ( gsc ) on - board monitor of all - sky x - ray image ( _ maxi _ ; * ? ? ?
* ) and the burst alert telescope ( bat ) on - board _ swift _ @xcite , respectively .
the _ maxi _ light curves are obtained by the image fitting method described in @xcite .
the epoch of the _ suzaku _ observation is indicated with the down arrow ( left ) in figure [ maxi_bat ] . as noticed from the figure , it corresponds to the peak of the hard x - ray flux above 15 kev during the 2012 september - october outburst .
detailed behavior in the soft band ( @xmath14 kev ) was more complex , showing large variability for @xmath1510 days before the suzaku observation .
the inset in figure [ maxi_bat ] plots the _ maxi _ light curve in 1/4 day bins around the _ suzaku _ observation .
the data between mjd = 5619956200.25 are not available because the source was out of the field - of - view of _ maxi_. figure [ hardness ] plots the hardness - intensity diagram corresponding to the epoch shown in figure [ maxi_bat ] .
it first evolved in the opposite direction to the normal track of bhbs in outbursts @xcite , i.e. it moved from left to right along the upper branch of the hardness - intensity diagram .
we note that this is a well known behavior of the vhs in several bhbs ( e.g. , * ? ? ?
* ; * ? ? ?
_ suzaku _ @xcite carries two types of instruments , the x - ray imaging spectrometer ( xis ) and the hard x - ray detector ( hxd ) .
the xis is a focal plane x - ray ccd camera coupled to the x - ray telescope ( xrt ) and covers the 0.212 kev band . at the time of our observation ,
two front - side illuminated xiss ( fi - xiss ) , xis0 and xis3 , and one back - side illuminated xis ( bi - xis ) , xis1 , were available .
the hxd is a collimated type non - imaging instrument , consisting of silicon pin diodes and gadolinium silicon oxide ( gso ) crystal scintillators , which cover the 1070 kev and 40600 kev band , respectively . to minimize photon pile - up events ,
all xiss were operated with a burst option of 0.1 sec exposure per frame .
the 1/4 window mode was employed for xis0 and xis1 , while the full window mode is used for xis3 .
the net exposure time for xis0 and xis1 is @xmath163 ks and that for xis3 is @xmath160.5 ks .
the hxd was operated in the standard mode , and its net exposure time corrected for dead time is @xmath1540 ks .
we utilize the `` cleaned '' event files reduced by the pipeline processing version 2.8.16.34 , and analyze them in a standard manner with heasoft version 6.12 and calibration database ( caldb ) released on 2012 september 2 . to extract the xis light curves and spectra , we accumulate photon events in a circular region with a radius of @xmath17 centered at the source position .
the background of the xis is neglected because the contribution is less than 1% of the source counts . as for the background of the hxd
, we utilize the so - called `` tuned '' background files provided by the _
suzaku _ team .
the modeled cosmic x - ray background ( cxb ) is further subtracted from the pin data but not from the gso data since its contribution is less than 0.1@xmath18 of the total gso background .
figure [ 4u_lc ] plots the _ suzaku _ light curves of 4u 163047 in three energy bands , 210 kev ( xis ) , 1060 kev ( pin ) , and 50200 kev ( gso ) .
the averaged 210 kev flux measured by xis0 is about @xmath19 .
the pin and gso light curves are corrected for dead time .
figure [ psd ] shows the normalized power spectral density ( psd ) obtained from the pin light curve with 0.001 sec binning .
this psd is calculated over the entire observation duration by using the powspec tool in the xronos software package version 5.22 .
as noticed , no significant quasi periodic oscillation ( qpo ) is detected in the range of 0.0010.02 hz .
the time - averaged spectra of the xis and the hxd are plotted in the uppermost panel of figure [ 4u_spec ] .
we create the response matrix files and auxiliary response files of the xis with the ftool xisrmfgen and xissimarfgen @xcite , respectively . as for the response file of the hxd
, we use ae_hxd_pinxinome11_20110601.rsp for pin , and ae_hxd_gsoxinom_20100524.rsp and ae_hxd_gsoxinom_crab_20100526.arf for gso . to account for possible calibration uncertainties ,
1@xmath18 systematic errors are included in each spectral bin of all spectra .
considering possible calibration uncertainties in the low energy tail in the pulse height amplitude distribution , we increase the errors to 3% at energies below 1.7 kev in the xis spectra . in order to compare our results in the vhs with those in the hss ,
we re - analyze the xis spectra taken between 2006 february 8 15:09:56 to february 9 03:20:18 ( obsid=400010010 ) published by @xcite .
the main reason is to obtain the mcd parameters in the hss by taking into account dust - scattering effects , which were neglected in @xcite .
all xiss were operated with the 1/4 window mode and 1-s burst option in this observation .
the net exposure is 22 ks for each xis .
the xis spectra are integrated within a circular region of a radius of @xmath20 , while central region of @xmath21 are excluded in order to minimize the effect of pileup .
the hxd was operated in the standard mode , and its net exposure time corrected for dead time is @xmath1522 ks .
the same systematic errors as those applied for the 2012 data are included .
the flux in the 210 kev band is about @xmath22 .
in this section , we present the results of detailed spectral analysis of the _ suzaku _ data of 2012 october .
the spectral fit is performed by using xspec version 12.8.0 .
we simultaneously fit the spectra of three xiss , hxd / pin , and hxd / gso , using the energy range of 1.29.0 kev for xis0 and xis3 , 1.28.0 kev for xis1 , 1360 kev for pin , and 7 0200 kev for gso .
the xis data in the 1.71.9 kev energy band were excluded in the spectral fit to avoid the large calibration uncertainties . the cross normalization factor between xis0 and hxd ( both pin and gso ) is fixed at 1.164 while those between xis0 , xis1 , and xis3 are set to be free .
we extend the energy range of the spectral fit to 0.011000 kev when using convolution models .
because 4u 163047 suffers from heavy absorption ( @xmath23 @xmath8 ) by the interstellar medium , we have to take into account the effects by dust scattering , which become more significant at lower energies .
a part of the direct emission from the object is scattered out by interstellar dust , while photons emitted toward slightly different angles are scattered in the line of sight , thus making the dust - scattering halo around the point source .
when one can not integrate all the emission including both direct and halo components , the scattered - in and scattered - out photons are not canceled each other .
the situation becomes more complex for a variable source because there is a time delay between the scattered - in and direct photons due to their different path lengths .
hence , we have to take into account these effects in the spectral analysis of a heavily absorbed object unless the spectrum is constant _ and _ the integrated image region is sufficiently large to fully cover the dust - scattering halo .
following the work by @xcite , who applied the same procedure to _ suzaku _ data of grs 1915 + 105 , we first estimate the fraction of dust scattering halo of 4u 1630 - 47 contained in our xis spectra . according to @xcite , who studied the detailed image profile of the dust scattering halo of the bright galactic source gx 13 + 1 observed with chandra
, the surface density of the halo profile can be approximately represented by a power law with @xmath24 below @xmath25 and @xmath26 above @xmath25 .
we then run the ftool xissimarfgen by assuming this profile and find that about 65% of the total photons of the dust - scattering halo is included in the vhs spectra within our integration region . in spectral fit ,
we utilize the dscat model , a local model on xspec developed by @xcite .
it calculates the fractions of the scattered - in and scattered - out components as a function of energy , by referring to the cross section of the dust scattering given by @xcite . ] .
the free parameters of the dscat model are the hydrogen column density ( @xmath27 ) and the fraction of the halo ( scattered - in ) component contained in the xis spectra ( hereafter `` scattering fraction '' ) , which is fixed at 0.65 .
this is not applied to the spectra of the hxd , which has a sufficiently large field of view . for simplicity
, we neglect the effect of time variability .
the maximum time delay of the scattered - in component to the direct one is estimated to be @xmath161.5 days at the outer integration radius of @xmath17 by assuming that the scatterers are located at the half distance of @xmath12 kpc .
the _ maxi _ light curve ( figure [ maxi_bat ] ) shows that the soft x - ray flux was constant within 10% for 1.5 days before the _ suzaku _ observation .
we confirm that our conclusions are not affected ( see section [ truncation ] ) . in the spectral fit presented below ,
we multiply to an intrinsic model both the wabs and dscat models , which take into account the interstellar photo - electric absorption @xcite and dust - scattering effects , respectively .
the hydrogen column density of the dscat model is linked to that of the wabs model .
the estimated contribution of the halo component in the xis0 spectrum is plotted by the magenta curve in the upper limit of figure [ 4u_spec ] .
we find that the net effects by dust scattering are @xmath1683% at 2 kev and @xmath1666% at 3 kev compared with the case when neglected .
thus , consideration of the dust - scattering effects is critical when we discuss the intrinsic source spectrum at low energies . to begin with
, we try simple phenomenological models to fit the xis+hxd spectra in the vhs .
we find a single power law model gives a very poor fit ( @xmath28 ) with the best - fit photon index of @xmath29 .
the model leaves systematic hard x - ray excess above @xmath1620 kev that monotonically increases with energy and soft excess below @xmath162 kev .
the residuals indicate that the overall spectral shape is approximated by a power law flatter than @xmath29 on which an additional soft component is required .
next , we fit the spectra with a mcd plus a power law model ( diskbb + pow in the xspec terminology , hereafter referred as `` model a '' ) .
this is a conventional model often adopted to fit the spectra of bhbs in the hss when a power law component is rather weak .
we find that this model well reproduce the _ suzaku _ spectra in the vhs with @xmath30 , even though the power law component dominates the total flux at all energies from 1.2 kev to 200 kev , unlike the case of the hss .
we obtain a power law photon index of @xmath31 and innermost disk temperature and radius of @xmath32 kev and @xmath33 km from the mcd component , where @xmath34 and @xmath35 .
the model description and best - fit parameters are given in tables [ model ] and [ result ] , respectively .
figure [ 4u_spec](a ) shows the residuals between the data and model in units of @xmath36 .
the best - fit unabsorbed spectrum of model a is shown in figure [ bestfitmodel](a ) , where different components are separately plotted .
the absorbed 1.2200 kev flux is estimated @xmath37 , which corresponds to the absorbed luminosity of @xmath38 assuming isotropic emission .
this high luminosity and the dominant steep power law component confirm that the object was in the vhs .
ll model a & wabs*dscat*(diskbb+powerlaw ) + model b & wabs*dscat*(simpl(d)*diskbb+simpl(c)*diskbb+ + & rdblur*rfxconv*simpl(c)*diskbb ) + model c & wabs*dscat*(simpl(d)*diskbb+simpl(c)*diskbb+nthcomp+ + & rdblur*rfxconv*(simpl(c)*diskbb+nthcomp ) ) + model d & wabs*dscat*(simpl(d)*dkbbfth(d)+simpl(c)*dkbbfth(d)+dkbbfth(c)+ + & rdblur*rfxconv*(simpl(c)*dkbbfth(d)+dkbbfth(c ) ) ) + hereafter , we apply more physically - motivated models , where the power law component is modeled as comptonization of seed photons emitted from the disk . essentially the same models are employed to fit the _ suzaku _ spectrum of the bhb gx 3394 in the vhs by @xcite .
first , we utilize the simpl model @xcite multiplied to the diskbb component .
the simpl model is an empirical model in which a fraction of the photons in an input seed spectrum is compton - scattered into a power - law component .
we find that this model alone does not reproduce the xis+hxd spectra , yielding @xmath39 . to improve the fit
, we further consider the reflection of the comptonized component from the accretion disk by using the rfxconv model @xcite .
this is a convolution model that combines the reflected spectra of @xcite from a constant density ionized disk with the ireflect model ( a convolution version of the pexriv model ; * ? ? ?
* ) . to consider general relativistic smearing effect , we utilize the model rdblur @xcite
following @xcite , we fix the iron abundance to be solar , the inner and outer radius @xmath40 , @xmath41 , and the power law index of emissivity @xmath42 , which are very difficult to determine from the fit .
we confirm that other choices of the parameters between @xmath43 or @xmath44 give almost the same results within the errors .
we refer this model including the disk reflection component as `` model b '' , whose description in the xspec terminology is given in table [ model ] .
we find that model b gives a much better fit with @xmath45 than that without a reflection component .
the best - fit parameters are summarized in table [ result ] .
the fitting residuals and unabsorbed best - fit model are plotted in figures [ 4u_spec](b ) and [ bestfitmodel](b ) , respectively .
llcccc & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 + + diskbb & @xmath51 & @xmath52 & @xmath53 & @xmath54 & + & @xmath55 & @xmath56 & @xmath57 & @xmath58 & + + nthcomp & @xmath59 & & & & @xmath60 + or & @xmath61 & & & @xmath62 & @xmath63 + dkbbfth & @xmath64 & & & @xmath65 & @xmath66 + & norm & & & @xmath67 & @xmath68 + & @xmath69 & & & & @xmath70 + + & derived @xmath71 & & & @xmath72 & @xmath73 + + pl & @xmath74 & @xmath75 & @xmath76 & ( 2.1 ) & ( 2.1 ) + or & norm & @xmath77 & & & + simpl & @xmath78 & & @xmath79 & @xmath80 & @xmath81 + + rfxconv & @xmath82 & & @xmath83 & @xmath84 & @xmath85 + & @xmath86 & & @xmath87 & @xmath88 & @xmath89 + + & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 + + photon flux & @xmath95 & & & @xmath96 & @xmath97 + inner radius & @xmath98 & & & @xmath99 & @xmath100 the simpl model used in model b assumes no high energy cutoff , and hence should be regarded to represent comptonized emission by non - thermal electrons . however , previous studies of bhbs in the vhs suggest that the comptonizing corona contains both thermal and non - thermal electrons ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it is not clear whether it consists of two separate components or is a hybrid plasma whose energy distribution is maxwellian at low energies but a power - law at high energies . for simplicity , here we consider the former case and leave the detailed modeling of comptonization by a hybrid plasma for future work .
we note , however , that while a hybrid electron distribution can fit the spectrum , such a homogeneous source model can not explain the result by @xcite , who show that the spectrum of the fastest variability is different to that of the time averaged spectrum .
accordingly , we add a thermal comptonization model @xcite ( model c in table [ model ] ) to model b. as in model b , we include the disk reflection component of the comptonized emission ( both thermal and non - thermal ones ) .
the nthcomp model has four parameters , the seed photon temperature , which we tie at the disk temperature of diskbb , the electron temperature , the photon index @xmath61 , and its normalization .
we set the upper limit of the electron temperature to be 300 kev , as the nthcomp model does not include full relativistic corrections . since we can not constrain both @xmath61 in nthcomp and @xmath101 in simpl , we fix @xmath102 , a typical value for a non - thermal power law component seen in the hss @xcite .
we note that the choice of @xmath101 is not sensitive to the overall fitting result as our data is available up to only 200 kev .
we find that model c gives a better fit than model b , yielding @xmath108 .
the improvement from model b is significant at 99.9995% level from an f - test , i.e. the data require the presence of another comptonized component at high confidence , although the electron temperature of the nthcomp component is not well constrained , with a lower limit of 73 kev .
the best fit parameters are summarized in table [ result ] .
figure [ 4u_spec](c ) plots the residuals , while figure [ bestfitmodel](c ) shows the unabsorbed best - fit model spectrum .
it is seen that the thermal compton component dominates below @xmath1550 kev .
+ in reality , it is expected that the hot comptonizing corona surrounds only an inner part of the disk . in this picture ,
model c is unphysical because it does not take into account any energetic coupling between the disk and corona ; the summed emissivity of the disk blackbody and thermal comptonization components in an inner region is larger than the extrapolation from the outer disk that is not comptonized ( see * ? ? ?
* ; * ? ? ?
this would lead an incorrect estimate of the innermost disk radius .
to overcome this situation , we replace diskbb and nthcomp to dkbbfth @xcite ( model d in table [ model ] ) .
the dkbbfth model assumes that the energy released by gravity is dissipated locally , which is emitted purely from the standard disk at @xmath109 but is split into the disk and corona at @xmath110 .
thus , the inner part of the disk covered by the corona is less luminous than it would have been if the corona did not exist .
the dkbbfth model has five parameters , the disk temperature , the outer radius of the corona @xmath111 , the photon index of the comptonized component , the temperature of the corona , and the normalization .
we set all these parameters to be free in the fitting .
since this model only takes into account thermal comptonization , we further add the simpl model to describe non - thermal comptonization .
there we set the disk emission to be the seed photons of simpl and fix the photon index at 2.1 .
we note that the number of free parameters in model d is the same as in model c. the fit shows a significant improvement ( @xmath112 ) from model c. thus , we conclude that model d gives the best description of the _ suzaku _ spectra among the three physically - motivated models examined above .
the best - fit parameters are summarized in table [ result ] and the fitting residuals are plotted figure [ 4u_spec](d ) .
the best - fit unabsorbed model is shown in figure [ bestfitmodel](d ) .
the dashed - dotted curve shows the intrinsic disk emission that would be observed when all the luminosity of the dkbbfth component were emitted as the standard disk .
to compare with the results in the vhs , we also perform spectral fit to the _ suzaku _ spectra of 4u 163047 taken in 2006 february 8 when the object was in the hss ( see section [ obs ] ) .
we simultaneously fit the spectra of the 4 xiss and hxd / pin .
we utilize the energy band of 1330 kev of the pin spectrum . only the xis spectra below 6 kev
are utilized to avoid pile - up effects that produce an artificial hard excess above @xmath157 kev and to ignore the strong absorption lines from highly ionized iron ions ( see * ? ? ? * for details ) .
the energy ranges of 1.71.9 kev and 2.22.4 kev are excluded , where calibration uncertainties are large .
systematic errors of 1% and 3% are included in each spectral bin above 1.9 kev and below 1.7 kev , respectively . in the same way as for the vhs spectra
, we multiply the dscat*wabs model to the intrinsic continuum to take into account dust scattering effects , which could affect the continuum spectral parameters reported by @xcite . using the xissimargen ftool
, we estimate the scattering fraction to be 0.78 .
the cross normalization factor of hxd / pin relative to xis0 is fixed at 1.164 , while those of xis1 , xis2 , and xis3 are set to be free .
thus , we can directly compare the absolute flux between the two states by using the xis0 as the reference .
we apply a model consisting of the mcd component and its comptonization ( simpl*diskbb in the xspec terminology ) . the photon index in simpl
is fixed at 2.1 ( see fig 2b of * ? ? ?
although the model is rather simple , the fit is found to be reasonably good with @xmath113 .
we obtain the innermost disk temperature and radius of @xmath114 kev and @xmath115 km , respectively . the derived innermost radius is larger than that reported by @xcite , @xmath116 km .
the difference is attributable to the dust scattering effects .
the hydrogen column density is determined to be @xmath117 , which is consistent with the @xcite result and our result in the vhs .
the small scattered fraction @xmath118 means that the mcd component was dominant in this observation .
the absorbed 1.230 kev flux is estimated to be @xmath119 , which gives an absorbed luminosity of @xmath120 by assuming isotropic emission .
we performed near - infrared photometric observations of 4u 163047 in the @xmath121 ( 1 .25 @xmath122 ) , @xmath123 ( 1.63 @xmath122 ) , and @xmath124 ( 2.14 @xmath122)-bands on 2012 october 1 ( one day before the _ suzaku _ observation ) by using sirius camera @xcite on the 1.4-m irsf telescope at the south african astronomical observatory ( saao ) . we obtained 25 object - frames with an exposure of 15 sec per frame , and thus the net exposure time was 375 sec .
the seeing in full width half maximum in the @xmath121-band was @xmath125 ( 3.5 pixels ) .
we perform standard data reduction ( i.e. , dark subtraction , flat - fielding , sky subtraction , and combining dithered images ) with the irsf pipeline software on iraf ( the image reduction and analysis facility , distributed by the national optical astronomy observatory ) version 2.16 .
we combine all the object frames obtained in one night to maximize signal - to - noise ratio .
the near infrared counterpart has been identified by @xcite , but we are not able to measure the flux by aperture photometry since 4u 163047 is contaminated by nearby stars in the infrared images .
thus , we perform psf ( point spread function ) photometry using the daophot @xcite package in iraf .
we obtain the near - infrared magnitudes of @xmath126 mag in the @xmath123-band and @xmath127 mag in the @xmath124-band , while we are unable to constrain the @xmath121-band flux because of the severe source confusion .
these magnitudes are almost the same as those reported in the 1998 outburst @xcite .
we estimate the extinction in each band as @xmath128 mag and @xmath129 mag by using the relation between the hydrogen column density and optical extinction @xmath130 , @xmath131 \times 10^{21 } [ \rm{cm^{-2}}]$ ] @xcite , and the conversion factor from @xmath130 into @xmath132 or @xmath133 given by @xcite .
the extinction - corrected irsf fluxes ( @xmath134 mag , @xmath135 mag ) are plotted as the red points in figure [ irsfobs ] .
note that the extremely blue color ( @xmath136 mag ) , which is hard to explain by any stellar or continuum source , most probably comes from the errors in the photometry , @xmath27 , and/or the @xmath27 to @xmath130 conversion .
we find that the last one can produce the largest uncertainty .
figure 3 of @xcite shows that there is a deviation by a factor of @xmath152 in the relation between @xmath27 and @xmath130 for some objects .
accordingly , we consider an extreme case where the @xmath130 to @xmath27 ratio is twice smaller than the above as a lower limit of the extinction correction .
this gives the corrected fluxes of @xmath137 mag and @xmath138 mag , which are plotted as the blue points in figure [ irsfobs ] .
thus , the extinction - corrected magnitudes are estimated to be in the range of @xmath139 and @xmath140 . in figure [ irsfobs ] , the intrinsic thermal emission from the standard disk based on model d ( section [ modeld ] ) that would be observed without comptonization ( i.e. , the red dashed line in figure 6(d ) )
is plotted by the green solid curve .
as noticed , even the lower limits of the observed near - infrared fluxes are larger than the expected fluxes from the mcd component .
this suggests that the fluxes are dominated by other components .
we estimate the absolute magnitudes to be @xmath141 mag and @xmath142 mag for a distance of 10 kpc .
thus if the near - infrared fluxes completely come from the companion star , it would be an early type star ( o7-b2 ) , supporting the suggestion by @xcite , or a k type giant star .
these magnitudes should be regarded as the upper limit of the true luminosity of the companion star , however , as we also expect other origins that could be responsible for the near - infrared fluxes , including ( 1 ) the irradiation of the outer disk by the x - ray emission , ( 2 ) synchrotron radiation from the jets , and/or ( 3 ) cyclo - synchrotron emission from the hot flow @xcite .
unfortunately , the large uncertainties in the extinction - corrected fluxes make it very difficult to discriminate these possibilities .
we note that the observed ( uncorrected ) near - infrared fluxes in the vhs are almost the same as those reported by @xcite when the source was in the hss with a similar x - ray flux level .
since jet ejection is believed to be quenched in the hss , this comparison suggests that the synchrotron emission from the jets , if any , is not a dominant source of the near infrared fluxes in the vhs .
using _ suzaku _ , we have obtained so - far the best quality broad band x - ray spectra of 4u 163047 in the vhs , covering the 1200 kev band with good energy resolution by x - ray ccd below 10 kev .
the observation epoch corresponds to when the hard x - ray flux above 15 kev was peaked during the 2012 september - october outburst . in the spectral analysis
, we carefully take into account the effects by dust scattering .
the continuum is found to be well approximated by a phenomenological model consisting of a mcd and a power law component dominating the total flux in the entire 1.2200 kev band ( model a ) . however , we regard that this modeling is unnatural because ( 1 ) no low energy turn - over is included , which should exist if the power law component is originated by comptonization of seed photons emitted from the disk , and ( 2 ) the derived column density is significantly larger than that obtained in the hss .
we thus consider three physically - motivated models where the disk photons are compton - scattered by a hot corona surrounding it . among them , model d , where the comptonizing corona is energetically coupled with the underlying standard disk below a transition radius , best reproduce the spectra .
a significant reflection component from the inner disk with a solid angle of @xmath143 is detected .
this is consistent with the recent result by @xcite , although the state was different from ours .
below , we discuss the inner disk geometry on the basis of the result of model d. no significant emission or absorption lines are detected in our spectra .
this fact gives tight constraints on the presence of the baryonic jets discovered by @xcite 4 days before the _ suzaku _ observation , and that of the disk wind seen in the hss @xcite .
we find that the standard disk in the vhs is most likely truncated before reaching the isco . from model d ,
the best physically self - consistent model considered here , we derive the innermost radius of the standard disk is @xmath144 km in the vhs .
this is significantly larger than that found in the hss , @xmath115 km , which is found to be constant during this state and hence most probably corresponds to the isco @xcite .
since we directly compare the results obtained with the same detector ( xis0 ) , the comparison is free from any cross - normalization uncertainties between different instruments .
also , a common model is employed to represent the interstellar absorption and dust scattering , making the relative systematic error least .
when we assume an extreme case that the soft x - ray flux producing the dust scattering halo is by 10% lower than the suzaku epoch ( see section [ dust ] ) , we obtain an even larger innermost radius .
thus , the conclusion of disk truncation is robust against uncertainty in the correction for dust scattering . a truncated disk in the vhs is also suggested from the bhb gx 3394 by applying essentially the same model as our model d @xcite .
hence , disk truncation would be a common feature of the vhs , although the innermost radius is less than twice that of the isco in both objects .
we note that the `` true '' disk radius , @xmath145 , is estimated as @xmath146 , where @xmath147 is the color - to - effective temperature ratio and @xmath148 is a correction factor for the stress - free boundary condition at the isco @xcite .
to explain the @xmath4 value in the vhs by the isco , @xmath147 must be smaller than that in the hss . as discussed in @xcite ,
this case is quite unlikely under strong irradiation by the corona in the vhs .
relativistic baryonic jets of 4u 163047 were detected with _
xmm - newton _ on 2012 september 28 @xcite when the source was also in the vhs
. the _ xmm - newton _ spectrum shows doppler - shifted fe xxvi lines at 7.3 kev and 4 kev . assuming that they are twin jets of the same physical parameters emitted to opposite directions
, they estimate the jet speed , the inclination angle , and the cone angle to be @xmath149 , @xmath150 , and @xmath151 , respectively . the parameters are quite similar to those of ss 433 @xcite , the unique galactic microquasar exhibiting steady relativistic jets with a velocity of 0.26@xmath152 , which most likely contains a black hole @xcite . in contrast , our _ suzaku _ spectra , taken only four days after the _ xmm - newton _ observation , do not show such emission - line features . the equivalent width of the 7.3 kev emission line detected by @xcite is constrained to be @xmath1533 ev ( 90% confidence limit ) , which is about 1/12 of that in the _ xmm - newton _ spectra .
this suggests that the relativistic jets detected by @xcite had vanished in four days . in the case of ss 433
, the length of the x - ray emitting region of the jets is @xmath154 cm , corresponding to a traveling time of only @xmath155 s @xcite .
thus , if we assume the similar conditions as ss 433 , the jets of 4u 163047 would vanish in an hour after the ejection stops .
we can not , however , exclude the possibilities that a jet might still exist , such as compact jets often seen in the lhs @xcite .
the variation of the x - ray flux can provide clues to reveal the condition for the ejection of baryonic jets from a black hole . according to the _ maxi _ and _ swift_/bat
light curves ( figure [ maxi_bat ] ) , the _ suzaku _ observation was performed just after the hard x - ray flux showed a sudden increase from mjd@xmath15656201 .
the same trend is confirmed in the soft band , though less clear because of the data gap of _ maxi_. the source was by @xmath1530% fainter when the jets were detected with
_ xmm - newton _ ( mjd=56198 ) than in the _ suzaku _ epoch ( mjd=56202 ) .
thus , phenomenologically , it would be possible that the launch of the baryonic jets somehow ceased as more accretion power was dissipated into radiation .
however , we estimate the kinetic power of the jets to be @xmath157 erg s@xmath158 on the basis of the jet model of ss 433 @xcite , assuming that it is proportional to the luminosity of the ionized iron - k emission lines . this huge power can not be accounted for by the observed increase in the luminosity at mjd@xmath15656201 ( @xmath159 erg s@xmath158 ) unless the radiation efficiency is very low .
thus , a significant decrease in the intrinsic mass accretion rate would be required . to obtain firm observational clues on the jet formation mechanism ,
we need more frequent monitoring observations of bhbs in high luminosity states with high - quality x - ray spectroscopy .
the _ suzaku _ spectra of 4u 163047 in the vhs do not show any significant absorption line features of highly ionized iron ions that were detected in the hss by @xcite .
the upper limits for the equivalent widths of the absorption lines are 1.3 ev and 1.4 ev ( 90% confidence limits ) at 6.72 kev and 6.99 kev , corresponding those of and observed in the hss , respectively .
there are two possibilities to account for this fact ; ( 1 ) there is actually no disk wind in the vhs or ( 2 ) it exists but simply become invisible by being almost fully ionized by the strong x - ray irradiation . to examine the latter possibility , we perform simulations of a photo - ionized gas using xstar version 2.2.1bk . following @xcite , we first reproduce the physical parameters ( the ionization stage and column density ) of the disk wind detected in the hss by assuming the same spectrum and luminosity as observed
. then , we change the x - ray spectrum and luminosity to those in the vhs by keeping the other parameters same .
we find that the ion fractions of and becomes @xmath161/20 and @xmath161/4 of those in the hss .
the predicted equivalent width of is larger than the observed upper limit , suggesting that photo - ionization effects alone can not explain the absence of the absorption lines in the _ suzaku _ spectra . however , it is still plausible that a disk wind with a smaller density ( and/or at a smaller distance ) than that in the hss is present in the vhs ( see @xcite for discussion of thermodynamical stability , and @xcite for discussion of launch radius and density ) .
further observations of bhbs with good energy resolution covering different states and luminosities are necessary to establish the critical conditions to launch a disk wind . 1 .
the time - averaged _ suzaku _ spectra are well described by a physically self - consistent model consisting of thermal and non - thermal comptonization of the inner disk emission with energetic coupling between the disk and corona .
a strong relativistic reflection component from the accretion disk is required .
we properly take into account the effects of dust scattering .
2 . comparing the result of the _ suzaku _ spectra in the hss observed in 2006 february
, we find evidence that the accretion disk is slightly truncated before reaching to the isco in the vhs .
3 . our spectra do not show any doppler - shifted line emissions from the relativistic jets that were detected four days before our observation with _ xmm - newton_. this suggests that the jets were a transient event and vanished between these observations .
we do not detect absorption line features from highly ionized iron ions that were previously present in the high / soft state .
the upper limits on the equivalent width of these lines are not compatible with over - ionization by the higher luminosity irradiation of same wind .
if a wind still remain , it has changed in launch radius and/or density from that seen in the hss .
5 . from near - infrared data taken with irsf one day before the _ suzaku _ observation
, we obtain @xmath160 mag in the @xmath123-band and @xmath161 mag in the @xmath124-band for the counterpart .
since the extinction - corrected fluxes are larger than the extrapolated fluxes from the mcd component , contribution from a companion star , which would be an early type star or a k - type giant star if it dominates the total fluxes , and/or other origins such as irradiation of the outer disk by the x - ray emission are required .
we are grateful to the _ suzaku _ project for performing this too observation upon our request .
we thank tatsuhito yoshikawa for his help on the analysis of the irsf data , and shinya yamada and shin mineshige for discussions .
this work was partly supported by the grant - in - aid for scientific research no.23540265 ( y.u . ) , no.24111717 ( k.t . ) , and for young researchers ( m.s . ) .
abe , y. , fukazawa , y. , kubota , a. , kasama , d. , & makishima , k. 2005 , , 57 , 629 anders , e. , & grevesse , n. 1989 , , 53 , 197 augusteijn , t. , kuulkers , e. , & van kerkwijk , m. h. 2001 , , 375 , 447 axelsson , m. , done , c. , & hjalmarsdotter , l. 2014 , , 438 , 657 begelman , m. c. , mckee , c. f. , & shields , g. a. 1983 , , 271 , 70 belloni , t. , parolin , i. , del santo , m. , et al .
2006 , , 367 , 1113 burrows , d. n. , hill , j. e. , nousek , j. a. , et al .
2005 , , 120 , 165 chakravorty , s. , lee , j. c. , & neilsen , j. 2013 , , 436 , 560 daz trigo , m. , miller - jones , j. c. a. , migliari , s. , broderick , j. w. , & tzioumis , t. 2013 , , 504 , 260 done , c. , & kubota , a. 2006 , , 371 , 1216 done , c. , gierliski , m. , & kubota , a. 2007 , , 15 , 1 draine , b. t. 2003 , , 598 , 1026 ebisawa , k. , makino , f. , mitsuda , k. , et al .
1993 , , 403 , 684 esin , a. a. , mcclintock , j. e. , & narayan , r. 1997 , , 489 , 865 fabian , a. c. , rees , m. j. , stella , l. , & white , n. e. 1989 , , 238 , 729 fender , r. , corbel , s. , tzioumis , t. , et al . 1999 , , 519 , l165 fender , r. p. , belloni , t. m. , & gallo , e. 2004 , , 355 , 1105 gierliski , m. , zdziarski , a. a. , poutanen , j. , et al . 1999 , , 309 , 496 gierliski , m. , & done , c. 2003 , , 342 , 1083 gierliski , m. , done , c. , & page , k. 2009 , , 392 , 1106 gladstone , j. , done , c. , & gierliski , m. 2007 , , 378 , 13 hiroi , k. , ueda , y. , hayashida , m. , et al .
2013 , , 207 , 36 ishisaki , y. , maeda , y. , fujimoto , r. , et al .
2007 , , 59 , 113 king , a. l. , walton , d. j. , miller , j. m. , et al . 2014 ,
arxiv:1401.3646 kolehmainen , m. , done , c. , & daz trigo , m. 2011 , , 416 , 311 kotani , t. , kawai , n. , matsuoka , m. , & brinkmann , w. 1996 , , 48 , 619 kotani , t. , ebisawa , k. , dotani , t. , et al .
2000 , , 539 , 413 kubota , a. , tanaka , y. , makishima , k. , et al .
1998 , , 50 , 667 kubota , a. , makishima , k. , & ebisawa , k. 2001 , , 560 , l147 kubota , a. , & makishima , k. 2004 , , 601 , 428 kubota , a. , & done , c. 2004 , , 353 , 980 kubota , a. , dotani , t. , cottam , j. , et al .
2007 , , 59 , 185 kubota , k. , ueda , y. , fabrika , s. , et al .
2010 , , 709 , 1374 kuulkers , e. , wijnands , r. , belloni , t. , et al . 1998 , , 494 , 753 magdziarz , p. , & zdziarski , a. a. 1995 , , 273 , 837 makishima , k. , maejima , y. , mitsuda , k. , et al .
1986 , , 308 , 635 makishima , k. , takahashi , h. , yamada , s. , et al .
2008 , , 60 , 585 martins , f. , & plez , b. 2006 , , 457 , 637 matsuoka , m. , kawasaki , k. , ueno , s. , et al .
2009 , , 61 , 999 mitsuda , k. , inoue , h. , koyama , k. , et al . 1984 , , 36 , 741 mitsuda , k. , bautz , m. , inoue , h. , et al .
2007 , , 59 , 1 morrison , r. , & mccammon , d. 1983 , , 270 , 119 motta , s. , homan , j. , muoz darias , t. , et al .
2012 , , 427 , 595 muoz - darias , t. , coriat , m. , plant , d. s. , et al .
2013 , , 432 , 1330 nagayama , t. , nagashima , c. , nakajima , y. , et al .
2003 , , 4841 , 459 nakahira , s. , negoro , h. , shidatsu , m. , et al .
2014 , , submitted namiki , m. , kawai , n. , kotani , t. , & makishima , k. 2003 , , 55 , 281 oosterbroek , t. , parmar , a. n. , kuulkers , e. , et al .
1998 , , 340 , 431 parmar , a. n. , stella , l. , & white , n. e. 1986 , , 304 , 664 parmar , a. n. , williams , o. r. , kuulkers , e. , angelini , l. , & white , n. e. 1997 , , 319 , 855 ponti , g. , fender , r. p. , begelman , m. c. , et al .
2012 , , 422 , l11 predehl , p. , & schmitt , j. h. m. m. 1995 , , 293 , 889 rahoui , f. , chaty , s. , rodriguez , j. , et al .
2010 , , 715 , 1191 rahoui , f. , coriat , m. , corbel , s. , et al .
2012 , , 422 , 2202 rieke , g. h. , & lebofsky , m. j. 1985 , , 288 , 618 ross , r. r. , & fabian , a. c. 2005 , , 358 , 211 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 shidatsu , m. , ueda , y. , tazaki , f. , et al .
2011 , , 63 , 785 shidatsu , m. , ueda , y. , nakahira , s. , et al .
2013 , , 779 , 26 smith , r. k. , edgar , r. j. , & shafer , r. a. 2002 , , 581 , 562 steiner , j. f. , narayan , r. , mcclintock , j. e. , & ebisawa , k. 2009 , , 121 , 1279 stetson , p. b. 1987 , , 99 , 191 tamura , m. , kubota , a. , yamada , s. , et al .
2012 , , 753 , 65 tomsick , j. a. , lapshov , i. , & kaaret , p. 1998
, , 494 , 747 tomsick , j. a. , & kaaret , p. 2000 , , 537 , 448 tomsick , j. a. , corbel , s. , goldwurm , a. , & kaaret , p. 2005
, , 630 , 413 tomsick , j. a. , yamaoka , k. , corbel , s. , et al .
2009 , , 707 , l87 trudolyubov , s. p. , borozdin , k. n. , & priedhorsky , w. c. 2001 , , 322 , 309 ueda , y. , inoue , h. , tanaka , y. , et al .
1998 , , 492 , 782 ueda , y. , honda , k. , takahashi , h. , et al .
2010 , , 713 , 257 veledina , a. , poutanen , j. , & vurm , i. 2013 , , 430 , 3196 wainscoat , r. j. , cohen , m. , volk , k. , walker , h. j. , & schwartz , d. e. 1992 , , 83 , 111 zdziarski , a. a. , johnson , w. n. , & magdziarz , p. 1996 , , 283 , 193 zdziarski , a. a. , poutanen , j. , paciesas , w. s. , & wen , l. 2002 , , 578 , 357 ycki , p. t. , done , c. , & smith , d. a. 1999 , , 309 , 561 | we report the results from an x - ray and near - infrared observation of the galactic black hole binary 4u 163047 in the very high state , performed with _
suzaku _ and irsf around the peak of the 2012 september - october outburst .
the x - ray spectrum is approximated by a steep power law , with photon index of 3.2 , identifying the source as being in the very high state .
a more detailed fit shows that the x - ray continuum is well described by a multi - color disc , together with thermal and non - thermal comptonization .
the inner disc appears slightly truncated by comparison with a previous high / soft state of this source , even taking into account energetic coupling between the disc and corona , although there are uncertainties due to the dust scattering correction .
the near - infrared fluxes are higher than the extrapolated disc model , showing that there is a contribution from irradiation in the outer disk and/or the companion star at these wavelengths .
our x - ray spectra do not show the doppler shifted iron emission lines indicating a baryonic jet which were seen four days previously in an xmm - newton observation , despite the source being in a similar state .
there are also no significant absorption lines from highly ionized irons as are seen in the previous high / soft state data .
we show that the increased source luminosity is not enough on its own to make the wind so highly ionized as to be undetectable .
this shows that the disc wind has changed in terms of its launch radius and/or density compared to the high / soft state . |
geometric invariant theory @xcite has been very successful in forming moduli spaces of ( semi)stable coherent sheaves over polarised algebraic varieties @xmath0 @xcite .
moduli space is constructed as a quotient of a subset of a quot scheme by a group , and the hilbert - mumford criterion is applied to 1-parameter subgroups .
their weights are found to be dominated by positive linear combinations of weights of particularly simple 1-parameter subgroups corresponding to a degeneration of a sheaf @xmath4 into a splitting @xmath5 , for some subsheaf @xmath6 .
thus stability of sheaves is governed by subsheaves ; calculating the corresponding weights leads to the notion of slope stability ( the exact form of the slope depending on the linearisation used on quot ) .
forming moduli spaces of ( semi)stable varieties themselves using git has proved much more difficult , and has mainly been accomplished for canonically polarised varieties using the chow linearisation , due to work of mumford @xcite , gieseker @xcite , and , in a little more generality , viehweg @xcite .
roughly speaking one expects varieties polarised by their canonical bundle @xmath7 to be automatically stable since their moduli functor is already separated because of the birational invariance of spaces of sections of powers of the canonical bundle . in the general case no geometric criterion for ( in)stability has emerged .
this is because the hilbert - mumford criterion has not been successfully simplified or interpreted for varieties ; instead viehweg proved deep positivity results to produce the group - invariant sections of the appropriate line bundle directly ( for varieties with semi - ample canonical bundle ) .
kollar @xcite and others have turned to other methods for producing moduli of varieties , but new impetus to understanding stability has come from the link between k - stability and the existence of khler - einstein and constant scalar curvature khler metrics @xcite to which we apply our methods in @xcite .
our approach uses the hilbert - mumford criterion , as pioneered by mumford @xcite .
he calculates the relevant weights in terms of blow - ups of @xmath8 in subschemes supported on thickenings of @xmath9 ; one of our main results ( corollary [ ref ] ) reduces the calculation to a sum of weights of blow - ups _ in the scheme - theoretic central fibre_. ( in a sense which is made clear by the proof the construction turns the horizontal thickenings of mumford vertical " , into the central fibre . )
taking just one such blow - up in a subscheme @xmath3 gives the deformation to the normal cone of @xmath10 " , analogous to the simple 1-parameter subgroups that arise in the git of sheaves .
this gives a numerical condition for @xmath10 to destabilise @xmath11 , and so a notion of slope stability " , by analogy with the sheaf theory which we now review briefly .
for a sheaf @xmath4 over @xmath0 , the _ reduced _ hilbert polynomial @xmath12 is the monic version of the hilbert polynomial @xmath13 : @xmath14 and the _ slope _ of @xmath4 is its leading nontrivial coefficient , @xmath15 ( this differs from the usual definition deg@xmath16 by unimportant terms that depend only on the geometry of @xmath0 . )
then we say that @xmath4 is _ semistable _ if for all proper coherent subsheaves @xmath6 , @xmath17 for _ gieseker semistability _ , @xmath18 means there exists an @xmath19 such that @xmath20 for all @xmath21 ; for _ slope semistability _ the inequality is at the level of the @xmath22 coefficients : @xmath23 if the inequalities are all strict then @xmath4 is _ stable _ , and this agrees with the appropriate git notions for different choices of linearisation ( i.e.choice of equivariant line bundle on the quot scheme ; in fact jun li s line bundle @xcite is only semi - ample , but he extends git to this case . ) due to different choices of linearisation and asymptotics there are also many notions of stability ( hilbert , chow , asymptotic hilbert , asymptotic chow and k - stability ) for varieties , defined in the next section , and so many versions of slope .
we first describe the one relevant to k - stability . given a subscheme @xmath10 of a polarised algebraic variety @xmath0 , define @xmath24 the hilbert - samuel polynomial for _ fixed _
@xmath25 , @xmath26 defines @xmath27 which are polynomials in @xmath25 ( see section [ k ] ) and so extend by the same formulae to @xmath28 .
then analogously to ( [ mu ] ) we define the _ k - slope _ of @xmath29 ( with respect to @xmath30 and @xmath31 $ ] ) to be @xmath32 setting @xmath33 defines the slope of @xmath11 ( precisely : of @xmath34 with respect to @xmath30 and @xmath35 ) as @xmath36 independently of @xmath35 .
here @xmath37 are the coefficients of the hilbert polynomial @xmath38 and , for @xmath11 normal , @xmath39 ( [ normal ] ) .
setting @xmath40 , we say that @xmath0 is _ k - slope semistable _ if for all proper @xmath3 , @xmath41.\ ] ] cf .
( [ mu2 ] ) .
the definition of slope stability is slightly trickier ( definition [ defstab ] ) ; for this reason we work with @xmath42 instead of @xmath43
. then ( theorem [ thm : kstableslopestable ] ) @xmath11 is slope ( semi)stable if it is k-(semi)stable .
the @xmath44 correction term " in the definition of slope arises from the difference between the hilbert polynomial of a 2-component normal crossing variety and the sum of those of its components ( whereas for sheaves , @xmath45 for any @xmath6 ) .
simon donaldson pointed out that his analysis of stability of toric varieties @xcite throws up a boundary term similar to our @xmath46 ; we explain this in @xcite . similarly for @xmath47 ( embedded in projective space by the space of sections of its polarisation @xmath48 for ease of exposition ; see section [ sec : chow ] for the general case ) , a subscheme @xmath49 and an _ integer _
@xmath50 , we define the _ chow slope _ of @xmath29 to be @xmath51 setting @xmath33 gives @xmath52 ; then chow ( semi)stability implies chow slope ( semi)stability ( theorem [ thmchow ] ) : @xmath53 asymptotic chow is more like gieseker stability and so our slope criterion for that ( [ chow ] ) is more complicated .
section [ k ] describes the various stability notions uniformly , while section [ kslope ] introduces slope stability via the deformation to the normal cone .
arbitrary 1-parameter subgroups are studied in section [ testconfigs ] , calculating their weights in terms of a sequence of simple blow ups in corollary [ ref ] . in section [ converse ] these weights are shown to be those of a deformation to the normal cone under certain circumstances , giving a partial converse to
stability @xmath54 slope stability " .
we used to think this could be done in general , but the failure of the _ thickenings _ of certain flat families of subschemes to themselves be flat prevents us from carrying out the programme in full .
we study when the thickenings _ are _ flat , and get round the problem with a basechange trick in some situations .
this deals with the curve case completely , i.e.stability and slope stability are equivalent there , giving geometric proofs of the k- and asymptotic chow stability of curves of genus @xmath55 .
as far as we know all previous proofs used analysis and combinatorics respectively .
section [ sec : chow ] is devoted to chow stability and section [ egs ] to examples .
many more examples , such as projective bundles , appear in @xcite , in particular showing that k- and slope stability are also equivalent for projective bundles over a curve . * acknowledgements . * we would like to thank brian conrad , kevin costello , dale cutkosky , simon donaldson , david eisenbud , daniel huybrechts , frances kirwan , miles reid and balzs szendri for useful conversations .
the book @xcite and paper @xcite have been very useful to us .
the authors were supported by an epsrc phd studentship and a royal society university research fellowship respectively .
the second author would also like to thank oscar garcia - prada and peter newstead for supporting a visit to csic , madrid , where much of this work was done .
throughout this paper @xmath10 will denote a closed subscheme of a proper irreducible polarised scheme @xmath0 of dimension @xmath56 ; for much of the paper @xmath11 will be a normal irreducible variety .
by @xmath57 we mean the subscheme which is the @xmath58-fold thickening of @xmath10 defined by its ideal sheaf @xmath59 .
we denote the blow up along @xmath10 by @xmath60 , with exceptional divisor @xmath4 .
then @xmath61 and there exists a @xmath62 such that @xmath63 for all @xmath64 . for convenience we often suppress pullback maps , mix multiplicative and additive notation for line bundles , and use the same letter to denote a divisor and the associated line bundle .
for example on @xmath65 , we denote @xmath66 by @xmath67 .
worse still , where it does not cause confusion , @xmath68 may also denote @xmath69 . for brevity
we often denote sheaf cohomology on a space @xmath11 by @xmath70 ; this never means local cohomology .
any vector space @xmath71 with a @xmath72-action splits into one - dimensional weight spaces @xmath73 , where @xmath74 acts on @xmath75 by @xmath76 .
the @xmath77 are the _ weights _ of the action , and @xmath78 is the _ total weight _ of the action ; i.e. the weight of the induced action on @xmath79 . on any family over @xmath80 $ ]
, @xmath81 will denote the pullback of the standard coordinate on @xmath82 .
a line bundle @xmath30 is _ semi - ample _
@xcite if its high powers are globally generated ( i.e. basepoint free ) . in this paper
all semi - ample line bundles will have the additional property that the contraction they define is birational ; that is @xmath30 is the pull back of an ample line bundle from a birational scheme .
( it is important that this contraction can be trivial , i.e. semi - ample includes ample . ) given a ( semi-)ample line bundle @xmath30 on @xmath11 , the _ seshadri constant _ of @xmath10 is @xmath83 given a pair of ideals @xmath84 , we say that @xmath85 _ saturates _
i if there exists @xmath86 such that @xmath87 .
equivalently , on the blow up @xmath88 of @xmath11 along @xmath89 with exceptional divisor @xmath4 , the natural inclusion @xmath90 is an isomorphism .
( this equivalence is a tautology : the zero set of the sections of @xmath91 has homogeneous ideal the saturation of @xmath92 ( @xcite exercise ii.5.10 ) ; this is all of @xmath93 if and only if the zero set is empty , if and only if the sections generate @xmath94 . ) similarly , give a line bundle @xmath30 on @xmath11 , the global sections of @xmath95 generate a subsheaf @xmath96 , defining an ideal @xmath85 .
we say that _ the global sections of @xmath97 saturate _
@xmath89 if @xmath85 saturates @xmath89 .
equivalently , the global sections of @xmath97 generate the line bundle @xmath98 on @xmath99 .
this is weaker than ( i.e. is implied by ) @xmath97 being globally generated .
fix a polynomial @xmath100 of degree @xmath101 and consider any @xmath101-dimensional proper polarised scheme @xmath0 whose hilbert polynomial equals @xmath100 : @xmath102 where @xmath103 and , for smooth @xmath11 , @xmath104 fix @xmath105 such that @xmath106 is both very ample on @xmath11 , and @xmath107 then @xmath106 embeds @xmath11 in @xmath108 , defining a point of the hilbert scheme @xmath109 of subschemes of @xmath110 with hilbert polynomial @xmath111 .
then there is a @xmath112 such that for _ all _ points @xmath113 of hilb and @xmath114 we have an exact sequence @xmath115 this ( see for example @xcite ) defines hilb as a closed subscheme of the grassmannian @xmath116 so @xmath117 and a choice of isomorphism @xmath118 give us a point @xmath119 in @xmath120 , with different choices of isomorphism corresponding ( up to scale ) to the orbits of @xmath121 on @xmath122 .
a @xmath123 acting on @xmath124 induces an action @xmath125 and so one on @xmath126 .
it is this action that commutes with the action on @xmath127 induced by that on @xmath110 .
points in the orbit of @xmath128 correspond to the projective transformations of @xmath11 . from ( [ eq : seshilb ] ) ,
the fibre over @xmath129 of the hyperplane bundle on @xmath120 is naturally isomorphic to @xmath130 @xmath0 is hilbert ( semi)stable with respect to @xmath131 if the point @xmath132 is git ( semi)stable for the action of @xmath121 , linearised on ( [ eq : line ] ) , for all @xmath133 . _ asymptotic hilbert stability _
is defined to mean hilbert stability for all sufficiently large @xmath131 . by picking a different line bundle on the hilbert scheme
( i.e. a different projective embedding of hilb the beautiful chow embedding @xcite ) we also get the notion of _ chow stability _ with respect to @xmath131 and _ asymptotic chow stability_. we need the concept of a _ test configuration _ , as defined in the foundational paper @xcite .
a _ test configuration _ for a polarised variety @xmath134 is 1 . a proper flat morphism @xmath135 , 2 . an action of @xmath72 on @xmath136 covering the usual action of @xmath72 on @xmath82 , 3 .
an equivariant very ample line bundle @xmath137 on @xmath136 , such that the fibre @xmath138 is isomorphic to @xmath134 for one , and so all , @xmath139 .
a test configuration is called a _ product configuration _ if @xmath140 , and a _ trivial configuration _ if in addition @xmath72 acts only on the second factor .
we will often need a weaker concept which we call a _ semi test configuration _ where @xmath137 is just globally generated .
in the situation of ( [ rlarge ] ) , a 1-parameter subgroup of @xmath141 is equivalent to the data of a test configuration for @xmath117 .
the action of a 1-parameter subgroup of @xmath141 on @xmath142 clearly gives a test configuration @xmath143 for @xmath117 with @xmath137 the pull back of the ( equivariant ) line bundle @xmath144 .
conversely the subgroup can be recovered from the test configuration via the induced @xmath72-action on the dual of the vector space @xmath145 . since @xmath146 is torsion - free ( by flatness ) over the curve @xmath82
it is a vector bundle , and @xmath136 sits inside the projectivisation of its dual by the very ampleness of @xmath137 .
its general fibre has dimension @xmath147 , so @xmath145 does too .
pick a trivialisation of @xmath146 over @xmath82 , identifying it with @xmath148 .
this has a diagonal @xmath72-action , inducing one on @xmath149 which yields the original test configuration ; thus these two operations are mutual inverses .
( note that in fact @xmath150 .
the map @xmath151 is just restriction ; we need to define its inverse @xmath152 .
any element of @xmath145 can be lifted to a meromorphic section @xmath153 of @xmath137 on @xmath136 that is regular on @xmath154 .
the projection of its polar locus to @xmath82 does not contain @xmath155 and so is a finite number of points in @xmath82 . choosing a polynomial @xmath62 with high order zeros at these points such that @xmath156
, @xmath157 is a holomorphic section with the same class as @xmath153 in @xmath145 since @xmath158 .
then @xmath159 $ ] defines the required class in @xmath160 . )
denote the weight of the induced @xmath72-action on @xmath161 by @xmath162 .
( we enumerate by @xmath163 since @xmath164 is a test configuration for @xmath165 .
the confused reader may set @xmath166 temporarily . ) for @xmath167 is @xmath168 and @xmath169 is a polynomial of degree @xmath170 in @xmath171 by the equivariant riemann - roch theorem .
( it is important that we do _ not _ modify @xmath172 to be this polynomial for small @xmath173 , so for instance @xmath174 really is the weight on @xmath145 . ) to make the @xmath72-action special linear on @xmath145 we first pull back the family by the cover @xmath175 , making the action of weight @xmath176 .
composing with the trivial action which scales the @xmath137-fibres with weight @xmath177 scales @xmath178 with weight @xmath179 , cancelling out the previous action .
( the extra factor of @xmath131 is to make the formula ( [ normalised ] ) nicer , and is natural if @xmath180 is the @xmath131th twist of a test configuration for @xmath0 . ) since this new action acts with zero total weight on @xmath181 , it acts on the line @xmath182 ( compare ( [ eq : line ] ) which was for @xmath117 with no higher cohomology of @xmath106 ) with _ normalised weight _ @xmath183 the normalised weight is a polynomial @xmath184 of degree @xmath170 in @xmath173 for @xmath185 , with coefficients which are also polynomial of degree @xmath170 in @xmath131 for @xmath186 : @xmath187 for @xmath186 .
the normalisation means that the coefficient of @xmath188 vanishes : @xmath189 .
the hilbert - mumford criterion relates this weight to stability as follows .
[ thm : instability ] a polarised variety @xmath0 is _ stable _ if and only if @xmath190 where @xmath191 and @xmath192 mean the following for the different notions of stability : * hilbert stable with respect to @xmath131 : for any nontrivial test configuration for @xmath193 for all @xmath185 , * asymptotically hilbert stable : for all @xmath186 , any nontrivial test configuration for @xmath117 has @xmath194 for all @xmath185 , * chow stable with respect to @xmath131 : for any nontrivial test configuration for @xmath117 , the leading @xmath195 coefficient @xmath196 of @xmath197 is positive : @xmath198 , * asymptotically chow stable : for all @xmath186 , any nontrivial test configuration for @xmath117 has @xmath198 , * k - stable : for all @xmath186 , any nontrivial test configuration for @xmath117 has leading coefficient @xmath199 of @xmath196 ( the _ donaldson - futaki invariant _ of the test configuration ) positive : @xmath200 .
furthermore the result holds if we replace stable " with semistable " and strict inequalities with non strict inequalities throughout .
finally @xmath0 is _ polystable _ if it is semistable and any destabilising test configuration ( i.e. one which is not strictly stable ) is a product configuration .
the increasing number of test configurations that have to be tested in the second , fourth and fifth definitions as @xmath201 currently prevent us from adding k - stability to the left of the following consequences of theorem [ thm : instability ] .
+ asymptotically chow stable @xmath54 asymptotically hilbert stable @xmath54 asymptotically hilbert semistable @xmath54 asymptotically chow semistable @xmath54 k - semistable .
+ the relation between k - stability and asymptotic chow stability is analogous to the relation between slope stability and gieseker stability for vector bundles , and a geometric criterion for asymptotic chow stability would show it was implied by k - stability ; we only have a necessary condition ( theorem [ chow ] ) .
the fact that chow stability is controlled by the coefficient @xmath196 is due to mumford @xcite .
the definition of k - stability above is due to donaldson , adapting tian s original differential - geometric definition to allow nonnormal central fibres @xmath154 ( though what is called properly semistable in @xcite and stable in @xcite is what we call k - polystable ) .
tian @xcite defines a line bundle ( the cm polarisation " ) on @xmath202 such that k - stability is exactly stability in the sense of the hilbert - mumford criterion for this line bundle @xcite .
k - stability is probably not a bona fide git notion : tian s polarisation may not be ample , and the number of test configurations increases as @xmath131 tends to infinity .
however it is k - polystability that is conjecturally related to the existence of constant scalar curvature khler metrics ; we apply our methods to these in @xcite .
the definition of polystability says that any destabilising test configuration comes from a @xmath72-action on @xmath11 with the appropriate weight 0 ( i.e.donaldson-futaki invariant 0 in the k - polystability case ) .
this corresponds to an orbit in hilb which is closed in the semistable points but with possibly higher dimensional stabilisers ; equivalently the orbit in the total space of the dual of the polarising line bundle over hilb is closed .
there seems to be no universally accepted name for this ; we use polystability by analogy with bundles .
a test configuration @xmath143 for @xmath117 can be twisted to get another , @xmath164 , for @xmath165 ; for @xmath133 this will have no higher cohomology . since the definition of k - stability is unchanged if @xmath30 is replaced by some power , for this notion we can allow @xmath137 to be an ample @xmath203-line bundle in the definition of a test configuration .
letting @xmath204 denote the donaldson - futaki invariant and writing the unnormalised weight @xmath172 as @xmath205 , we see that @xmath206 and @xmath207 is the coefficient of @xmath208 in the expansion of @xmath209 . when the central fibre @xmath154 is smooth , @xmath210 , where @xmath211 is the usual calabi - futaki invariant of @xmath212 and the vector field generated by the @xmath213-action @xcite .
+ without loss of generality we now take @xmath166 .
an arbitrary test configuration @xmath214 for @xmath0 is , by definition , @xmath72-isomorphic to the trivial test configuration @xmath215 away from the central fibre .
it is therefore @xmath72-birational to @xmath215 and so is dominated by a blow up @xmath143 of @xmath8 in a @xmath72-invariant ideal @xmath89 supported on ( a thickening of ) the central fibre @xmath9 : @xmath216 here we use the canonical @xmath72-action on @xmath217 inherited from that on @xmath89 , and its linearisation on the line bundle @xmath218 , where @xmath4 denotes the exceptional divisor .
@xmath219 and the horizontal arrow in ( [ any ] ) is an equivariant map of equivariant polarisations ( although @xmath220 may be only semi - ample ) , whereas @xmath62 does not respect the polarisation .
mumford ( @xcite section 3 of the proof of theorem 2.9 ) essentially shows that any test configuration is of this form , where @xmath221 is of the form @xmath222.\ ] ] here the ideals @xmath223 correspond to subschemes @xmath224 of @xmath0 , so @xmath225 is @xmath72-invariant and corresponds to a subscheme of @xmath8 supported on ( the @xmath131-fold thickening of ) the central fibre @xmath9 .
this can be proved by writing the @xmath72-invariant ideal @xmath89 as a sum of weight spaces : defining @xmath226 in terms of the weight-@xmath58 piece @xmath227 we get the weight space decomposition ( [ ideals ] ) . or
, embedding a test configuration in some @xmath228 , mumford s result applies directly .
for example , given a @xmath72-action on @xmath11 with repulsive fixed point set " @xmath10 ( that part of the fixed point set with negative weight spaces in its normal cone ) , there is a blow up of @xmath8 supported on @xmath229 in which the proper transform of @xmath9 can be blown down to give back @xmath8 but with a nontrivial @xmath72-action . since @xmath143 is a _ semi _ test configuration ( and because later we will want to replace an ample @xmath0 with a semi - ample @xmath230 for some resolution of singularities @xmath62 )
we will prove many results in the generality of semi test configurations .
general test configurations ( [ any ] ) will be studied in section [ testconfigs ] ; next we look at the simplest case ( beyond product configurations ) of @xmath231 in ( [ any ] ) : the _ deformation of @xmath11 to the normal cone of @xmath10_.
given any proper subscheme @xmath3 we get a canonical test configuration @xmath136 , the blow up of @xmath8 along @xmath229 with exceptional divisor @xmath232 .
this deformation to the normal cone has central fibre @xmath233 , where @xmath234 is the blow up of @xmath11 along @xmath10 with exceptional divisor @xmath4 . when @xmath10 and @xmath11 are both smooth @xmath235 is the projective completion of the normal bundle @xmath211 of @xmath10 in @xmath11 , glued along @xmath236 to the blow up of @xmath11 .
for more on the deformation to the normal cone see @xcite ; for a diagram see section [ testconfigs ] .
let @xmath237 be the composite of the projections @xmath238 .
for @xmath30 an ample line bundle on @xmath11 and @xmath35 a positive rational number let @xmath239 be the @xmath240-line bundle @xmath241 .
this restricts to @xmath30 on the general fibre of @xmath242 , and is ample for @xmath35 sufficiently small .
[ ample ] fix @xmath30 an ample ( respectively semi - ample ) line bundle on @xmath11 . for @xmath243 ( [ def : seshadri ] ) , the @xmath203-line bundle @xmath239 is ample ( semi - ample ) on @xmath136 . if @xmath244 then @xmath245 is nef .
if in addition the global sections of @xmath246 saturate ( see section [ notation ] ) for @xmath185 then @xmath245 is semi - ample .
note that if @xmath247 and @xmath68 is globally generated and the sections of @xmath248 saturate , then @xmath249 is globally generated on @xmath8 by the sections @xmath250 .
algebraically this is the statement that on @xmath8 , @xmath251 saturates @xmath252 .
geometrically it says that the global sections saturating @xmath248 generate @xmath253 on @xmath136 away from the zero section @xmath254 of the cone @xmath255 , over which the sections @xmath256 generate .
putting @xmath257 now gives the third claim .
for the first two we claim that we may assume that @xmath30 is ample by replacing @xmath0 by @xmath258 if necessary . for @xmath173
sufficiently large that @xmath259 is globally generated , @xmath260 generates a subsheaf @xmath261 and so a subscheme @xmath262 such the pullback of @xmath263 to @xmath11 is @xmath29 .
thus it is sufficient to prove the result for @xmath264 and then pullback to get the result for @xmath265 . on @xmath234 , @xmath266 is in the ample cone for small @xmath35 and on its boundary for @xmath267 , so by kleiman @xcite @xmath266 is ample for rational @xmath268 .
so @xmath269 is globally generated for @xmath185 ; equivalently the global sections of @xmath248 saturate @xmath270 ( for @xmath173 sufficiently large that the pushdown of @xmath271 to @xmath11 is @xmath270 ) .
@xmath68 is also globally generated so again this implies that @xmath249 is globally generated .
thus @xmath239 is nef for @xmath272 , but it is ample for small @xmath35 since @xmath273 is ample .
thus by kleiman again , @xmath239 is actually ample and @xmath245 is nef .
the obvious @xmath72-action on @xmath8 ( acting trivially on the @xmath11 factor ) lifts to an action on the deformation to the normal cone @xmath274 which , on the central fibre @xmath275 , is trivial on @xmath234 .
when @xmath10 and @xmath11 are both smooth , @xmath276 acts on @xmath277 as @xmath278 .
[ degen ] fix @xmath30 ample and @xmath243 .
then @xmath274 is a flat family of polarised schemes , and , for all @xmath279 , the total weight of the induced action on @xmath280 is @xmath281 from the definition of the blow up of a subscheme , @xmath282 , where @xmath283{\mathcal o}_x\ \subset\ { \mathbb c}[t]\otimes{\mathcal o}_x . \label{s}\end{aligned}\ ] ] it follows that for all @xmath185 , the pushdown of @xmath284 to @xmath8 is @xmath285{\mathcal o}_x$ ] , with the higher pushdowns zero . by the ampleness of @xmath286 ( [ ample ] ) , for @xmath185 we have the vanishing of @xmath287h^i_x(l^k),\ ] ] for @xmath86 . in particular then , @xmath288 now @xmath289 where by ( [ s ] ) , @xmath290 replacing @xmath29 by @xmath291 does not change the blow up ( just the corresponding exceptional line bundle ) so that , for @xmath185 , @xmath292 the vanishing ( [ vanish ] ) of @xmath293 for @xmath294 means that the dimension of this is @xmath295 which is @xmath296 . by ( @xcite theorem iii.9.9 ) this proves flatness of the family .
now ( [ wspace ] ) is the weight space decomposition with respect to the @xmath72-action : @xmath72 acts on @xmath81 with weight @xmath297 and trivially on @xmath11 and so on @xmath29 .
therefore @xmath298 where the second and last equalities follow from the vanishing ( [ vanish ] ) of @xmath299 for @xmath294 and @xmath86 .
we may only take @xmath267 if the global sections of @xmath246 saturate for @xmath185 . in this case
@xmath239 is semi - ample , pulled back from a contraction @xmath62 from @xmath300 to @xmath301 , which we call @xmath302 . by construction
, @xmath303 is then an isomorphism . by lemma 2.13 of @xcite
, @xmath304 is also a flat family , so @xmath305 is a test configuration for @xmath185 .
[ ss ] let @xmath10 be a proper subscheme of an irreducible polarised algebraic variety @xmath0 , and suppose that @xmath306 and the global sections of @xmath246 saturate for @xmath307 .
letting @xmath214 be the contraction of @xmath274 constructed above , the weight of the action on @xmath308 is @xmath309 by ( [ s ] ) , for @xmath185 , @xmath310h^0_x(l^k),\ ] ] yielding @xmath311 by the isomorphism @xmath303 this is the weight space decomposition of @xmath312 , which by flatness of @xmath304 and ampleness of @xmath313 is @xmath314 .
so the total weight is @xmath315 just as in ( [ noco ] ) this is @xmath316 .
then replacing @xmath317 by @xmath318 gives an error bounded by @xmath319 , where @xmath320 .
so the result follows from lemma [ error ] below . [ error ] fix @xmath265 and @xmath243 , or
@xmath31 \cap{\mathbb q}$ ]
if @xmath244 and the global sections of @xmath321 saturate for @xmath185 . then @xmath322 fix @xmath323 . then @xmath324 is ample on @xmath136 , while @xmath239 is generated by its global sections and so nef . so we can apply fujita vanishing ( @xcite theorem 1.4.35 ) to give @xmath325 such that @xmath326 has no higher cohomology for any @xmath327 .
for @xmath328 , setting @xmath329 shows that @xmath330 has no higher cohomology .
so for @xmath185 we have @xmath331h^i_x(l^k ) , \label{dec}\end{aligned}\ ] ] for @xmath86 .
that is , @xmath332 for @xmath333 .
so the sum becomes @xmath334 where we have taken @xmath173 sufficiently large that the pushdown of @xmath335 is @xmath336 ( and higher pushdowns are zero ) for @xmath337 .
a corollary of fujita vanishing is that @xmath338 for any coherent sheaf @xmath339 and nef divisor @xmath340 ( @xcite theorem 1.4.40 ) .
applying this on @xmath234 in turn to @xmath341 and @xmath342 shows that each @xmath343 . since @xmath344 is fixed , then , we get a similar bound on the whole sum .
[ weight ] for @xmath31\cap{\mathbb q}$ ] define @xmath345 which , for @xmath131 sufficiently large _ for fixed @xmath3 , _ is @xmath346 for @xmath347 , @xmath106 globally generated and @xmath348 saturated by its sections , @xmath117 is unstable if @xmath349 , where @xmath18 is defined as in theorem [ thm : instability ] , depending on the type of stability .
we say that @xmath117 is destabilised by @xmath10 .
similarly for strict instability and @xmath350 .
the conditions on @xmath131 are just to ensure that @xmath351 ( or @xmath352 if @xmath257 ) is a genuine test configuration ; for k - stability we are free to twist by higher @xmath131 and use proposition [ ample ] to remove these conditions .
the result is a direct consequence of theorems [ degen ] and [ ss ] and the identity @xmath353 ( [ normalised ] ) .
recall from the introduction the definition of the @xmath37 and @xmath27 ( [ hs ] ) .
choose @xmath62 so that @xmath354 for @xmath64 , then for @xmath355 , @xmath356 but by riemann - roch , @xmath357 is a polynomial of total degree @xmath101 . writing @xmath358 where @xmath359 is homogeneous of degree @xmath360 is therefore a degree @xmath361 polynomial in @xmath25 which can be extended to all real @xmath25 .
[ integrals ] for @xmath11 irreducible the weights @xmath362 of theorems [ ss ] and [ degen ] can be written @xmath363 using ( [ definitionofa_i ] ) for @xmath364 we split up @xmath365 as @xmath366 using the fact that @xmath57 has dimension @xmath367 ( since @xmath11 is irreducible ) and @xmath368 is a polynomial with @xmath369 . for a smooth function @xmath370 , as @xmath371 with @xmath372 ,
the trapezium rule gives @xcite @xmath373 this can be proved by taylor s theorem , or directly for polynomials by noting that for @xmath374 we have the identity @xmath375 applying this to @xmath376 , ( [ prox ] ) now approximates @xmath365 by @xmath377 since @xmath378 .
subtracting @xmath379 gives the result .
note that by riemann - roch on @xmath234 , @xmath380 for @xmath381 by the ampleness of @xmath382 , so for @xmath31 $ ] , @xmath383 .
therefore we can define the _ slope _ ( or k - slope ) of @xmath29 ( [ def : slope ] ) by @xmath384 and that of @xmath11 ( [ def : x ] ) , @xmath385 this gives the following obvious definition of k - slope stability , which , like k - stability , is independent of replacing @xmath30 by @xmath106 ( on replacing @xmath386 and @xmath387 by @xmath388 and @xmath389 ) .
[ defstab ] we say that @xmath0 is slope ( semi / poly)stable if for every proper subscheme @xmath10 of @xmath11 , the following holds : * * _ slope semistability _ * : @xmath390 for all @xmath31 $ ] , * * _ slope stability _ * : @xmath391 for all @xmath272 , and also for @xmath267 if @xmath244 and global sections of @xmath246 saturate for @xmath185 , * * _ slope polystability _ * : @xmath392 is slope semistable , and if @xmath393 is any pair such that @xmath394 , then @xmath267 is rational and , on the deformation to the normal cone of @xmath10 , @xmath239 is pulled back from a product test configuration @xmath395 .
[ thm : kstableslopestable ] if a polarised variety @xmath0 is k - stable then it is slope stable .
if it is k - polystable ( respectively k - semistable ) then it is slope polystable ( semistable ) . from the deformation to the normal cone of @xmath10
we get test configurations @xmath396 ( for @xmath397 ) , or @xmath398 ( for @xmath267 if @xmath244 and the global sections of @xmath246 saturate ) .
its total weight @xmath172 is given by theorems [ degen ] and [ ss ] respectively .
writing @xmath399 , its donaldson - futaki invariant ( [ futaki ] ) is , by proposition [ integrals ] , @xmath400 this is a strictly positive multiple of @xmath401 , and k - stability ( k - semistability ) implies it is strictly positive ( nonnegative ) .
this gives the result so long as , in the semistable / polystable case , the test configuration is not trivial .
but the central fibres of both @xmath274 and @xmath214 have nontrivial @xmath72-actions .
this result means that slope instability provides an obstruction to the existence of constant scalar curvature khler metrics ( and so also khler - einstein metrics ) ; see @xcite where there are also numerous examples .
letting @xmath402 be the coefficients of @xmath403 , we define the slope of @xmath404 ( in slightly misleading notation ) to be @xmath405 notice that we can rephrase slope stability in the equivalent ways @xmath406 due to the implications @xmath407 for @xmath408 , on setting @xmath409 and @xmath410 ( so @xmath411 ) . [ normal ] on the blow up @xmath234 we have the formula @xmath412 and so @xmath369 , since the intersection @xmath413 can be calculated equally on @xmath11 or @xmath234 .
the @xmath414 are the coefficients of the polynomial @xmath415 _ if @xmath11 is normal _ , then @xmath416 ( @xcite proof of corollary iii.11.4 ) .
@xmath4 has dimension @xmath417 and @xmath418 is supported on points over which the fibre has dimension @xmath419 , so the support of @xmath420 has codimension at least @xmath421 .
hence @xmath422 , and @xmath423 so @xmath424 also .
( the same argument shows that if @xmath10 has dimension @xmath58 , then @xmath425 for @xmath426 ( and @xmath427 for @xmath11 normal ) . if @xmath3 are both smooth in a neighbourhood of @xmath10 , then @xmath428 so @xmath425 for all @xmath429 . ) since @xmath430 for @xmath431 , @xmath368 is a decreasing function in @xmath25 : @xmath432 .
in fact from ( [ a0 ] ) , @xmath433 by the ampleness of @xmath382 .
in particular @xmath434 for @xmath381 , showing that @xmath435 ( [ qslope ] ) is finite . for @xmath11
normal then , @xmath436 is strictly less than @xmath437 for small @xmath35 , and the slope inequality is automatically satisfied . in all of the examples we have considered
@xcite , one need only test the slope inequality at @xmath267 .
if this held in general it would simplify the definition of stability .
szkelyhidi @xcite has shown by example that for the modification of k - stability relevant to extremal metrics this is not the case . [
simplify ] suppose that @xmath3 is a strictly destabilising subscheme in the sense that it violates the slope inequality ( [ defstab ] ) . then at least one of the connected components of @xmath10 strictly destabilises .
similarly if @xmath10 is a thickening @xmath438 of @xmath439 then @xmath440 strictly destabilises .
suppose @xmath441 with @xmath442 and @xmath443 disjoint
. then @xmath444 ( [ qslope ] ) since @xmath445 .
suppose @xmath10 is strictly destabilises .
then there is a @xmath31\cap{\mathbb q}$ ] such that @xmath446 this implies that for some @xmath447 , @xmath448 , and by lemma [ seshadridisconnected ] @xmath449 , so @xmath450 is strictly destabilising . finally if @xmath451 destabilises then so does @xmath452 since @xmath453 and @xmath454 as @xmath455 for @xmath456 ( [ a0 ] )
. [ seshadridisconnected ] if @xmath457 then @xmath458 .
let @xmath459 be the blowup of @xmath11 along @xmath460 with exceptional divisor @xmath461 , where @xmath462 is the subset of @xmath4 sitting over @xmath463 .
let @xmath464 , so by definition @xmath465 is nef . if @xmath466 is an irreducible curve contained in @xmath467 then @xmath468 by ampleness of @xmath30 and the fact that @xmath469 .
if @xmath466 is not contained in @xmath467 then @xmath470 . hence by the kleiman criterion , @xmath471 is nef .
but this line bundle is the pullback of @xmath472 from @xmath473 , so the latter line bundle is also nef ( @xcite example 1.4.4(ii ) ) , proving that @xmath474 . [ pts ]
let @xmath0 be a smooth polarised variety of dimension @xmath101 and @xmath475 be the seshadri constant of some point @xmath62 in @xmath11 .
then @xmath62 strictly destabilises if and only if @xmath476 we have @xmath477 since @xmath478 can be represented by a cycle on @xmath11 missing @xmath62 , so by a cycle on @xmath234 missing @xmath4 .
similarly using @xmath479 , @xmath480 yielding ( [ qslope ] ) , @xmath481 so @xmath62 strictly destabilises if and only if @xmath482 . as this is linear in @xmath35 it holds for some @xmath483 if and only if it holds for @xmath257 . substituting @xmath484 and @xmath485 gives the result .
one can of course also calculate the slope of a smooth point more directly by working locally with @xmath486 .
we now get [ pts2 ] if @xmath11 is smooth then no point strictly destabilises . for any line bundle @xmath487
we say that @xmath488 _ generates @xmath153-jets at @xmath62 _ if @xmath489 is a surjection , where @xmath490 denotes the maximal ideal at @xmath62 .
we define @xmath491 to be the maximum integer @xmath153 such that @xmath488 generates @xmath153-jets at @xmath62 ( and set @xmath492 if no such @xmath153 exists ) .
demailly ( @xcite lemma 8.6 ) proves that given any @xmath493 , if @xmath494 generates @xmath495-jets at @xmath62 , then @xmath496 generates @xmath153-jets at @xmath62 . setting @xmath497 ( and noting that @xmath498 ) we see that for @xmath499 and any @xmath500
, @xmath501 applying this again to @xmath502 ( with @xmath503 ) implies that @xmath504 and so inductively @xmath505 for @xmath506 . setting @xmath507 , @xmath508
fix any rational number @xmath509 , so that @xmath510
. then by the identity ( @xcite lemma 7.6 ) @xmath511 applied to @xmath512 , we can find a @xmath185 such that @xmath513 .
thus by ( [ generationofadjointbundles ] ) , @xmath514 has a section ( not vanishing at @xmath62 ) , and so @xmath515 .
therefore @xmath516 .
if @xmath517 we may as well assume that @xmath30 is the hyperplane bundle as slope stability is invariant under rescaling .
so @xmath518 for any point @xmath62 and @xmath519 , which gives equality in ( [ pts ] ) , so @xmath520 is at best k - semistable ( since @xmath521 is generated by global sections ) .
in fact @xmath520 is k - polystable in the sense of definition [ defstab ] , and the deformation to the normal cone of @xmath522 has semi - ample @xmath239 for @xmath523 , pulled back from a blow down to @xmath524 .
so the degeneration is a product family with a nontrivial @xmath72-action , and the semistability we are seeing comes from a @xmath72-action on @xmath520 with donaldson - futaki invariant zero .
if instead of k - instability we are interested in asymptotic stability , then the relevant slope function is more complicated .
let @xmath0 be a polarised manifold , @xmath131 be a positive integer , and @xmath10 a subscheme of @xmath11 such that on the blow up @xmath88 of @xmath11 along @xmath10 , @xmath525 ( e.g. if @xmath3 are both smooth in a neighbourhood of @xmath10 . )
then for _ all _ @xmath173 and @xmath25 ( with @xmath526 ) we have the exact formula @xmath527 letting @xmath528 denote the bernoulli numbers , define @xmath529 and @xmath530 for @xmath531 .
then we define the _ asymptotic chow slope _ @xmath532 of @xmath29 for @xmath533 to be @xmath534 where @xmath535 i.e. @xmath536 , @xmath537 defining the asymptotic chow slope of @xmath11 to be the slope of the empty subscheme , @xmath538 we say ( with great difficulty ) that @xmath11 is asymptotically chow slope strictly destabilised by @xmath10 if , for all @xmath186 , @xmath539 [ chow ] if a polarised variety is asymptotically chow slope strictly destabilised by @xmath10 satisfying ( [ condition ] ) then it is asymptotically chow strictly unstable . the proof is almost the same as for theorem [ thm : kstableslopestable ] .
we calculate the coefficient @xmath196 of @xmath195 in @xmath197 ( [ normalised ] ) , @xmath540 ( instead of its @xmath541 coefficient the donaldson - futaki invariant ) for the deformation to the normal cone . by theorem [ thm : instability ] , if @xmath542 then @xmath0 is chow unstable with respect to @xmath131 . by the continuity of @xmath543 we may take @xmath268 , so that @xmath174 is calculated by theorem [ degen ] and @xmath544 by proposition [ integrals ] , giving @xmath545 instead of the trapezium rule ( [ easyeuler ] ) we use the fact @xcite that for any polynomial @xmath370 , @xmath546 ( the case @xmath374 follows from the definition of @xmath547 , implying the general case by linearity . ) the theorem follows by applying this to the polynomial @xmath548 .
we begin with an important technical result allowing us to calculate weights of a test configuration @xmath398 in terms of a semi test configuration @xmath549 that dominates it .
so let @xmath214 be an equivariantly polarised flat @xmath72-family with general fibre @xmath0 , and fix another flat @xmath72-family @xmath550 with a birational @xmath72-equivariant map @xmath551 [ stein ] if @xmath11 is normal then there exists an @xmath552 such that @xmath553 @xmath62 factors through the its stein factorisation ( @xcite corollary iii.11.5 ) as @xmath554 @xmath555 induces an equivariant isomorphism between @xmath556 and @xmath557 for @xmath185 , and @xmath558 is flat over @xmath82 since @xmath556 has no @xmath81-torsion by the flatness of @xmath136 . thus replacing @xmath136 by @xmath558 and @xmath62 by @xmath559 if necessary
, we may assume that @xmath62 is finite and so @xmath560 is ample .
moreover , the general fibre of @xmath558 is @xmath0 by the normality of @xmath11 , so @xmath549 is now a genuine test configuration for @xmath561 .
so we have a @xmath72-equivariant morphism of polarised families @xmath562 , and we wish to relate the total weights of the @xmath72-actions on @xmath314 and @xmath563 ; but the latter is now isomorphic to @xmath564 for @xmath185 by ampleness and flatness .
@xmath565 induces an exact sequence @xmath566 for some cokernel @xmath567 supported on @xmath568 ( since @xmath62 is an isomorphism away from the central fibres ) .
so @xmath569 for some @xmath493 .
we first give the argument for @xmath570 to illustrate the more technical general case .
tensoring with @xmath571 and pushing down to @xmath82 gives an exact sequence on @xmath82 , @xmath572 where @xmath573 satisfies @xmath574 .
therefore its restriction to @xmath575 is also isomorphic to @xmath576 , and tor@xmath577 , giving the exact diagram @xmath578 where the vertical arrows are restriction to @xmath575 , and the flatness of @xmath579 and ampleness of @xmath580 give the two central terms and ensure that tor@xmath581 .
these are maps of @xmath72-modules , except that the left hand @xmath576 has weight shifted by @xmath297 ( i.e. is isomorphic to @xmath582 as a @xmath72-module ) .
we see this by exhibiting an explicit weight-@xmath583 isomorphism @xmath584 from @xmath576 to the kernel of the lower @xmath565 .
given @xmath585 , choose a lift @xmath586 .
@xmath587 has zero image in @xmath576 so is in the image of some @xmath588 .
@xmath589 maps under @xmath565 to @xmath590 , i.e. zero , so is an element @xmath591tor@xmath592 . since this map involved multiplication by @xmath81 it has weight @xmath297 .
so in this case of @xmath570 , @xmath593 , and @xmath594 , where @xmath595rank@xmath596 since @xmath576 is supported on the @xmath101-dimensional central fibre . for general @xmath153
we still have the exact sequence @xmath597 where again the first term @xmath598 is isomorphic to the last @xmath599 .
for instance if we pick an isomorphism @xmath600/(t^{j_i})$ ] , each @xmath601 , then @xmath602/(t)$ ] maps isomorphically via @xmath603 to @xmath604 ) /(t^{j_i})=\ker\big(q_k{\stackrel{\times t\,}{\longrightarrow}}q_k\big)$ ] .
more invariantly , filter @xmath605 by @xmath606 , and so @xmath599 by @xmath607 with graded pieces @xmath608 .
then there exists a weight-@xmath609 multiplication by @xmath610 " map taking @xmath611 into @xmath314 whose sum over @xmath612 gives an isomorphism to tor@xmath613 .
the map is defined roughly as before .
given @xmath614 , lift to @xmath615 and then to @xmath586 .
the image of @xmath616 in @xmath576 is zero by construction , so is in the image of @xmath588 .
@xmath617 maps to @xmath618 , i.e. to zero since @xmath619 .
thus @xmath620 for some @xmath621 . @xmath622 and @xmath623 clearly map to zero via this construction , so @xmath584 is well defined .
the upshot is that @xmath624 @xmath625 is a polynomial in @xmath173 ( since both weights @xmath626 are ) , so is @xmath627 for some @xmath552 because @xmath628 as before .
[ assumenormal ] in particular if @xmath11 is normal then for k - stability we need only consider _ normal _ test configurations .
this is because any test configuration @xmath214 is dominated by its normalisation , which also has general fibre @xmath11 if @xmath11 is normal .
the pullback of @xmath313 is ample , so some twist gives another test configuration which is _ less stable _ than @xmath214 , in the sense that it has the same weight to leading order and a smaller donaldson - futaki invariant , by proposition [ stein ] above .
similarly for chow stability we may compute the chow weight of any test configuration on its normalisation .
of course most of our test configurations have nonnormal central fibre , however . given any test configuration @xmath214 ( [ any ] )
we now build inductively the semi test configuration @xmath629 ( [ ideals ] ) that dominates it , starting from the deformation to the normal cone @xmath630 of @xmath631 .
so let @xmath632 denote @xmath633 , i.e. the blow up in @xmath634 : @xmath635 recall that the central fibre of @xmath636 is @xmath637 , where @xmath99 is the blow up of @xmath11 along @xmath631 with exceptional divisor @xmath638 , and @xmath232 is the exceptional divisor of @xmath639 .
@xmath232 is a projective cone over @xmath631 ( proj of the graded algebra over @xmath631 with @xmath173th graded piece @xmath640 ) the projective completion of the normal cone to @xmath641 .
its zero section @xmath642 is a copy of @xmath631 which fits into a flat family with the @xmath643 in each fibre @xmath644 , which we see as follows .
the proper transform @xmath645 is defined by the graded sheaf of ideals generated by @xmath646 in the graded sheaf of algebras @xmath647 .
that is , @xmath648 is generated by the sections of @xmath649 .
it is abstractly isomorphic to the blow up of @xmath650 along its intersection with @xmath634 , but this is a divisor in @xmath650 , so @xmath651 .
the central fibre @xmath652 is defined by the graded sheaf of ideals generated by @xmath653 .
similarly the proper transform of @xmath654 is the blow up of @xmath654 along its intersection @xmath655 with @xmath634 ; that is @xmath656 .
it is defined by the graded sheaf of ideals generated by @xmath657 , with central fibre @xmath658 isomorphic to @xmath442 and defined by @xmath659 we now form @xmath660 by blowing up @xmath636 in @xmath661 .
since @xmath662 is just @xmath663 ( [ ideals ] ) , we have basically shown that @xmath660 dominates @xmath664 .
precisely , we have maps @xmath665 , and set @xmath666 to be the exceptional divisor of @xmath667 , and @xmath4 to be the exceptional divisor of @xmath664 . while @xmath668 is relatively ample for @xmath669
, it is only semi - ample for @xmath670 : [ p ] @xmath671 factors through a map @xmath672 . under this map , @xmath673 . for @xmath173 sufficiently large , @xmath674 , in which the ideal @xmath675 defines the @xmath173th power of the ideal of @xmath661 ( [ i1 ] ) .
that is , @xmath676 .
therefore the sections @xmath677 generate @xmath678 and so define a regular map from @xmath660 to @xmath679 , i.e. to @xmath680 in fact the contraction @xmath681 just collapses the restriction @xmath682 of the cone @xmath683 down to @xmath442 , but we will not need this . similarly in @xmath660
there is a copy @xmath684 of @xmath442 , sitting in a flat family with @xmath685 as the central fibre of the proper transform @xmath686 . in the coordinate ring @xmath687
( we are recycling the symbol @xmath688 and shall do so again below ) pulled back from @xmath136 by the above map @xmath681 ( [ p ] ) , @xmath689 is generated by @xmath690 , since this is the largest ideal that localises to @xmath691 when we invert @xmath81 and work on @xmath692 .
thus @xmath693 is generated by @xmath694 .
so there is also a @xmath695 , isomorphic to @xmath443 , inside it .
@xmath696 has ideal generated by @xmath697 , since this is the largest ideal that localises to @xmath698 when we invert @xmath81 and work on @xmath692 .
thus its central fibre @xmath699 is defined by the ideal @xmath700 blowing it up gives @xmath701 , with exceptional divisor @xmath702 ( and we denote the pullbacks to @xmath701 of @xmath703 by the same notation ) .
then , just as in proposition [ p ] , the pushdown of @xmath704 to @xmath8 is generated by @xmath705 by ( [ i3 ] ) , so @xmath706 factors through @xmath707 .
inductively we obtain @xmath708 as the blow up @xmath709 along @xmath710 , the central fibre of the proper transform of @xmath711 .
the coordinate ring of @xmath712 over @xmath8 has @xmath173th graded piece @xmath713 , and the ideal of the proper transform of @xmath714 is generated by @xmath715 , since this is the largest ideal that localises to @xmath716 when we invert @xmath81 and work on @xmath692 .
therefore the ideal of its central fibre @xmath717 is generated by @xmath718 thus the pushdown of @xmath719 contains @xmath720 , giving the following , as in proposition [ p ] .
[ pr ] @xmath708 factors through a map @xmath721 , where @xmath722 ( [ ideals ] ) . under this map , @xmath723 .
@xmath724 in turn any test configuration @xmath214 is dominated by a map @xmath725 from some @xmath726 ( [ any ] ) , giving @xmath727 .
denote by @xmath728 the semi - ample line bundle @xmath729 on @xmath730 , so that @xmath731
. then by the above and proposition [ stein ] we have [ ref ] the total weight on ( the @xmath173th twist of ) an arbitrary test configuration @xmath214 can be calculated on @xmath732 by @xmath733 [ seshadriir ] the seshadri constant @xmath734 of the ideal @xmath225 ( [ ideals ] ) is @xmath735 .
if @xmath736 then for @xmath185 , the sections @xmath737 saturate @xmath738 ( [ ideals ] ) .
therefore the seshadri constant of @xmath225 ( which is @xmath55 by the semi ampleness of @xmath220 in ( [ any ] ) ) is at least @xmath739 .
we prove the opposite inequality inductively .
the induction begins for @xmath740 by proposition [ ample ] ; suppose it is true for @xmath131 .
that is , @xmath382 is nef on @xmath741 precisely when @xmath742 $ ] .
equivalently , pulling back by @xmath743 , @xmath744 is nef on @xmath730 if and only if @xmath745 $ ] .
pick @xmath35 such that @xmath746 is nef on @xmath747 ; we must show that @xmath748 and @xmath749 . to show the former , we claim that since @xmath746 is nef on @xmath747 , @xmath750 is nef on @xmath730 .
given any irreducible proper curve @xmath751 not entirely contained in @xmath752 , let @xmath753 denote its proper transform in @xmath747 .
then @xmath754.c\ge[l - c(e_1+\ldots+e_{r+1})].\overline c\ge0 $ ] since @xmath755 . on the other hand
if @xmath756 , then there is an isomorphic copy @xmath757 such that @xmath754.c=[l - c(e_1+\ldots+e_{r+1})].c'\ge0 $ ] .
so indeed @xmath750 is nef on @xmath730 and so @xmath748 by induction .
secondly , fix @xmath35 such that @xmath750 is nef on @xmath730 .
@xmath758 lies in the central fibre @xmath759 , fitting into a flat family with @xmath760 away from the central fibre .
seshadri constants are lower semicontinuous in polarised families , so the seshadri constant ( with respect to @xmath750 ) of @xmath752 _ inside the central fibre @xmath759 _ is at most @xmath761 ( with respect to @xmath30 , since this is @xmath750 restricted to a general fibre ) .
the seshadri constant of @xmath752 inside the whole of @xmath730 can only be smaller still ; therefore if @xmath762 is nef then @xmath749 .
we could contract each @xmath763 using @xmath764 ; this would give an isomorphism in a neighbourhood of @xmath765 by ( @xcite proposition ii.7.3 ) since @xmath766 is ample _ and _ @xmath767 is globally generated ( by sections of @xmath768 on @xmath11 ) .
thus we could proceed inductively with these contracted @xmath763s with _ ample _ line bundles on them , but since we can not currently seem to get significantly better estimates by working with ample bundles we proceed with the semi - ample @xmath769 and contract at the last , @xmath131th stage . by corollary [ ref ]
we lose nothing by ignoring this contraction and simply calculating the weight on @xmath730 .
so , modulo the contraction , we have exhibited any test configuration ( [ any ] ) as a finite number of blow ups ( starting with @xmath8 ) _ in subschemes @xmath770 supported in the scheme theoretic central fibre _ that themselves sit in flat families with the @xmath771 .
we calculate the weight on such a blow up in theorem [ big ] , for which we need two preliminary results .
the following proposition is the appropriate generalisation to general test configurations of the case @xmath772 used in ( [ st ] ) , ( [ qt ] ) and ( [ error ] )
. we will apply it to the flat families @xmath773 and their thickenings @xmath774 , when these thickenings are also flat .
[ central ] fix flat families @xmath775 with central fibres @xmath776 , such that the thickenings @xmath777 are also flat over @xmath82 .
let @xmath778 ( respectively @xmath779 ) denote the ideal sheaf of @xmath780 ( @xmath776 ) .
then @xmath781 flatness of @xmath782 and @xmath783 imply the exactness of the bottom two rows of the following , from which follows the exactness of the top row . @xmath784 chasing through either the top two rows or the first two columns then shows that in @xmath785 , @xmath786 . also by flatness
there is a similar diagram with @xmath81 replaced by @xmath787 ( and the right hand column suitably modified ) showing that in fact @xmath788 applying @xmath789 gives a similar exact diagram without the right hand and lower zeros , so the same argument shows that @xmath790 the top row also gives @xmath791 , which implies that @xmath792 @xmath793 , so @xmath794 and @xmath795 where we use the fact that in an abelian category , for @xmath796 with @xmath797 , we have @xmath798 in @xmath799 .
we claim that @xmath800 except for @xmath801 when the right hand side becomes @xmath802 .
the inclusion @xmath803 is clear . for @xmath804 , consider the left hand side : @xmath805 an element of this can be written as @xmath806 , that is @xmath807 , where @xmath808 and so @xmath809 . so @xmath810 , which by ( [ one ] ) is @xmath811 .
thus @xmath812 may be taken to lie in @xmath813 . similarly
therefore @xmath815 , proving the inclusion .
so by ( [ three ] ) and ( [ four ] ) , equation ( [ five ] ) has become @xmath816 to check that the sum is direct , intersect the @xmath58th numerator with the others : @xmath817 by the same methods as before this lies in @xmath818 , which is the @xmath58th denominator , as required .
to apply the above result inductively to the @xmath819 requires the flatness of the thickenings @xmath820 in @xmath819 of the proper transforms of the @xmath821 .
this is only automatic for @xmath503 , but also holds for arbitrary @xmath173 if the @xmath463 and @xmath11 are all _ smooth _ , or , we shall show , if : @xmath822 this odd looking condition is clearly satisfied if , for instance , each @xmath463 is reduced .
[ flat ] suppose that @xmath11 and the @xmath463 satisfy ( [ reduced ] ) . then @xmath823 is flat over @xmath82 for each @xmath824 .
firstly , consider @xmath825 .
its ideal is defined by those functions which , on restriction to @xmath826 , lie in @xmath827 .
since @xmath81 is invertible there , this implies that if @xmath828 then @xmath829 .
therefore the structure sheaf of @xmath825 has no @xmath81-torsion .
it is also flat away from @xmath830 ( where it is @xmath831 ) ; thus @xmath825 is automatically flat over @xmath82 .
@xmath832 ; therefore to prove that @xmath833 is flat over @xmath82 it is sufficient to prove the opposite inclusion @xmath834 @xmath819 is obtained by blowing up @xmath835 in @xmath836 ; we claim the complement @xmath837 of the proper transform @xmath838 of the central fibre in @xmath819 is affine over @xmath8 with coordinate ring @xmath839 @xmath840\ + \
\frac{{\curly i}_{\!i-1}}t\ + \
\frac{{\curly i}_{\!i-2}+{\curly i}_{\!i-1}^2}{t^2 } \ + \
\frac{{\curly i}_{\!i-3}+{\curly i}_{\!i-2}{\curly i}_{\!i-1}+{\curly i}_{\!i-1}^3}{t^3}\ + \
\ldots,\end{gathered}\ ] ] with its obvious ring structure .
each @xmath841 is an @xmath34-module , so the whole ring inherits an @xmath842$]-module structure , corresponding to the projection to @xmath8 .
there are many ways to see this .
one is to note that , away from @xmath843 , the map to @xmath844 of theorem [ pr ] is an isomorphism .
this is because , in the notation of that section , sections of @xmath845 over @xmath8 do not contract the exceptional divisor @xmath462 of the blow up of @xmath835 in @xmath836 ( as noted in the remarks following the proof of corollary [ seshadriir ] ) .
let @xmath4 denote the exceptional divisor of @xmath844 , and @xmath846 the canonical section vanishing on @xmath4 . for @xmath185 ,
the sections of @xmath847 are the sections of @xmath848 ( that is , pull back sections of @xmath849 from @xmath8 to @xmath844 and divide by @xmath850 to get a regular section of @xmath847 ) .
@xmath851 defines the section @xmath852 of @xmath94 which trivialises @xmath94 over @xmath837 the complement of its zero locus @xmath838 . using this trivialisation identifies the functions on @xmath837 which have poles of order @xmath853 on @xmath838 with @xmath854 taking the limit as @xmath371 gives the regular functions ( [ coord ] ) .
alternatively , we can work inductively with the @xmath855 .
a similar analysis as above shows the coordinate ring of the complement of @xmath856 in the blow up of a family @xmath136 over @xmath82 in an ideal @xmath857 is @xmath858 thus we find the coordinate ring of @xmath859 is @xmath860\,+\,\frac{{\curly i}_0}t\,+\,\frac{{\curly i}_0 ^ 2}{t^2}\,+\,\frac{{\curly i}_0 ^ 3}{t^3}\,+\ldots\ ] ] inside this the ideal of @xmath686 is @xmath861 ( as this is the largest ideal that localises to @xmath862 $ ] on @xmath826 ) ; applying ( [ bi ] ) to this ideal @xmath89 gives the coordinate ring of @xmath863 as @xmath860\ + \
\frac{{\curly i}_1}t\ + \
\frac{{\curly i}_1 ^ 2+{\curly i}_0}{t^2}\ + \
\frac{{\curly i}_1 ^ 3+{\curly i}_0{\curly i}_1}{t^3}\ + \
\ldots\ ] ] but this is the @xmath864 case of ( [ coord ] ) , and inductively we recover it for all @xmath429 . in ( [ coord ] ) we have the ideal @xmath865 as this is the largest ideal that localises to @xmath866 $ ] on @xmath826 . in the @xmath867 case , @xmath868 unless @xmath869 , so @xmath870 differs from ( [ coord ] ) only in the first term @xmath871 $ ] : @xmath872 by ( [ inclusion ] ) we are left with showing that each term of ( [ kz ] ) is contained in the @xmath58th power of ( [ z ] ) : @xmath873 we now work locally , where the conditions ( [ reduced ] ) on @xmath874 imply that @xmath875 and @xmath876 for some @xmath877 _ which do not divide _ @xmath878 .
therefore , for any @xmath879 with @xmath880 , we have @xmath881 using the fact the @xmath34 is torsion - free .
this gives the desired inclusion . to apply proposition [ flat ]
will involve replacing @xmath11 by a blow up on which the pullbacks of the @xmath463 are divisors . pulling back the polarisation
, we find we are forced to work with a semi - ample line bundle . to this end , for any @xmath3 and _ semi - ample _
@xmath882 , we define @xmath883 ( cf .
( [ degen ] ) and ( [ ss ] ) ) for any @xmath884 such that @xmath248 is saturated by its global sections .
this equals @xmath885 by the following generalisation of lemma [ error ] applied to @xmath10 and @xmath33 .
[ wkerror ] for @xmath886 semi - ample , and @xmath248 saturated by global sections for @xmath185 , @xmath887 .
consider the blow up of @xmath888 in @xmath229 with exceptional divisor @xmath232 and line bundle @xmath889 .
this is semi - ample , generated by @xmath890 , where @xmath891 are the sections vanishing at @xmath892 respectively .
so its higher cohomology has total dimension bounded by @xmath893 . but by pushing down first to @xmath888 , then to @xmath11 , this higher cohomology can be computed as @xmath894 , of total dimension @xmath895 .
fix @xmath896 such that for @xmath185 , @xmath897 is a semi test configuration for @xmath561 . given @xmath898 a @xmath72-invariant subscheme ,
denote its general fibre by @xmath3 and central fibre @xmath899 .
let @xmath900 denote the blow up of @xmath136 along @xmath440 , with exceptional divisor @xmath4 and line bundle @xmath901 .
( as usual @xmath902 or @xmath903 if @xmath904 is saturated by global sections for @xmath185 . )
thus @xmath239 is semi - ample by proposition [ ample ] , making @xmath905 a semi test configuration for @xmath561 for @xmath185 . then we have the following generalisation of theorems [ degen ] and [ ss ] .
[ big ] in the above situation , suppose that the thickenings @xmath906 are flat over @xmath82 for all @xmath824 , and the @xmath72-action on @xmath907 has only weights which lie between @xmath908 and @xmath909 , for some @xmath910 .
then @xmath911 @xmath912 has only weights which lie between @xmath913 and @xmath909 , and @xmath914 if @xmath30 is ample then the first @xmath893 correction vanishes , and if in addition either @xmath915 or the @xmath72-action on @xmath916 is trivial then the second correction is @xmath917 .
@xmath918 is a polarised family over @xmath82 with general fibre @xmath919 .
it is flat because @xmath920 has no @xmath81-torsion , by the flatness of @xmath136 . for @xmath185 , by the flatness of @xmath921 and ampleness of @xmath313 , the central fibre has sections @xmath922 .
again by flatness and ampleness , this has the same dimension as for the general fibre , which is @xmath296 , which equals @xmath923 by semi - ampleness . in turn by flatness and semi - ampleness , this equals @xmath924 .
therefore the inclusion @xmath925 has codimension @xmath917 .
this inclusion is @xmath72-equivariant , and all weights on the right hand side lie between @xmath908 and @xmath909 by assumption .
therefore the total weights on the two vector spaces differ by at most @xmath893 , as claimed .
if @xmath30 is ample then by cohomology vanishing @xmath926 and the correction vanishes .
+ to streamline notation we fix the convention in the proof of the second result that @xmath927 so that , for instance , @xmath928 means @xmath929 and we can deal with all of the terms in proposition [ central ] uniformly . for @xmath185 ,
this gives @xmath930 since the weight on @xmath787 is @xmath931 and lies between @xmath932 and @xmath909 , this shows the weights on @xmath912 indeed lie between @xmath913 and @xmath909 . included in this @xmath72-module
is @xmath933 of dimension @xmath934 , which we have already noted above is the same as @xmath296 to @xmath917 .
the same working ( applied to @xmath935 instead of @xmath936 ) shows that @xmath937 also equals @xmath938 .
thus the @xmath913-bound on weights means we can instead calculate the weight on ( [ split2 ] ) at the expense of an @xmath893 error . if @xmath30 is ample and @xmath915 then @xmath239 is also ample , so @xmath939 and we can calculate the weight on ( [ split2 ] ) without error .
finally if @xmath903 and @xmath30 is ample then as in proposition [ error ] , ( [ hbl ] ) and ( [ split2 ] ) agree for all but @xmath940 ( independent of @xmath173 ) so if the @xmath72-action on @xmath916 is trivial then their weights differ by @xmath941 .
define @xmath942 , and @xmath943 let @xmath944 be its weight space decomposition .
given such a splitting @xmath945 of a vector space @xmath946 , the generic subspace @xmath947 is not generated by the pieces @xmath948 ; i.e. @xmath949 .
but if @xmath946 has a @xmath72-action whose weight space decomposition is @xmath945 , and each @xmath947 is @xmath72-invariant , then indeed @xmath950 this holds here since @xmath951 is @xmath72-invariant .
so the splitting @xmath952 ( [ split2 ] ) further splits as @xmath953 defining @xmath954 , the weight on the determinant of ( [ split9 ] ) is @xmath955 . by ( [ filt1 ] ) , @xmath956 , which by ( [ filt ] ) is @xmath957 .
so as in ( [ noco ] ) , @xmath958 . similarly , the fact that @xmath959 filters @xmath960 means that @xmath961 .
therefore @xmath962 equals @xmath963 we can replace @xmath317 by @xmath318 at the expense of @xmath893 by lemma [ wkerror ] ( and at the expense of @xmath917 , by proposition [ error ] , if @xmath30 is ample ) . but @xmath964 sits in a flat family with @xmath965 , in which euler characteristics are preserved , so the above weight is @xmath966 by ( [ wkchi ] ) . and
if @xmath30 is ample and if either @xmath915 or the @xmath72-action on @xmath916 is trivial then the correction is @xmath917 .
to apply theorem [ big ] to corollary [ ref ] to express weights of test configurations ( [ any ] ) as sums of @xmath967s requires flatness of the thickenings @xmath833 in @xmath819 . to help achieve this by applying proposition [ flat ] we assume that @xmath11 is reduced and pass to a blow up @xmath88 on which the ( pullbacks of the ) @xmath463 are cartier divisors @xmath968 : @xmath969 .
in fact by hironaka s resolution of singularities we may take the @xmath968 to have simple normal crossing ( snc ) support .
that is there are smooth reduced divisors @xmath970 such that @xmath971 has simple normal crossings and @xmath972 for each @xmath429 and some nonnegative integers @xmath973 .
we pull @xmath30 back to @xmath234 , and construct the families @xmath974 as before , using @xmath975 in place of @xmath976 .
there are equivariant surjective maps @xmath977 which identify ( by pullback ) the @xmath764 line bundles , defined by pulling back sections of @xmath978 on @xmath11 to sections of @xmath979 on @xmath234 .
[ bigger ] suppose that @xmath11 is normal , and fix an arbitrary test configuration @xmath214 ( [ any ] ) with associated subschemes @xmath980 ( [ ideals ] ) .
suppose that there is a resolution @xmath981 with divisors @xmath968 satisfying condition ( [ reduced ] ) .
( for instance if the connected components of the snc divisors @xmath968 have the same multiplicities locally , i.e. @xmath982 for some locally constant @xmath983 and all @xmath429 .
e.g. if the @xmath968 are all reduced , which is the @xmath984 case . )
then @xmath985 @xmath986 dominates @xmath732 which in turn dominates @xmath987 and so @xmath214 , and the space of sections of the @xmath173th power of the line bundle on the general fibre is @xmath296 for all three , by the normality of @xmath11 .
therefore by proposition [ stein ] and the first part of theorem [ big ] we may calculate @xmath988 at the expense of an @xmath893 error .
we apply theorem [ big ] inductively to @xmath989 with @xmath523 ( which is @xmath990 by corollary [ seshadriir ] ) and @xmath991 with central fibre @xmath992 .
the condition ( [ reduced ] ) on the @xmath968 guarantees that each @xmath993 is flat over @xmath82 by proposition [ flat ] . the induction starts with @xmath994 , for which all weights are zero so trivially satisfy the @xmath908 bound .
theorem [ big ] then ensures the induction continues to compute the weight as @xmath995 since @xmath11 is normal , all sections of @xmath996 on @xmath234 are pullbacks from @xmath997 . thus the same is true of those sections vanishing on the pullback of @xmath998 .
thus @xmath999 by their definition ( [ wkz ] ) , as required .
in particular we can now calculate the leading order term of the weight if the snc divisors @xmath968 in a resolution of singularities @xmath234 of @xmath1000 are reduced .
the next case we consider is when they have multiplicities which can vary with @xmath429 but are still locally constant over their snc support . here
the relevant flatness result does not hold , but we will find that it does after performing a basechange and normalisation .
so we consider the case when @xmath225 ( [ ideals ] ) is locally of the form @xmath1001 for some reduced snc divisor @xmath340 .
these @xmath1002 may vary over the different connected components of @xmath340 , but since the total weight is a sum over contributions from each connected component we can calculate the weight of each separately . pick a local function @xmath1003 generating the ideal @xmath1004 , so that ( [ ideal ] ) is @xmath1005 . in the concave hull of the points @xmath1006 in the @xmath1007-plane , we choose extremal vertices @xmath1008 so that they form a concave set with the same concave hull ( and @xmath1009 ) .
this defines a new ideal @xmath1010 with the same integral closure as ( [ ideal ] ) since in this situation taking integral closures corresponds to taking concave hulls , and an ideal saturates its integral closure . in the next theorem
we decorate @xmath1011 ( [ wkz ] ) as @xmath1012 with the value @xmath35 that determines the line bundle @xmath1013 on the blow up .
[ divs ] set @xmath1014 , and @xmath1015 . if @xmath30 is ample , then the total weight of the blow up in the ideal ( [ ideal ] ) is the sum over the connected components @xmath340 of @xmath1016 for some @xmath552 . if @xmath30 is semi - ample , the above expression is correct to @xmath893 .
we start by proving the weaker estimate for @xmath30 semi - ample .
first blow up @xmath8 in @xmath1017 ( i.e. locally in the ideal @xmath1018 ) , then in @xmath1019 ( recall this is the central fibre of @xmath1020 in the blow up ) , then @xmath1021 , etc . up to @xmath1022 . as in theorem [ pr ]
we denote by @xmath462 the pullback of the exceptional divisor of the @xmath429th blow up , and by @xmath1023 the canonical section of @xmath1024 vanishing on @xmath462 .
then we claim that the pushdown of @xmath1025 , to @xmath8 is the ideal @xmath1026 here
@xmath1027 " means all convex combinations in between " , i.e. the integral closure of the ideal generated by the named terms .
( so @xmath1028 includes all monomials @xmath812 such that there exist @xmath1029 with @xmath1030 . )
that is , we claim the sections of @xmath1031 over @xmath8 are @xmath1032 times by the above ideal .
this is standard but fiddly to prove , and is best done by newton diagram .
we give an unenlightening proof ; the reader is advised to skip straight to the newton diagram ( figure [ diag ] ) for the special case below .
we prove ( [ induction ] ) inductively alongside the claim that the complement of the divisor @xmath1033 ( which is the proper transform , in the @xmath58th blow up , of the central fibre of the @xmath1034th blow up ) is affine over @xmath8 with coordinate ring @xmath1035.\ ] ] ( in particular , at a smooth point of @xmath340 where @xmath1003 is a local coordinate in an analytic coordinate system @xmath1036 for @xmath11 , we see that this part of the @xmath58th blow up is @xmath1037 = { \operatorname{spec}\,}{\mathbb c}[y_1,\ldots , y_{n-1},z , t]$ ] , where @xmath1038 , and so is locally isomorphic to @xmath8 with the proper transform of @xmath1039 being @xmath1040 . ) for the first blow up , the sections of @xmath1041 are @xmath1042 ( there are no more because the exceptional divisor is a @xmath1043-bundle over @xmath340 ) .
this is in agreement with ( [ induction ] ) , and the proper transform of the central fibre is @xmath1044 . on its complement , @xmath1045 trivialises @xmath1041
; dividing by it identifies the sections @xmath1046 with @xmath1047 . taking the limit
as @xmath371 gives the coordinate ring @xmath1048 $ ] claimed ( [ affcoord ] ) .
so the induction starts at @xmath867 . at the @xmath58th stage the coordinate ring ( [ affcoord ] )
shows that @xmath1049 has ideal @xmath1050 , where @xmath1038 .
let @xmath237 denote the @xmath1051th blow up .
by the first step of the induction , @xmath1052 has coordinate ring augmented by @xmath1053 ; i.e. by induction @xmath1054\left[\frac z{t^{j+1}}\right]={\mathcal o}{^{\ } _ } { \!x\times{\mathbb c}}\left[\frac z{t^{j+1}}\right]$ ] .
@xmath1055 is the ideal @xmath1056 ; i.e. the sections of @xmath1057 over @xmath855 are @xmath1058 times by sections of this ideal .
multiplying by the trivialising section @xmath1059 of @xmath1031 shows that over our affine piece , @xmath1060 is the ideal @xmath1061.s_1^{-p_1}\!\ldots s_j^{-p_j}\subseteq{\mathcal o}(-p_1e_1-\ldots - p_je_j).\ ] ] thus the sections of @xmath1062 are the sections of @xmath1031 which lie in the above ideal , i.e. the intersection of ( [ induction ] ) with the ideal @xmath1063 .
but this is ( [ induction ] ) with @xmath58 replaced by @xmath1064 , completing the induction .
we first assume that the @xmath1065 are all integers .
then the ideal ( [ ideal2 ] ) has integral closure of the form ( [ induction ] ) , on taking @xmath1066 all equal to @xmath1067 , then @xmath1068 all equal to @xmath1069 , and so on , up to @xmath1070 all equal to @xmath1071 .
this is illustrated in figure [ diag ] , the newton diagram of the @xmath1007 plane , with the @xmath1065s being ( minus ) the gradients of the bold lines . taking the integral closure , i.e. including the monomials
@xmath1027 " in ( [ induction ] ) , corresponds to including all integral points both on and above the line to lie in the ideal . replacing @xmath1003 by @xmath1038 and
multiplying by the trivialising section @xmath1072 in the above working corresponds to the integral affine transformation that locally takes one corner of the bold line into the @xmath1007-axes .
so we may calculate the weight of this sequence of blow ups , @xmath1073 times in the central fibre of @xmath1020 with weight @xmath1074 , then @xmath1075 times in the central fibre of @xmath1020 with weight @xmath1076 , etc .
( the weight here just means the coefficient of the exceptional divisor in the line bundle @xmath266 we use , as usual . ) the flatness criterion ( [ reduced ] ) is trivially satisfied so that by theorem [ big ] we may calculate the weight to be that claimed .
this differs from the weight of the blow up in ( [ ideal ] ) by a @xmath1077 correction by proposition [ stein ] since the blow up in the integral closure ( [ ideal2 ] ) is the normalisation of the blow up in ( [ ideal ] ) .
if @xmath30 is ample then we can improve the estimate . in the first blow up
, @xmath1078 is the trivial product configuration , so the @xmath72-action on @xmath916 is trivial and we can use the better estimate of theorem [ big ] .
next we group the first @xmath1073 blow ups together as one blow up in @xmath1079 ; this has the advantage that we do no blowing down ( in fact the previous @xmath1073 blow ups blow down to this ) . since this blow up is the @xmath1080 basechange of the blow up in the ideal @xmath1081 ( with @xmath1074 ) , it has weight @xmath1082 , the same as the sum of the @xmath1073 blow ups we performed above .
similarly we group the next @xmath1083 blow ups together , blowing up with @xmath1076 in the ideal generated by @xmath1084 and the ideal of @xmath1020 , and calculate its weight as the @xmath1085 basechange of a blow up we already know .
inductively we end up with the same formula for the weights , but with the added @xmath917 accuracy of theorem [ big ] coming from the fact that ( after the first blow up ) there are no blow downs , so @xmath35 is less than the seshadri constant of the relevant @xmath1086 at each stage . finally ,
if the @xmath1065 are not integers , we simple replace @xmath81 by @xmath1087 in ( [ ideal ] ) , i.e. we basechange our test configuration , where @xmath1088 clears the denominators of all the @xmath1065 .
this replaces all the @xmath1065 by the integers @xmath1089 while multiplying the @xmath72-weight by @xmath1088 .
substituting the @xmath1089 into our formula for the weights gives the weight of this new test configuration with a @xmath1077 correction ( coming from taking the integral closure of this new ideal ; replacing the test configuration with its normalisation and using proposition [ stein ] ) ; dividing by @xmath1088 gives the weight of the original test configuration for any @xmath1065 .
this just leaves the case of where the divisors @xmath1090 in the snc divisors @xmath968 in a resolution @xmath234 intersect with differing multiplicities ( the simplest example being @xmath1091 and @xmath1092 locally ) .
this of course can not happen for curves , so we have [ kslopecurve ] a smooth curve @xmath0 is k-(semi / poly)stable if and only if it is slope ( semi / poly)stable .
smooth curves are normal , and for the resolution of singularities @xmath99 we of course take @xmath11 itself ,
so we can apply the stronger form of theorem [ divs ] .
since the donaldson - futaki invariant only uses the coefficients @xmath544 and @xmath1093 of @xmath1094 , this implies that the futaki invariant of an arbitrary test configuration ( [ any ] ) is @xmath1095 a positive linear combination of futaki invariants of ( the deformation to the normal cone of ) subschemes .
slope stability implies that these are all positive .
this result makes it trivial to understand k - stability for smooth curves ; see theorem [ smoothcurvesarekstable ] .
mumford s notion of chow ( semi)stability @xcite of @xmath1096 , for _ fixed _
@xmath344 , is the simplest form of stability to calculate ( as opposed to _
asymptotic _ chow stability ( [ chow ] ) , which is second only to asymptotic hilbert stability in difficulty ) .
it is also useful in algebro - geometric applications since it is a genuine git notion giving projective ( and so proper and separated ) moduli spaces . for any subscheme @xmath1097
, we define the polynomial @xmath368 by @xmath1098 for @xmath1099 as before .
we also define @xmath1100 by @xmath1101 ; by ( [ a0 ] ) , @xmath1102 . for any subscheme @xmath1103 and _ integer _
@xmath50 , we define the _ chow slope _ of @xmath29 to be @xmath1104.\ ] ] ( here we define max of the empty set to be @xmath1105 , and division by 0 to give @xmath1106 . ) setting @xmath33 gives @xmath1107 as @xmath1108 if @xmath1109 is the kodaira embedding of @xmath11 in @xmath1110 , i.e. if @xmath1111 is an isomorphism , then @xmath1112 for all @xmath1097 .
this can be arranged , for fixed @xmath0 , by taking a sufficiently large multiple @xmath1113 of the polarisation and setting @xmath1114 .
in this situation the above slopes can be written intrinsically in terms of @xmath1115 as @xmath1116 we say that @xmath1117 is _ chow slope stable _ if @xmath1118 for all nonempty @xmath1097 , and chow slope semistable if @xmath1119 .
we have @xmath1120 [ thmchow ] if @xmath1121 is chow ( semi)stable then it is chow slope ( semi)stable .
choose a basis of @xmath1122 compatible with the filtration @xmath1123 , so that the first @xmath1124 elements are contained in @xmath1125 .
the corresponding hyperplanes @xmath1126 @xmath1127 and subschemes @xmath1128 therefore satisfy @xmath1129 ( i.e.@xmath1130 ) .
choose weights @xmath1131 so that ideal on @xmath8 , @xmath1132 equals @xmath1133 which by construction is contained in the ideal @xmath1134 @xmath1117 is chow stable for the @xmath72-action which has weight @xmath1135 on the @xmath429th vector of our basis of @xmath1122 .
mumford ( @xcite theorem 2.9 ) shows that this is equivalent to the inequality @xmath1136 where @xmath1137 is the chow weight of the blow up of @xmath8 in the ideal @xmath89 ( [ wobbly ] ) , i.e. @xmath1138 is the total weight of the induced @xmath72-action on @xmath1139 .
( this is only one half of mumford s result , the harder part being his computation of @xmath1137 in terms of ideals on @xmath8 , which we do not use .
the reader wishing to compare our conventions with mumford s should rewrite the above as @xmath1140 , replace @xmath101 with @xmath131 , @xmath344 with @xmath101 , and @xmath1135 with @xmath1141 .
finally the sign arises from our convention ( [ convention ] ) that if @xmath1142 acts on @xmath71 then the induced action on its functions @xmath1143 is by @xmath1144 ; mumford calculates the weight of @xmath1145 . )
notice this is theorem [ thm : instability ] applied to ( [ normalised ] ) on setting @xmath1146 and @xmath1147 . since @xmath1148 , we know the chow weight on the blow up in @xmath1149 ( proposition [ integrals ] ) .
but @xmath89 is contained in @xmath1149 ( [ straight ] ) , so the weight on the latter is more negative ( for instance mumford s formula ( @xcite theorem 2.9 ) for the chow weight shows this ) , giving the inequality @xmath1150 thus ( [ mummy ] ) gives @xmath1151@xmath1152 which is @xmath1153 semistability is similar , replacing ( [ mummy ] ) by the non strict inequality .
setting @xmath523 gives the inequality @xmath1154 , and replacing @xmath10 by @xmath1155 then gives @xmath1156 .
we could have gotten this alternative slope - type inequality directly from stability for the @xmath72-action that had all of the above @xmath1135 equal to zero ( for @xmath1157 ) or one ( @xmath1158 ) .
however , it is strictly weaker than our slope inequality @xmath1159 ( [ thmchow ] ) since this last sum is clearly @xmath1160 . for @xmath11 semistable
, taking @xmath1161 ( so @xmath1162 ) shows that @xmath1163 , i.e. @xmath11 is not contained in any hyperplane and @xmath1122 injects into @xmath1164 . to make this injection
an isomorphism requires the assumption that @xmath1121 is a kodaira embedding .
notice that ( [ r1 ] ) is more - or - less the @xmath195 coefficient of ( [ weight ] ) ( with @xmath166 and rearranged ) .
there can be higher cohomology corrections since @xmath89 ( [ wobbly ] ) is not the same as @xmath1149 ( [ straight ] ) ; these disappear for @xmath186 .
we now want to understand to what extent slope chow stability should imply chow stability .
this involves demonstrating the inequality ( [ mummy ] ) for _ all _ linearly independent sequences of hyperplanes @xmath1165 and _ all _ choices of weights @xmath1166 . as before we set @xmath1167 .
the idea is to relate the weight of the associated @xmath72-action ( with weight @xmath1135 on the @xmath429th vector of our basis of @xmath1122 ) to a sum of weights of @xmath72-actions of the standard form considered in theorem [ thmchow ] .
the problem is that we have noted that we can only express weights as a sum of weights of the deformation to the normal cone of subschemes if either condition ( [ reduced ] ) holds or the multiplicities of @xmath841 are locally constant ( at least on some resolution where they are ncds ) .
but in either of these good cases we can demonstrate the general procedure of passing from slope stability to stability .
so let us assume for illustration that each @xmath841 has the local form @xmath1168 for some reduced cartier divisor @xmath340 and number @xmath1169 constant on @xmath340 .
( we have seen , as in theorem [ bigger ] , one can also pass to a resolution @xmath234 if necessary to make the @xmath463 sncds to calculate chow weights ( the @xmath195-coefficient of hilbert weights ) , so the assumption is not so restrictive , and is sufficient to deal with curves . )
that is , @xmath1170 has the local form @xmath1171 with @xmath1172 , and @xmath1173 ( so @xmath1174 is the smallest number with @xmath1175 ) .
these @xmath1169 and @xmath1176 may vary as @xmath340 ranges over the connected components of @xmath442 .
[ uniform ] if @xmath1177 is slope chow stable then the weight @xmath1137 ( [ mummy ] ) satisfies @xmath1178 , where @xmath1179 .
this is the inequality @xmath1180 required for chow stability ( [ mummy ] ) , modulo some @xmath1181 corrections ( which we estimated away in the k - stability analogue ) . in ( [ uniformcurve ] )
the result is strengthened slightly and the correction estimated on curves to prove their asymptotic chow stability .
the @xmath195-coefficient @xmath1137 of the weight of our @xmath72-action is the same as that on the blow up of @xmath994 in @xmath89 ( [ deali ] ) . in theorem [ divs ]
we calculated this to be a sum @xmath1182 over the connected components @xmath340 of its support , where @xmath1183 here @xmath1184 is defined uniquely by requiring that @xmath1185 lies on the boundary of the concave hull of the set of points @xmath1186 in the @xmath1187-plane . thus the @xmath1184 need not be integers but , for instance , @xmath1188 and @xmath1189 for all @xmath1190 .
more generally , @xmath1191 for all @xmath429 .
concavity of the @xmath1185 ensures that any term in the above sum with a zero in the denominator also has zero in the numerator ; the prime @xmath1192 on the summation sign signifies that we ignore these @xmath1193 terms ( the terms with @xmath1194 ) in the sum ; equivalently we set @xmath1195 .
in the first term we set @xmath1196 . the seshadri constant of @xmath340 is @xmath1197 for all @xmath429 since @xmath1198 is locally the intersection of a sequence of hyperplanes , so @xmath1199 is globally generated near @xmath340 , so @xmath1200 is too . therefore chow slope stability for @xmath1201 ( [ sss ] ) gives the inequalities @xmath1202 since all of the integrals in ( [ a ] ) have coefficients which are @xmath1203 by the concavity of the @xmath1185 , we obtain @xmath1204 where in the last line we have added back in the @xmath1194 terms since they are positive . since the last sum is also @xmath1205 , with a positive coefficient for each @xmath1184 , we can replace @xmath1184 by @xmath1206 to give @xmath1207 sum over the connected components @xmath340 and use @xmath1208 to give @xmath1209 ( in the above , any sum from @xmath429 to @xmath58 with @xmath1210 is to be interpreted as zero , and in passing from the first line to the second we have used the fact that @xmath1211 since @xmath1175 . ) the chow slope inequality ( [ sss ] ) can be rewritten @xmath1212 for some small @xmath1213 . this implies the weaker inequality ( which is all that we shall require for curves ) @xmath1214 if @xmath387 can be chosen uniformly in ( [ css ] ) for all @xmath3 we call @xmath11 _ uniformly chow slope stable with constant @xmath387_. this will help us to deal with the correction @xmath584 . for @xmath11 a curve we can improve the estimates of theorem [ uniform ] slightly and use clifford s theorem to bound the @xmath1181 terms : [ uniformcurve ] if a smooth curve @xmath0 of genus @xmath1142 and @xmath1215 is uniformly chow slope stable ( [ css ] ) with constant @xmath1216 then it is chow stable .
we follow the same proof as above with @xmath1217 replaced by @xmath1218 throughout , up to ( [ diverge ] ) , at which point we use estimates specific to curves to better bound the @xmath1181 terms .
that is , @xmath340 is locally a smooth point @xmath522 in @xmath11 , and for those @xmath429 with @xmath1219 ( i.e. those involved in the sum @xmath1220 ) , @xmath1221 therefore ( [ diverge ] ) becomes @xmath1222 we can now add back in those @xmath429 with @xmath1194 , as all terms are positive .
each @xmath1184 appears with positive coefficient in the result , since it can be rearranged as @xmath1223 .
so replacing @xmath1184 by @xmath1206 , summing over @xmath62 in the support of @xmath442 , and using @xmath1224 , gives @xmath1225 \\ & \le & \sum_{i=1}^n(\rho_{i+1}-\rho_i)\big[n+1-i\big ] + \frac12\sum_{i=1}^n(\rho_{i+1}-\rho_i)\big(h^1({\curly i}_i(1))+h^1({\curly i}_{i+1}(1))\big ) \\ & = & \sum_{i=1}^{n+1}\rho_i+\delta , \hspace{20 mm } \delta:=\frac12\sum_{i=1}^n(\rho_{i+1}-\rho_i ) \big(h^1({\curly i}_i(1))+h^1({\curly i}_{i+1}(1))\big)\end{aligned}\ ] ] i.e. @xmath1226 this @xmath584 is a tiny improvement over the one in theorem [ uniform ] , but can be bounded using clifford s theorem , which for our purposes says that @xmath1227 .
thus @xmath1228 for @xmath1229 and vanishes for @xmath1230 .
this yields @xmath1231 since @xmath1232 .
since the @xmath1135 are monotonic in @xmath429 , @xmath1233 adding @xmath1234 to both sides gives @xmath1235 and so @xmath1236 combined with ( [ d ] ) we find that uniform slope stability implies that @xmath1237 which implies the inequality @xmath1238 required ( [ mummy ] ) if @xmath1239 .
the condition @xmath1240 implies that @xmath1241 so @xmath1242 and @xmath1243 .
therefore the inequality is @xmath1244 .
smooth curves @xmath1177 of genus @xmath1245 are uniformly slope stable ( [ css ] ) with any constant @xmath1246 , and so are asymptotically chow stable
. we need only demonstrate the inequality ( [ css ] ) in the case of @xmath1247 a single reduced point , as the inequality for a multiple of @xmath522 is weaker and the inequality for arbitrary @xmath10 follows from adding the inequalities for its connected components .
by riemann - roch , @xmath1248 , so that @xmath1249 while @xmath1250 .
therefore @xmath1251 so long as @xmath1252 .
of course @xmath1253 , so @xmath0 is uniformly chow slope stable with any constant @xmath1246 .
theorem [ uniformcurve ] then gives chow stability , so long as @xmath1254 which is true for @xmath1245 and sufficiently large @xmath1255 .
mumford @xcite proves the sharper result that @xmath1177 is chow stable for @xmath1256 using a combinatorial argument .
our remaining examples all deal with k - slope stability . many more examples , calculations and applications are given in @xcite .
suppose that @xmath11 has at worst canonical singularities .
that is @xmath11 is normal , there is an integer @xmath983 such that @xmath1257 is cartier , and given any resolution of singularities @xmath1258 we have @xmath1259 where the @xmath1260 are the irreducible components of the exceptional set of @xmath1261 .
we can define intersection with the canonical class of @xmath11 on @xmath1262 by @xmath1263 . for
any subscheme @xmath3 let @xmath1264 be a resolution of singularities of @xmath234 , the blow up of @xmath11 along @xmath10 .
@xmath1265 is a resolution of singularities of @xmath11 , so ( [ canonical ] ) holds .
letting @xmath1266 , @xmath1267 is nef on @xmath1268 for @xmath1269 .
we wish to compute @xmath27 on @xmath1268 instead of @xmath99 .
since @xmath1270 with quotient supported in codimension one , we have an inclusion @xmath1271 with cokernel of dimension @xmath1272 by fujita vanishing , for @xmath1273 . as in ( [ normal ] ) , for @xmath185 , @xmath1274 since the support of @xmath1275 has codimension at least @xmath421 .
therefore @xmath1276 for some @xmath552 .
that is , for @xmath1273 , we have @xmath1277 and @xmath1278 where the second inequality follows from ( [ canonical ] ) and the fact that @xmath1279 is nef .
again as in ( [ normal ] ) , the first two terms of the euler characteristic of @xmath30 are the same as those of @xmath1280 , so equality holds in ( [ a_idefinedbyaresolution ] ) when @xmath1281 . that is @xmath484 and @xmath1282 since @xmath30 is trivial along the @xmath1260 . with these preliminaries ,
it becomes easy to prove the following .
[ cy ] * calabi - yaus and canonical models*. + let @xmath11 be an irreducible variety with at worst canonical singularities .
@xmath1283 if @xmath1284 is numerically trivial then @xmath0 is slope stable for all polarisations @xmath30 .
@xmath1283 if @xmath1284 is ample then the canonical polarisation @xmath1285 is slope stable . in both cases
, @xmath1286 is numerically equivalent to a nonnegative multiple @xmath1287 of the polarisation .
so by ( [ a_1definedbyaresolution ] ) , @xmath1288 . by ( [ a_idefinedbyaresolution ] ) , @xmath1289 which equals @xmath1290 since @xmath1279 is nef .
since @xmath455 for @xmath381 ( [ a0 ] ) , integration gives @xmath1291,\ ] ] which rearranges to give slope stability : @xmath1292 .
* for any divisor @xmath1294 there is a @xmath1295 such that if @xmath1296 and @xmath1297 is ample then @xmath0 is slope stable . *
if @xmath1298 is nef then @xmath0 is slope stable . *
if @xmath1299 is nef then @xmath0 is slope stable . in @xcite
we prove that no smooth @xmath10 can slope destabilise a smooth @xmath0 with these properties , and the proof extends to general @xmath10 and @xmath11 with canonical singularities using the preliminaries ( [ a_idefinedbyaresolution ] ) and ( [ a_1definedbyaresolution ] ) .
these results are to be expected due to the deep and difficult related results of viehweg @xcite , and the expectation that the minimal model programme can be carried out in all dimensions ( and so can be done in families ) @xcite .
in fact theorem [ cy ] can be proved in a round about way for smooth varieties with no holomorphic vector fields by the stability results of @xcite applied to the khler - einstein metrics of @xcite on such varieties .
similarly , if @xmath11 has no holomorphic vector fields and @xmath30 is ample and sufficiently close to @xmath1284 then an implicit function theorem argument applied to the khler - einstein metric provides a constant scalar curvature khler metric in @xmath1300 $ ] .
but the quick proofs above demonstrate that using slope stability it could become much easier to produce and compactify moduli of varieties with semi - ample canonical bundle @xcite .
[ prop : stabilityofpolarisedcurves ] let @xmath1301 be an irreducible polarised curve with arithmetic genus @xmath1302 .
the hilbert - samuel polynomial of a subscheme @xmath1303 can be written @xmath1304 and @xmath10 destabilises @xmath1305 if and only if it strictly destabilises , if and only if @xmath1306 .
the assumption on the genus implies that @xmath1307 . as @xmath1308
, @xmath1309 is a degree 1 polynomial vanishing at the origin , while @xmath1310 has degree 0 .
so writing @xmath1311 and @xmath1312 , @xmath1313 this has the same sign as @xmath1314 for all @xmath1315 , and tends to @xmath1316 as @xmath1317 .
destabilises if and only if it strictly destabilises if and only if this sign is negative .
for the first statement suppose @xmath1247 is a point of @xmath1305 with multiplicity @xmath1318 .
northcott has shown ( @xcite lemma 1 ) that @xmath1319 .
hence @xmath1320 and @xmath10 strictly destabilises . for the second statement suppose for a contradiction that @xmath10 is a destabilising subscheme of @xmath1305 . from proposition
[ simplify ] ( especially ( [ simplify2 ] ) ) and lemma [ seshadridisconnected ] we see that @xmath1321 for some connected component @xmath631 of @xmath10 and @xmath1322 ( we are not saying that @xmath631 destabilises since we may not be allowed to take @xmath1323 ) .
if the support @xmath1324 of @xmath631 is a smooth point then @xmath1325 must by a thickened point , which by proposition [ simplify ] shows that @xmath1326 .
but in the notation of ( [ hilbertsamuelofsubschemesofcurves ] ) , @xmath1327 , which by theorem 3.2 of @xcite implies that @xmath1328 , so by proposition [ prop : stabilityofpolarisedcurves ] @xmath1329 is in fact @xmath1330 . hence @xmath10
must be supported at one of the singular points , and we can reduce to looking at the local analytic model @xmath1331/(xy)$ ] .
let @xmath89 be an ideal of @xmath1332 which is supported at @xmath1333 .
( by abuse of notation we shall not distinguish between a polynomial in two variables and its class in @xmath1332 . )
we can pick a finite number of generators of @xmath89 of the form @xmath1334 , with @xmath1335 .
let @xmath1336;@xmath1337 since @xmath89 is supported at the origin .
pick an @xmath429 such that @xmath1338 and @xmath1339 ; then @xmath1340 ; similarly @xmath1341 .
therefore @xmath1342 is spanned by @xmath1343 . by the definition of @xmath62 and @xmath1344 ,
@xmath1345 is spanned by @xmath1346 , so the vectors @xmath1347 in @xmath1342 are linearly independent .
thus @xmath1348 writing @xmath1349 we have @xmath1350 .
hence @xmath1351 , so by proposition [ prop : stabilityofpolarisedcurves ] @xmath89 does not destabilise .
these results combined with theorem [ thm : kstableslopestable ] imply that curves with singularities of multiplicity greater than two are strictly k - unstable .
we can not deduce positive results about k - stability from the results of section [ converse ] , however , since @xmath1305 need not be normal . unless , that is , @xmath1305 is smooth : by corollary [ kslopecurve ] it is equivalent to prove the results for slope ( poly)stability . instead of using previous results
it is now easier to proceed directly .
any nonempty subscheme @xmath10 is a divisor of degree @xmath1353 , so @xmath1354 shows that @xmath1355 and @xmath1356 .
thus @xmath1357 for @xmath1245 , proving slope stability . for @xmath1352 , @xmath35 may take values up to and _ including _ @xmath1358 , since @xmath1359 is globally generated .
thus @xmath1360 with equality ( strict semistability ) only for @xmath1361 , _
i.e. _ @xmath10 a single point , and @xmath267 .
since the deformation to the normal cone of a single point on @xmath1043 blows down to @xmath1362 ( with a nontrivial @xmath72-action ) from which the relevant line bundle @xmath239 pulls back , we find @xmath1043 is in fact slope polystable .
_ on the ricci curvature of a compact khler manifold and the complex monge - ampre equation
. i. _ comm .
pure appl . math . *
31 * , 339411 .
zhang , s. ( 1996 ) . _ heights and reductions of semi - stable varieties .
_ compositio math . *
104 * , 77105 . | we make a systematic study of the hilbert - mumford criterion for different notions of stability for polarised algebraic varieties @xmath0 ; in particular for k- and chow stability . for each type of stability
this leads to a concept of slope @xmath1 for varieties and their subschemes ; if @xmath0 is semistable then @xmath2 for all @xmath3 .
we give examples such as curves , canonical models and calabi - yaus .
we prove various foundational technical results towards understanding the converse , leading to partial results ; in particular this gives a geometric ( rather than combinatorial ) proof of the stability of smooth curves . |
in the minimal standard model ( sm ) , neutrinos are massless
. however , small nonzero neutrino masses are required by the atmospheric and solar neutrino experiments .
a natural explanation for such tiny neutrino masses in the sm comes from an effective dimension-5 operator @xcite @xmath1 where @xmath2 are the usual left - handed lepton doublets transforming as @xmath3 under the standard electroweak @xmath4 gauge group and @xmath5 is the usual higgs doublet of the sm .
there are three realizations of this operator @xcite , the most popular one being the canonical seesaw @xcite mechanism which adds three singlet heavy neutral fermions @xmath6 to the sm lagrangian .
the neutrino mass matrix is then given by @xmath7 where @xmath8 is the @xmath9 dirac mass matrix linking @xmath10 with @xmath11 through the yukawa interactions @xmath12 .
the majorana masses of @xmath11 violate lepton number by two units .
therefore , in the early universe , a net lepton asymmetry may be generated @xcite through the out - of - equilibrium decay of the lightest @xmath11 ( call it @xmath13 ) .
the generated lepton asymmetry then gets converted into a baryon asymmetry through the interactions of the sm sphalerons @xcite which conserve @xmath14 , but violate @xmath15 , where @xmath16 and @xmath17 are baryon and lepton number respectively .
the existence of @xmath11 explains thus at the same time why both neutrino masses as well as the observed baryon asymmetry of the universe ( bau ) are nonzero and small . in supersymmetric theories the reheating temperature ( @xmath18 ) following inflation is likely to be rather low @xcite .
although some models @xcite may allow a higher reheating temperature , in the conventional models @xmath18 is bounded strongly from above by the possible overproduction of gravitinos . on the other hand , in the simplest version of the seesaw mechanism
the condition for thermal leptogenesis requires that the mass @xmath19 of the lightest right - handed neutrino @xmath13 should be much higher than @xmath18 , assuming that @xmath13 contributes to the left - handed neutrino masses dominantly .
since inflation would erase any pre - existing lepton asymmetry , the asymmetry generated by @xmath13 after inflation would be highly suppressed by its small number density and hence this mechanism will fail to explain the bau . to avoid this problem ,
several ideas have been discussed in the literature @xcite .
an attractive scenario is the extended seesaw mechanism . in this paper
we follow the same scheme and work with the canonical seesaw mechanism ( sm plus three @xmath11 ) in the next to minimal supersymmetric standard model ( nmssm ) , which has an extra singlet superfield @xmath20 . to distinguish @xmath11 from @xmath20 ,
an exactly conserved @xmath0 discrete symmetry is imposed , corresponding to @xmath21 .
we then add an extra heavy singlet superfield @xmath22 , together with a softly broken discrete symmetry @xmath23 , under which @xmath22 is odd and all others are even . as a result ,
the production and decay channels of @xmath22 are different , and the out - of - equilibrium decay of @xmath22 can take place much below the mass scale of the lightest right - handed neutrino @xmath13 .
the lower bound on the mass scale of @xmath22 can then be compatible with the upper bound on the reheating temperature after inflation .
the rest of this paper is arranged as follows . in section
ii we review the canonical leptogenesis and briefly recall the davidson - ibarra ( di ) bound on the mass scale of @xmath13 . in section
iii we discuss an extended seesaw model by introducing an additional heavy singlet fermion @xmath22 of mass less than that of @xmath13 . in section
iv we discuss how the thermal - leptogenesis constraint on the mass scale of @xmath22 can be lowered in comparison with the mass scale of @xmath13 . in section v we solve the required boltzmann equations numerically and show how the low mass scale of @xmath22 is compatible with thermal leptogenesis . in section
vi we state our conclusions .
in canonical leptogenesis the lightest right - handed neutrino @xmath13 decays into either @xmath24 and @xmath25 , or @xmath26 and @xmath27 .
thus a cp asymmetry can be established from the interference of the tree - level amplitudes with the one - loop vertex @xcite and self - energy corrections @xcite .
a net lepton asymmetry arises when the decay rate @xmath28 fails to compete with the expansion rate of the universe @xmath29 at @xmath30 , where @xmath31 is the effective neutrino mass parameter , @xmath32 is the effective number of relativistic degrees of freedom in the mssm and @xmath33 gev .
this means that a bound on @xmath34 may be established by first considering the out - of - equilibrium condition @xmath35 which gives @xmath36 however , for @xmath37 a reduced lepton asymmetry may still be generated , depending on the details of the boltzmann equations which quantify the deviation from equilibrium of the process in question . assuming a normal mass hierarchy in the right handed neutrino mass spectrum the cp asymmetry @xmath38 is given by @xmath39 ^ 2}{(m_d^\dagger m_d)_{11}}\ , .
\label{epsilon_1}\ ] ] the baryon - to - photon ratio of number densities has been measured @xcite with precision , i.e. @xmath40 to get the correct value of @xmath41 , one needs @xmath38 to be of order @xmath42 to @xmath43 . however
, using the di upper bound on the cp asymmetry parameter @xcite |_1| [ cp - bound ] one can get a lower bound on the mass scale of the lightest right - handed neutrino as @xmath44 where @xmath45 is the vacuum expectation value ( vev ) of the sm higgs and it is also assumed that the physical left handed neutrinos follow the normal mass hierarchy . since @xmath18 is not likely to exceed @xmath46 gev , this poses a problem for canonical leptogenesis . in order to overcome this problem we consider an extended seesaw model as follows .
in a recent paper @xcite we proposed a singlet mechanism to overcome the di bound shown in eq .
( [ di - bound ] ) .
we now present a realistic model where this mechanism can be implemented .
one more ingredient has been added to produce these singlet fields abundantly in a thermal bath .
we start with the nmssm model , which includes a singlet superfield @xmath20 in addition to the usual particles of the minimal supersymmetric standard model ( mssm ) . to implement the seesaw mechanism ,
we also include three right - handed neutrinos @xmath6 .
we then demonstrate how a minimal extension of this model may admit a very low scale of leptogenesis , thus overcoming the gravitino problem .
we include another singlet superfield @xmath22 and impose a @xmath47 discrete symmetry . under @xmath0 ,
the lepton superfields @xmath48 are odd , whereas the higgs superfields @xmath49 are even .
this corresponds to having an exactly conserved lepton number @xmath21 , or the usual @xmath50parity of the mssm . under @xmath23
, @xmath22 is odd and all others are even , but @xmath23 is allowed to be broken softly .
the most general superpotential invariant under @xmath47 is then given by @xmath51 we do not discuss quarks nor other interactions of the nmssm , which have been studied elsewhere .
we deal only with neutrino masses and leptogenesis .
we now break @xmath23 softly and the only possible such term is @xmath52 i.e. exactly as required to implement the singlet mechanism of ref . @xcite .
this allows @xmath22 to mix with @xmath11 to form a @xmath53 mass matrix in the basis @xmath54 $ ] , i.e. @xmath55 where @xmath56 , and without loss of generality we choose @xmath57 to be diagonal with masses @xmath58 . for small @xmath59 as well as
the usual seesaw assumption that the entries of @xmath60 are very small relative to @xmath61 , the heavy masses are roughly given by @xmath62 corresponding to the mass eigenstates @xmath63 and @xmath64 @xmath65 the light neutrino mass matrix is then ( m_)_ij -_k ( m_d)_ik ( m_k + ) ^-1 ( m_d)_kj .
[ neutrino_mass ] in the limit @xmath66 we recover the neutrino masses as in the canonical seesaw mechanism .
we assume thus @xmath67 and @xmath68 in the following .
in this model , the addition of @xmath22 allows the choice @xmath70 .
the induced couplings of @xmath22 to leptons are suppressed by factors of @xmath71 compared to those of @xmath11 .
the decay rate of @xmath22 is thus given by _
d^s = m_s _ i , [ gamma_d ] where we have assumed @xmath72 and the effective neutrino mass parameter is defined as _ i=. [ effective - mass ] assuming that @xmath73 and @xmath74 the above eq .
( [ gamma_d ] ) can be rewritten as _
d^s m_s _ 3 m_3 ( d_3/m_3)^2 .
[ final_gamma ] the out - of - equilibrium condition @xmath75 is thus suppressed by a factor @xmath76 in comparison to @xmath77 and hence can be satisfied at a lower mass depending on the value of @xmath78 . the value of @xmath79 , @xmath34 and @xmath80 can be approximately fixed from the low energy neutrino oscillation data .
so , the remaining free parameters are @xmath78 and @xmath69 on which the suppression factor @xmath81 depends .
1.25 in -.95 in -.5 in the cp asymmetry generated by the decays of @xmath22 comes from the interference of the tree - level and one - loop diagrams of fig .
( [ lepfg ] ) . both the numerator and denominator of eq .
( [ epsilon_1 ] ) are then suppressed by the same @xmath82 factor , and we obtain @xmath83 ^ 2}{(m_d^\dagger m_d)_{11}}\ , , \label{epsilon_s}\ ] ] where we have assumed that @xmath84 . thus we see that the cp asymmetry parameter is independent of the suppression parameter @xmath78 .
therefore , depending on the value of @xmath85 the @xmath17-asymmetry will saturate at different temperatures as implied by eq .
( [ suppression - factor ] ) .
this is shown in section v by numerically solving the required boltzmann equations . as demonstrated in fig .
( [ allowed_ms ] ) the value of @xmath69 can be lowered much less than the di bound on @xmath19 by an appropriate choice of @xmath78 .
this is because the low values of @xmath69 are not restricted by the low energy neutrino oscillation data for @xmath86 as we have seen from eq .
moreover the washout effects are suppressed for low values of @xmath87 .
so , a successful lepton asymmetric universe before the electroweak phase transition can be created even for a tev scale of @xmath22 .
in this model @xmath22 is produced through the decay of @xmath20 .
the corresponding yukawa coupling @xmath88 can be as large as of order unity .
therefore , @xmath22 can be brought to thermal equilibrium through the scattering processes : @xmath89 , @xmath90 and @xmath91 .
note that these processes never change the number density of @xmath22 , but they keep @xmath22 in kinetic equilibrium .
the decay rate of @xmath20 can be given by @xmath92 where @xmath93 is the boost factor and @xmath94 .
thus the inverse decay of @xmath20 is given by @xmath95 .
once the @xmath22 particles are produced , they decay through the channel : @xmath96 as shown in fig .
( [ lepfg ] ) which violates lepton number by two units .
apart from that , the other process which depletes the number density of @xmath22 is @xmath97 .
this is shown in fig .
( [ delta_l=1 ] ) . the subsequent decay of @xmath22 , below its mass scale , then produces the required baryon asymmetry through the leptogenesis route .
however , an exact lepton asymmetry can be estimated by solving the required boltzmann equations @xcite .
it is useful to define the boltzmann equations in terms of the dimensionless variables @xmath98 and @xmath99 , where @xmath100 is called the comoving density of @xmath22 while @xmath101 is the density of net lepton in a comoving volume and s = g _ * t^3 [ entropy ] is the entropy density .
the required boltzmann equations are given as = -(y_s - y_s^eq ) [ s - evolution ] and = _ s ( y_s - y_s^eq ) -y_l , [ l - evolution ] where @xmath102 , @xmath103 and @xmath104 simultaneously represent the decay , scattering and wash out rates that takepart in establishing a net lepton asymmetry in a thermal plasma .
the hubble parameter @xmath105 is given by h(z)= with h(m_s)=1.67 g_*^1/2 .
[ hubble_mod ] in a relativistic frame the decay rate ( [ final_gamma ] ) can be rewritten as _
d^s = m_s ( ) _ 3 m_3 ( d_3/m_3)^2 .
[ final_decay ] the @xmath103 in eq .
( [ s - evolution ] ) represents the processes which violate lepton number by one unit and is given by _ s=4^s_,s+2^s_,t .
[ gamma_s ] the @xmath106 in eq .
( [ l - evolution ] ) represents the lepton number violating processes by two units and is given by _ w=^s_id+2^l_,t+2^l_,s ( ) + 2^l_n1 + 2^l_n1,t , [ washout ] where @xmath107 . in eqs .
( [ gamma_s ] ) and ( [ washout ] ) the @xmath108 s are defined as @xmath109 where @xmath110 is the scattering density .
note that the other @xmath111 processes : @xmath112 , and of course higher order processes , which contribute to @xmath104 are suppressed in comparison to the processes shown in fig .
( [ washoutfig ] ) . in fig .
( [ decay_fig ] ) we have plotted the decay and inverse decay of @xmath20 and @xmath22 against @xmath113 . it is shown that the inverse decay of @xmath114 is not sufficient to bring @xmath22 into thermal equilibrium even if the suppression factor @xmath78 is as large as @xmath115 gev .
on the contrary , the decay rate of @xmath20 is sufficiently larger than the hubble expansion parameter .
hence @xmath22 can be brought to thermal equilibrium through the scattering process involving @xmath20 as long as @xmath116 .
therefore , at a temperature far above the mass scale of @xmath22 it is in thermal equilibrium and hence a net lepton asymmetry in the thermal plasma is zero . below the mass scale of @xmath22
the lepton number violating processes go out of thermal equilibrium and thus produce a net lepton asymmetry dynamically .
this is obtained by solving the boltzmann equations ( [ s - evolution ] ) and ( [ l - evolution ] ) .
we take the following initial conditions : y_s = y^eq_s and y_l=0 at z0 .
[ initial - condition ] the evolution of the number density of @xmath22 and the corresponding asymmetry with respect to @xmath113 are shown in figs .
( [ d38 ] ) , ( [ d37 ] ) and ( [ d36 ] ) . at any epoch
@xmath113 the value of @xmath100 and the corresponding asymmetry @xmath101 can be inferred from @xmath117 since the decay rate of @xmath22 is suppressed by a factor of @xmath85 , the asymmetry is produced at late times depending how small it is .
however , the value of @xmath85 can not be made indefinitely small . because a net @xmath17-asymmetry has to build up before the electroweak phase transition which is required to be converted to the @xmath16-asymmetry through the sphaleron transitions .
the final @xmath17-asymmetry is numerically obtained for three values of @xmath78 in figs .
( [ d38 ] ) , ( [ d37 ] ) and ( [ d36 ] ) . it is found that the final @xmath17-asymmetry is almost same apart from a numerical factor . this is because for the delayed decay of @xmath22 the wash out effects are comparatively small . while @xmath118 and @xmath119 are used in figs .
( [ d38 ] ) and ( [ d37 ] ) , it is of @xmath42 in fig .
( [ d36 ] ) . as seen from figs .
( [ d38 ] ) , ( [ d37 ] ) and ( [ d36 ] ) , for @xmath120 the value of @xmath101 is saturated at around @xmath121 respectively . assuming that a final @xmath17-asymmetry has to be produced before @xmath122 gev the minimum tolerable value of @xmath123 is obtained .
this indicates that for @xmath124 gev , @xmath125 and @xmath69 can be readjusted among themselves so as to get the suppression factor @xmath85 ranging from @xmath126 to @xmath127 .
this is shown in fig.([allowed_ms ] ) . the solid line in fig .
( [ allowed_ms ] ) is obtained for @xmath128 .
the region above to that are defined by @xmath129 .
so these values of @xmath78 are unnatural and are not allowed . while the region below to the solid line are defined by @xmath130 and hence is allowed for naturalness .
thus we see that a wide range of @xmath69 values from @xmath131 gev to @xmath115 gev are allowed that can produce the required lepton asymmetry before the electroweak phase transition .
we accomplished the baryogenesis via leptogenesis from the decay of an additional singlet @xmath22 in a supersymmetric extended nmssm .
the bound coming from the out - of - equilibrium condition could be evaded because the couplings of the singlets cancel out from the asymmetry , so the couplings could be small and can satisfy the out - of - equilibrium condition even at low scales . in the simplest seesaw models the couplings of the lightest right - handed neutrino could not be lowered much because that will not enable the thermal production of these fields .
however , in the present case there is one additional singlet field ( @xmath20 ) which can produce these @xmath22 fields , having large couplings to them , but itself not taking part in leptogenesis .
the work of em was supported in part by the u. s. department of energy under grant no .
de - fg03 - 94er40837 .
in this appendix we give the various scattering densities that have been used in the boltzmann eqs .
( [ s - evolution ] ) and ( [ l - evolution ] ) .
_ , s & = & ( ) ^2 _ 3 _
1^dx_1 k_1(z)^2 + _ , t & = & ( ) ^2 _ 3 _ 1^dx_1 k_1(z ) where @xmath45 is the vev of sm higgs and @xmath132 , @xmath133 being the mandelstam variable , and @xmath134 . _ n1 & = & _ 0^ dx_2 k_1 ( z ) + & & + _ n1,t & = & _ 0^dx_2 k_1 ( z ) where @xmath135 and = m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ , eds .
p. van nieuwenhuizen and d. z. freedman ( north - holland , 1979 ) , p. 315 ; t. yanagida , in _ proc . of the workshop on the unified theory and the baryon number in the universe _
o. sawada and a. sugamoto , kek report no .
79 - 18 ( tsukuba , japan , 1979 ) , p. 95 ; r. n. mohapatra and g. senjanovic , phys . rev
. lett . * 44 * , 912 ( 1980 ) .
m. y. khlopov and a. d. linde , phys .
* b138 * , 265 ( 1984 ) ; j. r. ellis , j. e. kim , and d. v. nanopoulos , phys .
* b145 * , 181 ( 1984 ) ; j. r. ellis , d. v. nanopoulos , and s. sarkar , nucl . phys .
* b259 * , 175 ( 1985 ) ; m. bolz , a. brandenburg and w. buchmuller , nucl .
b * 606 * , 518 ( 2001 ) .
a. pilaftsis , phys . rev . *
d56 * , 5431 ( 1997 ) ; m. bolz , w. buchmuller , and m. plumacher , phys . lett . *
b443 * , 209 ( 1998 ) ; m. ibe , r. kitano , h. murayama , and t. yanagida , phys . rev . * d70 * , 075012 ( 2004 ) ; t. dent , g. lazarides , and r. ruiz de austri , phys . rev . * d72 * , 043502 ( 2005 ) ; y. farzan and j. w. f. valle , phys . rev . lett . * 96 * , 011601 ( 2006 ) ; n. okada and o. seto , phys . rev . * d73 * , 063505 ( 2006 ) .
a. abada , s. davidson , f .- x .
josse - michaux , m. losada , and a. riotto , jcap * 0604 * , 004 ( 2006 ) .
see also e. nardi , y. nir , e. roulet , and j. racker , jhep * 0601 * , 164 ( 2006 ) ; o. vives , phys . rev . * d73 * , 073006 ( 2006 ) . n. sahu and u. a. yajnik , phys
. rev .
d * 71 * , 023507 ( 2005 ) , phys .
b * 635 * , 11 ( 2006 ) ; n. sahu and u. sarkar , phys .
d * 74 * , 093002 ( 2006 ) , [ arxiv : hep - ph/0605007 ] ; m. frigerio , t. hambye and e. ma , jcap * 0609 * , 009 ( 2006 ) ; e. j. chun , phys .
d * 72 * , 095010 ( 2005 ) ; l. boubekeur , t. hambye and g. senjanovic , phys .
lett . * 93 * , 111601 ( 2004 ) . | thermal leptogenesis in the canonical seesaw model in supersymmetry suffers from the incompatibility of a generic lower bound on the mass scale of the lightest right - handed neutrino and the upper bound on the reheating temperature of the universe after inflation .
this is resolved by adding an extra singlet superfield , with a discrete @xmath0 symmetry , to the nmssm ( next to minimal supersymmetric standard model ) .
this generic mechanism is applicable to any supersymmetric model for lowering the scale of leptogenesis .
ucrhep - t423 + november 2006 1.0 cm |
perovskite manganites , with a general formula @xmath1mno@xmath2 , where @xmath1 is the composite of trivalent rare - earth elements and divalent alkaline - earth elements , have attracted considerable attention since the discovery of the colossal magnetoresistance ( cmr ) effect.@xcite both experimental and theoretical studies in the last decade have unveiled a plethora of phases in manganite compounds , with very different macroscopic properties but very similar energies.@xcite the cmr and colossal electroresistance ( cer)@xcite effects , which correspond to the observation of drastic nonlinear responses of the manganites to external stimulations , can be understood as a result of the intense competition between the ferromagnetic ( fm ) metallic phase and antiferromagnetic ( afm ) charge - ordered ( co ) insulating phases.@xcite this phase competition is not only sensitive to applied external fields but also to the geometric and chemical environments of the surface of the system under study . in nano - sized materials
there is a high surface to volume ratio and , as a consequence , the surface effects play a crucial role .
this influence of the surface has been observed in a series of recent experiments .
on one hand , in materials where a fm state is stabilized in the bulk , an afm or spin - glass surface state is found.@xcite on the other hand , fm tendencies at the surface of nano - sized manganites presenting afm / co bulk order have also been observed.@xcite the first tendency toward surface afm ordering usually appears in nano - sized fm or ferrimagnetics materials,@xcite and can be understood within the following naive picture . generally , in strongly correlated electron materials , the charge conducting properties are determined by the ratio @xmath3 , where @xmath4 is the hubbard coulomb repulsion and @xmath5 is the electron hopping bandwidth . at the surface , the bandwidth @xmath5 will be suppressed due to the reduction in dimensionality , while the on - site @xmath4 will not change .
therefore , the enhancement of @xmath3 at the surface could prefer an insulating surface over a metallic one .
the insulating phase in manganites is usually afm .
however , finding a model and rational for the other tendency found experimentally , namely a fm tendency at the surfaces of ce manganites , is not straightforward . a theoretical model that can explain the fm tendency at the surface should be able to consider the following four experimental signatures
first , the co phase is significantly weakened .
the co transition peak in the magnetization ( @xmath6 ) _ vs. _ temperature ( @xmath7 ) curve is suppressed until it completely disappears with decreasing the size of the manganite systems.@xcite second , @xmath6 is enhanced at low @xmath7 and a fm - type hysteresis loop is observed.@xcite third , the exchange bias effect emerges , indicating a coupling between phases with different spin orders.@xcite and finally , the measurement of magnetocaloric properties and low-@xmath7 specific heat also suggest the existence of a fm contribution.@xcite theoretically , previous investigations have mainly paid attention to the surface effects of fm manganites @xcite whereas the study of the surface of afm manganites was rare . to understand these phenomena , it is reasonable to partition the nano - sized ( with typical scales @xmath8 nm ) system into two regions : an inner core and a surface shell .
the physical properties of the inner core should be comparable to those of the bulk material .
in contrast , the properties of the surface shell can be different from those of the bulk .
the lower coordination number at the surface , which effectively reduces the superexchange coupling , may give rise to a fm tendency at that surface .
following this idea , in previous investigations a core - shell model was proposed assuming an afm core wrapped by a fully fm surface shell.@xcite this model fits some experimental results well , but in this scenario it is already established from the start the nature of the phases at both the core and surface . as a consequence ,
it is too phenomenological to better understand the true physical origin of the afm - fm transition at the surface . in particular , the assumption of a fully fm shell conflicts with some experimental results : due to the weak value of the magnetization @xmath6 , the calculated thickness of the fm shell can be even thinner than one `` molecular '' layer , which indicates that assuming a fully developed fm spin order at the surface is not correct .
thus , theoretical studies using realistic microscopic hamiltonians and unbiased assumptions about the surface are necessary to clarify the experimental observations found at the surface of nano - sized afm / co manganites . in this paper ,
the core - shell model is incorporated into a two - orbital hamiltonian for manganites .
we show that the unscreened coulomb interactions lead to an increase of electron density on the surface .
monte carlo simulations reveal that this increase in the density drives the surface layer from an afm / co state to an unexpected phase - separated state , as opposed to a fully developed fm state .
this surface phase - separated state exhibits clear fm signatures , but they are weak , compatible with the experimental observations .
to better understand the physics at the surface , here we consider a two - orbital model hamiltonian for manganites that includes both finite superexchange coupling and the effect of jahn - teller phonons . as a well - accepted approximation for manganite models , we consider the limit of infinite hund coupling .
the hamiltonian reads : @xmath9 here , the first term is the two - orbital double exchange interaction . @xmath10 and @xmath11 denote the two mn @xmath12-orbitals @xmath13 ( @xmath14 ) and @xmath15 ( @xmath16 ) .
@xmath17 ( @xmath18 ) annihilates ( creates ) an @xmath12 electron in orbital @xmath13 of site @xmath19 with its spin parallel to the localized @xmath20 spin @xmath21 .
the hopping direction is denoted by * r*. as discussed in previous literature,@xcite the hopping amplitudes are @xmath22 , @xmath23 , @xmath24 and @xmath25 ( energy unit ) .
the infinite hund coupling generates the factor @xmath26 $ ] , where @xmath27 and @xmath28 are the angles of the @xmath20 spins in spherical coordinates .
the second term is the superexchange interaction between nearest - neighbor ( nn ) @xmath20 spins . in the third term
, @xmath29 is the chemical potential .
@xmath30 corresponds to a site dependent coulomb potential .
the origin and relevance of this term will be discussed in detail later in this section .
@xmath31 is the @xmath12 charge - density at site @xmath19 .
the fourth term stands for the electron - phonon coupling .
the @xmath32s are phonons corresponding to jahn - teller modes ( @xmath33 and @xmath34 ) and the breathing mode ( @xmath35 ) .
@xmath36 is the orbital pseudospin operator , giving @xmath37 and @xmath38 .
the last term is the elastic energy of the phonons . for simplicity ,
we have already assumed in this model that both the @xmath20 spins and the phononic degrees of freedom are classical variables .
the above described hamiltonian is solved via a combination of exact diagonalization and monte carlo ( mc ) techniques : classical @xmath20 spins and phonons evolve following the mc procedure ; and at each mc step , the fermionic sector of the hamiltonian is numerically exactly diagonalized .
the first @xmath39 mc steps are used for thermal equilibrium and another @xmath40 mc steps are used for measurement . more details of the hamiltonian and mc technique can be found in refs .
o sheet termination is considered here .
thus , the surface formula is @xmath41mno@xmath42 . ]
since a fm tendency is often found in half - doped @xcite or nearly half - doped@xcite nano - sized manganites , and since a ce - type afm / co phase usually appears in half - doped narrow band manganites,@xcite we solve the two - orbital model using densities corresponding to half - doped systems and for couplings where a bulk ce phase exists .
this can be done by tuning the couplings ( @xmath43 , @xmath44 ) to be within the ce regime of the phase diagram obtained in previous studies.@xcite although only one set ( @xmath43 , @xmath44 ) will be investigated in the studies described below , we believe our qualitative conclusions remain valid for other choices of parameter sets , as long as they are within the the ce region of the phase diagram .
although there are several possible surface directions in real cases , here only the simplest case , i.e. , the ( 001 ) surface , will be considered , similarly as in most former theoretical investigations in this context .
@xcite strictly speaking , the surface problem should be considered on a three - dimensional half - infinite cubic lattice , e.g. infinite in the @xmath45 and @xmath46 directions , but semi - infinite in the @xmath47 direction . in practice , to address this problem numerically the above two - orbital model hamiltonian is studied using a @xmath48 cubic lattice , as shown in fig .
1 . periodic boundary conditions ( pbc ) are applied in both the @xmath45 and @xmath46 directions . following the core - shell - model idea discussed in the previous section , in the @xmath47 direction the spins in the bottom layer ( @xmath49 )
are here _ fixed _ so that they have the same ce - type afm pattern as in the bulk . by this procedure ,
the properties of the half - infinite core are encoded in the bottom layer in our model .
hence , the rest @xmath50 layers are to be considered as the outer shell . to better justify the above assumption , @xmath51 must be large enough that surface effects are limited within the outer @xmath50 shell layers .
our results below suggest that in practice @xmath52 is enough for these purposes .
this is fortunate , since the numerical studies described here are rather cpu time consuming . for the surface layer ( @xmath53 )
, we have to take into account the `` termination '' procedure . in this paper , a clean - limit @xmath1o sheet
is considered as the termination , which makes the `` molecular composition '' of the surface layer to be @xmath41mno@xmath42 .
therefore , the outmost @xmath1o sheet transfers an extra @xmath54 electron per site to the nearby mn cations , and it is positively polarized .
the polarized @xmath1o sheet introduces an unscreened coulomb attraction on the surface . as a result
, those extra electrons must be localized near the surface .
the effect of this surface coulomb attraction has been taken into account here via an effective negative potential near the surface : @xmath55 when @xmath19 belongs to the surface layer and @xmath56 otherwise .
this may be justified because of the typical short thomas - fermi - like screening length found in many of these materials .
the boundary condition on the surface layer is then set to be open .
the consideration of the @xmath1o sheet has another important effect : it keeps the oxygen octahedrons complete for the outmost mn cations .
therefore , the phonon modes do not need to be changed , even for the surface layer .
for the phonons , pbc are used in all directions for simplicity .
this will not affect much the central physics discussed in this manuscript since the oxygen displacements along the @xmath47 axis are negligible in both the fm and ce type afm phases.@xcite in addition to the @xmath1o termination , the other possible choice for the surface is a clean - limit mno@xmath57 sheet as the termination , but this will not be considered in our current model .
however , the possible effects of this alternative termination will also be discussed in the next section . to contrast results with those of the open surface case
, the model will also be studied following assumptions that address the bulk material : this corresponds to using pbc in all directions , for both spins and phonons , and without freezing spins anywhere .
both simulations are performed at @xmath58 ( experimentally , it corresponds to @xmath59 k ) . to simulate the bulk material , we set @xmath56 for all sites . the parameters @xmath60 and @xmath61 are used to obtain a stable ce phase with average density @xmath62 .
note that it is well known that the ce phase is stable over a broad range of couplings of the half - doped manganites , thus this selection of parameters should not be considered arbitrary or fine tuned .
the stability of the ce phase is also confirmed in the present simulations by analyzing the spin and charge structure factors , which will be discussed in detail in the next section . as observed in fig .
2(a ) , this corresponds to choosing the chemical potential in the window @xmath63 .
these parameters will also be adopted in our subsequent simulation of the open surface model .
to obtain reasonable results from the simulation of the open surface model , we have to set the coulomb potential to an appropriate value .
actually , an optimal value of @xmath64 exists in order to fulfill the following three criteria : first , the average density per site of the entire @xmath48 system @xmath65 should be equal to @xmath66 to keep the charge neutral ; second , the chemical potential should remain within ( @xmath67 , @xmath68 ) , which is required to have the ce phase stable far from the surface ; third , the charge density in the bottom layer should be very close to @xmath69 to match the frozen ce type . in order to find the optimal @xmath64 ,
we tested several values from @xmath70 to @xmath71 , stepped by @xmath72 . in this range , @xmath73 was found to be a proper parameter at @xmath74 .
thus , @xmath73 was the value adopted in the following simulations .
charge - density as a function of the chemical potential using @xmath52 .
( a ) is for the case of the bulk material model .
the corresponding nn spin - correlation is also shown .
there is a @xmath75 plateau with the nn spin - correlation @xmath76 in the range @xmath63 , indicating a stable ce phase .
( b ) contrasts the results for the open - surface and the bulk material models near half - doping . ]
using as initial configuration a perfect ce spin pattern @xcite and random phonons , the mc simulation was carried out under the assumption @xmath73 for the open - surface model . the averaged @xmath12 charge - density as a function of @xmath29
is shown in fig .
2(b ) , allowing us to compare the results for the bulk material model with those of the open surface .
an interesting result is that the @xmath0=@xmath77 plateau in the bulk model , which corresponds to the stable ce phase , disappears in the open - surface model . the charge density increases with increasing chemical potential but it has large fluctuations ( see error bars ) , suggesting that the system presents a strong competition between phases with different densities . in the following ,
we will focus on the properties at @xmath74 , with the average total charge density close to @xmath78.@xcite to understand the origin of the different average charge density between bulk and open - surface models shown in fig .
2(b ) , it is important to analyze the @xmath12 charge - density at each layer .
the results are shown in fig .
the charge density is almost exactly @xmath69 at the bottom layer and fluctuates around @xmath69 in the two middle layers .
the most prominent change occurs at the surface layer , where the charge density increases to @xmath79 , due to the presence of the coulombic term .
as expected , the extra @xmath54 electron from the outmost @xmath1o sheet is mainly located in the first mn sheet , which offsets the coulomb interaction arising from the outmost @xmath1o sheet for the mn cations of the second layer .
this is consistent with @xmath80-@xmath81 calculations that show that for the @xmath1o termination the uncompensated electrons are accumulated mainly at the surface.@xcite this result is also consistent with the assumption that @xmath30 is nonzero only for the first layer .
charge - density for each layer ( counted from the surface ) .
( b ) the corresponding averaged nn spin - correlation , within ( integer index ) and between ( half - integer index ) layers . ]
it should be noted that the density @xmath79 usually corresponds to a fm phase in bulk manganites , as indicated in fig .
if the bulk phase diagram remains valid at the surface , we would expect a fm state there .
then , the fm tendency could be naturally explained as a result of a density - driven transition . to verify this possible fm tendency ,
the average nn spin - correlation as a function of layer index is shown in fig .
starting from the bottom layer ( @xmath52 ) , we find that the in - plane nn spin correlations are almost zero for layer @xmath52 and @xmath82 , implying that the fm and afm links have almost the same population .
the inter - layer correlation between these two layers is close to @xmath83 , indicating a fully afm connection between the two layers .
these are consistent with the picture that a ce afm state is stabilized in these two layers , given that the charge density is about @xmath69 .
interestingly , we do see an increase of both the in - plane and inter - layer nn spin - correlations as we approach the surface .
both the correlations in the first layer and between the first and second layers take positive values , and they display a clear fm tendency at the surface layer .
but the positive value for the in - plane correlation is rather small ( @xmath84 ) at the surface layer , which suggests that the state is only _ partially _ fm .
therefore , the idea of a density - driven transition is too simplistic and not quite correct .
this already shows an interesting conclusion of our research : the phase diagram at the surface can not be obtained by merely analyzing the bulk phase diagram at the appropriate charge density , but a special investigation is needed to clarify the surface s properties .
charge - distribution corresponding to the surface layer ( left ) and bottom layer ( right ) . here
are shown both a typical mc `` snapshot '' ( upper panels ) and mc averaged results ( lower panel , averaged over @xmath85 mc steps ) .
the size of the circles is in proportion to the local charge density @xmath86 . for the surface layer
, the dotted lines separate the high density ( @xmath87 ) and low density ( @xmath88 ) regions .
( b ) sketch of the orbital occupation ( based on the mc - averaged @xmath89 and @xmath90 ) corresponding to the above - described mc - averaged charge distribution .
( c ) mc - time evolution after thermal equilibrium of the @xmath12 charge - density ( average value for all the sites ) .
the frequent tunneling events are prominent among the possible densities , indicating tendencies toward electronic phase separation , which is a first - order transition when @xmath29 is varied . ] to visualize the nature of the surface state more explicitly , we study the distribution of local charge density on the surface layer . in a fm phase ,
the distribution of local charge density @xmath86 is approximately uniform , in contrast to the charge disproportion typical of an afm / co phase.@xcite the distribution of local charge density on the surface layer of the model studied here is presented in fig .
4(a ) , and the result at the bottom layer is also shown in the same figure for contrast . from the regular charge pattern ,
it is clear that the ce charge ordering in the bottom layer is very stable .
since the ce spin pattern is fixed at the bottom layer , the charge disproportion between large - density sites ( bridge - sites of the zig - zag chains ) and smalle - density sites ( corner - sites of the zig - zag chains ) will not be smeared by mc average . however , for the surface layer the charge distribution is not uniform , and it is not regularly distributed . in some sites , the densities are large and close to @xmath91 , but in other sites the densities can be as low as approximately @xmath77 .
this inhomogeneous distribution persists prominently even in the mc - averaged result , which rules out the possibility of observing a non - uniform charge state due to thermal fluctuations .
this inhomogeneity can also be confirmed from the orbital occupation , as shown in fig .
in contrast to the distinct ce type orbital - ordering in the bottom layer , the orbital distribution in the surface layer shows two regions corresponding to charge inhomogeneity . the high - density region shows an orbital ordering similar to the case in undoped manganites , in contrast to the low - density region which shows the orbital disorder similar to the case in fm manganites .
in addition , the mc - time evolution of the @xmath12 charge - density is presented in fig .
4(c ) , showing that the tunneling events are prominent among several possible densities .
these tunneling events are characteristic of a first - order phase transition varying @xmath29 , rather than standard thermal fluctuations.@xcite therefore , the inhomogeneous charge distribution at the surface should be attributed to tendencies in the model toward nanoscale electronic phase separation,@xcite similarly as those observed in bulk simulations in other regions of parameter space . ,
@xmath92 ) as a function of the layer index , showing the reduction of the ce - phase staggered charge - order tendencies as the surface is reached . ] to quantitatively reveal the competing phases that appear at the surface , the spin structure factor @xmath93 for each layer is calculated via a fourier transformation of the spin correlation function:@xcite @xmath94 by monitoring the @xmath95 dependent spin structure factor , we may detect possible weak spin - order signals in the complicated real - space spin pattern@xcite because each spin order corresponds to a unique set of characteristic @xmath95 vectors.@xcite in fig .
5(a ) , the @xmath93s for some possible components ( summed over all characteristic @xmath95 vectors for each spin order ) are shown layer by layer .
for the bottom layer , which is frozen into a ce spin pattern , both the c and e components contribute @xmath96 , respectively , as expected .
the ce phase tendencies become gradually weaker with decreasing layer index , indicated by the decreasing values and large fluctuations of the corresponding @xmath93s . at the surface ,
the c component is very weak ( @xmath97 ) and the dominating components here are fm ( @xmath98 ) and e ( @xmath99 ) . in this case , the e value does not match the c value anymore because the site regions where @xmath100 can also contribute to the e type order.@xcite in addition to the @xmath101 spin structure factor , the charge structure factor @xmath102 in each layer is also calculated to characterize the charge ordering : @xmath103 where @xmath104 is the average density of each layer . here , only @xmath105(@xmath92,@xmath92 )
is shown in fig .
5(b ) since all other components are very close to zero . for the ce phase at the bottom layer , @xmath105(@xmath92,@xmath92 ) is about @xmath106 , a result consistent with the expected charge - ordering with charge disproportion @xmath107 ( the charge difference between high- and low - density sites ) . decreasing the layer index , @xmath105(@xmath92,@xmath92 ) decreases monotonically as we move toward the surface , until it completely disappears at the surface .
this is also straightforward to understand since charge - ordering usually accompanies the afm phases instead of the fm ones , even in bulk manganites . in the above simulation
, the ground state is a robust co ce phase by fixing the proper @xmath43 and @xmath44 .
therefore , it is worth to address the other possible cases in real half - doped manganites . here
we will give a brief analysis based on the half - doped phase diagram obtained in previous works.@xcite on one hand , by decreasing @xmath43 only , the ground phase can change from the ce to a - type afm , then finally to fm phase .
this process corresponds to the experimental observed transition from the narrow band manganites to middle band one , then finally to wide band manganties.@xcite on the other hand , by decreasing @xmath44 only , the charge - ordering ( or the degree of charge disproportionation ) will become weaker and weaker , till completely turn to the fm phase when ( @xmath43 , @xmath44 ) crosses the phase boundary between ce and fm phases . in both cases ,
fm tendency will be enhanced . in short ,
our above simulation mainly aims at the family of narrow band manganties whose co ce phase is stable , e.g. nd@xmath108ca@xmath108mno@xmath2 and pr@xmath108ca@xmath108mno@xmath2 . in the simulations described above , only the @xmath1o sheet termination was considered .
but it is necessary to discuss the other choice already mentioned , namely a mno@xmath57-sheet termination . in the case of this mno@xmath57 termination
, the cubic symmetry is lost at the surface due to the breakdown of the oxygen octahedrons .
therefore , the @xmath109 energy levels of the mn cations at the surface are different from those with full oxygen octahedrons . in particular , the energy of the @xmath16 orbital will be lowered substantially .
an early model study by caldern _
et al _ showed that the mno@xmath57 termination would generate an afm surface for fm manganites because the @xmath12 density at the surface was enhanced from @xmath110 to @xmath91.@xcite however , an @xmath80-@xmath81 calculation by fang _
et al . _ showed the reverse result : decreased @xmath12 density at the surface by mno@xmath57 termination.@xcite therefore , whether the mno@xmath57 termination can generate the fm tendency , e.g. by enhancing the surface charge density from @xmath77 to @xmath79 as it occurs in the case of the @xmath1o termination , remains unclear and is an interesting subject of investigations .
however , this mno@xmath57 termination is far more complex than the case studied here , and beyond the scope of the present work .
it is important to remark that some other extrinsic factors , such as defects of cations / oxygen and recomposition of surface structures , may also affect the physical properties of real manganites .
even qualitatively considering these effects , our model still gives reasonable results . for a crude comparison with experiments ,
the fm component fraction can be estimated as @xmath111 of the surface layer ( fig .
therefore , to compare with experiments the fm fraction predicted by our study we should use the number @xmath112surface / volume ( at zero magnetic field and low @xmath7 ) , where surface / volume should be calculated based on the actual shape and size of the nano - sized clusters used experimentally . by this procedure , our theory agrees with the weak magnetization found in nd@xmath108ca@xmath108mno@xmath2 nanoparticles ( diameter @xmath113 nm , experimental magnetization @xmath114 emu / g @xmath115 of the saturation magnetization @xmath116 , while in our model our estimation gives @xmath117)@xcite and pr@xmath108ca@xmath108mno@xmath2 nanowires ( diameter @xmath118 nm , experimental magnetization @xmath119 emu / g @xmath120 , while our model estimation is @xmath120).@xcite it should be noted that the phenomenological core - shell model can not explain these very weak magnetizations.@xcite the agreement between our estimates and experimental data suggests that our model at least grasps the main physics of the surface effects in the nano - sized ce - phase manganites .
in conclusion , we have performed a monte carlo study of the ce - type afm / co phase in a 3d lattice with an open surface .
the @xmath1o sheet termination leads to the generation of an extra @xmath54 electron - per - site at the surface , here simulated by the introduction of an unscreened coulomb attraction at that surface . as a result
, the charge density on the surface mn - layer was enhanced from @xmath69 to @xmath121 .
the charge density @xmath121 usually corresponds to a fully fm phase in bulk manganites .
however , within the monte carlo simulations for small clusters discussed in this manuscript , the surface was found to have a nontrivial nanoscale electronic phase separated state . at the surface ,
the charge distribution was found to be inhomogeneous and coexisting with a weak fm spin correlation .
the studies of both the charge structure factor and spin structure factor confirmed the suppression of afm / co order and the enhancement of fm order near the surface .
our result is helpful to understand the weak fm tendencies observed in nano - sized afm / co manganites .
however , clearly these results have to be considered as just a first step toward the understanding of the phase diagram at the surfaces of manganites . larger clusters and other numerical / analytical techniques
should be used to confirm our results and further explore the physics unveiled here .
also , a systematic study varying parameters of the many tendencies expected in the anticipated rich phase diagram of these compounds at the surface should be carried out in future investigations .
we thank w. plummer , m. j. caldern , s. v. bhat , and a. biswas for useful comments .
this work was supported by the nsf grant dmr-0706020 and the division of materials science and engineering , u.s .
doe , under contract with ut - battelle , llc .
s.d . and j.m.l were supported by the national key projects for basic research of china ( 2006cb921802 ) .
was also supported by the china scholarship council and the scientific research foundation of graduate school of nanjing university .
bib s. jin , t. h. tiefel , m. mccormack , r. a. fastnacht , r. ramesh , and l. h. chen , science * 264 * , 413 ( 1994 ) .
e. dagotto , _ nanoscale phase separation and colossal magnetoresistance _
( berlin : springer , 2002 ) .
e. dagotto , t. hotta , and a. moreo , phys .
rep . * 344 * , 1 ( 2001 ) .
e. dagotto , new j. phys .
* 7 * , 67 ( 2005 ) .
j. burgy , a. moreo , and e. dagotto , phys .
lett . * 92 * , 097202 ( 2004 ) .
a. asamitsu , y. tomioka , h. kuwahara , and y. tokura , nature ( london ) * 388 * , 50 ( 1997 ) . c. en , g. alvarez , and e. dagotto , phys .
* 98 * , 127202 ( 2007 ) .
r. yu , s. dong , c. en , g. alvarez , and e. dagotto , phys .
b * 77 * , 214434 ( 2008 ) .
s. dong , h. zhu , and j .- m .
liu , phys .
b * 76 * , 132409 ( 2007 ) .
s. dong , c. zhu , y. wang , f. yuan , k. f. wang , and j .-
liu , j. phys . : condens .
matter * 19 * , 266202 ( 2007 ) .
t. zhang , g. li , t. qian , j. f. qu , x. q. xiang , and x. g. li , j. appl . phys . * 100 * , 094324 ( 2006 ) . t. zhu , b. g. shen , j. r. sun , h. w. zhao , and w. s. zhan , appl .
phys . lett . * 78 * , 3863 ( 2001 ) .
p. dey and t. k. nath , appl . phys
. lett . * 87 * , 162501 ( 2005 ) .
m. l. moreira , j. m. sores , w. m. de azevedo , a. r. rodrigues , f. l. a. machado , and j. h. d. arajo , physica b * 384 * , 51 ( 2006 ) .
m. h. zhu , y. g. zhao , w. cai , x. s. wu , s. n. gao , k. wang , l. b. luo , h. s. huang , and l. lu , phys .
b * 75 * , 134424 ( 2007 ) .
a. biswas , i. das , and c. majumdar , j. appl . phys . * 98 * , 124310 ( 2005 ) .
s. s. rao , k. n. anuradha , s. sarangi , and s. v. bhat , appl .
phys . lett . *
87 * , 182503 ( 2005 ) .
s. s. rao , s. tripathi , d. pandey , and s. v. bhat , phys .
b * 74 * , 144416 ( 2006 ) . c. l. lu , s. dong , f. gao , z. q. wang , j .-
liu , and z. f. ren , appl .
91 * , 032502 ( 2007 ) .
t. zhang , t. f. zhou , t. qian , and x. g. li , phys .
b * 76 * , 174415 ( 2007 ) . t. sarkar , b. ghosh , a. k. raychaudhuri , and t. chatterji , phys .
b * 77 * , 235112 ( 2008 ) . t. zhang , c. g. jin , t. qian , x. l. lu , j. m. bai , and x. g. li , j. mater . chem . * 14 * , 2787 ( 2004 ) .
v. markovich , i. fita , a. wisniewski , r. puzniak , d. mogilyansky , l. titelman , l. vradman , m. herskowitz , and g. gorodetsky , phys .
b * 77 * , 054410 ( 2008 ) . t. sarkar , a. k. raychaudhuri , and t. chatterji , appl .
* 92 * , 123104 ( 2008 ) .
a. biswas , t. samanta , s. banerjee , and i. das , j. appl . phys . * 103 * , 013912 ( 2008 ) . c. l. lu ,
k. f. wang , s. dong , j. g. wan , j .- m .
liu , and z. f. ren , j. appl .
* 103 * , 07f714 ( 2008 ) .
a. biswas , t. samanta , s. banerjee , and i. das , appl .
* 92 * , 212502 ( 2008 ) .
j. nogus , j. sort , v. langlais , v. skumryev , s. suriach , j. s. muoz , and m. d. bar , phys . rep . * 422 * , 65 ( 2005 ) . m. j. caldern , l. brey , and f. guinea , phys .
b * 60 * , 6698 ( 1999 ) .
h. zenia , g. a. gehring , and w. m. temmerman , new . j. phys .
* 9 * , 105 ( 2007 ) .
h. zenia and g. a. gehring , g. banach , and w. m. temmerman , phys .
b * 71 * , 024416 ( 2005 ) . s. dong , f. gao , z. q. wang , j .- m .
liu , and z. f. ren , appl .
. lett . * 90 * , 082508 ( 2007 ) . e. o. wollan and w. c. koehler , phys .
100 * , 545 ( 1955 ) .
s. yunoki , t. hotta , and e. dagotto , phys .
lett . * 84 * , 3714 ( 2000 ) .
k. pradhan , a. mukherjee , and p. majumdar , phys .
lett . * 99 * , 147206 ( 2007 ) .
z. fang and k. terakura , j. phys .
. jpn . * 70 * , 3356 ( 2001 ) . usually , the phononic displacements are prominent in co insulating phases but weak in fm metallic phases . for the ce phase ,
the @xmath122-dimensional ( @xmath45-@xmath46 plane ) nature of the spin pattern causes the oxygen displacements along the @xmath47 direction to be zero . for more details ,
see refs .
the ce zig - zag chains are arranged in the ( 001 ) plane . therefore , the ( 001 ) surface will not break the in - plane zig - zag chains topologically .
this choice can avoid the finite size effects along the [ 100 ] and [ 010 ] directions .
it is very difficult to obtain the exact @xmath0=@xmath78 density due to the phase separation tendencies in the grand canonical ensemble .
the mc - averaged density at @xmath74 with @xmath73 is @xmath123 .
this small discrepancy of charge density does not affect the conclusions of our manuscript .
s. dong , s. dai , x. y. yao , k. f. wang , c. zhu , and j
liu , phys .
b * 73 * , 104404 ( 2006 ) .
e. dagotto , s. yunoki , a. l. malvezzi , a. moreo , j. hu , s. capponi , d. poilblanc and n. furukawa , phys .
b * 58 * , 6414 ( 1998 ) . c. en , g. alvarez , and e. dagotto , phys .
b * 70 * , 064428 ( 2004 ) .
s. kumar and a. p. kampf , phys .
lett . * 100 * , 076406 ( 2008 ) .
( @xmath70 , @xmath92 ) and ( @xmath92 , @xmath70 ) for c type afm , ( @xmath124 , @xmath124 ) ( @xmath124 , @xmath125 ) ( @xmath125 , @xmath124 ) ( @xmath125 , @xmath125 ) for e type afm , and ( @xmath92 , @xmath92 ) for g type afm .
usually , the wavevector @xmath95 for the fm phase is simply ( @xmath70 , @xmath70 ) .
however , for an even number of electrons , the pbc ( in the @xmath45-@xmath46 plane ) can induce an artificial spiral phase with period @xmath126 , while using open boundary conditions the true fm phase is obtained .
@xmath127-@xmath128 for details . to take into account the convergence of the fake spiral phase to the fm phase in the bulk limit , ( @xmath129 , @xmath70 ) and ( @xmath70 , @xmath130
should also be considered as part of the fm phase when using a finite - size system .
j. riera , k. hallberg , and e. dagotto , phys .
79 * , 713 ( 1997 ) .
t. a. kaplan and s. d. mahanti , j. phys . : condens .
matter * 9 * , l291cl297 ( 1997 )
. j. zang , h. roder , a. r. bishop , and s. a. trugman , j. phys .
: condens .
matter * 9 * , l157 ( 1997 ) .
t. hotta , m. moraghebi , a. feiguin , a. moreo , s. yunoki , and e. dagotto , phys .
lett . * 90 * , 247203 ( 2003 ) .
r. kajimoto , h. yoshizawa , y. tomioka , and y. tokura , phys .
b. * 66 * , 180402(r ) ( 2002 ) . | most previous investigations have shown that the surface of a ferromagnetic material may have antiferromagnetic tendencies . however , experimentally the opposite effect has been recently observed : ferromagnetism appears in some nano - sized manganites with a composition such that the antiferromagnetic charge - ordered ce state is observed in the bulk .
a possible origin is the development of ferromagnetic correlations at the surface of these small systems . to clarify these puzzling experimental observations ,
we have studied the two - orbital double - exchange model near half - doping @xmath0=0.5 , using open boundary conditions to simulate the surface of either bulk or nano - sized manganites . considering the enhancement of surface charge density due to a possible @xmath1o termination ( @xmath1 = trivalent / divalent ion composite , o = oxygen ) ,
an unexpected surface phase - separated state emerges when the model is studied using monte carlo techniques on small clusters .
this tendency suppresses the ce charge ordering and produces a weak ferromagnetic signal that could explain the experimental observations . |
the field of quantum plasma physics is becoming of increasing current interest [ @xcite ] , motivated by its potential applications in modern technology ( e.g. metallic and semiconductor nanostructures - such as metallic nanoparticles , metal clusters , thin metal films , spintronics , nanotubes , quantum well and quantum dots , nano - plasmonic devices , quantum x - ray free - electron lasers , etc . ) .
in dense quantum plasmas and in the fermi gas of metals , the number densities of degenerate electrons are extremely high so that their wave functions overlap , and they therefore electrons obey the fermi - dirac statistics .
the collective oscillations in quantum plasmas have been studied by several authors in the past [ @xcite ] with applications to the fermi plasmas in metals and semiconductors , and to electrostatic oscillations in quantum pair plasmas [ @xcite ] .
@xcite studied experimentally the bohm - pines dispersion relation of the electron plasma oscillations by measuring the energy loss of electrons by the excitation of collective modes in metals .
the fermi degenerate dense plasma may also arise when a pellet of hydrogen is compressed to many times the solid density in the fast ignition scenario for inertial confinement fusion [ @xcite ] . since there is an impressive developments in the field of short pulse petawatt laser technology , it is highly likely that such plasma conditions can be achieved by intense laser pulse compression using powerful x - ray pulses . here
ultrafast x - ray thomson scattering techniques can be used to measure the features of laser enhanced plasma lines , which will , in turn , give invaluable informations regarding the equation of state of shock compressed dense matters .
recently , spectrally resolved x - ray scattering measurements [ @xcite ] have been performed in dense plasmas allowing accurate measurements of the electron velocity distribution function , temperature , ionization state , and of plasmons in the warm dense matter regime [ @xcite ] .
this novel technique promises to access the degenerate , the closely coupled , and the ideal plasma regime , making it possible to investigate extremely dense states of matter , such as the inertial confinement fusion fuel during compression , reaching super - solid densities . in this paper , we present a study of the dispersion properties of electrostatic oscillations in a dense quantum plasma , by employing the wigner - poisson model .
we point out the differences between different regimes comprising the relatively low density regime of semiconductor plasmas , and the higher density regimes corresponding to metallic electron densities and laser compressed plasmas , as well as plasmas in dense astrophysical objects such as white dwarf stars .
we here present a derivation of the dispersion relation for electrostatic waves in a degenerate quantum plasma .
the electron dynamics is governed by the wigner equation @xmath0 \\ & & \times\left [ \phi_1\left({\bf x}+\frac{\boldsymbol{\lambda}}{2},t\right ) -\phi_1\left({\bf x}-\frac{\boldsymbol{\lambda}}{2},t\right)\right]f_0({\bf v } ' ) , \label{eq1}\end{aligned}\ ] ] where the electrostatic potential @xmath1 is given by the poisson equation @xmath2 here @xmath3 is the magnitude of the electron charge , @xmath4 is the electron mass , @xmath5 is the planck constant divided by @xmath6 , and @xmath7 is the permittivity of free space .
furthermore , @xmath8 and @xmath9 denote the equilibrium electron distribution function and the electron number density , respectively , while @xmath10 , @xmath11 and @xmath12 denote the perturbed electron distribution function , the electrostatic potential , and the ion number density , respectively . assuming that @xmath10 , @xmath11 and @xmath12 are proportional to @xmath13 , where @xmath14 is the frequency and @xmath15 is the wave vector , we obtain from eq .
( [ eq1 ] ) and ( [ eq2 ] ) , respectively , @xmath16 \\ & & \times \left [ e^{i { \bf k}\cdot\boldsymbol{\lambda}/2}-e^{-i { \bf k}\cdot\boldsymbol{\lambda}/2 } \right]f_0({\bf v}')\phi_1(\omega,{\bf k } ) , \label{eq_b3}\end{aligned}\ ] ] @xmath17 since ions are non - degenerate in quantum plasmas , we have for @xmath18 , @xmath19 where @xmath20 is the ion susceptibility , @xmath21 is the ion thermal speed , and @xmath22 is the ion plasma frequency . rewriting ( [ eq_b3 ] ) as @xmath23 \right . \\
& & \left . - \exp\left[i \frac{m_e}{\hbar } ( { \bf v}-{\bf v } ' ) \cdot{\boldsymbol{\lambda}}-i { \bf k}\cdot\boldsymbol{\lambda}/2\right ] \right\}f_0({\bf v}')\phi_1(\omega,{\bf k}),\end{aligned}\ ] ] and performing the integration over @xmath24 space , we have @xmath25 \right . \\
& & \left .
-\delta\left[\frac{m_e}{\hbar } ( { \bf v}-{\bf v}')-\frac{\bf k}{2}\right ] \right\}f_0({\bf v}')\phi_1(\omega,{\bf k}),\end{aligned}\ ] ] where @xmath26 is the dirac delta function .
now , the integration can be performed over @xmath27 space , obtaining the result @xmath28\phi_1(\omega,{\bf k } ) .
\label{eq_b7}\ ] ] eliminating @xmath12 and @xmath10 in ( [ eq_b4 ] ) with the help of ( [ ni1 ] ) and ( [ eq_b7 ] ) , we obtain the dispersion relation @xmath29 where the ion susceptibility is given by ( [ chi_i ] ) and the electron susceptibility is given by @xmath30 d^3 u. \label{chi_e1}\ ] ] suitable changes of variables in the two terms in square brackets in eq .
( [ chi_e1 ] ) now give @xmath31 } -\frac{1}{[-\omega+{\bf k}\cdot ( { \bf u}+\frac{\hbar{\bf k}}{2 m_e } ) ] } \right]f_0({\bf u } ) d^3 u,\ ] ] which can be rewritten as @xmath32 this expression was also been derived by @xcite by using a series of canonical transformations of the hamiltonian of the system [ see for example the dispersion relation ( 57 ) in their paper ] , and by @xcite by using the method of self - consistent fields .
we now choose a coordinate system such that the @xmath33 axis is aligned with the wave vector @xmath15 .
then , ( [ eq_b10 ] ) takes the form @xmath34 we next consider a dense plasma with degenerate electrons in the zero temperature limit .
then , the background distribution function takes the simple form @xmath35 where @xmath36 is the speed of an electron on the fermi surface , and @xmath37 is the fermi energy .
the integration in ( [ eq_b11 ] ) can be performed over velocity space perpendicular to @xmath38 , using cylindrical coordinate in @xmath39 and @xmath40 , obtaining the result @xmath41 where @xmath42 it is interesting to note that the distribution , which is flat - topped in three - dimensions becomes parabola - shaped in the remaining velocity dimension after the integration over the two perpendicular velocity dimensions .
hence , the electron distribution function @xmath43 in ( [ f0 ] ) may support landau damping if the pole of the denominator in ( [ eq_b13 ] ) falls into the range of negative slope of @xmath43 in velocity space .
equation ( [ eq_b13 ] ) can be written as @xmath44 performing the integration over velocity space , we have from ( [ eq_b15 ] ) @xmath45 \log\left| \frac { \frac{\omega}{k}-v_{fe}+\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}+\frac{\hbar k}{2 m_e } } \right| \right . \\
& & \left .
+ \frac{m_e}{\hbar k v_{fe}}\left[v_{fe}^2-\left(\frac{\omega}{k}-\frac{\hbar k}{2 m_e}\right)^2\right ] \log\left| \frac { \frac{\omega}{k}-v_{fe}-\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}-\frac{\hbar k}{2 m_e } } \right| \right\}. \label{eq_b18}\end{aligned}\ ] ] in the derivation of ( [ eq_b18 ] ) , we have assumed that the waves are only weakly damped , so that , when integrating over poles , only the principal parts of the integrals are kept . in the limit @xmath46
, we have from ( [ eq_b18 ] ) @xmath47 where it holds that @xmath14 is real and @xmath48 .
hence in this `` semi - classical '' limit , we do not have landau damping . [ we call ( [ eq_b16 ] ) `` semi - classical '' , since it can be derived from the vlasov equation for electrons using the flat - topped background electron distribution function given by ( [ flat ] ) . ]
we here consider high - frequency ( @xmath49 ) oscillations so that @xmath50 in ( [ chi_i ] ) .
then , expanding ( [ eq_b18 ] ) for small wavenumbers up to terms containing @xmath51 , we obtain from ( [ disp_e ] ) the dispersion relation @xmath52 where @xmath53 and @xmath54 is the bohr radius . for a typical metal such as gold , which has a free electron number density of @xmath55
, we would have @xmath56 . for the free electron density in semiconductors , which is many orders of magnitude less than in metals
, @xmath57 is much smaller and can safely be dropped compared to unity . however , for electron plasma oscillations in dense matters , @xmath57 could be larger than unity . it should be noted that the term proportional to @xmath57 in ( [ eq_b19 ] ) was not discussed by @xcite and others , but was , however , obtained and discussed by @xcite in his study of collective electron oscillations in metals [ see eq .
( 10 ) in his paper , where , in his notation , it should be @xmath58 .
it is interesting to note that ( [ eq_b18 ] ) _ may admit _ landau damping above a certain critical wavenumber , @xmath59 and corresponding frequency @xmath60 .
this occurs if the denominator in the integral of ( [ eq_b15 ] ) vanishes within the integration limits @xmath61 . for the critical wavenumber and frequency , we have @xmath62 inserting this expression into ( [ eq_b18 ] ) we note that the term involving the logarithm on the second line of ( [ eq_b18 ] ) vanishes , and we obtain the critical wavenumber @xmath63 from @xmath64=0 .
\label{critical2}\ ] ] a careful examination of the dispersion relation for @xmath59 , should involve landau contours to correctly take into account landau damping .
here we are interested in low- and high - frequency waves in the weakly damped regime and have postponed the investigation of landau damping of the system to future studies .
in a quantum plasma system composed of mobile ions and inertialess electrons , we have possibility of low - phase speed ( in comparison with the fermi electron thermal speed ) ion - acoustic - like oscillations . for low - frequency ( @xmath65 ) waves , we have from ( [ eq_b18 ] ) @xmath66 . \label{chi_e}\ ] ] for small wavenumbers @xmath67 , we have the approximate electron susceptibility , up to terms containing factors of @xmath51 , @xmath68 using the dispersion relation @xmath69 we employ ( [ chi_i ] ) to obtain the frequency of ion acoustic waves as @xmath70 with @xmath71 given by ( [ chi_e ] ) or ( [ chi_e2 ] ) . for @xmath71 given by the approximate expression ( [ chi_e2 ] ) , we have @xmath72 where @xmath73 is the fermi ion acoustic speed .
we note that @xmath74 as @xmath75 .
we note that there is a critical density parameter in the system . when the inter - particle distance is smaller than the bohr radius , then the quantum statistical pressure dominates the wave dynamics , while in the opposite case , the quantum tunneling effects become important when the wavelength is comparable to the inter - particle distance .
this can be seen by normalize the system such that @xmath76 and @xmath77 .
then , eq . ( [ eq_b18 ] ) takes the form @xmath78 \log\left| \frac { \frac{\omega}{k}-1+\frac{k}{2\beta } } { \frac{\omega}{k}+1+\frac{k}{2\beta } } \right| \right . \\
& & \left .
+ \frac{\beta}{k}\left[1-\left(\frac{\omega}{k}-\frac{k}{2\beta}\right)^2\right ] \log\left| \frac { \frac{\omega}{k}-1-\frac{k}{2\beta } } { \frac{\omega}{k}+1-\frac{k}{2\beta } } \right| \right\ } , \label{eq_b20}\end{aligned}\ ] ] where @xmath79 in the scaled variables , ( [ eq_b16 ] ) and ( [ eq_b19 ] ) take the form @xmath80 and @xmath81 respectively .
we note that the limit @xmath82 to obtain ( [ eq_b16 ] ) from ( [ eq_b18 ] ) corresponds to @xmath83 to obtain ( [ eq_semi_scaled ] ) from ( [ eq_b20 ] ) . for @xmath84 ,
the quantum statistical pressure dominates , while for @xmath85 , the quantum tunneling effects dominate .
considering the value of @xmath86 for different physical systems , we note that @xmath87 corresponds to relatively low density degenerate plasma such as in semiconductors , while @xmath88 corresponds to typical free electron densities in metals .
the high density case @xmath89 corresponds to high density matter which may be obtained in laser compression schemes or which exist in white dwarf stars .
even though we have formally considered the semiclassical limit @xmath90 , it should be kept in mind that an upper limit for the validity of our theory is when the electron density becomes high enough that the fermi speed @xmath91 becomes comparable with the speed of light . in this limit
, the inter - particle distance @xmath92 approaches the compton length @xmath93 , and we have an electron number density of the order @xmath94 , corresponding to @xmath95 . for larger values of @xmath86 , the equilibrium equation of state for the electrons [ @xcite ] change from @xmath96 to @xmath97 . for this case , we need to include relativistic effects in the electron susceptibility . in the normalized variables , the condition ( [ critical2 ] ) for the critical wavenumber for the limit between undamped and landau damped high - frequency waves
is given by @xmath98=0,\ ] ] and the normalized critical frequency is obtained from ( [ critical1 ] ) as @xmath99 where @xmath100 and @xmath101 . finally , for the low - frequency case , the electron susceptibilities ( [ chi_e ] ) is in the normalized variables given by @xmath102.\ ] ] and the expression for small wavenumbers ( [ chi_e2 ] ) is given by @xmath103
versus @xmath104 ) for different values of @xmath105 .
the solid curves show solutions of ( [ disp_e ] ) using the exact susceptibility ( [ eq_b18 ] ) , the dashed curves show the expanded solution ( [ eq_b19 ] ) , and the dash - dotted curve shows solutions of ( [ disp_e ] ) using the `` semi - classical '' electron susceptibility ( [ eq_b16 ] ) . the dotted curve indicates the border between undamped waves and landau damped waves , given by ( 3.2 ) and ( 3.3 ) [ or ( 5.5 ) and ( 5.6)].,width=377 ] versus @xmath104 ) for the low - frequency ion oscillations , for different values of @xmath105 .
the solid curves show the wave frequency given by ( [ omega_ion ] ) using the low - frequency electron susceptibility ( [ chi_e ] ) , while the dashed curves uses the approximate electron susceptibility ( [ chi_e2 ] ) . for @xmath89
the solid and dashed curves are indistinguishable.,width=377 ] in figs . 1 and 2 , we show dispersion curves for the high - frequency ( @xmath49 ) waves for different values of @xmath86 , obtained from the solutions of the dispersion relation ( [ disp_e ] ) , by using the electron susceptibility ( [ eq_b18 ] ) , as well as the expansion ( [ eq_b19 ] ) and the limiting semi - classical case ( [ eq_b16 ] ) .
we have also indicated the border between undamped and landau damped waves , obtained from ( [ critical1 ] ) and ( [ critical2 ] ) .
we note that the dispersion curve for the semiclassical case in fig 1 always lies in the undamped regime , below the border between undamped and landau damped waves . for the undamped waves , the expansion ( [ eq_b19 ] ) approximates the exact dispersion relation within a few percent , and
can therefore be used instead of ( [ eq_b18 ] ) for most cases .
this holds especially for small wavenumbers , as can be seen in the closeup in fig .
the dispersion curves for the low - frequency ion - acoustic oscillations are plotted in fig .
3 , were we have depicted the @xmath14 in ( [ omega_ion ] ) for @xmath71 given by the exact expression ( [ chi_e ] ) and the approximate expansion ( [ chi_e2 ] ) , and for different values of @xmath86 .
we note that the dispersion curves show agreement at small wavenumbers but deviate significantly for larger wavenumbers in the case @xmath87 , while the agreement is better for @xmath88 and excellent for @xmath89 .
it should be kept in mind that when @xmath106 then the wavelength of the oscillations is comparable to the inter - particle distance , and there will be corrections due to the discrete nature of the ion background .
thus , the theory is not valid for oscillations with @xmath107 .
in this paper , we have here studied the dispersion properties of electrostatic oscillations in quantum plasmas for different parameters ranging from semiconductor plasmas , to typical metallic electron densities and to densities corresponding to compressed matter and dense astrophysical objects .
we have derived a simplified expansion that accurately approximates the exact dispersion relation for small wavenumbers .
the possibility of landau damping due to quantum tunneling effects at large wavenumbers has also been discussed , and conditions for landau damping has been derived .
the present results should be useful in understanding the salient features of electrostatic plasma oscillations in dense plasmas with degenerate electrons .
the latter are encountered in metals , in highly compressed intense laser - solid density plasma experiments , and in compact astrophysical objects ( e.g. interior of white dwarf stars ) .
watanabe h. 1956 experimental evidence for the collective nature of the characteristic energy loss of electrons in solids studies on the dispersion relation of plasma frequency _ j. phys .
. jpn _ * 11 * , 112119 . | we present a derivation of the dispersion relation for electrostatic oscillations ( esos ) in a zero temperature quantum plasma . in the latter , degenerate
electrons are governed by the wigner equation , while non - degenerate ions follow the classical fluid equations .
the poisson equation determines the electrostatic wave potential .
we consider parameters ranging from semiconductor plasmas to metallic plasmas and electron densities of compressed matter such as in laser - compression schemes and dense astrophysical objects . due to the wave diffraction caused by overlapping electron wave function due to the heisenberg uncertainty principle in dense plasmas
, we have possibility of landau damping of the high - frequency electron plasma oscillations ( epos ) at large enough wavenumbers . the exact dispersion relations for the epos are solved numerically and compared to the ones obtained by using approximate formulas for the electron susceptibility in the high- and low - frequency cases . |
in @xcite , szemerdi and trotter proved the following theorem : [ szemereditrotterthm ] the number of incidences between @xmath0 points and @xmath1 lines in @xmath11 is @xmath12 .
theorem [ szemereditrotterthm ] has seen a number of generalizations , including the following one due to pach and sharir in @xcite : [ pachsharirresult ] let @xmath13 be a collection of @xmath0 points in @xmath11 and @xmath14 a collection of @xmath1 curves with @xmath15 degrees of freedom
there exists a constant @xmath16 such that any two curves can meet in at most @xmath16 points and at most @xmath16 curves can contain any given @xmath15 points .
then the number of incidences between points in @xmath13 and curves in @xmath14 is @xmath17 where the implicit constant depends only on @xmath16 .
theorem [ pachsharirresult ] was proved using the crossing number inequality for planar graphs , which was first employed in incidence geometry by szkely in @xcite and later used by arnov in @xcite to obtain some partial incidence results in higher dimensions . in a different direction ,
tth @xcite generalized theorem [ szemereditrotterthm ] to complex points and lines in @xmath10 .
however , as of this writing ( 2012 ) , tth s paper is still in the midst of a lengthy review process while awaiting publication .
solymosi and tardos @xcite gave a simpler proof of the same result in the special case where the point set is a cartesian product of the form @xmath18 .
edelsbrunner and sharir @xcite obtained incidence results for certain configurations of points and codimension1 hyperplanes in @xmath2 , and aba and solymosi @xcite obtained incidence bounds for points and a general class of 2dimensional surfaces in @xmath19 , provided the points satisfied a certain homogeneity condition .
elekes and tth @xcite and later solymosi and tth @xcite obtained incidence results between points and hyperplanes in @xmath20 , again provided the points satisfied various nondegeneracy and homogeneity conditions . in @xcite , solymosi and
tao used the discrete polynomial ham sandwich theorem to obtain bounds for the number of incidences between points and bounded degree algebraic surfaces satisfying certain `` pseudoflat '' conditions ( i.e. they behaved similarly to hyperplanes ) . aside from an @xmath6 loss in the exponent , solymosi and tao s result resolved a conjecture of tth on the number of incidences between points and @xmath21flats in @xmath22 .
the discrete polynomial ham sandwich theorem was also used by the author in @xcite to obtain incidence results between points and 2dimensional surfaces in @xmath19 ( with no homogeneity condition ) , and by kaplan et al . in @xcite to obtain similar bounds on the number of incidences between points and spheres in @xmath23 in this paper , we combine the crossing number inequality and the discrete polynomial ham sandwich theorem to obtain a new result which can be seen either as a sharpening of the @xmath2 version of the solymosi - tao result from @xcite or a generalization of tth s result from @xcite . to the best of the author
s knowledge , this is the first time these two techniques have been used together .
[ gooddefn ] we call a collection @xmath14 of surfaces @xmath16_good _ if for each pair of distinct surfaces @xmath24 we have the following : 1 . [ gooddefndim2 ] @xmath25
@xmath26 is a real algebraic variety and @xmath27 .
3 . @xmath28 . here
, @xmath16 should be thought of as a fixed constant , while our collections of points and surfaces is allowed to be large .
all of the bounds in the rest of this paper will depend on @xmath16 , but for notational convenience we shall suppress this dependence , and we shall say a collection of surfaces is _ good _ if it is @xmath16good for some fixed value of @xmath16 .
[ incidencesdef ] let @xmath13 be a collection of points and @xmath14 a good collection of surfaces .
we define the set of incidences of @xmath13 and @xmath14 to be @xmath29 [ smoothandtangent ] let @xmath30 .
we define @xmath31 to be the set of smooth points of @xmath26 ( see e.g. @xcite ) . if @xmath32 we define @xmath33 to be the tangent plane to @xmath26 at @xmath34 .
this is an ( affine ) 2plane in @xmath2 .
[ admissibledefn ] we say a collection of incidences @xmath35 is @xmath36_admissible _ if the following requirements are satisfied : 1 . [ admissibledefnsmooth ] if @xmath37 then @xmath38 .
[ admissible2nd ] if @xmath26 and @xmath39 are distinct and @xmath40 , then @xmath41 .
[ admissible3rd ] for any @xmath15 points @xmath42 , there are at most @xmath16 surfaces @xmath30 such that @xmath43 for each @xmath44 .
as above , we shall call a collection of incidences @xmath15_admissible _ if it is @xmath36admissible for some value of @xmath16 .
finally , we may say a collection of surfaces is @xmath45 if the value of @xmath15 is either understood from the context or does not matter .
the most important example arises from complex points and lines in @xmath10 .
let @xmath46 be a collection of ( complex ) points in @xmath10 and let @xmath47 be a collection of complex lines in @xmath10 .
then if @xmath48 is the image of @xmath46 under the standard identification of @xmath10 with @xmath2 , and @xmath14 is the collection of affine 2planes in @xmath2 which are the images of the complex lines in @xmath47 , then @xmath14 is a @xmath49good collection of surfaces , and @xmath50 is a @xmath51admissible collection of incidences .
requirement [ admissible2nd ] in definition [ admissibledefn ] may seem artificial , but it is necessary to prevent the situation in which all of the surfaces meet in a common 1dimensional variety and all of the points lie on that variety .
if this were to occur , then the trivial bound of @xmath52 incidences would be sharp .
we are now ready to state our results .
[ surfacesinr4thm ] let @xmath48 be a collection of points , with @xmath53 let @xmath14 be a good collection of surfaces , with @xmath54 .
let @xmath35 be a collection of 2admissible incidences .
then we have the following bounds . *
if @xmath55 then @xmath56 while if @xmath57 @xmath58 * for all values of @xmath0 and @xmath1 , @xmath59 here , the constant @xmath60 may depend only on the constants appearing in definitions [ gooddefn ] and [ admissibledefn ] , while @xmath61 may also depend on @xmath62 .
the next theorem is an analogue of theorem [ pachsharirresult ] , though we lose an additional logarithmic factor .
let @xmath48 be a collection of points , with @xmath53 let @xmath14 be a good collection of surfaces , with @xmath54 .
let @xmath35 be a collection of @xmath15admissible incidences .
then we have the following bounds . *
if @xmath55 then @xmath63 while if @xmath57 @xmath64 * for all values of @xmath0 and @xmath1 , @xmath65 the constants @xmath60 and @xmath66 have the same dependencies as in theorem [ surfacesinr4thm ] .
if @xmath14 is a collection of affine @xmath8planes in @xmath2 with no two @xmath8planes meeting in a common line , then @xmath14 is a good collection . if @xmath48 , then any collection of incidences @xmath67 is 2admissible .
thus we have the following corollary : [ szemereditrotter2planes ] let @xmath48 be a collection of @xmath0 points and let @xmath14 be a collection of @xmath1 2planes in @xmath2 such that any two planes meet in at most one point
. then the number of incidences between points in @xmath13 and planes in @xmath14 is bounded by the quantities in .
we can use corollary [ szemereditrotter2planes ] to recover the szemerdi - trotter theorem for complex lines in @xmath10 , which was originally proved by tth in @xcite .
note that by point - line duality in @xmath10 , we can always assume that the number of lines is at least as great as the number of points .
thus we have : let @xmath13 be a collection of @xmath0 points and let @xmath14 be a collection of @xmath1 ( complex ) lines in @xmath10 . then the number of incidences between points in @xmath13 and complex lines in @xmath14 is @xmath68 as another special case , any collection @xmath14 of complex unit circles in @xmath10 ( regarded as 2dimensional real varieties in @xmath2 ) is a good collection .
furthermore , if @xmath13 is a collection of points then we can partition the collection of incidences between @xmath14 and @xmath13 into boundedly many classes @xmath69 such that each collection of incidences is 2admissible .
see corollary 2.7 of @xcite for details .
thus we have let @xmath13 be a collection of @xmath0 points and let @xmath14 be a collection of @xmath1 ( complex ) unit circles in @xmath10 .
then the number of incidences between points in @xmath13 and circles in @xmath14 is @xmath68 this result allows us to obtain an improved bound on the unit distance problem in @xmath10 : let @xmath70 be a finite set of points , with @xmath71 .
then @xmath72 for simplicity , we shall only sketch the proof of theorem [ surfacesinr4thm ] , i.e. the @xmath73 case , and we will concentrate on proving . furthermore , we shall assume that the surfaces in @xmath14 are 2planes .
the basic idea is as follows .
since each pair of points has at most two 2planes passing through both of them , we can use the cauchy - schwarz inequality to obtain a rudimentary bound on the cardinality of any collection of point - surface incidences .
we will call this the cauchy - schwarz bound . using the discrete polynomial ham sandwich theorem
, we find a polynomial @xmath74 of controlled degree whose zero set cuts @xmath2 into open `` cells , '' such that each cell contains roughly the same number of points from @xmath13 , and each surface from @xmath14 does not enter too many cells .
we can then apply the cauchy - schwarz bound within each cell .
this allows us to count the incidences occurring between surfaces and points in @xmath75 . in order to count the remaining incidences ,
we perform a `` second level '' polynomial ham sandwich decomposition on the variety @xmath76 .
this gives us a polynomial @xmath77 which cuts @xmath76 into a collection of 3dimensional cells , which are open in the relative ( euclidean ) topology of @xmath76 .
we then apply the cauchy - schwarz bound to each of these 3dimensional cells .
the only incidences left to count are those between surfaces in @xmath14 and points in @xmath78 .
let @xmath79 we can choose @xmath74 and @xmath77 in such a way that @xmath80 is a 2dimensional variety in @xmath2 .
let @xmath26 be a surface in @xmath81 then @xmath26 will intersect @xmath80 in a union of isolated points ( proper intersections ) and 1dimensional curves ( non - proper intersections ) ( the case where @xmath26 meets @xmath80 in a 2dimensional variety can be dealt with easily ) .
the number of isolated points in the intersection can be controlled by the degrees of the polynomials @xmath74 and @xmath77 ( we are working over @xmath82 where bzout s theorem need not hold , so we need to be a bit careful ) , and thus the number of incidences between points @xmath83 and surfaces @xmath30 such that @xmath84 is an isolated point of @xmath85 can be controlled .
the only remaining task is to control the number of incidences between points of @xmath86 and 1dimensional curves arising from the intersection of @xmath80 and surfaces @xmath30 . to simplify the exposition ,
pretend @xmath80 is a disjoint union of @xmath87 2planes , i.e. @xmath88 ( of course , we will not make this assumption when we prove the actual result ) .
then for each plane @xmath89 @xmath90 is a line on @xmath91 .
it remains to count the number of incidences between @xmath92 and @xmath93 the szemerdi - trotter theorem for lines in @xmath11 would give us the bound @xmath94 but if we sum over the @xmath87 values of @xmath95 , we have only bounded the number of incidences by @xmath96 since @xmath87 can be quite large , this is not sufficient . instead ,
recall szkely s proof in @xcite of the szemerdi - trotter theorem , which uses the crossing number inequality ( the crossing number inequality and all other graph - related quantities are defined in section [ graphtheorysection ] below ) . loosely speaking
, we consider the graph drawing @xmath97 on @xmath91 whose vertices are the points of @xmath92 , and two vertices are connected by an edge if there is a line segment from @xmath98 passing through the two points .
then the number of edges of the graph is comparable to the number of incidences between points and lines , and this is controlled by @xmath99 , where @xmath100 is the number of edge crossings and @xmath101 is the number of vertices of @xmath97 .
thus in place of , we have @xmath102 the key insight is that @xmath103 indeed , every pair of 2planes @xmath104 can intersect in at most one point , and since we assumed the planes @xmath105 composing @xmath80 were disjoint , the intersection point of @xmath106 can occur on @xmath91 for at most one index @xmath95 .
summing over all choices of @xmath95 , applying hlder s inequality , and inserting , we obtain the correct bound on the number of incidences between surfaces in @xmath14 and points lying on @xmath80 .
unfortunately , the assumption that @xmath80 is a disjoint union of 2planes need not be true .
thus we must cut @xmath80 up into pieces , each of which behaves like a 2plane , and we need to prove a more general form of the szemerdi - trotter theorem for families of curves and points that lie on suitable domains .
this is a purely topological argument , and it is done using the crossing number inequality .
the decomposition of @xmath80 into suitable domains relies on results from real algebraic geometry , and it works well provided @xmath80 is not of too high degree .
the degree of @xmath80 depends on the value of @xmath107 .
if @xmath108 is too large compared to @xmath109 , the degree of @xmath80 is too big , and we can not cut @xmath80 into suitable pieces without introducing error terms that are larger than the bounds we are trying to prove .
thus , instead of choosing `` optimal '' values for the degrees of @xmath74 and @xmath77 , we choose smaller values so that the degree of @xmath80 is not too big .
we can no longer apply the cauchy - schwarz bound to control the incidences occurring within each of the 3 and 4dimensional cells described above .
instead , we borrow an idea from solymosi and tao in @xcite , and we assume an appropriate bound on the number of incidences between collections of points @xmath110 and surfaces @xmath111 provided @xmath112 and @xmath113 have a better ratio of sizes than @xmath108 and @xmath109 . based on the way the cell decomposition works , this induction assumption will be satisfied for the collection of points and surfaces inside each cell of the decomposition , provided that @xmath114 for some constant @xmath5 , where the improvement obtained by the induction step depends on @xmath6 .
after finitely many applications of the induction step , we obtain a situation where the ratio of @xmath108 and @xmath109 is very favorable , and thus our decomposition of @xmath80 into `` 2plane - like pieces '' is very good . at this point , we can select @xmath74 and @xmath77 to have optimal degrees and use the arguments above where we apply the cauchy - schwarz bound within each cell rather than the induction hypothesis .
the proof of is similar , except we never apply the cauchy - schwarz bound ; we keep applying the induction hypothesis bounds until our cells are so small that we can use a trivial bound to control the number of incidences that occur with each cell .
this is similar to the techniques used by solymosi and tao .
however , solymosi and tao need to iterate their induction step roughly @xmath115 times , while with our techniques we only need to perform the induction step about @xmath116 times .
this allows us to get a slightly sharper bound . in order to prove theorem [ surfacesinr4thm]@xmath117
, we use the more intricate crossing - number based arguments of pach and sharir from @xcite in place of szkely s proof of the szemerdi - trotter theorem .
the author is very grateful to silas richelson and terence tao for helpful discussions .
the author was supported in part by the department of defense through the national defense science & engineering graduate fellowship ( ndseg ) program .
in this section we shall recall some basic definitions and also state several lemmas that will be needed in the body of the proof .
the proofs of these lemmas have been deferred to appendix [ proofappendix ] . throughout the paper
, @xmath118 will denote a sufficiently large absolute constant that is allowed to vary from line to line .
we will write @xmath119 to mean @xmath120 and we say that a quantity @xmath121 is @xmath122 if @xmath123 .
we say a quantity @xmath121 is @xmath124 if @xmath125 is @xmath126 , and we say a quantity @xmath121 is @xmath127 if @xmath128 is some function of @xmath129 with @xmath130 as @xmath131 ( @xmath121 may also be a function of additional variables . in this case
, we require that @xmath130 as @xmath131 for each fixed value of the other variables ) .
let @xmath132 be a multigraph with maximum edge multiplicity @xmath133 , and let @xmath134 be a drawing of @xmath132 , i.e. a collection of points and curves in @xmath11 such that every vertex of @xmath132 corresponds to a distinct point of @xmath134 , and every edge of @xmath132 corresponds to a curve in @xmath134 such that every two curves intersect in a discrete set , and no points ( i.e. vertices ) are contained in the relative interior of any curve ( i.e. edge ) .
we define @xmath135 to be the number of vertices of @xmath132 and @xmath136 to be the number of edges , and similarly for @xmath134 .
we define @xmath137 to be the _ crossing number _ of @xmath134 , i.e. the number of times two curves cross each other . since the intersection of any two curves is a discrete set , @xmath137 is finite .
[ crossingnumberineq ] let @xmath134 be a drawing of a multigraph with maximum edge multiplicity @xmath133 .
then either @xmath138 , or @xmath139 [ turanboundlemma ] let @xmath140 be fixed , and let @xmath141 be a bipartite graph with @xmath142 that contains no copy of @xmath143 .
then @xmath132 has at most @xmath144 edges .
symmetrically , @xmath132 has at most @xmath145 edges . unless otherwise noted , all polynomials will be ( affine ) real polynomials , i.e. elements of @xmath146 $ ] , and all varieties will be affine real varieties . definitions and standard results about real algebraic varieties can be found in @xcite .
a ( real ) algebraic variety @xmath147 is a set of the form @xmath148 where @xmath149 $ ] are polynomials . note that we do not require varieties to be irreducible .
if @xmath150 $ ] is an ideal , we define @xmath151 by abuse of notation , if @xmath152 $ ] , then we define @xmath153 .
if @xmath147 is a variety , we define @xmath154\colon\ p(x)=0\ \textrm{for all}\ x\in z\}.\ ] ] if @xmath147 is a real variety , then @xmath155 denotes the smallest complex variety containing @xmath76 .
note that if @xmath156 is a point , then @xmath157 is the image of @xmath84 under the canonical embedding @xmath158 .
conversely , if @xmath159 is a complex variety , then @xmath160 is its set of real points .
the properties of @xmath161 and @xmath162 will be discussed further in section [ realcplxvarietiessec ] .
let @xmath163 $ ] be a collection of non - zero real polynomials .
a _ strict sign condition _ on @xmath164 is a map @xmath165 .
if @xmath166 , we will denote the evaluation of @xmath167 at @xmath77 by @xmath168 .
if @xmath147 is a variety , and @xmath167 is a strict sign condition on @xmath164 , then we can define the _ realization of @xmath167 on z _ by @xmath169 we define @xmath170 and @xmath171 we call @xmath172 the collection of `` realizations of realizable strict sign conditions of @xmath164 on @xmath76 . ''
note that if some @xmath166 vanishes identically on @xmath76 then @xmath173 and thus @xmath174 .
[ defnofrealideal ] an ideal @xmath150 $ ] is _ real _ if for every sequence @xmath175 $ ] , @xmath176 implies @xmath177 for each @xmath178 .
the following proposition shows that real principal prime ideals and their corresponding real varieties have some of the `` nice '' properties of ideals and varieties defined over @xmath179 .
[ propertiesofprinciplerealidealprop ] let @xmath180 $ ] be a principal prime ideal
. then the following are equivalent : 1 .
@xmath181 is real .
[ nullz ] @xmath182 .
[ smalldim ] @xmath183 .
@xmath184 does not vanish identically on @xmath185 .
[ nosignchange ] the sign of @xmath74 changes somewhere on @xmath20 . while not every polynomial @xmath152
$ ] is a product of real ideals , the following lemma shows that for our applications , we can always modify our polynomials to ensure that this is the case . [ makeapolyprodrealideals ] let @xmath152 $ ] be a real polynomial .
then there exists a real polynomial @xmath186 such that @xmath187 @xmath188 and the irreducible components of @xmath186 generate real ideals .
the proof of this lemma has been deferred to appendix [ proofappendix ] .
we will make essential use of the real bzout s theorem ( see e.g. @xcite ) and of harnack s theorem ( see e.g. ( * ? ? ? *
p 57 ) ) .
[ realbezout ] let @xmath189 $ ] be real polynomials of degrees @xmath190 .
then the number of nonsingular intersection points of @xmath191 is at most @xmath192 .
[ planarharnak ] let @xmath193 be an algebraic plane curve .
then @xmath76 contains @xmath194 connected components . in @xcite ,
guth and katz used the polynomial ham sandwich theorem to prove the following `` discrete '' polynomial ham sandwich theorem .
[ discretehamsandthm ] fix @xmath195 .
let @xmath196 be finite families of points , with @xmath197 .
then there exists a polynomial @xmath198 $ ] of degree at most @xmath195 such that @xmath199 iterating theorem [ discretehamsandthm ] as in @xcite , for any discrete set we can find a polynomial @xmath74 that cuts @xmath2 into open `` cells , '' with each cell not containing too many points from the discrete set : [ hamsandwichthm ] let @xmath13 be a collection of points in @xmath2 , and let @xmath200 .
then there exists a non - zero polynomial @xmath74 of degree at most @xmath201 such that each connected component of @xmath202 contains at most @xmath203 points of @xmath13 . after applying lemma [ makeapolyprodrealideals ]
, we can ensure that the irreducible components of @xmath74 generate real ideals : [ hamsandwichthmstronger ] let @xmath13 be a collection of points in @xmath2 , and let @xmath200 .
then there exists a non - zero polynomial @xmath74 of degree at most @xmath201 such that each connected component of @xmath202 contains at most @xmath203 points of @xmath13 , and each irreducible component of @xmath74 generates a real ideal .
[ ex1 ] consider the following collection of 72 points : @xmath204 and let @xmath205 .
then the degree4 polynomial @xmath206 cuts @xmath2 into 16 open cells @xmath207 ( the cells are unbounded , but this is fine ) plus the set @xmath208 we can verify that the polynomials @xmath209 generate real ideals , so @xmath74 is a product of irreducible polynomials , each of which generates a real ideal .
we have @xmath210 for each @xmath211 .
thus @xmath74 satisfies the requirements of corollary [ hamsandwichthmstronger ] ( corollary [ hamsandwichthmstronger ] only specifies the degree of @xmath74 up to an implicit constant , so we can not verify that the degree is correct ) .
finally , note that we have @xmath212 [ ex2 ] let @xmath48 be a large collection of points that lie in general position on the 2plane @xmath213 , and let @xmath201 be much smaller than @xmath214 then we can verify that the polynomial @xmath215 satisfies the requirements of corollary [ hamsandwichthmstronger ] ; @xmath216 cuts @xmath2 into the two cells @xmath217 and @xmath218
. we have @xmath219 and @xmath220 this phenomenon is unavoidable : any polynomial @xmath74 satisfying the requirements of corollary [ hamsandwichthmstronger ] must contain a factor @xmath221 of the form @xmath222 for some @xmath223 ( provided the points of @xmath13 are in general position ) .
thus we must have @xmath224 so each of the cells of the decomposition @xmath225 will be empty .
this example is interesting because if @xmath226 is a collection of complex points and lines in @xmath10 , and @xmath227 is the collection of points and 2planes in @xmath2 associated to @xmath226 , then all of the points of @xmath13 will lie on a common 2plane , so the situation will resemble this example . theorem [ hamsandwichthm ] will be used to obtain the `` first level '' decomposition of the point set @xmath13 .
however , as seen in the above examples , many points may lie on the `` boundary '' @xmath228 and we will need to control the number of incidences between surfaces in @xmath14 and points on @xmath185 . to do this , we shall perform a second discrete polynomial ham sandwich decomposition on the algebraic set @xmath185 .
this shall be done with the following theorem : [ varianthamsandwichthm ] let @xmath13 be a collection of points in @xmath2 lying on the set @xmath216 for @xmath74 an irreducible polynomial of degree @xmath201 such that @xmath74 generates a real ideal .
let @xmath229 be a small constant , and let @xmath230 .
then there exists a collection of polynomials @xmath231 $ ] with the following properties : 1 .
[ varianthamsandwichthmitm1 ] @xmath232 .
[ varianthamsandwichthmitm2 ] @xmath233 .
[ varianthamsandwichthmitm3 ] none of the polynomials in @xmath164 vanish identically on @xmath76 .
[ varianthamsandwichthmitm4 ] the realization of each of the @xmath234 strict sign conditions of @xmath164 on @xmath76 contains at most @xmath235 points of @xmath13 .
[ varianthamsandwichthmitm5 ] each irreducible component of each polynomial @xmath166 generates a real ideal .
all implicit constants depend only on @xmath62 and the dimension @xmath21 .
theorem [ varianthamsandwichthm ] is proved in ( * ? ? ?
* a.3 ) .
we can continue examples [ ex1 ] and [ ex2 ] .
let @xmath236 @xmath201 , @xmath237 and @xmath76 be as in example [ ex1 ] above .
then @xmath238 is the only irreducible component of @xmath74 whose zero - set contains points from @xmath13 .
let @xmath239 , and let @xmath240 .
let @xmath241 then @xmath242 consists of the 8 octants of @xmath19 , where we identify @xmath19 with the hyperplane @xmath243 in @xmath2 .
each of the components of @xmath242 contains 2 points from @xmath244 , and @xmath245 i.e. every point of @xmath13 either lies in some cell of @xmath246 or some realization of a realizable strict sign condition of @xmath164 on @xmath247 .
let @xmath236 @xmath201 , @xmath237 and @xmath76 be as in example [ ex2 ] , and let @xmath248 be much smaller than @xmath249 let @xmath250 .
then @xmath172 consists of the sets @xmath251 and @xmath252 neither of these sets contain any points from @xmath253 ; indeed , @xmath254 .
thus @xmath164 satisfies the requirements of theorem [ varianthamsandwichthm ] , but none of the points of @xmath13 lie in any cell of @xmath246 nor in any realization of a realizable strict sign condition of @xmath164 on @xmath76 .
section [ proofofcrossingnumberineqlemmasection ] will be devoted to dealing with this type of situation .
the following lemma will be used to control the number of incidences that can occur between points and good surfaces that lie inside of an algebraic variety in @xmath2 .
a proof of the lemma can be found in appendix [ corofthomtypeboundlem2 ] .
[ controlofsmoothptslem ] let @xmath14 be a collection of good surfaces , let @xmath13 be a collection of points , let @xmath255 be a collection of admissible incidences , and let @xmath256 be a variety with @xmath257 .
suppose that for each @xmath258 we have @xmath259 , and that for each @xmath260 we have @xmath261 .
then @xmath262 in order to use the polynomial ham sandwich cell decomposition described above , we need to control how many cells a given surface @xmath30 can enter .
this is accomplished by the following theorem : [ thomtypeboundlem1 ] let @xmath198 $ ] be a real polynomial of degree @xmath201 whose irreducible components generate real ideals , and let @xmath26 be a 2dimensional irreducible real variety of degree @xmath263 .
then @xmath26 intersects at most @xmath264 connected components of @xmath265 , where the implicit constant depends only on @xmath118 .
we will need similar control on the number of times a surface @xmath30 can enter a 3dimensional cell on the variety @xmath185 .
[ corofthomtypeboundlem2 ] let @xmath198 $ ] be an irreducible real polynomial of degree @xmath201 that generates a real ideal , and let @xmath266 $ ] be a collection of real polynomials , such @xmath267 , and for each @xmath166 , the irreducible components of @xmath77 generate real ideals .
let @xmath26 be a 2dimensional surface of degree @xmath263 .
then @xmath26 meets at most @xmath268 realizations of realizable strict sign conditions of @xmath164 on @xmath185 .
the proofs of lemmas [ thomtypeboundlem1 ] and [ corofthomtypeboundlem2 ] can be found in appendix [ proofappendix ] .
a similar result to lemma [ corofthomtypeboundlem2 ] can be obtained from the recent work of barone and basu in @xcite .
[ propertiesofrealcplxvar ] let @xmath147 be a real variety and let @xmath269 @xmath270 be the irreducible components of @xmath161 .
then @xmath271 are the irreducible components of @xmath76 .
furthermore , for each @xmath272 @xmath273 so in particular @xmath274 is non - empty .
[ completecomplexintersectioncor ] if @xmath275 $ ] are irreducible and @xmath276 are real ideals such that @xmath277 , then @xmath278 is a complete intersection .
the proofs of theorems [ surfacesinr4thm ] and [ surfacesinr4thm]@xmath117 are very similar , so wherever possible we shall deal with both theorems at the same time . thus we will frequently divide our arguments into the cases @xmath73 ( corresponding to theorem [ surfacesinr4thm ] ) and @xmath279 ( corresponding to theorem [ surfacesinr4thm]@xmath117 ) .
let @xmath280 @xmath281 and @xmath15 be as in the statements of theorems [ surfacesinr4thm ] and [ surfacesinr4thm]@xmath117 .
first , it suffices to prove theorems [ surfacesinr4thm ] and [ surfacesinr4thm]@xmath117 in the special case where all the surfaces @xmath30 are irreducible .
if the surfaces are not irreducible , then there exists a constant @xmath16 such that each @xmath30 can be written @xmath282 , where each @xmath283 is an irreducible 2dimensional surface ( technically , @xmath26 may contain irreducible components that are 0 or 1dimensional , but by property [ admissibledefnsmooth ] of a admissible intersection , no points may lie on such components , so we may discard them ) . for each index @xmath284 , let @xmath285 ( where @xmath286 if @xmath26 contains fewer than @xmath287 irreducible components ) .
we can verify that each collection @xmath288 of surfaces is good , and that @xmath289 is admissible .
we have @xmath290 and the surfaces in each collection @xmath288 are irreducible .
thus if we can establish theorem [ surfacesinr4thm ] in the special case where all the surfaces are irreducible , we can apply the result to each term in to recover the general result .
thus we shall assume that all surfaces in @xmath14 are irreducible .
first , from lemma [ turanboundlemma ] , we have latexmath:[\[\begin{aligned } this immediately gives us and @xmath117 , and it also gives us , @xmath117 , , and if @xmath292 or @xmath293 for some small constant @xmath294 to be chosen later .
thus we may assume @xmath295 theorem [ surfacesinr4thm ] will follow from the following two theorems : [ basecasethm ] let @xmath14 be a good collection of surfaces , @xmath13 a collection of points , and @xmath296 a collection of @xmath15admissible incidences .
suppose @xmath297 then @xmath298 if @xmath73 , and @xmath299 if @xmath300 [ inductionthm ] let @xmath14 be a good collection of surfaces , @xmath13 a collection of points , and @xmath296 a collection of @xmath15admissible incidences .
then there exists the following : 1 .
a number @xmath125 satisfying @xmath301 is some small absolute constant .
numbers @xmath302 with each @xmath303 and @xmath304 and numbers @xmath305 with each @xmath306 and @xmath307 3 .
collections @xmath308 with @xmath309 , with @xmath310 4 .
collections @xmath311 with @xmath312 such that for each index @xmath95 , @xmath313 these collections have the property that @xmath314 where @xmath315 if @xmath73 , and @xmath316
if @xmath300 we will first prove the bounds and ( [ surfacesinr4thmintm]@xmath117 ) by establishing the following induction step .
there exists a small absolute constant @xmath317 with the following property .
if there exists some @xmath318 and a constant @xmath319 for which and ( [ surfacesinr4thmintm]@xmath117 ) hold for that value of @xmath320 , then there exists a constant @xmath321 for which and ( [ surfacesinr4thmintm]@xmath117 ) hold for all values of @xmath322 if @xmath323 then and ( [ surfacesinr4thmintm]@xmath117 ) hold by theorem [ basecasethm ] .
once this fact has been established , it suffices to note that for any @xmath229 , if @xmath324 then we obtain the bounds and ( [ surfacesinr4thmintm]@xmath117 ) for @xmath62 after at most @xmath121 iterations of the above procedure .
thus it suffices to establish the induction step .
suppose or ( [ surfacesinr4thmintm]@xmath117 ) holds for all @xmath325 .
let @xmath326 .
apply the decomposition from theorem [ inductionthm ] .
we can assume that for each index @xmath287 , @xmath327 where @xmath118 is the implicit constant from . if this is not the case for some @xmath287 , simply add admissible surfaces to @xmath288 at random .
doing so will not affect @xmath281 and it will increase @xmath328 by at most a factor of 2 .
now , for each index @xmath287 , we have @xmath329 i.e. @xmath330 thus @xmath331 where in the final line we used and the assumption that @xmath109 is sufficiently large to absorb the constant @xmath332 into the exponent ( if @xmath333 is not sufficiently small then the entire result is trivial ) . similarly , for each index @xmath334 , we have @xmath335 so @xmath336 now , recall that we selected @xmath62 so that @xmath337 then by our induction hypothesis , @xmath338 if @xmath73 , and @xmath339 if @xmath279 .
thus if @xmath73 , we obtain the following bound ( for clarity , we shall write the computation for general @xmath15 ) : @xmath340\\ & \leq c_{\rho_0}\big[c m^{\frac{k}{2k-1 } } a^{-\frac{4k-4}{2k-1 } } n^{\frac{2k-2}{2k-1}}a^{\frac{4k-4}{2k-1}}+m+ca^2n\big]\\ & \leq c_{\rho_0}\cdot c\big[m^{\frac{k}{2k-1 } } n^{\frac{2k-2}{2k-1 } } + m+a^2n\big ] , \end{split}\ ] ] and if @xmath341 @xmath342.\ ] ] similarly , if @xmath73 , @xmath343\\ & \leq c_{\rho_0}\big[c\sum_i m_i^{\frac{k}{2k-1}}(a_ia^3)^{-\frac{k-1}{2k-1 } } n^{\frac{2k-2}{2k-1 } } ( a_ia)^{\frac{2k-2}{2k-1}}+m + ca^2 n\big]\\ & \leq c_{\rho_0}\big[cn^{\frac{2k-2}{2k-1}}\sum_i m_i^{\frac{k}{2k-1}}(a_i / a)^{\frac{k-1}{2k-1}}+m + ca^2 n\big]\\ & \leq c_{\rho_0}\big[cn^{\frac{2k-2}{2k-1 } } m^{\frac{k}{2k-1 } } \big(\sum_i(a_i / a)\big)^{\frac{k-1}{2k-1}}+m + ca^2 n\big]\\ & \leq c_{\rho_0}\cdot c\big[n^{\frac{2k-2}{2k-1 } } m^{\frac{k}{2k-1 } } + m + a^2 n\big ]
. \end{split}\ ] ] and if @xmath279 , @xmath344 .
\end{split}\ ] ] now , by , @xmath345 and thus for @xmath73 we have @xmath346 while for @xmath279 we have @xmath347 , we obtain and ( [ surfacesinr4thmintm]@xmath117 ) and for all values of @xmath348 .
we can apply similar reasoning to establish and ( [ surfacesinr4thmallvalues]@xmath117 ) .
the key idea is that by iterating theorem [ inductionthm ] about @xmath349 times , we reduce to the case of counting incidences between collections of points and curves of bounded size , and we can apply trivial bounds . on the other hand , each application of theorem [ inductionthm ] introduces a multiplicative constant @xmath16 into our bounds , so our final bounds are sharp except for a @xmath350 factor .
we will now make this intuition precise . the case @xmath73 and @xmath279
can be dealt with simultaneously .
suppose holds for all collections @xmath351 with @xmath352 .
apply theorem [ inductionthm ] to the collection @xmath353 then @xmath354\\ & \leq c\cdot c_0 \big[\big(a^{4 - 4k } m^k ( \log m-4\log a)^{c_1(2k-1)}\big)^{\frac{1}{2k-1}}\big(a^2 n \big)^{\frac{2k-2}{2k-1 } } \\
& \qquad\qquad+ m + a^2n \big]\\ & \leq c\cdot c_0 m^{\frac{k}{2k-1}}n^{\frac{2k-2}{2k-1}}(1 - 4c_0)^{c_1}\log m + cm + ca^2n . \end{split}\ ] ] similarly , @xmath355 let @xmath356 be the supremum of the constant appearing on the final lines of and .
if we make @xmath60 large enough , @xmath357 using , and assuming @xmath0 is sufficiently large ( if @xmath0 is not sufficiently large then the entire result is trivial ) , we have @xmath358 thus , @xmath359 this closes the induction . in this section
, we shall prove the base case for our induction .
we shall prove the theorem by induction on @xmath0 .
for @xmath0 sufficiently small the result is immediate , since we can chose a large constant in the quasi - inequalities and ( [ basecaseeqn]@xmath117 ) .
let @xmath360 which by satisfies @xmath361 let @xmath74 be a squarefree polynomial of degree at most @xmath201 such that @xmath362 cuts @xmath2 into @xmath363 cells @xmath364 , such that @xmath365 .
let @xmath366 be the number of surfaces in @xmath14 that meet cell @xmath95 . by lemma [ thomtypeboundlem1 ] , @xmath367 and
thus if we apply lemma [ turanboundlemma ] within each cell , we obtain @xmath368 write @xmath369 where @xmath370 ( resp .
@xmath371 ) consists of those surfaces that are contained ( resp .
not contained ) in @xmath76 . by lemma [ controlofsmoothptslem ] , we have @xmath372 .
let @xmath373 be the square - free part of @xmath374 , i.e. @xmath375 is radical and @xmath376 .
since @xmath74 was square - free , @xmath377 is a complete intersection .
if we let @xmath378 be those surfaces contained in @xmath377 , then we must have @xmath379 , and thus applying lemma [ turanboundlemma ] , we have @xmath380 be those surfaces ( contained in @xmath76 ) that are not contained in @xmath381 .
we must now control @xmath382 and @xmath383 .
but note that @xmath76 and @xmath381 are both the zero - set of polynomials of degree @xmath384 , and thus the two collections of incidences can be dealt with in the same fashion .
we shall now control @xmath385 .
factor @xmath386 , with each @xmath387 irreducible of degree @xmath388 , and let @xmath389 .
let @xmath390 so that @xmath391 ; if the same point @xmath84 lies on several @xmath392 , place @xmath84 into just one of the sets @xmath393 ( the choice of set does not matter ) .
let @xmath394 we have @xmath395 thus by the induction hypothesis , we have @xmath396 if @xmath397 and @xmath398 if @xmath300 if @xmath294 is selected sufficiently small , the contribution from is acceptable .
let us now suppose @xmath399 .
use lemma [ varianthamsandwichthm ] to find a collection @xmath400 of polynomials with @xmath401 , where @xmath402 such that each strict sign condition of @xmath400 on @xmath392 contains @xmath403 points .
let @xmath404 be the number of points of @xmath393 that lie in the @xmath95th strict sign condition of @xmath400 on @xmath392 , and let @xmath405 be the number of surfaces in @xmath371 that meet the @xmath95th strict sign condition of @xmath400 on @xmath392 .
let @xmath406 .
by lemma [ corofthomtypeboundlem2 ] , @xmath407 thus @xmath408 it remains to control @xmath409 .
we shall defer this step until the next section .
thus , if @xmath399 then @xmath410 and thus by hlder s inequality , @xmath411 [ crossingnumberineqlemma ] let @xmath80 be a 2dimensional variety of the form : * @xmath412 , with @xmath413 . * @xmath414 @xmath415 * @xmath387 is irreducible generates a real ideal .
@xmath416 is a product of irreducible polynomials , each of generates a real ideal ( and thus @xmath392 and @xmath416 are hypersurfaces ) .
* @xmath416 is not divisible by @xmath387 .
let @xmath417 be a collection of points , with @xmath418 let @xmath14 be a collection of admissible surfaces , with @xmath54 such that no surface in @xmath14 is contained in @xmath80 .
let @xmath255 be a collection of @xmath15good incidences .
then the following holds : * if @xmath73 , then @xmath419 * if @xmath279 , then @xmath420 we shall defer the proof of lemma [ crossingnumberineqlemma ] to section [ proofofcrossingnumberineqlemmasection ] below . applying lemma [ crossingnumberineqlemma ] to the collection @xmath421 with
@xmath422 we see that if @xmath73 , then @xmath423 here we used the assumption @xmath424 and thus @xmath425 if @xmath279 then a similar computation shows @xmath426 combining , , and we obtain latexmath:[\[\label{pcapzsphrs2k2 } if @xmath73 , and latexmath:[\[\tag{\ref{pcapzsphrs2k2}$^\prime$ } if @xmath300 however , an identical argument can be used to obtain analogous bounds for @xmath429 . combining , , , , and either or ( [ pcapzsphrs2k2]@xmath117 ) , we obtain @xmath430 if @xmath73 , and latexmath:[\[\tag{\ref{boundonincidencesk2}$^\prime$ } if @xmath279 .
we shall now prove the induction step theorem .
the proof will follow a similar outline to that of theorem [ basecasethm ] , and we shall gloss over those steps that are identical . let @xmath432^{1/2}.\ ] ] by , @xmath125 satisfies @xmath433 as in the previous section , let @xmath74 be a squarefree polynomial of degree at most @xmath125 such that @xmath216 cuts @xmath2 into @xmath434 cells @xmath364 such that if @xmath435 then @xmath436 let @xmath437 . by lemma [ thomtypeboundlem1 ] , @xmath438
we have that the collections @xmath439 satisfy the requirements from and .
it remains to find appropriate collections of points and surfaces that contain @xmath440 .
write @xmath441 as in .
let @xmath442 by lemma [ controlofsmoothptslem ] , @xmath443 .
let @xmath373 be the squarefree part of @xmath374 .
let @xmath378 be those surfaces contained in @xmath377 , so @xmath379 .
let @xmath444 by lemma [ turanboundlemma ] , we have @xmath445 be those surfaces ( contained in @xmath76 ) that are not contained in @xmath381 .
we must now control @xmath446 and @xmath447 .
factor @xmath448 and partition @xmath449 as in the previous section .
let @xmath450 .
for each index @xmath287 , apply lemma [ varianthamsandwichthm ] to the data @xmath451 and denote the resulting collection @xmath452 let @xmath453 be those points of @xmath393 that lie in the @xmath95th strict sign condition of @xmath400 on @xmath392 , and let @xmath454 be those surfaces in @xmath371 that meet the @xmath95th strict sign condition of @xmath400 on @xmath392 .
we can verify that @xmath455 and @xmath456 satisfy the requirements from .
we shall perform a similar decomposition for the points of @xmath457 , and denote the resulting collections @xmath458 , @xmath459 .
again , @xmath460 and @xmath461 satisfy the requirements from , and thus the collection @xmath462 satisfies the properties from . finally , it remains to control the `` boundary incidences , '' @xmath463 where @xmath464 and @xmath465 is defined similarly .
but by lemma [ crossingnumberineqlemma ] , @xmath466 if @xmath73 , and latexmath:[\[\tag{\ref{controlincidences3k2}$^\prime$ } if @xmath279 .
the analogous bounds hold for @xmath468 thus if we let @xmath469 we can verify that our collections @xmath470 and @xmath471 , satisfy the conclusions of theorem [ inductionthm ] . before we begin the proof , let us first recall szkely s proof in @xcite of the szemerdi - trotter theorem , which uses the crossing number inequality .
let @xmath46 be a collection of points and @xmath47 a collection of lines in @xmath11 .
let @xmath472 be the number of incidences between points in @xmath46 and lines in @xmath47 .
suppose that all of the points ( and thus all of the incidences ) are contained in some large disk @xmath473 consider the following graph drawing @xmath134 : the vertices of @xmath134 are the points of @xmath46 and the points where a line from @xmath47 meets @xmath474 .
the edges of @xmath134 are the line segments connecting two vertices that arise from lines in @xmath47 .
to each incidence @xmath475 , we can associate an edge of @xmath134 in such a way that the same edge is assigned to at most two incidences .
thus @xmath476 now , delete all of the edges involving a vertex on @xmath474 , and delete the vertices on @xmath474 .
we have deleted at most @xmath477 edges .
let @xmath478 be the resulting graph drawing
. then @xmath479 . by the crossing number inequality , @xmath480
thus @xmath481 we wish to do the same thing on the real algebraic varieties @xmath482 . in order to do that , we will have to formulate the above proof in a slightly more general setting .
[ defnofcurveptincidences ] let @xmath125 be a collection of open curves ( i.e. homeomorphic images of @xmath483 ) and @xmath121 a collection of points on a planar domain @xmath484 then we define the number of incidences between curves in @xmath125 and points in @xmath121 to be @xmath485 and we define the number of crossings of curves in @xmath121 to be @xmath486 [ stforlinesonhomeo ] let : * @xmath487 be a smooth 2dimensional manifold that is homeomorphic to an open subset of @xmath11 .
* @xmath125 be a collection of 1dimensional open curves lying on @xmath487 .
* @xmath121 be a collection of points on @xmath487 .
* let @xmath16 be a constant so that for any two points @xmath488 there are at most @xmath16 curves @xmath489 with @xmath490 and @xmath491 for each @xmath284 .
then @xmath492 where the implicit constant depends only on @xmath16 .
note that in definition [ defnofcurveptincidences ] , a point is incident to an ( open ) curve if it lies on the _ closure _ of that curve . on the other hand ,
a crossing of two curves is a point common to the relative interior of both curves .
similarly , in lemma [ stforlinesonhomeo ] we require that for any two points , there are at most @xmath493 curves which contain those points in their relative interiors there may be arbitrarily many curves whose closures contain the two points .
we also have the slightly weaker bound for `` curves with @xmath15 degrees of freedom '' : let @xmath494 and @xmath495 be as in lemma [ stforlinesonhomeo ] , except in place of the final item we require : for any @xmath15 points @xmath496 , there are at most @xmath16 curves from @xmath125 that contain @xmath497
. then @xmath498 lemmas [ stforlinesonhomeo ] and [ stforlinesonhomeo]@xmath117 are useful when combined with the following lemma : [ decompofylem ] let @xmath499 be a 2dimensional real variety of the following form : * @xmath412 , where each @xmath413 is a 2dimensional real algebraic set that is a ( set - theoretic ) complete intersection of two ( real ) hypersurfaces * @xmath500 with @xmath501 for some small absolute constant @xmath294 .
let @xmath48 be a collection of points , let @xmath14 be a collection of good surfaces , and let @xmath296 be a collection of admissible incidences .
suppose further that for each @xmath37 there does not exist an index @xmath287 for which @xmath84 is an isolated point of @xmath502 thus for each @xmath30 and for index @xmath287 , either @xmath503 , or @xmath84 lies on a 1dimensional component of @xmath504 .
then there exists an `` error set '' @xmath505 and an `` incidence model '' @xmath506 , where for each @xmath95 , * @xmath507 is homeomorphic to an open subset of @xmath11 .
* @xmath508 is a collection of open curves ( homeomorphic to @xmath483 ) contained in @xmath509 .
* @xmath510 is a collection of points .
* @xmath511 is an embedding .
@xmath512 and @xmath471 have the following properties : 1 .
@xmath512 does not increase crossing number : @xmath513 2 .
@xmath512 counts incidences : if @xmath514 , then there exists some index @xmath95 , some @xmath515 , and some @xmath516 such that @xmath517 , @xmath518 @xmath519 , and @xmath520 .
[ kdegfreedomitm ] the curves in @xmath508 have @xmath15 degrees of freedom : for each index @xmath95 , and for each collection of @xmath15 points @xmath521 there are at most @xmath16 curves in @xmath508 passing through each of @xmath497 , where @xmath16 is the constant from definition [ gooddefn ] .
4 . @xmath512 does not contain too many curves : @xmath522 5 .
@xmath471 is not too big : @xmath523 and [ decompofylem ] , we shall prove lemma [ crossingnumberineqlemma ] .
let @xmath524 @xmath525 @xmath236 and @xmath296 be as in the statement of lemma [ crossingnumberineqlemma ] .
we define @xmath526 note that since @xmath527 for any @xmath258 @xmath528 consists of those pairs @xmath529 such that for each index @xmath287 , either @xmath503 or @xmath84 lies on a 1dimensional component of @xmath504 .
we shall first deal with @xmath530 .
fix a choice of @xmath26 and @xmath287 .
since @xmath531 , @xmath532 consists of 0 and 1dimensional irreducible components . by harnack
s theorem ( see remark [ harnackremark ] below ) , @xmath532 can contain at most @xmath533 0dimensional components .
now , consider the collection @xmath534 of 1dimensional irreducible components of @xmath532 .
we have @xmath535 . for each curve @xmath536 , either @xmath537 , or @xmath538 is a discrete set .
let @xmath539 and @xmath540 denote those sets where the former ( resp .
latter ) occurs .
if @xmath541 then we must have @xmath542 , and thus for any point @xmath543 that lies on @xmath544 we must have @xmath545 .
if @xmath546 , then by bzout s theorem ( over @xmath179 ) , @xmath547 and thus there are at most @xmath548 points @xmath543 for which @xmath549 and @xmath550 .
summing over all curves in @xmath551 we conclude @xmath552 [ harnackremark ] while harnack s theorem deals with plane algebraic curves and @xmath532 is a space curve , since @xmath532 is a proper intersection , we can select a generic projection @xmath553 .
the image curve @xmath554 may have fewer connected components than the original curve , but the decrease is controlled by the number of singular points of @xmath554 , and this in turn is controlled by @xmath533 summing over all indices @xmath287 and all choices of @xmath30 and adding back the incidences that lie on 0dimensional components of @xmath532 for al choices of @xmath258 we obtain @xmath555 .
use lemma [ decompofylem ] on the data @xmath556 . denote the resulting incidence model @xmath512 and the resulting error set @xmath557 .
we have @xmath558 , , and follow from the definition of @xmath559 and the observation that if @xmath560 @xmath561 i.e. an open curve only has two ends .
now , if @xmath73 , apply lemma [ stforlinesonhomeo ] to each collection @xmath562 .
we have @xmath563 where in the final line we used and the fact that @xmath564 if @xmath341 apply lemma [ stforlinesonhomeo]@xmath117 to each collection @xmath562 .
we get @xmath565 this concludes the proof of lemma [ crossingnumberineqlemma ] , modulo the proofs of lemmas [ stforlinesonhomeo ] and [ decompofylem ] . since all of the quantities we wish to consider are invariant under homeomorphism , without loss of generality we can assume @xmath487 is an open subset of @xmath11 .
now , replace each curve @xmath566 with a slightly `` shrunk '' curve @xmath567 , so that @xmath568 does not meet any point from @xmath121 nor any curve from @xmath125 .
if @xmath569 denotes the set of shrunk curves , then @xmath570 delete from @xmath569 those curves @xmath567 that are incident to fewer than 2 points from @xmath121 , and denote the resulting set of curves @xmath571
. then @xmath572 consider the drawing @xmath134 of the multi - graph whose vertices are the points of @xmath121 and where two vertices are connected by an edge if the two corresponding points are joined by a curve from @xmath571 .
then @xmath134 is an admissible graph drawing .
the multigraph @xmath134 need not be a graph , since two vertices can be connected by several edges . however , the maximum edge multiplicity of @xmath134 is bounded by some constant @xmath16 .
furthermore , @xmath573 , so by theorem [ crossingnumberineq ] , @xmath574 the proof of lemma [ stforlinesonhomeo]@xmath117 is very similar to pach and sharir s proof in @xcite of the szemerdi - trotter theorem for curves with @xmath15 degrees of freedom , but there are a few key places where the proof differs ( and the final conclusion of lemma [ stforlinesonhomeo]@xmath117 is slightly weaker ) . for completeness , we have included a sketch of lemma [ stforlinesonhomeo]@xmath117 in appendix [ pachsharirappendix ] .
an earlier result similar to lemma [ stforlinesonhomeo]@xmath117 was proved by pach and sharir in @xcite . there , the authors required that the curves in @xmath125 be algebraic curves defined by @xmath15 real parameters , i.e. there is an assignment from these @xmath15 parameters to the curves such that any choice of parameters defines at most a constant number of curves , and the dependence of the curves on the parameters is algebraic of low degree .
while the curves we are dealing with are algebraic and are defined by @xmath15 parameters , the dependence of the curves on the parameters is of high degree , so the result from @xcite is not sufficient for our purposes .
we shall begin by constructing the `` error set '' @xmath471 : @xmath577 first , note that @xmath578 is an algebraic curve , and @xmath579 now , if @xmath580 and there exist two distinct surfaces @xmath581 such that both @xmath529 and @xmath582 are in @xmath583 then @xmath106 must contain a 1dimensional component of @xmath584 so in particular @xmath106 is not a discrete set , which contradicts the fact that @xmath14 is a good collection of surfaces .
thus for each @xmath585 there exists at most one @xmath30 for which @xmath586 .
thus it suffices to control @xmath587 however , is bounded by the degree of @xmath588 which is @xmath589 this establishes .
now for each @xmath258 we shall partition the curves of @xmath590 that do not lie in @xmath578 into classes @xmath591 . write @xmath85 as a union of irreducible real curves and isolated points . to each real curve @xmath592 , either @xmath593 or there is a unique index @xmath287 for which every 1dimensional connected component of @xmath592 lies in @xmath594 ( note that @xmath592 may have 0dimensional components , which need not lie in @xmath595 but by our assumptions on @xmath296 no incidences can occur on these components ) . if @xmath596 , place @xmath592 in @xmath591 for this choice of @xmath287 .
similarly , partition @xmath13 into collections @xmath597 , where @xmath598 if @xmath287 is the minimal index for which @xmath599 then if @xmath600 , @xmath601 , then there is a unique index @xmath287 such that @xmath598 and there exists a curve @xmath602 with @xmath549 . the only way this can fail to be the case is if @xmath603 and every irreducible curve @xmath604 containing @xmath84 lies in @xmath605 for some index @xmath606 . but
this would imply that @xmath84 is an isolated point of @xmath607 and by assumption @xmath296 does not contain any pairs @xmath608 for which this can occur .
thus @xmath609 we wish to consider each variety @xmath594 and the associated points and curves on that variety separately .
however , before we do this we need to ensure that the same intersection @xmath106 does not get counted as two distinct `` crossings '' on two different varieties ( say @xmath594 and @xmath610 ) . to prevent this from happening ,
we will keep track of the points where such an accident might occur . to this end , for each @xmath30 and each index @xmath287 let @xmath611 [ defnofcurvemult ] let @xmath602 , and let @xmath612 .
then we define @xmath613 as follows .
let @xmath567 be a generic projection of @xmath592 onto @xmath11 , let @xmath614 be the image of @xmath34 under the same projection , and let @xmath615 be the squarefree polynomial with @xmath616 .
then @xmath613 is the order of vanishing of @xmath615 at @xmath614 . in particular , if @xmath617 then @xmath618 . by the same arguments used above to obtain , @xmath619 on the other hand , we have @xmath620 indeed , the only way could fail is if the following occurs : there exist surfaces @xmath26 and @xmath621 a point @xmath622 , indices @xmath287 and @xmath623 and curves @xmath624 and @xmath625 such that @xmath626 and @xmath627 but if holds then @xmath34 is an isolated point of @xmath628 so @xmath629 and thus fails . this establishes .
thus , we can consider each variety @xmath594 and its associated point and curve sets @xmath630 individually .
we are reduced to proving the following lemma .
[ incidencemodelonecomponent ] suppose we are given the following data : * a 2dimensional real variety of the form @xmath413 with @xmath631 * a collection of points @xmath13 , a collection of good surfaces @xmath525 and for each @xmath30 , a collection @xmath591 of irreducible curves in @xmath504 such that for each @xmath632 @xmath633 .
* a collection of @xmath15admissible incidences @xmath634 such that for each @xmath260 , there exists @xmath602 with @xmath635 * for each @xmath258 a finite collection of `` bad '' points @xmath636 then there exists an incidence model @xmath637 satisfying : 1 .
[ incidencemodelonecomponentitm1 ] for each @xmath95 , @xmath508 is a collection of simple open curves , @xmath638 is a collection of points , @xmath509 is homeomorphic to a open subset of @xmath11 , and @xmath511 is an embedding .
[ incidencemodelonecomponentitm2 ] + ' '' '' + @xmath639 + i.e. the number of crossings between pairs of curves in the incidence model is controlled by the number of crossings of the curves from @xmath591 and @xmath640 that do not occur on the `` bad '' sets @xmath641 or @xmath642 as @xmath26 and @xmath39 range over all pairs of surfaces .
[ incidencemodelonecomponentitm3][2ndlemma2nditem ] if @xmath543 , @xmath30 , and @xmath632 then there exists some index @xmath95 , some @xmath643 and some @xmath516 such that @xmath644 @xmath645 and @xmath520 , i.e. the incidence model counts curve - point incidences .
[ incidencemodelonecomponentitm4 ] for each @xmath95 , the curves in @xmath508 have @xmath15 degrees of freedom ( in the sense of item [ kdegfreedomitm ] from lemma [ decompofylem ] ) .
[ incidencemodelonecomponentitm5 ] + ' '' '' + @xmath646 + where @xmath613 is as defined in definition [ defnofcurvemult ] .
once we have lemma [ incidencemodelonecomponent ] , we can prove lemma [ decompofylem ] as follows . for each index
@xmath647 apply lemma [ incidencemodelonecomponent ] to each collection @xmath648 @xmath649 and denote the resulting incidence model @xmath650 @xmath651 . from and , we have @xmath652 if @xmath37
then as noted above there exists a unique index @xmath287 and a curve @xmath602 with @xmath598 and @xmath549 .
but then by property [ 2ndlemma2nditem ] , there exists some index @xmath95 , some @xmath653 some @xmath654 such that @xmath655 finally , @xmath656 where on the final line we used the fact that @xmath657 .
thus the incidence model @xmath658 verifies the requirements of lemma [ decompofylem ] .
this concludes the proof of lemma [ decompofylem ] , modulo the proof of lemma [ incidencemodelonecomponent ] .
we shall select a very large ball @xmath659 containing all of the points from @xmath236 and we shall decompose @xmath660 into a union of 2dimensional @xmath661 manifolds . the manifolds will have the property that for a suitably chosen @xmath8plane @xmath662 any affine translate of @xmath663 will intersect a given manifold at most once ( such manifolds are called `` monotone '' in the computational geometry literature ) .
indeed , let @xmath664 [ xjfigure ] is a 2torus in @xmath2 .
in the figure , we have projected @xmath80 into @xmath19 with a projection @xmath665 chosen so that @xmath666 is a vertical line passing through the origin .
the set @xmath667 is denoted by the blue line.,title="fig : " ] if our choices for @xmath668 and @xmath669 in are generic , then @xmath667 may be empty or it may be a union of isolated points and 1dimensional curves .
we can see that each ( necessarily smooth ) connected component of @xmath670 is monotone : first , let @xmath671 be a generic projection in the direction of some vector @xmath672 , so @xmath673 is a line passing through the origin ( see figure [ xjfigure ] ) .
let @xmath195 be a vector so that @xmath674 .
then if @xmath487 is a ( necessarily bounded ) connected component of @xmath675 then @xmath676 is also bounded and connected .
it suffices to show that for any @xmath677 , @xmath678 is a smooth point of @xmath676 , and the line @xmath679 meets @xmath676 solely at the point @xmath678 .
first , suppose @xmath676 is not smooth , and let @xmath678 be a singular point
. then since @xmath487 is smooth and @xmath665 is a local diffeomorphism in a neighborhood of each pre - image point of @xmath680 , in a small neighborhood of @xmath678 , @xmath676 is a union of distinct , smooth 2manifolds , each of which contains @xmath678 .
thus , we can find a nearby point @xmath681 for which @xmath682 meets @xmath676 in at least two distinct points , call them @xmath683 and @xmath684 .
now , let @xmath685 be a smooth path in @xmath487 connecting a pre - image of @xmath683 to a pre - image of @xmath686 .
then @xmath687 is a smooth curve in @xmath688 and @xmath689 always lies in @xmath690 , the geometric tangent cone of @xmath487 . but @xmath690 never contains @xmath691 , which is a contradiction .
thus each compact connected component of @xmath670 is monotone .
we must now count how frequently a curve @xmath602 intersects @xmath692 .
this will be done in the next two lemmas .
[ controlofgammacapxj ] let @xmath280 @xmath693 and @xmath694 be as in the statement of lemma [ incidencemodelonecomponent ] .
select @xmath602 and let @xmath667 be as in . then @xmath695 first , note that if our choices of @xmath668 and @xmath669 in are generic , then for each curve @xmath696 , we have that @xmath697 will be a discrete set of points .
thus , we can guarantee the following things : 1 .
every point of @xmath698 is a smooth point of @xmath592 and a smooth point of @xmath667 . in particular , no point of @xmath698 is an isolated point of @xmath667 .
[ morselikeitem ] at every point @xmath699 we have the following property : let @xmath700 be a sufficiently small ball centered at @xmath701 , so that if @xmath702 is the grassmanian of ( complex ) 2planes in @xmath703 , then the tangent bundle @xmath704 is a smooth 2(complex)-dimensional submanifold of @xmath705 .
let @xmath706 then if @xmath707 is the image of @xmath678 in @xmath708 we have that @xmath709 and @xmath710 intersect transversely at @xmath711 so the intersection is a smooth curve in @xmath705 .
thus in particular , if we apply a small @xmath661 perturbation to the surface @xmath712 , then the image of the perturbed surface in @xmath705 will still intersect @xmath710 transversely
. first , observe : if @xmath713 is a reduced ideal , i.e. if @xmath714 on @xmath715 then @xmath716 where @xmath717(z).\ ] ] then , to compute @xmath718 , it suffices to count the number of intersection points in @xmath719 , and this is a ( set - theoretic ) complete intersection .
we are working over @xmath575 , so we can not appeal directly to bzout s theorem , but we can use arguments similar to those used to obtain . in order to make this argument work
, we will need to perturb @xmath387 and @xmath416 to make @xmath713 a reduced ideal .
doing so will cause @xmath594 to `` split '' ( in a small neighborhood of a smooth point ) into several sheets , and ( locally ) there will be one or more copy or copies of @xmath592 on each sheet . through
careful counting , we can recover the above result .
we will need several elementary results from intersection theory .
further details can be found in standard textbooks such as @xcite . throughout this section
, we will choose an embedding of @xmath703 into @xmath720 this will allow us to identify points in @xmath703 with those in @xmath721 and to identify ( complex , affine ) polynomials with homogeneous polynomials .
[ multiplicitydefn ] if @xmath722,$ ] we will let @xmath723 denote the projective ideal generated by @xmath724 ( here as elsewhere , @xmath724 is identified with its image under the above embedding from @xmath703 into @xmath725 ) .
if @xmath726 $ ] , let @xmath727 be the zero set of @xmath724 ( either in @xmath703 or in @xmath725 , depending on context ) . if @xmath728 then @xmath729 is the local ring obtained by localizing @xmath725 at the point @xmath34 ( again , we have identified @xmath34 with its image in @xmath725 ) .
if @xmath730 are polynomials , we say that @xmath731 intersect properly if @xmath732 if @xmath731 intersect properly , and @xmath733 is an irreducible component of @xmath734 , then we define @xmath735 where @xmath34 is a generic point of @xmath733 .
this definition of multiplicity has the following useful geometric property .
let @xmath731 intersect properly and let @xmath733 be an irreducible component of @xmath734 .
select a generic point @xmath736 and a generic hyperplane @xmath134 of dimension @xmath737 , such that @xmath738 .
then we can find a small ball @xmath700 containing @xmath34 such that @xmath34 is the only point in @xmath739 .
now , let @xmath740 where @xmath741 are generic real numbers with sufficiently small magnitude .
then @xmath742 is a union of @xmath743 points : if @xmath121 is a small ball centered around a generic point of @xmath733 , then if we perturb @xmath744 the intersection `` splits '' into @xmath743 sheets . let @xmath26 be the surface and let @xmath602 be the curve in the statement of lemma [ controlofgammacapxj ] .
let @xmath745 where @xmath195 is a generic vector in @xmath2 .
we can verify that @xmath746 is an irreducible 3dimensional variety of bounded degree : in short , select a rotation of @xmath2 so that @xmath195 is the @xmath747direction . by the tarski - seidenberg theorem ( see e.g. @xcite ) , the projection @xmath748 is a bounded degree algebraic variety , with ideal @xmath749.$ ] let @xmath750 where @xmath751 $ ] is the canonical embedding of @xmath752 into @xmath753 $ ] .
@xmath754 has codimension 1 , so we can write @xmath755 for some polynomial @xmath756 that generates a real ideal . now , @xmath757 is a 1dimensional curve , and thus @xmath757 is a proper intersection .
we also have @xmath758 is a 1dimensional ( complex ) curve and so @xmath758 is a proper intersection .
since @xmath759 we also have @xmath760 , so there exists an irreducible component @xmath761 such that ( as sets ) @xmath762 .
we define @xmath763 similarly . by proposition [ propertiesofrealcplxvar ] , @xmath764 is irreducible .
thus @xmath765 is well - defined .
we define @xmath766 where @xmath34 is a generic point of @xmath764 . since @xmath764 lies generically in @xmath767 there exists a unique irreducible component @xmath733 of @xmath768 that contains @xmath764 , and @xmath769 where the latter multiplicity is given by definition [ multiplicitydefn ] .
we will need a rigorous way to show that if two manifolds intersect transversely , then slight perturbations of the manifolds still intersect transversely .
the following definition will aid in this process .
[ defnofperturbation ] let @xmath487 and @xmath770 be smooth manifolds in some ambient space @xmath771 ( we will have either @xmath772 or @xmath773 .
we say @xmath487 and @xmath770 are @xmath6close
if there exists a diffeomorphism @xmath774 sending @xmath487 to @xmath775 and @xmath776 if @xmath770 is @xmath6close to @xmath487 , we say @xmath770 is an @xmath6perturbation of @xmath487 .
let @xmath777 where @xmath778 are chosen generically from the interval @xmath779 ; @xmath6 will be chosen later .
then @xmath780 is a reduced ideal .
we claim : if @xmath678 is a smooth point of @xmath592 ( and thus @xmath701 is a smooth point of @xmath764 ) , a smooth point of @xmath594 , and a smooth point of @xmath781 , then if @xmath700 is a sufficiently small ball centered at @xmath782 , then : * @xmath783 is a union of @xmath784 smooth disjoint 2manifolds , and each of these 2manifolds is a @xmath127perturbation of @xmath712 .
* @xmath785 is a union of @xmath786 smooth curves , and each of these curves is a @xmath127perturbation of @xmath787 these curves lie on the various connected components of @xmath788 the key observation is the following .
[ numberofperturbedverticalincidences ] select @xmath789 and let @xmath229 .
let @xmath121 be the ball centered at @xmath678 of radius @xmath62 .
then provided @xmath6 is sufficiently small ( depending on @xmath62 ) , we have @xmath790 the idea is to show that each of the @xmath791 1dimensional curves in @xmath792 intersects some 1dimensional curve from @xmath793 . we expect this to happen because the ( unperturbed ) curves intersect transversely , and the perturbation is very small .
indeed , let @xmath794 be a simple curve from @xmath795 then since @xmath796 is smooth , @xmath797 is a @xmath127perturbation of @xmath796 , in the sense of definition [ defnofperturbation ] .
let @xmath798 be the smooth 2manifold containing @xmath797 . then
since @xmath799 is smooth , @xmath800 is a @xmath127 perturbation of @xmath801 .
now , @xmath678 is one of finitely many intersection points of @xmath592 and @xmath667 , and each of these intersections are transverse .
thus if we select @xmath6 sufficiently small ( depending on both the transversality of the intersection and @xmath62 ) , then @xmath797 and @xmath800 must intersect . thus in particular
, @xmath797 intersects @xmath802 but there are @xmath803 such curves @xmath797 .
this establishes the lemma .
we claim : if @xmath6 is sufficiently small , then there exists a curve @xmath804 which is the `` image '' of @xmath764 under the above perturbation . indeed , we can select a small constant @xmath294 so that if @xmath805 is the set of points that are distance at least @xmath806 from any point of @xmath807 that does not lie in @xmath764 , then @xmath132 contains an open interval .
let @xmath808 be a point contained in the relative interior of this open interval , and let @xmath809 be a small ball centered at @xmath678 . then if we select @xmath6 sufficiently small , then @xmath810 is a union of curves , each of which is a @xmath294perturbation of @xmath811 .
let @xmath812 be the smallest algebraic set containing @xmath813 we have @xmath814 @xmath812 corresponds to the intuitive notation of the `` image '' of @xmath764 under the perturbation .
we can now bound the degree of @xmath815 let @xmath816 be a generic @xmath817plane . then @xmath818 .
but by the definition of multiplicity above , to each point @xmath819 , we can find a small ball @xmath121 centered at @xmath34 such that @xmath820 thus @xmath821 lemma [ numberofperturbedverticalincidences ] can be rephrased as the statement @xmath822 i.e. @xmath823 but this and imply that @xmath824 this concludes the proof of lemma [ controlofgammacapxj ] .
[ controlofgammacapysing ] let @xmath525 @xmath693 and @xmath694 be as in the statement of lemma [ incidencemodelonecomponent ] , and let @xmath825 then @xmath826 let @xmath746 be as defined in .
for each @xmath632 let @xmath827 and @xmath764 be defined as above
. then if @xmath828 there must be smooth points of @xmath594 in every ( euclidean ) neighborhood of @xmath34 ( @xmath602 implies @xmath592 lies generically on @xmath829 ) .
thus @xmath782 is a singular point of @xmath830 , so @xmath831 from this we can conclude that @xmath782 is a singular point of the ( not necessarily irreducible ) curve @xmath758 . furthermore , the number of intersection points of @xmath764 with @xmath832 ( counting multiplicity ) as @xmath592 ranges over all curves in @xmath591 is controlled by the number of singular points of @xmath758 that occur on @xmath833 ( again , counting multiplicity ) .
let @xmath834 be the union of all irreducible curves in @xmath758 that are not contained in @xmath833 .
then the number of singular points of @xmath758 that occur on @xmath833 is at most the number of singular points of @xmath835 plus the number of intersection points of @xmath797 with @xmath833 ( again , counting multiplicity ) .
the former quantity is at most @xmath836 and the latter is at most @xmath837 .
we can now complete the proof of lemma [ incidencemodelonecomponent ] . fix a choice of @xmath825 for
each @xmath632 delete the points that lie in @xmath838 the points of @xmath839 the points of @xmath840 and the points that lie on some @xmath841 we can verify that after these points have been removed , the remaining set is a disjoint union of connected 1dimensional manifolds .
furthermore , the number of 1manifolds is @xmath842 indeed , every time a point @xmath34 is removed , the number of manifolds can increase by at most @xmath843 .
if @xmath844 then @xmath845 , so the number of curves added by removing points of this type is at most @xmath846 we can apply similar bounds to control the number of manifolds that are added when the other types of points are excised .
now , each of these 1manifolds is homeomorphic to either the interval @xmath483 or the circle @xmath847 . for those manifolds that are homeomorphic to circles , remove two points at random to obtain two curves homeomorphic to @xmath483 .
our new collection of simple curves has the same cardinality ( up to a factor of two ) , and each curve lies entirely within a single component of @xmath670 . let @xmath848 be the connected components of @xmath849 and let @xmath508 consist of those simple curves lying in @xmath509 , as @xmath592 ranges over @xmath850 .
let @xmath851 and let @xmath511 be the canonical embedding of @xmath509 into @xmath2 .
from we have @xmath852 so requirement [ incidencemodelonecomponentitm5 ] is satisfied . in order to verify the remaining requirements we shall introduce some notation
let @xmath516 for some index @xmath95 .
then there exists a unique surface @xmath853 and a unique curve @xmath854 such that @xmath855
. we will define @xmath856 to be @xmath857 and we will define @xmath858 to be @xmath859 . we can verify that @xmath860 i.e. that @xmath861 since the vectors @xmath668 and @xmath669 from were chosen generically .
recall as well that by assumption , @xmath862 .
if @xmath863 @xmath864 then we can immediately verify that either there exists an index @xmath95 and a curve @xmath516 with @xmath865 and @xmath866 , or there exists an index @xmath95 , a curve @xmath516 , and a point @xmath867 in @xmath638 with @xmath868 and @xmath869 thus requirement [ incidencemodelonecomponentitm3 ] is satisfied . to verify requirement [ incidencemodelonecomponentitm2 ] , note that if @xmath870 and @xmath871 then @xmath872 and @xmath873 furthermore , @xmath874 are the unique pair of curves in @xmath302 corresponding to the triple @xmath875 to obtain requirement [ incidencemodelonecomponentitm4 ] , we shall apply a slight perturbation to the curves of @xmath302 as follows : for each curve @xmath516 and point @xmath876 , if @xmath877 but @xmath878 modify @xmath879 in a small neighborhood of @xmath84 so that @xmath880 .
we can always do this in such a way that the number of crossings between @xmath879 and the other curves is not affected , and @xmath879 remains unchanged in a small neighborhood of every point @xmath881 distinct from @xmath84 .
after this perturbation has been performed for every curve , then the collection @xmath882 satisfies requirement [ incidencemodelonecomponentitm4 ] . indeed , since @xmath296 is @xmath15admissible , if there exists an index @xmath95 , a collection of @xmath15 points @xmath521 and a collection of @xmath883 curves in @xmath508 such that each curve is incident to each of @xmath497 , then there must exist two curves @xmath884 from this collection of curves such that @xmath885 .
however , this implies that @xmath886 either contains a singular point of @xmath858 , or it contains a point of @xmath887 ( if @xmath888 ) .
but by our construction of @xmath508 , no such points may lie on any curve in @xmath508 .
thus requirement [ incidencemodelonecomponentitm4 ] is satisfied .
finally , recall that we already established requirement [ incidencemodelonecomponentitm1 ] above ( in the discussion preceding lemma [ controlofgammacapxj ] ) .
theorems [ surfacesinr4thm ] and [ surfacesinr4thm]@xmath117 have several obvious shortcomings , namely the fact that we do not have a uniform bound on the cardinality of a collection of @xmath8admissible incidences @xmath255 which does not depend on the ratio @xmath889 and in addition we loose a @xmath115 factor when bounding the cardinality of a collection of @xmath15admissible incidences .
this gives rise to our first conjecture : [ nolossconjecture ] let @xmath48 be a collection of points , let @xmath14 be a good collection of surfaces , and let @xmath296 be a collection of @xmath15admissible incidences
. then @xmath890 in order to prove conjecture [ nolossconjecture ] for the case @xmath397 it appears that we need a way to deal with incidences between points @xmath543 and surfaces @xmath30 for which @xmath891 is a 1dimensional curve , and @xmath84 lies on @xmath892 here @xmath80 is as defined in lemma [ crossingnumberineqlemma ] . for @xmath893
we need to deal with both this problem and the logarithmic loss introduced by emulating the pach - sharir proof for obtaining bounds on incidences between points and curves with @xmath15 degrees of freedom .
it seems plausible that a cleverer modification of the pach - sharir techniques might remove this logarithmic loss .
we can also ask similar questions in higher dimensions , and here things appear to be more difficult . in @xmath894 we define a family of surfaces @xmath14 to be _ good _ if it satisfies the requirements from definition [ gooddefn ] , except in place of requirement [ gooddefndim2 ] we require that every surface in @xmath14 has dimension @xmath21 .
we define a set of incidences @xmath255 to be @xmath15admissible if it satisfies the requirements from definition [ admissibledefn ] .
we then have the following conjecture .
[ conj1 ] let @xmath13 be a collection of points in @xmath895 @xmath21 even , and @xmath14 a collection of smooth @xmath896dimensional algebraic surfaces of degree @xmath493 such that every two surfaces meet in @xmath493 points and at most @xmath493 surfaces pass through any collection of @xmath15 points .
then latexmath:[\[\label{conjecturedhigherdimbound } the bound is what you would obtain if you performed a cell decomposition on the point set @xmath13 with a polynomial @xmath74 whose degree was chosen to minimize the number of incidences inside the cells , and you ignored incidences occurring on the boundary of the cells , i.e. if you only considered section [ firstdecomp ] from the proof of theorem [ basecasethm ] .
this assumption is satisfied if the points in @xmath13 are in general position in @xmath20 .
conjecture [ conj1 ] is an obvious generalization of a conjecture of tth ( * ? ? ?
* conjecture 3 ) , who proposed conjecture [ conj1 ] in the case where @xmath14 is a collection of affine @xmath21planes .
the conjecture seems plausible because a slight modification of the methods used by solymosi and tao in @xcite yield the following slightly weaker statement : let @xmath898 be as in conjecture [ conj1 ] .
then for any @xmath5 , we have @xmath899 more ambitiously , we could conjecture [ secondconj ] let @xmath13 be a collection of points in @xmath900 and @xmath14 a collection of smooth @xmath901dimensional algebraic surfaces , @xmath902 , of degree at most @xmath16 such that every @xmath903 surfaces meet in at most @xmath16 points and at most @xmath16 surfaces pass through any collection of @xmath15 points , @xmath904 .
then @xmath905 again , the bound in conjecture [ secondconj ] is the one we would obtain if we used the polynomial ham sandwich theorem to cut @xmath20 into cells and applied lemma [ turanboundlemma ] to control each cell , i.e. conjecture [ secondconj ] is saying that the number of incidences between surfaces and points on the boundary of the ( first level ) cell decomposition is controlled by ( the worst case behavior of ) the number of incidences between surfaces and points in the interior of the ( first level ) cell decomposition .
it appears that in order to resolve these conjectures some new techniques must be developed . in particular
, it appears that in order to control the number of incidences between points and @xmath901dimensional varieties , we need to perform @xmath901 polynomial ham sandwich decompositions , with each successive decomposition performed on the variety defined by the previous decompositions .
as @xmath901 increases , the number of cases to be considered increases dramatically , and certain difficulties such as the failure of the connected components of a complete intersection to themselves be complete intersection , the failure of a set - theoretic complete intersection to be a nonsingular complete intersection , etc .
become increasingly problematic .
a second difficulty is that the surfaces in @xmath14 need not form a complete intersection with the polynomial ham sandwich cuts .
this was already the main difficulty for proving theorems [ surfacesinr4thm ] and [ surfacesinr4thm]@xmath117 . in our proof , however , it was only necessary to perform two ham sandwich decompositions , and we were left with a problem about points and curves on a 2dimensional surface , so we could appeal to the crossing number inequality . in higher dimensions
there is no obvious analogue of the crossing number inequality , and thus it is not obvious how to proceed .
let @xmath906 . write @xmath907 as a product of irreducible factors .
place each irreducible factor that generates a real ideal in @xmath164 .
if @xmath387 is a factor that does not generate a real ideal then consider @xmath908 for @xmath909 a generic vector .
we have @xmath910 and @xmath911 apply the above procedure to @xmath908 .
this process will terminate after finitely many iterations .
let @xmath912 .
let @xmath543 , and let @xmath913 suppose there existed surfaces @xmath914 with @xmath915 recall that @xmath84 must be a smooth point of @xmath26 and of @xmath916 . since @xmath259 , we have @xmath917 .
similarly , @xmath918 .
since @xmath296 is @xmath15good , @xmath919 thus we have two affine 2planes , @xmath920 and @xmath921 which meet only at the point @xmath84 , but both are contained in the affine 3plane @xmath134 .
this can not occur .
thus for each point @xmath922 there can exist at most one surface @xmath30 with @xmath923 thus @xmath924 the proof is similar to the proof of lemma 13 in @xcite , so we will only give a brief sketch here .
first , select a large number @xmath373 so that every cell that meets @xmath26 does so within the ball centered at the origin of radius @xmath373 .
let @xmath925 , where @xmath926 .
thus the number of cells of @xmath185 that @xmath26 meets is at most the number of bounded cells of @xmath927 that @xmath26 meets . since the property of @xmath26 meeting a cell is open
, we can apply a small generic translation to @xmath26 and a small generic perturbation to @xmath186 , and doing so can only increase the number of bounded cells that @xmath26 meets .
now , we can find @xmath928 such that @xmath929 @xmath930 is a reduced ideal , and @xmath931 and @xmath932 are of bounded degree .
let @xmath909 be a generic vector in @xmath2 , and let @xmath933 .
then @xmath934 .
now , the number of cells of @xmath927 that @xmath26 enters is controlled by the number of ( necessarily non - singular ) intersection points of @xmath935 @xmath927 , and @xmath936 ( again , see lemma 13 in @xcite for details ) , and this is @xmath264 .
write @xmath937 as a union of irreducible curves , and denote this collection of irreducible curves by @xmath938 . by harnack
s theorem , @xmath939 can have at most @xmath940 components ( again , see remark [ harnackremark ] for details ) . now , for each irreducible curve @xmath941 and each @xmath166 , either @xmath942 or @xmath943 and thus @xmath944 .
we will call intersections of this type `` important '' intersections between @xmath935 @xmath76 and @xmath945 . if @xmath942 , then since @xmath942 , @xmath946 , and thus @xmath592 does not enter _ any _ realizations of realizable strict sign conditions of @xmath164 on @xmath76 . now ,
if @xmath947 is a realization of a realizable sign condition of @xmath164 on @xmath76 , and @xmath948 , then we can associate to the pair @xmath949 an important intersection of @xmath26 , @xmath76 , and @xmath945 for some @xmath166 in such a way that every important intersection is assigned to at most 2 pairs @xmath949 .
thus the number of realizations of realizable strict sign conditions of @xmath164 on @xmath76 is at most @xmath950 .
let @xmath487 be an open subset of @xmath11 , let @xmath951 be a finite collection of points , and let @xmath125 be a collection of simple open curves .
suppose there exists a constant @xmath16 such that the following holds : * for any @xmath15 points from @xmath121 , there are at most @xmath16 curves @xmath489 with @xmath952 and @xmath953 . * any two curves from @xmath125 intersect in at most @xmath16 points .
* the boundary of each curve is disjoint from @xmath121 and does not lie on any other curve .
if we attempt to draw a graph on the region @xmath487 in the same manner as in lemma [ stforlinesonhomeo ] , we run into the difficulty that many curves can connect two vertices , and a priori we do not have a way to bound the number of such curves , so we can not use theorem [ crossingnumberineq ] to obtain useful bounds on the number of edge crossings . to get around this problem ,
we will partition the points in @xmath121 into roughly @xmath955 diadic size ranges based on the degree of the points , i.e. the number of edges meeting each point .
we will use the crossing number inequality to control the number of edges in each piece of this diadic decomposition and then sum the contributions from each piece .
pach and sharir consider the case where @xmath956 is comparable to @xmath957 , and they are able to use some careful arguments to avoid introducing an additional @xmath955 factor into their bounds . unfortunately in our case
@xmath956 may be much smaller than @xmath957 so we can not use their techniques to avoid the @xmath955 factor that appears in . for @xmath958 ,
let @xmath959 be the number of curves in @xmath125 that contain @xmath84 .
define @xmath960 let @xmath961 be the degree ( up to a factor of 2 ) of the points in @xmath962 , and let @xmath963 .
note that @xmath964 , and thus it suffices to control @xmath965 .
define @xmath966 we have @xmath967 applying lemma [ turanboundlemma ] we have @xmath968 we shall now control @xmath969 for @xmath970 .
let @xmath971 be the graph drawing whose vertices are the points @xmath962 , and where if two vertices both lie on a curve @xmath566 with at most @xmath972 points of @xmath962 in between them , then we connect them by the piece of @xmath592 lying between them ( thus in general several edges will be drawn on top of each other , and two vertices may be connected by several different edges ) .
let @xmath973 be the graph obtained from @xmath971 by erasing every edge whose multiplicity exceeds @xmath974 for some large constant @xmath118 to be chosen later .
we have @xmath975 now , every point on a curve @xmath977 belongs to the relative interior of at most @xmath978 edges of @xmath973 lying on @xmath879 , so there is a subgraph @xmath979 with @xmath980 such that no two edges of @xmath981 overlap , i.e. @xmath981 is an admissible drawing of a multigraph with maximum edge multiplicity @xmath982 .
let @xmath983 for @xmath984 , @xmath985 but since @xmath986 , we get @xmath987 and thus @xmath988 99 m. ajtai , v. chvatal , m. newborn , e. szemerdi . crossing - free subgraphs .
discrete math_. 12:912 .
b. aronov , v. koltun , m. sharir .
incidences between points and circles in three and higher dimensions . _ discrete comput .
geom_. 33:185206 .
s. barone , s. basu . refined bounds on the number of connected components of sign conditions on a variety .
s. basu , r. pollack , m. roy .
_ algorithms in real algebraic geometry_. springer , berlin .
j. bochnak , m. coste , m. roy .
_ real algebraic geometry_. springer - verlag , berlin . 1998 .
k. clarkson , h. edelsbrunner , l. guibas , m. sharir , e. welzl .
combinatorial complexity bounds for arrangements of curves and surfaces .
_ discrete comput .
geom_. 5(1):99160 . 1990 .
j. coolidge .
_ a treatise on algebraic plane curves_. dover , new york .
h. edelsbrunner , m. sharir : a hyperplane incidence problem with applications to counting distances . _ the victor klee festschrift , dimacs ser .
theor . comput .
_ 4:253-263 . 1991 .
g. elekes , c. tth .
incidences of not - too - degenerate hyperplanes .
_ computational geometry ( scg05)_. acm , new york : 1621 .
w. fulton . _ intersection theory_. ergebnisse der mathematik und ihrer grenzgebiete .
3 . folge .
springer - verlag , berlin , second edition , 1998 .
l. guth , n. katz . on the erdos distinct distance problem in the plane .
r. harshorne .
_ algebraic geometry_. springer - verlag , new york .
a. iosevich , h. jorati , i. aba .
geometric incidence theorems via fourier analysis . _ trans .
_ 361(12):65956611 . 2009 .
h. kaplan , j. matouek , z. safernova , m. sharir .
unit distances in three dimensions . arxiv:1107.1077v1 .
h. kaplan , j. matouek , m. sharir .
simple proofs of classical theorems in discrete geometry via the guth
katz polynomial partitioning technique .
t. kvari , v. ss , p. turan . on a problem of k. zarankiewicz .
_ colloquium mathematicum_. 3:5057 .
i. aba , j. solymosi .
incidence theorems for pseudoflats .
_ discrete comput . geom .
_ 37(2):163174 . 2007 .
j , landsberg .
differential - geometric characterizations of complete intersections .
_ j. differential geom .
f. leighton .
_ complexity issues in vlsi , foundations of computing series_. mit press , cambridge , ma
j. milnor . on the betti numbers of real varieties .
ams_. 15:275280 .
j. pach , m. sharir . on the number of incidences between points and curves .
comput_. 7(1):121127 .
j. pach , m. sharir .
repeated angles in the plane and related problems .
_ j. combin .
a _ 59:1222 . 1992 .
j. solymosi , t. tao .
an incidence theorem in higher dimensions . arxiv:1103.2926v2 .
j. solymosi , g. tardos , on the number of @xmath15rich transformations .
_ proceedings of the 23th annual symposium on computational geometry _ , ( socg 2007 ) , acm , new york : 227231 . 2007 . j. solymosi , c. tth .
on distinct distances in homogeneous sets in the euclidean space . _ discrete comput .
_ 35:537549 . 2005
l. szkely .
crossing numbers and hard erds problems in discrete geometry .
probab comput_. 6(3):353358 .
e. szemerdi , w. trotter .
extremal problems in discrete geometry .
_ combinatorica_. 3(3):381392
r. thom , sur lhomologie des varits algbriques relles . _
differential and combinatorial topology _
, ( symposium in honor of marston morse ) , ed . s.s . cairns .
princeton univ . press .
255265 . 1965 . c. tth .
the szemeredi - trotter theorem in the complex plane .
arxiv : math/0305283v3 .
h. whitney
. elementary structure of real algebraic varieties . _
annals of math_. 66:545556 . 1957 .
j. zahl .
an improved bound on the number of point - surface incidences in three dimensions . | we show that under suitable non - degeneracy conditions , @xmath0 points and @xmath1 2dimensional algebraic surfaces in @xmath2 satisfying certain `` pseudoflat '' requirements can have at most @xmath3 incidences , provided that @xmath4 for any @xmath5 ( where the implicit constant in the above bound depends on @xmath6 ) , or @xmath7 . as a special case
, we obtain the szemerdi - trotter theorem for @xmath8planes in @xmath9 again provided @xmath4 or @xmath7 . as a further special case
we recover the szemerdi - trotter theorem for complex lines in @xmath10 with no restrictions on @xmath0 and @xmath1 ( this theorem was originally proved by tth using a different method ) . as a second special case
, we obtain the szemerdi - trotter theorem for complex unit circles in @xmath10 , which has applications to the complex unit distance problem .
we obtain our results by combining the discrete polynomial ham sandwich theorem with the crossing number inequality . |
the wealth of new data from the @xmath2 factories belle and babar has caused a great amount of phenomenological interest in @xmath2 decays in recent years .
already the much anticipated measurement of @xmath9 @xcite has been achieved , and many new results in the field of rare @xmath2 decays ( e.g. @xmath10 @xcite and @xmath11 @xcite ) are eagerly awaited .
less attention has been devoted to charmed ( @xmath12 ) meson decays , although the @xmath2 factories and forthcoming cleo c promise the largest sample of charmed mesons to date . @xmath12
mesons may be produced at the @xmath2 factories by two mechanisms : i ) continuum @xmath13 production and ii ) decay of @xmath2 mesons . while the majority of @xmath2 decays
involve some charmed particles , it is difficult to extract charm data from these decays due to the high multiplicity of particles in the final states . with the high luminosity of the @xmath2 factories ,
however , a lot of @xmath14 pairs that subsequently hadronize to @xmath12 mesons are produced directly in the collision of the primary @xmath15 beams .
both belle and babar will each have about @xmath16 @xmath13 continuum events in the anticipated data samples of @xmath17 fb@xmath18 , thus providing a rich testing ground for charm decays . at cleo - c ,
prospects are also very promising with the threshold production of @xmath12 mesons offering distinct advantages over the @xmath13 continuum production at the @xmath2 factories , which compensates for the lower luminosity of cleo - c @xcite .
these experiments will have substantially increased sensitivity to the purely leptonic decays of the charged @xmath19 mesons , @xmath20 and @xmath21 . in the sm
, such decays occur via @xmath22 annihilation in the @xmath23-channel and provide an opportunity to measure the decay constants ( @xmath24 ) for @xmath25 and @xmath19 . of these six leptonic decays , only @xmath26 and @xmath27 have been observed , from which @xmath28 is measured with an error @xmath29 .
the @xmath19 decays are cabbibo suppressed compared to @xmath25 , and none have been observed except for 1 event for @xmath30 .
the @xmath2 factories and cleo - c will offer improved measurements of @xmath26 and @xmath27 , which in turn will significantly reduce the error in the current measurements of @xmath28 .
observation of @xmath30 ( or the less accessible @xmath31 ) will provide the first serious measurement of @xmath32 .
in this paper we advocate searching for physics beyond the sm through these decays .
of special interest are @xmath1 and @xmath0 for which no experimental limits exist , but could readily be searched for at the above experiments . in the sm
these are severely helicity suppressed by @xmath33 and have brs of order @xmath34 and @xmath35 , respectively .
hence such decays have been largely overlooked since ( in the context of the sm ) they can not offer a measurement of the decay constant with present or upcoming data samples . however , the smallness of their brs enables these decays to play a new role of probing models beyond the sm .
we show that slepton contributions in the minimal supersymmetric standard model ( mssm ) with explicit @xmath3 parity violation can enhance these brs to @xmath4 , a result which strongly motivates a search in these channels .
although the effect of new physics in purely leptonic decays is sometimes tainted by the uncertainty in the decay constant , the tiny sm branching ratios for @xmath1 and @xmath0 assure that mere observation of these decays at the aforementioned machines would be an unambiguous signal of physics beyond the sm .
our work is organised as follows . in section 2
we introduce the @xmath19 meson annihilation decays . in section 3
we show how these decays can be enhanced in @xmath3 parity violating susy models .
section 4 presents our numerical results and section 5 contains our conclusions .
to date the primary interest in measuring the purely leptonic decays @xmath36 has been to obtain information on the charged @xmath12 meson decay constants @xcite . in the sm
these decays proceed via annihilation to a @xmath22 in the @xmath37channel ( see fig.1 ) . due to helicity suppression ,
the rate is proportional to @xmath38 , and the phase space suppression is particularly severe for @xmath39 .
the br is given by : @xmath40 exchange [ wexchange],width=264 ] @xmath41 & @xmath42 & @xmath43 + @xmath44 & @xmath45 & @xmath46 @xcite + @xmath47 & @xmath48 & @xmath43 + @xmath49 & @xmath50 & @xmath43 + @xmath51 & @xmath52 & @xmath53 @xcite + @xmath54 & @xmath55 & @xmath56 @xcite + where @xmath57 or @xmath23 .
the sm predictions for the brs and the current experimental status of the various searches are shown in table 1 .
channels from the ones given e.g. in the babar book .
this is because the rate depends very strongly on the mass of the @xmath12 , for which we are using newer values .
we also use a different value for the decay constant of the @xmath58 , see section 4 for details . ]
one can see that the decays involving @xmath59 have tiny brs , while those involving @xmath60 and @xmath39 have brs in the range @xmath61 .
of the six possible decays , only two have been measured with any sort of accuracy , yielding a world average of @xmath62 mev @xcite .
additionally there is a very imprecise measurement of @xmath30 based on 1 observed event , giving @xmath63 mev @xcite . of the three decays which have not been searched for
, @xmath1 and @xmath0 have particularly clean signatures . with the expected large samples of @xmath19 and @xmath25 mesons at belle , babar and cleo - c @xcite
, these experiments should be sensitive to br@xmath64 .
cleo - c aims to accumulate 30 million @xmath19 events ( 6 million fully tagged ) and 1.5 million @xmath25 events ( 0.3 million fully tagged ) by the end of 2004 .
belle and babar expect @xmath65 @xmath13 continuum events by the end of 2005 .
mere observation of these decays would be an patent signal of physics beyond the sm . in the next section
we show that susy particles in @xmath3 parity violating extensions of the mssm can enhance these decays to experimental observability .
thus in addition to offering measurements of the charged @xmath12 meson decay constants , the purely leptonic decays of @xmath66 mesons assume a new role of probing physics beyond the sm . in @xcite
cleo - c is considering increasing the selection efficiency of the @xmath67channel by waiving the @xmath67tag requirement , stating that no @xmath68contamination is to be expected due to the small sm rate of the respective channel .
we strongly encourage also performing an analysis with a muon identification tag , because the @xmath68channel can substantially contribute to the total leptonic annihilation decay rate as will be shown later in this paper .
the related decays @xmath69 , are known to have larger brs than @xmath70 in the sm @xcite .
this is because the presence of a photon in the final state removes the helicity suppression .
the analysis of @xcite finds br@xmath71 and br@xmath72 ) the `` effective '' sm prediction for br@xmath73 ) should include the contribution from br@xmath74 ) with a soft photon ( i.e. one which can not be detected experimentally ) , whose infra - red singularity cancels with the radiative corrections to br(@xmath70 ) : @xmath75 however , the soft photon contribution is only a small fraction of the total rate for br(@xmath69 ) , and so br@xmath76 ) would still be below the expected experimental sensitivity of @xmath77 .
hence we suggest that observation of br@xmath78 could only be attributed to physics beyond the sm .
we note that inclusive measurements of the br(@xmath79 ) have been performed , to which any enhanced @xmath80 would have contributed .
however , the error in the measurements of @xmath81 @xcite and @xmath82 @xcite still allow for contributions from @xmath80 of the order of a percent or more .
the main motivation for @xmath3 parity violating susy @xcite is to account for the observed neutrino oscillations without increasing the particle content of the mssm @xcite .
the superpotential is given by : @xmath84 bilinear terms @xmath85 are also possible , but have negligible impact on the annihilation decays we consider .
since the @xmath86 term can mediate proton decay , it is customary to assume that the @xmath87 couplings vanish due to some discrete symmetry ( e.g. baryon parity ) .
the simplest approach to @xmath3 parity violating phenomenology is to assume that a single @xmath3 parity violating coupling in the _ weak basis _
@xmath88 ) is dominant with all others negligibly small .
it was shown that such an approach leads to several non - zero @xmath3 parity violating couplings in the _ mass basis _
@xmath89 due to quark mixing @xcite : @xmath90 here we have assumed that all quark mixing lies in the up type sector , so that the mixing matrix is the usual kobayashi maskawa matrix @xmath91 . this simplification avoids the appearance of the right handed quark mixing matrix and gives the most conservative limits on the @xmath3 parity violating couplings , which would otherwise be constrained more severely from the decay @xmath92 @xcite .
a realistic @xmath3 parity violating model would have many non
zero couplings _ in the weak basis _ and
so in general would have a very rich phenomenology provided the couplings are not too small .
it has been emphasised before that the purely leptonic decays are very sensitive at tree level to @xmath3 parity violating trilinear interactions , and thus these decays constitute excellent probes of the model e.g. the decays @xmath93 may be enhanced up to current experimental sensitivity @xcite .
the relevant feynman diagrams for the decays @xmath94 are depicted in fig .
[ sparticleexchange ] and consist of @xmath23 and @xmath95channel exchange of sparticles .
decay in @xmath3 parity violating models : sparticle exchange [ sparticleexchange],width=340 ] these additional channels modify the sm rate ( [ smrate ] ) by @xmath96 where @xmath97 stems from @xmath3 parity conserving susy charged higgs exchange @xcite which we will not consider further since it is also proportional to the lepton mass and is relatively unimportant for the lighter leptons on which we focus .
the @xmath3 parity violating susy contributions are given by : @xmath98 where @xmath99 for @xmath100 , respectively .
these formulae were derived in @xcite for leptonic decays of @xmath2 mesons .
the helicity suppressed contribution from the @xmath101 term can be mediated by just one @xmath102 coupling ( if @xmath103 and @xmath104 ) , or by two different couplings ( if @xmath105 and/or @xmath106 ) . the dominant @xmath107 term ( which is not helicity suppressed ) requires one non
zero @xmath108 _ and _ one non zero @xmath102 . in the next section
we will vary the @xmath3 parity violating couplings inside their allowed ranges to determine the obtainable brs .
the contribution of the @xmath95channel diagrams has been considered in @xcite for @xmath26 and @xmath27 . here
only single coupling limits were considered and weak limits are derived for @xmath109 and @xmath110 .
analogous @xmath95channel exchange diagrams also occur for the neutral @xmath111 meson decays @xmath112 @xcite,@xcite .
strong upper limits ( @xmath113 ) on br@xmath114 ) have been obtained from the tevatron .
these decays have no @xmath23channel contributions of the type @xmath115 due to the absence of the coupling @xmath116 in the lagrangian .
we wish to focus on the decays @xmath117 , in particular the helicity unsuppressed @xmath23channel contributions mediated by combinations of @xmath115 .
although these decays might be problematic at the tevatron due to the missing energy of @xmath118 , they can be readily searched for at the @xmath15 machines belle , babar and cleo - c .
in our analysis we make use of the latest limits on the @xmath3 parity violating couplings @xmath108 and @xmath102 .
single coupling bounds are listed in @xcite .
further input parameters are @xmath119 for the @xmath19 meson and @xmath120 for the @xmath121 .
the parameters for rare @xmath122 decays are taken from @xcite . many of the bounds on @xmath3 parity violating couplings relevant for our analysis are of the same order of magnitude ; the products of these couplings are @xmath123 and can mediate @xmath70 with brs up to @xmath124 . however , most combinations of @xmath115 which mediate @xmath70 would strongly contribute to the kaon decays , @xmath125 , via @xmath126 exchange in the @xmath23channel @xcite .
the limits on such combinations is @xmath127 , which is @xmath128 better than the product of the single coupling limits , and at first sight would seem to rule out the possibility of a sizably enhanced @xmath70 mediated by @xmath115 combinations .
however , large brs for @xmath70 can occur if the neutrino is @xmath129 .
this is because the corresponding decay in the kaon sector would involve a @xmath122 lepton in the final state , which is kinematically impossible .
the only possibility for a lepton flavour violating kaon decay mediated by these combinations of couplings would be @xmath130 . here
the additional suppression factors ( e.g. off shell propagators , additional vertices ) easily weaken the limits on the relevant @xmath115 couplings to @xmath123 where the single coupling bounds become more restrictive than the bounds from kaon decays .
in addition the final state @xmath131 has not been searched for .
therefore the most promising combinations which enhance @xmath132 and @xmath133 are @xmath134 and @xmath135 respectively . for these combinations
the use of the single coupling bounds is justified .
the single coupling bounds @xmath136 and @xmath137 ( for sparticle mass @xmath138 gev ) can induce br@xmath139 .
although as stated above , this combination of couplings is safe from rare kaon decay bounds , the same combination can induce the lepton flavour violating @xmath122 decay @xmath140 @xcite . ) and br(@xmath140 ) on the product of @xmath141 couplings @xmath142 .
[ ddcorrelation],width=566 ] the dependence of br(@xmath143 ) and br(@xmath140 ) on the product of the @xmath144 couplings @xmath142 is shown in fig.[ddcorrelation ] .
the @xmath145 and @xmath146axes give the respective branching ratios and @xmath142 is varied along the diagonal line .
the plot is logarithmic along both axes .
the experimental bound on br(@xmath147 ) prior to summer 2002 @xcite ( rightmost vertical dashed line ) is less restrictive for @xmath142 , and therefore for br(@xmath143 ) , than the product of the individual coupling bounds .
the br(@xmath143 ) attainable with the individual coupling bounds is indicated by the upper horizontal dashed line .
however , a much improved experimental bound on br(@xmath140 ) has recently been published @xcite , and this limit is indicated by the left vertical dashed line . ) were released and therefore derives a rather weak bound @xmath148 , which is less restrictive than the single coupling limit .
this limit improves to @xmath149 , an order of magnitude better than the single coupling limit , with the new data from @xcite . ]
this new limit restricts the enhancement of br(@xmath143 ) from @xmath144 couplings quite substantially ; only brs of @xmath77 remain attainable , while the older bound on lepton flavour violating @xmath122decays allowed brs of @xmath150 .
cleo - c expects 30 million @xmath19 events ( 6 million tagged ) , so even brs of @xmath77 or smaller could be observed .
alternatively , lack of observation would further improve the limit on @xmath142 . ) and br(@xmath151 ) on the product of @xmath141 couplings @xmath152 .
[ dscorrelation],width=566 ] the situation is much more favourable for @xmath153 , which is correlated with the less well measured @xmath154 .
the single coupling bounds @xmath136 and @xmath155 give br@xmath156 . the plot analogous to fig.[ddcorrelation ] for the dependence of br(@xmath153 ) and br(@xmath154 ) on the product of @xmath144 couplings @xmath152 is shown in fig.[dscorrelation ] .
the current experimental bound on br(@xmath154 ) @xcite ( vertical dashed line ) restricts @xmath152 ( and therefore br(@xmath153 ) ) slightly more than the product of the single coupling limits , indicated in the top right hand corner of the graph .
the lower horizontal dashed line indicates a hypothetical limit on br(@xmath153 ) of @xmath157 which seems realistic in light of an expected number of 1.5 million @xmath58 events ( 0.3 million fully tagged ) after one year of running of cleo - c .
this limit would restrict the product of couplings @xmath152 about a factor of 10 better than present experiments . to compete with this accuracy , the experimental limit on br(@xmath154 )
would have to improve by about two orders of magnitude which does not seem attainable in the current runs of the @xmath2 factories .
we therefore believe that even if searches for @xmath158 do not detect an enhancement in these channels , they would still be useful for setting new limits on products of @xmath144 couplings .
the poorly measured decay @xmath30 may also be enhanced by the combination @xmath159 .
the current error allows for a sizeable enhancement over the sm prediction of the order 36 , depending on the value of the decay constant .
the sm prediction currently lies at the lower end of the experimentally allowed interval .
if subsequent measurements should tend towards the upper end of the current interval , non zero @xmath3 parity violating couplings would be a possible explanation for this deviation .
conform measurements on the other hand would allow for better limits on @xmath159 .
cleo - c aims to measure the br(@xmath30 ) to a precision of a few @xmath160 . for the decays
@xmath161 we do not find sizably enhanced brs .
the decays @xmath162 and @xmath163 can also be mediated by products of two @xmath102 couplings ( right diagram in fig.[sparticleexchange ] ) , but because of the helicity suppression , only the @xmath122channel can receive a sizeable contribution . even in these cases , the @xmath144 contributions can only become as large as the uncertainties of the sm predictions .
therefore neither a large enhancement , nor improvements of the limits are possible , except for single coupling limits on @xmath164 and @xmath165 as shown in @xcite .
in the context of the mssm we have studied the effects of @xmath3 parity violating couplings ( @xmath166 ) on the purely leptonic decays @xmath167 .
we showed that slepton mediated contributions proportional to combinations of the type @xmath115 can strongly enhance the previously unmeasured decays @xmath168 to the sensitivity of current @xmath2 factories and forthcoming cleo - c .
maximum values for br(@xmath1 ) and br(@xmath0 ) of @xmath4 and @xmath169 respectively were found .
mere observation of these decays would be an unequivocal signal of physics beyond the sm .
in simple @xmath3 parity violating models with a single dominant @xmath115 combination , there would be a correlation with the decays @xmath170 and @xmath154 , which would be similarly enhanced to the sensitivity of current and planned experiments .
such correlated signals would provide strong evidence for @xmath3 parity violating interactions .
k. abe _ et al . _ [ belle collaboration ] , phys .
lett . * 87 * , 091802 ( 2001 ) .
b. aubert _ et al . _ [ babar collaboration ] , phys .
lett . *
87 * , 091801 ( 2001 ) .
s. bertolini , f. borzumati , a. masiero and g. ridolfi , nucl .
b * 353 * , 591 ( 1991 ) ; t. hurth , arxiv : hep - ph/0106050 and refs therein .
a. g. akeroyd , y. y. keum and s. recksiegel , phys .
b * 507 * , 252 ( 2001 ) ; a. g. akeroyd and s. recksiegel , phys .
b * 525 * , 81 ( 2002 ) ; a. arhrib , c. k. chua and w. s. hou , eur .
j. c * 21 * , 567 ( 2001 ) ; a. ali and e. lunghi , hep - ph/0206242 .
i. shipsey , hep - ex/0207091 + ( see also http://www.lns.cornell.edu/public/cleo/spoke/cleoc/ ) .
s. sldner - rembold , hep - ex/0109023 .
j. z. bai _ et al .
_ [ bes collaboration ] , phys .
b * 429 * , 188 ( 1998 ) .
a. heister _ et al .
_ [ aleph collaboration ] , phys . lett .
b * 528 * , 1 ( 2002 ) .
e. m. aitala _ et al . _
[ e791 collaboration ] , phys .
b * 462 * , 401 ( 1999 ) .
m. chadha _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 58 * , 032002 ( 1998 ) .
g. abbiendi _ et al . _
[ opal collaboration ] , phys .
b * 516 * , 236 ( 2001 ) .
g. burdman , t. goldman and d. wyler , phys .
d * 51 * , 111 ( 1995 ) ; d. atwood , g. eilam and a. soni , mod .
a * 11 * , 1061 ( 1996 ) ; c. q. geng , c. c. lih and w. m. zhang , mod .
lett . a * 15 * , 2087 ( 2000 ) ; g. l. wang , c. h. chang and t. f. feng , hep - ph/0102251 . g. p. korchemsky , d. pirjol and t. m. yan , phys . rev .
d * 61 * , 114510 ( 2000 ) .
j. z. bai _ et al . _
[ bes collaboration ] , phys .
d * 56 * , 3779 ( 1997 ) .
g. abbiendi _ et al .
_ [ opal collaboration ] , eur . phys .
j. c * 8 * , 573 ( 1999 ) .
h. k. dreiner , in `` kane , g.l .
( ed . ) : perspectives on supersymmetry '' 462 - 479 , + ( hep - ph/9707435 ) ; g. bhattacharyya , nucl .
suppl . * 52a * ( 1997 ) 83 ; o. c. kong , hep - ph/0205205 .
b. c. allanach , a. dedes and h. k. dreiner , phys .
d * 60 * , 075014 ( 1999 ) . c. s. aulakh and r. n. mohapatra , phys .
b * 119 * , 136 ( 1982 ) ; l. j. hall and m. suzuki , nucl .
b * 231 * , 419 ( 1984 ) ; r. hempfling , nucl .
b * 478 * , 3 ( 1996 ) ; h. p. nilles and n. polonsky , nucl .
b * 484 * , 33 ( 1997 ) ; e. j. chun , s. k. kang , c. w. kim and u. w. lee , nucl
. phys .
b * 544 * , 89 ( 1999 ) ; o. c. kong , mod .
a * 14 * , 903 ( 1999 ) ; s. k. kang and o. c. kong , hep - ph/0206009 . f. borzumati and j. s. lee , hep - ph/0207184 . k. agashe and m. graesser , phys . rev .
d * 54 * , 4445 ( 1996 ) .
s. w. baek and y. g. kim , phys .
d * 60 * , 077701 ( 1999 ) .
h. k. dreiner , g. polesello and m. thormeier , phys .
d * 65 * , 115006 ( 2002 ) .
a. g. akeroyd and s. recksiegel , phys .
b * 541 * , 121 ( 2002 ) .
a. g. akeroyd and s. recksiegel , hep - ph/0209252 .
w. s. hou , phys .
d * 48 * , 2342 ( 1993 ) .
f. ledoit and g. sajot , + http://qcd.th.u-psud.fr/gdr_susy/gdr_susy_public/entete_note_publique g. burdman , e. golowich , j. l. hewett and s. pakvasa , phys .
d * 52 * , 6383 ( 1995 ) .
g. burdman , e. golowich , j. hewett and s. pakvasa , phys .
d * 66 * , 014009 ( 2002 ) .
d. choudhury and p. roy , phys .
b * 378 * , 153 ( 1996 ) .
j. p. saha and a. kundu , phys .
d * 66 * , 054021 ( 2002 ) .
k. g. hayes _
et al . _ ,
d * 25 * , 2869 ( 1982 ) .
s. chen _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 66 * , 071101 ( 2002 ) .
g. bonvicini _ et al . _
[ cleo collaboration ] , phys .
* 79 * , 1221 ( 1997 ) . | the purely leptonic decays @xmath0 and @xmath1 , for which no experimental limits exist , are highly suppressed in the standard model
. mere observation of these decays at the @xmath2 factories belle / babar or forthcoming cleo - c would be a clear signal of physics beyond the sm . we show that @xmath3 parity violating slepton contributions can give rise to spectacular enhancements of the decay rates , resulting in branching ratios as large as @xmath4 , which strongly motivates a search in these channels .
= 10000 * hep - ph/0210376 + kias p02071 + kek
th852 + october 2002 * 1.cm 0.8 cm * .3 cm a.g .
akeroyd@xmath5 , s. recksiegel@xmath6 * * _ @xmath7 korea institute for advanced study , 207 - 43 cheongryangri - dong , + dongdaemun - gu , seoul 130 - 012 , republic of korea _ * * _ @xmath8 theory group , kek , tsukuba , ibaraki 305 - 0801 , japan _ * 2.0 cm 1.0 cm 1.0 cm * keywords : rare d decay * |
Subsets and Splits