article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
electromagnetic forbidden , both magnetic dipole ( m1 ) and electric quadrupole ( e2 ) , transitions of mo vi are important for temperature and density estimations of tokamak plasmas @xcite , especially in the collision - radiative model @xcite .
long lifetimes of metastable states are dominated by these forbidden transitions and these states are generally difficult to observe in the laboratory plasmas due to strong collisions .
however , these forbidden transitions of mo vi have been observed in laboratory in electron spin resonance experiment @xcite and therefore , they must be one of the sources of density estimations in astrophysical plasmas where collisions are very low due to high dilute interstellar medium @xcite
. accurate estimation of abundances of molybdenum in the atmosphere of the evolved stars are important to understand the stellar nucleosynthesis @xcite .
hexavelent molybdenum , isoelectronic to rubidium with 4p@xmath24d as ground state configuration , is generated by electron impact in the atomic collision process .
the electron - impact ionization of multiply charged mo ions , relevant to astrophysics and laboratory plasma research , have also been investigated @xcite .
recently , fisker et al .
have given the possibility of the origin of the lightest isotope of molybdenum in proton rich type ii supernova @xcite .
the necessity of accurate estimation of allowed dipole transition strengths to find out their mixing these effects in the dipole forbidden transitions in mo vi is explicitly discussed by t. yamamoto @xcite .
again , the transition strength between the fine structure states of 4d lavel can reflect the electronic structure of mo vi in crystal @xcite .
a few calculations have been carried out to study the on electric dipole ( e1 ) transitions in mo vi over the last few decades using the mean - field theory @xcite .
more recently , j. reader @xcite has estimated the e1 transition probabilities among low - lying states by estimating transition strengths in the semiempirical approach with the experimental excitation energies .
for this single reference system , mo vi , we have performed relativistic coupled - cluster ( rcc ) calculation with single ( s ) , double ( d ) and partial triple ( t ) excitations in the framework of fock space multi - reference ( fsmr ) .
both the excitation energies and transition probabilities are determined using this rcc method using which lifetimes of many low - lying states are estimated .
the oscillator strength for e1 transition from @xmath3 to @xmath4 is given as @xmath5[eq7 ] where @xmath6 is the excitation energy between the upper and lower states and @xmath7 is the degeneracy factor of the upper state with total angular momentum @xmath8 .
the single particle reduced matrix elements for the e1 , e2 and m1 transition operators are given in @xcite .
the emission transition probabilities ( in sec@xmath9 ) for the e1 , e2 and m1 channels from states _ f _ to _ i _ can be expressed as @xmath10 @xmath11 @xmath12 where @xmath13 is the transition strength for the coressponding operator @xmath14 ( in a.u . ) and @xmath15 ( in ) is the corresponding transition wavelength .
the lifetime of a particular excited state @xmath16 can be computed by the reciprocal of the total transition probability , @xmath17 ( in sec@xmath9 ) , arising from all possible states @xmath18 due to spontaneous electromagnetic transitions , i.e. @xmath19 the fsmrcc method is one of the most powerful highly correlated many - body approaches due to its all order structure to account the correlation effects @xcite .
the fsmrcc , which is mainly meant for multi - reference systems , is used here for the one valence electron and has been described in details elsewhere @xcite . here
we present the method briefly .
we first consider the dirac - coulomb hamiltonian for a closed - shell @xmath20 electron system which is given by @xmath21+\sum_{i < j}^{n}\frac{1}{r_{ij}}\ ] ] with all the standard notations often used .
the theory for a single valence system is based on the concept of common vacuum for both the closed shell @xmath20- and open shell @xmath22-electron systems , which allows us to formulate a direct method to determine energy differences ( electron attachment energy or negative of the ionization potential ) .
also , the holes and particles are defined with respect to the common vacuum for both the electron systems .
model space of an ( n , m ) fock - space contains determinants of @xmath23 holes and @xmath24 particles distributed within a set of orbitals known as _ active _ orbitals .
for example , in the present article , we are dealing with ( 0,1 ) fock - space which is a complete model space ( cms ) by construction and is given by @xmath25 s are the expansion coefficients of @xmath26 , and @xmath27 s are the model space configurations made of df orbitals .
the dynamical electron correlation effects are introduced through the _ valence - universal _ wave - operator @xmath28 @xcite @xmath29 where @xmath30 at this juncture , it is convenient to single out the core - cluster amplitudes @xmath31 and call them @xmath32 .
the rest of the cluster amplitudes will henceforth be called @xmath33 .
since @xmath28 is in normal ordered , we can rewrite eq .
( [ eq2.5 ] ) as @xmath34 the `` valence - universal '' wave - operator @xmath28 in eq .
( [ eq2.7 ] ) is parameterized in such a way that the states generated by its action on the reference space satisfy the fock - space bloch equation @xmath35 where @xmath36 here , p is the projection operator of model space .
( [ eq2.8 ] ) is valid for all ( k , l ) starting from k = l=0 ( i.e. , the _ core _ problem ) to some desired _ parent _ model space , with k = m ,
l = n . in this present calculation , we truncate eq .
( [ eq2.6 ] ) at @xmath37 and @xmath38 . in this work ,
single ( @xmath39 ) and double ( @xmath40 ) excitations are considered for @xmath32 and @xmath33 clusters operator .
therefore , the total correlated wavefunction of the system with single valence orbital _ v _ , can be written as @xmath41 important triple excitations , correspond to the correlation to the valence orbitals , are included in the open shell fsmrcc - sd calculations by an approximation that is similar in spirit to fsmrcc - sd(t ) @xcite . the approximate valence
triple excitation amplitudes are given by @xmath42 where @xmath43 are the amplitudes corresponding to the simultaneous excitations from core orbitals @xmath44 and valence @xmath45 to virtual orbitals @xmath46 , and @xmath47 , respectively .
@xmath48 and @xmath49 are the connected composites involving @xmath50 and @xmath32 , and @xmath50 and @xmath51 , respectively , where @xmath50 is the two electron coulomb ( @xmath52)integral and @xmath53 s are the orbital energies . the transition matrix element due to any operator @xmath14 can be expressed as @xmath54
in the present calculation , the radial wavefunctions of df orbitals of closed shell mo vii are obtained using gaussian type orbitals ( gto ) basis with finite nuclear size as discussed in our earlier paper @xcite .
we have used universal basis set , where the exponent @xmath55 is related with two parameters @xmath56 and @xmath57 , same for all the symmetries , expressed as @xmath58 we have considered @xmath56 and @xmath57 as 0.00625 and 2.72 , respectively , after obtaining best fit of the bound orbital energies and evaluating the expectation values of different radial functions ( @xmath47 , @xmath59 , @xmath60 ) generated with gtos and grasp2 @xcite . in the df calculations ,
we have taken 22 , 20 , 17 , 15 , and 12 number of gtos for s , p , d , f , and g type symmetries , respectively , to generate the atomic orbitals . in fig .
[ fig1 ] , we have given the relative errors obtained for different orbitals in the calculations of these quantities using the above chosen parameters .
since these errors are very small , it shows that there is a good agreement between results .
we assume that both the bound and continuum orbitals generated using the above parameters will describe well both inside and outside of the nucleus .
therefore , we have considered all the orbitals obtained using gtos for the rest of the calculations .
the number of the df orbitals for different symmetries used in the present calculation is based on convergent criteria of core correlation energy of mo vii for which it satisfies numerical completeness .
the number of df orbitals considered for s , p , d , f , and g type symmetries in the rcc calculations are 12 , 11 , 10 , 9 and 8 , respectively ; and among them 9 , 8 , 7 , 5 , and 5 are bound orbitals , respectively , including all the core orbitals .
the @xmath32 amplitudes are first determined by solving the closed shell rcc equations for the closed - shell system ( mo vii ) , then @xmath33 amplitudes are solved from the open - shell equations for the single - valence states of mo vi . the relative energies and expectation values of @xmath47 , @xmath59 and @xmath60 of the df gto orbitals to the df grasp orbitals.,scaledwidth=80.0% ] table i summarizes the calculated excitation energies ( ee ) and fine structure splitting ( fs ) of low - lying excited states and their comparison with the recent experimental results @xcite .
the average deviation is around 0.5% for ee .
we have also presented the contribution from the partial triple excitations to the ee ( e@xmath61 ) , which is around 0.3% to the total ee .
we have examined the first order excitation energy corrections due to breit interaction using large scale relativistic ci calculations .
maximum contribution is coming for 5s state , which is around @xmath62 , whereas , contributions to 5p and 4f states are around @xmath63 and @xmath64 ; respectively . for 5d state , it is as small as 54 @xmath65 consistent with the result obtained by pan and beck @xcite .
since , the contributions due to breit interaction are relatively small , we do not consider them here selfconsistently to evaluate wavefunctions .
.excitation energies ( ee ) and fine structure splitting ( fs ) ( in cm@xmath9 ) of mo vi and their comparison with experimental results .
contribution from absolute value of partial triple excitation to the ee ( e@xmath61 ) are also presented .
[ cols= " < , > , > , > , > , > " , ] [ tab : results1 ] @xmath67 ref .
forbidden transition probabilities among the low - lying states of mo vi relevant for astro- and plasma physics are calculated using highly correlated relativistic coupled - cluster method for the first time in literature to the best of our knowledge .
the lifetime of the 4d @xmath68 state is found to be around 5 second , which will be useful in many physical processes .
contributions of different correlation terms are discussed and found strong effect from higher order core - polarization . in the near future
, present work will motivate experimentalists to verify our results due to its importance in many areas in physics .
we are grateful to prof b p das and dr rajat k chaudhuri , indian institute of astrophysics , bangalore for providing the cc code .
one of us ( narendra ) would like to recognize the support of council of scientific and industrial research ( csir ) , india .
feldman u 1981 _ phys .
_ * 24 * 681 .
herter t , briotta d a , gull g e , shure m a and houck j r 1982 _ astrophys .
j. _ * 262 * 164 .
fournier k b , pacella d , may m j , finkenthal m and goldstein w h 1997 _ nucl . fusion _ * 37 * 825 .
quinet p 2002 _ j. phys .
b _ * 35 * 19 .
baur j , kunzer m , dombrowski k f , kaufmann u , schneider j , baranov p g , and mokhov e n 1997 _ inst .
ser . _ * 155 * 933 .
charro e and martin i 2003 _ astrophysics j. _ * 585 * 1191 .
orlov m ya and shavrina a v 1988 _ astrophysics_*28 * 354 .
hathiramani d , hartenfeller u , aichele k , hofmann g , steidl m , stenke m , voelpel r and salzborn 1998 _ j. phys .
b _ * 31 * 2101 .
fisker j l , hoffman r d and pruet j 2009 _ astrophys .
j. _ * 690 * l135 .
yamamoto t 2008 _ x - ray spectrom . _ * 37 * 572 .
szotek z , temmerman w m , svane a , petit l , and winter h 2003 _ phys .
* 104411 . migdalek j and baylis w e 1979 _ j. quant .
transfer _ * 22 * 127
. zilitis v a 2007 _ optics and spectroscopy _ * 103 * 895 .
pan l and beck d r 2006 _ phys .
_ * 73 * 607 .
reader j 2010 _ j. phys .
b _ * 43 * 074024 . lindgren i and morrison j 1985 _ atomic many - body theory _ * 3
* ed . g. e. lambropoulos and h. walther ( berlin : springer ) .
lindgren i and mukherjee d 1987 _ phys .
rep . _ * 151 * 93 .
haque a and mukherjee d 1984 _ j. chem .
phys . _ * 80 * 5058 .
pal s , rittby m , bartlett r j , sinha d and mukherjee d 1987 _ chem .
lett . _ * 137 * 273 ; 1988 _ j. chem .
phys . _ * 88 * 4357 .
raghavachari k , trucks g w , pople j a and head - gordon m 1989 _ chem .
lett . _ * 157 * 479 ; urban m , noga j , cole s j and bartlett r j 1985 _ chem .
lett . _ * 83 * 4041 .
johnson w r , plante d r and sapirstein j 1995 _ adv . at .
_ * 35 * 255 .
dixit g , nataraj h s , sahoo b k , chaudhuri r k and majumder s , 2008 _ j. phys .
b _ * 41 * 025001 ; 2008 _ phys .
a _ * 77 * 012718 .
parpia f a , fischer c f and grant i p 2006 _ comp .
comm . _ * 175 * 745 .
cowan r d 1981 _ the theory of atomic structure and spectra _ ( berkeley , ca : university of california press ) . | relativistic coupled - cluster ( rcc ) calculations have been performed to estimate the electromagnetic forbidden transition probabilities , oscillator strengths and lifetimes of many low - lying states of five times ionized molybdenum ( mo vi ) .
contributions from the breit interaction up to the first order of perturbation have been examined .
our results are in good agreement with the available other reported theoretical and experimental results .
a long lifetime about @xmath0 of the first excited state , 4d @xmath1 , has been predicted which can be a very useful criteria in the doping process of thin films .
correlations trends from various rcc terms to the transition amplitude calculations are discussed . |
the hierarchical nature of large - scale structure formation is spectacularly revealed by observations of approximately equal mass mergers between pairs of galaxy clusters .
these major mergers can subject galaxies to an environment that can drive abnormal rates of galaxy evolution , particularly at times close to pericentric passage @xcite .
for example , it has been suggested that the core passage phase of a merger may be important in both triggering and truncating starbursts @xcite , although it remains unclear which merger - specific mechanisms are responsible .
simulations suggest that the high pressure environment a galaxy encounters during the core passage phase of a merger may result in the collapse of giant molecular clouds ( gmcs ) within a galaxy , leading to an initial burst of star formation @xcite while the subsequent stripping of the interstellar medium terminates gmc formation , halting further star formation @xcite .
evidence for these processes should reveal itself in the vicinity of observed shock fronts which mark regions of severe merger activity . in this context , @xcite find that the shock front in the `` bullet '' cluster ( 1es0657 - 558 * ? ? ?
* ) has not had an appreciable effect on the star formation in the galaxies in its vicinity .
thus , while it appears that the intense core passage phase of a major merger may play a significant role in shaping the star - forming properties of the galaxies , more detailed observations are required to understand the dominant processes . in this vein , the merging cluster abell 2744 ( z=0.3064 , hereafter a2744 ; * ?
* hereafter o11 ) is an excellent candidate for testing the effects of major mergers on cluster galaxies .
recently , combined _ chandra _ x - ray and anglo australian telescope aaomega optical observations to constrain the dynamics of the merger in a2744 .
the new data allowed a refinement of the previous merger scenarios @xcite and indicate that a high velocity ( @xmath0@xmath1 ) , near head on , major merger along a roughly north - south axis at @xmath2 to our line of sight is shortly past core passage , along with an infalling group to the northwest ( see also * ? ? ? * ) .
the _ chandra _
data reveal evidence for a `` bullet - like '' shock front driven by the remnant core of the less massive subcluster .
its higher velocity dispersion and strong lensing features around its brightest member ( figure 1 ) indicate that the northern - most subcluster is the remnant of the more massive subcluster .
the properties of the major merger in a2744 make it a prime target for understanding how this extreme environment affects the resident galaxies . in this letter
we present a suggestive spatial coincidence between a merger - affected region revealed by the _
chandra _
analysis of and several peculiar galaxies discovered in archival _
hubble space telescope _
( hst ) imagery .
these `` jellyfish '' galaxies exhibit one - sided trails of extremely blue knots and filaments reminiscent of those first noticed by @xcite and ( * ? ? ? * see also @xcite ) .
these knots and filaments are interpreted as the manifestation of hot , young stars formed _ in - situ _ within gas which has been stripped from the parent galaxy ( * ? ? ?
* hereafter s10 ) .
we suggest that the major merger environment of a2744 is significantly influencing these ` galaxies in turmoil ' .
hst / acs observations of a2744 were taken during 2009 october using the f435w , f606w and f814w filters .
the observations consist of north and south pointings each having exposure times of 6624s for the f606w and f814w filters and 8081s for the f435w filter .
the data were retrieved from the archive and the pointings were combined using the multidrizzle task @xcite .
the rgb color image is shown in figure [ hst_plus_chan ] , where the f814w image is used for the red channel , f606w green and f435w blue . overlaid
are contours from our background - subtracted , exposure - corrected _ chandra _ x - ray image
. plotted as a magenta arc is the position of the shock edge described in .
we also make use of our aaomega spectra .
close inspection of the hst images revealed four galaxies ( marked by red boxes in figure [ hst_plus_chan ] ) with distinct trails of extremely blue knots and filaments , most conspicuous in the bluest f435w band and having magnitudes @xmath3 .
this band corresponds to the sloan @xmath4-band in the cluster rest frame , which is known to be sensitive to light from young , hot ob stars , hence active regions of star formation @xcite .
the top panels of figure [ rgbimages ] show images of these four galaxies , revealing their `` jellyfish '' morphology , which is due to trails and filaments of bright star - forming regions .
this interpretation is consistent with star - forming signatures in the optical spectra discussed below for three of the four galaxies .
all 3 galaxies are spectroscopically confirmed cluster members .
galaxy f0083 is a luminous ( @xmath5 ) , nearly face - on late - type spiral with numerous blue star - forming regions in its disk and an unresolved bright nucleus .
the disk structure is irregular with an asymmetry on the eastern side connecting to a faint tidal feature which connects to a nearby faint galaxy ( f606w mag @xmath6 ) , indicating an interaction has taken place .
there is a bright blue rim @xmath7kpc north of the galaxy center .
a trail of blue knots and filaments extends @xmath8kpc from the galaxy center to the southwest these knots and filaments
are not associated with the disk .
the spectrum ( figure [ rgbimages ] ) reveals balmer lines with narrow and broad components which are well fitted by two gaussians having rest - frame velocity dispersions of @xmath9@xmath1 ( @xmath10@xmath1 ) and @xmath11@xmath1 ( @xmath12@xmath1 ) , respectively , for @xmath13 ( @xmath14 ) .
the oiii doublet is also well fitted by broad and narrow components with dispersions @xmath15@xmath1 and @xmath16@xmath1 , respectively .
the combination of broad and narrow lines indicate that f0083 hosts a seyfert 1 nucleus . however , the ratios of the equivalent widths ( ew ) of the narrow components of log([nii]/@xmath13)=-0.51 and log([oiii]/@xmath14)=0.67 lie close to the agn / star - forming boundary , indicating that a fraction of the emission may be due to star formation @xcite .
a bright x - ray point source is associated with this galaxy ( figure [ hst_plus_chan ] ) .
f0237 ( originally cn104 ; * ? ? ?
* ) is a major merger involving two @xmath17 late - type galaxies ( classified by * ? ? ? * as sbc and sab ) the centers of which are separated by @xmath18kpc .
the multi - color image ( figure [ rgbimages ] ) reveals a faint tidal feature extending @xmath19kpc to the southeast , and a trail of @xmath20 blue knots extending @xmath21kpc to the southwest .
unlike f0083 , there are no blue knots within the galaxy disks .
the aaomega spectrum confirms @xcite poststarburst classification there is strong balmer absorption ( ew @xmath22 ) with no measurable oii or @xmath13 emission .
the fiber aperture covers only a portion of f0237 , leaving the extent of its poststarburst emission undetermined .
however , the lack of blue star - forming regions within the system indicates a dearth of ongoing star formation .
the faint ( @xmath23 ) , blue galaxy f1228 is the western - most of the highlighted galaxies in figure [ hst_plus_chan ] .
it hosts the least spectacular trail of blue knots , the most distant of which is @xmath24kpc from f1228 s main component .
the morphology is similar to the `` tadpole '' galaxies seen in the hubble ultra - deep field @xcite with a head that contains a number of bright knots of star formation , a blue ridge on its western side , and a tail pointing to the se containing the blue knots .
the spectrum ( figure [ rgbimages ] ) reveals strong @xmath13 emission with ew @xmath25 . the emission line ew ratios of [ nii]/@xmath13=-0.64 and [ oiii]/@xmath14=-0.04 are consistent with a star - forming origin @xcite .
the jellyfish galaxy closest to the bullet - like subcluster ( right - most panel of figure [ rgbimages ] ) has a disk - like morphology , with no discernible bulge or spiral arms .
the trail of extremely blue knots points to the nw and extends to a radius of @xmath26kpc .
this galaxy is below the brightness limit used in and thus does not have a spectrum or , therefore , a redshift .
+ we estimate the ages of the knots and filaments by comparing their colors to those predicted by the solar metallicity models of @xcite .
the models used correspond to the stellar populations produced by two extremal star formation histories a single burst , and an exponentially decaying star formation with a decay timescale of 20gyrs . the colors of the knots and filaments and the color evolution tracks derived from the models are shown in figure [ fig : colours ] .
there are two caveats .
first , the contribution of emission lines is ignored , although this is exepected to be minimal within the broad bands of the filters ( e.g. , see figure 9 * ? ? ?
second , we do not account for reddening .
however , dust obscuration tends to make stellar populations appear older , thus our age estimates are upper limits . from figure
[ fig : colours ] it can be seen that in the majority of cases the knots / filaments are bluer than predicted by the 100myr populations , thus we take this to be the upper limit on the times since the onset of star formation .
we have presented hst observations which reveal four `` jellyfish '' galaxies in the merging cluster a2744 .
low redshift jellyfish analogues are observed in nearby clusters @xcite . however , at redshifts comparable to a2744 , only three have been observed , all in separate clusters @xcite .
indeed , @xcite found only two examples in a survey of 13 intermediate redshift clusters .
the rarity of jellyfish galaxies in intermediate redshift clusters emphasizes the significance of the observations presented here ; we observe four such systems within a2744 . while the statistics at intermediate redshifts are sparse , compiled a sample of 13 coma cluster galaxies which harbor ultraviolet asymmetries / tails .
the orientation of the tails , which generally point away from the cluster center , suggested that these galaxies formed from an infalling population experiencing the cluster environment for the first time .
however , there are two key distinctions to be made when comparing the low redshift jellyfish population with that of a2744
. first , the knots in a2744 are bright ( @xmath27 ) when compared to prominent examples in coma ( gmp4060/rb199 , @xmath28 s10 ; * ? ? ?
* ) and virgo ( ic3418 , @xmath29 * ? ? ?
* ; * ? ? ?
* ) while only eso 137 - 001 in the merging cluster a3627 has knots with comparable brightness @xcite .
thus , the bright knots seen in a2744 are rare in low redshift clusters .
second , the orientation of the tails in a2744 are less well ordered than those seen in coma .
this does not preclude the existence of a faint , coma - like infalling population of jellyfish in a2744 , which may be revealed by deeper observations , but indicates that the picture outlined by is unlikely in this case for the jellyfish observed in a2744 .
thus , there appears to be a relatively larger number of jellyfish in a2744 compared to other intermediate redshift clusters and these jellyfish appear to be a different , brighter version of their low redshift counterparts .
is the major merger in a2744 driving the formation of an excess of these bright jellyfish galaxies ?
there is a strong spatial correlation between the jellyfish galaxies and features associated with the high speed bullet - like subcluster : the proximity of ( i ) f0083 and f0237 to the portion of the bullet - driven shock front revealed as an edge in the _ chandra _ observations , and ( ii ) the central jellyfish to the x - ray peak associated with the remnant gas core of the bullet - like subcluster . while we can not know the exact 3d locations of the jellyfish galaxies with respect to the icm structures
, the small projected distances suggest that the jellyfish galaxies may have recently been overrun by the shock front and/or the bullet - like subcluster gas .
this indicates that a mechanism related to an interaction with these icm features may be responsible for either the stripping of the gas leading to the tails or the triggering of the star formation in the tails , or both .
this assertion is supported by the young ages of the stellar populations in the knots and filaments , which suggest that the star formation was triggered @xmath30 ago ( section [ colours ] ) . on these timescales , a galaxy with velocity @xmath31@xmath1 travels @xmath32kpc , so we would expect there to still be a strong spatial coincidence between the jellyfish and the putative icm features responsible for triggering the star formation .
furthermore , consideration of the peculiar velocity of the bullet - like subcluster and of the three jellyfish galaxies with measured redshifts ( @xmath33@xmath1 ; lower panels , figure [ rgbimages ] ) indicates that the jellyfish galaxies are not members of the bullet - like subcluster , and that if they have interacted with the shock or the bullet - like subcluster s icm , then the relative velocity of the interaction was high of the order of the merger velocity @xmath34@xmath1 .
similarly , @xcite suggested that the jellyfish - like galaxy c153 in a2125 may be a result of enhanced ram pressure stripping caused by a high velocity encounter with the icm due to a cluster merger , while find hints that some of their jellyfish galaxies are associated with merger - related enhancements in the icm density .
the proximity of the jellyfish to merger - related icm features in this high speed merger suggests that the merger is responsible for the increased fraction and more extreme nature of the jellyfish in a2744 . a high - speed
, head - on merger creates much greater ram pressure and a powerful shock which can significantly enhance galaxy - icm processes when compared with those felt by a galaxy falling into a relaxed cluster .
enhanced ram pressure may strip gas more efficiently leading to a higher incidence of gaseous tails , and therefore a higher likelihood of forming the observed jellyfish phenomena .
this is consistent with the simulations of @xcite , who find the mass lost from a galaxy due to ram pressure during a major merger is substantially increased , and also with @xcite who suggest the enhanced ram pressure stripping due to an interaction with the infalling m49 group explains the strong ram pressure stripping observed in ngc 4522 .
furthermore , the increased icm pressure due to the merger shock may promote star formation in the tails and may also drive higher star formation rates , hence higher surface brightness features , compared with jellyfish galaxies in lower pressure environments .
the shock has a mach number @xmath35 meaning the pressure jumps by a factor of @xmath24 with respect to the pressure of the surrounding unshocked icm , @xmath36 .
thus , the static pressure due to the shock is @xmath37 which is an order of magnitude higher than the threshold pressure required to trigger the collapse of gmcs leading to star formation @xcite .
furthermore , there is a rapid increase in the ram - pressure the galaxy feels as it is overrun by the shock due to the high relative velocity and the factor of 3 increase in icm density for a mach 3 shock .
if , like the bullet cluster @xcite , the shock front is collisionless , it would be more abrupt than a collision - dominated shock ( relaxation time @xmath38myr ) , so the interaction is likely too fast ( @xmath38myr ) for the shock s ram pressure to be responsible for stripping the gas leading to the observed tails .
this is supported by the southwesterly orientation of the trails of material seen in f0083 and f0237 . if ram pressure from the shock was responsible , the tails should point southeast toward the near part of the shock front .
this high pressure shock - triggering scenario is consistent with the simulations of @xcite and @xcite who found that the star formation rates in ram - pressure stripped gaseous tails is largely driven by the icm pressure .
however , simulations specific to the scenario outlined above are required for confirmation . on the question of whether the increased ram pressure which occurs in mergers enhances or suppresses the star formation in cluster members @xcite ,
the observations presented here show that both effects may be in play .
considering f0083 , a number of blue knots are seen across the galaxy indicating that there is disk - wide star formation .
there is evidence for an interaction with a smaller galaxy ( figure [ rgbimages ] ) , which may have driven gas towards the galaxy center , triggering the agn activity .
however , such a minor interaction is unlikely to trigger disk - wide star formation .
thus , we suggest that the high icm pressure has compressed the gmcs and triggered disk - wide star formation @xcite . the asymmetric distribution in the star formation particularly the arc - shaped star - forming region @xmath7kpc to the north of the galaxy center on the side opposing the tail of star - forming regions is consistent with that expected due to the compression from ram pressure as a galaxy moves edge - on through the icm @xcite .
conversely , the spectrum of f0237 ( figure [ rgbimages ] ) shows strong balmer absorption and an absence of emission lines indicating a starburst has recently ( @xmath39gyr ago ) been abruptly truncated . here
, the major merger of two spiral galaxies is the most likely trigger for the initial starburst @xcite .
however , in the absence of external mechanisms the poststarburst phase is expected to occur at the later stages of the major merger after the galaxies have coalesced @xcite .
the premature truncation of the starburst may be due to the enhanced ram pressure felt by a galaxy during a cluster merger , which strips gas more rapidly .
furthermore , the high pressure cluster merger environment may increase the star formation leading to a more rapid consumption of the gas .
the difference in the star - forming properties of f0083 and f0237 may be attributed to the different mass and dynamical states of the two systems .
given its luminosity , f0083 is likely more massive than f0237 , meaning it is less susceptible to complete stripping of gas due to ram pressure .
further , strong tidal forces due to the major merger in f0237 may unbind gas , making it more susceptible to complete ram pressure stripping @xcite thereby halting star formation in the disks .
we have presented hst observations of four rare `` jellyfish '' galaxies in the merging cluster a2744 .
the knots and filaments are brighter than the majority of jellyfish analogues found in local clusters , indicating more vigorous star formation is occurring compared to that found in tails of low redshift jellyfish .
intriguingly , three of these jellyfish galaxies lie in close proximity to merger - related features in the icm .
this leads us to propose that the rise of the jellyfish in a2744 is due to the effects of the high pressure of the icm during the ongoing cluster merger .
in particular , we suggest that the star formation occurring in the tails of stripped gas has been triggered by the rapid , dramatic increase in pressure during an interaction with the shock .
these observations support the hypothesis that galaxies undergo accelerated evolution in their star - forming properties during the violent core passage phase of a cluster merger .
we thank the referee , curtis struck , for his helpful comments .
m.s.o . and
w.j.c acknowledge the financial support of the australian research council .
was partly supported by nasa grant nas8 - 03060 . , e. , mellier , y. , radovich , m. , missonnier , g. , didelon , p. , & morin , b. 2002 , in astronomical society of the pacific conference series , vol .
281 , astronomical data analysis software and systems xi , ed .
d. a. bohlender , d. durand , & t. h. handley , 228+ , c. , pracy , m. b. , couch , w. j. , bekki , k. , lewis , i. , glazebrook , k. , baldry , i. k. , baugh , c. m. , bland - hawthorn , j. , bridges , t. , cannon , r. , cole , s. , colless , m. , collins , c. , dalton , g. , de propris , r. , driver , s. p. , efstathiou , g. , ellis , r. s. , frenk , c. s. , jackson , c. , lahav , o. , lumsden , s. , maddox , s. , madgwick , d. , norberg , p. , peacock , j. a. , peterson , b. a. , sutherland , w. , & taylor , k. 2004 , , 355 , 713 , a. m. , fruchter , a. s. , hook , r. n. , & hack , w. 2002 , in the 2002 hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system , ed . s. arribas , a. koekemoer , & b. whitmore , 337+ , j. , coe , d. , dupke , r. , massey , r. , zitrin , a. , cypriano , e. s. , okabe , n. , frye , b. , braglia , f. g. , jimnez - teja , y. , bentez , n. , broadhurst , t. , rhodes , j. , meneghetti , m. , moustakas , l. a. , sodr , jr . , l. , krick , j. , & bregman , j. n. 2011 , , 417 , 333 , r. j. , lucey , j. r. , hammer , d. , hornschemeier , a. e. , carter , d. , hudson , m. j. , marzke , r. o. , mouhcine , m. , eftekharzadeh , s. , james , p. , khosroshahi , h. , kourkchi , e. , & karick , a. 2010 , , 408 , 1417 | we identify four rare `` jellyfish '' galaxies in hubble space telescope imagery of the major merger cluster abell 2744 .
these galaxies harbor trails of star - forming knots and filaments which have formed in - situ in gas tails stripped from the parent galaxies , indicating they are in the process of being transformed by the environment .
further evidence for rapid transformation in these galaxies comes from their optical spectra , which reveal starburst , poststarburst and agn features .
most intriguingly , three of the jellyfish galaxies lie near icm features associated with a merging `` bullet - like '' subcluster and its shock front detected in _
chandra _
x - ray images .
we suggest that the high pressure merger environment may be responsible for the star formation in the gaseous tails .
this provides observational evidence for the rapid transformation of galaxies during the violent core passage phase of a major cluster merger . |
optical afterglow studies of gamma - ray bursts ( grbs ) provide a powerful tool for unveiling the interstellar medium ( ism ) properties of their host galaxies .
long grbs are connected to the death of a massive star , and their host galaxies are subsequently expected to be sites of heavy star - formation extending back to the early universe .
metallicities determined from absorption lines of the ism show higher values than those of , for example , qso absorbers ( fynbo et al . 2006a ; savaglio et al . 2006 ; savaglio 2009 ) , though this might be only a sightline effect ( fynbo et al .
kinematics of absorption lines give some indications for galactic outflows ( e.g. thne et al .
2007 ) , which are an expected phenomenon in star - forming galaxies .
usually , absorption lines in grb afterglow spectra are constant in time . in
a few cases metal absorption lines together with their corresponding fine structure transitions , caused by uv pumping from the intense afterglow radiation , vary in intensity following the change in the ionizing flux of the burst .
modeling of the line variability allows us to better constrain the distance from the grb to the absorbing material , which typically ranges from @xmath3pc to a few kpc ( e.g. vreeswijk et al .
2007 , delia et al . 2009 ) .
the presence of neutral material in the spectra implies distances on the order of a few hundred pc to several kpc ( e.g. prochaska et al .
2007 , thne et al . 2007 ) . for strong
nv absorption lines , it has been suggested that those lines must arise from material close to the grb site , at @xmath4100pc ( prochaska et al .
grb 080310 showed variation in all resonant fe ii transitions which can only be explained with photoionization by the burst at a distance of @xmath5 100pc ( a. de cia et al . in prep . ) .
excluding those few special cases , optical grb afterglow spectroscopy seems to predominantly probe the ism of the host galaxy and not the material around the grb progenitor . studying the circumburst environment , however , would allow us to verify the predictions of current progenitor models .
prompt emission x ray data showed some evidence for variations in the absorbing column density ( amati et al .
2000 ) . in the optical , absorption features from circumburst material
are difficult to detect due to the flash ionization of the surrounding material by the grb out to several tens of parsecs ( robinson et al .
2010 ) , even though the grb is predicted to reside in a dense environment .
in addition , any remaining absorbing material would be hidden under the stronger absorption lines from the line - of - sight ism in the host galaxy .
there are several potential signatures which could demonstrate that we are observing material close to the grb instead of the general ism of the galaxy : time variability of the absorption line strength ( perna & loeb 1998 ) , a high ionization state of the medium ( prochaska et al .
2008 ) , or absorption lines at large velocities compared to the redshift of the grb .
the last has been observed in a few cases ( fox et al .
2008 , mller et al .
2002 ) , but it can not be excluded that those systems are intervening absorbers and hence not connected to the grb .
it was also suggested that a large difference between the optical and x - ray column densities might be a hint of observing highly ionized material , eventually close to the grb ( watson et al .
2007 , campana et al . 2010 ) .
an ideal laboratory would be a grb progenitor that exploded in a relatively isolated environment with little ism along the line of sight .
so far , no conclusive evidence for such conditions has been found , probably because such physical conditions are rare and only a few afterglows have time resolved afterglow spectroscopy . in this paper
we present the first example for such a scenario , grb 090426 .
grb 090426 was discovered by the bat ( burst alert telescope ) @xmath6-ray telescope onboard the _ swift _ satellite ( gehrels et al .
2004 ) on april 26 , 2009 , 12:48 ut ( cummings et al . 2009 ) and had a duration of @xmath7s ( sato et al .
soon after its discovery an optical counterpart was found by _ swift _
( cummings et al . 2009 ) and the redshift was determined from an optical spectrum to be @xmath8 ( levesque et al .
both its restframe duration ( 0.3s ) as well as its observed duration put grb 090426 into the short burst category according to the commonly used classification based on the duration of the prompt emission . in 2
we describe the optical spectra and afterglow observations . in
3 we discuss the characteristics of grb 090426 itself and some issues regarding its classification . 4 describes the properties of the ism in the line of sight as derived from optical and x ray observations . in
5 we deal in detail with the ly@xmath1 line variation and model its behaviour with a photoionisation code .
6 describes the properties of the host galaxy of grb 090426 .
discussion and conclusions are reported in 7 .
we obtained both photometric and spectroscopic observations of the afterglow and the host galaxy .
the imaging data are listed in tab .
[ logafterglow ] , and the spectroscopic observations are detailed in tab .
we observed the optical afterglow with mosca and alfosc at the not as well as fors2 at the vlt from 2009 apr .
26 to june 19 in several bands .
in addition , we include acquisition images from the spectroscopic observations .
photometry was performed using sdss stars ( abazajian et al .
2009 ) as reference and the transformations described by jester et al .
( 2005 ) to convert the sdss to v and r@xmath9 magnitudes . in order to account for the contribution of the host galaxy system to the light curve independent of the seeing , we used aperture photometry centered on the afterglow with radius of 2 .
we obtained a spectrum of the afterglow with fors2 at the vlt starting on 2009 apr 27.048 ut ( 12.3 hr after the burst ) using grisms 600v and 600ri together with a 10 slit and an exposure time of 30min per grism .
the spectrum covers a wavelength range from 3500 to 8600 with a resolution of @xmath5 11 .
furthermore , we reanalyzed the spectrum from levesque et al . (
2010 ) taken with lris at keck 1.1 hr after the burst .
this spectrum has a similar resolution and wavelength coverage compared to the fors2 spectrum .
* the spectra were reduced using standard tasks in iraf , with the different exposures and grisms from each instrument combined , flux calibrated , and then normalized .
flux calibration prior to normalizing the spectrum makes the normalization more reliable , especially bluewards of ly @xmath1 and in the ly @xmath1 forest , since the afterglow continuum follows a smooth power law . * after the fading of the afterglow an object was found @xmath10 sw of the grb position and initially suggested as the candidate host galaxy ( g1 ) .
we took spectra of g1 with fors2 at the vlt on june 18 2009 using grism 300v ( 3300 8700 ) and exposure times of 2@xmath11s .
the acquisition image for the spectroscopic observations of g1 revealed another object underlying the grb position ( therefore suggested as the real host galaxy , h ) , for which we obtained another spectrum on june 19 2009 with the same settings and exposure times .
we also obtained 2 @xmath12 30min spectra with x - shooter at the vlt ( dodorico et al .
2006 ) on june 2 2009 , covering a wavelength range from 9800 to 24800 .
the slit was placed across the grb position ( and the host ) and g1 . [ cols="<,<,<,<,<,<",options="header " , ] the variability of ly@xmath1 strongly suggests that the material lies close to the grb progenitor .
such an effect has been predicted from photoionization modeling ( perna & loeb 1998 ) , but has never before been observed .
the high column densities of the high ionization lines and the absence of their low ionization counterparts support the scenario that we are observing a large quantity of matter close to and ionised by the grb .
this is further supported by the large difference between the optical and x ray column density , which can be explained by photoionization of the medium by the grb ; x ray absorption is insensitive to this phenomenon since it is caused by excitation of the electrons in the inner shells of the atoms ( campana et al .
2010 ) .
we compare the lines found in the spectrum of grb 090426 to a sample of 77 low - resolution afterglow spectra of long grbs observed by _ swift _
( fynbo et al . 2009 ) .
the majority of those spectra in the right wavelength range to detect ly@xmath1 absorption ( 33 ) showed a strong damped ly@xmath1 ( dla ) system with a median value of log n@xmath13/@xmath14 @xmath15 21.6 . only 6 other spectra had column densities below log n@xmath13/@xmath14 = 20.0 , the lower limit for the definition of a dla : grb 050908 ( log n@xmath13/@xmath14 = 17.8 ) , grb 060124 ( log n@xmath13/@xmath14 = 18.5 ) , grb 060607a
( log n@xmath13/@xmath14 @xmath5 17.2 ) , grb 070411 ( log n@xmath13/@xmath14 = 19.3 ) , grb 071020 ( log n@xmath13/@xmath14 @xmath4 20.0 ) , and grb 080310 ( log n@xmath13/@xmath14 = 18.8 ) .
all of them securely belong to the long burst group with @xmath16s between 25 and 100s .
five of these bursts are also highlighted in fig .
[ 090426:hr ] .
christensen et al .
( 2010 ) produced a composite using 60 of the 77 spectra from fynbo et al .
( 2009 ) mentioned above , and determined ews and column densities for all detected lines in the composite .
the following lines , which are usually strong in the environments of long - duration grbs , are not detected in our spectrum : siii @xmath17 1260,1304 and 1526 , c ii @xmath171534 , and al iii @xmath17 1670,1855 and 1863 .
the ews of c iv and si iv in the spectrum of grb 090426 are almost twice as high as in the composite grb spectrum .
considering that the hi column density of grb 090426 is very low compared to the approximate value for the composite ( log @xmath18/@xmath14= 21.7 ) , the relative abundance of these species are rather extreme for this burst . if we take only a composite of sub - dla bursts ( 3 spectra ) , we see the following : a fit of the ly@xmath1 absorption line gives log @xmath18/@[email protected] ( dominated by the spectrum of grb 070411 ) .
all the usual strong low ionization lines are also present in this composite such as si ii @xmath17 1260 , 1304 and 1526 , si ii * @xmath17 1264 , c ii @xmath17 1334 , and al iii @xmath17 1670 .
si iv @xmath17 1402 is not very strong , but falls in a noisy area of the composite , with ew@xmath20= [email protected] .
in contrast , civ 1549 is also quite strong here , with ew@xmath20 of [email protected] .
the values in fig .
[ 090426:comp ] are not derived from the composite spectrum since this is dominated by the grb 070411 spectrum ; instead instead we took the average of the ews of all sub - dlas . comparing the ionisation of c and si to those determined from the sample of fynbo et al .
2009 ( see fig . [ 090426:comp ] ) , grb 090426 occupies a rather different region than most burst in the sample .
however , it seems that other grbs with low n@xmath13 also tend to show a rather high ionisation state of the observed medium .
grb 060607a is the most extreme case , although it is not included in the plot since only civ has been listed in fynbo et al .
( 2009 ) . from fig . 3 in fox
et al . ( 2008 ) , who analysed the uves spectrum of this burst , we see that this burst , together with the lowest reported log n@xmath13/@xmath14 of 16.8 ( fox et al .
2008 , whereas fynbo et al .
2009 report a value of @xmath5 17.2 ) for a grb , has no cii or siii while the high ionisation species are rather strong .
grb 050809 has relatively strong low ionisation lines , but the value in fig .
[ 090426:comp ] is influenced by the nondetection of siiv @xmath17 1403 while the other line of the doublet at @xmath17 1393 is detected .
this seems to confirm that there is a correlation between the hydrogen column density and the ionisation of the medium , not surprising since a large hydrogen column density would screen the uv radiation from the grb .
however , it might also imply that some bursts occur in a rather isolated region of their host galaxy , which would allow us to truly probe ionised gas from the surroundings of the burst itself . /@xmath14
@xmath4 19 are labeled .
grb 090426 has a high ionization for both elements , similar to other bursts with low log n@xmath13.,title="fig : " ] /@xmath14 @xmath4 19 are labeled .
grb 090426 has a high ionization for both elements , similar to other bursts with low log n@xmath13.,title="fig : " ]
photoionization models allow to us determine the distance of the observed material from the grb by considering the ionization state , column densities , and their variations with time ( e.g. lazzati et al . 2001 ) . here
, we use the photoionisation code by perna & lazzati ( 2002 , see also lazzati & perna 2003 ) specifically written for application to grb observations .
the code assumes that the grb is initially surrounded by a cold ( @xmath21 k ) absorbing material with solar metallicity , either in the form of a spherical shell * with a thickness of 0.1 the shell radius * or a spherical cloud . the grb spectrum and its temporal evolution are the input in the code , which computes the evolution of the absorbing material s ionization state as a function of both time and the distance of the absorber from the burst .
at each time , the resulting wavelength - dependent opacity integrated over the line - of - sight is outputted and compared to the observations .
the total absorption at early times is a measure of the total absorbing column , while the speed of the evolution is a measure of the distance of the absorber .
this analysis is similar to what has been done for the combined soft - x - ray photoionization opacity in grb 050904 ( campana et al .
grb 090426 is better suited for this kind of analysis since we observe variation in the hi absorption .
since hydrogen is the most abundant element in the ism , constraining the geometry of the absorption region with the evolution of the ly@xmath1 absorption gives us a direct result that does not depend on metallicity considerations .
the four panels of figure [ fig : ew ] show the measured column densities of the h , c , si , and n ions with the best absorption model overlaid .
we obtain the best fit model for a uniform cloud surrounding the grb .
the cloud properties are fit using only the hi absorption , with the other ions and the soft - x - ray continuum absorption only used as consistency checks .
the absorber that best reproduces our ly@xmath1 measurements is a cloud of radius @xmath22 pc ( which corresponds to the size of a large molecular cloud ) and an initial hi column density @xmath23 @xmath14 .
such values correspond to a total mass of the molecular cloud of @xmath24 and an average density @xmath25 @xmath26 .
the errors on the values from the modeling are on the order of @xmath27 .
* the model does not rule out a solution in the form of a shell ; however , since the thickness of the shell is 0.1r , the density would be about 10 times larger as a result .
a wind blown medium would be fully ionised ( see robinson et al .
2010 ) at very early times and is therefore ruled out here . *
figure [ fig : clev ] shows the 1- , 2- , and 3- @xmath28 contour levels of the ly@xmath1 fitting in the @xmath29 plane ( thin solid lines ) .
overlaid is the area that is consistent with the combined constraint of siiv column evolution and soft x - ray continuum absorption .
the siiv constraint allows for regions of high column density ( right of the thick solid line ) while the soft x - ray absorption allows only for low column density regions ( left of the dashed line ) .
the shaded area is in full agreement with the ly@xmath1 result .
a much looser constraint comes from the nv fitting , while the civ column density can not be fit by our model . the civ excess might be an indication of a carbon - enriched medium , possibly from earlier sn explosions , but we caution the reader due to the large errors affecting the civ line ( see sec .
[ spectrum ] ) .
the agreement of the photoionization model predictions with the evolution of the column densities ultimately confirms that all the material we observe is close to the grb site .
such material has never been probed before in grb afterglow spectra .
in addition to the afterglow spectroscopy , we obtained observations of the host galaxy and its neighbour . using the acquisition images of the spectroscopic observations , we determined the magnitudes of the two galaxies g1 and h ( the host ) to be @xmath30 and @xmath31 , respectively , corresponding to absolute magnitudes of @xmath32(1500)= 21.05 and 20.43 . at the host redshift
, this implies relative luminosities of 1.5 and 0.8 @xmath33 ( reddy & steidel 2009 ) . fig .
[ 090426:host ] shows a color image of the host galaxy complex ( obtained from our acquisition images ) and indicates the position of the afterglow . in the fors2 observations ,
both galaxies show clear ly@xmath1 emission at redshifts of @xmath34 ( h ) and @xmath35 ( g1 ) , indicating that the host has about the same redshift as the grb while the companion is at a slightly higher redshift . at that redshift ,
the physical distance between their cores is only 9kpc in projection , which implies that they are likely interacting and possibly in the process of merging .
the ly@xmath1 emission of the host galaxy gives a star formation rate of 3.1 @xmath36 yr@xmath37 , on the lower end of what is typically observed in ( long duration ) grb hosts ( savaglio et al .
2009 ) . in the 2-dimensional x - shooter spectrum
we detect several faint emission lines in the near - ir , including [ oiii ] @xmath17 5007 , 4959 and h@xmath38 .
both [ oiii ] lines show two components separated by about 400 km s@xmath37 and offset in the spatial direction , which we attribute to the host ( h ) and g1 .
the second component of h@xmath38 is coincident with an atmospheric emission line . comparing the afterglow spectra and the emission lines from the hot gas of the two galaxies , the kinematics turn out to be rather complex ( see fig . [ 090426:kinematics ] ) .
the siiv transitions at @xmath17 1393 and 1403 each show two equally strong velocity components with a separation of about 300 kms@xmath37 . for nv and civ ,
the two components can not be separated since the doublets are closer together ( civ is also affected by a sky line ) .
the red component of the siiv lines is at about the same velocity as the center of the ly@xmath1 absorption .
the almost equally strong blue si iv component is coincident with the blue wing of ly@xmath1 , which could be fitted with a small additional component at that velocity but with only 1/100 to 1/1000 the column density of the main component .
the ly@xmath1 emission lines of the host and g1 are slightly redshifted compared to their nebular emission lines , which could imply an outflow of hydrogen gas as commonly observed in , for example , lyman break galaxies ( steidel et al .
furthermore , emission and absorption line velocities from the host do not coincide , which suggests a complex velocity field due to the physical motion of the material seen in absorption .
this is to be expected in the merging process of two galaxies .
it also supports the suggestion that the material we see in absorption is actually close to the grb and not only coming from the ism of the host galaxy .
the spectrum of grb 090426 is the first compelling case of observed photoionised material from the star - forming region of the progenitor itself .
its low hydrogen column density is varying , the absorption lines show a highly ionised medium of high column density , and the large discrepancy between x ray and optical absorption gives another indication for photoionised material being present .
grb spectra have shown a large range of hydrogen column densities , usually explained by different sightlines of the grb through its host galaxy ( fynbo et al .
2009 ; jakobsson et al .
2006 ) , with some column densities even lower than the one measured for grb 090426 .
however , if grb 090426 exploded in , for example , the halo of the galaxy or a very low density ism like a globular cluster , we would also expect low column densities of the metal absorption lines .
a location in the halo had been suggested for grb 070125 ( cenko et al .
2008 ) and grb 071003 ( perley et al .
2008 ) , both of which show a bright but featureless continuum with extremely weak absorption lines .
grbs with log @xmath18/@xmath14 @xmath4 20.0 tend to show a higher fraction of ionised versus neutral material , possibly explained by a lack of uv shielding by the lower density of hydrogen ; however , some of them still contain a relatively large column density of neutral material .
in addition , the afterglow showed that the grb must have exploded in a non - low - density medium ( levesque et al .
the strange properties of the spectrum of grb 090426 can therefore not simply be explained by a sightline effect .
considering the fact that this grb exploded in what seems to be a galaxy merger , another scenario might explain the strange spectrum we observe .
the lack of absorption from the normal ism of the host suggests that the progenitor did not reside in a star - forming region inside its host galaxy .
a possible scenario is therefore that the progenitor was located in an isolated star - forming region outside the host itself , e.g. in something like a tidal tail created by the merging of the two galaxies .
thus , the sightline to the grb would only intersect the material of the star - forming region , ionised by the grb flux and eventually piled up by the wind from massive wolf - rayet stars , and not the dense ism of the galaxy .
we note , however , that the afterglow position is , in projection , close to the center of the galaxy .
another strange feature about this burst was its relatively short duration paired with a soft high energy emission spectrum .
naturally , the question arises if the unusual high - energy properties and those of the afterglow spectrum could be connected . from fig .
[ 090426:hr ] we see that all other low column density grbs occupy the normal parameter space for long bursts concerning hr and duration , so there seems to be no obvious correlation between hr or duration and low column density .
since the material observed in the afterglow spectrum is relatively far away from the burst compared to the prompt emission , which is expected to be produced by internal shocks within the jet of the grb , a connection is not to be expected . concerning the progenitor of the burst , antonelli et al .
( 2009 ) and levesque et al .
( 2010 ) had already concluded that , despite its short duration , the burst was likely due to the collapse of a massive star .
arguments include the star - forming host galaxy , the high afterglow luminosity , consistency with the amati relation , and the strength of the absorption lines . in past years
, the classification of bursts purely on its duration or another single observational fact has also been questioned .
evidence for a collapsar progenitor has been observed even in the absence of a supernova signature ( grb 060505 and grb 060614 , see e.g. fynbo et al .
2006b ) , which we can not verify for grb 090426 since the sn would be too faint to detect with current instruments , and a range of high redshift bursts have been detected with short intrinsic duration ( see e.g. l et al . 2010 ) .
suggestions for how to produce a collapsar with such a short intrinsic duration include viewing the grb from off - axis , such that we see only a small part of the jet ( lazzati et al .
2009 ) , or the possibility that the central engine had been turned off before the jet reached the surface of the str , with the destruction wave catching up with the jet and leading to a shorter prompt duration ( mizuta & aloy 2009 ) .
no other grb observed so far has shown such extreme properties of the absorption lines , despite a number of spectra with relatively low column densities . however
, none of them had time resolved spectral observations that would allow a line variability study ( with the exception of grb 080310 , a potentially similar case to grb 090426 concerning the properties of the afterglow spectrum ; see a. de cia et al . in prep . ) .
a larger number of rapid response observations and time series of spectra using a high sensitivity and medium- to high - resolution spectrographs ( in order to fit reliable column densities ) might allow us to study a few more of those rare events where we directly see the interaction of the grb with its environment .
extremely rapid response , on the timescale of a few minutes , might even allow us to observe the material ejected by the progenitor itself .
observations are based on eso proposals 083.d-0606(c ) , 083.d-0606 , and commissioning data of the x - shooter instrument .
we thank the paranal staff , especially g. carraro , p. lynam , e. mason , j. smoker , f. j. selman and s. stefl for performing our observations .
also based on observations with the nordic optical telescope under program 39 - 023 ( pi jakobsson ) , operated on the island of la palma jointly by denmark , finland , iceland , norway , and sweden in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
we thank the observers zita banhidi and sren frandsen for performing the not observations .
ct and eml thank josh bloom for his guidance in obtaining the keck afterglow spectrum .
ct thanks alex kann for the discussion on the brightness of short vs. long afterglows .
ct & sc are partially supported by asi grants .
the dark cosmology centre is funded by the danish national research foundation .
lc is supported by the dfg cluster of excellence origin and structure of the universe " .
eml s participation was funded in part by a ford foundation predoctoral fellowship .
maa is partly founded through grants aya2007 - 67626-c03 - 01 and prometeo-2009 - 103 . | long duration gamma - ray bursts are commonly associated with the deaths of massive stars .
spectroscopic studies using the afterglow as a light source provide a unique opportunity to unveil the medium surrounding it , probing the densest region of their galaxies .
this material is usually in a low ionisation state and at large distances from the burst site , hence representing the normal interstellar medium in the galaxy . here
we present the case of grb 090426 at @xmath0 , whose optical spectrum indicates an almost fully ionised medium together with a low column density of neutral hydrogen . for the first time
, we also observe variations in the ly@xmath1 absorption line .
photoionisation modeling shows that we are probing material from the vicinity of the burst ( @xmath2 pc ) .
the host galaxy is a complex of two luminous interacting galaxies , which might suggest that this burst could have occurred in an isolated star - forming region outside its host galaxy created in the interaction of the two galaxies .
[ firstpage ] gamma - rays : bursts x rays : general x ray : ism |
roughly speaking , randomness is the fact that , even using all the information that we have about a physical system , in some situations it is impossible , or unfeasible , for us to predict exactly what will be the future state of that system .
randomness is a facet of nature that is ubiquitous and very influential in our and other societies @xcite . as a consequence , it is also an essential aspect of our science and technology .
the related research theme , that was motivated initially mainly by gambling and led eventually to probability theory @xcite , is nowadays a crucial part of many different fields of study such as computational simulations , information theory , cryptography , statistical estimation , system identification , and many others @xcite .
one rapid - growing area of research for which randomness is a key concept is the maturing field of quantum information science ( qis ) .
our main aim in this interdisciplinary field is understanding how quantum systems can be harnessed in order to use all nature s potentialities for information storage , processing , transmission , and protection @xcite .
quantum mechanics @xcite is one of the present fundamental theories of nature .
the essential mathematical object in this theory is the density operator ( or density matrix ) @xmath1 .
it embodies all our knowledge about the preparation of the system , i.e. , about its state . from the mathematical point of view , a density matrix is simply a positive semi - definite matrix ( notation : @xmath2 ) with trace equal to one ( @xmath3 ) . such kind of matrix can be written as @xmath4 , which is known as the spectral decomposition of @xmath1 . in the last equation
@xmath5 is the projector ( @xmath6 and @xmath7 , where @xmath8 is the @xmath9dimensional identity matrix ) on the vector subspace corresponding to the eigenvalue @xmath10 of @xmath1 . from the positivity of @xmath1
follows that , besides it being hermitian and hence having real eigenvalues , its eigenvalues are also non - negative , @xmath11 .
once the trace function is base independent , the eigenvalues of @xmath1 must sum up to one , @xmath12 .
thus we see that the set @xmath13 possesses all the features that define a probability distribution ( see e.g. ref .
@xcite ) .
the generation of pseudo - random quantum states is an essential tool for inquires in qis ( see e.g. refs .
@xcite ) and involves two parts .
the first one is the generation of pseudo - random sets of projectors @xmath14 , that can be cast in terms of the creation of pseudo - random unitary matrices .
there are several methods for accomplishing this last task @xcite , whose details shall not be discussed here . here
we will address the second part , which is the generation of pseudo - random discrete probability distributions @xcite , dubbed here as pseudo - random probability vectors ( prpv ) . in this article
we go into the details of three methods for generating numerically prpv .
we present the problem details in sec .
[ sec_problem ] . the sec .
[ sec_iid ] is devoted to present a simple method , the iid method , and to show that it is not useful for the task regarded in this article . in sec .
[ sec_norm ] the standard normalization method is discussed .
the bias of the prpv appearing in its naive - direct implementation is highlighted . a simple solution to this problem via random shuffling of the prpv components
is then presented . in sec .
[ sec_trigo ] we consider the trigonometric method .
after discussing some issues regarding its biasing and numerical implementation , we study and compare the probability distribution generated and the computer time required by the last two methods when the dimension of the prpv is varied .
the conclusions and prospects are presented in sec .
[ conclusions ] .
by definition , a discrete probability distribution @xcite is a set of non - negative real numbers , @xmath15 that sum up to one , @xmath16 in this article we will utilize the numbers @xmath17 as the components of a probability vector @xmath18 despite the nonexistence of consensus regarding the meaning of probabilities @xcite , here we can consider @xmath17 simply as the relative frequency with which a particular value @xmath19 of a physical observable modeled by a random variable @xmath20 is obtained in measurements of that observable under appropriate conditions .
we would like to generate numerically a pseudo - random probability vector @xmath21 whose components @xmath22 form a probability distribution , i.e. , respect eqs .
( [ eq : prob1 ] ) and ( [ eq : prob2 ] ) .
in addition we would like the prpv to be unbiased , i.e. , the components of @xmath21 must have similar probability distributions . a necessary condition for fulfilling this last requisite
is that the average value of @xmath17 ( notation : @xmath23 ) becomes closer to @xmath24 as the number of prpv generated becomes large . at the outset
we will need a pseudo - random number generator ( prng ) . in this article
we use the mersenne twister prng @xcite , that yields pseudo - random numbers ( prn ) with uniform distribution in the interval @xmath25 $ ] .
we observe however that the results reported in article can be applied straightforwardly also when dealing with true random numbers @xcite .
a simple way to generate an unbiased pseudo - random probability vector @xmath0 is as follows . if we create @xmath26 _ independent _ pseudo - random numbers @xmath19 with _ identical _ probability _ distributions _ in the interval @xmath25 $ ] ( so the name of the method ) and set @xmath27 we will obtain a well defined discrete probability distribution , i.e. , @xmath28 and @xmath29 .
besides , as @xmath21 is unbiased , the mean value of @xmath17 approaches @xmath24 as the number of samples grows .
for one million random samples generated using the iid method . in the inset
is shown the probability distribution for the components of the probability vector @xmath30 . ] nevertheless , we should note that the sum @xmath31 shall be typically greater than one .
this in turn will lead to the impossibility for the occurrence of prpvs with one of its components equal ( or even close ) to one . as can be seem in fig .
[ fig_stat_prob_dist_iid ] , this problem becomes more and more important as @xmath26 increases .
therefore , this kind of drawback totally precludes the applicability of the iid method for the task regarded in this article .
lets us begin our discussion of this method by considering a probability vector with dimension @xmath32 , i.e. , @xmath33 .
if the prng is used to obtain @xmath34 $ ] and we impose the normalization to get @xmath35 , we are guaranteed to generate an uniform distribution for @xmath36 $ ] for both @xmath37 and @xmath38 . if @xmath39 then @xmath40 and the prng is used again ( two times ) to obtain @xmath34 $ ] and @xmath41 $ ] .
note that the interval for @xmath42 was changed because of the _ normalization _ of the probability distribution , which is also used to write @xmath43 . as @xmath44 is equiprobable in @xmath25 $ ] ,
for a large number of samples of the prpv , its mean value will be @xmath45 .
this shall restrict the values of the other components of @xmath21 , shifting the `` center '' of their probability distributions to @xmath46 , biasing thus the prpv .
of course , if one increases the dimension of the prpv , the same effect continues to be observed , as is illustrated in the table below for @xmath47 prpv generated for each value of @xmath26 . [ cols="^,^,^,^,^,^,^",options="header " , ] as was the case with the normalization method , for the trigonometric method we need to generate @xmath48 prn per prpv , @xmath49 for the angles and @xmath49 for the random permutation . nevertheless , because of the additional multiplications in eq .
( [ eq : trig_param ] ) , the computation time for the last method is in general a little greater than that for the former , as is shown in fig .
[ fig_time_d ] . .
the calculations were realized using a processor 1.3 ghz intel core i5 . ]
one may wonder if the normalization and trigonometric methods , that are at first sight distinct , lead to the same probability distributions for the prpv s components and also if they produce an uniform distribution for the generated points in the probability hyperplane .
we provide some evidences for positive answers for both questions in figs .
[ fig_comparison ] and [ fig_sample_prob_space ] , respectively . .
we see that the two methods yield , for all practical purposes , the same probability distributions for the components of the prpv . ] , with exception of the slight overpopulated corners , we get a fairly uniform distribution of points in the probability space . ]
in this article we discussed thoroughly three methods for generating pseudo - random discrete probability distributions .
we showed that the iid method is not a suitable choice for the problem studied here and identified some difficulties for the numerical implementation of the trigonometric method .
the fact that in a direct application of both the normalization and trigonometric methods one shall generate biased probability vectors was emphasized .
then the shuffling of the pseudo - random probability vector components was shown to solve this problem at the cost of the generation of additional @xmath49 pseudo - random numbers for each prpv .
it is worthwhile recalling that pure quantum states in @xmath50 can be written in terms of the computational basis @xmath51 as follows : @xmath52 where @xmath53 and @xmath54 .
the normalization of @xmath55 implies that the set @xmath56 is a probability distribution .
thus the results reported in this article are seem to have a rather direct application for the generation of _ unbiased pseudo - random state vectors_. we observe however that the content presented in this article can be useful not only for the generation of pseudo - random quantum states in quantum information science , but also for stochastic numerical simulations in other areas of science .
an interesting problem for future investigations is with regard to the possibility of decreasing the number of prn , and thus the computer time , required for generating an unbiased prpv .
this work was supported by the brazilian funding agencies : conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) and instituto nacional de cincia e tecnologia de informao quntica ( inct - iq ) .
we thank the group of quantum information and emergent phenomena and the group of condensed matter theory at universidade federal de santa maria for stimulating discussions .
we also thank the referee for his(her ) constructive comments .
j. maziero , r. auccaise , l.c .
celeri , d.o .
soares - pinto , e.r .
deazevedo , t.j .
bonagamba , r.s .
sarthour , i.s .
oliveira , r.m .
serra , quantum discord in nuclear magnetic resonance systems at room temperature , braz .
* 43 * , 86 ( 2013 ) m. wahl , m. leifgen , m. berlin , t. rhlicke , h .- j .
rahn , o. benson , an ultrafast quantum random number generator with provably bounded output bias based on photon arrival time measurements , appl .
. lett . * 98 * , 171105 ( 2011 ) | the generation of pseudo - random discrete probability distributions is of paramount importance for a wide range of stochastic simulations spanning from monte carlo methods to the random sampling of quantum states for investigations in quantum information science . in spite of its significance ,
a thorough exposition of such a procedure is lacking in the literature . in this article
we present relevant details concerning the numerical implementation and applicability of what we call the iid , normalization , and trigonometric methods for generating an unbiased probability vector @xmath0 .
an immediate application of these results regarding the generation of pseudo - random pure quantum states is also described . |
low mass dwarfs ( @xmath0 ) are the dominant stellar component of the milky way , composing @xmath1 of all stars @xcite and nearly half the stellar mass of the galaxy . however ,
despite their abundance , m dwarfs were studied in relatively small samples , due to their dim intrinsic brightnesses ( @xmath2 ) .
however , the situation has been radically altered in the last decade , as deep surveys covering large areas of the sky were carried out .
these projects , such as the sloan digital sky survey ( sdss , * ? ? ?
* ) and the two
micron all sky survey ( 2mass , * ? ? ?
* ) , can trace their roots back to photographic surveys , epitomized by the national geographic society - palomar observatory sky survey ( poss - i , * ? ? ?
* ) and its successor , poss - ii @xcite .
however , the photometric precision achieved by modern surveys distinguishes them from their photographic predecessors .
for example , sdss has imaged 1/4 of the sky to @xmath3 and 2mass imaged the entire sky to @xmath4 , with typical uncertainties of a few percent .
the resulting databases contain accurate multi - band photometry of tens of millions of low mass stars , enabling exciting new science .
while there is a rich heritage of historical investigations , this article will focus on results derived from sdss data .
a multitude of studies have focused on the intrinsic properties of low mass stars using sdss observations .
the field luminosity function ( lf ) and corresponding mass function ( mf ) was measured using over 15 million stars @xcite .
the mass radius relation of m dwarfs has been studied with eclipsing binary systems @xcite .
average photometric colors and spectroscopic features have also been quantified @xcite .
multiple studies have attempted to estimate the absolute magnitude of low mass stars in sdss @xcite . however , due to the lack of precise trigonometric parallaxes for many of the m dwarfs in sdss , absolute magnitudes and distances are derived by secondary means .
chromospheric activity , driven by magnetic dynamos within the stars , has been observationally traced with h@xmath5 emission measured in sdss spectroscopy @xcite .
flare rates have been measured for thousands of stars @xcite and will be incorporated into predicting the flaring population observed by next
generation surveys @xcite . while interesting objects in their own right , m dwarfs are also powerful tools for studying the milky way .
their ubiquity , combined with lifetimes much greater than a hubble time @xcite , make them ideal tracers of galactic structure and kinematics .
using photometry of over 15 million m dwarfs , @xcite measured the scale heights and lengths of the thin and thick disks .
a complementary study by @xcite used higher mass stars to measure the stellar density profiles .
the local gravitational potential has been probed using the kinematics of @xmath6 7000 m dwarfs along one line of sight with proper motions , sdss spectroscopy and photometry @xcite .
this study has been expanded to the entire sdss footprint , using a database of @xmath6 25,000 m dwarfs with velocities and distances @xcite .
furthermore , @xcite employed proper motions and distances of @xmath6 2 million m dwarfs to estimate velocity distributions .
below , i highlight several studies that have used sdss observations to study both the intrinsic properties of low mass stars _ and _ use them to study the milky way . in
[ sec : observations ] , the technical details of sdss photometric and spectroscopic observations are briefly described . in [ sec : lf ] , a new investigation measuring the m dwarf field lf and mf and galactic structure parameters is detailed .
kinematics within the milky way are explored in [ sec : kinematics ] .
the importance of placing large , deep surveys of m dwarfs in a galactic context is discussed in [ sec : case_study ] .
concluding remarks and avenues for future investigations are discussed in [ sec : conclusions ] .
the sloan digital sky survey ( sdss , * ? ? ? * ) was a photometric and spectroscopic survey covering @xmath6 10,000 sq .
deg . centered on the northern galactic cap conducted with a 2.5 m telescope @xcite at apache point observatory ( apo ) .
photometry was acquired in five filters ( @xmath7 , * ? ? ?
* ; * ? ? ?
* ) down to a limiting magnitude @xmath3 @xcite .
calibration to a standard star network @xcite was obtained through concurrent observations with the `` photometric telescope '' ( pt ; * ? ? ?
* ; * ? ? ?
absolute astrometric accuracy was estimated to better than 0.1 @xcite , and has been used to measure proper motions with a precision of 3 mas yr@xmath8 @xcite . when the northern galactic cap was not visible at apo , a 300 sq .
deg . stripe ( stripe 82 ) centered on zero declination was scanned repeatedly .
observations of stripe 82 were used to empirically quantify photometric precision @xcite . over 357 million unique photometric objects
have been identified in the latest public data release ( dr7 , * ? ? ?
the photometric precision of sdss is unrivaled for a survey of its size , with typical errors @xmath9 0.02 mag @xcite . when the conditions at apo were not photometric , sdss operated in a spectroscopic mode .
twin fiber fed spectrographs obtained medium resolution ( @xmath10 ) optical spectra , covering 3800 - 9200 .
objects with sdss photometry were targeted for spectroscopic followup by complex algorithms .
the primary targets were galaxies @xcite and quasars @xcite .
however , over @xmath6 70,000 m dwarfs have been identified within the sdss spectroscopic database @xcite from targeted and serendipitous ( i.e. , failed quasar ) observations .
in the following section , the method and results of @xcite are detailed .
the @xcite study produced two distinct yet intertwined results : it combined the investigation of an intrinsic stellar property ( the mf and lf ) with a measurement of galactic structure and self consistently described the local low
mass stellar population .
the wide , deep , precise photometric coverage of sdss enabled this type of analysis .
previous investigations of the lf and mf have relied on one of two techniques : 1 ) nearby , volume limited studies of trigonometric parallax stars ( e.g. * ? ? ?
* ) ; or 2 ) pencil beam surveys of distant stars over a small solid angle ( e.g. * ? ? ?
* ) . in both cases ,
sample sizes were limited to a few thousand stars , prohibiting detailed statistical measurements . using a sample drawn from sdss , 2mass and guide star catalog photometry and supplemented with sdss spectroscopy
, @xcite performed the largest field low mass lf and mf investigation prior to the @xcite study . @xcite
measured the lf with a sample covering 30 sq .
deg . and containing @xmath11 low mass stars .
they also constrained their contamination rate to @xmath9 a few percent after obtaining spectra of every red point source in a 1 sq .
deg . calibration region .
please refer to @xcite for a detailed description of sample selection .
m dwarfs were selected from sdss data release 6 ( dr6 ; * ? ? ? * ) footprint using @xmath12 and @xmath13 photometric colors .
noisy photometric observations were removed using flag cuts , yielding a final database containing over 15 million low mass stars .
the @xcite photometric sample comprised a data set three orders of magnitude larger ( in number ) than any previous lf study .
furthermore , it covered over 8,400 sq@xmath14 deg@xmath14 , nearly 300 times larger than the sample analyzed by @xcite .
the large sky coverage of sdss presented the main challenge in measuring the lf .
most previous studies either assumed a uniform density distribution ( for nearby stars ) or calculated a galactic density profile , @xmath15 along one line of sight . with millions of stars
spread over nearly 1/4 of the sky , numerically integrating galactic density profiles for each star was computationally prohibitive . to address this issue , @xcite employed the following technique for measuring the luminosity function .
first , absolute magnitudes were assigned and distances to each star were computed using new @xmath16 and @xmath17 color absolute magnitude relations ( cmrs ) estimated from nearby stars with @xmath7 photometry and accurate trigonometric parallaxes .
next , a small range in absolute magnitude ( 0.5 mag ) was selected and the stellar density was measured as a function of galactic radius ( _ r _ ) and galactic height ( _ z _ ) .
this range in absolute magnitude was selected to provide high resolution of the lf , with a large number of stars ( @xmath18 ) in each bin . finally , a galactic profile was fit to the @xmath19 density maps , solving for the shape of the thin and thick disks and the local stellar density .
the lf was then constructed by combining the local density of each absolute magnitude slice .
an example of one density map and the corresponding fit is shown in figure [ fig : maps ] .
the measured `` raw '' lf is subject to various systematic effects .
unresolved binarity , metallicity gradients within the milky way , interstellar extinction and malmquist bias can all influence a star s estimated absolute magnitude and distance .
understanding these effects is important , since the photometric data set contains millions of stars and poisson errors are no longer significant within a given lf bin . rather , systematic errors in absolute magnitudes are the dominant source of uncertainty . in the @xcite study ,
the uncertainty in each lf bin was calculated by repeating the entire analysis multiple times , using five different prescriptions for the cmr , which were selected to account for metallicity gradients , extinction and reddening .
the effects of unresolved binarity and malmquist bias were quantified using monte carlo realizations of the milky way .
each model was populated with synthetic systems that were consistent with the observed galactic structure and lf .
binaries were included within the simulations , using four different prescriptions for mass ratio distributions .
the mock stellar catalog was analyzed with the same pipeline as the actual observations .
the differences between the input and `` observed '' galactic structure and lf were used to systematically correct the observed `` raw '' values from sdss data . the single
star lf was also estimated during this step .
the detailed analysis of malmquist bias and unresolved binaries and their effect on the resultant lf are given in @xcite .
the final adopted system and single
star @xmath20 lfs are presented in figure [ fig : lf_sing_system ] .
the uncertainty in each lf bin was computed from the full spread due to differences among cmrs and binary prescriptions .
the system lf ( black circles ) rises smoothly to @xmath21 , with a fall off to lower masses . the single
star lf ( red circles ) follows a similar track , but maintains a density increase of @xmath22 compared to the system lf at the same absolute magnitude .
this implies that lower luminosity stars are easily hidden in binary systems , but isolated low luminosity systems are not common . using the mass magnitude relations of , the field system and single
star lfs were converted to mfs , shown in figure [ fig : reid_zheng ] .
the system mf exhibits similar behavior to the lf , peaking near 0.3 @xmath23 . the single
star mf is compared to the local `` eight parsec '' mf of @xcite in figure [ fig : reid_zheng ] .
the @xcite sample consisted of 558 stars spread over 3@xmath24 steradians .
there is broad agreement over most of the mass range probed , with the sdss mf falling below the local mf at higher masses .
the discrepancy is probably due to the assumed cmr , which may have overestimated the brightness of higher mass m dwarfs ( see contribution by hawley et al .
this volume ; * ? ? ? * ) . to first order
, this would lead to overestimated distances and underestimated local stellar densities .
future investigations will rederive the lf and mf with updated cmrs .
the sdss mf was fit with two functions : a broken power law and lognormal .
both adequately represent the underlying data points , but a lognormal is slightly preferred due to the smaller number of free parameters .
the right panel of figure [ fig : reid_zheng ] compares the sdss system mf with the mf reported by @xcite .
the @xcite hst survey was a pencil beam survey of 1,400 m dwarfs over @xmath6 1 sq . deg .. their system mf agrees with most of the sdss mf , especially at lower masses .
the @xcite power law fit agrees favorably with the low mass component of the sdss broken power law fit . at higher masses
, there is a disagreement , which is likely due to cmr differences .
when searching for variations between mfs , it is crucial that mf @xmath25 is compared , rather than best fit functional forms .
this point , first introduced in @xcite and reiterated by is important .
the comparison of @xmath26 and not @xmath25 may have exaggerated the differences between the local census ( i.e. * ? ? ?
* ) and distant samples ( i.e. * ? ? ?
* ) , although other factors , such as unresolved binarity ( e.g. * ? ? ?
* ; * ? ? ?
* ) also contributed .
finally , the sdss single - star mf is compared to selected seminal works in the field in figure [ fig : big_shots ] .
while photometric studies can be used to study the distribution of m dwarfs throughout the galaxy and infer their fundamental properties , they offer a static glimpse of the population .
the dynamics of stars through the milky way , as measured by radial velocities and proper motions , can be used to infer the underlying local mass distribution and gravitational potential . in particular , kinematic studies of m dwarfs
have been useful for studying the properties of the thin and thick disks . using a sample of @xmath6 7,000
m dwarfs along one line of sight , @xcite examined the @xmath27 velocity distributions as a function of distance from the plane .
thin and thick disk components were identified with the use of probability plots @xcite . in general , both the thin and thick disks dispersions increase with height above the plane , with the @xmath28 velocity dispersion being the smallest of the three directions . low
mass stars follow the galactic potential across the entire mass range . in other words , early and late type m dwarfs , despite differing in mass by nearly a factor of ten , have similar velocity dispersions at the same height from the galactic midplane .
similar trends with height were also quantified by @xcite using proper motions and distances of @xmath6 2 million sdss m dwarfs . the analysis from @xcite
was expanded to the entire sdss footprint by @xcite , using the largest spectroscopic sample of m dwarfs ever assembled @xcite .
@xcite determined mean velocities , dispersion and relative scalings between the thin and thick disk .
they searched for velocity gradients radially in the disk , and observed a similar increase in dispersion away the plane ( see figure [ fig : disps ] ) .
@xcite kinematically estimated the local fraction of thick disk stars to be @xmath29 , similar to the photometric result from @xcite .
this study exemplifies the utility of low mass stars as probes of the local milky way .
in addition to their utility as tracers of the local milky way , low
mass dwarfs have intrinsic properties , such as chromospheric activity @xcite , that can be examined in a broader context .
additionally , metallicity may be studied using subdwarfs @xcite , readily identified by their spectra , which show enhanced calcium hydride ( cah ) absorption .
subdwarfs have been easily detected in large - scale surveys such as the sdss ( e.g. , * ? ? ?
* ; * ? ? ?
the study of m dwarf chromospheric activity in a galactic context is described below .
similar investigations were severely limited before the advent of large spectroscopic databases , such as sdss .
it highlights the importance of studying m dwarfs in a broader context within the milky way , as their distributions within the galaxy may be used as a valuable constraint on intrinsic properties .
low mass stars harbor strong magnetic dynamos @xcite , which in turn heat their chromospheres , producing balmer line emission . for early - type stars ( m0-m4 ) ,
the dynamos are probably powered by stellar rotation , but the driving mechanism at later types is not well known .
ca h @xmath30 k emission is also a tracer of activity in m dwarfs @xcite , but is often not observationally accessible for these dim , red stars . prior to sdss , surveys that measured the fraction of active m dwarfs often consisted of @xmath31 1000 stars , with very few stars ( @xmath32 ) in the faintest luminosity bins @xcite . since previous samples were flux
limited , they were heavily biased towards observing nearby , young stars .
the first major study of activity in m dwarfs with sdss data was performed by @xcite .
this study observed many stars at distances 5 - 10 times larger than previous studies due to the multi - fiber capabilities of the sdss spectrographs .
the differences between the @xcite study and the past surveys were immediately obvious .
first , some spectral types ( m7 and m8 ) which were previously observed to have active fractions near 100@xmath33 had lower active fractions , near 70@xmath33 .
the active fraction distributions for all spectral types declined with distance from the galactic midplane with a spectral type dependent slope .
@xcite suggested this was an age effect , as stars further from the plane also exhibited larger velocity dispersions .
thus , older stars have undergone more dynamic interactions and also possess weaker magnetic fields .
the m dwarf age - activity relation was calibrated by @xcite , which determined the activity lifetime for m dwarfs as a function of spectral type . by using a simple model that fit both the velocity and active fraction distributions within the galaxy
, @xcite were able to constrain the ages of stars , a fundamental stellar property .
it is worth noting that without large samples of m dwarfs , such as those available with sdss , this type of analysis would not be possible .
the @xcite study also highlights the potential biases that are encountered when limiting samples to the solar neighborhood .
the sun is very close to the midplane ( 15 pc ; * ? ? ?
* ) , and nearby stars are predominately young , biasing local samples .
current models of low mass stars can not accurately reproduce observable features .
this has led to uncertainty in determining fundamental parameters , such as mass , age and luminosity .
thus , empirical relations have been developed to estimate these properties .
magnitude relations determined from eclipsing binaries , have proven extremely useful , but other fundamental parameters , most notably age and metallicity have not been fully calibrated .
recent advances in measuring metallicity are promising ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and see the review of the metallicity splinter session in this volume ) .
in particular , @xcite have calibrated an infrared technique that delivers 0.15 dex precision from @xmath34 band spectra . however , for the vast majority of m dwarfs with optical spectra , a similar level of precision has not been achieved . further observations of m dwarfs with known metallicities
( i.e. , those in wide binaries with fgk companions ( i.e. , * ? ? ? * ) or cluster members ) need to be obtained before a suitable calibration can be determined
. however , once precise metallicities can be measured for m dwarfs , they will be a powerful probe of the milky way s chemical history .
age is notoriously difficult to estimate for field stars ( of any mass ) .
however , age activity relations ( i.e. * ? ? ?
* ) , suggest that observational tracers , such as chromospheric activity , can constrain the ages of m dwarfs to a few gyr .
further work is needed to produce precise estimates of m dwarf ages .
the next generation of synoptic surveys , such as lsst and panstarrs may be useful in this endeavor and provide a link between rotation periods and stellar age .
i have briefly detailed some of the investigations that examine both intrinsic m dwarf properties and galactic structure and kinematics .
the requisite data for such studies , namely precise , deep , multi - band photometry over large areas of the sky , will become more common in the future , as additional large surveys come online . within the next decade
, surveys such as panstarrs , lsst and gaia , will produce new data sets that are ideal for similar investigations . while databases containing millions of low mass stars will be the norm for future studies , it is important to note that systematic errors will be the dominant source of error in many areas . in sdss
, nearly all the observed m dwarfs do not have reliable trigonometric parallaxes , forcing the use of photometric parallax estimates . reliably calibrating and testing these photometric parallaxes
will only become more important as surveys push to larger distances .
other systematics , such as how metallicity affects the absolute magnitude of m dwarfs , will also be crucial to future investigations .
finally , the importance of sdss spectroscopy can not be overstated .
the large spectroscopic samples of sdss m dwarfs have enabled many novel investigations .
significant spectroscopic followup of the next generation of surveys should be a high priority .
i would like to thank the cool stars 16 soc for the opportunity to present this work .
i also thank suzanne hawley , andrew west and kevin covey for their insight and many enlightening conversations over the years .
i thank sebastian pineda for kindly providing a figure .
i would also like to acknowledge the comments of neill reid , ivan king , niall deacon and gilles chabrier , who all contributed to work presented in this article .
, r. l. , & abell , g. o. 1963 , the national geographic society - palomar observatory sky survey ( basic astronomical data : stars and stellar systems , edited by k. a. strand . published by the university of chicago press , chicago , il usa , 1968 , p.481 ) , 481 | modern sky surveys , such as the sloan digital sky survey and the two micron all sky survey , have revolutionized the study of low mass stars . with millions of photometric and spectroscopic observations ,
intrinsic stellar properties can be studied with unprecedented statistical significance . low
mass stars dominate the local milky way and are ideal tracers of the galactic potential and the thin and thick disks .
recent efforts , driven by sdss observations , have sought to place the local low - mass stellar population in a broader galactic context .
i highlight a recent measurement of the luminosity and mass functions of m dwarfs , using a new technique optimized for large surveys . starting with sdss photometry ,
the field luminosity function and local galactic structure are measured simultaneously . the sample size used to estimate
the lf is nearly three orders of magnitude larger than any previous study , offering a definitive measurement of this quantity .
the observed lf is transformed into a mass function and compared to previous studies .
ongoing investigations employing m dwarfs as tracers of galactic kinematics are also discussed .
sdss spectroscopy has produced databases containing tens of thousands of low mass stars , forming a powerful probe of the kinematic structure of the milky way .
sdss spectroscopic studies are complemented by large proper motion surveys , which have uncovered thousands of common proper motion binaries containing low mass stars .
additionally , the sdss spectroscopic data explore the intrinsic properties of m dwarfs , including metallicity and magnetic activity .
the highlighted projects demonstrate the advantages and problems with using large data sets and will pave the way for studies with next generation surveys , such as panstarrs and lsst . |
one of the most important open question in extragalactic studies regards cosmic reionization and clarifying which sources drive this important phase transition in the early universe . while much evidence suggests that the process might be driven by galaxies ( e.g. , robertson et al .
2013 , 2015 ; mitra et al . 2015 ; bouwens et al .
2015 ) , others have suggested that quasars could provide the dominant contribution ( giallongo et al .
2015 ; madau & haardt 2015 ; mitra et al .
2016 ) .
the important issues appear to be whether large numbers of faint quasars exist at high redshift ( e.g. , willott et al . 2010 ; mcgreer et al .
2013 ) , whether faint galaxies show an appreciable ( @xmath95% ) escape fraction ( e.g. , siana et al . 2010 , 2015 ;
vanzella et al .
2012 , 2016 ; nestor et al .
2013 ) , and what the total emissivity is in the rest - frame @xmath10 in faint galaxies beyond the limits of current surveys in the _ hubble _ ultra deep field ( hudf : beckwith et al . 2006 ; ellis et al .
2013 ; illingworth et al .
important issues for the latter question are the precise values of the faint - end slopes and the faint - end cut - off to the @xmath10 luminosity function .
depending on the value of the faint - end slope and the luminosity where a cut - off in the lf occurs ( kuhlen & faucher - gigure 2012 ; bouwens et al .
2012 ; robertson et al . 2013 ; bouwens 2016 ) , the total emissivity from galaxies in the @xmath10 can vary by factors of @xmath42 - 10 . one potentially promising way to constrain the total luminosity density in the rest - frame @xmath10 is by taking advantage of the impact of gravitational lensing by galaxy clusters for magnifying individual sources .
this can bring extremely faint galaxies into view such that they can be detected with current telescopes ( e.g. , bradac et al .
2009 ; maizy et al .
2010 ; coe et al .
there has been a significant investment in this approach by _ hst _ in the form of the _ hubble _ frontier fields program ( coe et al .
2015 ; lotz et al .
2016 ) , which is investing 840 orbits into reaching @xmath429-mag in 7 optical+near - ir bands , as well as two uvis channels in a supporting effort ( alavi et al .
2016 ; siana 2013 , 2015 ) .
already , analyses of sources behind the hff clusters have resulted in the identification of @xmath0 - 8 sources first to @xmath11 mag ( atek et al .
2015a , b ) and later to @xmath4@xmath12 mag ( kawamata et al . 2016 ; castellano et al .
2016a , b ; livermore et al . 2016 [ hereinafter , l16 ] ) . at @xmath13 - 3
, it has been similarly possible ( alavi et al .
2014 , 2016 ) to probe to @xmath4@xmath12 mag taking advantage of very deep wfc3/uvis observations over abell 1689 and various clusters in the hff program . based on these deep searches , the volume density of galaxies at @xmath616 mag have been estimated , with quoted faint - end slopes for @xmath13 - 3 lfs that range from @xmath14 to @xmath15 ( alavi et al .
2014 , 2016 ) and from @xmath15 to @xmath16 for @xmath0 - 8 lfs ( atek et al .
2015a , b ; ishigaki et al . 2015 ; laporte et al . 2016 ; castellano et al . 2016b ; l16 ) , respectively . in spite of the great potential that lensing clusters have for probing the faint end of the @xmath10 lf , successfully making use of data over these clusters to perform this task in an accurate manner
is not trivial .
the entire enterprise is fraught with sources of systematic error .
one of these sources of systematic error concerns the assumed size distribution of extremely faint galaxies ( bouwens et al .
2016 ; grazian et al . 2011
; oesch et al .
small differences in the assumed half - light radii have the potential to change the inferred faint - end slopes by large factors , i.e. , @xmath17 depending on whether the mean size of extremely faint galaxies is 120 mas , 30 mas , or 7.5 mas ( e.g. , see figure 2 from bouwens et al .
fortunately , we found that most of the extremely faint sources seem consistent with being almost unresolved , i.e. , with intrinsic sizes of @xmath1810 - 30 mas ( bouwens et al .
2016 ; see also kawamata et al .
2014 ; laporte et al .
2016 ) , making this issue much more manageable in terms of its impact ; but it still remains an uncertainty . a second source of systematic error arises from errors in the magnification maps , since this can have a profound impact on the lfs derived . finally , there are issues related to subtraction of the foreground cluster light , contamination from individual sources in the clusters ( e.g. , globular clusters ) , and from other less important systematic effects that affect determinations of the volume densities in the field vs. the cluster .- band , which is useful for discriminating between @xmath0 and @xmath19 galaxies , while the hudf and candels ( grogin et al . 2011 ; koekemoer et al .
2011 ) programs do feature deep integrations in this filter . the availability or not of deep observations in the @xmath20 band could impact the @xmath0 and @xmath19 samples and lf results derived from these data sets in different ways . ] even without such considerations
, it is easy to see that systematics could be a concern for lf studies from lensing clusters , simply by comparing several recent lf results from clusters with similar results based on deep field studies using the hudf .
to give one recent example , alavi et al .
( 2016 ) reported a faint - end slope @xmath21 of @xmath22 for the @xmath10 lf at @xmath23 based on an analysis of sources behind 3 lensing clusters , while parsa et al .
( 2016 ) reported a faint - end slope of @xmath24 based on a deep @xmath23 search over the hudf .
these results differ at a significance level of @xmath49@xmath8 taking at face value the quoted statistical errors .
this is but one example of the large differences frequently present between lf results derived from deep field studies and those derived on the basis of lensing clusters ( see figure 1 for several other examples ) .
in addition to the clear scientific importance of the faint - end slope @xmath21 for computing the total ionizing emissivity from faint galaxies , the observations also allow us to test for a possible flattening or turn - over of the @xmath10 lf at low luminosities .
many cosmological hydrodynamic simulations of galaxy formation predict a flattening in the @xmath10 lf at @xmath25 or @xmath26 due to less efficient atomic and molecular hydrogen cooling , respectively ( muoz & loeb 2011 ; krumholz & dekel 2012 ; kuhlen et al .
2013 ; jaacks et al . 2013 ; wise et al . 2014 ; liu et al .
2016a ; gnedin 2016 ; finlator et al . 2016 ) , while other simulations predict a flattening in the range @xmath27 to @xmath25 mag due to the impact of radiative feedback ( oshea et al . 2015
; ocvirk et al .
2016 ) . meanwhile , by combining abundance matching and
detailed studies of color - magnitude diagram of low - luminosity dwarfs in the local universe , evidence for a low - mass turn - over in the luminosity function has been reported at @xmath12 mag ( boylan - kolchin et al .
2015 ; see also boylan - kolchin et al .
current observations likely provide us with some constraints in this regime . however , given the significant systematics that appear to be present in current determinations of the faint - end slope @xmath21 from lensing clusters ( vs. field results ) , it is not at all clear that current constraints on the form of the @xmath10 lf at @xmath615 mag are reliable , particularly at @xmath28 . in the present paper ,
we take the next step in our examination of the impact of systematic errors on derived lf results from lensing clusters , after our previous paper on this subject , i.e. , bouwens et al .
( 2016 ) , where the emphasis was on the uncertain sizes of faint sources . here
the focus will be more on the uncertainties in lf results that arise from errors in the gravitational lensing models .
as we will demonstrate explicitly , the recovered lf from a straightforward analysis tends to migrate towards a faint - end slope of @xmath29 ( or slightly steeper ) , if uncertainties in the magnification factor are large .
the impact of these uncertainties is to wash out features in the lf , particularly at low luminosities . given that magnification factors @xmath30 necessarily become uncertain when these factors are high , i.e. , @xmath31 and especially @xmath32 , accurately constraining the shape of the lf at extremely low luminosities and also detecting a turn - over or flattening is very challenging .
the purpose of this paper is to look at the constraints we can set on the faint end of the @xmath0 @xmath10 lf with a thorough assessment of the possible systematic errors . in doing so , we will look for possible evidence for a turn - over in the lf at very low luminosities and if not present , what constraints can be placed on the luminosity of a turn - over .
evidence for a turn - over will be evaluated through the introduction of a _ curvature parameter _ which we constrain through extensive markov - chain monte - carlo ( mcmc ) trials .
the confidence intervals we obtain on the shape of the @xmath10 lf at faint magnitudes will provide theorists with some important constraints for comparison with models and cosmological hydrodynamic simulations .
most importantly , these results provide balance to some discussion in the literature arguably making premature claims for the lf rising steeply to very low luminosities . to keep the focus of this paper on our new techniques , we restrict our analysis to just the @xmath0 lf from the first four hff clusters .
the plan for this paper is as follows .
2 summarizes the data sets we use to select our @xmath0 sample and derive constraints on the @xmath0 lf .
3 provides some useful context for the issue of errors in the magnification models and shows the general impact it would have on lf results .
4 and 5 present new lf results at @xmath0 using our new forward - modeling methodology .
6 compares our new results with previous reported lf results , as well as results from various theoretical models .
finally , in 7 , we summarize and conclude .
we refer to the hst f814w , f850lp , f105w , f125w , f140w , and f160w bands as @xmath33 , @xmath20 , @xmath34 , @xmath35 , @xmath36 , and @xmath37 , respectively , for simplicity . through this paper , a standard `` concordance '' cosmology with @xmath38 km s@xmath39 mpc@xmath39 , @xmath40 and @xmath41
is assumed . this is in good agreement with recent cosmological constraints ( planck collaboration et al .
magnitudes are in the ab system ( oke & gunn 1983 ) .
ccccccc + glafic & y & y & glafic & y & 0.03@xmath42 & oguri ( 2010 ) ; ishigaki et al . 2015
; kawamata et al .
( 2016 ) + cats & y & y & lenstool & y & 0.1@xmath42 & jullo & kneib ( 2009 ) ; richard et al .
( 2014 ) ; jauzac et + & & & & & & al .
( 2015a , b ) + sharon / johnson & y & y & lenstool & y & 0.06@xmath42 & johnson et al .
( 2014 ) + zitrin - nfw & y & y & zitrin & y & 0.06@xmath42 & zitrin et al .
( 2013 , 2015 ) + + + grale & n & y & grale & n & 0.22@xmath42 & liesenborgs et al .
( 2006 ) ; sebesta et al .
( 2016 ) + bradac & n & y & bradac & n & 0.2@xmath42 & bradac ( 2009 ) + zitrin - ltm & y & n & zitrin & n & 0.06@xmath42 & zitrin et al .
( 2012 , 2015 ) .
in our selection of @xmath0 galaxies , we make use of the v1.0 reductions of the deep _ hst _ optical and near - ir _ hst _ observations available over the first four clusters in the hff program : abell 2744 , macs0416 , macs0717 , and macs1149 ( a. koekemoer et al . 2016 , in prep ) .
the optical observations include @xmath418 , @xmath410 , and @xmath442 orbits of acs observations in the f435w , f606w , and f814w bands from 0.4@xmath30 m to 0.9@xmath30 m .
near - ir observations over these fields total 34 , 12 , 10 , and 24 orbits in the f105w , f125w , f140w , and f160w , reaching to roughly a @xmath43 limiting magnitude of 28.8 to 29.0 mag .
subtraction of foreground light from cluster galaxies and cluster galaxies was performed using galfit ( peng et al .
2002 ) and the median - filtering algorithm of sextractor ( bertin & arnouts 1996 ) run at two different grid scales .
there are many similarities of our procedure to that from merlin et al .
the only areas clearly inaccessible to us in our searches for faint @xmath0 galaxies occur directly under the cores of bright stars or galaxies in the cluster .
our procedure performs at least as well as any other procedure currently in use ( merlin et al .
2016 ; l16 ) .
while the approach used by l16 appears to be a good one , there is no evidence it is better than the procedures presented by merlin et al .
( 2016 ) or the approach we have developed .
a complete description of both our photometric procedure and selection criteria for identifying @xmath0 galaxies is provided in r.j .
bouwens et al .
( 2016 , in prep ) .
it remains similar to that done in many of our previous papers ( e.g. , bouwens et al .
we briefly summarize our criteria here for selecting a robust and large sample of @xmath0 galaxies .
we select all sources that satisfy the following @xmath33-dropout color criteria @xmath44 and which are detected at @xmath96.5@xmath8 adding in quadrature the s / n of sources in the @xmath34 , @xmath35 , @xmath36 , and @xmath37 band images measured in a 0.35@xmath42-diameter aperture .
the above color selection criterion also explicitly excludes the inclusion of @xmath45 @xmath34-dropout galaxies .
as the above criteria identify sources at both @xmath0 and @xmath19 , we compute the redshift likelihood function @xmath46 for each source and only include those sources where the best - fit photometric redshift is less than 6.3 .
sources are further required to have a cumulative probability @xmath1835% at @xmath47 to keep contamination to a minimum in our high - redshift samples .
our sample of 160 @xmath0 candidate galaxies is the largest compilation reported to date .
each of the hff clusters we examine in this study have at least seven independent lensing models available , with both convergence @xmath48 and shear @xmath49 maps ( table [ tab : lensingmodel ] ) .
we estimate the magnification of sources based on publicly - available models by first multiplying the @xmath48 and @xmath49 maps of each cluster by @xmath50 and then computing the magnification @xmath30 as follows : @xmath51 for our magnification estimates for individual sources , we take the median of the model magnifications from the cats ( jullo & kneib 2009 ; richard et al .
2014 ; jauzac et al . 2015a , b ) , glafic ( oguri 2010 ; ishigaki et al .
2015 ; kawamata et al .
2016 ) , sharon / johnson ( johnson et al .
2014 ) , and zitrin parametric nfw models ( zitrin - nfw : zitrin et al . 2013 , 2015 ) .
the parametric models generally provided the best estimates of the magnification for individual sources in the hff comparison project ( meneghetti et al .
2016 ) , but we emphasize that many non - parametric magnification models also performed very well .
we present in figure [ fig : sample ] the distribution of absolute magnitudes and magnification factors we estimate for sources in our @xmath0 sample .
we set an arbitrary maximum magnification factor of 100 , given the lack of predictive power for magnification maps at such high values ( see 3.1 ) .
this constraint affects very few sources .
our selection includes sources ranging from @xmath52 mag to @xmath53 and with magnification factors ranging from 1.2 to 50 , with the bulk of the sources having absolute magnitudes of @xmath54 and magnification factors of @xmath42 .
we should emphasize that the inferred luminosities and total magnitudes we report for sources are intended to provide a rather complete accounting for light in individual sources .
they are based on scaled - aperture photometry following the kron ( 1980 ) method , with a correction for flux on the wings of the psf ( e.g. , see bouwens et al .
however , in comparing our total magnitude measurements with the magnitude measurements from other groups ( e.g. , l16 : see 6.1.2 ) , we have found some sources being reported to have apparent magnitude measurements fainter by @xmath40.7 - 1.0 mag than what we measure for the same sources ( including the very faint @xmath55 source reported by l16 that anchors their lf at the faint end : see 6.1.2 and 6.2 ) .
if we quote the luminosities of sources in our study using a similar procedure as to what l16 appear to use where individual sources are often @xmath40.7 - 1.0 mag fainter than we find our probe would extend to @xmath56 mag ( indicated in figure [ fig : sample ] with the red bin ) , essentially identical to that claimed by l16 ( see also castellano et al .
2016 and kawamata et al .
we emphasize , however , that the low luminosities claimed by measuring magnitudes in this way exaggerate how faint the hff program probes .
we discuss this further in 6.1.2 and 6.2 .
we prefer our photometric scheme for accounting for the total light in faint sources .
an important aspect of the present efforts to provide constraints on the @xmath0 lf will be our explicit efforts to include a full accounting of the uncertainties present in the magnification models we utilize .
we begin by looking first at the general size of errors in the magnification models and second at how the errors would impact lfs derived from lensing clusters . in making use of various gravitational lensing models to derive constraints on the prevalence of extremely faint galaxies at high redshift , it is important to obtain an estimate of how predictive the lensing models are for the true magnification factors .
one way of addressing this issue is the fully end - to - end approach pursued by meneghetti et al .
( 2016 ) and involves constructing highly - realistic mock data sets , analyzing the mock data sets using exactly the same approach as are used on the real observations , and then quantifying the performance of the different methods by comparing with the actual magnification maps . while each of the methods did fairly well in reproducing the magnification maps to magnification factors of @xmath410 , the best performing methods for reconstructing the magnification maps of clusters were the parametric models , with perhaps the best reconstructions achieved by the glafic models , the sharon / johnson models , and the cats models .
an alternate way of addressing this issue is by comparing the public lensing models against each other .
here we pursue such an approach .
we treat one of the models as the truth and then to quantify the effectiveness of the other magnification models taken as a set for predicting that model s magnifiction map .
we consider both the case that the true mass profile of the hff clusters is considered ( 1 ) to lie among parametric class of models built on nfw - type mass profiles and ( 2 ) to lie among the non - parametric class of models which allow for more freedom in the modeling process .
we take the former models to include the glafic , cats , sharon / johnson , and zitrin - nfw models , and the latter to include the bradac [ 2009 ] , grale [ liesenborgs et al .
2006 ; sebesta et al .
2016 ] , and zitrin - ltm [ zitrin et al . 2012 , 2015 ] .
a brief description of the general properties of the public lensing models can be found in table [ tab : lensingmodel ] . in performing this test
, we assume that the median of the magnification models provides our best means for predicting magnifications in the model we are treating as the truth . the truth model is always excluded when constructing the median magnification map for this test .
the results of this exercise are presented in the upper panel of figure [ fig : predpow ] for both the parametric and non - parametric model cases .
the geometric mean of the predictive power for each of the parametric ( glafic , cats , sharon / johnson , zitrin - nfw ) and non - parametric ( grale , bradac , zitrin - ltm ) models is presented as an assessment of the predictive power of each approach for similar classes of models .
the upper and lower bounds of the shaded regions give the predictive power for models corresponding to best and worst performing cluster , while the solid line in the center of the shaded region gives the median performance .
the lower panel of figure [ fig : predpow ] shows the position - to - position scatter around the median magnification in the model treated as the truth . from this exercise ,
it is clear that the magnification maps have excellent predictive power to magnification factors of @xmath410 in all cases and perhaps to even higher magnification factors assuming that the magnification profiles of hff clusters are as well behaved as in the parametric models .
the scatter , however , is already very large at magnification factors of 10 .
we will extend this exercise in a future work ( r.j .
bouwens et al .
2016 , in prep ) .
the exercise we perform in this section shows similarities in philosophy to the analyses that priewe et al .
( 2016 ) pursue , in comparing magnification models over the hff clusters with each other to determine the probable errors in the individual magnification maps .
one prominent conclusion from that study was that differences between the magnification maps was almost always larger than the estimated errors in the magnification for a given map , pointing to large systematics in the construction of the individual maps .
this provides some motivation for the tests we perform here and in future sections in this paper , and confirmation of the importance of this study .
other powerful tests of the predictive power of the magnification maps , and the challenges , were provided by observations of sne ia ( rodney et al .
2015 ) .
the purpose of this subsection is illustrate the impact of magnification errors on the derived lfs from the hff clusters .
two different example lfs are considered for this exercise : ( 1 ) one with a faint - end slope of @xmath57 and a turn - over at @xmath11 and ( 2 ) another with a fixed faint - end slope @xmath21 of @xmath57 and no turn - over .
how well can we recover these lfs given uncertainties in the magnification maps ?
we can evaluate this by generating a mock catalog of sources for each of the first four clusters from the hff program using one set of magnification models ( `` input '' models ) and then attempting to recover the lf using another set of magnification models ( `` recovery '' models ) . in computing the impact of lensing ,
the redshifts are fixed to @xmath58 for all sources .
the input magnification models are taken to be the glafic models for this exercise .
these catalogs include positions and magnitudes for all the individual sources in each cluster .
the selection efficiencies of sources are accounted for when creating the mock catalogs , as estimated in appendix a. in performing this exercise , we ignore errors in our estimates of the selection efficiencies and small number statistics at the faint end of the lf .
one can try to recover the input lfs from these mock catalogs , using various magnification models .
sources are binned according to luminosity using the `` recovery '' magnification model .
the selection volumes available in each luminosity bin are also estimated as described in appendix a using this `` recovery '' magnification model . to demonstrate the overall self - consistency in our approach
, we show the recovered lfs using the same magnification model as we used to construct the input catalogs in the top two panels in figure [ fig : illust ] .
what is the impact if different lensing models are used to recover the lf than those used to construct the mock catalogs ?
the lowest two rows of panels in figure [ fig : illust ] show the results using the latest magnification models by grale , cats , zitrin - ltm , and the median of the cats , sharon / johnson , and zitrin - nfw models where available .
these simulation results demonstrate that the recovery process appears to work very well for input lfs with faint - end slopes of @xmath57 ( _ left panels _ in figure [ fig : illust ] ) independent of the magnification model , with all recovered lfs showing a very similar form to the input lfs .
very different results are , however , obtained in our attempts to recover input lfs with a turn - over at @xmath115 mag ( _ right panels _ in figure [ fig : illust ] ) using magnification models that are different from the input model .
for all four magnification models we consider , the recovered lfs look very similar to the lf example we just considered .
all recovered lfs show a steep faint - end slope to @xmath59 mag .
what is striking is that they do not reproduce the turn - over present in the input model at @xmath11 mag .
there are some differences in the recovered lfs depending on how similar the input magnification model is to the recovery model , with effective faint - end slopes of @xmath57 , @xmath60 , and @xmath61 achieved with the grale and zitrin - ltm models , the cats models , and the median parametric models , respectively .
the glafic magnification model is not used when constructing the median parametric magnification model .
both examples demonstrate that the faint - end slope for the recovered lfs tend to gravitate towards a value of @xmath57 .
the faint - end slope of @xmath57 results from magnification errors ( that become large for @xmath62 ) trading off with selection volume errors so as to scatter sources towards ( 1 ) higher luminosities and lower volume densities or ( 2 ) lower luminosities and higher volume densities . in either case
, the scatter runs approximately parallel to a faint - end slope @xmath21 of @xmath57 ( or slightly steeper depending on the intrinsic sizes of sources : see appendix b ) .
two other examples of the impact of large magnification errors on lf results are presented in figures [ fig : illustsize ] and [ fig : illust15 ] in appendices b and c , utilizing an input lf with a faint - end slope of @xmath63 . for each of these examples
, the recovered lf closely matches the input lf ; dramatically , however , faintward of @xmath11 mag ( and even @xmath116 mag for some models ) , the recovered lfs steepen and asymptote again towards a faint - end slope of @xmath4@xmath57 ( or steeper if sources are resolved ) , even if the actual slope of the lf is much shallower ( or the lf turns over ! ) .
each of these examples demonstrate that , regardless of the input lf , a faint - end slope @xmath21 of @xmath29 will be recovered whenever the magnification uncertainties are large .
one can not , therefore , use the consistent recovery of a steep faint - end slope based on a large suite of lensing models to argue that the actual lf maintains a steep form to extremely low luminosities ( as was done by l16 using their figure 11 ) .
the presented examples show this is not a valid argument .
how then can one interpret lf results from lensing clusters when a steep lf @xmath64 is found ?
as we have demonstrated , such a result could be indicative of the lfs truly being steep or simply an artifact of large magnification uncertainties .
to determine which is the case , the safest course of action is to simulate all steps in the lf recovery process , to determine the impact of magnification uncertainties on the shape of the lf , and finally to interpret the recovered lfs from the observations . while we showed a few examples here
, we formalize the process in the next section .
the purpose of the present section is to describe a new methodology for quantifying the constraints on the @xmath10 lf to very low luminosities , given the uncertainties in the magnification maps .
the development of such a procedure is useful given the challenges presented in the previous section .
we will apply this formalism in 5 .
the lf recovery results presented in 3 ( figure [ fig : illust ] ) illustrate the impact that errors in the magnification maps can have on the recovered lfs .
input lfs , of very different form , can be driven towards a faint - end slope @xmath21 of @xmath57 at the faint end , after accounting for the impact of magnification errors .
the results from 3.2 demonstrate the importance of forward modeling the entire lf recovery process to ensure that both the results and uncertainties are reliable .
we then utilize our forward - modeling approach to derive constraints on the @xmath0 lf .
the basic idea behind our approach follows closely from the simulations we ran in the previous section and is illustrated in figure [ fig : flowchart ] .
we begin by treating one of the public magnification models as providing an exact representation of reality . in conjunction with an input lf ,
those models are used to create a mock data set for the four hff clusters considered .
the mock data set is then interpreted using other magnification models for the clusters to determine the distribution of sources vs. @xmath10 luminosity @xmath65 and also to recover the @xmath10 lf . as illustrated by the lf recovery experiments presented in 3.2 ,
the recovery could be done with the models individually or by using some combination of models like the median .
there are many advantages to using the present procedure to derive accurate errors on the overall shape of the @xmath10 lf .
probably the most significant of these is inherent in the end - to - end nature of the present procedure and our relying significantly on forward modeling to arrive at accurate errors on the observational results . through the construction of many mock data sets using plausible magnification models and recovery using other similarly plausible models ,
the proposed procedure allows us to determine the full range of allowed lfs .
in addition , the advocated procedure provides us with a natural means to account for `` noise '' in the lensing model magnification maps .
the presence of lensing model `` noise '' is obvious looking at the range in magnification factors across the various models ( e.g. , compare the glafic and grale critical lines in figure [ fig : flowchart ] or see the lower panel in figure [ fig : predpow ] ) .
such noise can even be present in a median magnification model created from the combination of many individual models , as there will be regions where the high - magnification regions of the lensing maps simply overlap due to chance coincidence .
analogous to considerations of low - significance sources in imaging observations , the robustness of specific magnification factors in the median map can be assessed , by considering comparisons with independent determinations of the same map . through the treatment of one of the public magnification models as the truth
, the present forward - modeling approach effectively formalizes such a technique to determine the robustness of specific features in the magnification maps .
the advantage of the current procedure is that this robustness can be determined using the full magnification maps available for each cluster ( and not just at a limited number of positions where candidate high - magnification @xmath31 sources happen to be found in the real observations ) , while also allowing us to derive more reliable likelihood distributions ( with realistic errors ) .
another advantage of our forward modeling procedure is that it explicitly incorporates source selection .
this is important , since the selection efficiency @xmath66 could depend on the magnification factor @xmath30 in the sense that the most magnified sources would also be the most incomplete , as is likely the case for the brightest and most extended objects , due to the impact of lensing shear on source detection ( oesch et al .
if the same situation applied to the faintest sources in the hffs ( and one did not utilize a procedure that included forward modeling or an explicit correction ) , the recovered lfs would be biased .
this is due to the fact that the actual surface density of the sources on the sky is @xmath67 , but it is assumed to be proportional to @xmath68 and @xmath69 at high magnifications @xmath31 ( figure [ fig : predpow ] ) .
the recovered lf would therefore be higher by the factor @xmath70 .
we would expect such an issue to affect recovered lfs , in all cases where sources have non - zero size ( since the selection efficiency would then depend on the magnification factor ) .
for example , if the completeness of sources shows an inverse correlation with the magnification factor as oesch et al .
( 2015 ) find , i.e. , @xmath71 ( e.g. , as in their figure 3 ) , a direct approach would recover a faint - end slope that is @xmath72 steeper than in reality ( figure [ fig : illustsize ] from appendix b ) , at very low luminosities where @xmath31 ( where @xmath18@xmath73@xmath9 is typically less than @xmath18@xmath74@xmath9 ) . remarkably , there is no evidence that this issue is even considered in many recently derived lfs , which is worrisome given the size assumptions which were made .
this is most problematic for analyses pushing to very low luminosities , i.e. , @xmath75 mag , while quoting tiny statistical uncertainties ( e.g. , l16 who quote statistical uncertainties on @xmath21 of @xmath70.04 vs. this bias which is @xmath48@xmath3 larger ) .
we perform our forward modeling simulations at the catalog level , to ensure that the time requirements on these simulations are manageable .
this involves the construction of catalogs of sources with precise positions and apparent magnitudes .
both in the construction of the mock catalogs and in recovering the lf from these catalogs , the selection efficiency @xmath66 must be accounted for , which is in general a function of the apparent magnitude @xmath76 and magnification factor @xmath30 , i.e. , @xmath77 . for the lowest luminosity @xmath78 - 8 galaxies
, there is little evidence to suggest that these sources show significant spatial extension ( bouwens et al .
2016 ) , which implies that we can credibly treat their selection efficiencies as just a function of the apparent magnitudes , i.e. , @xmath79 .
we describe our procedure for estimating @xmath79 in this case in appendix a. in putting together the mock suite of observations for each lf parameter set we are considering , i.e. , @xmath80 , @xmath21 , and a third parameter @xmath81 to be introduced in the next section , @xmath42@xmath310@xmath82 sources are inserted at random positions ( uniformly in the source plane ) across the 4 hff cluster fields we are using ( figure [ fig : flowchart ] ) . in constructing the mock set of observations , we include sources in the catalogs in proportion to their estimated selection efficiencies @xmath77 from appendix a. all sources in the input catalogs
are assumed to have the same input redshift @xmath58 . during the recovery process
, sources are placed into individual bins in @xmath10 luminosity using a `` recovery '' magnification model , which we take to be the median of the parametric magnification models . during the recovery process
, the redshift is taken to have the same mean value as assumed in constructing the mock catalogs , i.e. , @xmath83 , but with a @xmath84 uncertainty of 0.3 .
this is to account for the impact of uncertainties in the estimated redshifts of individual sources on the recovered lfs .
we use the results of the simulations we run for each parameter set ( each with @xmath42@xmath310@xmath82 sources ) to establish the expectation values for the number of sources per luminosity bin ( in the same 0.5-mag intervals used in the previous section ) .
we then compute the likelihood of a given parameterization of the lf by comparing the observed number of galaxies per bin in luminosity ( considering all 4 clusters at the same time ) with the expected numbers assuming a poissonian distribution , as @xmath85 where @xmath86 is the expected number of sources in a given bin in intrinsic @xmath10 luminosity .
the observed number of sources in a given bin in @xmath10 luminosity @xmath87 is computed from the median parametric magnification maps , as done in 2 ( figure [ fig : sample ] ) . a flow chart showing one example of our forward - modeling approach
is provided in figure [ fig : flowchart ] .
use of a parametric form to the lf is particularly useful for examining the overall lf constraints on the shape of the @xmath10 lf in lensed fields , due to uncertainties that exist on the magnification factors , and hence luminosities , of individual sources used in the construction of the lf .
this makes it difficult to place sources in specific bins of the @xmath10 lf ( resulting in each bin showing a larger error ) . for the parametric modeling
we do of the lf , we start with a general schechter form for the lf : @xmath88 however , we modify the basic form of the schechter function by multiplying the general schechter form by the following expression faintward of @xmath89 mag : @xmath90 positive values of @xmath81 result in the lf turning over faintward of @xmath89 mag , while negative values of @xmath81 result in the lf becoming steeper faintward of @xmath89 mag . an illustration of this parameterization is provided in figure [ fig : param ] .
our use of @xmath89 mag allows us to test for possible curvature in the shape of the lf at @xmath616 mag , as predicted by some models ( e.g. , kuhlen et al .
2013 ; jaacks et al . 2013 ;
ocvirk et al .
@xmath89 mag is also just faintward of luminosities probed in field studies ( i.e. , @xmath91 mag : bouwens et al .
finally , since @xmath92 mag corresponds to the luminosity of @xmath31 faint sources in the hffs , our fitting for a curvature parameter @xmath81 allows us to investigate how well the shape of the lf can be recovered in the regime where the magnification factors are large . with this parameterization ,
the turn - over luminosity @xmath93 can be easily shown to be @xmath94 assuming that @xmath95 .
we now make use of the formalism we presented above and our selection of 160 @xmath0 galaxies behind the first four hff clusters by r.j .
bouwens et al .
( 2016 , in prep ) to set constraints on the form of the uv lf at extremely faint magnitudes .
+ cccccc + glafic & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + cats & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + grale & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + bradac & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + sharon / johnson & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + zitrin - nfw & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + zitrin - ltm & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + mean & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + mean parametric & @xmath120.94 & [email protected] & @[email protected] & 0.00 & + + + glafic & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + cats & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + grale & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + bradac & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + sharon / johnson & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + zitrin - nfw & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + zitrin - ltm & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + mean & @xmath120.94 & [email protected] & @[email protected] & [email protected] & + mean parametric & @xmath120.94 & [email protected] & @[email protected] & [email protected] & + + + glafic & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + cats & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + grale & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + bradac & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + sharon / johnson & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + zitrin - nfw & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + zitrin - ltm & @xmath120.94 & [email protected] & @[email protected] & [email protected] & @[email protected] + mean & @xmath120.94 & [email protected] & @[email protected] & [email protected] & + mean parametric & @xmath120.94 & [email protected] & @[email protected] & [email protected] & + + + atek et al .
( 2015 ) & @xmath96 & 0.28@xmath97 & @xmath98 + livemore et al .
( 2016 ) & @xmath99 & 0.23@xmath100 & @xmath12.10@xmath101 & & @xmath102 + + + bouwens et al .
( 2015 ) & @xmath103 & 0.50@xmath104 & @xmath105 + finkelstein et al .
( 2015 ) & @xmath106 & @xmath107 & @xmath108 we begin by looking at the constraints we can set on the shape of the @xmath0 lf by restricting our analysis to @xmath0 samples found behind the first four hff clusters .
such an exercise is useful , since it allows us to examine the lf constraints we obtain from our hff search results , entirely independent of results in the field . in deriving best - fit lf results , we use the following approach .
we fix @xmath109 to the same value bouwens et al .
( 2016 ) found at @xmath0 , i.e. , @xmath110 mag , set the curvature @xmath81 to be zero , and then fit for @xmath80 and @xmath21 .
we use a markov chain monte carlo ( mcmc ) algorithm where we start with the field lf results from bouwens et al .
( 2015 ) , i.e. , @xmath111 mpc@xmath112 and @xmath113 to determine how the likelihood of various parameter combinations varies as a function of @xmath80 and @xmath21 .
for each set of parameters @xmath21 and @xmath80 , we repeat the simulations described in 4.2 to calculate the likelihood of those parameters . for those simulation
, we consistently use magnification maps from one team to create the mock observations and then recover the results using the median magnification maps formed from the parametric models . on the basis of the grid of likelihoods we derive
, we determine the most likely values for @xmath80 and @xmath21 , while also determining covariance matrix which best fits the same likelihood grid . from the covariance matrix
, we estimate errors on @xmath80 and @xmath21 .
to determine the impact that errors in the magnification maps can have on the derived values for @xmath80 and @xmath21 , we repeat the exercise from the above paragraph seven times . in each case , we treat the magnification maps from a different team as the truth and proceed to derive constraints on @xmath80 and @xmath21 using the results from the first four hff clusters . as two of the hff clusters we examine do not have zitrin - nfw magnification maps available , i.e. , macs0717 and macs1149 , we make use of the zitrin - ltm - gauss models instead . the results are presented in table [ tab : lfparm ] .
the faint - end slope @xmath21 results we obtain from the hff clusters alone inhabit the range @xmath114 to @xmath115 depending on which magnification model we treat as reality .
the faint - end slope we estimate averaging the results from all of the models is @xmath116 , while the faint - end slope @xmath21 we find using the parametric models is @xmath117 .
the quoted uncertainty includes median statistical error added in quadrature with the standard deviation among the faint - end slope determinations for the different magnification models .
these results are interesting in that they are consistent with our own results over the field , i.e. , bouwens et al .
( 2015 ) , where @xmath118 , as well as other estimates in the literature ( yan & windhorst 2004 ; bouwens et al .
2007 ; calvi et al .
2013 ; bowler et al .
2015 ; finkelstein et al .
2015 ) which generally lie in the range @xmath119 to @xmath120 .
ccccc + @xmath116.75 & @xmath11.88 & @xmath11.82 & @xmath11.69 & @xmath11.63 + @xmath116.25 & @xmath11.71 & @xmath11.64 & @xmath11.50 & @xmath11.42 + @xmath115.75 & @xmath11.54 & @xmath11.47 & @xmath11.30 & @xmath11.23 + @xmath115.25 & @xmath11.38 & @xmath11.30 & @xmath11.13 & @xmath11.05 + @xmath114.75 & @xmath11.29 & @xmath11.18 & @xmath10.95 & @xmath10.84 + @xmath114.25 & @xmath11.31 & @xmath11.12 & @xmath10.73 & @xmath10.54 + @xmath113.75 & @xmath11.43 & @xmath11.13 & @xmath10.49 & @xmath10.18 + @xmath113.25 & @xmath11.65 & @xmath11.18 & @xmath10.22 & 0.25 + @xmath112.75 & @xmath11.95 & @xmath11.29 & 0.07 & 0.73 + @xmath112.25 & @xmath12.34 & @xmath11.45 & 0.37 & 1.26 + @xmath111.75 & @xmath12.82 & @xmath11.66 & 0.70 & 1.84 + @xmath111.25 & @xmath13.37 & @xmath11.93 & 1.04 & 2.48 + @xmath110.75 & @xmath14.01 & @xmath12.24 & 1.40 & 3.17 + @xmath110.25 & @xmath14.74 & @xmath12.60 & 1.79 & 3.91 + + + @xmath116.75 & @xmath11.85 & @xmath11.79 & @xmath11.66 & @xmath11.60 + @xmath116.25 & @xmath11.67 & @xmath11.60 & @xmath11.46 & @xmath11.39 + @xmath115.75 & @xmath11.50 & @xmath11.42 & @xmath11.27 & @xmath11.20 + @xmath115.25 & @xmath11.34 & @xmath11.27 & @xmath11.11 & @xmath11.04 + @xmath114.75 & @xmath11.36 & @xmath11.21 & @xmath10.90 & @xmath10.75 + @xmath114.25 & @xmath11.57 & @xmath11.27 & @xmath10.65 & @xmath10.35 + @xmath113.75 & @xmath11.93 & @xmath11.42 & @xmath10.36 & 0.16 + @xmath113.25 & @xmath12.43 & @xmath11.66 & @xmath10.04 & 0.75 + @xmath112.75 & @xmath13.08 & @xmath11.99 & 0.31 & 1.42 + @xmath112.25 & @xmath13.88 & @xmath12.40 & 0.68 & 2.18 + @xmath111.75 & @xmath14.81 & @xmath12.90 & 1.09 & 3.02 + @xmath111.25 & @xmath15.88 & @xmath13.48 & 1.52 & 3.95 + @xmath110.75 & @xmath17.10 & @xmath14.17 & 1.98 & 4.95 + @xmath110.25 & @xmath18.46 & @xmath14.93 & 2.47 & 6.04 it is worthwhile emphasizing the value of the test we perform in the previous paragraph .
as we consider the use of searches behind lensing clusters for constraining the faint end of the @xmath10 lfs , it is essential that we derive constraints from the lensing clusters in isolation of those obtained from field searches to verify that no major systematics appear to be present in the lf results from the lensing clusters .
this is relevant , since recent determinations of the faint - end slope @xmath21 to the lfs from field and cluster search results seem to show a substantial discrepancy ( figure 1 ) .
we should emphasize however that the uncertainties on the faint - end slope @xmath21 we quote , i.e. , @xmath70.03 or @xmath70.04 , are the formal statistical errors and likely smaller than the true errors , which also include systematic uncertainties .
one particularly important systematic uncertainty regards the dependence of the lf results on the unknown size distribution , through the incompleteness effects ( e.g. , bouwens et al . 2016 ; grazian et al . 2011 ) . if the sizes of faint galaxies are sufficiently different from point sources such that the completeness is overestimated by a factor of 2 , the faint - end slope we derive would be too shallow by @xmath121 .
+ we now proceed to derive constraints on the overall form of the @xmath10 lf at @xmath0 combining constraints from the field with those available from the hff clusters . for simplicity , we keep the characteristic luminosity @xmath109 fixed to the value @xmath120.94 mag that we found in our earlier wide - area field study ( bouwens et al .
2015 ) , given the lack of substantial information in the hff cluster program for constraining this parameter due to the small volume probed .
in combining constraints from the field and from the hff clusters , we need to allow for some error in the normalization of both the field and hff cluster results , as a result of large - scale structure variations ( `` cosmic variance '' : robertson et al .
2014 ; see also somerville et al .
2004 and trenti & stiavelli 2008 ) and also small systematic errors in the estimates of the volume densities of galaxies in each of our probes .
we assume an uncertainty of @xmath420% in the volume density in both the field and hff lf results .
the 20% uncertainty we assume to be present in the normalization of the lf results at both the bright and faint ends includes a @xmath410% uncertainty in our estimates of the selection volume and @xmath410% systematic error due to uncertainties in the total magnitude measurements ( reflecting a @xmath40.1 mag systematic error in the magnitude measurements ) .
the inclusion of such an error is relevant given the existence of real errors in the estimated volume densities of galaxies using photometric criteria .
small differences appear to be guaranteed , given that the filters available for the selection of galaxies from the field are different ( in particular including a deep `` z''-band filter ) from those available over the hff clusters ( which do not include a deep `` z '' band filter ) .
important factors contributing to these uncertainties are ( 1 ) likely differences between the assumed sizes and seds of galaxies vs. redshift in the observations vs. those in the simulations and ( 2 ) uncertainties in the contamination rate of observed samples.0.1 mag level , as evident looking at the broad range of magnituede measurements in skelton et al .
see e.g. their figures 35 - 36 .
the situation is likely even more challenging for galaxies behind lensing clusters due to the substantial foreground light from the clusters themselves , and in fact the total magnitudes measured with different procedures and by different groups are found to differ at the @xmath40.2 mag to 0.25-mag level . see 6.1.2 .
this translates into normalization differences of @xmath420% to @xmath425% assuming a faint - end slope of @xmath29 for lf results . ]
similar to our lf results using the hff observations alone , we derive confidence regions on the @xmath0 lf results using a mcmc - type procedure where we explore a limited region in parameter space @xmath80-@xmath21-@xmath81 space and calculate the likelihood of each point in parameter space using the forward modeling simulations we describe in 4.1 .
our calculated likelihoods explicitly include a marginalization across the 20% volume density uncertainties we assume .
these likelihoods are then multiplied by the likelihoods on the same schechter parameters derived by bouwens et al .
( 2015 ) at @xmath0 using results from the full candels program , the hudf , and the hudf parallels ( again after marginalizing the bouwens et al .
2015 over @xmath80 to account for a 20% uncertainty in the volume density of sources ) .
finally , this 3d likelihood grid is fit to derive the most likely values for @xmath80 , @xmath21 , and @xmath81 and also the covariance matrix . as in our determinations of the lf parameters from the hff programs alone , we repeat this exercise seven different times , treating each of the magnification models from different teams as reality and recovering the lf results using the median magnification map .
we present our constraints on each of the lf parameters in table [ tab : lfparm ] .
the faint - end slope @xmath21 we estimate averaging all of our models and just the parametric models are @xmath116 and @xmath116 , respectively . as in our determinations using only the hff observations themselves , the faint - end slopes @xmath21 we derive at @xmath0 are fairly consistent with lf results in the field .
our obtaining consistent results for all seven of the magnification model families we consider is not especially surprising , given the fact that individual sources would be expected to scatter in almost the same direction as the dominant slope of the lf , i.e. , @xmath29 ( the expected slope of the lf from scatter ) vs. @xmath122 ( the actual slope of the lf ) . despite general agreement on the most likely value for @xmath21 at @xmath0
, we find a broad range of values for the curvature parameter @xmath81 , from 0.07 to 0.21 , with @xmath84 uncertainties ranging from 0.16 to 0.32 .
none of the magnification models we considered point towards our having even modest evidence , i.e. , @xmath123 , for the @xmath0 lf showing a turn - over at the faint end .
we find the smallest uncertainties on @xmath81 assuming that the glafic magnification models represent reality .
slightly larger uncertainties on @xmath81 are found assuming that the cats , sharon / johnson , and zitrin - nfw models represent reality , while the largest uncertainties on @xmath81 are found , assuming that the zitrin - ltm , bradac , and grale models represent reality .
the size of the uncertainty on @xmath81 is a function of how similar the various magnification models are to the median magnification model formed from the parametric models .
given the similarity of assumptions utilized in the glafic , cats , sharon / johnson , and zitrin - nfw models , it is not surprising that their magnification maps agree best with median magnification maps constructed using similar assumptions .
the results in table [ tab : lfparm ] for the different magnification models indicate the general range of constraints one could obtain on the form of the @xmath0 lf : the results for the non - parametric models indicate the errorbars on the lfs if the mass distribution in clusters do not strictly follow the assumptions made in the parametric models , while results for the parametric models indicate the likely error bars , if the mass profiles in the hff clusters generally do adhere to those assumptions .
ccccc + @xmath124 & 26.13 & 26.17 & 26.25 & 26.28 + @xmath125 & 26.28 & 26.33 & 26.42 & 26.47 + @xmath126 & 26.35 & 26.40 & 26.54 & 26.65 + @xmath127 & 26.36 & 26.42 & 26.93 & 28.65 + @xmath128 & 26.36 & 26.42 & 31.10 & 41.23 + + + @xmath124 & 26.13 & 26.18 & 26.26 & 26.29 + @xmath125 & 26.29 & 26.33 & 26.43 & 26.47 + @xmath126 & 26.33 & 26.39 & 26.57 & 27.00 + @xmath127 & 26.33 & 26.39 & 27.39 & 31.62 + @xmath128 & 26.33 & 26.39 & 34.35 & 55.95 uncertainties in the redshifts of the lensed @xmath0 galaxies also contribute to the error in @xmath81 . to estimate the impact
, we kept the redshifts of lensed background souces fixed while rerunning the forward - modeling simulations from our mcmc chain . comparing the uncertainties we derive in @xmath81 to the uncertainties we derive including errors in the redshift
, we find a typical decrease of 0.01 in the uncertainty on @xmath81 , i.e. , from 0.16 to 0.15 in the case of the glafic simulations .
to help visualize what our present lf results mean , we have made use of our parametrized constraints to derive 68% and 95% confidence intervals on the volume density of galaxies as a function of the @xmath10 luminosity @xmath65 .
these results are presented both in figure [ fig : lf6all ] and also in table [ tab : modelswlf ] .
+ one question that has recently been of significant interest in the literature regards whether there is a flattening or turn - over in the @xmath10 lf at the faint end .
this question is important , since the answer could indicate to us whether there is a sufficient number of extremely faint galaxies to produce the photons necessary for driving the reionization of the universe .
fortunately , using the likelihood contours for @xmath81-@xmath21-@xmath80 and eq .
[ eq : mt ] , we can directly determine the brightest point in the lf where a turn - over is permitted ( with the 68% confidence intervals ) .
the results do depend somewhat on which magnification model we assume to be representative of reality ( and therefore which of the panels we consider from figure [ fig : lf6all ] ) .
nevertheless , we find that the hff observations allow for a turn - over in the lf as bright as @xmath129 mag to @xmath130 mag ( within the 68% confidence intervals ) .
the allowed turn - over luminosities we estimate assuming different magnification models are presented in table [ tab : lfparm ] and also in figure [ fig : turnover ] .
+ besides the lensing models , the unknown sizes of extremely faint @xmath0 galaxies represents a potentially significant uncertainty for constraints on the @xmath0 lf . if a modest fraction of faint , highly - magnified galaxies behind hff clusters had intrinsic half - light radii of @xmath90.06@xmath42 , the successful detection of these sources would be challenging . in a companion paper ( bouwens et al .
2016 ) , we looked into the question of the intrinsic sizes of extremely faint galaxies , taking advantage of the large samples of extremely faint , highly magnified @xmath13 - 3 + @xmath78 - 8 galaxies available behind the first four hff clusters .
we found no evidence to indicate that these galaxies were resolved , looking at ( 1 ) the prevalence of high magnification sources as a function of the predicted shear and ( 2 ) their stacked profile along the expected shear axis .
however , the absence of evidence for extended faint galaxies behind the hff clusters is not itself an indication ( or evidence ) that such extended sources do not exist
. the size distribution could be bimodal or show an extended tail towards large values .
one admittedly adhoc way of accounting for the unknown size distribution is to suppose that our completeness estimates using point sources could be up to a factor of 2 too low , e.g. , as we would derive if @xmath450% of the very faint ( @xmath132 mag ) sources in our samples had sizes of @xmath450 mas or larger .
this contrasts with the other case that we consider , where we have assumed that our completeness estimates , using point sources , are accurate .
given that we do not know which is the case , we repeat our determinations of the lf parameters from the previous section , but now include another step in the process where we marginalize our likelihood estimates over the unknown completeness of our faint sample ( assuming a uniform prior on the completeness lying somewhere between 50% and 100% of the point source case ) .
these results are presented in table [ tab : lfparm ] .
the results are fairly similar to our fiducial results but give faint - end slopes @xmath21 that are @xmath133 - 0.02 steeper . as specified earlier in this section , if we assume that our completeness estimates are a factor of 2 too high and ignore the field results , we derive a faint - end slope which is steeper by @xmath121 . lc @xmath120.75 & [email protected] + @xmath120.25 & [email protected] + @xmath119.75 & [email protected] + @xmath119.25 & [email protected] + @xmath118.75 & [email protected] + @xmath118.25 & [email protected] + @xmath117.75 & [email protected] + @xmath117.25 & [email protected] + @xmath116.75 & [email protected] + @xmath116.25 & [email protected] + @xmath115.75 & 0.0415@xmath134 + @xmath115.25 & 0.0600@xmath135 + @xmath114.75 & 0.0817@xmath136 + @xmath114.25 & 0.1049@xmath137 + @xmath113.75 & 0.1270@xmath138 + @xmath113.25 & 0.1451@xmath139 + @xmath112.75 & 0.1567@xmath140 to determine if galaxies produce enough ionizing photons to drive the reionization of the universe , we require constraints on the luminosity density in the rest - frame @xmath10 that include the contribution from all galaxies
. we can compute confidence intervals on the luminosity densities in the rest - frame @xmath10 by using the constraints from the analyses performed in the previous section and then marginalizing over @xmath80 , @xmath21 , and @xmath81 .
we compute the results to a number of different limiting luminosities @xmath65 , i.e. , @xmath141 mag , @xmath11 mag , @xmath12 mag , @xmath142 mag , and @xmath143 mag , mag , is only included for illustrative value and takes as its inspiration results from oshea et al .
( 2015 ) and ocvirk et al .
( 2016 ) which predict sources to such faint magnitudes . ]
the first four of which commonly appear in the literature , in considering whether galaxies might drive cosmic reionization , particularly including the contribution at very low luminosities .
we have presented the results we obtain in figure [ fig : lumd ] and also in table [ tab : lumd ] to these faint - end limits .
our results imply a luminosity density of @xmath144 ergs s@xmath39 hz@xmath39 mpc@xmath112 to @xmath11 mag . not surprisingly , our lf results allow for essentially an arbitrarily high contribution from very faint galaxies to the @xmath10 luminosity density , particularly including a contribution from galaxies to @xmath143 mag .
these results also provide fairly firm @xmath84 and @xmath145 lower bounds on our results .
we find @xmath84 and @xmath145 lower limits on luminosity density in the rest - frame @xmath10 of @xmath146 ergs s@xmath39 hz@xmath39 mpc@xmath112 and @xmath147 ergs s@xmath39 hz@xmath39 mpc@xmath112 , respectively .
the @xmath84 and @xmath145 lower bounds we find on the luminosity density integrating to arbitrarily faint luminosities is not especially higher than what we find integrating to @xmath11 mag .
these results indicate that it is not yet possible to argue for the discovery of a significant additional reservoir of photons from galaxies faintward of @xmath11 mag ( atek et al . 2015a , b ) , as has been the claim in one recent study ( l16 ) . finally , to conclude this section , we derive a binned representation of the results from the first four hff clusters
. binned representations of the lf results from the hffs should be very reliable at high luminosities , where errors in the magnification maps are smaller .
binned representations retain the advantage that they are a much more model independent probes of the lf shape as a function of luminosity . in our binned representation of the lf
, we adopt bins of width 0.5 mag and determine the binned lf results @xmath148 to be as follows : @xmath149 where @xmath150 is the number of sources in absolute magnitude bin @xmath76 after correcting for the estimated magnification .
we derive the selection volumes @xmath151 in a given magnitude bin using the following equation : @xmath152 where @xmath153 denotes the area , @xmath154 denotes the volume , @xmath155 denotes the estimated completeness , and @xmath30 denotes the magnification factor .
our estimates of the completeness are provided in appendix a ; the completeness @xmath155 appears not to be a strong function of the magnification factor @xmath30 in data sets as deep as the hffs ( see bouwens et al .
2016 ) .
we derived simple stepwise constraints on the @xmath10 lf brightward of @xmath89 mag , by dividing the number of sources in each absolute magnitude bin by the computed selection volume .
the results are presented in table [ tab : swlf ] and figure [ fig : lf6comp ] .
faintward of @xmath89 mag , we include results in table [ tab : swlf ] and figure [ fig : lf6comp ] by taking the geometric mean of the best - fit lf results using the glafic , cats , and sharon / johnson models .
we present those results to luminosities plausibly probed by the present study .
@xmath84 errors on the results at the faint end of the @xmath0 lf are similarly taken to be the geometric mean of the confidence intervals on the lf fits for the same parametric magnification models .
the purpose of this paper is to present new constraints on the form of the @xmath0 lf to low luminosities utilizing new constraints from the first four clusters available from the hff program . before looking into comparisons of our new observational constraints with theory ,
it is useful first to compare the present results with previous results where available to try to understand differences where they might occur .
we first consider a comparison with the most recent results of atek et al .
( 2015 ) who make use of observations available over the first three hff clusters abell 2744 , macs0416 , and macs0717 and selected galaxies using an @xmath33-dropout selection criteria which would identify galaxies from @xmath0 to @xmath19 . a comparison with the most recent determination of the @xmath0 - 7 lf from atek et al .
( 2015 ) is provided in figure [ fig : lf6comp ] . in comparing against the atek et al .
( 2015 ) lf determinations , we incorporate a @xmath40.3-mag brightward shift of the atek et al .
( 2015 ) lf to correct for differences in our apparent magnitude measurements for individual sources . as already noted in one of the companion papers to this study ( bouwens et al .
2016 ) , overall the agreement appears to be quite good , at least insofar as the stepwise points are concerned . the best - fit @xmath80 and luminosity density that atek et al . ( 2015 ) estimate to @xmath11 mag , i.e. , @xmath156 ergs s@xmath39 hz@xmath39 mpc@xmath112 , is @xmath40.18 dex lower than what we find .
this is a small but readily understandable difference that arises because atek et al .
( 2015b ) provide a constraint on the lf at a higher mean redshift than we do , i.e. , @xmath157 vs. @xmath0 , and also include in their determinations results from field surveys , i.e. , candels or the hudf , which probe @xmath19 vs. our @xmath0 probe .
given that the integrated luminosity density to a limit of @xmath141 mag changes by @xmath40.2 dex per unit redshift , our larger luminosity density estimate is entirely expected . to ensure that comparisons with the lf results from l16 were made using a consistent luminosity scheme , we carefully cross - matched sources from our catalogs with those from l16 and comparing our measured apparent magnitudes with those inferred from their study using the tabulated absolute magnitudes , redshifts , and magnification factors .
comparing the total magnitudes we infer for sources using our scaled aperture scheme to the l16-inferred magnitudes , we find a 0.52-mag median difference , with l16-inferred magnitudes fainter than ours , both for relatively bright @xmath158 sources and fainter @xmath159 sources .
if we instead estimate total magnitudes for sources by taking the flux in fixed apertures that would enclose 70% of the flux for point sources , as performed by hudf12 team ( schenker et al .
2013 ; mclure et al .
2013 ) and derive an inverse variance - weighted total magnitude from the @xmath34 , @xmath35 , @xmath36 , and @xmath37 bands , we find differences of 0.25-mag in the median , with the l16-inferred magnitudes being fainter , comparing magnitudes for the faintest sources ( i.e. , @xmath928 mag ) .
the l16 magnitudes show a similar offset relative to the published photometry of atek et al .
( 2015a ) .
given that the hudf12 apparent magnitude measurement scheme should give a fairly conservative lower limit on the total fluxes for individual candidates , these comparisons suggest that l16 systematically underestimate the luminosity of individual sources in their catalog by at least @xmath40.25 mag , if not more ( taking our scaled - aperture magnitudes as the baseline ) . in the next subsection , we present evidence that l16 underestimated the flux in the faintest @xmath0 candidate ( in this case by @xmath41 mag ) .
given this range in values , we adopt a shift of the binned @xmath0 lf of l16 brightward by 0.4 mag to compare volume density measurements at luminosities closer to what we measure . after doing so , we find that the l16 stepwise results appear to be a factor of @xmath160 - 4@xmath3 higher than our own results in the luminosity range @xmath141 to @xmath161 mag ( figure [ fig : lf6comp ] ) . after considering different explanations for these differences
, it would appear that they are due to the large intrinsic half - light radii that l16 assume ( median of 0.09@xmath42 ) . in bouwens et al .
( 2016 ) , we demonstrated through extensive simulations that such size assumptions would result in much higher inferred volume densities for sources
( by factors of @xmath95 ) than if smaller sizes ( i.e. , @xmath1810 mas ) were assumed . see figure 2 from bouwens et al .
( 2016 ) . with our recent paper ( bouwens et al .
2016 ; see also kawamata al .
2014 ; laporte et al .
2016 ) , it is now clear that the use of such large sizes is not realistic .
several lines of evidence indicate that the intrinsic half - light radii of very low luminosity sources ( i.e. , @xmath616 mag ) are very small , i.e. , @xmath1620.03@xmath42 , with many sources appearing to have sizes @xmath1620.01 ( bouwens et al .
small sizes imply small incompleteness corrections and therefore much lower volume densities for faint galaxies .
it is important to verify that the sizes assumed in selection volume simulations are realistic , given how sensitive the selection volume estimates ( and hence lfs ) for faint galaxies are to this issue ( see figure 11 of bouwens et al .
2016 ) .
the very high volume densities l16 find for @xmath163 mag galaxies in their @xmath0 lf likely impacted the analyses they performed regarding a possible turn - over at the faint end , as we explain in 6.2 .
it would be very difficult for l16 to find evidence for a brighter turn - over being allowed , if one considered lf results that included an apparently steepening slope at @xmath141 mag ( see dotted feature in figure [ fig : lf6comp ] ) .
for similar reasons , the present constraints on the faint - end slope @xmath21 , i.e. , @xmath164 are shallower than obtained by l16 . given that the constraints can potentially depend on the gravitational lensing model assumed , it is perhaps best to compare faint - end slopes assuming the same gravitational lensing model . if we take their formal constraints as measured assuming the glafic gravitational lensing model , our estimated faint - end slope @xmath21 is @xmath116 vs. the l16 faint - end slope at @xmath0 of @xmath165 , a difference of 3.5@xmath8 .
the actual differences between our faint - end slope @xmath21 estimates and l16 s estimates are likely even larger than 3.5@xmath8 , given that l16 combine their hff measurements with the finkelstein et al .
( 2015 ) field constraints which prefer @xmath166 .
re - estimating the faint - end slope @xmath21 from the l16 hff results alone , we infer faint - end slopes of @xmath167 to @xmath168 ( see appendix d here or 6.2 of bouwens et al . 2016 ) .
this is significantly ( @xmath1694.5@xmath8 ) steeper than our mean estimate using the hff data alone @xmath116 .
one issue we examined in this study concerned the existence of a possible turn - over in the @xmath0 @xmath10 lf at the faint end .
since we could not find compelling evidence for a turn - over at the faint end of the lf ( 5.3 ) , we proceeded to examine what constraints we can place on the presence of a turn - over as well as the luminosity at which a turn - over could occur . using our search results
, we concluded that any possible turn - over in the lf would need to occur at @xmath170 mag or @xmath171 mag ( within the 68% confidence intervals ) , depending on the assumptions we made about errors in the magnification maps .
the present conclusions seem fairly similar to the findings of castellano et al .
( 2016b ) , who had previously concluded that the @xmath10 lf at @xmath0 appeared unlikely to show a turn - over at @xmath172 mag ( 68% confidence ) , after factoring in the uncertainties in the magnification maps .
l16 had concluded based on their analysis of @xmath0 galaxies behind the first two hff clusters that they have `` positive '' and `` strong '' evidence that any turn - over in the lf must be fainter than @xmath173 mag and @xmath5 mag , respectively , using the criteria @xmath174 and @xmath175 , respectively ( where we draw these numbers from their figure 12 ) .
bic denotes the bayesian information criterion ( schwarz 1978 ) .
given their application of bic , bic is effectively just equal to @xmath176 , and the above limits on the turn - over luminosity translate to nominal confidence levels of 84% and 98.5% , respectively . on the basis of the l16 fit results assuming different magnification models , l16 reported an uncertainty in their allowed turn - over luminosity of @xmath177 mag and @xmath178 mag , respectively . while our conclusions regarding the lack of a bright turnover do not impact the l16 claim , there are other reasons for being concerned about the validity of their turn - over constraints and their claim that it can not occur at brighter magnitudes .
l16 claimed evidence against a turn - over in the @xmath0 lf 3.1 mag fainter than what we do ( taking the glafic - vs .- median magnification map uncertainties to be typical ) , despite their examining just half the present data set and with a smaller @xmath0 sample .
how could they claim stronger constraints ?
remarkably , we believe the reason lies not in the faintest source they report , but instead in the very high volume densities they estimate for the @xmath10 lf in the range @xmath141 to @xmath11 ( which lie in significant excess of our own determinations , by factors of @xmath160 - 4 : see figure [ fig : lf6comp ] ) . when compared to brighter points in the lf , the volume density of sources they report over the luminosity range @xmath141 to @xmath11 is sufficiently high ( and possessing such small error bars ) as to suggest a lf form which steepens further at lower luminosities , i.e. , @xmath179 , rather than one which retains a fixed faint - end slope and then flattens towards lower luminosities , i.e. , @xmath95 ( using our formalism ) . given the inclusion of these points , with a `` concave up '' feature at @xmath180 mag ( _ indicated in figure [ fig : lf6comp ] with the short dotted line feature _ ) , in their lf fits , we suspect it would be very difficult indeed for l16 to find evidence for a turn - over at intermediate luminosities ( since all the `` evidence '' would seem to point in the opposite direction ) . given the statistical weight of this feature at @xmath180 to @xmath11 mag and the points brighter than @xmath141 mag , l16 would find that they needed to probe very faint indeed to find a luminosity where a turn - over was allowed .
even if we ignore such issues , l16 present a second piece of evidence to argue against a possible turn - over at @xmath26 mag and this involves a @xmath0 candidate reported to be at @xmath181 mag and on whose basis they estimate a volume density of @xmath46 galaxies per mpc@xmath182 at @xmath183 mag .
if robust , this candidate could also argue against a turn - over in the lf at higher luminosities .
this @xmath0 candidate from l16 is also detected in our catalogs .
however , it fails to meet our selection criteria , due to the fact that the integrated likelihood of the candidate lying at @xmath47 is @xmath450% and therefore we do not include it in our @xmath0 sample .
however , the source may still correspond to a @xmath0 galaxy and therefore it is worthwhile to consider .
on the basis of our own photometry , we estimate the source to have a total magnitude of 28.9 mag . using the magnification factor of 60.6@xmath184 l16 estimate for the source based on current magnification models which are publicly available and a redshift of @xmath0
, we estimate an absolute magnitude of @xmath185 mag , which is @xmath41 mag brighter than what l16 estimate for the same source . ,
magnification factor 60.6 , and absolute magnitude measurement @xmath55 for this source ) is 29.9 mag , i.e. , corresponding to a @xmath145 detection in the data .
since this is @xmath40.6 mag fainter than this source could have if the source was a reliable @xmath95@xmath8 detection in the observations , which would be implied by their requirement for 3.5@xmath8 detections in two wfc3/ir bands , the absolute magnitude l16 report appears to be incorrect .
the correct magnitude for this source must be @xmath40.6 - 0.7 mag brighter than what l16 report , i.e. , @[email protected] mag .
] while it is unclear why our own magnitude estimates are in such large tension with those from l16 , we can nevertheless examine the implications this source would have for our @xmath10 lf results if it is indeed at @xmath0 .
taking this single source and dividing by the selection volume in the range @xmath186 , we estimate a volume density of @xmath187 mpc@xmath112 mag@xmath39 .
interestingly enough , this is completely consistent , as illustrated in figure [ fig : faintz6 ] ( _ left panel _ ) , with the lf constraints we find at @xmath188 mag treating the glafic magnification model as reality and recovering the lf results using the median magnification model .
it is also consistent ( _ right panel _ to figure [ fig : faintz6 ] ) with the lf results we derive using the other magnification models , e.g. , zitrin - ltm , which suggest a magnification factor of 19 instead of the median magnification factor of @xmath460.6 found by l16 and which imply volume densities of @xmath40.06 mpc@xmath112 mag@xmath39 .
it is not surprising that we recover lower volume densities than recovered by l16 , given our use of smaller sizes for the faint @xmath92 mag population as now appears to be appropriate ( bouwens et al .
smaller sizes translate into a higher estimated completeness and selection volume .
we also checked that the inclusion of this source only had a moderate impact on the allowed luminosity of a turn - over .
if we include this source in the observed counts , we find our constraints on a possible turn - over in the @xmath10 lf change by 0.8 mag ( becoming fainter ) assuming that glafic - vs .- median model are typical of the true magnification errors and by @xmath40.4 mag ( becoming fainter ) assuming that the bradac - vs .- median or grale - vs .- median models are more typical of the errors . if we ask which luminosities we can exclude for a turn - over at 95% confidence , the excluded luminosities are @xmath189 mag for the glafic - vs .- median case ( 2.7 mag brighter than what l16 report for this limit ) .
as we have seen in this section , it is not possible at present to definitively rule out the existence of a turn - over in the @xmath10 lf at @xmath26 mag and especially at @xmath190 mag on the basis of current observational data nor , in light of the new results presented in this paper , could l16 actually do so from their data .
the observational constraints we have obtained here are obviously of great value for comparing against the predictions for the form of the lf at the faint end , as provided by many different teams using simulations , theoretical models , and on the basis of observations of the nearby universe .
we compare our lf constraints with the following cosmological simulation or theoretical model results : + _ renaissance [ oshea et al .
2015 ] : _ oshea et al .
( 2015 ) report some of the first results from the `` renaissance '' simulations .
the `` renaissance '' simulations are zoom - in simulations of a @xmath191 volume of the universe , powered by the enzo code ( bryan et al .
2014 ) , and self - consistently following the evolution of gas and dark matter , including @xmath192 formation and destruction from photodissociation .
star formation and supernovae physics are included and ionizing and uv radiation are produced as predicted by starburst99 ( leitherer et al .
individual dark - matter particles in the simulations have masses of @xmath193 @xmath194 , meaning that the smallest halos that are resolved in the simulation are @xmath195 @xmath196 ( @xmath470 particles / halo ) .
many details of the physical implementation of the implementation of the physics and also sub - grid recipes are provided in xu et al .
( 2013 , 2014 ) and chen et al .
( 2014 ) . in the `` renaissance '' simulations , flattening in the uv lf is a direct result of the decreasing fraction of baryons converted to stars in the lowest mass halos , due to the impact of radiative feedback and less efficient cooling processes . while it is not yet possible to follow the results of these simulations to @xmath0 , results are available at @xmath197 and this is the redshift we use for comparisons .
+ _ coda [ ocvirk et al .
2016 ] : _ the cosmic dawn ( coda ) simulations are full gravity + hydrodynamic simulations of a large @xmath198 volume of the universe using the ramses code ( teyssier 2002 ) .
the simulations include standard prescriptions for star formation and supernovae explosions following standard recipes ( ocvirk et al .
2008 ; governato et al .
2009 , 2010 ) .
one new feature of the coda simulations is the inclusion of radiative transfer into the simulations , in the sense that hydrodynamics and radiative transfer are now fully coupled . as a result , the effects of photoionization heating on low - mass galaxies are fully included in the coda simulations .
ocvirk et al .
( 2016 ) report that radiative feedback plays a big role in suppressing star formation in low mass galaxies and modulating the faint - end of the lf .
+ _ croc [ gnedin 2014 , 2016 ] : _ the lf results for the cosmic reionization on computers ( croc ) are based on gravity + hydrodynamical simulations using adaptive refinement treement ( art ) code ( kratsov 1999 ; kravtsov 2002 ; rudd et al .
the croc simulations include a wide variety of physics processes , including gas cooling and heating processes , molecular hydrogen chemistry , star formation , stellar feedback , radiative transfer of ionizing and uv light from stars .
these simulations are conducted in 20h@xmath39 mpc boxes at a variety of resolutions .
the effective slope of croc lfs continue to flatten towards fainter magnitudes and reach a peak at @xmath199 mag .
however , the peak at @xmath199 mag is reported not to be a robust prediction of the simulation and to depend on the minimum particle size in the simulations .
the impact of radiative feedback is less important in the croc simulations than in coda .
+ _ finlator et al .
( 2015 , 2016a , 2016b ) : _ the finlator et al .
( 2015 , 2016a , 2016b ) lf results are based on a cosmological simulation of galaxy formation in a @xmath200 mpc@xmath182 volume of the universe including both gravity and hydrodynamics as implemented in the gadget-3 code ( springel 2005 ) . to this code
, gas cooling is added through collisional excitation of hydrogen and helium as in katz et al .
( 1996 ) , and metal line cooling is implemented using the collisional ionization equilibrium tables of sutherland & dopita ( 1993 ) . star formation is added using the kennicutt - schmidt law , with supernovae feedback included following the ezw prescription from dav et al .
( 2013 ) and metal enrichment from supernovae as implemented as in oppenheimer & dav ( 2008 ) .
flattening in the finlator et al .
( 2015 , 2016a , 2016b ) lfs occurs mostly due to less efficient gas cooling at lower halo masses .
+ _ dragons [ liu et al .
2016a ] : _ the lf results from liu et al .
( 2016a ) are based on the dark - ages reionization and galaxy - formation observables from numerical simulations ( dragons ) project which build semi - numerical models of galaxy formation on top of halo trees derived from n - body simulations done over different box sizes to probe a large dynamical range .
the semi - numerical models include gas cooling physics , star formation prescriptions , feedback and merging prescriptions , among other components of the model . the turn - over in the lf results of liu et al .
( 2016a ) at @xmath199 mag correspond to the approximate halo masses @xmath201
@xmath194 where the gas temperature is @xmath202 k. above this temperature , atomic cooling processes become efficient . in earlier data sets ,
muoz & loeb ( 2011 ) had looked at what constraints could be placed on this mass using earlier lfs of bouwens et al .
. + _ jaacks et al .
( 2013 ) : _ the simulation results in jaacks et al .
( 2013 ) are powered by the gadget-3 ( springel 2005 ) gravity+hydrodynamics code run in three box sizes ( 10 , 33.75 , and 100 @xmath203 mpc ) and three different particle sizes ( 9@xmath310@xmath82 , 2@xmath310@xmath204 , and 3@xmath310@xmath205 h@xmath39 @xmath194 ) .
radiation cooling is included in the simulations by h , he , and metals ( choi & nagamine 2009 ) , a uv background self - shielding effect , and heating by a uniform uv background ( faucher - gigure et al .
supernovae feedback is implemeneted by a momentum - drive wind model ( choi & nagamine 2011 ) .
star formation in the simulations is governed by sfr - vs-@xmath192 model of krumholtz et al .
( 2009 ) rather than in terms of the total density in cold gas .
the implementation of this in the gadget-3 code is as described in thompson et al .
( 2013 ) and is similar to the implementation of the same recipe by kuhlen et al .
( 2013 ) in the enzo code . as a result of the lower gas density in molecular hydrogen @xmath192 in fainter , lower - mass galaxies , the lfs predicted by jaacks et al .
( 2013 ) show a turn - over at @[email protected] mag .
+ _ dayal et al . (
2014 ) : _ the lf results from dayal et al .
( 2014 ) are based on a semi - analytic model which follows the evolution of galaxies in merger tree constructed from extended press - schechter theory ( lacey & cole 1993 ) . the star formation rate in individual proceed at such a rate as to balance the impact of supernovae feedback in expelling all the gas from a galaxy .
flattening in the @xmath10 lf is partially the result of a similar flattening in the halo mass function , as well as lower efficiency for star formation in the lower - mass halos that contribute to the low luminosity end of the lf .
+ _ yue et al . (
2016 ) : _ yue et al .
( 2016 ) derive their lf results assuming a non - evolving stellar mass - halo mass relation .
yue et al .
( 2016 ) adopt a very similar approach to what mason et al .
( 2015 ) employ in predicting the galaxy lf ( see also trenti et al .
2010 and tacchella et al .
yue et al .
( 2016 ) start with the halo mass function , break up the star formation history of each halo into segments according to which the halo grows in mass by a factor of two , and then assume that the sfr must be such to maintain a constant stellar mass - halo mass relation which they calibrate to the @xmath78 lf of bouwens et al .
yue et al .
( 2016 ) then look into the impact that radiative feedback could have during the epoch of reionization .
yue et al .
( 2016 ) derive their lf results assuming that halos below some fixed circular velocity would have their star formation quenched .
+ finally , we also include a comparison with the empirical results on the faint end of the lf at high redshift : + _ boylan - kolchin et al .
( 2015 ) : _ boylan - kolchin et al .
( 2015 ) arrive at constraints on the faint - end of the lf at @xmath19 by leveraging deep probes of the color - magnitude relationship of nearby dwarf galaxies which allow one to estimate the luminosity of these sources at @xmath19 . by comparing the distribution of inferred luminosities of these dwarfs with the expected numbers extrapolating @xmath19 lfs to @xmath142 mag , boylan - kolchin et al .
( 2015 ) infer a break in the lf at @xmath25 mag from a faint - end slope of @xmath29 to @xmath207 .
+ we present comparisons of the predicted lfs from both sophisticated hydrodynamical simulations , various semi - analytic models , and the empirical results of boylan - kolchin et al .
( 2015 ) in figures [ fig : theory ] and [ fig : theory2 ] .
overall , we find reasonable agreement between our observational results and the predicted lfs from both hydrodynamic simulations and various semi - empirical theoretical models .
the predicted lfs from the coda simulation ( ocvirk et al .
2016 ) and from two semi - analytical models yue et al .
( 2016 ) and dayal et al .
( 2016 ) against which we compare fall slightly below our observational constraints at @xmath208 mag by @xmath40.2 - 0.3 dex , but otherwise are in reasonable agreement with our results .
in particular , we find that our observational constraints allow for the existence of a flattening or turn - over in the @xmath0 lf at @xmath615 mag as predicted in the theoretical models , due to a variety of physical processes , including a greater role for radiative feedback ( oshea et al . 2015 ; ocvirk et al . 2016 ; yue et al .
2016 ) and less efficient cooling in lower mass halos where atomic cooling processes would be less important ( wise et al . 2014
; gnedin 2016 ; liu et al .
the present results suggest that these physical processes can impact the shape of the lf at @xmath75 mag , as is predicted , and there is no fundamental disagreement with observational results to @xmath209 mag ( contrary to reports from l16 ) .
our observational results are also fully consistent with the abundance matching constraints obtained by boylan - kolchin et al .
( 2015 ) which suggest a break in the faint end slope of the lf at @xmath113 mag .
the toy lf from boylan - kolchin et al .
( 2015 ) is almost entirely contained with our 68% confidence intervals , suggesting that the flattening they infer from analyses of nearby dwarf galaxies is fully consistent with the hff observations of faint @xmath0 galaxies .
here we have made use of a large selection of 160 lensed @xmath0 galaxies from the first four hff clusters to constrain the shape of the @xmath0 uv lf to very low luminosities .
not only does the present sample of lensed @xmath0 galaxies represent the most comprehensive such sample to date , but the present sample probes to equivalently low luminosities as other work ( if we account for the fact that others report luminosities up to 0.7 - 1.0 mag fainter than what we find for the same sources [ though @xmath40.5 mag is more typical ] : see 2 and 6.1-6.2 ) .
our analysis provides a much more realistic assessment of the impact of the large magnification uncertainties inherent at high magnifications .
this allows us to set improved constraints on the faint - end slope @xmath21 and also to investigate whether the @xmath10 lf shows a turn - over at very low luminosities .
one particular emphasis of this analysis was to include a full account of systematic errors in deriving accurate constraints on the shape of the @xmath0 @xmath10 lf .
we looked especially at the impact of errors in the magnification maps , but we also considered the impact of uncertainties in the estimated completeness based on the size distribution building upon results in a companion paper ( bouwens et al .
2016 ) . to explore the impact of errors in the magnification map on lf results , we have developed a new forward modeling approach which involves using one set of magnification models and a candidate lf to create mock catalogs over each of the hff clusters .
these catalogs are then analyzed using the same type of magnification models as are used to interpret the real observations ( figure [ fig : flowchart ] ) .
the likelihood of a given lf can then be assessed by comparing the observed counts with the expected counts derived from the simulations .
our quantification of the general form of the @xmath0 lf includes not only the normal schechter parameters , but also a curvature parameter @xmath81 which we apply faintward of @xmath116 mag to characterize the form of the lf to very low luminosities .
our new simulation results using forward modeling demonstrate the basic impact of magnification errors on the lf results .
we show that scatter due to magnification errors results in the lf showing a faint - end slope of @xmath4@xmath57 or steeper in the very low luminosity regime where the magnification factors are high @xmath31 - 30 .
this effect is so pervasive that it can eliminate any indication of a turn - over ( even if present in reality ) at the faint end of the lf ( see figure [ fig : illust ] ) . at lower magnification factors ,
i.e. , @xmath210 , where the predictive power of the magnification models is best ( e.g. , see figure [ fig : predpow ] and also meneghetti et al .
2016 ) , the impact appears to be most manageable in terms of the overall impact on the lf results . for higher magnification factors ,
i.e. , at @xmath31 and especially at @xmath211 , the predictive power of the magnification models is much poorer , resulting in large uncertainties in the magnification factors . as a result
, it can be difficult to determine whether the lf shows a turn - over at @xmath20615 mag , whether it steepens further at @xmath20615 mag , or whether it continues with a fixed faint - end slope to @xmath5 mag ( see figure [ fig : illust ] ) .
taking advantage of our forward modeling procedure , we derive new constraints on the faint - end slope of the lf and arrive at a value of @xmath116 using the hff observations alone ( table [ tab : lfparm ] ) and @xmath116 combining our constraints with the field results of bouwens et al .
both constraints are consistent with our previous determination @xmath212 ( bouwens et al . 2015 ) using the hudf+hudf - parallel+candels data alone .
we nevertheless caution that the faint - end slope could be up to @xmath121 steeper if our search for faint @xmath0 galaxies is @xmath131 more incomplete than we assume .
this is a direct result of the unknown size distribution of extremely faint galaxies .
we use our new constraints to derive 68% and 95% confidence regions on the faint - end form of the @xmath0 lf , presenting our results in figure [ fig : lf6all ] .
we find no evidence for a turn - over in the lf at the faint end .
nevertheless , we can place constraints on how faint it must be , though the result does depend on the assumed size of the errors in the magnification models . if the true errors in the models are similar to the differences between the glafic model and the median parametric model ,
our results strongly indicate that a turn - over can not occur brightward of @xmath129 mag ( 68% confidence ) .
however , if differences between the non - parametric models and the median parametric models are typical , then a turn - over can not occur brightward of @xmath130 mag ( 68% confidence ) .
our results are fully consistent with recent observational results from boylan - kolchin et al .
( 2015 ) and theoretical models ( oshea et al .
2015 ; gnedin 2016 ; liu et al .
2016a ; ocvirk et al .
2016 ) predicting some flattening in the @xmath10 lfs at @xmath615 mag .
the faint - end slope @xmath21 we derive at @xmath0 is @xmath116 and 3.5@xmath8 shallower than the l16 faint - end slope @xmath213 ( statistical error only ) .
meanwhile , our constraints on the turn - over are consistent with the findings by atek et al .
( 2015 ) and castellano et al .
( 2016b ) , but are weaker than what l16 report . despite having larger samples than l16 and considering twice as many hff clusters ( while probing to comparably low luminosities ) , we only find evidence against a turn - over brightward of @xmath214 and @xmath215 mag at 68% confidence , vs. the @xmath216 mag reported by l16 at nominally slightly higher confidence .
we speculate that l16 s stronger claims against a turn - over ( and steeper faint - end slope results ) arose as a result of artifacts in their determinations of the lfs resulting from likely inaccurate size assumptions ( see 6.1 - 6.2 and bouwens et al .
2016 ) and also a possible bias in their measurements of the total luminosities of sources ( 6.1.2 ) .
the new formalism we have developed to derive lf results in the presence of errors in the magnification map has significant utility and can be applied to other _
_ observations that have been obtained with the hff program . in the immediate future ,
we plan to make use of our new forward - modeling methodology to derive lf results at @xmath0 , @xmath19 , @xmath45 , and @xmath217 from the full hff program .
these results would provide us perhaps our most complete information on the faint - end form of the lfs before jwst and provide us with clues as to how the overall ionizing emissivity evolves with cosmic time .
we acknowledge many useful conversations on the algorithms presented in this paper with marijn franx .
we are greatly appreciative to pratika dayal , kristian finlator , nicholas gnedin , chuanwu liu , brian oshea , pierre ocvirk , and bin yue for sending us the predictions they derive for the lf results at high redshift .
we thank kristian finlator for discussing with us at length different turn - over mechanisms that would impact the @xmath10 lf at low luminosities .
austin hoag kindly sent us the v2 bradac models for hff cluster macs0416 .
rse acknowledges financial support from the european research council under advanced grant fp7/669253 .
alavi , a. , siana , b. , richard , j. , et al .
2014 , , 780 , 143 alavi , a. , siana , b. , richard , j. , et al .
2016 , , in press , arxiv:1606.00469 atek , h. , richard , j. , kneib , j .-
p . , et al .
2014 , , 786 , 60 atek , h. , richard , j. , kneib , j .-
, et al . 2015a , , 800 , 18 atek , h. , richard , j. , jauzac , m. , et al .
2015b , , 814 , 69 beckwith , s. v. w. , stiavelli , m. , koekemoer , a. m. , et al .
2006 , , 132 , 1729 bertin , e. and arnouts , s. 1996 , , 117 , 39 bouwens , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , , 670 , 928 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al . 2012 , , 754 , 83 bouwens , r. , bradley , l. , zitrin , a. , et al . 2014b , , 795 , 126 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2015 , , 803 , 34 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2015 , , 811 , 140 bouwens , r. 2016 , understanding the epoch of cosmic reionization : challenges and progress , 423 , 111 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2016 , , submitted , arxiv:1608.00966 bowler , r. a. a. , dunlop , j. s. , mclure , r. j. , et al .
2015 , , 452 , 1817 boylan - kolchin , m. , bullock , j. s. , & garrison - kimmel , s. 2014 , , 443 , l44 boylan - kolchin , m. , weisz , d. r. , johnson , b. d. , et al .
2015 , , 453 , 1503 dunlop , j. s. , mclure , r. j. , et al .
2015 , , 452 , 1817 brada , m. , treu , t. , applegate , d. , et al .
2009 , , 706 , 1201 bryan , g. l. , norman , m. l. , oshea , b. w. , et al .
2014 , , 211 , 19 castellano , m. , amorn , r. , merlin , e. , et al .
2016a , , 590 , a31 castellano , m. , yue , b. , ferrara , a. , et al . 2016b , , 823 , l40 chen , p. , wise , j. h. , norman , m. l. , xu , h. , & oshea , b. w. 2014 , , 795 , 144 choi , j .- h . , & nagamine , k. 2009 , , 393 , 1595 choi , j .- h . , & nagamine , k. 2011 , , 410 , 2579 coe , d. , bradley , l. , & zitrin , a. 2015 , , 800 , 84 dav , r. , katz , n. , oppenheimer , b. d. , kollmeier , j. a. , & weinberg , d. h. 2013 , , 434 , 2645 dayal , p. , ferrara , a. , dunlop , j. s. , & pacucci , f. 2014 , , 445 , 2545 diego , j. m. , broadhurst , t. , molnar , s. m. , lam , d. , & lim , j. 2015 , , 447 , 3130 duncan , k. , & conselice , c. j. 2015 , , 451 , 2030 dunlop , j. s. , rogers , a. b. , mclure , r. j. , et al . 2013 , , 432 , 3520 ellis , r. s. , mclure , r. j. , dunlop , j. s. , et al . 2013 , , 763 , l7 faucher - gigure , c .- a . , lidz , a. , zaldarriaga , m. , & hernquist , l. 2009 , , 703 , 1416 finkelstein , s. l. , papovich , c. , salmon , b. , et al .
2012 , , 756 , 164 finkelstein , s. l. , ryan , r. e. , jr .
, papovich , c. , et al . 2014 , arxiv e - prints finlator , k. , thompson , r. , huang , s. , et al .
2015 , , 447 , 2526 finlator , k. , oppenheimer , b. d. , dav , r. , et al . 2016a , , 459 , 2299 finlator , k. , prescott , m. k. m. , oppenheimer , b. d. , et al . 2016b ,
, in press , arxiv:1609.06348 giallongo , e. , grazian , a. , fiore , f. , et al .
2015 , , 578 , a83 gnedin , n. y. 2014 , , 793 , 29 gnedin , n. y. 2016 , , 825 , l17 grazian , a. , castellano , m. , koekemoer , a. m. , et al .
2011 , , 532 , a33 grogin , n. a. , kocevski , d. d. , faber , s. m. , et al .
2011 , , 197 , 35 illingworth , g. d. , magee , d. , oesch , p. a. , et al .
2013 , , 209 , 6 ishigaki , m. , kawamata , r. , ouchi , m. , et al .
2015 , , 799 , 12 jaacks , j. , thompson , r. , & nagamine , k. 2013 , , 766 , 94 jauzac , m. , jullo , e. , eckert , d. , et al . 2015a , , 446 , 4132 jauzac , m. , richard , j. , jullo , e. , et al .
2015b , , 452 , 1437 johnson , t. l. , sharon , k. , bayliss , m. b. , et al .
2014 , , 797 , 48 jullo , e. , & kneib , j .-
2009 , , 395 , 1319 katz , n. , weinberg , d. h. , & hernquist , l. 1996 , , 105 , 19 kawamata , r. , ishigaki , m. , shimasaku , k. , oguri , m. , & ouchi , m. 2015 , , 804 , 103 kawamata , r. , oguri , m. , ishigaki , m. , shimasaku , k. , & ouchi , m. 2016 , , 819 , 114 koekemoer , a. m. , faber , s. m. , ferguson , h. c. , et al .
2011 , , 197 , 36 kravtsov , a. v. 1999 , ph.d .
thesis , 5564 kravtsov , a. v. , klypin , a. , & hoffman , y. 2002 , , 571 , 563 kravtsov , a. v. 2013 , , 764 , l31 kron , r. g. 1980 , , 43 , 305 krumholz , m. r. , mckee , c. f. , & tumlinson , j. 2009 , , 699 , 850 krumholz , m. r. , & dekel , a. 2012 , , 753 , 16 kuhlen , m. , & faucher - gigure , c .- a .
2012 , , 423 , 862 kuhlen , m. , madau , p. , & krumholz , m. r. 2013 , , 776 , 34 lacey , c. , & cole , s. 1993 , , 262 , 627 laporte , n. , infante , l. , troncoso iribarren , p. , et al .
2016 , , 820 , 98 leitherer , c. , schaerer , d. , goldader , j. d. , et al . 1999 , , 123 , 3 liesenborgs , j. , de rijcke , s. , & dejonghe , h. 2006 , , 367 , 1209 liu , c. , mutch , s. j. , angel , p. w. , et al .
2016a , , 462 , 235 liu , c. , mutch , s. , poole , g. , et al .
2016b , arxiv:1608.00819 livermore , r. , finkelstein , s. , lotz , j. 2016 , apj , submitted , arxiv:1604.06799 lotz , j. m. , koekemoer , a. , coe , d. , et al .
2016 , arxiv:1605.06567 madau , p. , & haardt , f. 2015 , , 813 , l8 maizy , a. , richard , j. , de leo , m. a. , pell , r. , & kneib , j. p. 2010
, , 509 , a105 mason , c. a. , trenti , m. , & treu , t. 2015 , , 813 , 21 mcgreer , i. d. , jiang , l. , fan , x. , et al .
2013 , , 768 , 105 mclure , r. j. , dunlop , j. s. , bowler , r. a. a. , et al .
2013 , , 432 , 2696 meneghetti , m. , natarajan , p. , coe , d. , et al . 2016 , arxiv:1606.04548
merlin , e. , amorn , r. , castellano , m. , et al .
2016 , , 590 , a30 merten , j. , meneghetti , m. , postman , m. , et al .
2015 , , 806 , 4 [ mitra et al.(2015)]2015mnras.454l .. 76 m mitra , s. , choudhury , t. r. , & ferrara , a. 2015 , , 454 , l76 mitra , s. , choudhury , t. r. , & ferrara , a. 2016 , arxiv:1606.02719 muoz , j. a. , & loeb , a. 2011 , , 729 , 99 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 nestor , d. b. , shapley , a. e. , kornei , k. a. , steidel , c. c. , & siana , b. 2013 , , 765 , 47 ocvirk , p. , gillet , n. , shapiro , p. r. , et al .
2015 , in press , arxiv:1511.00011 oesch , p. a. , bouwens , r. j. , illingworth , g. d. , et al .
2015 , , 808 , 104 oguri , m. 2010 , , 62 , 1017 oppenheimer , b. d. , & dav , r. 2008 , , 387 , 577 oshea , b. w. , wise , j. h. , xu , h. , & norman , m. l. 2015 , , 807 , l12 parsa , s. , dunlop , j. s. , mclure , r. j. , & mortlock , a. 2016 , , 456 , 3194 peng , c. y. , ho , l. c. , impey , c. d. , & rix , h .- w .
2002 , , 124 , 266 planck collaboration , ade , p. a. r. , aghanim , n. , et al .
2015 , arxiv:1502.01589 [ pc15 ] priewe , j. , williams , l. l. r. , liesenborgs , j. , coe , d. , & rodney , s. a. 2016 , , submitted , arxiv:1605.07621 richard , j. , jauzac , m. , limousin , m. , et al .
2014 , , 444 , 268 robertson , b. e. , furlanetto , s. r. , schneider , e. , et al .
2013 , , 768 , 71 robertson , b. e. , ellis , r. s. , dunlop , j. s. , et al .
2014 , , 796 , l27 robertson , b. e. , ellis , r. s. , furlanetto , s. r. , & dunlop , j. s. 2015 , , 802 , l19 rodney , s. a. , patel , b. , scolnic , d. , et al .
2015 , , 811 , 70 rogers , a. b. , mclure , r. j. , dunlop , j. s. , et al .
2014 , , 440 , 3714 rudd , d. h. , zentner , a. r. , & kravtsov , a. v. 2008 , , 672 , 19 schenker , m. a. , robertson , b. e. , ellis , r. s. , et al .
2013 , , 768 , 196 schwarz , g. , 1978 , the annals of statistics , 6 , 461 sebesta , k. , williams , l. l. r. , mohammed , i. , saha , p. , & liesenborgs , j. 2016 , , 461 , 2126 shibuya , t. , ouchi , m. , & harikane , y. 2015 , , 219 , 15 siana , b. , teplitz , h. i. , ferguson , h. c. , et al .
2010 , , 723 , 241 siana , b. 2013 , hst proposal , 13389 siana , b. 2015 , hst proposal , 14209 siana , b. , shapley , a. e. , kulas , k. r. , et al .
2015 , , 804 , 17 skelton , r. e. , whitaker , k. e. , momcheva , i. g. , et al .
2014 , , 214 , 24 somerville , r. s. , lee , k. , ferguson , h. c. , et al .
2004 , , 600 , l171 springel , v. 2005 , , 364 , 1105 sutherland , r. s. , & dopita , m. a. 1993 , , 88 , 253 tacchella , s. , trenti , m. , & carollo , c. m. 2013 , , 768 , l37 teyssier , r. 2002 , , 385 , 337 thompson , r. , nagamine , k. , jaacks , j. , & choi , j .- h . 2014 , , 780 , 145 trenti , m. , & stiavelli , m. 2008 , , 676 , 767 trenti , m. , stiavelli , m. , bouwens , r. j. , et al .
2010 , , 714 , l202 vanzella , e. , guo , y. , giavalisco , m. , et al .
2012 , , 751 , 70 vanzella , e. , de barros , s. , vasei , k. , et al .
2016 , , 825 , 41 wilkins , s. m. , bunker , a. j. , stanway , e. , lorenzoni , s. , & caruana , j. 2011 , , 417 , 717 wise , j. h. , demchenko , v. g. , halicek , m. t. , et al .
2014 , , 442 , 2560 xu , h. , wise , j. h. , & norman , m. l. 2013 , , 773 , 83 xu , h. , ahn , k. , wise , j. h. , norman , m. l. , & oshea , b. w. 2014 , , 791 , 110 yue , b. , ferrara , a. , & xu , y. 2016 , arxiv:1604.01314 zitrin , a. , broadhurst , t. , bartelmann , m. , et al .
2012 , , 423 , 2308 zitrin , a. , meneghetti , m. , umetsu , k. , et al .
2013 , , 762 , l30 zitrin , a. , fabris , a. , merten , j. , et al .
2015 , , 801 , 44
here we describe our procedure for estimating the effective selection volumes for faint galaxies behind the hff clusters we examine .
our baseline treatment is to model faint galaxies as point sources in estimating their selection volumes .
this choice is motivated by our demonstration in bouwens et al .
( 2016 ) that the faintest @xmath13 - 8 galaxies behind the hff clusters appeared consistent with having essentially point - source spatial profiles , with no discernible extension along the expected shear axes .
also , no discernible dependence was found for the measured surface densities as a function of the predicted shear in the high - magnification regions . nonetheless , as such small sizes for galaxies are unexpected ( e.g. , liu et al .
2016b ; kratsov 2013 ) , we also consider the impact of larger sizes ( and a larger incompleteness ) on the lf results throughout the paper . as point sources ,
the only quantity of importance from the lensing model is the magnification factor ; the form of the deflection ( or shear ) map has no impact on the results .
this simplifies the selection volume simulations enormously , since it means we can estimate the selection volumes for extremely faint sources in the presence of lensing in exactly the same way we would estimate the selection volumes in the absence of lensing
. the only quantity of importance in estimating the selection volumes is the apparent magnitudes of the sources .
bouwens et al .
( 2016 ) discuss this in their 6.3 .
we adopt a median value of @xmath218 for the @xmath10-continuum slope for galaxies in our simulations to match that found in the observations ( e.g. , bouwens et al .
2014 : see also wilkins et al .
2011 ; bouwens et al .
2012 ; finkelstein et al . 2012 ; dunlop et al . 2013
; rogers et al . 2014 ; duncan & conselice 2015 ) . adopting these assumptions for the color of the sources and a point - source assumption for the size
, we create artificial images for each source over the full suite of passbands and insert these images into the real observations .
we then do object detection and photometry using the same procedure as we use in constructing our catalogs ( 3 ) and then apply our selection criteria . in this way
, we derive the completeness for sources in different regimes .
selection volumes are computed by multiplying the cosmological volume element by the estimated completeness and integrating over redshift .
the impact of magnification errors on the derived lf can also depend on source size , as discussed in 4.1 .
this can occur as a result of the fact that higher magnification sources are more generally more difficult to detect , if they are spatially extended , than if their magnification is lower .
in other words , the selection efficiency @xmath66 is a function of the magnification factor @xmath30 .
the issue is that while the actual surface density of sources in our catalogs is proportional to @xmath67 , the selection volumes we estimate for these sources is @xmath68 .
this results in the recovered volume density for these sources being higher than the true surface density by the factor @xmath18@xmath67@xmath9@xmath219@xmath18@xmath68@xmath9 .
@xmath4 @xmath18@xmath74@xmath9 , no bias is present in the recovery volume density of sources .
however , when @xmath18@xmath73@xmath9 is less than @xmath18@xmath74@xmath9 , as is often the case at higher magnifications @xmath31 ( figure [ fig : predpow ] ) , the recovered surface density of sources can be significantly higher than reality .
for example , assuming that @xmath220 as in figure 3 of oesch et al .
( 2015 ) , the lf asymptotes towards a faint - end slope of @xmath168 . to illustrate the impact of source size and different assumptions about @xmath221 , we consider two different cases : the first involving point sources where @xmath66 is independent of @xmath30 and the second involving extended sources where @xmath221 is proportional to @xmath222 .
similar to the simulations run in 3.2 , we use the glafic model to construct mock catalogs and then recover the lf using either the grale model ( _ red line _ ) or the median parametric magnification model ( _ blue line _ ) . in the two cases ,
we incorporate the different dependencies of @xmath66 on the magnification factor @xmath30 for both the catalog construction and recovery of the lf .
the input lf for the simulations has a faint - end slope of @xmath63 , with no turn - over at the faint end .
the results are presented in figure [ fig : illustsize ] , and the differences between the two size cases are immediately obvious . in the point source case , the faint - end slope asymptotes to @xmath57 . however , in the case of extended sources , the faint - end slope instead asymptotes to @xmath168 . in either case , the lfs asymptotically approach these slopes faintward of @xmath89 mag assuming the grale model and faintward of @xmath11 assuming the median parametric magnification model .
this example should reinforce how difficult it is to obtain accurate constraints on the shape of the lf at @xmath75 mag and thus to detect the existence of a flattening or turn - over in the lf .
not only do the results depend on the magnification level to which magnification maps retain their predictive power ( e.g. , see figure [ fig : predpow ] ) , but the results also depend significantly ( i.e. , @xmath72 ) on the size distribution for faint sources . we emphasize that the impact this has on the lf shape is distinct from the effect already discussed in the companion study to the present one ( bouwens et al .
2016 ) , where the faint - end slope @xmath21 of the lf could be biased if the sizes and hence selection volumes were improperly estimated .
this bias explicitly arises _ because of errors in the magnification map _ and due to mismatches between @xmath18@xmath67@xmath9 and @xmath18@xmath68@xmath9 . if faint sources are slightly resolved ( after magnification ) as assumed in many recent studies of faint galaxies this bias has the potential to be quite significant at absolute magnitudes @xmath65 of @xmath75 mag where @xmath223 .
amazingly , however , previous work appear to have neither recognized the importance of such a bias , not made use of procedures that would allow for its correction , even though given the size assumptions in e.g. l16 , this bias would constitute an important effect .
errors in the magnification maps can have a substantial impact on the shape of the @xmath0 lf faintward of @xmath11 mag .
we already illustrated this in 3.2 of this paper using a lf which turned over at @xmath11 mag ( figure [ fig : illust ] ) .
here we show the impact of these magnification errors using a @xmath0 lf with a faint - end slope @xmath63 , which we intentionally take to be substantially shallower than @xmath57 ( the direction in which magnification errors drive the apparent faint - end slope ) .
again , we use a forward modeling procedure where we create the mock catalogs using the glafic magnification maps and then alternatively recover the lfs with the cats , grale , zitrin - ltm , and median parametric magnification maps . the result is shown in figure [ fig : illust15 ] , and excellent recovery of the lf is observed brightward of @xmath224 mag for all magnification maps .
the best performance is achieved using the median magnification map ; however , we note that faintward of @xmath11 mag , the recovered lf still diverges from the input lf , rapidly transitioning faintward of @xmath11 to a faint - end slope of @xmath57 .
in utilizing the data from the hff clusters to map out the faint - end of the uv lf and derive faint - end slope results , it is valuable to perform this exercise using only the hff samples to preserve the independence of the faint - end slope determinations from those derived from the field ( i.e. , the hudf ) . by doing so , one can conduct fair comparisons between faint - end slope results @xmath21 derived using lensing clusters and from the field to test for consistency . towards this end
, we have taken the binned lf results from l16 on the hff clusters and fit the results to a power law to estimate a faint - end slope @xmath21 .
the results are presented in figure [ fig : livfit ] as the solid lines and give slopes of @xmath225 , @xmath226 , and @xmath227 at @xmath0 , @xmath19 , and @xmath45 , respectively .
interestingly , these results are mostly steeper than the faint - end slopes inferred from field searches at the relevant redshifts .
part of this difference could be due to the modest bias towards steeper slopes as a result of the large sizes l16 use in estimating the selection volumes ( bouwens et al .
2016 : see their figure 2 ) .
we can obtain even stronger constraints on the faint - end slope @xmath21 results from the hff clusters by having at least one point on the bright end of the lf from field searches to use as an anchor .
there is not much search volume available behind clusters to constrain this part of the lf , and so including this information is useful .
we therefore refit the @xmath0 , @xmath19 , and @xmath45 lf results on the hff clusters from l16 , anchoring the fit results to the field lf results from the same group , i.e. , finkelstein et al .
( 2015 ) , at @xmath228 mag using their best - fit schechter function .
we chose @xmath228 mag somewhat arbitrarily to be close enough to the knee of the lf , i.e. , @xmath229 mag , while not being so bright as to be affected by uncertainties in the assumed characteristic luminosity or form of the bright end of the lf ( schechter vs. power - law : i.e. , bowler et al .
2015 ) .
the fits are presented in figure [ fig : livfit ] as the dotted lines .
the faint - end slopes in this case are @xmath230 , @xmath231 , and @xmath232 for the @xmath0 , @xmath19 , and @xmath45 lfs , respectively .
these values are consistent with the faint - end slope @xmath21 results we derive from the hff results from l16 without a bright anchor point . | the _ hubble _ frontier fields ( hff ) program has substantial potential for constraining both the properties and prevalence of faint galaxies in the early universe . yet
the accuracy of results derived in high magnification regions using lensing clusters is limited due to systematics .
we present a new forward - modeling formalism to incorporate the impact of magnification uncertainties into luminosity function results , by exploiting the availability of many independent magnification models for the same clusters .
one public magnification model is treated as the truth and used to construct a mock set of lensed high - redshift galaxies , which can then be analyzed using another magnification model ( typically a median model ) to construct a luminosity function . using the most comprehensive and faintest selection of lensed @xmath0 galaxies to date ( drawn from the first 4 hff clusters ) , we then derive direct constraints on the shape of the uv lf to @xmath113.5 mag and with much greater uncertainties to @xmath112.5 mag .
our simulations reveal that for highly - magnified sources ( @xmath230@xmath3 ) the systematic uncertainties become very large , reaching _ several orders of magnitude _ at 95% confidence at @xmath4@xmath5 mag .
the volume densities we derive for faint ( @xmath617 mag ) sources are @xmath43 - 4@xmath3 lower than one recent report , with a faint - end slope @[email protected] ( 3.5@xmath8 shallower ) . introducing a new curvature parameter to model the faint end of the lf , we demonstrate that current observations permit ( within the 68% confidence intervals ) a turn - over in the @xmath0 uv lf as bright as @xmath115.3 and @xmath114.2 mag , respectively , if differences between the full and parameterized set of lensing models in the hff effort are representative .
we discuss the implications of such a turn over in the context of recent theoretical predictions . |
leptogenesis @xcite , a cosmological consequence of the see - saw mechanism @xcite , provides an attractive explanation for the baryon asymmetry of the universe , one of the most long - standing cosmological puzzles .
a lepton asymmetry is produced in the decays of the very heavy rh neutrinos predicted by the see - saw mechanism . in order for the ( @xmath4 conserving )
sphaleron processes to be able to convert part of the lepton asymmetry into a baryon asymmetry , very high temperatures , @xmath5 , are required in the early universe @xcite . in comparison with other models of baryogenesis
, leptogenesis offers the unique advantage of relying on an ingredient of physics beyond the standard model , neutrino masses , already confirmed by the experiments . furthermore and very interestingly
, a quantitative analysis @xcite shows that the values of the atmospheric and of the solar neutrino mass scales , inferred from neutrino mixing experiments , favor leptogenesis to work in a mildly ` strong wash - out regime ' : inverse processes are strong enough to wash - out any contribution to the final asymmetry depending on the initial conditions but not too strong to prevent successful leptogenesis . in this way
the observed matter - antimatter asymmetry can be unambiguously explained within a minimal extension of the sm where rh neutrinos with a majorana mass term and yukawa couplings are added to the sm lagrangian and the see - saw limit is assumed .
no particular assumptions on the initial conditions are required , in complete analogy with what happens in the calculation of the primordial nuclear abundances within standard big bang nucleosynthesis . in a typical @xmath6-dominated scenario
, the asymmetry is dominantly produced from the decays of the lightest rh neutrino @xmath6 .
a necessary ( but not sufficient ) condition is the assumption of a mild hierarchy in the heavy neutrino mass spectrum , such that @xmath7 , the mass of the next - to - lightest rh neutrino , is approximately three times larger than @xmath8 , the mass of the lightest rh neutrino @xcite . in an unflavored analysis ,
a stringent lower bound on @xmath8 holds @xcite . in the decoupling limit ,
when the @xmath6-decay parameter @xmath9 and assuming an initial thermal abundance , one finds .
here we are using a slightly higher value that is obtained , as we will see , when the reduced experimental error on the baryon asymmetry and on the atmospheric neutrino mass scale is taken into account . ]
@xmath10 @xcite . however , there are different drawbacks for the saturation of this lower bound that is anyway strongly model dependent . a more significant and stringent lower bound , @xmath11 ,
is obtained at the onset of the strong wash - out regime , @xmath12 @xcite , where the final asymmetry does not depend on the initial conditions .
this lower bound implies an associated lower bound @xcite , @xmath13 , on the value of the temperature at the beginning of the standard radiation dominated regime , the reheating temperature within inflation . as we said , the assumption of a mild hierarchy in the heavy neutrino mass spectrum is not sufficient to guarantee that the @xmath6-dominated scenario holds .
it is indeed possible that , for a proper choice of the see - saw parameters , a @xmath14-dominated scenario holds , where the final asymmetry is dominated by the contribution from the decays of the next - to - lightest rh neutrino @xmath14 @xcite . in this case
the lower bound on @xmath8 does not hold any more and is replaced by a lower bound on @xmath7 , however still implying a lower bound on @xmath15 .
even when flavor effects @xcite are taken into account , violation relevant for leptogenesis , it was first discussed in @xcite . ]
these lower bounds do not get relaxed @xcite .
in particular , flavor effects do not help to alleviate the conflict with the upper bound on the reheating temperature coming from the avoidance of the gravitino problem when a supersymmetric framework is considered @xcite . on the other hand ,
flavor effects relax the lower bound on @xmath8 for @xmath16 @xcite and , interestingly , it has been shown that the dirac phase and , more efficiently , the majorana phases can strongly enhance the amount of the relaxation @xcite .
besides this effect , there is an even more interesting role played by the majorana and dirac phases when flavor effects are taken into account . in an unflavored analysis ,
the final asymmetry does not depend on the low - energy phases and this represents a limit to the possibility of further tightening the link between leptogenesis predictions and low - energy neutrino experiments .
in @xcite it has been shown that , accounting for flavor effects , an asymmetry can be generated even when the total @xmath2 asymmetry vanishes .
this is possible because flavor effects introduce an additional source of @xmath2 asymmetry stemming from low - energy phases .
the flavor composition of the anti - lepton produced in the decay of the rh neutrino can be indeed different from the one of the @xmath2 conjugated lepton . in this way
a new intriguing scenario arises , where the majorana and the dirac phases , potentially observable in low - energy neutrino experiments , could act as the unique source of @xmath2 violation responsible for the observed matter - antimatter asymmetry of the universe .
first calculations have been presented in @xcite for particular values of @xmath8 and within a two rh neutrino scenario , corresponding to a specific choice of the see - saw orthogonal matrix @xcite .
in @xcite it has been first shown that successful leptogenesis stemming only from low - energy phases is possible and the lower bound on @xmath8 , with its dependence on @xmath17 and on the initial conditions , has been calculated for a specific choice of the see - saw orthogonal matrix in the hl .
it has been found that for values of @xmath17 in the strong wash - out regime , the allowed region is very constrained when just majorana phases are switched on , and it is even worse when only the dirac phase is switched on , even for @xmath3 close to its experimental upper bound . compared to the usual cases where high - energy phases contribute to @xmath2 violation as well , the lower bounds on @xmath8 and on the reheating temperature get much more stringent , especially in the strong wash - out regime and in particular for values of @xmath17 in the range favored by neutrino mixing experiments .
therefore , one can say that the asymmetry production from low - energy phases is somehow secondary compared to the usual case when leptogenesis proceeds from the high - energy phases contained in the see - saw orthogonal matrix .
this conclusion has been recently confirmed also in @xcite in the context of mssm . in @xcite ,
the results have been generalized for an arbitrary choice of the orthogonal matrix and of the low - energy phases but without a study of the dependence on @xmath17 and on the initial conditions . in this paper
we focus on the particularly interesting case of @xmath0-leptogenesis , where the dirac phase , which has realistic chances to be observed in neutrino mixing experiments for not too small values of @xmath3 , is the only non - vanishing phase .
we study the dependence of the final asymmetry on the initial conditions , finding the onset of the strong wash - out regime and showing the dependence of the @xmath8 lower bound on the important decay parameter @xmath17 , whose value is related to the values of the neutrino masses and at the same time determines the efficiency of the asymmetry production ( involving both the production of the heavy neutrinos and the wash - out ) .
we first obtain that , in the hl , the possibility to explain the observed asymmetry is only marginally allowed and just limited to the less relevant weak wash - out regime , when the correct condition for the validity of the fully flavored regime is taken into account @xcite
. then we point out that this obstacle can be nicely circumvented going beyond the hl .
indeed , like in the unflavored case @xcite , the flavored @xmath2 asymmetries , and consequently the final @xmath4 asymmetry , get enhanced and the lower bounds on @xmath8 and on @xmath15 get relaxed .
the possibility of @xmath0-leptogenesis beyond the hl has been already studied in @xcite within resonant leptogenesis @xcite , where the heavy neutrino mass differences are equal to the resonance widths , for initial vanishing abundance and in @xcite in the context of radiative leptogenesis @xcite with minimal flavor violation @xcite .
we perform a general analysis in the degenerate limit ( dl ) , where at least one of the degeneracies @xmath18 .
we show that in this case the strong wash - out regime always holds and the lower bound on @xmath8 can be expressed through @xmath19 and the quantity @xmath20 . in the most extreme case of resonant leptogenesis
this turns both into a lower bound on @xmath1 and into an upper bound on the absolute neutrino mass scale that depend on each other . in this way we find that @xmath0-leptogenesis can indeed explain the observed matter - antimatter asymmetry of the universe in the strong wash - out regime and therefore , like leptogenesis from high - energy phases , exhibits the same virtue of independence of the initial conditions
in section 2 we introduce the general framework and set the notation . in section 3
we present the results in the hl .
we confirm , in a more general way , the conclusions of @xcite , showing that the allowed region for @xmath0-leptogenesis is quite restricted , especially in the strong wash - out regime and considering that the asymmetry production has to switch off for @xmath21 , when the unflavored case is recovered and @xmath2 violation from low - energy phases turns off .
we also verify that this conclusion holds even when the asymmetry production from the two heavier rh neutrinos is taken into account . on the other hand , we show that a @xmath14-dominated scenario can also be realized in @xmath0-leptogenesis .
we conclude that one needs to go beyond the hl for successful @xmath0-leptogenesis in the strong wash - out regime and in any case not to be just marginally allowed .
therefore , in section 4 we study the dl showing that successful @xmath0-leptogenesis is possible and we find a condition that relates @xmath19 to @xmath22 ( @xmath23 and @xmath24 ) and to @xmath25 .
we also find an upper bound on the absolute neutrino mass scale dependent on @xmath3 that makes @xmath0-leptogenesis falsifiable independently of the rh neutrino spectrum .
in section 5 we draw the conclusions .
adding to the standard model three rh neutrinos with a majorana mass term @xmath26 and yukawa couplings @xmath27 , after spontaneous breaking a dirac mass term , @xmath28 , is generated by the vev @xmath29 of the higgs boson . in the see - saw limit ,
@xmath30 , the spectrum of neutrino mass eigenstates splits in two sets , a very heavy one , @xmath31 and @xmath32 with masses respectively @xmath33 and almost coinciding with the eigenvalues of @xmath26 , and a light one , with masses @xmath34 corresponding to the eigenvalues of the light neutrino mass matrix given by the see - saw formula @xcite , m_= - m_d 1 m m_d^t .
neutrino mixing experiments measure two light neutrino mass squared differences . in a normal scheme one has @xmath35 and @xmath36 , whereas in an inverted scheme one has @xmath37 and @xmath38 . for @xmath39 @xcite
the spectrum is quasi - degenerate , while for @xmath40 @xcite is fully hierarchical . in the early universe ,
the decays of the heavy neutrinos into leptons and higgs bosons produce , in general , a lepton number that is partly converted into a baryon number by sphaleron ( @xmath4 conserving ) processes if the temperature is higher than about @xmath41 @xcite .
an important role is played by the decay parameters of the heavy neutrinos defined as @xmath42 , the ratios of the decay widths to the expansion rate when the rh neutrinos start to become non - relativistic at @xmath43 . for @xmath44 the bulk of the @xmath45 decays occurs when they are non - relativistic and the inverse decays are not effective anymore . in this case all decays occur out - of - equilibrium and the wash - out of the asymmetry is weak . on the other hand , for @xmath46 ,
the heavy neutrinos decays are balanced by inverse processes . in this case
the heavy neutrino abundance tracks quite closely the equilibrium abundance and the wash - out of the asymmetry is potentially , but not necessarily , strong .
the answer depends on a detailed description of flavor effects that are triggered by the charged lepton yukawa interactions with a rate @xmath47 @xcite , where the @xmath48 s are the charged lepton yukawa couplings in the diagonal basis .
if @xmath49 , is different by that lepton produced by the decay of a rh neutrino @xmath50 and therefore is not in general fully absorbed by the @xmath50 inverse decay @xcite . ] during all the relevant period of the asymmetry generation , then the lepton state coherence is preserved between decays and inverse decays and the unflavored regime , where flavor effects are negligible , holds .
this requirement implies @xcite [ unflavored ] m_1510 ^ 11gev . in the unflavored regime
the condition @xmath46 is also sufficient for the wash - out regime to be strong .
it is important to stress that in this regime the only source of @xmath2 violation is due to a different total decay rate into leptons and anti - leptons and , as it is well known , it stems uniquely from high - energy phases .
therefore , in the unflavored regime , @xmath0-leptogenesis is not viable . if the charged lepton yukawa interactions are in equilibrium ( @xmath51 ) and faster than inverse decays , [ condition ] _
_ i_id^i , during the relevant period of the asymmetry generation , then the lepton quantum states lose coherence between the production at decay and the subsequent absorption in inverse processes . in this way
the higgs bosons interact incoherently with leptons of each flavor . in the limit case ,
when the quantum state becomes completely incoherent and is fully projected in one of the flavor eigenstates , each lepton flavor can be treated as a statistically independent particle species and a ` fully flavored regime ' is obtained .
one has to distinguish a two - flavor regime , for @xmath52 , such that the condition eq .
( [ condition ] ) is in any case satisfied only for @xmath53 , and a three - flavor regime , where the condition eq .
( [ condition ] ) applies also to @xmath54 . in the fully ( two or three ) flavored regime
, classic boltzmann equations can be used like in the unflavored regime , with the difference , in general , that now each single flavor asymmetry has to be tracked independently . in the fully flavored regime
there are two new effects compared to the unflavored regime @xcite .
these can be understood introducing the projectors and writing them as the sum of two terms , p_i & & |l_i|l_|^2 = p_i^0 + p_i2 + of the wash - out compared to the unflavored regime and is described by the tree level contribution @xmath55 setting the fraction of the total asymmetry , produced in @xmath45-decays , that goes into each single flavor @xmath56 . in the fully flavored regime , each single inverse decay involves an independent lepton flavor eigenstate and therefore does not wash out , in general , as much asymmetry as that one produced in each single decay but an amount reduced by @xmath57 .
the second effect is an additional @xmath2 violating contribution coming from a different flavor composition between @xmath58 and @xmath59 .
this can be described in terms of the projector differences @xmath60 , such that @xmath61 . indeed , defining the flavored @xmath2 asymmetries , _
i -_i-_i _ i+_i , where @xmath62 and @xmath63 , these can be now written as _
i=_ip^0_i+ p_i2 , where @xmath64 are the total @xmath2 asymmetries . in the last expression one can see that the first term is the usual contribution due to a different decay rate into lepton and anti - leptons and the second is the additional contribution due to a possible different flavor composition between @xmath58 and @xmath59
. taking into account only decays and inverse decays with proper subtraction of the resonant contribution from @xmath65 and @xmath66 processes @xcite , the set of effective classic boltzmann equations valid in the fully three - flavored regime can be written as [ flke ] dn_n_idz & = & -d_i(n_n_i - n_n_i^eq ) ( i=1,2,3 ) + & = & _ i_id_i(n_n_i - n_n_i^eq ) -_ip_i^0 w_i^idn _ _ ( = e , , ) , where @xmath67 and where we indicated with @xmath68 any particle number or asymmetry @xmath69 calculated in a portion of co - moving volume containing one heavy neutrino in ultra - relativistic thermal equilibrium , so that @xmath70 . defining @xmath71 and @xmath72 ,
the decay factors are given by d_i = k_ix_iz , where @xmath73 is the expansion rate .
the total decay rates , @xmath74 , are the product of the decay widths times the thermally averaged dilation factors @xmath75 , given by the ratio @xmath76 of the modified bessel functions .
the equilibrium abundance and its rate are also expressed through the modified bessel functions , n_n_i^eq(z_i)= 12z_i^2k_2 ( z_i ) , = -12z_i^2k_1 ( z_i ) .
finally , the inverse decays wash - out terms are given by [ wid ] w_i^id(z ) = 14k_ik_1(z_i)z_i^3 .
we are neglecting the non resonant contributions from @xmath65 and @xmath66 processes , a good approximation for @xmath77 , as we will always consider . we are also neglecting @xmath78 scatterings @xcite , giving a correction to a level less than @xmath79 @xcite and spectator processes @xcite that , at least for a hierarchical heavy neutrino spectrum , produce a correction to a level less than @xmath80 @xcite . in the degenerate limit
it can not be excluded that the effect of spectator processes is more relevant and further studies are required .
we are also neglecting thermal corrections @xcite , that can give relevant ( though with big theoretical uncertainties ) corrections in the weak wash - out regime but negligible ones in the more important strong wash - out regime .
the evolution of the @xmath81 s can be worked out in an integral form , n_(z)=n_^in e^-_ip_i^0_z_in^zdzw_i^id(z ) + _ i_i_i(z ) , with the 9 efficiency factors given by [ ef ] _
p^0_i)=-_z_in^zdzdn_n_idz e^-_ip_i^0_z^zdzw_i^id(z ) .
the total final @xmath4 asymmetry is then given by @xmath82 . finally , assuming a standard thermal history and accounting for the sphaleron converting coefficient @xmath83 , the final baryon - to - photon number ratio can be calculated as [ etab ] _
b = a_sphn_b - l^fn_^rec 0.9610 ^ -2n_b - l^f , to be compared with the measured value @xcite [ etabobs ] _
b^cmb = ( 6.1 0.2)10 ^ -10 . notice that the efficiency factors depend only on the @xmath57 but not on the differences @xmath84 .
notice also that , in the two - flavor regime , the individual electron and muon asymmetries are replaced by one kinetic equation for the sum , @xmath85 , where the individual flavored @xmath2 asymmetries and projectors have also to be replaced by the their sum , namely @xmath86 and @xmath87 @xcite .
the calculation is therefore somehow intermediate between the one - flavor approximation and the three - flavor regime , though the results are very similar to the three - flavor regime @xcite .
the flavored @xmath2 asymmetries are given by the following expression @xcite [ veia ] _
i= _ ji \ { im+ } , where [ xi ] ( x)= 23x .
a parametrization of the dirac mass matrix , particularly fruitful within leptogenesis , is obtained in terms of the see - saw orthogonal matrix @xmath88 @xcite [ opar ] m_d = ud_m^1/2d_m^1/2 , where we defined @xmath89 and @xmath90 .
the matrix @xmath91 diagonalizes the light neutrino mass matrix @xmath92 , such that @xmath93 , and it can be identified with the lepton mixing matrix in a basis where the charged lepton mass matrix is diagonal .
moreover , neglecting the effect of the running of neutrino parameters from high energy to low energy @xcite , one can assume that the @xmath91 matrix can be identified with the pmns matrix , partially measured in neutrino mixing experiments .
for normal hierarchy we adopt the parametrization @xcite @xmath94 where @xmath95 , @xmath96 and , neglecting the statistical errors , we will use @xmath97 and @xmath98 , compatible with the results from neutrino mixing experiments .
moreover , we will adopt the @xmath99 range @xmath100 , allowed from a global @xmath101 analysis for unitary @xmath91 @xcite , an approximation that holds with great precision in the see - saw limit with @xmath102 . within the convention we are using , @xmath103 , the case of inverted hierarchy corresponds
is obtained performing a cyclic permutation of columns in the pmns matrix parametrization eq .
( [ umatrix ] ) , such that the @xmath104-th column becomes the @xmath105-th .
since we are interested in understanding whether a non - vanishing dirac phase can be the only source of @xmath2 violation for successful leptogenesis , we will set the majorana phases to zero .
we will comment later on the effects of turning on the majorana phases .
it will also prove useful to introduce the following parametrization for the see - saw orthogonal matrix , [ second ] (_21,_31,_32 ) = r_12(_21 ) r_13(_31 ) r_23(_32 ) , where [ r ] notice that , using the orthogonal parametrization , the decay parameters @xmath106 can be expressed as linear combinations of the neutrino masses @xcite [ kimi ] k_i = m _ = _ jm_jm_|_ji^2| , where @xmath107 are the effective neutrino masses @xcite and where @xmath108 is the equilibrium neutrino mass @xcite given by @xmath109 barring huge phase cancellations and special forms for @xmath88 , typically the @xmath106 s span within the range @xmath110 $ ] where @xmath111 and @xmath112 . before entering into a detailed analysis focusing on @xmath0-leptogenesis
, we want to discuss some general features concerning the fully flavored regime and in particular the possibility to have important deviations from the unflavored case .
for definiteness and simplicity , we refer to the two - flavor case within the @xmath6-dominated scenario , so that @xmath113 .
consider first the ` democratic case ' , where @xmath114 and @xmath115 .
summing the two equations for @xmath53 and @xmath116 one obtains a closed equation for the total asymmetry where the only effect compared to the unflavored case is that the wash - out is reduced by a factor two and the final asymmetry is obtained by replacing @xmath117 .
therefore , in the strong wash - out regime ( @xmath118 ) , since @xmath119 @xcite , one has approximately a factor two enhancement .
let us now consider @xmath120 , still with @xmath114 .
since approximately @xmath121 and at the same time @xmath122 , one has that the final asymmetry stays approximately constant compared to the democratic case with the two contributions from the @xmath123 and @xmath124 flavors comparable with each other .
therefore , for vanishing @xmath125 , flavor effects produce just @xmath126 corrections compared to the unflavored approximation .
this conclusion changes when non - vanishing @xmath125 are considered . in this case
there are two remarkable possibilities .
the _ first possibility _ is the so called one - flavor dominated scenario , relying on the fact that , for @xmath127 , one has @xmath128 @xcite .
therefore , considering now for example @xmath129 , one has that the asymmetry in the tauon flavor @xmath130 , showing that there can be a large enhancement compared to the unflavored case in the strong wash - out regime .
this brings to a strong relaxation of the lower bounds on @xmath8 and @xmath15 at @xmath118 , though , as we already said , not to a relaxation of the usual lowest bounds at @xmath9 or at @xmath131 .
it should also be said that , as shown in @xcite , the applicability of the one - flavor dominated scenario is strongly limited by the condition of validity of the fully flavored regime eq .
( [ condition ] ) .
the _ second possibility _ relies on the observation that , contrarily to @xmath132 , the @xmath125 s depend on the low - energy phases as well and , even though @xmath133 , they do not vanish if the dirac or the majorana phases do not vanish .
therefore , one can have a final asymmetry originating just from low - energy phases @xcite .
this scenario represents , potentially , the most important novelty introduced by flavor effects compared to the unflavored picture and in the following sections we will study it in detail , focusing on the case of @xmath0-leptogenesis , when only the dirac phase is switched on while @xmath88 is real and the two majorana phases vanish . before concluding this section , we want to notice that one can have @xmath134 not only when the see - saw orthogonal matrix is real , but also when the absolute neutrino mass scale increases @xcite . in this way
, the low - energy phases can play an important role in circumventing the upper bound on the neutrino masses holding in the unflavored regime @xcite .
it is however still to be assessed whether the fully flavored regime can offer a sufficient description to solve this issue . in @xcite
the bound was found to be completely nullified by flavor effects .
in @xcite it has been pointed out how this conclusion relies on a extension of the fully flavored regime beyond the regime of its validity given by the eq .
( [ condition ] ) . in @xcite
the authors find that in the fully flavored regime , thanks to spectator processes , the bound holding in the unflavored regime , even though not nullified , is anyway relaxed to @xmath135 .
let us consider first @xmath0-leptogenesis in the hl , such that @xmath136 @xcite . in the unflavored regime , this assumption typically implies a @xmath6-dominated scenario , where the final asymmetry is dominated by the contribution from the lightest rh neutrino decays , [ n1ds ] n^f_b - l .
n_b - l^f |_n_1 _
indeed , in general , in the hl one has two effects .
the first effect is that the asymmetry production from the two heavier rh neutrinos , @xmath14 and @xmath32 , is typically later on washed out by the @xmath6 inverse processes and @xmath137 .
the second effect is a consequence of the fact that the total @xmath2 asymmetries vanish in the limit when all particles running in the loops become massless and this yields typically @xmath138 .
however , for a particular choice of the see - saw parameters , @xmath139 and @xmath140 , the contribution to the final asymmetry from the next - to - lightest rh neutrino @xmath14 is not only non - negligible but even dominant , giving rise to a @xmath14-dominated scenario @xcite .
indeed for @xmath139 different things happen simultaneously .
first , @xmath14 , even though decoupled from @xmath6 , is still coupled to @xmath32 and in the hl the total @xmath2 asymmetry @xmath141 does not vanish , since it receives a non suppressed contribution from graphs where @xmath32 runs in the loops . on the other hand ,
now one has @xmath133 , since @xmath6 is essentially decoupled from the other two rh neutrinos . at the same time one also has @xmath142 , so that the wash - out from @xmath6 inverse processes is negligible .
the final result is that @xmath143 and the final asymmetry is dominantly produced from @xmath14-decays .
therefore , in the unflavored approximation and in the hl , a condition @xmath144 in the @xmath88-matrix parametrization ( cf .
( [ second ] ) ) is sufficient to have a negligible asymmetry production from the two heavier rh neutrinos and to guarantee that the @xmath6-dominated scenario holds .
this condition is even not necessary for @xmath145 , since in this case , due to the fact that @xmath146 , one has necessarily @xmath118 and a wash - out from @xmath6-inverse processes is anyway strong enough to suppress a possible contribution to the final asymmetry produced from @xmath14-decays .
when flavor effects are taken into account , the domain of applicability of the @xmath6-dominated scenario reduces somehow .
there are two aspects to be considered .
the first aspect is that the wash - out from @xmath6 inverse processes becomes less efficient . indeed
the projectors @xmath147 can considerably reduce the wash - out of the asymmetry produced in the flavor @xmath56 from @xmath14-decays @xcite .
this turns the condition @xmath145 into a looser condition @xmath148 .
another effect is that @xmath6 inverse processes can make part of the asymmetry produced in @xmath14 decays somehow orthogonal to the the wash - out from @xmath6 inverse processes @xcite .
recently , it has been also pointed out that spectator processes can lead to a reduction of the wash - out from @xmath6-inverse processes as well @xcite . in this way
the assumption @xmath149 is not valid in general .
the second aspect concerns the flavored @xmath2 asymmetries . in the hl , from the general expression
( [ veia ] ) , one has @xmath150,\\
\label{ve2a } \ve_{2\alpha}&\simeq & \frac{3}{16 \p ( h^{\dag}h)_{22 } } \left\{\frac{m_2}{m_3}{\rm i m } \left[h_{\a 2}^{\star } h_{\a 3}(h^{\dag}h)_{2 3}\right ] -\frac{2}{3 } { \rm im}\left[h_{\a 2}^{\star } h_{\a 1}(h^{\dag}h)_{1 2}\right ] \right\ } , \\
\ve_{3\alpha}&\simeq & -\frac{1}{8\,\p ( h^{\dag}h)_{33 } } \sum_{j\neq 3 } \left\{{\rm im}\left[h_{\a 3}^{\star } h_{\a j}(h^{\dag}h)_{j 3}\right]\right\}.\end{aligned}\ ] ] different comments are in order .
the @xmath151 s , like @xmath132 , vanish for @xmath152 while the @xmath153 s , like @xmath141 , do not . on the other hand , in the hl ,
the @xmath153 s , contrarily to @xmath141 , are not suppressed when @xmath154 ( a particular example is given by @xmath155 ) but , like @xmath141 , they vanish for @xmath156 .
this observation @xcite can also potentially contribute to enlarge the domain of applicability of the @xmath14-dominated scenario when flavor effects are taken into account .
another interesting observation is that the @xmath157 s , contrarily to @xmath158 , do not vanish in the hl .
this could open the door even to a @xmath32-dominated scenario , though this is possible only for @xmath159 , when flavor effects are effective in @xmath32 decays .
therefore , when flavor effects are taken into account , the conditions of applicability of the @xmath6-dominated scenario become potentially more restrictive than in the unflavored case .
there is a clear choice of the parameters , for @xmath156 and @xmath160 , where the @xmath6-dominated scenario holds . indeed in this case , in the hl
, one has that @xmath153 and @xmath158 are suppressed .
this can be considered somehow opposite to the case @xmath152 , where the @xmath14-dominated scenario holds @xcite . in general
, one can say that the asymmetry produced from the two heavier rh neutrinos is non - negligible if two conditions are satisfied .
( i ) the asymmetry generated from @xmath161-decays at @xmath162 has to be non - negligible compared to the asymmetry generated at @xmath163 from @xmath6-decays .
this depends on an evaluation of the @xmath2 asymmetries @xmath164 and of the wash - out due to the same @xmath161-inverse processes .
( ii ) the asymmetry produced from @xmath161-decays has not to be afterwards washed - out by @xmath6-inverse processes .
notice that this second condition is subordinate to the first condition . in the particular case of @xmath0-leptogenesis
, one has @xmath165 .
this means that the first condition can be satisfied only if @xmath166 and this constitutes already an important limitation . in the following , we will consider different particular cases , verifying whether the production from the two heavier rh neutrinos can be neglected or not .
we will find that the situation is actually similar to what happens in the unflavored case where , except for the case @xmath167 , a @xmath6-dominated scenario holds .
let us therefore start showing in detail how to calculate the contribution to the final asymmetry from @xmath6-decays .
the expression eq .
( [ ef ] ) for the @xmath168 s can be specialized as [ ef1 ] _
1(z;k_1,p^0_1)= -_z_in^zdzdn_n_1dz e^-p_1 ^ 0_z^zdzw_1^id(z ) . from the eq .
( [ ef ] ) , extending an analytic procedure derived within the one - flavor approximation @xcite , one can obtain simple analytic expressions for the @xmath169 s . in the case of an initial thermal abundance ( @xmath170 ) , defining @xmath171 , one has [ k1a ] _ 1^f ( k_1 ) ( 1-e^-k_1z_b(k_1)2 ) , where z_b(k_1 ) 2 + 4k_1 ^ 0.13e^-2.5k_1 . in the case of initial vanishing abundance ( @xmath172 )
one has to take into account two different contributions , a negative and a positive one , so that _ 1^f = _
-^f(k_1,p_1 ^ 0)+ _ + ^f(k_1,p_1 ^ 0 ) , whose analytic expressions , used to obtain all presented results , can be found in @xcite . the condition for the validity of the fully flavored regime eq .
( [ condition ] ) can be specialized and re - cast like [ full ] m_1 .
this condition neglects the effect of @xmath78 scatterings and of coherent scatterings , the first contributing with inverse decays to preserve the quantum state coherence , the second , conversely , in projecting it on the flavor basis @xcite .
both of them can be as large as the effect from inverse decays .
moreover , in a rigorous quantum kinetic description , it is likely that other subtle effects contribute to the determination of the exact value of @xmath8 below which the fully flavored regime can be assumed .
therefore , the condition ( [ full ] ) should be regarded as a very qualitative one . in the plots showing the @xmath8 lower bound , we will then distinguish four regions .
all plots will be cut at @xmath173 , since above this value , according to the condition ( [ unflavored ] ) , the unflavored regime is recovered and the asymmetry production has to switch off . on the other hand , when the condition eq .
( [ full ] ) is satisfied , one can expect the fully flavored regime to hold .
there is an intermediate regime where a transition between the fully flavored regime and the unflavored regime takes place .
this regime will be indicated in all plots with a squared region .
this signals that , even though we still show the results obtained in the fully flavored regime , important corrections are expected , especially when @xmath8 gets close to @xmath174 . since this region describes a transition toward the unflavored regime , where the asymmetry production has to switch off , these corrections are expected to reduce the final asymmetry , making more stringent the lower bounds shown in the plots .
furthermore , since within current calculation , large corrections to the condition eq .
( [ full ] ) can not be excluded , we will also indicate , with a hatched region , that area where the condition eq .
( [ full ] ) holds but a very conservative condition , [ conservative ] m_1 does not . in this region some corrections to the presented results can not be excluded but the fully flavored regime should represent a good approximation . we anticipate that , in the @xmath6-dominated scenario , successful leptogenesis always requires @xmath175 , where the two - flavor regime holds .
therefore , considering that we are assuming @xmath176 , the eq .
( [ n1ds ] ) can be specialized into [ final ] .
n_b - l^f |_n_1 ( _ 1^f-_1,e+^f)_1 , showing that , in order to have a non - vanishing final asymmetry it has to be @xmath177 .
the tree - level projectors can be expressed , through the orthogonal parametrization eq .
( [ opar ] ) , like [ p01a ] p^0_1=|_ju_j_j 1|^2 _ jm_j|^2_j1| , that , from the eq .
( [ kimi ] ) , also implies [ k1alpha ]
k_1= |_ju_j_j 1|^2 .
let us now calculate the flavored @xmath2 asymmetry @xmath178 from the general expression eq .
( [ ve1a ] ) . in terms of the orthogonal parametrization eq .
( [ opar ] ) , this can be re - cast as @xcite [ e1alom ] r_1=-_h , l m_lm_atm im[u_hu_l^_h1_l1 ] , where we defined @xmath179 , with specialized into @xcite r_1=-_h < l ( m_l - m_h)m_atm _h1_l1im[u_hu_l^ ] . taking @xmath53 and specifying the matrix elements @xmath180 , from the eq .
( [ umatrix ] ) , one has [ r1tau ] r_1=-m_atm[a_12+a_13+a_23 ] , where a_12 & = & - ( m_2-m_1 ) m_atm^2 _11_21 im[(s_12s_23-c_12c_23s_13e^i ) + & & ( c_12s_23+s_12c_23s_13e^-i ) e^-i2(_2-_1 ) ] , + a_13 & = & ( m_3-m_1)m_atm^2 _11_31c_23c_13 im[(s_12s_23-c_12c_23s_13e^i ) e^i2_1 ] , + a_23 & = & - ( m_3-m_2 ) m_atm^2 _21_31c_23c_13 im[(c_12s_23+s_12c_23s_13e^i ) e^i2_2 ] . in the case of @xmath0-leptogenesis ( @xmath181 )
these expressions further specialize into a_12 & = & ( m_2-m_1)m_atm^2 _11_21s_23c_23 + a_13 & = & -(m_3-m_1)m_atm^2 _11_31c_23 ^ 2c_12c_13 , + a_23 & = & - ( m_3-m_2)m_atm^2 _21_31c_23 ^ 2s_12c_13 , where remember that @xmath182 .
it is now instructive to make some general considerations .
looking at the expression eq .
( [ final ] ) , one can see that , in order for the final @xmath4 asymmetry not to vanish , two conditions have to be simultaneously satisfied : @xmath183 and @xmath184 .
these two conditions are a specialization of the sakharov necessary conditions to the case of @xmath0-leptogenesis .
indeed , the first is the condition to have @xmath2 violation and , as one could expect , from the expressions found for the terms @xmath185 , one can have @xmath183 only if @xmath186 .
the second condition is a specialization of the condition of departure from thermal equilibrium in quite a non - trivial way .
indeed , in the case of @xmath0 leptogenesis , in a full out - of - equilibrium situation where only decays are active , no final asymmetry is generated since @xmath133 , implying that there is an equal number of decays into lepton and anti leptons .
however , the presence of inverse processes can remove this balance , yielding a different wash - out rate for the @xmath124 asymmetry and for the @xmath187 asymmetry , such that , if @xmath188 , one has a net lepton number dynamical generation . from the expression ( [ k1alpha ] ) , one can see that this is possible independently of the value of the dirac phase that , therefore , is directly responsible only for @xmath2 violation and not for lepton number violation , exactly as in neutrino mixing , where indeed lepton number is conserved .
it should also be noticed that the @xmath151 s are expressed through quantities @xmath189 $ ] , that are invariant under change of the pmns matrix parametrization @xcite .
therefore , the final asymmetry depends correctly only on physical quantities . maximizing the asymmetry over all involved parameters for fixed @xmath8 and @xmath17 and imposing @xmath190 ( cf .
( [ etab ] ) and ( [ etabobs ] ) ) , a lower bound on @xmath8 is obtained @xcite m_1 m_1^min(k_1 ) , where we introduced the quantity gives the lower bound on @xmath8 in the unflavored case for initial thermal abundance and in the limit @xmath9 .
because of the improved determination of @xmath191 from the 3 years wmap data @xcite and from new data from neutrino oscillation experiments , in particular from the minos experiment , the error on @xmath192 is halved compared to the previous estimation in @xcite . ] [ barm1 ] _ 1 n_^recv^2a_sph _
b^cmbm_atm = ( 6.250.4)10 ^ 8gev 510 ^ 8gev .
the last inequality gives the @xmath99 value that we used to obtain all the results shown in the figures .
we also defined @xcite [ xi1a ] _ 1 _ = , e+_1 , _ 1 , that gives the deviation introduced by flavor effects compared to the unflavored approximation in the hierarchical light neutrino case ( @xmath193 ) .
notice that @xmath194 , implying @xmath195 as well .
therefore , the maximum asymmetry is obtained for @xmath196 and @xmath197 . the calculation of the contribution to the asymmetry from @xmath14-decays proceeds in an analogous way .
again this can always be calculated in the two - flavor regime , since , in the hl , successful leptogenesis always implies @xmath198 .
therefore , one can write an expression similar to the eq .
( [ final ] ) for the contribution to the final asymmetry from @xmath14-decays , [ final2 ] .n_b - l^f|_n_2(_2^f-_2,e+^f)_2 .
the difference is now in the calculation of the efficiency factors that are suppressed by the wash - out of the @xmath6 inverse processes . in the hl
this additional wash - out factorizes and @xcite _ 2^f(k_2 ) e^-38k_1 , where @xmath199 . for the calculation of the tree - level projectors
@xmath200 an expression analogous to the eq .
( [ p01a ] ) holds .
the calculation of the contribution to the final asymmetry from @xmath32-decays proceeds in a similar way and analogous expressions hold .
the only non trivial difference is that now , in the calculation of the efficiency factors , one has also to include the wash - out from the @xmath14 inverse processes , so that [ k3a ] _
3^f(k_3 ) e^-38(k_1+k_2 ) .
notice that in the calculation of @xmath201 ( @xmath202 ) we are not including a possible effect where part of the asymmetry in the flavor @xmath116 produced in @xmath14 ( @xmath32-decays ) is orthogonal to @xmath6 inverse decays @xcite and is not washed out .
this wash - out avoidance does not apply to the asymmetry in the @xmath124 flavor .
therefore , as we have verified , in all cases we have considered the effect is negligible , since a @xmath124-dominated scenario is always realized .
let us now calculate the final asymmetry in some interesting cases .
the first case we consider is @xmath156 , implying @xmath203 in the eq .
( [ r1tau ] ) . as we said already , it is easy to check from the eq .
( [ ve2a ] ) that @xmath204 and therefore there is no asymmetry production from @xmath14-decays even if @xmath205 . on the other hand , one obtains r_3= -23(m_3-m_1 ) m_atm _31c_12c_23 ^ 2c_13 , essentially the same expression as for @xmath206 but with @xmath207 replaced by @xmath208 .
therefore , for @xmath159 , one has to worry about a potential non - negligible contribution from @xmath32 decays .
however , when the wash - out from @xmath6 and @xmath14 inverse processes is taken into account , see eq .
( [ k3a ] ) , we always find that the contribution from @xmath32-decays is negligible and the @xmath6-dominated scenario holds .
the results are shown in fig . 1 for @xmath197 , @xmath209 and @xmath210 , a choice of values that approximately maximizes the final asymmetry and yields the lower bound @xmath211 . in the left panel
we show the tree level projectors @xmath212 and the @xmath213 s .
it can be seen how for @xmath214 one has @xmath215 , while for @xmath216 one has @xmath217 . in the central panel @xmath218 and the @xmath219 s
are plotted and one can see how for @xmath220 a @xmath124-dominance is realized . finally , in the right panel , we show @xmath211 and we compare it with the lower bound in the unflavored approximation obtained for @xmath156 ( in this case @xmath88 can not be real ) @xcite .
one can see how , at @xmath214 , the asymmetry production rapidly dies , so that @xmath221 and @xmath222 .
notice that we plotted the lower bound both for initial thermal @xmath6-abundance and for initial vanishing @xmath6-abundance .
we also indicated @xmath223 , defined as that value of @xmath17 such that for @xmath224 the dependence on the initial conditions can be neglected and the strong wash - out regime holds .
one can notice that the intermediate regime between a fully flavored regime and the unflavored regime , the squared area , is quite extended . in this regime corrections to the results we are showing , obtained in the fully flavored regime ,
are expected , in a way that the unflavored regime should be recovered for @xmath225 . in this limit
the asymmetry production has to switch off and therefore one expects that the lower bound on @xmath8 has to become more restrictive and eventually , for @xmath225 , the allowed region has to close up .
therefore , one can see that there is no allowed region in the strong wash - out regime .
the hatched area , where corrections can not be excluded within current theoretical uncertainties , cuts away almost completely any allowed region even in the weak wash - out regime . in conclusion , the allowed region where one can safely rely on the fully flavored regime according to current calculations , is very restricted and confined only to a small region in the weak wash - out regime . the second case we consider is the limit @xmath226 . in this limit
one has necessarily @xmath227 , implying @xmath228 , and also @xcite [ ss ] = ( ccc 0 & 0 & 1 + & -_31 & 0 + _31 & & 0 ) .
notice that this particular form of @xmath88 corresponds to set @xmath229 and @xmath230 in the eq .
( [ r ] ) .
now in the expression for @xmath206 ( cf .
( [ r1tau ] ) ) one has @xmath231 and therefore r_1 ( 1-m_2m_atm ) _31c_23 ^ 2c_13s_12 . if @xmath232 , there is no contribution from the next - to - lightest rh neutrino decays anyway , since these occur in the unflavored regime where @xmath233 . on the other hand , if @xmath205 , then one has to worry about a ( flavored ) asymmetry generation from @xmath14-decays .
a calculation of @xmath153 shows that the first term in the eq .
( [ ve2a ] ) vanishes while the second term gives r_2=-23m_atm ( 1-m_2m_atm)_31 c_23 ^ 2s_12 .
this is an example of how the second term in the eq .
( [ ve2a ] ) is not suppressed in the hl like the first term .
however , like for the contribution from @xmath32-decays in the case @xmath156 , when the wash - out from @xmath6-inverse processes is taken into account one finds @xmath234 and a @xmath6-dominated scenario is realized anyway .
notice that there is a strong dependence whether one assumes a normal or an inverted hierarchy . for normal hierarchy
the results are shown in fig . 2 for @xmath235 and @xmath236 . for inverted hierarchy
the asymmetry is so suppressed that there is no allowed region .
this means that for any choice of the parameters one always obtains @xmath237 .
notice that results for @xmath0-leptogenesis , in this particular case where @xmath226 , have been recently presented in @xcite for vanishing initial @xmath6 abundance .
for example in @xcite the authors obtain a lower bound @xmath238 imposing the existence of an allowed region for @xmath239 while we would obtain @xmath240 .
the difference is probably due to a ( @xmath241 ) more conservative lower bound that we are using on @xmath192 ( see eq .
( [ barm1 ] ) ) , a difference in the employed value of @xmath108 ( see eq .
( [ d ] ) ) , only partly understood in terms of the different convention for the higgs v.e.v @xmath29 .
there is also a difference in the employed efficiency factor in the strong wash - out regime that , in our case , is about a factor 2 larger .
another likely minor source of difference is that we are not accounting for the effect of spectator processes encoded in the matrix @xmath242 that relates the @xmath243 asymmetries to the @xmath244 asymmetries @xcite .
however , notice that here we do not want to emphasize too much a precise value of this lower bound on @xmath3 , since we believe this is anyway affected by much larger theoretical uncertainties on the validity of the fully flavored regime .
it is however a good way to compare our results with those presented in @xcite .
the third case we consider is @xmath155 .
this time one has @xmath245 in the eq.([r1tau ] ) . in the case of normal hierarchy
the @xmath2 asymmetry , compared to the case @xmath156 , is suppressed by a factor @xmath246 , while it is essentially the same for inverted hierarchy .
the projectors present very similar features to the case @xmath156 .
one can also again calculate , for @xmath205 , the contribution from @xmath14-decays to the final asymmetry and again one finds that the first term in the eq .
( [ ve2a ] ) vanishes , while the second produces a term @xmath247 , so that r_2= 23(m_2-m_1 ) m_atm _21s_23c_23 .
when the efficiency factors are taken into account , one finds that only in the case of normal hierarchy the contribution to the final asymmetry from @xmath14-decays can be comparable to that one from @xmath6-decays .
however , in this case both productions are suppressed and there is no allowed region anyway in the end . in the case of inverted hierarchy ,
the contribution from @xmath14-decays is always negligible compared to that one from @xmath6-decays .
notice , moreover , that @xmath248 and therefore there is no contribution from @xmath32-decays . in conclusion , for @xmath155 ,
the lower bound on @xmath8 for normal hierarchy is much more restrictive than in the case @xmath156 , while it is very similar for inverted hierarchy .
a production from the two heavier rh neutrinos can be neglected and the @xmath6-dominated scenario always holds when the asymmetry is maximized .
the last interesting case is @xmath152 . from the eq .
( [ e1alom ] ) one can easily check that @xmath249 .
one can also easily check that , contrarily to the case @xmath155 , the second term in the eq .
( [ ve2a ] ) vanishes while the first term does not and yields r_2 = ( m_3-m_2)m_atm _32s_12c_23 ^ 2c_13 .
notice that this time @xmath250 and actually , more generally , one can see that this expression is obtained from the eq .
( [ r1tau ] ) for @xmath206 in the case @xmath156 , just with the replacement @xmath251 . at the same time
one has @xmath252 so that the wash - out from @xmath6-inverse processes vanishes for @xmath253 . for @xmath159 one has to worry about a possible contribution to the asymmetry also from @xmath32-decays .
a straightforward calculation shows that @xmath254 and therefore an asymmetry is produced at @xmath255 .
we verified , once more , that the wash - out from @xmath14-inverse processes is always strong enough that the contribution to the final asymmetry from @xmath32-decays is negligible . in complete analogy with the unflavored case @xcite
, one has that the lower bound @xmath211 is replaced by a lower bound @xmath256 obtained for @xmath257 and shown in the right panel of fig .
3 . one can see that also in this case , within the validity of the condition eq .
( [ full ] ) , the allowed region is constrained to a small portion falling in the weak wash - out regime .
assuming the very conservative condition of validity for the fully flavored regime , outside the squared and hatched regions , there is no allowed region even in the weak wash - out regime .
one can wonder whether there is some choice of @xmath88 , beyond the special cases we analyzed , where the final asymmetry is much higher and the lower bound on @xmath8 much more relaxed , especially in the strong wash - out regime .
we have checked different intermediate cases and we can exclude such a possibility .
therefore , the lower bound shown in fig .
1 has to be considered , with good approximation , the lowest bound for any choice of real @xmath88 .
another legitimate doubt is whether , going beyond the approximations we made , the lower bound in fig .
1 can be considerably relaxed .
however , the inclusion of non resonant @xmath65 or @xmath78 scattering does not produce large corrections .
recently the effect of the off - diagonal terms in the @xmath242 matrix has been considered , but it has been shown that it does not produce any relevant change in the final asymmetry @xcite .
relevant corrections , as already pointed out , can come only from a full quantum kinetic treatment , that should describe accurately the transition between the unflavored regime and the fully flavored regime .
the same kind of considerations holds for the @xmath14-dominated scenario , realized for @xmath152 .
as soon as @xmath88 deviates from @xmath258 , the wash - out from @xmath6 inverse processes comes into play suppressing the final asymmetry and at the same time @xmath259 gets also suppressed .
therefore , the lower bound on @xmath7 is necessarily obtained for @xmath152 in complete analogy with the unflavored approximation @xcite . in conclusion @xmath0-leptogenesis in the hl
is severely constrained , confirming the conclusions of @xcite and @xcite .
in particular , imposing independence of the initial conditions , then not even a marginal allowed region seems to survive .
notice moreover that all plots have been obtained for @xmath260 , the current @xmath261 upper bound value .
assuming that for values of @xmath8 above the condition eq .
( [ full ] ) the unflavored regime is quickly recovered and therefore that the asymmetry production quickly switches off , then a one - order - of - magnitude improvement of the upper bound on @xmath3 would essentially completely rule out @xmath0-leptogenesis in the hl , even the marginally allowed regions falling in the weak wash - out regime . therefore , in the next section , we will consider the effect of close heavy neutrino masses in enhancing the @xmath2 asymmetries and relaxing the lower bounds on @xmath262 and the related one on @xmath15 . in the end of this section
we want to mention that in the more general case of real @xmath88 with non - vanishing majorana phases , an upper bound @xmath263 has been obtained in the fully flavored regime @xcite .
this bound clearly applies also to @xmath0-leptogenesis , but in this case , considering the results we have obtained and the expected quantum kinetic corrections to the fully flavored regime , the issue is actually whether an allowed region exists at all in the hl , even for @xmath193 .
therefore , we do not even try to place an upper bound on @xmath264 in the hl . in the next section
, we will show that actually for @xmath0-leptogenesis an upper bound on @xmath264 holds even in the resonant limit , where the @xmath2 asymmetries are maximally enhanced .
in this section we show that going beyond the hl the lower bound on @xmath8 ( or on @xmath7 ) can be considerably relaxed .
nevertheless , we will see that some interesting constraints on the involved parameters still apply . for simplicity
, we can assume a full three - flavor regime holding for @xmath8 ( or @xmath7 ) @xmath265 , when also the muon - yukawa interactions are faster than inverse decays .
therefore now , when we sum over the flavor index , it has to be meant @xmath266 .
this assumption simplifies the calculation , since we do not have to describe a transition between the two and the three - flavor regime and because we can completely neglect the effect , envisaged in @xcite , for which part of the asymmetry produced from @xmath14-decays is not touched by @xmath6-inverse decays . indeed in a two - flavor regime ,
even though in the hl we have found that this effect is negligible in all cases we considered because a @xmath124-dominance is always realized , in the dl it can become more relevant because the asymmetry is not necessarily produced dominantly in the @xmath124-flavor . in order to go beyond the hl , it is convenient to introduce the quantities _ ji = -1 .
we are interested in the degenerate limit ( dl ) , where at least one @xmath19 is small enough that both the asymmetry production from decays and the wash - out from inverse processes of the @xmath45 s and of the @xmath267 s can be approximately treated as if they occur at the same temperature , so that they can be simply added up .
the dl is a good approximation for @xmath268 @xcite .
if @xmath269 and @xmath270 then one has a partial dl and in this case the efficiency factors can be approximated , for thermal initial abundance , as @xcite [ k1 ] _
i^f_j^f(k_i+k_j ) . in all considered cases , it will be always verified @xmath271 , so that the strong wash - out regime always applies and there is no need to consider the case of initial vanishing abundance .
another possibility is to have a partial dl with @xmath272 so that @xmath273 . in this case one
has to take into account the wash - out from the lightest rh neutrino and therefore [ k2 ] _ i^f_j^f(k_i+k_j)e^-38k_1 . finally , in the full dl , one has @xmath274 and [ k3 ] _ 1^f_2^f_3^f ( k_1+k_2+k_3 ) .
let us now calculate the flavored @xmath2 asymmetries . in the case of real @xmath88
, implying real @xmath275 , the general expression eq.s ( [ veia ] ) can be conveniently specialized as [ veiarealo ] _ i= _
ji(h^h)_i jim . in the dl one has approximately @xmath276 and consequently _ i _
ji(h^h)_i j im_ji^-1
. we can again express the neutrino yukawa coupling matrix through the orthogonal representation .
this time the presence of the factor @xmath277 does not allow to remove the sum on @xmath278 , as it has been possible in the hl in order to derive the eq .
( [ e1alom ] ) .
however , considering the same special cases studied in the hl , only one term @xmath279 survives and we can write _ i _
n , h < lm_nm_atm _ni_nj [ _hi_lj - _li_hj ] im[u^_hu_l ] .
the same expression holds for @xmath280 simply exchanging the @xmath104 and @xmath278 indexes .
we can always choose @xmath281 , so that @xmath282 . in all the particular cases we will consider
it is realized @xmath283 , for @xmath284 , and moreover the following simplifications apply : _
nm_n_ni_nj=(m_q - m_p)_ji_jj _ h < l[_hi_lj - _li_hj ] = , with @xmath285 . except for @xmath226 , in the other cases one has @xmath286 and @xmath287 .
the final asymmetry can then be expressed as n_b - l^f _ ( _ i+_j ) _ ^f(k_i+k_j , k_k)= 2k_atm(k_i+k_j)k_ik_j ( m_q - m_p)m_atm^2 _ji + & & [ g ] _
_ ^f(k_i+k_j , k_k ) im[u^_pu_q ] and where @xmath288 is given by one of the three expressions eq .
( [ k1 ] ) , eq .
( [ k2 ] ) or eq .
( [ k3 ] ) , according to the particular case .
it is interesting to notice that in the full dl the expression ( [ k3 ] ) holds and as a consequence of the orthogonality of @xmath88 , one has k_1+k_2+k_3=_km_km_|u_k|^2 . in the degenerate limit ,
since @xmath91 is unitary , this quantity tends to @xmath289 , independently of the flavor , and therefore the sum on the flavors in the eq .
( [ g ] ) tends to vanish .
this will contribute , as we will see , to place a stringent upper bound on the absolute neutrino mass scale in the full dl .
it is also worthwhile to notice that the sign of @xmath25 can not be predicted from the sign of the observed final asymmetry , since the sign of @xmath290 depends on the sign of @xmath291 that is undetermined .
notice also that @xmath292/\d}$ ] does not depend on @xmath25 but nevertheless there is a dependence of @xmath290 on @xmath0 and on @xmath1 coming from the tree level projectors @xmath293 in the sum @xmath294 .
however , in any case , for @xmath295 one has @xmath296 , since the final asymmetry has to vanish when @xmath3 or @xmath297 vanish .
the function @xmath298 can be maximized over @xmath291 .
indeed for @xmath193 , since @xmath299 and @xmath300 , one has @xmath301 .
increasing @xmath264 there is a suppression due to the fact that @xmath302 and @xmath303 decreases monotonically .
therefore , for any @xmath264 , there is a lower bound on @xmath8 given by [ lbm1 ] m_1 m_1^min(m_1,_13 , ) _
the @xmath2 asymmetries , and consequently the final asymmetry , are maximally enhanced in the extreme case of resonant leptogenesis @xcite when the heavy neutrino mass degeneracy is comparable to the decay widths .
this implies approximately to have @xmath304 with @xmath305 , that would correspond to have @xmath306 in the unflavored case with maximal phase .
this can be taken as a conservative limit that implies , maximizing over @xmath0 , a lower bound [ lbsint ] _ 13 _ 13^min = d_b^cmbn_^rec a_sph max_[g_max(m_1,_13^min , ) ) ] .
notice that , within the validity of perturbation theory , one can not specify which is the exact value of @xmath307 , that means the value of @xmath19 above which the expression for the @xmath2 asymmetries given in the eq .
( [ ve1a ] ) are valid @xcite and therefore there is an uncertainty in the calculation of the maximum enhancement of the asymmetries in the resonant regime .
let us now specialize the expressions for the four special cases we have already analyzed in the hl .
remember that in this case one has @xmath308 implying @xmath248 , a consequence of the fact that the heaviest rh neutrino decouples .
moreover @xmath227 , such that terms @xmath309 can be neglected , @xmath228 and @xmath310 for normal hierarchy or @xmath311 for inverted hierarchy .
therefore , there is actually no dependence on @xmath264 in @xmath290 that we can indicate simply with @xmath312 and that is given by the expression ( [ g ] ) with @xmath313 and @xmath314 or explicitly g(_21,_13 , ) & & 2(k_1+k_2)k_atmk_1k_2 ( 1-m_2m_atm ) _21 + [ gm3 ] & & _ ( k_1+k_2 ) im[u^_2u_3 ] . in the case of normal hierarchy
@xmath315 slightly decreases when @xmath25 increases and so the maximum is found for @xmath316 and in this case the dependence on @xmath1 and on @xmath0 disappears . replacing @xmath317 with @xmath17 , in fig .
4 we have plotted @xmath318 for central values of @xmath319 and @xmath320 . including the errors , one finds @xmath321 .
the ( @xmath99 ) lower bounds on @xmath8 for normal hierarchy , from the general expression ( [ lbm1 ] ) , is then given by [ lbm1minf ] m_1 0.9 10 ^ 10gev_21|| .
in the case of inverted hierarchy the situation is somehow opposite , since for @xmath322 the electron flavor contribution vanishes in the eq .
( [ gm3 ] ) and there is an exact cancellation between the @xmath323 and @xmath324 contributions .
consequently , the asymmetry increases for increasing values of @xmath1 and thus the maximum is found for @xmath325 while @xmath326 . in this case
one has that @xmath327 \simeq ( 9\pm 2)\times 10^{-8}$ ] , that plugged in the eq .
( [ lbm1 ] ) gives at @xmath99 m_1610 ^ 15gev_21 .
it should be remembered that these conditions have been obtained in the three - flavor regime and in the dl and therefore are valid for @xmath328 .
this implies @xmath329 for normal hierarchy and @xmath330 for inverted hierarchy .
analogously the general expression ( [ lbsint ] ) gives , for normal and inverted hierarchy respectively , the following ( @xmath99 ) lower bounds on @xmath3 : [ lbsintminf ] _ 13 3.310 ^ -7d _ 130.06d . in this particular case , the next - to - lightest rh neutrino
is decoupled from the other two and this implies that @xmath331 for any @xmath56 and that the @xmath151 s do not depend on @xmath7 , in particular they do not get enhanced if @xmath332 .
therefore , one has necessarily to consider @xmath333 , implying a full dl with all three degenerate rh neutrino masses .
the function @xmath290 is now obtained from the general expression ( [ g ] ) for @xmath334 and @xmath335 , or explicitly g(m_1,_31,_13 , ) & & 2k_atm(k_1+k_3)k_1k_3 ( m_3-m_1)m_atm^2 _31 + & & [ gr13 ] _ ( k_1+k_2+k_3 ) im[u^_1u_3 ] .
it is interesting to notice that in this case an @xmath336-dominance is realized .
moreover , one has that the dependence of @xmath337 on @xmath1 and @xmath0 is slight and the maximum is again for @xmath316 and for @xmath193 and one finds @xmath338 for normal hierarchy and @xmath339 for inverted hierarchy , so that the general expression ( [ lbm1 ] ) for the lower bound on @xmath8 gives , at @xmath99 for normal and inverted hierarchy , m_1 5.510 ^ 9gev_31|| m_1 5 10 ^ 11gev_31|| , while the general expression ( [ lbsint ] ) in resonant leptogenesis gives _
132.310 ^ -7d _
131.510 ^ -5d . increasing @xmath264 , the value of @xmath340 decreases and the lower bound on @xmath3 in resonant
leptogenesis becomes more and more restrictive .
this dependence is shown in fig .
5 both for normal ( left panel ) and inverted ( right panel ) hierarchy and for @xmath341 ( solid line ) and @xmath342 ( short - dashed line ) . very interestingly , imposing the experimental ( @xmath99 ) upper bound @xmath343 , one obtain the upper bound @xmath344 , depending on the value of @xmath307 .
this upper bound will become more stringent if the experimental upper bound on @xmath3 will improve , as expected in future experiments in the case of no discovery .
the most stringent experimental upper bound that can be hopefully reached in future with neutrino factories is approximately @xmath345 @xcite .
this asymptotical upper bound is also shown in fig . 5 and
would imply an upper bound @xmath346 for normal hierarchy and @xmath347 for inverted hierarchy .
therefore , an interesting interplay between two measurable quantities is realized and this makes @xmath0-leptogenesis falsifiable independently of the rh neutrino mass spectrum . in the more conservative case of normal hierarchy , see left panel of fig . 5 , a good approximation is given by the fit m_10.6 ( _ 13- 2.310 ^ -7 ) ^0.25 ev .
it is interesting that this upper bound holds in the extreme case of resonant leptogenesis and therefore holds for any rh neutrino spectrum .
however , we have to verify whether it holds also for a different choice of @xmath88 .
the situation for @xmath155 is quite different compared to the previous cases .
now one has @xmath335 and @xmath348 and it is possible to have both a partial dl with @xmath349 and a full dl . in the first case , the general expression eq .
( [ g ] ) becomes g(m_1,_21,_13 , ) & & 2k_atm(k_1+k_2)k_1k_2 ( m_2-m_1)m_atm^2 _21 + & & [ gr21 ] _ ( k_1+k_2 ) im[u^_1u_2 ] .
this time the contribution from the electron flavor vanishes .
furthermore , for normal hierarchy , there is an almost perfect cancellation between the @xmath123 and the @xmath124 contribution . in the left panel of fig .
6 we show the lower bound on @xmath3 versus @xmath264 and one can see how , compared to the previous case @xmath156 , this is much more restrictive .
in particular , imposing @xmath350 , one obtains now a much more stringent upper bound @xmath351 . on the other hand , for inverted hierarchy , the cancellation between the @xmath123 and the @xmath323 flavor does not occur and one has a lower bound on @xmath3 , for @xmath352 , shown in the right panel of fig .
6 , that is very similar to what has been obtained in the case @xmath156 .
however , now there is no flavor cancellation for increasing values of @xmath264 , because @xmath353 does not tend to a common value like @xmath354 .
therefore , one can see in fig . 6
that this time the upper bound on @xmath264 is much looser , both compared to normal hierarchy and compared to @xmath156 . in the full dl , the flavor cancellation at large @xmath264 occurs and the results
are shown in fig .
one can see how now for normal hierarchy the upper bound on @xmath264 is even much more restrictive and , for inverted hierarchy , one has a situation that is similar to the case @xmath156 .
in this case the lightest rh neutrino decouples and @xmath249 , independently of @xmath8 . therefore , there is no contribution to the final asymmetry from @xmath6 decays . on the other hand @xmath153 and @xmath157 do not vanish and therefore there is a contribution from the decays of the two heavier rh neutrinos . still @xmath6 inverse processes have to be taken into account since they contribute to the wash - out .
there are two different possibilities . in a full dl
the wash - out from @xmath6 inverse decays just cumulates with the wash - out from the two heavier .
therefore , this time , in the expression eq .
( [ g ] ) , one has @xmath355 and @xmath334 and @xmath356 , explicitly g(m_1,_32,_13 , ) & & 2k_atm(k_2+k_3)k_2k_3 ( m_3-m_2)m_atm^2 _32 + & & [ g32a ] _ ( k_1+k_2+k_3 ) im[u^_2u_3 ] . in fig .
8 we show the dependence of the @xmath3 lower bound on @xmath264 . this time there is a bigger suppression than in the case @xmath156 , both for normal and for inverted hierarchy .
in the case @xmath273 one has g(m_1,_32,_13 , ) & & 2k_atm(k_2+k_3)k_2k_3 ( m_3-m_2)m_atm^2 _32 + & & [ g32b ] _ ( k_2+k_3 ) e^-38k_1 im[u^_2u_3 ] .
the dependence of the lower bound on @xmath3 on @xmath264 is shown in fig .
9 for normal hierarchy . in this case
the upper bound on @xmath264 is now slightly less stringent than in the previous cases . for inverted hierarchy the final asymmetry production is so suppressed that there is no allowed region .
we can conclude this section noticing that these results show that @xmath0-leptogenesis can be falsified . in the case of normal hierarchy ,
the current upper bound @xmath357 implies @xmath358 , while , in future , a potential upper bound @xmath359 would imply @xmath360 , with a more precise determination depending on the possibility of improving the current estimation of the parameter @xmath307 in resonant leptogenesis .
the most attractive feature of @xmath0-leptogenesis is that a non - vanishing dirac phase , the only see - saw phase that we can realistically hope to discover in future , acts as the only source of @xmath2 violation responsible for the matter - antimatter asymmetry of the universe . we think that this feature , despite of the objections that we are going to discuss , provides a strong motivation for @xmath0-leptogenesis .
as we have seen , successful @xmath0-leptogenesis implies stringent conditions on the rh neutrino masses , something quite interesting since they escape conventional experimental information .
in particular we have seen that , except for a marginal allowed region in the weak wash - out regime , the hl is non - viable .
we also observed that a definite conclusion on the existence of such a marginal allowed region , requires a full quantum kinetic treatment but in any case corrections are expected to shrink this already quite restricted allowed region .
therefore , @xmath0-leptogenesis motivates models with degenerate rh neutrino masses , with the most extreme limit represented by resonant leptogenesis . even in this extreme limit
imposing successful @xmath0-leptogenesis , interesting conditions follow on quantities accessible in low - energy neutrino experiment : @xmath3 , the absolute neutrino mass scale , normal or inverted scheme , the dirac phase itself .
therefore , an interesting aspect of @xmath0-leptogenesis is that it is falsifiable independently of the heavy neutrino mass spectrum .
there are some objections to @xmath0-leptogenesis .
there is no clear theoretical motivation for @xmath0-leptogenesis , more generally to choose a real orthogonal @xmath88 matrix .
apparently , sequential dominated models @xcite could represent an interesting theoretical framework . indeed in @xcite it was shown that these models correspond to have an @xmath88 matrix that slightly deviates from the unit matrix or from all the other five that can be obtained from the unit matrix exchanging rows or columns .
however , it has been noticed @xcite that in the limit @xmath361\rightarrow 0 $ ] total @xmath2 asymmetries @xmath362 do not necessarily vanish .
therefore , in this limit and taking vanishing majorana phases , one does not necessarily obtain @xmath0-leptogenesis .
writing @xmath363 $ ] , the correct condition to enforce @xmath364 is to take the limit @xmath365 .
this is a more demanding limit than @xmath361\rightarrow 0 $ ] and it is currently not motivated by generic sequential dominated models .
this limit is not motivated either by radiative leptogenesis @xcite within the context of the minimal flavor violation principle @xcite , as recently considered in @xcite .
therefore , there is no theoretical justification for @xmath0-leptogenesis at the moment .
another possible objection to @xmath0-leptogenesis is that it can not be distinguished from the general scenario , where all phases are present , even if a non - vanishing dirac phase is discovered . indeed
a dirac phase would give in this case a subdominant contribution .
this objection is however related also to the first one . indeed , since a theoretical model motivating @xmath0-leptogenesis is required anyway
, one can hope to find some specific prediction that makes the model testable and @xmath0-leptogenesis together with it .
dirac phase leptogenesis would then become distinguishable from the general scenario , though in an indirect way .
this last objection can be also considered within a more particular case where @xmath88 is still real but majorana phases are present together with the dirac phase .
it has been noticed that the contribution to the final asymmetry from majorana phases is in general dominant compared to that one coming from the dirac phase @xcite . in the right panel of fig .
1 we have compared the result on the @xmath8 lower bound for @xmath156 obtained in @xmath0-leptogenesis with the result when @xmath366=\d=0 $ ] but @xmath367 ( dotted lines ) .
one can see that in the second case the lower bound is @xmath368 times more relaxed .
this result can be easily understood analytically @xcite and actually it can be also observed that there can be exact cancellations between the contribution to the final asymmetry from the majorana phases and from the dirac phase .
the presence of cancellations can be somehow regarded as a limit to @xmath0-leptogenesis main motivation , since even though a dirac phase will be discovered , it is not guaranteed that the observed asymmetry can be explained .
this objection is however quite weak since it would be quite strange if nature disposed a sufficient source of @xmath2 violation but set up a second source that exactly cancels with the first one while the observed asymmetry is , in the end , explained still by a third one , for example the phases in @xmath88 . on the other hand
, we can say that it would be certainly positive for @xmath0-leptogenesis if in future experimental upper bounds on the majorana phases are placed , for example from @xmath369 decay , thus constraining the contribution to the final asymmetry from majorana phases @xcite .
this can be also regarded as a further prediction coming from @xmath0-leptogenesis . in conclusion , we have studied in detail a specific scenario of leptogenesis that is interesting especially in view of the many next planned experiments aiming at a discovery of @xmath2 violation in neutrino mixing . despite some important remarks and objections
, we think that @xmath0-leptogenesis realizes a very interesting link between a long - standing cosmological puzzle and @xmath2 violation in neutrino oscillations , one of the most relevant experimental topics in high - energy physics during next years . *
acknowledgments * + it is a pleasure to thank s. petcov for discussions during the neutrino oscillation workshop 2006 , held in conca della specchiulla ( italy ) . we also wish to thank g. raffelt for interesting discussions and comments . this work was supported , in part , under the marie curie project `` leptogenesis , seesaw and guts , '' contract no . meif - ct-2006 - 022950 .
p. minkowski , phys .
b * 67 * ( 1977 ) 421 ; t. yanagida , in _ workshop on unified theories _ ,
kek report 79 - 18 ( 1979 ) p. 95 ; m. gell - mann , p. ramond , r. slansky , in _
( north holland , amsterdam , 1979 ) eds .
van nieuwenhuizen , d. freedman , p. 315
glashow , in _ 1979 cargese summer institute on quarks and leptons _ ( plenum press , new york , 1980 ) p. 687
; r. barbieri , d. v. nanopoulos , g. morchio and f. strocchi , phys .
b * 90 * ( 1980 ) 91 ; r. n. mohapatra and g. senjanovic , phys .
* 44 * ( 1980 ) 912 .
f. x. josse - michaux and a. abada , arxiv : hep - ph/0703084 . m. y. khlopov and a. d. linde , phys .
b * 138 * ( 1984 ) 265 .
j. r. ellis , j. e. kim and d. v. nanopoulos , phys .
b * 145 * ( 1984 ) 181 ; k. kohri , t. moroi and a. yotsuyanagi , phys .
d * 73 * ( 2006 ) 123511 . j. a. casas and a. ibarra , nucl .
b * 618 * ( 2001 ) 171 .
s. antusch and a. m. teixeira , jcap * 0702 * ( 2007 ) 024 .
s. pascoli , s. t. petcov and a. riotto , phys .
d * 75 * ( 2007 ) 083511 .
s. pascoli , s. t. petcov and a. riotto , arxiv : hep - ph/0611338 .
g. c. branco , r. gonzalez felipe and f. r. joaquim , phys .
b * 645 * ( 2007 ) 432 .
a. pilaftsis , int .
j. mod .
a * 14 * ( 1999 ) 1811 .
s. uhlig , arxiv : hep - ph/0612262 .
r. gonzalez felipe , f. r. joaquim and b. m. nobre , phys .
d * 70 * ( 2004 ) 085009 ; g. c. branco , r. gonzalez felipe , f. r. joaquim and b. m. nobre , phys .
b * 633 * ( 2006 ) 336 .
v. cirigliano , b. grinstein , g. isidori and m. b. wise , nucl .
b * 728 * ( 2005 ) 121 .
k. s. babu , c. n. leung and j. t. pantaleone , phys .
b * 319 * ( 1993 ) 191 ; s. antusch , j. kersten , m. lindner and m. ratz , nucl
. phys .
b * 674 * ( 2003 ) 401 .
s. eidelman et al . , phys .
b592 * , 1 ( 2004 ) ( url : http://pdg.lbl.gov/ ) .
m. fujii , k. hamaguchi and t. yanagida , phys .
d * 65 * ( 2002 ) 115012 .
p. di bari , aip conf .
* 655 * ( 2003 ) 208 [ arxiv : hep - ph/0211175 ] .
p. di bari , [ arxiv : hep - ph/0406115 ] .
w. buchmller , p. di bari and m. plmacher , phys .
b * 547 * ( 2002 ) 128 .
a. de simone and a. riotto , jcap * 0702 * ( 2007 ) 005 .
o. vives , phys .
d * 73 * ( 2006 ) 073006 .
g. engelhard , y. grossman , e. nardi and y. nir , arxiv : hep - ph/0612187 .
t. shindou and t. yamashita , arxiv : hep - ph/0703183 . j. f. nieves and p. b. pal , phys .
d * 36 * ( 1987 ) 315 .
a. anisimov , a. broncano and m. plumacher , nucl .
b * 737 * ( 2006 ) 176 .
p. huber , m. lindner and w. winter , nucl .
b * 645 * ( 2002 ) 3 .
see s. f. king , rept .
* 67 * ( 2004 ) 107 and references therein . | we discuss the conditions for a non - vanishing dirac phase @xmath0 and mixing angle @xmath1 , sources of @xmath2 violation in neutrino oscillations , to be uniquely responsible for the observed matter - antimatter asymmetry of the universe through leptogenesis .
we show that this scenario , that we call @xmath0-leptogenesis , is viable when the degenerate limit ( dl ) for the heavy right - handed ( rh ) neutrino spectrum is considered .
we derive an interesting joint condition on @xmath3 and the absolute neutrino mass scale that can be tested in future neutrino oscillation experiments . in the limit of hierarchical heavy rh neutrino spectrum ( hl ) ,
we strengthen the previous result that @xmath0-leptogenesis is only very marginally allowed , even when the production from the two heavier rh neutrinos is taken into account .
an improved experimental upper bound on @xmath3 and ( or ) an account of quantum kinetic effects could completely rule out this option in the future .
therefore , @xmath0-leptogenesis can be also regarded as a motivation for models with degenerate heavy neutrino spectrum .
c i u
# 1 # 1#1 # 1 # 1#2#3phys .
lett .
* b # 1 * ( # 2 ) # 3 # 1#2#3nucl .
phys .
* b # 1 * ( # 2 ) # 3 # 1#2#3phys .
rev .
lett . *
# 1 * ( # 2 ) # 3 # 1#2#3phys .
rev . * d # 1 * ( # 2 ) # 3 # 1#2#3z .
phys .
* c # 1 * ( # 2 ) # 3 # 1#2#3class . and quantum grav
. * # 1 * ( # 2 ) # 3 # 1#2#3commun . math .
phys . *
# 1 * ( # 2 ) # 3 # 1#2#3j . math .
phys . *
# 1 * ( # 2 ) # 3 # 1#2#3ann . of phys .
* # 1 * ( # 2 ) # 3 # 1#2#3phys . rep . * # 1c * ( # 2 ) # 3 # 1#2#3progr .
theor .
phys . *
# 1 * ( # 2 ) # 3 # 1#2#3int .
j. mod .
phys . * a # 1 * ( # 2 ) # 3 # 1#2#3mod .
phys .
lett . * a # 1 * ( # 2 ) # 3 # 1#2#3nuovo cim . *
# 1 * ( # 2 ) # 3 # 1#2#3_ibid .
_ * # 1 * ( # 2 ) # 3 |
do events have already been simulated by a variety of models , ranging from simple conceptual ones to earth system models of intermediate complexity ( emics ) .
conceptual models are most suitable to perform large numbers of long - term investigations because they require very little computational cost .
however , they are often based on ad - hoc assumptions and only consider processes in a highly simplified way .
in addition to that , the number of adjustable parameters is typically large compared to the degrees of freedom in those models .
this implies that seemingly good results can often be obtained merely by excessive tuning .
nevertheless , conceptual models can provide important help for the interpretation of complex climatic processes .
the gap between conceptual models and the most comprehensive general circulation models ( gcms ) , which are not yet applicable for millennial - scale simulations because of their large computational cost , is bridged by emics @xcite .
emics include most of the processes described in comprehensive models ( in a more reduced form ) , and interactions between different components of the earth system ( atmosphere , hydrosphere , cryosphere , biosphere , etc . )
are simulated .
the number of degrees of freedom typically exceeds the number of adjustable parameters by orders of magnitude . since
many emics are fast enough for studies on the multi - millennial time scale , they are the most adequate tools for the simulation of do events .
the simple conceptual model which we use here is an extended version of the model described by @xcite ( in the supplementary material of that publication ) . here
we use the model to demonstrate and analyse two apparently counterintuitive resonance phenomena ( _ stochastic resonance _ and _ ghost resonance _ ) that can exist in a large class of highly nonlinear systems .
due to the complexity of many of those systems it is often impossible to precisely identify the reasons for the occurrence of these resonance phenomena .
our conceptual model , in contrast , has a very clear forcing - response relation as well as a very low computational cost and thus provides a powerful tool to explore these phenomena and to test their robustness .
furthermore , we describe and discuss the applicability of the model for improved statistical analyses ( i.e. monte - carlo simulations ) on the regularity of do events . in the following the key assumptions of the conceptual model are first discussed . in the supplementary material
we then compare the model performance under a number of systematic forcing scenarios with the performance of a more comprehensive model ( the emic climber-2 ) , compare supplementary information file and supplementary figs . 1 - 6 . in the framework of the conceptual model we finally demonstrate and interpret two hypotheses that were previously suggested in order to explain the recurrence time of do events , and we discuss how these hypotheses could be tested
our conceptual model is based on three key assumptions : 1 .
do events represent repeated transitions between two different climate states , corresponding to warm and cold conditions in the north atlantic region .
these transitions are rapid compared to the characteristic life - time of the two climate states ( i.e. in first order approximation they occur instantaneously ) and take place each time a certain threshold is crossed .
3 . with every transition between the two states the threshold overshoots and afterwards approaches equilibrium following a millennial - scale relaxation process .
this implies that the conditions for a switch between both states ameliorate with increasing duration of the cold and warm intervals .
our three assumptions are supported by paleoclimatic evidence and/or by simulations with a climate model : 1 . since long , do events have been regarded as repeated oscillations between two different climate states @xcite . it has been suggested that these states are linked with different modes of operation of the thc @xcite .
this seminal interpretation has since then influenced numerous studies and is now generally accepted @xcite .
indirect data indicate that the glacial thc indeed switched between different modes of operation @xcite which , according to their occurrence during cold and warm intervals in the north atlantic , were labelled stadial and interstadial modes . a third mode named heinrich mode ( because of its presence during the so - called heinrich events ) is not relevant here .
high - resolution paleoclimatic data show that transitions from cold conditions in the north atlantic region to warm ones often happened very quickly , i.e. on the decadal - scale or even faster @xcite .
the opposite transitions were slower , i.e. on the century - scale @xcite , but nevertheless still faster that the characteristic life - time of the cold and warm intervals ( which is on the centennial to multi - millennial time scale , compare fig .
the abruptness of the shifts from cold conditions to warm ones has commonly been interpreted as evidence for the existence of a critical threshold in the climate system that needs to be crossed in order to trigger a shift between stadial and interstadial conditions @xcite .
such a threshold could be provided by the thc ( more precisely , by the process of deep - water formation ) : when warm and salty surface water from lower latitudes cools on its way to the north atlantic , its density increases .
if the density increase is large enough ( i.e. if the surface gets denser than the deeper ocean water ) , surface water starts to sink .
otherwise , surface water can freeze instead of sinking .
the onset of deep - water formation can thus hinder sea - ice formation and facilitate sea - ice melting ( due to the vertical heat transfer between the surface and the deeper ocean ) .
a switch between two fundamentally different modes of deep - water formation can thus dramatically change sea ice cover and can cause large - scale climate shifts .
such nonlinear , threshold - like transitions between different modes of deep - water formation are at present considered as the most likely explanation for do events @xcite .
3 . the time - evolution of greenland temperature during the warm phase of do events has a characteristic saw - tooth shape ( fig .
[ fig : figure1 ] ) .
highest temperatures typically occur during the first decades of the events .
these temperature maxima are followed by a gradual cooling trend over centuries / millennia , before the system returns to cold conditions at the end of the events .
this asymmetry supports the idea that the system overshoots in some way during the abrupt warming at the beginning of the events and that the subsequent cooling trend represents a millennial - scale relaxation towards a new equilibrium @xcite .
we note that the time - evolution of greenland temperature provides no clear evidence for an overshooting during the opposite transitions ( i.e. from the warm state back to the cold one ) .
this , however , is not necessarily in contradiction to our assumption : this lack of an overshooting in the temperature fields does not necessarily mean that the ocean - atmosphere system did not overshoot , since greenland temperature evolution in the stadial state might have been dominated by factors other than the thc ( respectively its stability ) , e.g. by greenland ice accumulation , which would mask the signature of the thc in the ice core data .
+ we will show later ( in sect . 3.3 . )
that the assumption of an overshooting in the stability of the system is in fact strengthened by the analysis of model results obtained with the coupled model climber-2 . in that model the overshooting results from the dynamics of the transitions between the two climate states : in the stadial state deep convection occurs south of the greenland - scotland ridge ( i.e. at about 50 @xmath0n ) . in the interstadial state
, however , deep convection takes place north of the ridge ( i.e. at about 65 @xmath0n ) .
the onset of deep convection north of the greenland - scotland ridge , which releases accumulated energy to the atmosphere ( i.e. heat that is stored mainly in the deep ocean ) , in first place starts do events in the model .
this heat release leads to a reduction of sea ice , which in turn further enhances sea surface densities between 50 @xmath0n and 65 @xmath0n ( e.g. by increased local evaporation and reduced sea ice transport into that area ) .
as a result deep convection also starts between 50 @xmath0n and 65 @xmath0n , and much more heat can be released to the atmosphere . without a further response of the thc
the system would return quickly ( within years or decades , i.e. with the convective time scale ) to its original state . in climber-2 , however , the changes in deep convection trigger a northward extension and also an intensification of the ocean circulation ( i.e. an overshooting of the atlantic meridional overturning circulation ; compare @xcite ) , which maintains the interstadial climate state since it is accompanied by an increase in the salinity and heat flux to the new deep convection area ( at about 65 @xmath0n ) . in response to the overshooting of the overturning circulation , the system relaxes slowly ( within about 1000 years , i.e. with the advective time scale ) towards the stadial state .
we note that the advective time scale corresponds to the millennial relaxation time in our conceptual model .
the model climber-2 also supports the validity of our overshooting assumption during the opposite transition ( from the warm state back to the cold one ) , as we will show in sect .
we would like to stress that our interpretation of the processes during do events is , of course , not necessarily true since we can only speculate that the underlying mechanism of the events is correctly captured by climber-2 .
we implement the above assumptions in the following way ( compare fig .
[ fig : figure2 ] ) : first we define a discrete index s(t ) that indicates the state of the system at time t ( in years ) . since we postulate the existence of two states , s can only take two values ( s=1 : warm state , s=0 : cold state ) .
we further define a threshold function t(t ) that describes the stability of the system at time t ( i.e. the stability of the current model state ) .
dynamics of our conceptual model .
shown is the time evolution of the model , in response to a forcing that is large enough to trigger switches between both model states .
top : forcing f ( black ) and threshold function t ( red ) .
bottom : model state s ( grey ; s=0 corresponds to the cold state , s=1 to the warm one ) and state variable s ( green ) . at time
@xmath1 the forcing falls below the threshold function and a shift from the cold state into the warm one is triggered . with this transition ,
the threshold function switches to a non - equilibrium value ( representing an overshooting of the system ) and afterwards approaches equilibrium following a millennial - scale relaxation . at time
@xmath2 the forcing exceeds the threshold function , and a transition from the warm state back into the cold one is triggered . with this transition
, the threshold function switches to another non - equilibrium value and approaches equilibrium following another millennial - scale relaxation , until the forcing again falls below the threshold function and the next switch into the warm state is triggered .
note that the state variable s is chosen to be identical to the threshold function t. for convenience , discontinuities in t and s are eliminated by linear interpolation .
t , s and f are normalised in the figure . ,
width=302 ] second we define rules for the time evolution of the threshold function t. when the system shifts its state , we assume a discontinuity in the threshold function : with the switch from the warm state to the cold one ( at time t in fig .
[ fig : figure2 ] ) t takes the value a@xmath3 . likewise , with the switch from the cold state into the warm one ( at time t in fig .
[ fig : figure2 ] ) t takes the value a@xmath4 .
as long as the system does not change its state the evolution of t is assumed to be given by a relaxation process : @xmath5 ( s labels the current model state , @xmath6 denotes the relaxation time in that state , b@xmath7 is a state - dependent constant that labels the equilibrium value of t in each model state ) .
these assumptions result in the following expression for the threshold function t : @xmath8 note that in the above expression the index s again denotes the current state of the model ( i.e. s=0 stands for the cold state and s=1 for the warm one ) , @xmath9 labels the time of the last switch from the warm state into the cold one , and @xmath10 indicates the time of the last switch from the cold state into the warm one .
third we assume that transitions from one state to the other are triggered each time a given forcing function f(t ) crosses the threshold function t. more precisely , we assume that when the system is in the cold state ( @xmath11 = 0 $ ] ) and the forcing is smaller than the threshold value ( @xmath12 < t[t''+1]$ ] ) the system switches into the warm state ( @xmath13 = 1 $ ] ) .
this shift marks the start of a do event . likewise ,
when the system is in the warm state ( @xmath14 = 1 $ ] ) and the forcing is larger than the threshold value ( @xmath15 > t[t'+1]$ ] ) the system switches into the cold state ( @xmath16 = 0 $ ] ) .
that shift represents the termination of a do event .
if none of these conditions is fulfilled , the system remains in its present state ( i.e. @xmath17 = s[t]$ ] ) . to simplify the comparison of the model output with paleoclimatic records
we further define a state variable s , which represents anomalies in greenland temperature during do events . for simplicity
we assume that the state variable is equal to the threshold function : @xmath18 ( i.e. we assume that greenland temperature evolution during do events is closely related to the current state of the thc , respectively to its stability ) .
we stress that this assumption is of course highly simplified , because greenland temperature is certainly not only influenced by the thc but also by other processes such as changes in ice accumulation during do oscillation . however , this assumption is not crucial for the dynamics of our model , since the timing of the switches between both model states is solely determined by the relation between the forcing function f and the threshold function t. this means that even if we included a more realistic relation between t and s , the timing of the simulated climate shifts would be unchanged and the model dynamics would thus essentially be invariant . note
that @xmath9 and @xmath10 are not adjustable ; they rather represent internal time markers .
thus , six adjustable parameters exist in our model as described here , namely @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 and @xmath24 .
our choice for these parameters is shown in table [ tab : table1 ] . with these parameter values
the system is bistable ( i.e. no transition is ever triggered in the absence of any forcing , since @xmath25 and @xmath26 ) and almost symmetric .
that means that the average duration of the simulated warm and cold intervals is almost equal . when compared with greenland paleotemperature records this situation most likely corresponds to the time interval between about 27000 and 45000 years before present , during which the duration of the cold and warm intervals in do oscillations was also comparable ( fig .
1 ) . the model can , however , also represent an unstable ( for @xmath27 and @xmath28 ) or a mono - stable system , in which the stable state is either the warm one ( for @xmath27 and @xmath26 ) or the cold one ( for @xmath25 and @xmath28 ) ; when compared to the ice core data this situation is closer to the time interval between 15000 and 27000 years before present , since during that time the system was preferably in its cold state and the forcing apparently crossed the threshold only infrequently and during short periods of time . [
tab : table1 ] cll parameter & chosen value + a@xmath3 & -27 + a@xmath4 & 27 + b@xmath3 & -9.7 + b@xmath4 & 11.2 + @xmath23 & 1200 + @xmath24 & 800 + in order to test our conceptual model we compare its performance under a number of systematic forcing scenarios with the performance of the far more comprehensive model climber-2 ( a short description of that model is given in the appendix ; a detailed description exists in the publication of @xcite ) .
analogous to @xcite , we investigate the response of both models to a forcing that consists of two century - scale sinusoidal cycles . in the conceptual model , the forcing is implemented as the forcing function f. in the emic , the forcing is added to the surface freshwater flux in the latitudinal belt 50 - 70 @xmath29n , following @xcite and @xcite .
this anomaly changes the vertical density gradient in the ocean and can thus trigger do events .
switches from the cold state into the warm one are excited by sufficiently large ( order of magnitude : a few centimetre per year in the surface freshwater flux into the relevant area of the north atlantic ) negative freshwater anomalies ( i.e. by positive surface density anomalies that are strong enough to trigger buoyancy [ deep ] convection ) , and the opposite switches are triggered by sufficiently large positive freshwater anomalies ( i.e. by negative surface density anomalies that are strong enough to stop buoyancy [ deep ] convection ) .
this justifies our choice for the logical relations that govern the dynamics of the transitions in the conceptual model ( i.e. @xmath30 as the condition for the switch from the cold state to the warm one , @xmath31 for the opposite switch ) .
a detailed comparison between both model outputs is presented in the supplementary material .
we here only summarise the main results : we find a general agreement between both models , which is robust when the forcing parameters are varied over some range ( supplementary figs . 1 - 6 ) .
the conceptual model reproduces the existence of three different regimes ( _ cold _ , _ warm _ , _ oscillatory _ ) in the output of the emic and also their approximate position in the forcing parameter - space . by construction
only the nonlinear component in the response of the emic to the forcing is reproduced by the conceptual model ( this component represents the saw - tooth shape of do events ) .
a second , more linear component is not included in the conceptual model ( this component represents small - amplitude temperature anomalies which are superimposed on the saw - tooth shaped events in the emic ) . in particular , the conceptual model very well reproduces the timing of the onset of do events in the emic .
the fact that our conceptual model , despite its simplicity , agrees in so many aspects with the much more detailed model climber-2 suggests that it indeed captures the key features in the dynamics of do events in that model .
we would like to stress that the output of the emic indeed supports our assumption of an overshooting in the stability of the system during the transitions between both climate states : when driven by a periodic forcing ( with a period of 1470 years ) , the emic can show periodic oscillations during which it remains in either of its states for more than one forcing period ( i.e. for considerably more than 1470 years , compare supplementary fig .
this implies that ( at least in the emic ) the conditions for a return to the opposite state indeed ameliorate with increased duration of the cold or warm intervals .
if the thresholds in the model were constant ( or gradually increasing with increasing duration of the simulated cold / warm intervals ) , in contrast , the duration of the cold and warm intervals during the simulated oscillation could never be longer than 1470 years : if a periodic forcing does not cross a constant ( or gradually increasing ) threshold within its first period , it never crosses the threshold , due to the periodicity of the forcing .
strongly nonlinear systems can show complex and apparently counterintuitive resonance phenomena that can not occur in simple linear systems . in this section
we use our conceptual model to demonstrate and to discuss two of these phenomena , i.e. stochastic resonance ( sr ) and ghost resonance ( gr ) . since the explanation of the 1470-year cycle ( and in fact even its significance ) is still an open question ,
we further discuss how future tests could distinguish between the proposed mechanisms .
stochastic resonance .
the input consists of : 1 .
a sub - threshold sinusoidal signal with a period of 1470 years and an amplitude of 4.5 msv ( about 40 percent of the threshold value b@xmath4 above which do events occur in the model ) , 2 . a random gaussian - distributed signal with white noise power signature ( standard deviation @xmath32 = 8 msv ) and a cutoff frequency of 1/(50 years ) .
the cutoff is used since no damping exists in the model and it thus shows an unrealistically large sensitivity to high - frequency ( i.e. decadal - scale or faster ) forcing . a : total input ( black ) ,
periodic input component ( grey ) , model output ( green ) .
dashed lines are spaced by 1470 years .
b : relative frequency to obtain a spacing of 1470 years @xmath3310% ( triangles ) respectively @xmath3320% ( squares ) between successive events , as a function of the noise level @xmath32 . , width=321 ] in linear systems that are driven by a periodic input , the existence of noise generally reduces the regularity of the output ( e.g. the coherence between the input and the output ) .
this is not necessarily the case in nonlinear systems : excitable or bistable nonlinear systems with a threshold and with noise , which are driven by a sinusoidal sub - threshold input , can show maximum coherence between the input and the output for an intermediate noise level , for which the leading output frequency of the system is close to the input frequency .
this phenomenon is called stochastic resonance ( sr ) @xcite .
sr has been suggested to explain the characteristic timing of do events @xcite , i.e. the apparent tendency of the events to occur with a spacing of about 1470 years or integer multiples thereof .
it has further been demonstrated that do events in the model climber-2 can be subject to sr @xcite .
here we apply our conceptual model to reproduce these results and to reanalyse the underlying mechanism .
we use an input that is composed of : ( i ) a sinusoidal signal with a period of 1470 years , ( ii ) additional white noise .
figures [ fig : figure3 ] and [ fig : figure4 ] show that for a suitable noise level the model can indeed show do events with a preferred spacing of about 1470 years or integer multiples thereof .
the reason for this pattern in the output is easily understandable in the context of the model dynamics : do events in the model are triggered by pronounced minima of the total input ( total input = periodic signal plus noise ) .
these minima generally cluster around the minima of the sinusoidal signal , and the start of the simulated events thus has a tendency to coincide with minima of the sinusoidal signal ( fig .
[ fig : figure3]a ) .
some minima of the sinusoidal signal , however , are not able to trigger an event , because the magnitude of the noise around these minima is too small so that the threshold function is not reached by the total input .
consequently , a cycle is sometimes missed , and the spacing of successive events can change from about 1470 years to multiples thereof . unlike the model climber-2 ( which has a complex relationship between the input and the output and also a large computational cost ) our conceptual model can be used for a detailed investigation of the sr , e.g. because the dynamics of the model is simple and precisely known and because probability measures ( such as waiting time distributions ) can be explicitly computed .
in fact , the resonant pattern in the conceptual model ( fig .
[ fig : figure3 ] ) is due to two time - scale matching conditions : the noise level is such that the average waiting time between successive noise - induced transitions is comparable to _ half _ of the period of the periodic forcing , and also comparable to the relaxation times @xmath23 and @xmath24 of the threshold function ( compare fig . [
fig : figure4]b ) .
this situation is different from the usual sr , in which thresholds ( or potentials ) are constant in time ( apart from the influence of the periodic input ) . in the usual sr , only one time - scale matching condition exists @xcite , namely the one that the average waiting time between successive noise - induced transitions ( i.e. the inverse of the so - called kramers rate ) is comparable to _ half _ of the period of the periodic forcing . in order to investigate the implications of the second condition we simulate histograms for four different scenarios in the conceptual model ( fig .
[ fig : figure4 ] ) : 1 .
noise - only input , constant threshold ( fig .
[ fig : figure4]a ) ; 2 .
noise - only input , overshooting threshold ( fig .
[ fig : figure4]b ) ; 3 .
noise plus periodic input , constant threshold ( fig .
[ fig : figure4]c ) ; 4 . noise plus periodic input ,
overshooting threshold ( fig .
[ fig : figure4]d ) .
we note that 3
. corresponds to the usual sr , while 4 . describes our _ overshooting stochastic resonance_.
as can be seen from the histograms , the existence of the millennial - scale relaxation process leads to a synchronisation in the sense that the waiting times between successive events are confined within a much smaller time interval ( about 1000 - 4500 years with the overshooting , compared to about 0 - 10000 years without overshooting ) .
this confinement is plausible since the transition probability between both model states strongly depends on the magnitude of the threshold , which declines with increasing duration of the cold or warm intervals : when the standard deviation of the noise level is chosen such that the average waiting time between successive noise - induced transitions is comparable to the relaxation times @xmath23 and @xmath24 , as in fig .
[ fig : figure4 ] , the overshooting relaxation strongly reduces the transition probability for waiting times much smaller than the relaxation time ( since the corresponding values of the threshold function are large ) and increases the transition probability for waiting times of the order of the relaxation time or larger ( since the corresponding values of the threshold function are considerably smaller ) . the probability to find an only century - scale spacing between successive events is thus small , because the corresponding transition probabilities are small . on the other hand , the probability to find a multi - millennial spacing is also small , because the states are already depopulated before ( i.e. the probability to obtain lifetimes considerably larger than the relaxation time is almost zero ) .
this explains why the possible values for the spacing between successive do events are restricted to a much smaller range than in the usual sr ( i.e. in the case with constant thresholds ) .
this synchronisation effect is indeed not unique to the conceptual model : the output of the coupled model climber-2 shows a similar pattern ( with possible waiting times between successive do events of e.g. about 1500 - 5000 years or about 1000 - 3000 years , depending on the noise level ; compare fig .
4a - d in the publication of @xcite ) .
this similarity is of course not surprising , since the conceptual model is apparently able to mimic the events in the emic and since an overshooting in the stability of both states clearly also exists in climber-2 ( compare sect .
we note that in the gisp2 ice core data , do events in the time interval 27000 - 45000 years before present ( which , as discussed in sect .
3.2 , is the best analogue to the `` background climate state '' in our conceptional model , since the duration of the cold and warm intervals in the data is comparable in that interval ) have spacings of about 1000 - 3000 years ( compare fig . 1 ) and were reported to cluster around values of either about 1470 years or about 2940 years @xcite . because the sr mechanism could explain such a pattern ( compare fig . [
fig : figure4]d ) it has originally been proposed .
however , this mechanism requires a sinusoidal input with a period of about 1470 years , which has so far not been detected . in linear systems which are driven by a periodic input ,
the frequencies of the output are always identical to the input frequencies .
this is not necessarily the case in nonlinear systems .
for example , nonlinear excitable ( or bistable ) systems that are driven by an input with frequencies corresponding to harmonics of a fundamental frequency ( which itself is not present in the input ) can show a resonance at the fundamental frequency , i.e. at a frequency with zero input power .
this phenomenon , which was first described in order to explain the pitch of complex sounds @xcite and later observed experimentally e.g. in laser systems @xcite , is called ghost resonance ( gr ) .
gr and sr can indeed occur in the same class of systems , e.g. in bistable or excitable systems with thresholds
. however , unlike sr , gr requires a periodic driver with more than one frequency .
although many geophysical systems might be subject to gr ( since the relevant processes often have thresholds ) , the occurrence of this mechanism has so far not expressly been demonstrated in geoscience . ghost resonance .
top : forcing ( black ) and model response ( green ) . middle : amplitude spectrum of the forcing .
bottom : amplitude spectrum of the model response .
we use two sinusoidal forcing cycles , with frequencies of 7/(1470 years ) and 17/(1470 years ) , respectively , and with equal amplitudes .
these two cycles coincide every 1470 years and create peaks of particularly pronounced magnitude , spaced by exactly that period .
thus , despite the fact that there is no spectral power at the corresponding frequency ( see middle panel ) , the forcing repeatedly crosses the threshold at those intervals .
consequently , the response of the conceptual model ( i.e. the time evolution of the state variable s ) shows strictly repetitive do events with a period of 1470 years ( as indicated by the dashed lines , which are spaced by 1470 years ) . despite the lack of a 1470-year spectral component in the forcing
, the output shows a very prominent peak at the corresponding frequency .
, width=321 ] here we discuss a hypothesis that was recently proposed to explain the 1/(1470 years ) leading frequency of do events @xcite
. the underlying mechanism of the hypothesis is in fact the first reported manifestation of gr in a geophysical model system . according to that hypothesis the 1/(1470 years )
frequency of do events could represent the nonlinear climate response to forcing cycles with frequencies close to harmonics of 1/(1470 years ) .
our conceptual model illustrates the plausibility of this mechanism : we use a bi - sinusoidal input with frequencies of 7/(1470 years ) and 17/(1470 years ) , i.e. with frequencies corresponding to the 7th and the 17th harmonic of a 1/(1470 years ) fundamental frequency , and with equal amplitudes . a spectral component corresponding to the fundamental frequency
is not explicitly present in the input .
since the two sinusoidal cycles correspond to harmonics of the missing fundamental , the input signal repeats with a period of 1470 years . for an appropriate range of input amplitudes
, the output of the conceptual models shows periodic do events with a period of 1470 years ( fig .
[ fig : figure5 ] ) .
unlike the input , the model output exhibits a pronounced frequency of 1/(1470 years ) , corresponding to the leading frequency of do events and to the fundamental frequency that is absent in the input .
this apparent paradox is explained by the fact that the two driving cycles enter in phase every 1470 years , thus creating pronounced peaks spaced by that period . because the magnitude of these peaks results from constructive interference of the two driving cycles , it is indeed robust that a threshold process can be much more sensitive to these peaks than to the two original driving cycles @xcite .
the main strength of the gr mechanism is that unlike the sr mechanism it can relate the leading frequency of do events to a main driver of natural ( non - anthropogenic ) climate variability , since proxies of solar activity suggest the existence of solar cycles with periods close to 1470/7 ( = 210 ) years ( de vries or suess cycle ) and 1470/17 ( @xmath3486.5 ) years ( gleissberg cycle ) @xcite .
so far , however , no empirical evidence for this mechanism has been found @xcite , nor has it been shown yet that changes in solar activity over the solar cycles are sufficiently strong to actually trigger do events . in order to investigate the stability of this mechanism
we further add a stochastic component ( i.e. white noise ) to the forcing .
in this case the events are of course not strictly periodic anymore .
similar to the sr case , an optimal ( i.e. intermediate ) noise level exists for which the waiting time distribution of the simulated events exhibits a maximum at a value of 1470 years , corresponding to the period of the fundamental frequency of the two input cycles ( fig .
[ fig : figure6 ] ) .
in contrast to the sr case , in which a fairly simple waiting time distribution with a few broad maxima of century scale width is obtained ( compare fig . [
fig : figure4]d ) , we now find a much more complex pattern with a large number of very sharp lines of only decadal scale width . since the waiting time distributions of both mechanisms are considerably different , it could at least in principle
be possible to distinguish between both mechanisms by analysing the distribution of the observed do events . in practise , however , this approach is complicated by the fact that only about ten events appear to be sufficiently well dated for this kind of analysis , and even their spacing has an uncertainty of about 50 years @xcite , which is already of the same order as the width of the peaks in fig .
[ fig : figure6]b .
we note that the mechanism that is described in fig .
[ fig : figure6 ] is known as ghost stochastic resonance ( gsr ) , and its occurrence and robustness has already been reported before in other systems with thresholds and multiple states of operation @xcite . at least in our system , however , this mechanism is even more complex than the other two types of resonance ( sr , gr ) .
it is beyond the scope of this paper to describe the gsr mechanism in more detail .
ghost stochastic resonance .
the input consists of : 1 .
two sinusoidal forcing cycles with frequencies of 7/(1470 years ) and 17/(1470 years ) , respectively , and with an amplitude of 8 msv , 2 . a random gaussian - distributed signal with white noise power signature and a cutoff frequency of 1/(50 years ) , as in fig . [ fig : figure3 ] .
in a , the relative frequency to obtain a spacing of 1470 years @xmath331% ( triangles ) respectively @xmath332% ( squares ) between successive events is shown as a function of the noise level @xmath32 .
b shows the distribution of the spacing @xmath35 t between successive events ( standard deviation of the noise : 5.5 msv ) . , width=302 ] the most direct way to test which of the proposed mechanisms if any provides the correct explanation for the timing of do events would be to reconstruct decadal - scale density anomalies in the north atlantic in connection with the events
this is not possible since even the most highly resolved oceanic records do not allow to reconstruct the variability of the glacial ocean on that time scale .
thus , only indirect tests can be performed .
the identification of the postulated 1/(1470 years ) forcing frequency , which has so far not been detected , would certainly give further support for the sr mechanism . and
in order to support the gr mechanism , it remains crucial to demonstrate that century - scale solar irradiance variations are indeed of sufficiently large amplitude to trigger repeated transitions ( with a preferred time scale of about 1470 years ) between the two glacial climate states .
this could be tested with climate models .
an elegant and simple test is to make use of the observation that do events in the earth system model climber-2 represent the nonlinear response to the forcing , and that an additional ( and much smaller ) linear response is superimposed on the events . in the absence of any threshold crossing ( e.g. in the holocene ,
during which do events did not occur ) the response to the forcing , in contrast , does not show a strong nonlinear component .
this suggests that holocene climate archives from the north atlantic region might be able to reveal what triggered glacial do events .
this approach has two major advantages : first , more reliable ( e.g. better resolved and dated ) records are available to solve this issue .
second , linear analysis methods can be used for that purpose , e.g. linear correlations . in the context of the gr mechanism , for example
, the existence of a pronounced correlation between holocene climate indices from the north atlantic and solar activity proxies ( reconstructed e.g. from @xmath36c variations in precisely dated tree rings ) would be expected . up to
now at least one study exists that supports this prediction of a linear relationship between century - scale solar forcing and north atlantic climate variability throughout the holocene : proxies of drift ice anomalies in the north atlantic show a persistent correlation and a statistically significant coherency with `` rapid ( 100- to 200-year ) , conspicuously large - amplitude variations '' in solar activity proxies @xcite , in accordance with the proposed gr mechanism . the most challenging test , however , is the direct analysis of the glacial climate records .
we are convinced that one of the main difficulties in this approach is the high degree of nonlinearity of the events , which according to our interpretation has so far not been adequately addressed in many previous studies . for example , several attempts have already been made in order to investigate the 1470-year cycle by means of linear spectral analysis methods , and significance levels have commonly been calculated by assuming a red noise background . to us
this assumption seems to be oversimplified , since the system responds at a preferred time scale even when driven by white noise ( compare fig .
[ fig : figure4]b ) .
we thus suspect that the significance levels obtained by this method are unrealistically high .
we further think that the reported lack of a clear phase relation between solar activity proxies and do events @xcite can not rule out the idea that solar forcing synchronised do events , since in the case of an additional stochastic forcing component ( i.e. in the gsr case ) the events are triggered by the _ combined _ effect of solar forcing and noise .
thus , the observed lack could also imply that only some of the events were in first place triggered by the sun , whereas others were caused mainly by random variability ( e.g. by noise ) .
a new and promising approach , which is based on a monte - carlo method , has recently been proposed in order to test the significance of the glacial 1470-year climate cycle : @xcite define a certain measure in order to distinguish between different hypotheses for the timing of do events , and they explicitly calculate the value of this measure for the series of events observed in the ice core .
they then compare the calculated value with the values obtained by several hypothetic processes , e.g. by a random process ( for which assumptions concerning the probability distribution of the recurrence times of the events have to be made ) .
significance levels are obtained from the ( numerically estimated ) probability distributions of the measure as generated by the considered process .
although we do not share their conclusions ( because we think that more adequate measures can be chosen , which give considerably different results ) we think that this approach is elegant because significance levels are not calculated based on linear theories .
the method is thus also applicable to highly nonlinear time series .
a major hurdle in this method is that for each considered process the probability distribution of the waiting times which is unknown for almost all processes somehow has to be specified .
for example , ditlevsen et al .
use a simple mathematical ( i.e. an exponential ) distribution in order to mimic random do events . in order to improve their novel approach ,
some method is thus needed to calculate waiting time distributions in response to any possible input .
comprehensive models are not applicable , due to their large computational cost . our conceptual model , in contrast , is well designed for that purpose because it is combines the ability to mimic the complex nonlinearity of do events as described by an accepted earth system model with the extremely low computational cost of a very simple ( zeroth order ) model .
we thus think that our work is an important step in order to develop improved statistical analysis methods which are able to cope with the extreme nonlinearity of do events .
[ sec : end ] we here discussed do events in the framework of a very simple conceptual model that is based on three key assumptions , namely ( i ) the existence of two different climate states , ( ii ) a threshold process and ( iii ) an overshooting in the stability of the system at the start and the end of the events , which is followed by a millennial - scale relaxation .
these assumptions are in accordance with paleoclimatic records and/or with simulations performed with climber-2 , a more complex earth system model . in a couple of systematic tests we showed ( in the supplementary material )
that despite its simplicity , our model very well reproduces do events as simulated with climber-2 , whose dynamics is based on a ( albeit reduced ) description of the underlying hydro-/thermodynamic processes .
the correspondence between both models thus strengthens our interpretation that the conceptual model can successfully mimic key features of do events , and that these can be regarded as a new type of non - equilibrium oscillation ( i.e. as an _ overshooting relaxation oscillation _ ) between two states of a nonlinear system with a threshold .
although we discussed our model dynamics in the context of the ( thermohaline ) ocean circulation , our model does not explicitly assume that do events are linked with changes in the ocean circulation : threshold behaviour and multiple states exist in many compartments of the climate system ( not only in the ocean , but e.g. also in the atmosphere and in the cryosphere ) .
our model thus can not rule out a leading role of non - oceanic processes in do oscillations .
the millennial time scale of the events ( which is represented in our model by the assumption of a millennial - scale relaxation ) , however , corresponds to the characteristic time scale of the thermohaline circulation and thus points to a key role of the ocean in do oscillations .
the main strength of our model is its simplicity : due to the obvious relationship between forcing and response , the model can demonstrate why even a simple bistable ( or excitable ) system with a threshold can respond in a complex way to a simple forcing , which consists of only one or two sinusoidal inputs and noise .
we applied our model to discuss two highly nonlinear and apparently counterintuitive resonance mechanisms , namely stochastic resonance and ghost resonance . in doing
so we reported a new form of stochastic resonance ( i.e. an _ overshooting stochastic resonance _ ) , in which the overshooting of the system leads to a further synchronisation effect compared to the usual stochastic resonance .
our study provides the first explicitly reported manifestation of ghost resonance in a geophysical ( model ) system .
since threshold behaviour and multiple equilibria are not unique to do events but exist in many geophysical systems , we would indeed expect that ghost resonances could be inherent in many geosystems and not just in our model .
in addition to its applicability to demonstrate and interpret nonlinear resonance mechanisms , and to test their stability , we further illustrated the ability of our conceptual model to simulate probability measures ( e.g. waiting time distributions , which are required in order to test the significance and the cause of the proposed glacial 1470-year climate cycle by means of monte - carlo simulations ) .
because it combines the ability to reproduce essential aspects of do events with the extremely low computational cost of a conceptual model ( which is up to 10@xmath37 times lower than in the earth system model climber-2 ) , we think that our model represents an important advance in order to develop adequate nonlinear methods for improved statistical analyses on do events .
the earth system model climber-2 , which we used for our analysis , has dynamic components of the atmosphere , of the oceans ( including sea ice ) and the vegetation .
dynamic ice sheets were not included in our study .
climber-2 is a global model with coarse resolution : for the atmosphere and the continents the spatial resolution is 10@xmath29 in latitude , and 7 sectors are considered in longitude . the ocean is zonally averaged with a latitudinal resolution of 2.5@xmath29 for the three large ocean basins .
a detailed description of the model is given in the publication of @xcite .
do events in the model represent abrupt switches between two different climate states ( _ stadial _ [ i.e. cold ] and _ interstadial _ [ i.e. warm ] ) , corresponding to two different modes of the thc : in the interstadial mode , north atlantic deep water ( nadw ) forms at about 65 @xmath0n and much of the north atlantic is ice - free . in the stadial mode ,
nadw forms at about 50 @xmath0n and a considerably larger area of the north atlantic is ice - covered .
we note that for the climatic background conditions of the last glacial maximum ( lgm ) only the stadial mode is stable in the model whereas the interstadial mode is excitable but unstable @xcite . moreover , the stability of both modes depends on the actual climate state ( e.g. on the configuration of the laurentide ice sheet and on the freshwater input into the north atlantic ) , and the stability properties of the system change when the background conditions are modified ( more precisely , the system can be bistable or mono - stable ) .
transitions between both modes can be triggered by anomalies in the density field of the north atlantic , for example by variations in the surface freshwater flux ( since the density of ocean water increases with increasing salinity ) . in our study
we thus implement the forcing as a perturbation in the freshwater flux ( in the latitudinal belt 50 - 70 @xmath29n ) : we start the model with the climatic background conditions of the last glacial maximum ( lgm ) .
following earlier simulations @xcite we then add a small constant offset of 17 ( @xmath38 ) to the freshwater flux .
for this climate state ( which we label perturbed lgm ) the thc is in fact bistable and do events can be triggered more easily than for lgm conditions .
this perturbed lgm state gives us the background conditions for the model simulations as presented in this paper .
the authors thank r. calov , a. mangini , s. rahmstorf , k. roth and a. witt for discussion , and p. ditlevsen ( in particular for observing the difference between the usual stochastic resonance and our overshooting stochastic resonance ) and two anonymous reviewers for helpful comments .
h. braun was funded by deutsche forschungsgemeinschaft , dfg project number ma 821/33 .
alley , r. b. and clark , p. u. , the deglaciation of the northern hemisphere : a global perspective , ann .
earth planet .
, 27 , 149182 , 1999 .
alley , r. b. , clark , p. u. , keigwin , l. d. , and webb , r. s. , making sense of millennial - scale climate change , in : mechanisms of global climate change at millennial time scales , edited by : clark , p. u. , webb , r. s. , and keigwin , l. d. , pp .
385394 , agu , washington , dc , 1999 .
alley , r. b. , anandakrishnan , s. , and jung , p. , stochastic resonance in the north atlantic , paleoceanography , 16 , 190198 , 2001a .
alley , r. b. , anandakrishnan , s. , jung , p. , and clough , a. , stochastic resonance in the north atlantic : further insights , in : the oceans and rapid climate change : past , present and future , edited by seidov , d. , maslin , m. , haupt , b. j. , pp . 5768 , agu , washington , dc , 2001b .
alley , r. b. , marotzke , j. , nordhaus , w. d. , overpeck , j. t. , peteet , d. m. , pielke jr .
, r. a. , pierrehumbert , r. t. , rhines , p. b. , stocker , t. f. , talley , l. d. , and wallace , j. m. , abrupt climate change , science , 299 , 20052010 , 2003 .
benzi , r. , parisi , g. , sutera , a. , and vulpiani , a. , stochastic resonance in climatic change .
tellus , 34 , 1016 , 1982 .
bond , g. , kromer , b. , beer , j. , muscheler , r. , evans , m. n. , showers , w. , hoffmann , s. , lotti - bond , r. , hajdas , i. , and bonani , g. , persistent solar influence on north atlantic climate during the holocene .
science , 294 , 21302136 , 2001 .
braun , h. , christl , m. , rahmstorf , s. , ganopolski , a. , mangini , a. , kubatzki , c. , roth , k. , and kromer , b. , possible solar origin of the 1,470-year glacial climate cycle demonstrated in a coupled model , nature , 2005 , 438 , 208211 .
broecker , w. s. , peteet , d. m. , and rind , d. , does the ocean - atmosphere system have more than one stable mode of operation ?
nature , 315 , 2126 , 1985 .
broecker , w. s. , bond , g. , klas , m. , bonani , g. , and wolfli , w. , a salt oscillator in the glacial atlantic ? 1 .
the concept , paleoceanography , 5 , 469477 , 1990 .
buldu , j. m. , chialvo , d. r. , mirasso , c. r. , torrent , m. c. , and garcia - ojalvo , j. , ghost resonance in a semiconductor laser with optical feedback , europhysics letters , 64 , 178184 , 2003 .
centurelli , r. , musacchioa , s. , pasmanterc , r. a. , and vulpiani , a. , resemblances and differences in mechanisms of noise - induced resonance , physica a , 360 , 261273 , 2006 .
chialvo , d. r. , calvo , o. , gonzalez , d. l. , piro , o. , and savino , g. v. , subharmonic stochastic synchronization and resonance in neuronal systems , physical review e , 65 , 050902 , 2002 .
chialvo , d. r. , how we hear what is not there : a neuronal mechanism for the missing fundamental illusion , chaos , 13 , 1226 - 1230 , 2003 .
claussen , m. , mysak , l. a. , weaver , a. j. , crucifix , m. , fichefet , t. , loutre , m .- f . , weber , s. l. , alcamo , j. , alexeev , v. a. , berger , a. , calov , r. , ganopolski , a. , goose , h. , lohmann , g. , lunkeit , f. , mokhov , i. i. , petoukhov , v. , stone , p. , wang , z. , earth system models of intermediate complexity : closing the gap in the spectrum of climate system models .
dyn . , 18 , 579586 , 2002 .
clemens , s. c. , millennial - band climate spectrum resolved and linked to centennial - scale solar cycles .
rev . , 24 , 521531 , 2005 .
dansgaard , w. , clausen , h. b. , gundestrup , n. , hammer , c. u. , johnsen , s. f. , kristinsdottir , p. m. , and reeh , n. , a new greenland deep ice core , science , 218 , 12731277 , 1982 .
ditlevsen , p. d. , kristensen , m. s. , andersen , k. k. , the recurrence time of dansgaard - oeschger events and possible causes , j. clim . , 18 , 25942603 , 2005 .
ditlevsen , p. d. , andersen , k. k. , svensson , a. , the do - climate events are probably noise induced : statistical investigation of the claimed 1470 years cycle , clim .
past , 3 , 129134 , 2007 .
gammaitoni , l. , hnggi , p. , jung , p. , and marchesoni , f. , stochastic resonance .
, 70 , 223288 , 1998 .
ganopolski , a. and rahmstorf , s. , simulation of rapid glacial climate changes in a coupled climate model , nature 409 , 153158 , 2001 .
ganopolski , a. and rahmstorf , s. , abrupt glacial climate changes due to stochastic resonance , phys .
, 88 , 038501 , 2002 .
grootes , p. m. , stuiver , m. , white , j. w. c. , johnsen , s. , and jouzel , j. , comparison of oxygen isotope records from the gisp2 and grip greenland ice cores , nature , 366 , 552554 , 1993 .
grootes , p. m. and stuiver , m. , oxygen 18/16 variability in greenland snow and ice with 10@xmath39 to 10@xmath40-year time resolution .
j. geophys .
102 , 2645526470 , 1997 .
keeling , c. d. and whorf , t. p. , the 1,800-year oceanic tidal cycle : a possible cause of rapid climate change .
usa , 97 , 38143819 , 2000 .
leuenberger , m. , schwander , j. , and johnsen , s. , 16@xmath0c rapid temperature variations in central greenland 70,000 years ago , science , 286 , 934937 , 1999 .
muscheler , r. and beer , j. , solar forced dansgaard / oeschger events ?
, 33 , l20706 , 2006 .
oeschger , h. , beer , j. , siegenthaler , u. , stauffer , b. , dansgaard , w. , langway , jr .
, c. c. , late glacial climate history from ice cores , in : climate processes and climate sensitivity , edited by hansen , j. e. and takahashi , t. , pp . 299306 , agu , washington , dc , 1984 .
peristykh , a. n. and damon , p. e. , persistence of the gleissberg 88-yr solar cycle over the last 12,000 years : evidence from cosmogenic isotopes .
j. geophys .
res . , 108 ; 1003 , 2003 .
petoukhov , v. , ganopolski , a. , brovkin , v. , claussen , m. , eliseev , a. , kubatzki , c. , and rahmstorf , s. , climber-2 : a climate system model of intermediate complexity .
part i : model description and performance for present climate , clim .
16 , 117 , 2000 .
rahmstorf , s. , ocean circulation and climate during the past 120,000 years , nature , 419 , 207214 , 2002 .
rahmstorf , s. , timing of abrupt climate change : a precise clock , geophys .
, 30 , 15101514 , 2003 .
rahmstorf , s. and alley , r. b. , stochastic resonance in glacial climate , eos , 83 , 129135 , 2002 .
rial , j. a. , abrupt climate change : chaos and order at orbital and millennial scales , glob .
change , 41 , 95109 , 2004 .
sakai , k. and peltier , w. r. , dansgaard - oeschger oscillations in a coupled atmosphere - ocean climate model , j. clim .
, 10 , 949970 , 1997 .
sarnthein , m. , winn , k. , jung , s. j. a. , duplessy , j. c. , labeyrie , l. , erlenkeuser , h. , and ganssen , g. , changes in east atlantic deepwater circulation over the last 30,000 years : eight time slice reconstructions , paleoceanography , 9 , 209267 , 1994 .
schulz , m. , on the 1470-year pacing of dansgaard - oeschger warm events .
paleoceanography , 17 , 10141022 , 2002 .
schulz , m. , paul , a. , and timmermann , a. , relaxation oscillators in concert : a framework for climate change at millennial timescales during the late pleistocene .
lett . , 29 , 21932197 , 2002 .
severinghaus , j. p. and brook , e. , abrupt climate change at the end of the last glacial period inferred from trapped air in polar ice , science , 286 , 930934 , 1999 .
stuiver , m. and braziunas , t. f. , sun , ocean , climate and atmospheric co2 : an evaluation of causal and spectral relationships , holocene , 3 , 289305 , 1993 .
taylor , k. c. , mayewski , p. a. , alley , r. b. , brook , e. j. , gow , a. j. , grootes , p. m. , meese , d. a. , saltzman , e. s. , severinghaus , j. p. , twickler , e. s. , white , j. w. c. , whitlow , s. , and zielinski , g. a. , the holocene - younger dryas transition recorded at summit , greenland , science , 278 , 825827 , 1997 .
timmermann , a. , gildor , h. , schulz , m. , and tziperman , e. , coherent resonant millennial - scale climate oscillations triggered by massive meltwater pulses . j. clim . , 16 , 25692585 , 2003 .
van kreveld , s. a. , sarnthein , m. , erlenkeuser , h. , grootes , p. , jung , s. , nadeau , m. j. , pflaumann , u. , and voelker , a. , potential links between surging ice sheets , circulation changes and the dansgaard - oeschger cycles in the irminger sea , 60 - 18 kyr , paleoceanography , 15 , 425442 , 2000 .
wagner , g. , beer , j. , masarik , j. , muscheler , r. , kubik , p. w. , mende , w. , laj , c. , raisbeck , g. m. , and yiou , f. , presence of the solar de vries cycle ( @xmath34205 years ) during the last ice age , geophys .
, 28 , 303306 , 2001 .
yiou , p. , fuhrer , k. , meeker , l. d. , jouzel , j. , johnsen , s. , and mayewski , p. a. , paleoclimatic variability inferred from the spectral analysis of greenland and antarctic ice - core data . j. geophys .
, 102 , 2644126454 , 1997 . | here we use a very simple conceptual model in an attempt to reduce essential parts of the complex nonlinearity of abrupt glacial climate changes ( the so - called dansgaard - oeschger events ) to a few simple principles , namely ( i ) the existence of two different climate states , ( ii ) a threshold process and ( iii ) an overshooting in the stability of the system at the start and the end of the events , which is followed by a millennial - scale relaxation . by comparison with a so - called earth system model of intermediate complexity ( climber-2 ) , in which the events represent oscillations between two climate states corresponding to two fundamentally different modes of deep - water formation in the north atlantic , we demonstrate that the conceptual model captures fundamental aspects of the nonlinearity of the events in that model .
we use the conceptual model in order to reproduce and reanalyse nonlinear resonance mechanisms that were already suggested in order to explain the characteristic time scale of dansgaard - oeschger events . in doing
so we identify a new form of stochastic resonance ( i.e. an _ overshooting stochastic resonance _ ) and provide the first explicitly reported manifestation of _ ghost resonance _ in a geosystem , i.e. of a mechanism which could be relevant for other systems with thresholds and with multiple states of operation .
our work enables us to explicitly simulate realistic probability measures of dansgaard - oeschger events ( e.g. waiting time distributions , which are a prerequisite for statistical analyses on the regularity of the events by means of monte - carlo simulations ) .
we thus think that our study is an important advance in order to develop more adequate methods to test the statistical significance and the origin of the proposed glacial 1470-year climate cycle .
[ sec : intro ] do events as seen in the gisp2 ice - core from greenland @xcite .
the figure shows greenland temperature changes over the interval between 10000 and about 40000 years before present .
do events ( 0 - 10 ) manifest themselves as saw - tooth shaped warm intervals .
dashed lines are spaced by 1470 years .
, width=302 ] time series of north atlantic atmospheric / sea surface temperatures during the last ice age reveal the existence of repeated large - scale warming events , the so - called dansgaard - oeschger ( do ) events @xcite . in climate records from the north atlantic region
the events have a characteristic saw - tooth shape ( fig .
[ fig : figure1 ] ) : they typically start with a warming by up to 10 - 15 k @xcite over only a few years / decades .
temperatures remain high for centuries / millennia until they drop back to pre - events values over a century or so .
a prominent feature of do events is their millennial time scale : during marine isotope stages ( mis ) 2 and 3 , successive events in the gisp2 ice core were reported to be often spaced by about 1470 years or multiples thereof @xcite , compare fig .
[ fig : figure1 ] .
a leading spectral peak corresponding to the 1470-year period was found @xcite , and this spectral component was reported to be significant at least over a certain time interval @xcite .
we note , however , that the statistical significance of this pattern is still under debate @xcite , in particular because of the lack of adequate nonlinear analysis methods .
it was proposed that do events represent rapid transitions between two fundamentally different modes of the thermohaline ocean circulation ( thc ) @xcite , most likely corresponding to different modes of deep - water formation @xcite .
the origin of these transitions is also under debate : in principle they could have been caused by factors from outside of the earth system @xcite , but they could also represent internal oscillations of the climate system @xcite .
several nonlinear resonance mechanisms have been suggested in order to explain the characteristic timing of do events , including coherence resonance @xcite and stochastic resonance @xcite . |
gas - particle flows are commonly encountered in the chemical and energy conversion industries in the form of fluidized beds , risers , and other pneumatic conveying units .
the flow behaviors of gas - particle systems are commonly analyzed using continuum models that treat the particle and fluid phases as interpenetrating continua @xcite .
finite - volume simulations of these continuum , or ` two - fluid ' models , reveal that the scale of the gas - particle flow structures that are observed is a strong function of the grid size used in the simulation , with grid size independent results being obtained when the ratio between the grid size and particle size is @xmath0
grid cell sizes of ten particle diameters are not computationally affordable when performing two - fluid model simulations of large fluidized bed reactors that are often tens of meters tall and many meters in cross - section . as a result , coarse - grid simulations of large fluidized beds are common practice .
however , these coarse - grid simulations effectively neglect the presence of fine scale gas - particle flow structures that are manifested by the two - fluid model . to enable accurate coarse - grid simulation of the two - fluid model equations one must account for the effect of the fine - scale gas - particle flow structures in the coarse - grid simulation @xcite .
this approach is embodied by the recent development of a _ filtered _ two - fluid model approach for non - reacting gas - particle flows where effective closures for the _ meso - scale _ fluid - particle drag force , particle phase stress , and particle phase viscosity were extracted from fine - grid simulations of the two - fluid model equations @xcite .
while _ filtered _ two - fluid models for non - reacting monodisperse gas - particle flows have been developed , verified against fine - grid simulations , and validated against experimental observations @xcite , the extension of these models to reacting and polydisperse gas - particle flows remains an open problem .
fine - grid two - fluid model simulations of reacting gas - particle flow have been shown to yield quantitative agreement with experimental observations of conversion in solid - catalyzed ozone decomposition processes when grid - independent solutions are obtained @xcite . however , due to the computational expense of performing grid - independent simulations , coarse numerical resolution is often used to simulate reacting gas - particle flows in the continuum model framework . by not accounting for the fine - scale structure these coarse - grid simulations
have been shown to over - predict the conversion of ozone in a bubbling fluidized bed @xcite .
in addition , a few studies have also demonstrated that small - scale gas - particle flow structures that form in devices like fluidized beds are responsible for the wide variability in mass transfer models that is present within the fluidization literature @xcite .
the primary objective of this work is to demonstrate the need for _ filtered _
models for reacting gas - particle flows by performing fine - grid simulations of a first - order , isothermal , solid catalyzed reaction in a periodic domain and filtering the results .
it will be shown that the presence of particle clustering in fine - grid simulations leads to an effective reaction rate that is substantially smaller than what would be predicted via coarse - grid simulation .
we define the ratio of the reaction rate in the fine - grid simulation to that in the coarse - grid simulation as the _ cluster - scale _ effectiveness factor .
it will be shown that this _ cluster - scale _ effectiveness factor is a strong function of dimensionless filter size @xmath1 , and other model parameters . here
, @xmath2 is the filter size , @xmath3 is the gravitational acceleration vector , and @xmath4 is the terminal settling velocity of an isolated particle .
finally , it is shown that grid resolutions finer than those used to deduce _ filtered _
models for non - reacting gas - particle flows are necessary in order to obtain grid - independent _ filtered _
closures for reacting gas - particle flows .
this observation is supported by a recent work by @xcite . due to this grid size dependence
we propose a _ filtered _ reaction rate model based upon extrapolated effectiveness factor data obtained from fine - grid simulations .
the two - fluid model consists of balance equations for gas- and particle - phase mass and momentum with an additional transport equation governing evolution of the particle - phase granular temperature .
the evolution equations that constitute this eulerian framework are presented in table 1 , in addition to the constitutive relations that close the balance equations .
the constitutive relations given in table 1 were chosen to be consistent with earlier studies in filtered model development within our research group @xcite . here
we note that while we have restricted our attention to a specific set of constitutive relations , it has been shown that the clustering and bubbling phenomena manifested by the two - fluid model framework are robust to changes in constitutive relations @xcite .
however , we do expect that the constitutive relations comprising the microscopic two - fluid models will have quantitative effects on the filtered models that arise from fine - grid simulations .
therefore , while we have employed a certain set of constitutive relations in this study , filtered models can be developed from fine - grid , two - fluid model simulations using a different set of constitutive relations as well .
the effective diffusion coefficient @xmath5 and effective gas - phase viscosity @xmath6 appearing in eqs .
( [ eq : cont_species ] ) and ( [ eq : gas_stress ] ) will differ from the corresponding molecular properties as they also account for the enhanced scalar and momentum transport that occurs due to additional _ pseudo - turbulent _ transport occurring as a result of microscopic interactions between individual particles and the fluid .
however , two - fluid model simulations have revealed that the solutions manifested by the two - fluid model are insensitive to the value of @xmath6 @xcite .
therefore , we set @xmath6 equal to the molecular viscosity @xmath7 .
in addition , the effective diffusion coefficient @xmath5 is kept on the order of the bulk molecular diffusivity @xmath8 , which is generally @xmath9 for many different gas species . here
we investigate the role of clustering on a model , isothermal , solid - catalyzed , first - order chemical reaction @xmath10 where @xmath11 is the particle volume fraction , and @xmath12 is the mass fraction of species @xmath13 in the gas phase . for first order reaction kinetics , the effective reaction rate constant @xmath14 based on the bulk concentration @xmath15
is related to the intrinsic reaction rate constant @xmath16 via @xmath17 where @xmath18 is the _ intra - particle _ effectiveness in the absence of mass transport limitations , @xmath19 is the thiele modulus , @xmath20 is the volume to surface area ratio , and @xmath21 is the biot number for mass transport given as @xmath22 @xcite . here , @xmath23 is the convective mass transport coefficient , @xmath20 is the particle radius , and @xmath24 is the _ intra - particle _ diffusivity .
it is important to note that the effective reaction rate constants used in two - fluid model simulations incorporate the combined effect of mass transport resistance and _ intra - particle _ effects as well . in this study
all gas - solid flow computations are performed for fcc catalyst particles fluidized by air @xmath25 the physical properties for solid and fluid phases are given in table 2 .
5 inx @xmath26 @xmath27 @xmath28 @xmath29 @xmath30 @xmath31=-\nabla\cdot \boldsymbol{q } -\boldsymbol{\sigma}_s:\nabla \boldsymbol{v}-j_\text{vis}-j_\text{coll}+\gamma_\text{slip } \label{eq : gran_temp}\ ] ] + @xmath32 @xmath33 @xmath34 @xmath35 @xmath36 @xmath37 @xmath38 @xmath39 @xmath40 @xmath41 5 inx @xmath42 @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 @xmath48 @xmath49 @xmath50 @xmath51 @xmath52 @xmath53 6cmp2cmx @xmath54 & @xmath55 m + @xmath56 & @xmath57 @xmath58 + @xmath59 & @xmath60 @xmath58 + @xmath7 & @xmath61 @xmath62 + @xmath63 & @xmath64 +
the _ filtered _ two - fluid models are obtained by performing a spatial average of the microscopic two - fluid model equations .
the consequence of the filtering approach is that the fine - scale gas - particle flow structure that occurs on a length scale smaller than the filter size is accounted for through residual terms that must be constituted from theoretical considerations or fine - grid two - fluid model simulation results . in the case of non - reacting monodisperse gas - particle flows ,
_ filtered _ models with accompanying constitutive relations for residual terms have been shown to yield quantitatively similar macroscopic behaviors to those observed in fine - grid simulations of the same flow problem , with a dramatic savings in computational time @xcite . here
we follow , the filtering procedure given in @xcite .
the particle volume fraction and gas - species mass fractions obtained from fine - grid simulations can be represented as @xmath65 , where @xmath66 and _ t _ represent the location and time variables , respectively .
the _ filtered _ particle volume fraction @xmath67 is given as @xmath68 where @xmath69 is a weight function , and @xmath70 is the spatial location of the filter center . in this
work @xmath71 will be used to indicate filtered variables .
we require that @xmath72 , and in this study we use a top hat filter for @xmath69 . _
filtered _ gas - species mass fraction @xmath73 , particle phase velocity @xmath74 , and fluid phase velocity @xmath75 are given as @xmath76 @xmath77 @xmath78 upon filtering the microscopic species balance equation for component @xmath13 we obtain the following _ filtered _ species balance equation @xmath79 in eq .
( [ eq : filt_chi_eq ] ) we invoke the definition of filtered variables given in eqs .
( [ eq : filt_chi])([eq : filt_vel_g ] ) . due to
the fact that the reaction we are considering in this work is isothermal and produces no volume change , the _ filtered _ momentum balance equations remain unchanged from those derived in the work of @xcite , and as such , they will not be presented here for the sake of brevity . in eq .
( [ eq : filt_chi_eq ] ) , there are several terms that appear as _ filtered _ products of microscopic variables that must be constituted to solve the _ filtered _ two - fluid model equations . to facilitate filtered model simulations
we must decompose the products of microscopic variables into mean and fluctuating parts . here , we define fluctuating variables as follows @xmath80 @xmath81 where @xmath82 , @xmath83 , and @xmath84 represent species @xmath13 mass fraction , gas velocity , and volume fraction fluctuations , respectively . inserting the decomposition of mean and fluctuating parts into eq .
( [ eq : filt_chi_eq ] ) , we obtain the following filtered species balance equation @xmath85 the first term appearing on the right hand side of eq .
( [ eq : filt_chi_eq2 ] ) can be modeled as a dispersive term given by the following constitutive equation @xmath86 where @xmath87 is a dispersion coefficient .
dispersion coefficients defined in this way have been presented in the research literature @xcite . the final term on the right hand side of eq .
( [ eq : filt_chi_eq2 ] ) represents a _ filtered _ reaction rate that must also be constituted in terms of filtered variables alone .
a straightforward method for constituting this reaction rate is to define a _
cluster - scale _ effectiveness factor @xmath88 which is given as @xmath89 however , in this paper we present _ filtered _ models and associated constitutive relations for the _ non - locally _ corrected _ cluster - scale _ effectiveness factor @xmath90 defined as @xmath91 where @xmath92 .
the approach of removing _ non - local _ effects in the determination of filtered quantities was recently advanced by @xcite , and we apply the same procedure here for determining @xmath90 .
the _ non - locally _ corrected effectiveness factor is constructed by removing the dominant gradient terms that contribute the evaluation of the filtered product @xmath93 , which is done to allow the construction of _ filtered _ constitutive relations in terms of local _ filtered _ variables alone .
we present a derivation of the _ non - locally _ corrected effectiveness factor in [ sec : nonl_deriv ] .
it will be shown below that both @xmath88 and @xmath90 depend on volume fraction in a qualitatively similar way , but differ quantitatively .
all data presented in future sections and resulting _ filtered _ models will pertain to @xmath90 , thus necessitating one to track the gradients in filtered species mass fraction and volume fraction to enable the use of the resulting filtered models in coarse - grid simulations of reacting gas - particle flow .
dimensional analysis of the parameters governing gas - particle flow suggests that the _ cluster - scale _ effectiveness factor is a function of five independent dimensionless quantities , which are @xmath94 @xmath95 one can also readily define additional dimensionless parameters like @xmath96 and the coefficient of restitution @xmath63 using dimensional analysis .
however , earlier works of @xcite have shown that _ filtered _
quantities do not display any significant dependence on @xmath97 or @xmath63 , and as such , they were not included in the dimensional analysis presented here .
moreover , we found that @xmath98 and @xmath99 had a negligible effect on the _ cluster - scale _ effectiveness factor , while the _ meso - scale _ thiele modulus @xmath100 , schmidt number @xmath101 , and the dimensionless filter size @xmath102 substantially alter the _ cluster - scale _ effectiveness factor .
therefore , all work presented below will only interrogate the effect of @xmath100 , @xmath101 , and @xmath102 on @xmath90 . here the term _ meso - scale _ thiele modulus is used to distinguish it from the thiele modulus based on _ intra - particle _
diffusivity @xmath24 given in eq .
( [ eq : micro_eta ] ) .
when performing periodic domain simulations of reacting gas - particle flows , one is faced with the following dilemma : due to the imposition of periodic boundary conditions , the concentration of any reactant undergoing an irreversible reaction within the periodic domain decay to zero . in order to facilitate gas - particle flow simulations in periodic domains , an alternate simulation strategy must be developed .
for the special case of a first - order reaction , it can readily be shown that one can track the evolution of @xmath103 rather than @xmath15 itself , and relate the _ filtered _ value of @xmath104 directly to the _ cluster - scale _ effectiveness factor .
the benefit of tracking the evolution of @xmath104 lies in the fact that it has a non - zero statistically steady value , even though @xmath15 will decay to zero .
consider taking the species balance equation given by eq .
( [ eq : cont_species ] ) with the reaction rate expression given by eq .
( [ eq : rate_law ] ) and averaging it over the entire periodic domain , resulting in the following equation @xmath105 where @xmath106 is the volume of the periodic domain .
if we now define the variable @xmath103 and plug @xmath104 into the species balance equation given by eq .
( [ eq : cont_species ] ) we obtain the following evolution equation @xmath107 due to the fact that the reaction rate expressions in this problem are first order , we are able to track the evolution of @xmath104 without considering the time progression of @xmath15 or @xmath108 . the presence of non - zero source and sink terms on the right hand side of eq .
( [ eq : kappa_eq ] ) force statistically steady value of @xmath104 to be non - zero regardless of the reaction rate constant used .
therefore , in all work presented in following sections for the evolution of @xmath104 is solved , and the _ cluster - scale _ effectiveness factor is redefined in terms of @xmath104 as @xmath109 all simulation results presented in this work were generated using the multiphase flow with interface exchanges ( mfix ) software that relies on a variable time step , staggered grid , finite - volume method for the solution of the two - fluid model equations @xcite .
the iterative solution to the two - fluid model equations is obtained using the semi - implicit method with pressure linked equations , or the simple algorithm . due to the strong coupling between solid and fluid phases that arises due to the fluid - particle drag force
, the partial elimination algorithm of spalding was used to effectively decouple the solution of the solid and fluid phase balance equations @xcite .
in addition , a second order superbee discretization is employed for the convective terms that are present in the conservation equations for continuity , momentum and granular energy transport to limit the effects of numerical diffusion .
to illustrate the behavior of the _ cluster - scale _ effectiveness factor as a function of @xmath110 , simulation results are presented that were obtained using grid sizes that are sixteen times as large as the particle diameter .
this grid resolution was chosen to follow the simulations employed in the study of @xcite .
_ filtered _ variables were obtained from these fine - grid simulations by moving filters of different sizes throughout the periodic domain . due to the statistical homogeneity of periodic domains we are able to collect thousands of samples and average them regardless of spatial position . moreover , by _ filtering _ these fine - grid simulation results for various domain - averaged volume fractions the filter - averaged volume fraction dependence of the _ cluster - scale _ effectiveness factor can be ascertained .
this _ filtering _ procedure follows that outlined in the work of @xcite .
the characteristic @xmath110 dependence of @xmath88 and @xmath90 is presented in figure [ fig : eta_phi_dep ] .
it is clear from figure [ fig : eta_phi_dep ] that both @xmath88 and @xmath90 are strong functions of @xmath110 retaining an inverted bell shape that approaches unity in the limit of small and large particle volume fraction .
however , there are noticeable quantitative differences between @xmath88 and @xmath90 in figure [ fig : eta_phi_dep ] with the maximum deviation being near the minimum in both curves . the minimum value of @xmath111 , while the minimum value of @xmath112 .
therefore , at this resolution , the non - local correction to the cluster - scale effectiveness factor contributes as much as @xmath113 to the value of the _ cluster - scale _ effectiveness factor near the minimum in the curves given in figure [ fig : eta_phi_dep ] .
while the quantitative differences between @xmath88 and @xmath90 are functions of @xmath102 and other model parameters , this example provides a scale of the difference between the two effectiveness factors . in this work we have decided to model @xmath90 because it is directly related the small scale fluctuations in @xmath110 and @xmath108 , while @xmath88 is dependent on non - local variations in @xmath110 and @xmath108 ( see appendix ) . physical intuition for the decrease in the _ cluster - scale _ effectiveness factor from unity
can be obtained by observing the characteristic clustering patterns that are observed at different domain - averaged volume fractions , which are superimposed in figure [ fig : eta_phi_dep ] . at low volume fractions
there are a few small isolated clusters throughout the periodic domain , and as such , the cluster - scale effectiveness factor will begin to deviate from unity . as the volume fraction increases the frequency of these clusters increases thus making the effective contacting between gas and particle phases poor .
near the minimum value in the @xmath90 curve , clusters begin to span the periodic domain , and the solid phase changes from the dispersed to the continuous phase . as volume fraction continues to increase
the gas - particle flow becomes more homogeneous , and due to this homogeneity the cluster - scale effectiveness factor increases toward unity in the limit of high particle volume fraction . versus @xmath110 is shown with snapshots of the particle volume fraction field different domain - averaged volume fractions superimposed .
here , @xmath114 , @xmath115 , and @xmath116.,width=5 ] in figure [ fig : eta_curves ] ( a ) the _ cluster - scale _ effectiveness factor is shown as a function of @xmath110 for four different @xmath100 values at fixed values of the @xmath101 and @xmath102 .
the depth of the inverted bell shape curve in @xmath110 is an increasing function of @xmath100 .
the values of @xmath100 given in figure [ fig : eta_curves ] ( a ) are substantially smaller than unity , but there is a marked change in @xmath90 from unity .
this may seem peculiar , but the small values of @xmath100 arise as a result of the fact that the length scale used to determine @xmath100 is the particle diameter @xmath54 .
a more fitting length scale is that associated with a cluster .
however , the size of a cluster emerges as a result of fine - grid two - fluid model simulations , and can not be considered as an input parameter .
it is for this reason that we choose the particle diameter as the relevant length scale for @xmath100 rather than the length scale of a cluster . in figure
[ fig : eta_curves ] ( b ) the @xmath110 dependence of the _ cluster - scale _ effectiveness factor is given for four different values of the dimensionless filter size @xmath102 at fixed @xmath100 and @xmath101 .
the departure of the effectiveness factor from unity is also an increasing function of @xmath102 .
however , figure [ fig : eta_curves ] ( c ) shows that the change in the _ cluster - scale _ effectiveness factor from unity is a decreasing function of @xmath101 when keeping @xmath100 and @xmath102 constant . while one might expect that increasing the @xmath101 value should increase the departure in @xmath90 from unity , we observe a decrease in this departure due to the fact that we are demanding @xmath100 to remain constant in figure [ fig : eta_curves ] ( c ) . in these simulations
the variation of @xmath101 is achieved by varying the value of @xmath8 . in order to maintain a constant @xmath100 the effective rate constant @xmath117
must be changed .
therefore , varying @xmath101 at constant @xmath100 requires the variation in the effective rate constant .
it is this coupled variation that brings about the somewhat surprising the dependence of @xmath90 on @xmath101 .
when deducing a _ filtered _ model from a fine - grid simulation , it is necessary to ensure that _ filtered _ statistics are independent of grid size @xmath118 . to that end , the dependence of @xmath119 on @xmath110 is presented for four different grid resolutions in figure [ fig : eta_grid_dep ] ( a ) at fixed @xmath100 , @xmath102 , and @xmath101 .
the dependence of @xmath90 on grid size @xmath118 is substantial , and only begins to saturate when the grid size is around @xmath120 particle diameters .
figure 2 ( b ) shows the variation of the dimensionless _ filtered _ drag coefficient @xmath121 as a function of @xmath110 for four different grid resolutions . here
the dimensionless _ filtered _ drag coefficient is defined consistent with the earlier work of @xcite @xmath122 where the subscript @xmath123 indicates that the drag coefficient is inferred from the drag force and pressure fluctuation terms directed parallel to gravity . at grid sizes of sixteen
particle diameters @xmath121 seems to exhibit grid independence for @xmath124 , with some grid dependence emerging at higher volume fractions .
figures 3 ( a ) and ( b ) illustrate the sensitive grid dependence of the _ cluster - scale _ effectiveness factor when compared to other _ filtered _ quantities like the fluid - particle drag coefficient .
consequently , even finer grid resolutions are required when simulating reacting gas - particle flow when compared to non - reacting systems .
this observation further supports the need for the development of _ filtered _
models for accurate coarse - grid simulation of reacting systems . due to the sensitivity of @xmath90 with respect to grid size
, we seek to develop a model for the _ cluster - scale _ effectiveness factor extrapolated to the limit of infinite grid resolution . in figures [ fig : rescaled_plots ] ( a ) and ( c ) , plots of @xmath125 versus @xmath126 are given for the finest two grid resolutions presented in this study . here , @xmath127 is the maximum value of @xmath128 , and @xmath129 is the _ filtered _ volume fraction at which the maximum value of @xmath128 occurs . from inspection of figures [ fig : rescaled_plots ] ( a ) and ( c ) it is clear that the grid dependence of the volume fraction variation in @xmath128 can be removed by rescaling @xmath128 by @xmath127 and @xmath110 by @xmath129 for all @xmath130 .
in addition , it is demonstrated in figures [ fig : rescaled_plots ] ( b ) and ( d ) that the grid dependence of the volume fraction variation in @xmath128 can be removed for all @xmath131 by plotting @xmath125 versus @xmath132 .
here @xmath133 is the volume fraction at which the value of @xmath128 reaches zero . due to
the grid - independent behavior observed in figures [ fig : rescaled_plots ] ( a)@xmath25(d ) a filtered model utilizing a piecewise description of the variation in @xmath125 with particle volume fraction about @xmath129 is developed , while extrapolating the values of @xmath127 , @xmath129 , and @xmath133 to infinite resolution . in figure
[ fig : extrap_b ] ( a ) and ( b ) the grid size dependence of @xmath127 is presented for two different values of @xmath100 and a variety of different filter sizes .
the values of @xmath127 are clearly saturating as the value of @xmath118 approaches zero . in order to provide an accurate value of @xmath127 to use in our _ filtered _ reaction rate model , the value of @xmath127 at infinite resolution
is determined via a richardson extrapolation @xcite .
the values of @xmath129 and @xmath133 were observed to vary linearly with grid resolution , and a linear extrapolation was performed to ascertain the infinitely resolved estimates of @xmath129 and @xmath133 .
in the previous section , grid independence of the _ cluster - scale _ effectiveness factor was observed by plotting @xmath125 against scaled volume fraction coordinates that differ depending on whether @xmath110 is greater than or less than @xmath129 .
motivated by this observation , we seek to model the volume fraction variation of @xmath125 via a piecewise function about @xmath129 . in figures [ fig : vol_fr_scaling_delta_left ] ( a)@xmath25(c ) plots of @xmath125 versus @xmath126 are presented for different values of @xmath100 , @xmath102 , and @xmath101 .
the shape of the curve of @xmath125 is clearly altered by changes in @xmath100 and @xmath101 , with only mild changes in the volume fraction dependence of @xmath125 at different grid values of @xmath102 . in figure
[ fig : vol_fr_scaling_delta_left ] ( d ) the variation in @xmath125 with @xmath126 is shown to collapse by plotting all data with the same value of @xmath134 together .
figures [ fig : vol_fr_scaling_delta_right ] ( a)@xmath25(c ) , show the variation of @xmath125 with @xmath135 for different values of @xmath100 , @xmath102 , and @xmath101 .
the @xmath101 variation in the dependence of @xmath125 on @xmath135 is clear , while the @xmath100 and @xmath102 dependence is substantially weaker . in figure
[ fig : vol_fr_scaling_delta_right ] ( d ) we illustrate that the dependence of @xmath125 on @xmath135 for different @xmath100 , @xmath102 , and @xmath101 values can be collapsed onto a single curve provided the value of @xmath134 is kept constant . from figures [ fig : vol_fr_scaling_delta_left ] ( d ) and [ fig : vol_fr_scaling_delta_right ] ( d ) it is clear that the relevant scaling parameter governing the shape of the dependence of @xmath125 on @xmath110 on both sides of @xmath129 is dictated by a single parameter given as @xmath134 .
the following functional forms are used to describe the @xmath110 dependence of @xmath125 : @xmath136 @xmath137 here , @xmath138 and @xmath139 are least - squares fit parameters that depend only on @xmath134 , see figures [ fig : fit_params ] ( a ) and ( b ) .
both model parameters are given by smooth functions of @xmath134 presented in table [ tab : table3 ] .
it should be noted that the observed fluctuation in @xmath139 about the model curve in figure [ fig : fit_params ] ( b ) can be attributed to the uncertainty in determining @xmath133 from the simulation results for @xmath90 . in figure
[ fig : b_dep_delta ] ( a ) @xmath127 is presented as a function of @xmath102 for a variety of different @xmath100 and @xmath101 values .
the variation of @xmath127 with @xmath100 , @xmath102 , and @xmath101 is evident by inspection of figure [ fig : b_dep_delta ] ( a ) .
however , by replotting all data as a function of @xmath140 alone one can collapse all @xmath127 values onto a single master curve in figure [ fig : b_dep_delta ] ( b ) .
a curve fit is presented for @xmath127 in terms of @xmath140 in table 3 .
linear extrapolation of @xmath129 to infinite resolution reveals that @xmath129 varies between @xmath141 , with no discernable trend in @xmath100 , @xmath102 , or @xmath101 .
this variation in @xmath129 arises as a result of the fact that the _ cluster - scale _ effectiveness factor is a weak function of @xmath110 in the region around @xmath129 and thus determining the value of the @xmath129 is subject to error . since no discernable trend in @xmath129 was observed , we recommend the use of @xmath142 because it represents the ensemble average of the different @xmath129 values obtained .
moreover , since @xmath128 only varies slightly in the vicinity of @xmath127 , small errors in @xmath129 will not influence the quantitative behavior our _ filtered _ model substantially .
in addition , the extrapolated value of @xmath133 was found to vary between @xmath143 with no systematic dependence on @xmath100 , @xmath102 , or @xmath101 . as a result of this fluctuation
we choose @xmath144 because this is consistent with earlier work in our group suggesting that _ filtered _ model corrections for gas - particle hydrodynamics become negligible at volume fractions of @xmath145 and higher @xcite . the _ cluster - scale _ effectiveness model developed in this work predicts that effective reaction rates observed in coarse - grid simulations be larger than those predicted in fine grid simulations if the effects of fine scale structure are not accounted for via the _ cluster - scale _ effectiveness factor .
therefore , in order to accurately perform continuum model simulations of reacting gas - particle flows on coarse spatial grids , _ filtered _
models for effective reactions rates are necessary . without these corrections , coarse - grid continuum model simulations will consistently over - estimate the conversion observed in a solid catalyzed reaction ( for example see @xcite ) . therefore , in order to perform accurate coarse - grid numerical simulations of a first - order , isothermal , solid - catalyzed gas - phase reaction we suggest the use of the following _ filtered _ species balance equation @xmath146 where the constitutive relation for @xmath90 is given in table [ tab : table3 ] , and a constitutive relation for the effective dispersion coefficient @xmath87 can be inferred from the work of @xcite . in section [ sec : prelim_study ] , it was shown in that the non - local correction to the effectiveness factor contribute much as @xmath147 to the observed value of @xmath88 . we have observed that this contribution decreases as a function of increasing grid resolution and filter size .
therefore , in the limit of large filter sizes we expect that the contribution of the non - local term @xmath148 in eq .
( [ eq : filtered_model ] ) will be weak , and can thus be neglected .
however , for intermediate filter sizes we recommend the inclusion of this non - local correction . 5 inx @xmath149 @xmath150 @xmath151 @xmath152 while in this work the evolution of species mass fraction was solved in the gas phase alone , one could envision a model for a solid - catalyzed , gas phase reaction where the species mass fraction in both gas and solid phases is tracked separately and coupled through a mass transport term between particle and fluid - phases . in a model framework of this type
, the effective interphase mass transfer rates will be decreased due to the presence of clustering in the gas - particle flow , and any reduction in reactant conversion that arises can be attributed to decreased mass transfer efficiency .
we expect that such an observed decrease in mass transfer coefficient will be on the order of the _ cluster - scale _ effectiveness factor developed in this work . as an example , consider a mass transfer operation taking place between fcc particles and air , where the effective mass transfer coefficient is calculated to be @xmath153 using a model developed by @xcite for fixed beds
. using @xmath154 one can determine the characteristic rate constant for mass transfer @xmath155 , where @xmath156 and @xmath157 are the surface area and volume of an fcc particle , respectively .
for an fcc particle @xmath158 , and using the _ cluster - scale _ effectiveness factor model developed in eq .
( [ eq : b_max ] ) for a filter size of @xmath159 we predict that @xmath160 . using this minimum value of the _ cluster - scale _
effectiveness factor we predict that @xmath161 @xmath25 a nearly 20 fold decrease in the effective mass transfer coefficient ! indeed decreased effective mass transfer coefficients on this order have been found in energy minimization multi - scale model ( emms ) simulations of mass transport processes in gas - particle flows @xcite .
the need for the development of _ filtered _ two - fluid models for reacting gas - particle flows is demonstrated by considering a model first - order , isothermal solid - catalyzed gas - phase reaction .
it is shown that constitutive relations for the _ filtered _ reaction rate and _ filtered _ species dispersion must be postulated to close the _ filtered _ species balance equation . due to the fact that the model gas - phase reaction in this work is isothermal and produces no volume change , only _ filtered _ species balance equations must be developed , without the need to alter the existing _ filtered _ gas and solid momentum balance equations given in the earlier works of @xcite and @xcite for non - reacting , monodisperse gas - particle flows . using fine - grid continuum model simulations of reacting gas - particle flows
we extract the _ cluster - scale _ effectiveness factor , defined as the ratio of the fine - grid reaction rate to the reaction rate observed in a coarse grid simulation .
cluster - scale _ effectiveness factor is observed to retain an inverted bell shaped dependence on volume fraction approaching unity in both the low and high volume fraction limits . at intermediate volume fractions a decrease in the _ cluster - scale _ effectiveness factor from unity is observed with the magnitude of this reduction being a strong function of _ meso - scale _ thiele modulus @xmath100 , and dimensionless filter size @xmath102 with only a weak dependence on schmidt number @xmath101 .
due to the sensitivity of the _ cluster - scale _ effectiveness factor with grid size , we determine an asymptotic form for the cluster - scale effectiveness factor relying on a richardson extrapolation of the minimum in the cluster scale effectiveness factor .
the extrapolated values of the cluster - scale effectiveness factor are shown to collapse when plotted as a function of @xmath140 , with the volume fraction dependence of the cluster - scale effectiveness factor collapsing for fixed @xmath134 .
table [ tab : table3 ] presents a curve fit of our collapsed results for use in coarse - grid simulations of reacting gas - particle flows with first - order reaction kinetics .
the reduction in effective reaction rates observed in this study can be used to rationalize the over - prediction in reactant conversion that was seen in the two - fluid model simulations of @xcite .
finally , we note that the filtered model developed in this work is limited to first - order reaction kinetics , and was restricted to two - dimensional periodic domain simulations . future work should extend such analyses to other reaction kinetics , and three - dimensional bounded domains .
however , we expect that the characteristic volume fraction , thiele modulus , filter size , and schmidt number dependence of @xmath90 will be qualitatively similar to the results presented here .
@xmath162 fluid - particle drag coefficient @xmath20 particle radius + @xmath138 fit parameter for plot of @xmath125 versus @xmath126 + @xmath139 fit parameter for plot of @xmath125 versus @xmath132 + @xmath128 @xmath119 + @xmath21 biot number for mass transport @xmath163 + @xmath127 maximum value of @xmath128 @xmath54 particle diameter + @xmath5 effective diffusivity of reacting species in continuum model + @xmath8 molecular diffusivity of reacting species + @xmath24 _ intra - particle _ species diffusivity + @xmath87 _ meso - scale _ dispersion coefficient + @xmath63 coefficient of restitution of the particle phase + @xmath164 fluid - particle drag force experienced by the fluid + @xmath3 gravitational acceleration vector + @xmath165 radial distribution function at contact + @xmath69 weight function for filtering + @xmath166 collisional dissipation of granular energy + @xmath167 viscous dissipation of granular energy + @xmath16 intrinsic reaction rate constant + @xmath168 effective rate constant for mass transfer + @xmath117 effective reaction rate constant + @xmath169 _ meso - scale _ effective rate constant for mass transfer + @xmath23 convective mass transport coefficient + @xmath170 @xmath92 + @xmath171 gas - phase pressure + @xmath172 granular energy conduction vector + @xmath173 rate of production of species _
i _ + @xmath97 reynolds number for the gas - phase + @xmath156 surface area of a particle + @xmath174 rate of deformation tensor + @xmath101 _ meso - scale _ schmidt number @xmath175 + @xmath176 granular temperature + @xmath177 gas velocity + @xmath178 particle velocity + @xmath4 terminal settling velocity of an isolated particle + @xmath106 volume of a periodic domain + @xmath157 volume of a particle + @xmath70 generic spatial position vector + @xmath70 spatial position of filter center + @xmath66 generic spatial position vector + _ greek letters : _
+ @xmath179 fluid - particle friction coefficient + @xmath180 production of granular energy through interphase slip + @xmath118 dimensional grid size + @xmath2 dimensional filter size + @xmath102 dimensionless filter size @xmath1 + @xmath181 gas volume fraction + @xmath182 @xmath183 + @xmath18 intraparticle effectiveness factor + @xmath88 _ cluster - scale _ effectiveness factor + @xmath90 _ non - locally corrected _
_ cluster - scale _ effectiveness factor + @xmath104 ratio of @xmath15 to its domain - averaged value + @xmath184 conductivity of granular energy + @xmath185 bulk viscosity of particle phase + @xmath7 molecular gas - phase shear viscosity + @xmath6 effective gas - phase shear viscosity + @xmath186 shear viscosity of particle phase + @xmath59 gas density + @xmath56 particle density + @xmath187 gas - phase stress tensor + @xmath188 particle - phase stress tensor + @xmath11 particle volume fraction + @xmath100 _ meso - scale _ thiele modulus @xmath189 + @xmath190 thiele modulus @xmath191 + @xmath15 mass fraction of gas species _
the authors would like to acknowledge the financial support from exxonmobil research @xmath192 engineering co. and the u.s .
department of energy , office of fossil energy s carbon capture simulation initiative through the national energy technology laboratory .
in section [ sec : filt_two_fluid ] the cluster scale effectiveness factor is defined in two ways given by eqs .
( [ eq : effect_fac])@xmath25([eq : effect_fac2 ] ) . here , the effectiveness factor given in eq .
( [ eq : effect_fac2 ] ) is derived relying on the definition of the effectiveness factor given in eq .
( [ eq : effect_fac ] ) .
let the particle phase volume fraction @xmath11 and species mass fraction @xmath15 at any point be represented as follows @xmath193 where @xmath66 is a spatial variable associated with _ filtered _ and microscopic variables , respectively , neither of which are located at the filter center . assuming the _ filtered _
variables can be given by smooth functions of space the _ filtered _ value of @xmath11 and @xmath15 can be approximated via the following taylor series @xmath194 @xmath195 the taylor expansions given in eqs .
( [ eq : phi_tayl_filt ] ) and ( [ eq : tayl_filt ] ) can be plugged into eq .
( [ eq : decomp ] ) to yield expressions for @xmath11 and @xmath15 that depend only on the location of the filter center @xmath70 and the microscopic spatial variable @xmath66 .
utilizing eqs .
( [ eq : decomp])@xmath25([eq : tayl_filt ] ) the _ filtered _ product of @xmath11 and @xmath15 can be expressed as @xmath196 where @xmath92 . removing the gradient term that appears on the right hand side of eq .
( [ eq : non_l_rem ] ) provides a method to define _ filtered _ constitutive relations for @xmath90 that depend on local _ filtered _ quantities alone , without considering gradients in local _ filtered _ variables .
this approach of removing non - local effects in the process of _ filtered _ model development has recently been advanced when considering the development of _ filtered _ fluid - particle drag models for monodisperse gas - particle flows @xcite , and this formulation is extended here to reacting gas - particle flows . combining the definition of @xmath88 and @xmath90 given in eqs .
( [ eq : effect_fac ] ) and ( [ eq : effect_fac2 ] ) with eq .
( [ eq : non_l_rem ] ) one can arrive at the following two expressions @xmath197 @xmath198 therefore , @xmath90 provides a direct measure of the product of local fluctuations in @xmath110 and @xmath108 , while the value of @xmath88 is influenced by gradients in local filtered quantities . in order to interrogate the effect of local fluctuations alone , we choose to model @xmath90 in this study .
igci , y. , pannala , s. , benyahia , s. , sundaresan , s. , 2011 .
validation studies on filtered model equations for gas - particle flows in risers .
industrial @xmath192 engineering chemistry research , in press .
kashyap , m. , gidaspow , d. , oct .
computation and measurements of mass transfer and dispersion coefficients in fluidized beds .
powder technology 203 ( 1 ) , 4056 .
http://linkinghub.elsevier.com / retrieve / pii / s0032591010% 001531[http://linkinghub.elsevier.com / retrieve / pii / s0032591010% 001531 ] zimmermann , s. , taghipour , f. , dec . 2005 .
cfd modeling of the hydrodynamics and reaction kinetics of fcc fluidized - bed reactors .
industrial @xmath192 engineering chemistry research 44 ( 26 ) , 98189827 . | using the kinetic - theory - based two - fluid models as a starting point , we develop _ filtered _ two - fluid models for a gas - particle flow in the presence of an isothermal , first - order , solid - catalyzed reaction of a gaseous species . as a consequence of the _ filtering _ procedure , terms describing the _ filtered _ reaction rate and _ filtered _ reactant dispersion need to be constituted in order to close the _ filtered _ species balance equation . in this work , a constitutive relation for _ filtered _ reaction rate is developed by performing fine - grid , two - fluid model simulations of an isothermal , solid - catalyzed , first - order reaction in a periodic domain .
it is observed that the _ cluster - scale _ effectiveness factor , defined as the ratio between the reaction rate observed in a fine - grid simulation to that observed in a coarse - grid simulation , can be substantially smaller than unity , and it manifests an inverted bell shape dependence on _ filtered _ particle volume fraction in all simulation cases .
moreover , the magnitude of the deviation in the _ cluster - scale _ effectiveness factor from unity is a strong function of the _ meso - scale _ thiele modulus and dimensionless filter size .
thus coarse - grid simulations of a reacting gas - particle flow will overestimate the reaction rate if the _ cluster - scale _ effectiveness factor is not accounted for .
reactive flows , fluidization , fluid mechanics , two - fluid model |
cyg x-3 is a very unusual x - ray binary ( see @xcite for a review ) which does not fit well into any of the established classes of x - ray binaries .
a 4.8 hour modulation , thought to be the orbital period of the binary , is observed at x - ray ( @xcite , @xcite ) and infrared ( @xcite , @xcite ) wavelengths .
the 4.8 hour orbital period is typical of a low mass x - ray binary , but infrared observations suggest that the mass donating companion is a high mass wolf - rayet star ( @xcite ) . due to the distance ( @xmath1 8.4 kpc ; see @xcite ) , and its location in the galactic plane , no optical counterpart has been found .
cyg x-3 is also a strong source of radio emission with three types of behavior : ( a ) quiescence , ( b ) major flaring with quenching ( very low radio fluxes ) , and ( c ) minor radio flaring with partial quenching ( see waltman et al .
1994 , 1995 , 1996 ) . during the major radio outbursts
there is strong evidence of jet - like structures moving away from cyg x-3 at a velocity of either @xmath0 0.3c or @xmath0 0.6c , depending upon whether the jets are one - sided or double - sided ( @xcite , @xcite ) .
recent vlba images after a large cyg x-3 flare showed one - sided ejection with v @xmath2 0.8c ( @xcite ) . the soft x - ray emission ( below 20 kev ) from cyg
x-3 has been observed to undergo high and low states in which the various spectral components change with no apparent correlation with other properties ( @xcite ) . from ginga observations ( @xcite ) , it was found that the giant radio flares occur only when the soft x - ray flux is high .
hard x - ray ( hxr ) observations ( above @xmath0 50 kev ) , prior to 1991 , found the hxr emission to vary by less than 20% ( @xcite , @xcite ) .
recent compton gamma - ray observatory ( cgro ) observations of cyg x-3 made using osse have shown hxr emission to be variable by a factor of 3 ( @xcite ) . in this letter , we report results obtained from the nearly continuous monitoring of cyg x-3 for a period of almost 5 years in both the hxr ( cgro / batse ) and the radio wavelength bands .
cyg x-3 has been monitored in the hxr and radio as part of a study to understand the high - energy properties of binary x - ray sources and how they relate to the production of jets .
the batse experiment onboard cgro ( @xcite ) was used to monitor the hxr emission from cyg x-3 .
the batse large area detectors ( lads ) can monitor the whole sky almost continuously in the energy range of 20 kev2 mev with a typical daily 3 @xmath3 sensitivity of about 100 mcrab . to produce the cyg x-3 light curve ,
single step occultation data were obtained using a standard earth occultation analysis technique for monitoring hxr sources ( @xcite ) .
interference from known bright sources ( such as cyg x-1 ) and single steps with large deviations from the median flux measurement ( the result of the hxr flaring behavior of cyg x-1 ) were removed .
the single occultation step data were then fit with a power law with a fixed photon spectral index of @xmath4 to determine a flux in the 20100 kev band .
the spectral index was chosen based on spectral fitting of the batse data during periods of high levels of emission and is consistent with the index found by @xcite from osse observations .
varying the spectral index , for an acceptable range of spectral indices , results in a 1015 % variation in the measured flux .
a lomb - scargle periodogram ( @xcite , @xcite ) was created for the single step occultation data to search for periodicities in the data .
the known 4.8 hour period of cyg x-3 was detected in the batse data set at a very high significance ( @xcite ) , with no additional periodicities being found in the data .
sums of single step occultation data of @xmath5 2 day duration were found to uniformly sample all phases of the 4.8 hour period .
1 shows a history of the 20100 kev hxr emission from cyg x-3 .
the batse data were binned with each point containing 40 occultation steps , with each occultation step having an integration time of 110 seconds .
this corresponds to about three days of observations per data point .
cyg x-3 ranges from less than batse s three day detection limit of @xmath0 40 mcrab to intensities at least 7 times brighter .
the two most distinctive features in the hxr light curves are ( a ) extended periods of high hxr flux ( 0.04 0.08 @xmath6 ) and ( b ) low flux periods where the flux level drops below the detection limit of batse .
cyg x-3 was observed on a daily basis as part of the naval research laboratory green bank interferometer ( gbi ) monitoring program .
the gbi consists of two 25.9 m diameter antennas separated by a 2.4 km baseline .
the antennas each observe simultaneously at 2.25 and 8.3 ghz with 35 mhz of total bandwidth at each frequency .
the observing and data reduction techniques are described in @xcite , @xcite , and @xcite .
the average integration time for individual scans is approximately 10 minutes with up to 12 scans made on the source per day .
typical errors ( 1 @xmath3 ) in the gbi data set are flux density dependent @xmath7 4 mjy ( 2 ghz ) or 6 mjy ( 8 ghz ) for fluxes less than 100 mjy ; 15 mjy ( 2 ghz ) or 50 mjy ( 8 ghz ) for fluxes @xmath0 1 jy . in fig .
1 the 2.25 ghz radio flux densities are shown overlaid on a log scale for the same time period as the hxr measurements .
what is striking to the eye is an overall anticorrelation between the radio and the hxr emission . at the times
when the radio has become quiescent the hxr increases and goes into a state of high emission . with the occurrence of radio
flares the hxr emission drops and during the major flaring activity the behavior of the hxr and radio changes to a correlation ( see fig .
2 ) . to test fully the existence of these ( anti)correlations we calculated correlation coefficients ( @xmath8 ) and probability functions of the spearman rank - order . to assure that each pair of points is statistically independent , rejection of some fraction of the data to eliminate any internal correlation is necessary . for each of the two time series we divided the data into two sets of observations consisting of alternating measurements from the original time series .
these series were tested and if an internal correlation was found , we then rejected half of the data points and repeated the procedure with the remaining data .
we continued this process until at least one of the two different time series showed no internal correlation , then tested the matching data points in the time series of the two wavelength bands for a correlation . in all cases we consider a correlation to exist if the probability of no correlation was @xmath9 or less .
table 1 shows the results of correlation testing using 3 day averages of the data for different sets of the data . for each type of activity
is given ( a ) the spearman correlation coefficient , ( b ) probability of no correlation , ( c ) initial size of the data set , ( d ) size of the data set after data rejection , and ( e ) the wavelength band ( r : radio , x : hxr , b : both ) in which there is no internal correlation .
_ entire data sets : _ during times of high hxr flux the radio data show less variability and as flaring occurs , the hxr starts to drop .
one can define a variability index by taking the standard deviation ( @xmath3 ) in the radio data , over the time averaged interval , and dividing it by the mean value of the radio flux ( @xmath10s@xmath11 ) for the time averaged interval .
table 1 and fig .
3 shows the existence of a strong anticorrelation between the hxr flux and the variability index of the radio flux . _
quiescence radio activity : _ for the periods of quiescent radio emission in our data sets we see that there is a very strong anticorrelation between the hxr and the radio fluxes . _ major flaring activity : _ for major flaring activity one can see there exists a strong correlation between the hxr and the radio fluxes ( see fig .
osse observations made during the first major radio flare activity period showed a factor of 3 drop in the hxr as the radio became quenched prior to a major flare and later observations showed a higher flux level as cyg x-3 returned to quiescence ( @xcite ) in agreement with the batse data . _ minor flaring activity : _ during the minor flaring there was no significant correlation found between the hxr and radio fluxes .
4 is a radio versus hxr flux plot for the radio quiescent periods and the major radio flare activity . in this plot
we see two branches which describe the correlated behavior which we have found .
there is a horizonal branch which describes the quiescent radio state and a vertical branch which describes the major radio flaring .
it is of interest to note that where the two branches overlap one finds the minor radio flaring data points .
for the first time , we report correlated hxr and radio emission in cyg x-3 ( see table 2 for a summary ) .
the high degree of anticorrelation , during radio quiescence , between the wavelength bands indicates that the same high energy particle production mechanism may be responsible for the particles which produce both the radio ( synchrotron ) and hard x - ray ( compton processes ) .
however the efficiency of inverse - compton processes ( @xcite ) in cooling the high energy electrons required to produce the synchrotron radio emission make it unlikely the hxr and radio originate from the same emission region
. these high energy particles may be the result of either the production of a `` hot '' corona ( @xcite ) or the presence of a low - level jet ( @xcite ) , during a period of low mass transfer in the system .
the major radio flaring activity represents a distinct change in the behavior of the system .
the quenched radio emission preceding large to medium radio flares clearly corresponds to low hxr states in the batse data .
such episodes have been interpreted ( @xcite ) , based on the increase in soft x - rays ( @xcite ) , as an indication of a large increase in mass flow within an accretion disk prior to a major radio flare .
the nature of this radio emission during the preflare episodes is not understood , but possibly both the hxr and radio fluxes are generated in a region ( initially optically thick ) near the central object .
it is apparent that the quenched radio emission and the low hxr state are tied to the formation of the jet .
when a large radio flare occurs , there is a corresponding increase in the hxr flux .
as the radio flare decays , the hxr also drops on a similar timescale .
the decay in the hxr may be interrupted by subsequent flares or a transition to a radio quiescent state .
the radio flare can be interpreted as a consequence of an ejection event from a bipolar or one - sided jet ( @xcite , @xcite ) where the radio flare is due to the growth in solid angle and expansion of the developing jets .
the hxr emission is most likely the result of inverse compton processes , but the site of this emission is less certain .
early observations of the galactic jet sources , grs 1915 + 105 ( @xcite ) and gro j1655 - 40 ( @xcite ) , which showed a correlation between the hxr and radio , gave rise to models in which the hxr emission arose from the jet ( @xcite ) .
but later observations of these two sources have shown hxr outbursts with which there was no accompanying radio jets ( gro j1655 - 40 ; @xcite ) or there was an anticorrelation between the radio flare and the hxr flux ( grs 1915 + 105 ; @xcite ) . while there may be uncertainty on the site of the hxr emission
it is clear that there is hxr response for all of the major flares ( and associated jets ) in cyg x-3 . during minor radio flaring periods there
occur smaller mini - flares " and a drop in the overall hxr flux which is reflected in the anti - correlation between hxr flux and the radio variability index .
it is possible that conditions exist here similar to those present during the major radio outbursts , although the relatively low intensity in hxrs makes it difficult to check if the faster variability seen in the radio is also reflected at the same timescales in hxrs .
@xcite have suggested that the x - ray and radio activity are mediated by the dense wind from the companion star . in the quiescent radio state ,
the wind density is predicted to be low .
changes in the wind may determine whether the source changes from a radio quiescence / hxr high state to radio flaring / hxr low state .
basic research in radio astronomy at the naval research laboratory is supported by the office of naval research .
the green bank interferometer was operated by the national radio astronomy observatory under contract to the u.s .
naval observatory during the time period of these observations .
the national radio astronomy observatory is operated by associated universities , inc . under a cooperative agreement with the national science foundation .
lccccc hxr vs. @xmath3/@xmath10s(2250)@xmath11 & 0.489 & 2 x @xmath12 & 546 & 136 & r hxr vs. @xmath3/@xmath10s(8300)@xmath11 & 0.405 & 1 x @xmath13 & 546 & 136 & r hxr vs. 2.25 ghz & 0.477 & 1 x @xmath14 & 254 & 127 & x hxr vs. 8.3 ghz & 0.491 & 4 x @xmath12 & 254 & 127 & x hxr vs. 2.25 ghz ( major ) & 0.738 & 5 x @xmath14 & 40 & 40 & b hxr vs. 8.3 ghz ( major ) & 0.663 & 3 x @xmath13 & 40 & 40 & b hxr vs. 2.25 ghz ( minor ) & 0.168 & 9 x @xmath9 & 243 & 243 & r hxr vs. 8.3 ghz ( minor ) & 0.190 & 4 x @xmath15 & 243 & 121 & b lll & there exists a anticorrelation between & 836110175 & hxr and the variability in the radio . & the hxr and radio show a strong & 83618400 , 85008557 , 86358705 , & anticorrelation .
it is during this & 89609050 , 91709330 , 95709610 , & state that the hxr reaches its & 96309800 , 98609975 , 1003810175 & highest levels .
& the hxr and radio correlate with & 84158500 , 88608876 , 93859432 & the hxr vanishing during quenching & and recovering during the major & flare .
& the hxr and radio show no & 85578635 , 87058860 , 88768960 , & correlation for this activity . &
90509170 , 93309385 , 94329570 , & & 96109630 , 98009860 , 997510038 | using cgro / batse hard x - ray ( hxr ) data and ghz radio monitoring data from the green bank interferometer ( gbi ) , we have performed a long term study ( @xmath0 1800 days ) of the unusual x - ray binary cyg x-3 resulting in the discovery of a remarkable relationship between these two wavelength bands .
we find that , during quiescent radio states , the radio flux is strongly anticorrelated with the intensity of the hxr emission .
the relationship switches to a correlation with the onset of major radio flaring activity . during major
radio flaring activity the hxr drops to a very low intensity during quenching in the radio and recovers during the radio flare .
injection of plasma into the radio jets of cyg x-3 occurs during changes in the hxr emission and suggests that disk - related and jet - related components are responsible for the high energy emission . |
electric properties provide a unique window on the structure and dynamics of one - neutron halo nuclei . two prominent examples of one - neutron halo nuclei are @xmath6be and @xmath7c @xcite . due to their weakly - bound nature
, the length scales @xmath2 of the halo and @xmath1 of the core are separated . as a consequence
, observables can be calculated using halo eft @xcite which exploits this separation of scales between @xmath1 and @xmath2 .
the halo eft formalism has been successfully used to study various reactions and properties of halo - like systems .
some examples include @xmath3-wave @xmath8 resonance in @xmath9be @xcite , universal properties of two - neutron halo nuclei @xcite , radiative neutron / proton capture on @xmath10li @xcite and on @xmath10be @xcite and the ground state of @xmath11c @xcite . in this work ,
we follow the approach presented in ref .
@xcite , where electric properties of @xmath12be are calculated using halo eft .
the @xmath13 ground state of @xmath12be is predominantly an @xmath3-wave bound state and the @xmath14 first excited state predominantly a @xmath15-wave bound state .
we apply the halo eft framework to @xmath16c , which has also a @xmath13 ground state .
in contrast to @xmath12be , the first excited @xmath17 state of @xmath16c is predominantly a @xmath4-wave bound state .
we include the strong @xmath4-wave interaction by introducing a new dimer field and compute electric form factors and the e2 transition strength . in a similar way , electromagnetic form factors for the ground state of @xmath16c and the radiative neutron capture on @xmath18c to the @xmath13 ground state of @xmath16c are calculated in refs .
@xcite , respectively .
the neutron separation energy of the @xmath13 state of @xmath16c is 1218 kev and the neutron separation energy of the @xmath17 state is 478 kev @xcite .
the first excitation of the @xmath18c nucleus is 6.1 mev above the @xmath19 ground state .
compared to the bound states of @xmath18c the weakly - bound states of @xmath16c indicate their one - neutron halo nature . converting these energy scales into the relevant distance
scales @xmath1 and
@xmath2 in this system we can carry out our eft expansion in powers of @xmath20 . since the ratio is not particularly small , effects beyond lo are not negligible in our theory .
we follow the halo eft formalism developed in ref .
the relevant degrees of freedom are the @xmath18c core , represented by a bosonic field @xmath21 , and the halo neutron , represented by a spinor field @xmath22 .
the strong @xmath3-wave and @xmath4-wave interactions are included through the incorporation of auxiliary spinor fields @xmath23 and @xmath24 , which correspond to the @xmath13 and @xmath17 states , respectively
. with the convention that repeated spin indices are summed , the non - relativistic lagrangian is written as : @xmath25 \sigma_s + d_{m}^\dagger \left[c_2 \left(i \partial_t + \frac{\nabla^2}{2m_{nc}}\right)^2 + \eta_2 \left(i \partial_t + \frac{\nabla^2}{2m_{nc}}\right ) + \delta_2 \right ] d_{m}\\ \label{eq : lagrangian } & - g_0 \left[c^\dagger n_s^\dagger \sigma_s + \sigma_s^\dagger n_s c\right ] - g_2 \left[d^\dagger_{m } \left[n \overset{\leftrightarrow}\nabla^2 c\right]_{\frac{5}{2},m } + \left[c^\dagger \overset{\leftrightarrow}\nabla^2 n^\dagger \right]_{\frac{5}{2},m } d_{m } \right ] \ , \end{aligned}\ ] ] where @xmath26 is the galilean invariant derivative , @xmath27 denotes the neutron mass , @xmath28 the core mass and @xmath29 is the total mass of the n-@xmath18c system and we project out the @xmath30 part of the strong @xmath4-wave interaction . , and the thin , solid line the neutron .
the thin double line represents the bare dimer propagator , and the thick double line with the gray circle is the dressed dimer propagator .
the left panel shows the diagrammatic representation of the dyson equation for the dressed dimer propagator and the right panel the neutron - core scattering amplitude with the dressed dimer propagator.,scaledwidth=99.0% ] the dressed @xmath4 propagator and the @xmath4-wave scattering amplitude are computed from the diagrams in fig .
[ fig : dimer - propagator ] in the same way as in ref .
the dyson equation for the @xmath4-wave is illustrated in the left panel of fig .
[ fig : dimer - propagator ] with the one - loop self - energy @xmath31 .
we use the power divergence subtraction ( pds ) as regularization scheme @xcite . the @xmath4-wave scattering amplitude in the two - body center - of - mass frame with @xmath32 and @xmath33 for on - shell scattering
is matched to the effective range expansion ( ere ) .
since we have to renormalize arising divergences , the effective range parameters @xmath34 and @xmath35 are required at lo , which leads to the power - counting scheme below .
furthermore , we need the additional higher order kinetic term with coupling constant @xmath36 for the bare propagator in our lagrangian @xcite .
the assumed power counting for the @xmath37 propagator follows the ideas of refs .
we apply the constraint to have the minimal number of fine tunings possible in our power - counting scheme .
this requirement is motivated by the fact that every additional fine tuning makes our scenario less likely to be found in nature , as discussed in ref .
@xcite . however , the number of fine tunings for higher partial waves @xmath38 is larger than proposed in ref . @xcite . since arising divergences in the calculation of the dimer propagator have to be renormalized properly , the number of ere parameters needed at lo for the @xmath39-th partial wave is @xmath40
thus , for the @xmath39-th partial wave and @xmath41 we need @xmath39 fine - tuned parameters .
this suggests that higher partial wave bound states are less likely to appear in nature which is consistent with experimental observations . in the case of the @xmath42 propagator ,
two out of three ere parameters need to be fine - tuned , because @xmath43 and @xmath44 are both unnaturally large .
this scenario is more likely to occur than three unnaturally large parameters and three fine - tunings such that @xmath45 .
higher ere terms are suppressed by powers of @xmath0 .
thus , the relevant fit parameters in our eft at lo are @xmath46 , @xmath47 , @xmath48 , and @xmath35 .
electromagnetic interactions are included via minimal substitution in the lagrangian , @xmath49 .
additionally , we have to consider further , local , gauge - invariant operators involving the electric field @xmath50 and the fields @xmath21 , @xmath22 , @xmath51 and @xmath4 .
depending on the observable and respective partial wave they contribute at different orders in our eft .
photon to couple to the @xmath18c - neutron @xmath4-wave bound state field up to nlo . diagram ( a ) emerges from the minimal substitution in the strong @xmath4-wave interaction and contributes at nlo while diagram ( b ) arises since @xmath48 , @xmath35 and the local , gauge - invariant operator @xmath52 contribute at lo .
diagram ( c ) : the two lo diagrams contributing to the irreducible vertex that determines the @xmath3-to-@xmath4 state transition in halo eft.,scaledwidth=95.0% ] the result for the electric form factor of the @xmath3-wave bound state is given in ref .
@xcite for @xmath12be and in ref .
@xcite for @xmath16c .
we follow the approach of the first reference and compute the @xmath4-wave form factors by calculating the contribution to the irreducible vertex for @xmath53 interactions up to nlo shown in fig .
[ fig : formfactorse2 ] .
the computation is carried out in the breit frame , where @xmath54 , and the irreducible vertex for the @xmath55 photon coupling to the @xmath4 state in cartesian coordinates yields : @xmath56 \ , \end{aligned}\ ] ] with the three momentum of the virtual photon @xmath57 and three different @xmath4-wave tensors for each form factor @xmath58 , @xmath59 and @xmath60 .
the neutron spin is unaffected by the charge operator up to the order considered here and we get results for the electric @xmath61 , quadrupole @xmath62 and hexadecupole @xmath63 form factors . the local , gauge - invariant operator @xmath52 is needed at lo to counter the arising divergences for the charge radius and quadrupole moment . at lo
we obtain the following results for the charge and quadrupole radii and the quadrupole and hexadecupole moment : + @xmath64 with @xmath65 .
since the quadrupole radius @xmath66 and the hexadecupole moment @xmath67 differ only by a constant factor at lo , we obtain a smooth correlation between both observables @xmath68 .
the diagrams contributing to the irreducible vertex for the e2 transition from the @xmath69 state to the @xmath70 state at lo are shown in fig .
[ fig : formfactorse2 ] .
the photon has a four momentum of @xmath71 and its polarization is denoted by @xmath72 .
the computation of both diagrams yields a vertex function @xmath73 , where @xmath74 is the total angular momentum projection of the @xmath70 state and @xmath3 denotes the spin projection of the @xmath69 state .
we calculate the irreducible vertex in coulomb gauge such that we have @xmath75 for real photons and obtain for b(e2 ) : @xmath76 ^ 2 \ , \end{aligned}\ ] ] where the effective charge is given by @xmath77 @xcite . | we exploit the separation of scales in weakly - bound nuclei to compute e2 transitions and electric form factors in a halo eft framework .
the relevant degrees of freedom are the core and the halo neutron .
the eft expansion is carried out in powers of @xmath0 , where @xmath1 and @xmath2 denote the length scales of the core and halo , respectively .
we include the strong @xmath3-wave and @xmath4-wave interactions by introducing dimer fields .
the dimer propagators are regulated by employing the power divergence subtraction scheme and matched to the effective range expansion in the respective channel .
electromagnetic interactions are included via minimal substitution in the lagrangian .
we demonstrate that , depending on the observable and respective partial wave , additional local gauge - invariant operators contribute in lo , nlo and higher orders .
@xmath5 this work has been supported by deutsche forschungsgemeinschaft ( sfb 1245 ) . |
the mounting evidences suggest that the universe is expanding with acceleration @xcite .
this reveals that the universe is dominated by dark energy with negative pressure , whose energy density fraction is about 2/3 at present .
the nature and origin of dark energy is still mysterious , leaving theoretic physicts searching for novel answers .
althoug a cosmological constant may be responsible for the accelerating expansion of the universe and be consistent with current observational data , the extremely smallness of the value of the cosmological constant brings a big challenge to us .
it is also possible that the mechanism at work is dynamical .
a dynnamical candidate to provide dark energy may be supplied by a slowly rolling scalar field , widely referred as `` quintessence field '' @xcite .
there are also models of a scalar field with non - canonical kinetic term , known as `` k - essence '' @xcite or `` tachyonic '' models @xcite .
other alternative dark energy models include the holographic dark energy model @xcite and the extra - dimensional motivated models @xcite . among the cosmological community ,
the consensus about our universe is the so called `` concondance model '' : 95% percent of the universe is composed of dark components .
recently the chaplygin gas model with exotic equation of state @xmath5 was proposed to explain both dark enrgy and dark matter @xcite .
the chaplygin gas model was later generalzied to the generalized chaplygin gas ( gcg ) model with equation of state @xmath6 @xcite .
the novel feature of this model is that it unifies dark energy and dark matter in one model .
sandvik and coauthors claimed that the matter power spectrum essentially ruled the gcg model out @xcite .
however , their analysis does not include the effect of the baryons .
bea and collabrators showed that it is important to include baryons in the study of large scale structure and the conclusion changed when baryons were included @xcite .
it should also be noted that the results in @xcite was based on linear theory of perurbations and neglected any non - linear effects .
the gcg model has been extensively studied in the literature @xcite .
the gcg model was also studied in the framework of modified friedman equation in @xcite . in this paper
, we consider both dust like matter and gcg as the source .
the dust like matter may be just baryons or a portion of dark matter .
unlike other studies , we allow the parameter @xmath7 to be less than @xmath8 and the parameter @xmath9 to be in the region @xmath10 in addition to the usual region @xmath11 .
some physically reasonable conditions are then applied to the model to constrain the parameters .
therefore , the model considered here is more general in addition to be physical .
in a homogeneous and isotropic universe , the friedmann - robertson - walker ( frw ) space - time metric is @xmath12.\ ] ] for a null geodesic , we have @xmath13 where @xmath14 with both an ordinary pressureless dust matter and gcg as sources , the friedmann equations read @xmath15 where the hubble parameter @xmath16 , dot means derivative with respect to time , @xmath17 is the matter energy density , a subscript 0 means the value of the variable at present time . by using the gcg equation of state @xmath6
, we get the solution to eq .
( [ cos2 ] ) as @xmath18^{1/(1+\alpha)}.\ ] ] because @xmath19 , so @xmath20 .
substitute this expression into eq .
( [ eqrhoch ] ) , we get @xmath21 . therefore , eq . ( [ eqrhoch ] ) can be expressed in terms of @xmath22 and @xmath23 as @xmath24^{1/(1+\alpha)}.\ ] ] it is obvious that the generalized chaplygin gas behaves like the cosmological constant when @xmath25 and it behaves like the dust matter when @xmath26 . at early times , i.e. , the cosmological radius @xmath27 is small , @xmath28 , which corresponds to a dust like dominated universe . at late times , i.e. , the cosmological radius @xmath27 is large , @xmath29 , which corresponds to a cosmological constant like dominated universe .
therefore the generalized chaplygin gas interpolates between a dust dominated phase in the past and a de - sitter phase at late times .
this distinct feature makes the model an intriguing candidate for the unification of dark matter and dark energy .
the gcg equation of state can be derived from the generalized born - infeld action @xcite @xmath30^{\alpha/(1+\alpha)},\ ] ] where @xmath31 . from the above lagrangian
, we can easily get @xmath32 .
if we take gcg as a quintessence field , i.e. , if we take @xmath33 , then if gcg is the only source in a spatially flat universe , the potential for gcg is @xmath34 , \end{split}\ ] ] here we set @xmath35 . by using eq .
( [ eqrhoch1 ] ) , we get @xmath36^{-\alpha/(1+\alpha)}.\ ] ] if @xmath37 , then to the first order of expansion , @xmath38 and @xmath39 are @xmath40,\ ] ] @xmath41.\ ] ] from the above expressions , we see a mixture of a cosmological constant with a type of dark energy described by a constant equation of state parameter @xmath9 .
so the physical meaning of @xmath9 may be given in this sense .
follow chiba and nakamura @xcite , we require the following physically reasonable conditions : ( 1 ) the current total energy density is non - negative ; ( 2 ) the total energy density is not increasing ; ( 3 ) the present sound speed @xmath42 of the system satisfies @xmath43 because of causality and local stability .
the first condition gives @xmath44 , where @xmath45 .
the second condition tells us that @xmath46 , where the deceleration parameter @xmath47 .
the second condition is equivalent to the requirement that the effective equation of state for the total source @xmath48 .
the third condition gives @xmath49 , where the jerk parameter @xmath50 .
the deceleration parameter @xmath51 and the jerk parameter @xmath52 are similar to the state finder parameters used in @xcite . for the generalized chaplygin gas model , we get @xmath53 @xmath54 where @xmath55 and @xmath56 . by using the above eqs .
( [ eqj0 ] ) and ( [ eqq0 ] ) , we get the following constraints @xmath57 @xmath58 @xmath59 furthermore , we require that @xmath60 and @xmath61 .
to get accelerated expansion , we also require that @xmath62 .
therefore , we have one additional constraint @xmath63 in the literature , the parameters are usually constrained to be @xmath33 and @xmath64 .
some authors also considered the possibility of @xmath65 . from eqs .
( [ cond1]-[cond4 ] ) , we see that it is possible to get @xmath66 if @xmath67 . in this paper , we consider the parameter space to be : @xmath68 $ ] , @xmath69 and @xmath70 , @xmath71 $ ] .
in addition to the constraints ( [ cond1]-[cond4 ] ) on the parameters , we also require that the energy density of gcg given in eq .
( [ eqrhoch1 ] ) is not negative .
combining eqs .
( [ cos1 ] ) , ( [ cos2 ] ) and ( [ eqrhoch1 ] ) , we get the transition redshift @xmath72 when the expansion of the universe underwent the transition from deceleration to acceleration @xmath73^{\alpha/(1+\alpha)}\\ = -2w_{c0}(1+z_t)^{-3}-(1+w_{c0})(1+z_t)^{3\alpha}.\end{split}\ ] ]
in this section , we use the 157 gold sample supernova ia ( sn ia ) data compiled in @xcite to fit the model .
the parameters in the model are determined by minimizing @xmath74 ^ 2\over \sigma^2_i},\ ] ] where the extinction - corrected distance moduli @xmath75 , the redshift @xmath76 , the luminosity distance is @xmath77\\ = \frac{1+z}{h_0\sqrt{|\omega_k| } } { \rm sinn}\left[\sqrt{|\omega_k|}\int_0^z \frac{dz'}{e(z')}\right],\end{split}\ ] ] the dimensionless hubble parameter @xmath78 , @xmath79 is defined as @xmath80\{@xmath81 , @xmath82 } if @xmath83\{@xmath84 , @xmath8 } respectively , and @xmath85 is the total uncertainty in the observation .
the nuisance parameter @xmath86 is marginalized over with a flat prior assumption .
since @xmath86 appears linearly in the form of @xmath87 in @xmath88 , so the marginalization by integrating @xmath89 over all possible values of @xmath86 is equivalent to finding the value of @xmath86 which minimizes @xmath88 if we also include the suitable integration constant .
because we assume a flat prior on @xmath86 , therefore alternatively we marginalize @xmath86 by minimizing @xmath90 $ ] over @xmath91 , where @xmath92 . to get the marginalized likelihood of a parameter , we marginalize all other parameters by integrating the probability distribution @xmath89 over all possible values of the other parameters .
we first consider the special case @xmath25 , i.e. , the lcdm ( lambda cold dark matter ) model .
the best fit to the riess gold data is @xmath93 with @xmath94 .
the akaike information criterion ( aic ) is @xmath95 and the bayesian information criterion ( bic ) is @xmath96 @xcite .
next we consider the spatially flat case @xmath97 , the best global fit gives that @xmath98 , @xmath99 and @xmath100 with @xmath101 .
due to the constraints ( [ cond1]-[cond4 ] ) , it is difficult to find contours for the parameters .
the aic is @xmath102 and the bic is @xmath103 . for @xmath104 and @xmath67 , we get the local best fit : @xmath105 , @xmath106 and @xmath107 with @xmath108 .
the aic is @xmath109 and the bic is @xmath110 .
the probability distributions of @xmath111 , @xmath22 and @xmath9 are shown in figs .
[ wprob]-[aprob ] .
it is obvious that the distributions have bimodal characteristics .
therefore the models @xmath112 and @xmath113 fit the supernova data almost equally well .
the best fit of the full model gives that @xmath114 , @xmath115 , @xmath116 and @xmath117 with @xmath118 .
apparently , the full gcg model with curvature term fits worse than the spatially flat gcg model and the full model tends to be curved lcdm model .
if we think that the dust like matter source consists baryons only , then we can add a prior @xmath119 @xcite on the gcg model . in this case , for a spatially flat universe , the best fit parameters are @xmath1 , @xmath120 and @xmath121 with @xmath122 . substituting the best fit parameters into eq .
( [ eqzt ] ) , we find the transition redshift is @xmath4 .
the above results are summarized in table [ tab1 ] .
therefore , the current sn ia data does not favor gcg model over lcdm model .
furthermore , more parameters fail to give better fit .
.the comparison between different models .
gcg1 refers to the globally best fit spatially flat gcg model , gcg2 refers to the locally best fit spatially flat gcg model and gcg3 refers to the spatially flat gcg model with the assumption that the dust like matter source is only baryons .
[ cols="<,^,^,^",options="header " , ] for the spatially flat lcdm model , the best fit result is @xmath123 with @xmath124 . due to the sparse of the data ,
the result is not as good as that from sn ia . for the spatially flat gcg model ,
the best fit results are : @xmath125 , @xmath126 and @xmath127 .
again , the data does not favor gcg model over lcdm model .
in this paper , we studied the gcg model which provides a phenomenological mechanism of unifying dark matter and dark energy .
we explored a larger parameter space for the gcg model . instead of studying the usual parameter range @xmath128 , we extended the parameter space to be @xmath0 with some physical constraints on the parameters .
we found that the parameters had bimodal distributions in general .
the constraints from type ia sn data and the age data of clusters do not favor the gcg model over the simplest lcdm model from the standards of the akaike information criterion and the bayesian information criterion .
therefore , the current observations are consistent with both the gcg model and the lcdm model .
moreover , the flat model fits the observational data better .
the only benefit of the gcg model is that it may provide a phenomenological mechanism of unifying dark matter and dark energy .
the author yg thanks the hospitality of the abdus salam international center of theoretical physics where part of the work was done .
the author yg is also thankful to the icts of the university of science and technology of china for organizing the string / m theory , particle physics and cosmology workshop where this work was presented and discussed .
the work is fully supported by cqupt under grants a2003 - 54 and a2004 - 05 .
sp99 s. perlmutter _ et al _ , astrophy .
j. 517 ( 1999 ) 565 ; p.m. garnavich _ et al _ , astrophys .
j. 493 ( 1998 ) l53 ; a.g . riess _
et al _ , astron .
j. 116 ( 1998 ) 1009 .
et al _ , astrophys . j. suppl .
148 ( 2003 ) 175 .
riess , astrophys .
j. 607 ( 2004 ) 665 .
c. wetterich , nucl .
b 302 ( 1988 ) 668 ; b. ratra and p. j. e. peebles , phys . rev .
d 37 ( 1988 ) 3406 ; rev
phys . 75 ( 2003 ) 559 ; r. r. caldwell , r. dave and p.j .
steinhardt , phys .
80 ( 1998 ) 1582 ; i. zlatev , l. wang and p. j. steinhardt , phys .
82 ( 1999 ) 896 . c. armendariz - picon , t. damour and v. mukhanov , phys
b 458 ( 1999 ) 209 ; c. armendariz - picon , v. mukhanov and p.j .
steinhardt , phys .
85 ( 2000 ) 4438 ; phys .
d 63 ( 2001 ) 103510 .
a. sen , jhep 0204 ( 2002 ) 048 ; g.w .
gibbons , phys .
b 537 ( 2002 ) 1 ; t. padmanabhan and t.r .
choudhury , phys .
d 66 , 081301 ( 2002 ) ; j.s .
bagla , h.k .
jassal and t. padmanabhan , _ ibid .
_ 67 , 063504 ( 2003 ) ; t. padmanabhan and t.r .
choudhury , mon . not .
344 , 823 ( 2003 ) ; p.s .
piao , r.g .
cai and y.z .
zhang , phys .
d 66 ( 2002 ) 121301 .
a. cohen , d. kaplan and a. nelson , phys .
82 ( 1999 ) 4971 ; p. hoava and d. minic , phys .
85 ( 2000 ) 1610 ; s. thomas , phys .
( 2002 ) 081301 ; s.d.h .
hsu , phys .
b 594 ( 2004 ) 13 ; m. li , hep - th/0403127 ; k. ke and m. li , hep - th/0407056 ; b. wang , e. abdalla and r.k .
su , hep - th/0404057 ; r.g .
cai , jcap 0402 ( 2004 ) 007 ; q - g .
huang and y. gong , jcap 0408 ( 2004 ) 006 ; y. gong , phys .
d 70 ( 2004 ) 064029 .
dvali , g. gabadadze and m. porrati , phys .
b 485 , 208 ( 2000 ) ; c. deffayet , g.r . dvali and g. gabadadze , phys .
d 65 , 044023 ( 2002 ) ; z.h .
zhu , m. fujimoto , astrophys .
j. 581 , 1 ( 2002 ) ; 585 , 52 ( 2003 ) ; 602 , 12 ( 2004 ) ; z.h .
zhu , m. fujimoto and x.t .
he , astrophys .
j. 603 , 365 ( 2004 ) ; z.h .
zhu and j.s .
alcaniz , astro - ph/0404201 ; b. wang , y. gong and r.k .
su , hep - th/0408032 ; y. gong , x.m . chen and c.k .
duan , mod .
lett . a 19 , 1933 ( 2004 ) .
a. kamenshchik , u. moschella and v. pasquier , phys .
b 511 , 265 ( 2001 ) .
m. bento , o. bertolami and a.a .
sen , phys .
d 66 , 043507 ( 2002 ) .
sandvik , m. tegmark , m. zaldarriaga and i. waga , phys . rev .
d 69 , 123524 ( 2004 ) .
bea , p.p .
avelino , j.p.m . de carvalho and c.j.a.p .
martins , phys . rev .
d 67 , 101301 ( 2003 ) .
fabris , s.v.b .
gonsalves and p.e .
de souza , grg 34 , 53 ( 2002 ) ; n. bilic , g.b .
tupper and r. violler , phys .
b 535 , 17 ( 2002 ) ; v. gorini , a. kamenshchik and u. moschella , phys . rev .
d 67 , 063509 ( 2003 ) ; u. alam , v. sahni , t.d .
saini and a.a .
starobinsky , mon . not .
344 , 1057 ( 2003 ) ; v. gorini , a. kamenshchik , u. moschella and v. pasquier , phys . rev .
d 69 , 123512 ( 2004 ) ; gr - qc/0403062
. t. multamki , m. manera and e. gaztaaga , phys . rev .
d 69 , 023004 ( 2004 ) ; o. bertolami , a.a .
sen , s. sen and p.t .
silva , astro - ph/0402387 ; j.g .
hao and x.z .
li , astro - ph/0404154 ; s. nesseris and l. perivolaropoulos , phys .
d 70 , 043531 ( 2004 ) ; s.k .
srivastava , astro - ph/0407048 ; p.p .
_ et al_. , astro - ph/0208528 ; d. carturan and f. finelli , phys . rev .
d 68 , 103501 ( 2003 ) ; a.k.d .
evans , i.k .
wehus , .
grn and .
elgary , astro - ph/0406407 ; l.p .
chimento , phys . rev .
d 69 , 123517 ( 2004 ) ; z.h .
zhu , astron .
& astrophys .
423 , 421 ( 2004 ) .
y. gong and c.k .
duan , class .
quantum grav .
21 , 3655 ( 2004 ) ; mon . not .
352 , 847 ( 2004 ) ; t. barreiro and a.a . sen , astro - ph/0408185 .
bento , o. bertolami and a.a .
sen , phys .
b 575 , 172 ( 2003 ) . t. chiba and t. nakamura , prog .
100 , 1077 ( 1998 ) .
v. sahni , t.d .
saini , a.a .
starobinsky and u. alam , jetp lett .
77 , 202 ( 2003 ) .
h. akaike , ieee trans .
control , 19 , 716 ( 1974 ) ; g. schwarz , annals of statistics , 5 , 461 ( 1978 ) ; p.s .
corasaniti _ et al_. , astro - ph/0406608 .
bennett , astrophys .
148 , 1 ( 2003 ) .
s. capozziello , v.f .
cardone , m. funaro and s. andreon , astro - ph/0410268 .
r. rebolo _ et al .
_ , astro - ph/0402466 .
_ , astrophys . j. , 553 , 47 ( 2001 ) . | the generalized chaplygin gas model with parameter space @xmath0 is studied in this paper .
some reasonable physical constraints are added to justify the use of the larger parameter space .
the type ia supernova data and age data of some clusters are then used to fit the model .
we find that the parameters have bimodal distributions .
for the generalized chaplygin gas model , we also find that less free parameters fit the data better .
the best fit model is the spatially flat model with baryons .
the best fit parameters are : @xmath1 , @xmath2 and @xmath3 .
the transition redshift is @xmath4 . |
it is now widely accepted that black holes reside in the centre of nearly all galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
accretion onto these black holes is believed to be the mechanism responsible for powering active galactic nuclei .
the black holes in these galaxies can accrete matter , either from the interstellar medium close to the event horizon or from thick disks of gas . in order
to be accreted , gas needs to fall in with an angular momentum that does not exceed that needed for disk capture . in the vicinity of the sphere of influence of the black hole ,
the keplerian motion of the accreting material is exceedingly high , thus processes that redistribute the angular momentum of the accreting gas determine the accretion rate .
galactic mergers , gravitational torques from bars and other structures can efficiently channel gas into the nuclear regions ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
closer to the sphere of influence of the black hole , however , other mechanisms , such as turbulence driven by magnetic instabilities , are efficient sources of angular momentum transport ( e.g. * ? ? ?
* ; * ? ? ?
there are different modes of accretion .
the ` standard ' accretion mode occurs when matter is accreted onto the black hole through a radiatively efficient process .
this is generally associated with optically thick and geometrically thin accretion disks , and thought to be associated with quasar activity .
this process is likely to be prevalent where there is a plentiful supply of cold gas , which can accrete efficiently towards the black hole , and as such the fuel , in the form of the cold gas , should also be able to condense and provide the fuel for star formation ( e.g. * ? ? ?
indirect evidence for such a scenario comes from the similar evolution of the agn activity in the universe and the star - formation rate density , both increasing rapidly from @xmath5 and steadily turning over at @xmath6 ( e.g. * ? ? ?
furthermore , such a close connection between agn and star - formation activity may provide a relatively straightforward explanation of the local correlation between galaxy mass and black - hole mass @xcite .
accretion in an agn may also be radiatively inefficient .
when mass accretion is too low ( @xmath7 ) , the inflowing gas may not be able to radiate the gravitational potential energy liberated due to accretion .
magnetohydrodynamic simulations show that the energy dissipated during accretion could heat up only the ions in the disk ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
on the other hand , radiative losses , synchrotron radiation , and comptonization of low - energy photons take energy from the electrons , cooling them .
the characteristic time - scale for the hot ions and the cold electrons to achieve thermal equilibrium through coupling depends on the gas temperature in the vicinity of the black hole , and is inversely proportional to the number density of ions .
when the accretion rate is low , the ionic density is also small , and this time - scale is of the same order as , or larger than , the inflow time .
thus a two - temperature plasma is expected to form when @xmath7 . in this regime ,
advection dominated accretion flows ( adafs ) can be formed .
this radiatively inefficient process is currently taken as the most likely explanation for accretion flows around black holes at low accretion rates
. the critical accretion rate below which it becomes radiatively inefficient is estimated to be @xmath8 @xcite , where @xmath9 is a dimensionless parameter that measures the efficiency of angular momentum transport in disks . with the standard value of @xmath10 ( e.g. * ? ? ?
* ) , the critical accretion rate is situated around @xmath11 normalised to the eddington accretion rate . due to the inability of the gas to cool down radiatively in this regime
, the gas can be driven to a higher temperature , and thus cause a vertical thickening of the disk .
adafs are thus thought to be characteristic of hot thick disks .
what causes the different modes of radiatively efficient and inefficient accretion remains uncertain .
it has been suggested that the nature of the gas being accreted might determine the type of accretion , with cold gas producing a stable accretion disk and thus a radiatively efficient accretion , and hot gas producing flows that would result in an adaf ( e.g. * ? ? ?
the black hole spin has also been suggested to play a role in determining the accretion mode ( e.g. * ? ? ?
radio galaxies are a class of agn that can be divided into two sub - categories according to the ratio between the intensity of high and low excitation emission lines in their optical spectra : low excitation galaxies ( legs ) and high excitation galaxies ( hegs ) .
the two distinct classes were first noted by @xcite and further categorised by @xcite and @xcite .
@xcite defined legs as objects that obey the following requirements : have [ o iii]-line equivalent widths @xmath12 , have a ratio @xmath13/[o{\sc iii}]>1 $ ] , or both .
it is widely accepted that legs could be the result of radiatively inefficient accretion processes .
@xcite compared a large number of local radio - loud agn and concluded that the heg and leg populations show different accretion rate distributions , consistent with the idea that radiative efficiency is determined purely on accretion rate .
this is a subject that impacts on the most relevant issues for models of agn and their roles in galaxy evolution . for instance , if the heg / leg distinction is indeed related to the accretion process , then the fact that their radio properties are similar requires that the power being channeled into the jets is independent of the accretion rate .
moreover , given that radiatively efficient accretion is related to the ` quasar mode ' form of feedback processes in agn , and radiatively inefficient accretion associated with ` radio mode ' agn feedback in semi - analytic models ( e.g. * ? ? ?
* ) , the heg and leg characterisation could be used to diagnose the different modes at play in each system .
legs would for instance be the perfect laboratories to investigate the ` radio mode ' form of feedback , currently taken as being the principal mechanism responsible for shutting down star formation in the most massive systems at @xmath14 . in this paper
we investigate the heg / leg division in radio - loud sources and their different accretion processes , extending the study of @xcite to high - redshift sources ( @xmath15 ) . moving to high redshift
is not only complementary in terms of looking back time but also in terms of luminosity as it provides us with the opportunity to study the rarer more luminous objects in each class .
section [ sec : data ] describes the data selection .
section [ sec : sed_method ] describes the sed fitting method used . in section [ sec : sed_fits ] and [ sec : ind ] the sed fits are shown and the physical values extracted from them are presented .
the results are discussed in section [ sec : discuss ] , and conclusions are presented in section [ sec : conc ] . throughout this paper
we adopt the following values for the cosmological parameters : @xmath16 , @xmath17 and @xmath18 .
we use _ spitzer space telescope _ observations and multiple data collected from the literature , ranging from mid - infrared to optical wavelengths , of the sample of radio galaxies presented by @xcite .
this sample consists of 27 radio sources selected from the spectroscopically complete 3crr @xcite , 6ce @xcite , 6c * @xcite , 7crs @xcite and toot @xcite surveys to have narrow lines and a redshift of @xmath19 .
the narrow - line selection was intended to ensure the exclusion of quasars from the sample ( though 3c343 has since been classified as a quasar and 3c22 is classified as a weak quasar ) .
+ the _ spitzer _ data consist of observations with mips @xmath20 and irac @xmath21 , @xmath22 , @xmath23 and @xmath24 bands . the mips and irac observations took place in august 2006 and august 2007 , as described by @xcite .
the photometric measurements for the irac data were performed using a 7 arcsec diameter aperture with aperture corrections of 1.112 , 1.113 , 1.125 and 1.218 for irac channels 1 , 2 , 3 and 4 respectively , unless there was a nearby source in which case we used a 3.5 arcsec diameter aperture with appropriate aperture corrections .
for mips we extracted the photometry in a 13 arcsec diameter aperture with aperture correction of 1.167 .
a summary of these data is shown in table[table : spitzer ] .
l c c c c c c c c c c object & @xmath25 & err@xmath26 & @xmath27 & err@xmath28 & @xmath29 & err@xmath30 & @xmath31 & err@xmath32 & @xmath33 & err@xmath34 + & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 & @xmath35 + 3c280 & 277.26 & 38.69 & 524.69 & 87.58 & 1004.40 & 126.74 & 2200.45 & 204.88 & 9229.46 & 258.67 + 3c268.1 & 35.45 & 13.94 & 78.20 & 33.91 & 108.84 & 43.09 & 209.22 & 64.37 & 940.96 & 123.57 + 3c356 & 108.00 & 11.00 & 110.00 & 11.00 & 122.00 & 14.00 & 434.00 & 47.00 & 4060.00 & 192.00 + 3c184 & 129.88 & 44.40 & 145.48 & 46.09 & 242.00 & 55.00 & 288.00 & 76.00 & 742.00 & 184.00 + 3c175.1 & 119.60 & 25.54 & 109.55 & 40.21 & 92.03 & 42.56 & 124.70 & 53.55 & 836.90 & 166.56 + 3c22 & 1399.41 & 86.86 & 2431.72 & 188.43 & 3647.24 & 239.86 & 5852.97 & 333.61 & 13744.13 & 310.40 + 3c289 & 38.91 & 14.62 & 44.00 & 25.53 & 26.81 & 25.20 & 37.15 & 26.55 & 3650.16 & 168.21 + 3c343 & 151.35 & 28.62 & 142.75 & 45.74 & 211.85 & 60.29 & 683.25 & 114.58 & 7294.40 & 230.19 + 6ce1256 + 3648 & 62.50 & 18.44 & 61.01 & 29.99 & 55.72 & 32.24 & 133.86 & 52.70 & 1351.94 & 124.20 + 6ce1217 + 3645 & 118.37 & 25.34 & 120.80 & 42.11 & 125.26 & 51.99 & 150.98 & 53.52 & 313.64 & 104.94 + 6ce1017 + 3712 & 26.52 & 12.15 & 31.88 & 21.83 & 43.08 & 30.52 & 129.35 & 53.11 & 1136.21 & 131.80 + 6ce0943 + 3958 & 72.68 & 19.90 & 87.66 & 35.95 & 141.66 & 49.31 & 244.95 & 70.77 & 1996.79 & 153.62 + 6ce1257 + 3633 & 81.58 & 21.05 & 70.04 & 32.11 & 36.46 & 27.63 & 110.35 & 48.18 & 850.67 & 108.30 + 6ce1019 + 3924 & 159.69 & 29.41 & 117.47 & 41.57 & 75.93 & 38.14 & 23.26 & 28.80 & 334.56 & 181.18 + 6ce1011 + 3632 & 74.20 & 20.11 & 79.07 & 34.14 & 89.07 & 40.89 & 200.03 & 64.69 & 1326.94 & 150.39 + 6ce1129 + 3710 & 81.67 & 21.08 & 62.57 & 30.39 & 40.67 & 35.62 & 72.84 & 41.39 & 856.71 & 119.07 + 6c*0128 + 394 & 85.50 & 21.58 & 92.74 & 36.95 & 65.62 & 36.02 & 74.97 & 42.26 & 134.85 & 86.14 + 6ce1212 + 3805 & 74.06 & 20.10 & 61.81 & 30.24 & 30.77 & 31.14 & 55.14 & 37.46 & 257.42 & 92.98 + 6c*0133 + 486 & 53.85 & 17.20 & 64.35 & 30.86 & 50.99 & 41.43 & 75.81 & 37.92 & 82.15 & 75.92 + 5c6.24 & 111.17 & 24.57 & 102.82 & 38.89 & 86.34 & 41.21 & 118.65 & 50.69 & 728.22 & 119.30 + 5c7.23 & 54.77 & 17.32 & 48.11 & 26.75 & 67.04 & 35.75 & 41.12 & 34.62 & 623.52 & 160.69 + 5c7.82 & 114.37 & 24.92 & 83.15 & 34.99 & 45.29 & 31.63 & 55.92 & 37.71 & 422.33 & 154.47 + 5c7.242 & 123.35 & 25.87 & 104.96 & 39.33 & 69.57 & 36.58 & 41.92 & 34.64 & 1994.72 &
161.08 + 5c7.17 & 30.88 & 13.10 & 22.64 & 18.50 & 56.68 & 33.88 & 5.14 & 22.51 & 1371.30 & 176.80 + toot1267 & 114.48 & 24.91 & 95.55 & 37.47 & 88.35 & 39.65 & 141.41 & 53.90 & 1079.06 & 121.87 + toot1140 & 50.48 & 16.63 & 64.88 & 30.92 & 47.60 & 31.18 & 28.71 & 28.25 & 136.24 & 80.29 + toot1066 & 49.37 & 16.50 & 35.92 & 23.09 & -3.85 & 14.10 & 16.21 & 22.64 & 298.16 & 86.86 + [ table : spitzer ] to increase the number of data points in the near - infrared and optical bands , we used as many photometric values as possible from the literature .
the values available were extracted with a range of different apertures and , where possible , we chose apertures of @xmath36 , or larger , which approximately encompass the whole of the emission from the galaxies in our sample . for consistency , when magnitude values were only available with smaller apertures we applied a correction to transform them into magnitudes with a @xmath37 aperture or larger . given that the majority of the photometric data in the literature is presented in apertures of 8 or 9 arcsec diameter , and that this is a good approximation to the total flux of the source , we preferred to convert all the smaller aperture magnitudes to their equivalent value at 8 or 9 arcsec .
all the magnitudes measured using smaller apertures that we converted to 8 or 9 arcsec were measured either in 4 or 5 arcsec apertures , thus the corrections are small ( e.g. * ? ? ? * ) . to find relations between 4 and 5 arcsec and 8 or 9 arcsec
, we gathered a large sample of galaxies ( @xmath38 galaxies ) , from the papers referenced in tables 1 to 4 of the appendix , which had photometry available for several different apertures . with this sample
, we computed linear relations between the magnitudes at 4 arcsec and 9 arcsec and between 5 arcsec and 8 arcsec .
we then used these linear fits to determine all magnitudes in 8 or 9 arcsec apertures .
we also applied a galactic extinction correction to all the magnitude values that had not been corrected based on the maps of @xcite .
following @xcite we also subtracted the flux of the dominant emission lines @xmath39 , [ o ] , [ ne ] , mg , [ ne ] and c ] , from j , r , f702w , f606w , v , b , and g bands , as these lines can contribute a significant percentage to the total flux of the band where they lie . where the line fluxes were not available we used the [ oii ] fluxes from @xcite and determined the remaining emission lines using the line ratios given by and @xcite .
when the flux of h@xmath9 is not known , we used the average ratio estimated by for high redshift radio galaxies ( @xmath40 ) from the 3cr survey . to correct for emission line contamination
we then determined the location of each emission line for the redshift of the given source and account for the shape of the filter at that wavelength .
given the uncertainties affecting the aperture , emission - line contamination and galactic extinction corrections , we assume an additional @xmath41 per cent error in the flux density for each magnitude value that had any of these corrections applied .
all these corrections are detailed in tables [ table : jhk ] , [ table : hst ] [ table : ubvri ] , and [ table : ugriz ] of the appendix .
the photometric data span a wavelength range from @xmath42 to @xmath43 . at optical wavelengths ,
the emission from radio galaxies - obscured agn - is dominated by the stellar emission of the host galaxy .
at mid - infrared wavelengths , dust in the ` torus ' region that re - emits the radiation from the central agn is the main source of emission , at least for those sources with a relatively powerful and radiatively efficient agn .
this can be thought of as a composite of the central emission passing through a screen of dust .
we follow the approach of @xcite and model the radio galaxies with a galaxy template that dominates the emission in the optical part of the spectrum and a quasar template , extinguished by a dust extinction law , to fit the emission in the infrared spectral region . to replicate the host galaxy emission , we consider a bruzual and charlot stellar synthesis model ( * ? ? ?
* hereafter bc03 ) , as well as the maraston et al . `
fuel - consumption ' stellar population synthesis model ( * ? ? ?
* hereafter m05 ) . for both of these
we assume a salpeter initial mass function and solar metallicity . since radio galaxies are almost exclusively hosted by elliptical galaxies or recent merger remnants ( e.g. * ?
* ; * ? ? ?
* ) , we expect them to have a strong short episode of star formation in the beginning of their formation and for the star formation rate to quickly drop subsequently .
therefore , we consider that a single simple stellar population suffices to reproduce the host galaxy s star formation history , and a synthesis of ssps is not required .
to select the stellar age of our template we note that our objects all have @xmath15 , when , according to the adopted cosmology , the universe was @xmath44 old .
we , thus , constrain the range of possible ages with an upper limit for the stellar population age of @xmath45 . as for a lower limit
, previous studies have shown that for passively - evolving early - type galaxies in cluster environments , the bulk of stars form at @xmath46 and in low - density environments at @xmath47 .
this yields a stellar age of @xmath48 for cluster environments and @xmath49 for field galaxies . even though our sample consists of powerful radio galaxies , which tend to inhabit high - density environments as suggested by various lines of observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , including x - ray observations that show x - ray cavities identified with clusters of galaxies to be spatially coincident with non - thermal radio emission from radio - loud agn ( e.g. @xcite , @xcite ) , we adopt a conservative lower limit for the stellar age of @xmath50 .
we thus use templates with @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 , and @xmath45 of age for the sed fitting of the galaxies in our sample . to model the infrared emission of the radio galaxies , we need to account for the dusty torus absorbing the radiation from the central source and re - emitting it at longer wavelengths .
this can be done , as an approximation , by assuming that a radio galaxy is equivalent to a quasar with a layer of obscuring dust in front of it , where this layer can have diverse column densities depending on how obscured the galaxy is ( e.g. * ? ? ?
* ; * ? ? ?
indeed , @xcite find that they can reproduce a diversity of rest - frame @xmath57 seds of radio galaxies by combining various amounts of extinction of agn light with host galaxy starlight .
this representation is consistent with the orientation - dependent unified scheme and it is intended as a simplification only , as a more careful approximation should fully account for radiative transfer effects in a dusty medium ( e.g * ? ? ?
this approach would , however , imply a large number of free parameters with too few data points to constrain them , resulting in a high degeneracy , making it difficult to infer physically meaningful quantities .
thus , we build , as a simplification , a composite model of a quasar and a screen of dust . as a quasar template ,
we use the radio - quiet quasar sed by @xcite , which was constructed based on the mean energy distribution of a sample of 29 radio - quiet quasars .
we normalise this template by forcing the bolometric luminosity to be 1 . for this
, we used the median value of the b - band luminosity , @xmath58 , which , according to table 17 of @xcite , should be * @xmath59 . *
more recent works such as @xcite have shown a composite model of a jet , a truncated thin accretion disk and an adaf to be a good fit for low - luminosity agns , and therefore possibly better suited for legs .
@xcite however report an insufficient ir - emission produced by the truncated thin disk in order to match the observations .
given these issues , we opt to use the model by @xcite for all the objects in the sample for consistency . for a dust template
, we used the extinction laws derived by @xcite , which replicate how the extinction caused by the different types of dust varies with wavelength .
these templates are described by the following summation : @xmath60 where @xmath61 , @xmath62 and @xmath63 vary for each term and for each dust type ( see table 4 of @xcite for a full description ) .
the six terms involved represent the background , far - ultraviolet and far - infrared extinctions and the @xmath64 , @xmath65 m and @xmath66 m features . for reddened quasars ,
smc dust has been found to be appropriate ( e.g. * ? ? ?
* ; * ? ? ?
* ) , since dust in the host galaxy of high - redshift galaxies tends to have lower metallicity and thus can be better approximated by smc dust type .
however , for more obscured quasars , and in particular for the majority of the objects in our sample , we find that smc dust does not provide as good a fit as mw dust .
this could be explained by the fact that the dust intersecting the line of sight of the central emission in obscured galaxies comes from the inner region due to the galaxy s edge - on orientation , and the central regions of galaxies are usually more metal rich , as there is in general a metallicity gradient in galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , we adopted a mw dust type for all the galaxies in our sample .
we fit our model to the photometric data points by applying an extinction curve to the intrinsic quasar light template and then adding it to the galaxy light model : @xmath67 with @xmath68 where @xmath69 is the quasar light template and @xmath70 is the quasar light model already affected by extinction . to convert the extinction law @xmath71 to magnitudes ( @xmath72 ) we multiply equation by the term @xmath73 , where @xmath74 is the ratio of total - to - selective extinction defined by the equation : @xmath75 where @xmath76 is the colour excess .
we multiply each template by a coefficient which we allow to vary along a range of physically motivated values .
the extinction law is multiplied by the visual extinction coefficient @xmath77 spanning values of @xmath78 .
the resolution we use to run through these values varies from galaxy to galaxy depending on how precise the fit is , and we make it finer where the dispersion of the fit is smaller , as detailed below . after multiplying the extinction law in flux units by the transmitted quasar light , we force it to be as close as possible to the @xmath79 flux density value .
this means we do not allow the bolometric luminosity of the model to vary freely . for each given @xmath80 ,
it is fixed by the observed flux density at @xmath81 .
the reasoning for this is to constrain the fit , as we found that without this step , the higher degeneracy of the fits often provided poorer matches to the data in terms of reduced-@xmath82 . the galaxy template , @xmath83 ,
is multiplied by a mass normalisation factor , @xmath84 , which we allow to vary between @xmath85 and @xmath86 , an interval that comprises the typical values of stellar mass content in early - type galaxies and certainly the masses of the majority of radio galaxies studied to date ( e.g. * ? ? ?
* ) : @xmath87 we then perform a grid search in @xmath77 and @xmath88 to obtain the best fit .
given that the data from the literature were gathered from several different instruments , we opted to use a single set of filter transmission curves : for @xmath89 , @xmath90 and @xmath91 bands we used the filters from the visible and infrared survey telescope for astronomy ( vista ; see e.g. * ? ? ?
* ) for the hst f606w , f702w and f814w passbands , we used the respective hst filter response profiles ; for @xmath92 , @xmath93 , @xmath94 , @xmath95 and @xmath96 bands we used the filters used on the auxiliary - port camera ( acam ) mounted on the william herschel telescope ( wht ) ; finally , for the @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 bands , given that most data at these wavelengths were extracted from the sdss public release , we used the sdss bandpass filters . for the small number of photometric data points that were not observed with the exact same filters we use , the errors are smaller or of the same order of magnitude as the uncertainties on the data . using the filter responses
, we evaluate the flux density that the model produces for each band using the following : @xmath102 where @xmath103 is the flux density of the model before it has been convolved with the filter response ; @xmath104 is the flux density of the model after the filter response has been taken into account ; @xmath105 and @xmath106 are the wavelength where the filter response starts and ends respectively ; and @xmath107 is the filter transmission curve . to determine the flux @xmath108 of the data points , we multiply the flux density values by the mean frequency of each band .
the mean frequency is determined by computing the effective wavelength of each band , given by : @xmath109 and converting it to @xmath110 . + we determine the @xmath82 distribution of each model using , @xmath111 where @xmath112 is the number of flux / magnitude data points available , and @xmath113 is the error associated with each data point .
we repeat this process for all the different stellar models we are considering ( ages 0.5 , 1 , 2 , 3 , 4 , 5 , and 6gyr for both bc03 and m05 models ) and choose the model that has the lowest @xmath82 amongst these .
+ sources that are detected below a 2@xmath113 level in the imaging data are plotted as upper limits at a 2@xmath113 level .
we use their measured photometric flux for the sed fitting and place their error bars between zero and the detection limit for the @xmath82 evaluation .
however , photometric bands for which both the observed flux density and the model are lower than the flux density limit make no contribution to the @xmath82 . | understanding the evolution of accretion activity is fundamental to our understanding of how galaxies form and evolve over the history of the universe .
we analyse a complete sample of 27 radio galaxies which includes both high - excitation ( hegs ) and low excitation galaxies ( legs ) , spanning a narrow redshift range of @xmath0 and covering a factor of @xmath1 in radio luminosity . using data from the _ spitzer space telescope _
combined with ground - based optical and near - infrared imaging , we show that the host galaxies have masses in the range of @xmath2 with hegs and legs exhibiting no difference in their mass distributions .
we also find that hegs accrete at significantly higher rates than legs , with the heg / leg division lying at an eddington ratio of @xmath3 , which is in excellent agreement with theoretical predictions of where the accretion rate becomes radiatively inefficient , thus supporting the idea of hegs and legs being powered by different modes of accretion .
our study also shows that at least up to @xmath4 , hegs and legs are indistinguishable in terms of their radio properties . from this result
we infer that , at least for the lower radio luminosity range , another factor besides accretion rate must play an important role in the process of triggering jet activity .
[ firstpage ] galaxies : active - galaxies : jets - infrared : galaxies - radio continuum : galaxies - quasars : general - galaxies : nuclei |
extensive lattice simulations have shown that qcd exhibits a phase transition at a temperature of the order of the infrared scale @xmath0 .
this transition is believed to be a confinement - deconfinement transition for heavy quarks , and a chiral transition for light quarks .
the high temperature phase is characterized by a small coupling constant @xmath1 .
unfortunately , perturbation theory in @xmath1 is plagued with infrared singularities since the magnetic sector is non - perturbative at all temperatures .
high temperature lattice @xcite and instanton simulations @xcite seem to concur on the fact that in the vicinity of the critical temperature , @xmath2 , the chromoelectric and chromomagnetic condensates are still substantial .
a number of theoretical arguments , both perturbative @xcite and nonperturbative @xcite , suggests that in the euclidean formulation , the high temperature phase of qcd exhibits strong @xmath3 correlations .
it has even been argued that the global color symmetry might be spontaneously broken with the appearance of an @xmath4 condensate @xcite .
also , lattice simulations show that chiral symmetry is restored at the critical temperature , so that the quark condensate can be set to zero above @xmath2 . in the present analysis
we will incorporate these three important observations that are pertinent to qcd near and above @xmath2 ; the quark condensate vanishes , the chromoelectric and chromomagnetic correlations remain non - zero , and strong correlations develop in the static part of the @xmath4 gauge field that can be described by a local @xmath5 condensate . in the spirit of the qcd sum rule method , we will not address the complicated dynamical questions of how the gluon condensates come about , but simply assume that they exist .
we will also restrict our discussion to only those with dimension 2 , @xmath6 , and 4 , @xmath7 and @xmath8 , following an earlier suggestion by two of us @xcite .
the value of the condensates can only be estimated using non - perturbative calculations such as lattice qcd , and at high temperature , the scale is expected to be set by the mass gap in the magnetic sector which is @xmath9 . that chromoelectric and chromomagnetic condensates could be present in the high temperature phase
is not very surprising , and their size can be roughly estimated using the bulk energy and momentum measurements at finite temperature on the lattice .
the analysis in @xcite gives the values : @xmath10 the presence of an @xmath11 condensate is more controversial @xcite , so we will give a summary of the arguments presented in @xcite for why such a condensate should be considered .
the basic observation is that the wilson line ( or polyakov loop ) operator @xmath12 , is a gauge invariant operator that is a good order parameter for the pure glue theory .
although this operator is non - local , it is possible to relate it to a string of local operators by choosing the static gauge @xmath13 .
for example , in an su(2 ) theory we have @xmath14 , and one could now try to use factorization to extract @xmath15 . here
we will simply equate the condensate to the magnetic mass , we will typically use , @xmath16 , which is about @xmath17 for @xmath18 this estimate is also supported by a lattice calculation in pure yang - mills @xcite .
indeed , an analysis of the gluon - propagator in four - dimensions in landau - gauge gives @xmath19 on a @xmath20 lattice , which is equivalent to @xmath21 .
a direct reconstruction of this condensate from the dimensionally reduced theory has proven to be more subtle @xcite . in this letter , we present a calculation that suggests that the gluon condensates strongly affect the emission rates of low mass dileptons from the plasma
. our calculation will be very naive .
we will simply assume that the quark - antiquark annihilation processes that give rise to the dileptons occur in the presence of low momentum fluctuating background gluon fields .
by averaging over the background fields , we get leading corrections to the emission rates proportional to the various condensates .
note that in doing so we take the concept of a fluctuating background field at face value and perform the calculations directly in the time - like region assuming that the temperature is high enough for perturbation theory to be useful .
thus our method is very different from the qcd sum rules , where the background field method is just a convenient tool to calculate coefficient functions in the operator product expansion which is then used in the deep euclidean region . annihilation rate .
that calculation was inspired by the so called copenhagen vacuum based on fluctuating magnetic fluxtubes .
] however , since the emission rates can be expressed in terms of current - current correlation functions we can use the standard qcd sum rule _
techniques _ to calculate the condensate contributions . how this is done in detail ,
especially when it comes to gauge choice , is explained in @xcite . to calculate the contribution from the @xmath11 condensate we will use the following observation :
a constant potential @xmath22 is the same as an imaginary ( color ) chemical potential , @xmath23 , and thus one can obtain the term @xmath24 by using a grand canonical ensemble for the quarks with a chemical potential @xmath25 , _ averaged _ over the arbitrary directions , @xmath26 , in color space , and pick up the term @xmath27 .
( we have also checked that this procedure gives the same result as the standard fixed point gauge methods used in @xcite . ) at this point we want to stress that although our approach is inspired by the above theoretical arguments , it is basically phenomenological , and should primarily be judged by how successfully it can accommodate data .
the theoretical uncertainties are not limited to the assumptions we made about the non - perturbative corrections to correlations functions in a high temperature quark gluon plasma .
we also have to make several assumptions about the physics of the heavy ion collision , and we will use an oversimplified model , namely that of an expanding fireball containing _ only _ the plasma phase .
the main motivation for using such a simple model ( the details are given in section iii below ) is to demonstrate that the gluon condensate effects are important .
hopefully more realistic models including regions with mixed or hadronic phases , would also exhibit this sensitivity to the condensates . also in the transition regime
, the pure and mixed quark condensates may still be relevant .
having said all this , we will demonstrate that a naive fire ball model combined with a simple parametrization of the production amplitudes based on the above ideas , and with the condensate parameters at the values suggested by the scale arguments , is quite successful in explaining the dilepton data from the ceres and helios relativistic heavy ion experiments .
the rest of this paper , where this claim is substantiated , is organized as follows : in section [ sec : emirate ] , we calculate the effects of the leading order condensates on the dilepton emission rates from a thermalized quark - gluon phase . in section [ sec : results ] , we use a simple fire - ball model to estimate the dilepton yield , and compare to the ceres and helios experiments . in section
[ sec : ptcut ] , we show that our emission rates are also consistent with the recent analysis of the ceres results using different @xmath28 cuts .
in a translationally invariant , but non - perturbative , quark - gluon phase , the rate @xmath29 of dileptons produced in a unit four volume is directly related to the electromagnetic current - current correlation function @xcite . for a pair of leptons with mass @xmath30 and momenta @xmath31 , the rate per unit invariant momentum @xmath32 is , & = & - ( 1 + ) ( 1-)^1/2 ( q ) , where @xmath33 is the fine structure constant , and ( q ) = d^4 x e^-iqx ( * p*e^ia_4^a*q*^a e^-(-)/t ^(x ) _ ( 0 ) ) . [ eq:2 ] here
@xmath34 is the hadronic part of the electromagnetic current , @xmath35 the hadronic hamiltonian , @xmath36 the free energy , @xmath37 the temperature , and @xmath38 a global color generator .
( as mentioned above the direction , @xmath26 , is arbitrary but fixed . )
@xmath39 is related to the retarded current - current correlation function by , ( q ) = im^r ( q ) [ eq2a ] where ^r(q ) = id^4 x e^iqx ( * p*e^ia_4 ^ 3 * q*^3 e^-(-)/t * r*^(^(x ) _ ( 0 ) ) ) , [ eq2b ] and @xmath40 is the retarded product .
the leading contribution to the retarded function is condensate independent , and is just the `` born '' @xmath41 annihilation term , which for massless quarks becomes , ^r & = & ( n_c_q = u , d , s e_q^2 ) q^2 , [ eq : wr ] where @xmath42 is the number of colors , @xmath43 the charge of the quarks , and @xmath44 . at @xmath45 this is just the @xmath46 ratio well known from @xmath47 annihilation r_3(q^2 ) = _ i = - .
im^r_0 @xmath46 is @xmath48 .
as mentioned above , the contribution of the @xmath3 condensate to the retarded function @xmath49 can easily be calculated by picking up the term quadratic in the imaginary colored chemical potential @xmath50 .
the result is _ 11^r & = & ( n_c_q e_q^2 ) q^2 .
[ eq : wrnp ] according to the estimates in the previous section , @xmath51 is of the order of @xmath17 .
note that ( [ eq : wrnp ] ) has the same sign as ( [ eq : wr ] ) and will thus increase the dilepton yield . in the next section
, we will see that this effect can be rather dramatic in the invariant mass region studied in the ceres experiments .
the contribution of the chromoelectric and chromomagnetic condensates to the retarded function can be calculated using the methods in @xcite , and the result for @xmath52 is , _ 12^r & = & ( n_c_q e_q^2 ) .
[ eq : eb ] the temperature dependence of ( [ eq : eb ] ) is the same as that of ( [ eq : wrnp ] ) and can be understood from the condition of detailed balance .
note that the emission rate is enhanced by a chromomagnetic field and depleted by a chromoelectric field as expected pairs are repelled from each other in the cm frame in the presence of a chromoelectric field , and deflected in the same direction in the presence of a chromomagnetic field . ] , and also that all condensate contributions vanish at @xmath45 .
in fig . [ fig : plasma ] we show the dilepton emission rates at temperature @xmath53 mev , using the parameters , @xmath54 @xcite .
the contribution from the non - perturbative plasma , perturbative plus condensate contributions ( dotted and dash - dotted lines ) should be compared to the purely perturbative part ( thick solid line ) and , as a reference , to that of a gas of mesons ( thin solid line ) @xcite .
the condensate contribution to the rates are sizable in the 500 mev region , and the effect of the @xmath55 condensate , @xmath56 , dominates that of the chromoelectric and chromomagnetic ones , @xmath57 , for masses above 250 mev . the dependence on both condensates @xmath56 and @xmath57 is negligible at energies higher than 2 gev , and above 2.5 gev the perturbative plasma gives the same result as that of a hadron gas , as expected from duality . in the dalitz region ( @xmath58mev )
the effects of both the electric and magnetic condensates are appreciable and dominates the perturbative part by orders of magnitude . this observation maybe important for photon emission as well . in a heavy - ion collision
the electromagnetic emission occurs from various stages of the collision process , and hence at various temperatures . in this part
, we will focus on the emission rates in the low and intermediate mass region ( up to 1.5 gev ) .
we will assume that the dileptons emanate from a simple fire ball composed of the nonperturbative plasma phase only .
the fire ball expansion is assumed to be homogeneous , and we will use a standard parametrization of volume and temperature , v(t ) & = & v_0(1+)^3 t(t ) & = & ( t_i - t_)e^-t/+t _ .
[ fire ] the parameters in ( [ fire ] ) regarding the initial and freeze - out temperature will be taken from present transport calculations @xcite .
we will not try to justify this point except by saying that it allows us to define a space - time volume for the `` hot '' spot in the heavy - ion collision that is consistent with the one used for an interacting meson - phase @xcite .
hence @xmath59 fm / c , @xmath60 mev , @xmath61 mev , @xmath62 fm / c , and the value of @xmath63 is absorbed into the over - all normalization constant @xmath64 @xmath65 .
the freeze - out time will be set to @xmath66 fm / c . using ( [ fire ] ) and the above rate , the final expression for the integrated emission rate per unit rapidity @xmath67 and invariant mass @xmath68 is ( m ) & = & n_0 m _ 0^t_f.o .
dt v(t ) a(q_0,q^2 ) .
[ eq : rate ] the acceptance function @xmath69 enforces the detector cut @xmath70 , and @xmath71 mrad for ceres . in fig .
[ fig : ceres_plasma ] , we show the evolved emission rates for ceres .
the enhancement is sizable and goes in the direction of the data in the 500 mev range in ceres .
again , the nonperturbative qcd results are indicated by the dotted line for the condensates we quoted , the dashed - dotted line ( without the e- and b - condensates ) , while the perturbative qcd result is indicated by the solid line and the meson result by the thin solid line . for helios-3 , ( [ eq : rate ] )
can be used by integrating over @xmath67 with the cut @xmath72 . in fig .
[ fig : helios_plasma ] , we show the evolved emission rates for helios . for this energy range
the effect of chromoelectric and chromomagnetic condensates are small irrespective of the specific value for the thermal condensates as indicated in figs .
[ fig : plasma ] and [ fig : ceres_plasma ] .
the enhancement is compatible with the helios data , and slightly overshoots both the data and the meson gas in the 1 gev region .
another important check on the nonperturbative emission rates can be obtained by using the recent @xmath28 cut analysis by the ceres collaboration to probe the transverse momentum content of the dilepton emission rates in the fire - ball .
our results versus the data are shown in figs .
[ fig : plasma_pt_low]-[fig : plasma_pt_total ] , with and without the chromoelectric and chromomagnetic condensates . the @xmath28 dependent effects caused by the thermal condensates are most dramatic in the low @xmath28 region ( fig .
[ fig : plasma_pt_low ] ) , where there is a sizable condensate effect , mainly due to @xmath56 .
the condensate contributions are in the right direction around the low mass dielectron enhancement , and give important additions to the perturbative result . in the high @xmath28 region , fig .
[ fig : plasma_pt_high ] , the resulting condensate effects are rather small , and the curve is consistent with the present data .
therefore , the total transverse momentum content ( fig .
[ fig : plasma_pt_total ] ) is quite accurately reproduced by the addition of the gluon condensates .
in summary , we find it encouraging that not only the dilepton emission rates , but also the observed low mass - region @xmath28 enhancement is , at least partially , explained by our simple model , and we think that this further strengthens our proposal that nonperturbative condensate effects play a role in the plasma phase of heavy ion collisions .
we thank a number of our colleagues at stony - brook for discussions .
we also thank keijo kajantie and su houng lee for their comments on the manuscript . this work was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 88er40388 , and the swedish natural science research council .
m. campostrini and a. di giacomo , phys . lett .
* b197 * ( 1987 ) 403 ; s.h .
lee , phys . rev .
* d40 * ( 1989 ) 2484 . c. adami , t. hatsuda and i. zahed , phys
* d43 * ( 1991 ) 921 ; v. koch and g.e .
brown , nucl . phys .
* a560 * ( 1993 ) 345 ; t. schaefer and e. shuryak , rev .
phys . * 70 * ( 1998 ) 323 .
r. anishetty , j. phys . *
g10 * ( 1984 ) 423 ; k.j .
dahlem , z. phys . *
c29 * ( 1985 ) 553 ; v.m .
belayev and v.l .
eletsky , z. phys .
* c45 * ( 1990 ) 355 .
j. mandula and m. ogilvie , phys .
b201 * ( 1988 ) 117 .
s. nadkarni , phys . rev . *
d34 * ( 1986 ) 3904 .
hansson and i. zahed , _ qcd sum rules at high temperature " _
, suny-1990 report , unpublished ; i. zahed , _ light relativistic bound states in high temperature qcd " _ , in thermal field theories , eds .
h. ezawa et al . , elsevier 1991 , p. 357 ; i. zahed , _ dynamical excitations in high temperature qcd " _ , in from fundamental fields to nuclear phenomena , eds .
mcneil and price , world scientific 1991 , p. 238 .
k. kajantie , m. laine , k. rummukainen and m. shaposhnikov , hep - ph-9704416 ; and references therein .
hansson and i. zahed , nucl
* b 292 * ( 1987 ) 725 .
mclerran and t. toimela , phys .
d 31 * , 545 ( 1985 ) .
h. a. weldon , phys .
d 42 * , 2384 ( 1990 ) .
a. patkos and n. sakai , nucl . phys .
* b 168 * ( 1980 ) 521 .
lee , h. yamagishi , and i. zahed , hep - ph/9806391 , phys .
rev . * c * , in print .
ko , and g.e .
brown , nucl .
phys . * a 606 * , 568 ( 1996 ) .
r. rapp , g. chanfray , j. wambach , nucl .
phys . * a 617 * , 472 ( 1997 ) . | assuming that gluon condensates are important even above the deconfining phase transition , we develop a model for the dilepton yield from a quark gluon plasma . using a simple fire ball description of a heavy ion collision , and various estimates of the strengths of the gluon condensates , we compare our predicted dilepton yields with those observed in the ceres and helios experiments at cern .
the simple model gives an adequate description of the data , and in particular it explains the observed considerable enhancement of the yield in the low mass region .
h o s # 1 |
when studying the calculus of functions , the chain rule for the first derivative of a composition of functions is given by the familiar formula @xmath4 by repeatedly applying this formula , one can derive the classical fa di bruno formula for the @xmath0th derivative of the composite @xmath5 : @xmath6 where the sum is taken over all non - negative integer solutions to @xmath7 and @xmath8 .
an alternative higher order chain rule can be obtained by using the _ directional _ derivative of @xmath9 at @xmath10 along @xmath11 : @xmath12.\ ] ] huang , marcantognini , and young @xcite define _ higher order directional derivatives _
@xmath13 , functions of @xmath14-variables with @xmath15 , with which they obtain a concise formulation for the @xmath0th derivative of a composition @xmath16 when @xmath17 is a function of a single variable , or @xmath18 in the general case . for homotopy functors ( such as functors from spectra to spectra or spaces to spaces that preserve weak equivalences ) , goodwillie s calculus of functors builds a _ taylor tower _ of polynomial " functors and natural transformations : @xmath19 where
each @xmath20 is an @xmath0-excisive functor that approximates @xmath2 in a range of homotopy groups depending on the functor @xmath2 and the connectivity of the input @xcite .
the homotopy fiber of the map @xmath21 is the @xmath0th _ layer _ of the tower , @xmath22 . for homotopy functors from spectra to spectra ,
the layers are of the form @xmath23 , where @xmath24 is a spectrum with an action of the the @xmath0th symmetric group , @xmath25 .
this spectrum is called the @xmath0th derivative of @xmath2 .
several versions of chain rules for the derivatives of functor calculus have been developed .
notably , arone and ching @xcite derived a chain rule for the derivatives @xmath24 using the fact that for functors of spaces or spectra , the symmetric sequence @xmath26 is a module over the operad formed by the derivatives of the identity functor of spaces .
this generalized earlier work of klein and rognes @xcite that established a chain rule for first derivatives . for functors of spectra ,
a chain rule for the derivatives is given by ching @xcite .
for functors of spaces , yeakel has developed an alternative method for deriving these chain rules that does not require passage to spectra @xcite .
the chain rule results of the preceding paragraph use goodwillie s original formulation of the taylor tower . the second author and mccarthy @xcite defined taylor towers for not necessarily additive functors between abelian categories using the classical cross effect functors of eilenberg and mac lane @xcite . this approach has been generalized and applies to a wide variety of contexts @xcite , @xcite . in @xcite , johnson and mccarthy defined a notion of directional derivative @xmath27 for a functor @xmath2 valued in an abelian category and proved the analog of huang , marcantognini , and young s chain rule in degree one ( * ? ? ?
* proposition 5.6 ) : @xmath28 under the hypothesis that the functor @xmath3 preserves the zero object .
the present work was motivated by the goal of proving an analog of the higher order directional derivative chain rule of huang , marcantognini , and young for the abelian functor calculus of johnson and mccarthy . achieving this goal required first dealing with the question of how to define a higher order directional derivative in this context . to get a sense of the choices involved , consider the functor @xmath27 .
this is a functor of two variables , so the first choice to be made in defining a second order directional derivative is whether one should differentiate with respect to one of the variables ( a partial derivative ) or both variables simultaneously ( a total derivative ) .
for the first possibility , one notes that @xmath27 is already a linear functor with respect to its first variable , so that differentiating with respect to that variable yields nothing new .
so , for the partial derivative approach , we can restrict our attention to taking the directional derivative of @xmath27 with respect to @xmath29 to obtain a second order partial directional derivative @xmath30 that is a functor of @xmath31 variables . iterating this
process yields functors @xmath32 of @xmath14-variables defined in @xcite , which we refer to as _ iterated partial directional derivatives_. if we follow the total derivative approach by treating @xmath33 as a functor whose domain is a product category @xmath34 , then we obtain as a second order total directional derivative a functor @xmath35 whose source category is @xmath36 .
but , this leads to some redundancy , as we end up differentiating in some directions repeatedly .
the correct " approach , at least if one wants to define the analog of the huang - marcantognini - young higher order directional derivatives , lies somewhere in between .
to define our second order directional derivative @xmath37 of @xmath2 , we restrict @xmath38 along a diagonal functor @xmath39 that takes the triple of objects @xmath40 to the ordered pair of ordered pairs @xmath41 . with this definition
we are able to prove that for a pair of composable functors @xmath2 and
@xmath3 @xmath42 inductively , we define the _
@xmath0th order directional derivative _ @xmath43 and prove the desired analog of the huang - marcantognini - young chain rule equation : for a composable pair @xmath44 and @xmath45 of functors of abelian categories , there is a chain homotopy equivalence between the @xmath0th directional derivatives @xmath46 of course , this in - between " definition of the higher order directional derivatives raises a natural question how is it related to the more obvious choices for defining higher order directional derivatives ? in the case of @xmath47 , the functor @xmath43 is obtained by restricting along a diagonal functor @xmath48 as in the degree @xmath49 case .
the case of @xmath32 proves more interesting .
we show that @xmath43 and @xmath32 satisfy a fa di bruno - type relationship : for a functor @xmath2 between abelian categories , there is a chain homotopy equivalence @xmath50 where the sum is over non - negative integer solutions to the equation @xmath51 . a more concise formulation of this result , given in
[ sec : faa ] , makes use of the fact that the fa di bruno coefficients count the number of partitions of a set of @xmath0 elements . using these results
, we define ordinary ( non - directional ) derivatives for functors of modules over a commutative ring @xmath52 and prove a chain rule for these derivatives which is analogous to the huang - marcantognini - young rule from equation : for a composable pair of functors @xmath53 and @xmath54 of functors of abelian categories , the @xmath0th derivative @xmath55 is chain homotopy equivalent to @xmath56 where the sum is over non - negative integer solutions to the equation @xmath51 .
although our original motivation for this project was to prove higher order chain rules , we found that a significant part of the challenge in doing so involved placing these chain rules and the abelian functor calculus itself in an appropriate categorical context .
this was developed in two stages , using the concepts of kleisli categories and cartesian differential categories . for a functor @xmath57 between two abelian categories ,
the abelian functor calculus constructs a degree @xmath0 approximation @xmath58 , where @xmath59 is the category of chain complexes in @xmath60 concentrated in non - negative degrees . for a composable pair of functors @xmath57 and @xmath61 , this results in a pair of functors @xmath62 and @xmath63 which no longer appear to be composable . in @xcite ,
the authors form the composition @xmath64 by prolonging " @xmath20 to a functor @xmath65 , composing the prolongation with @xmath66 and taking the total complex of the resulting double complex .
we observe in section 3 that this process is precisely the composition in the kleisli category associated to a particular pseudomonad @xmath67 acting on the @xmath49-category of abelian categories , and use @xmath68 to denote that kleisli category . in functor calculus ,
the use of terms such as calculus " and derivatives " has been justified by pointing out strong formal resemblances to the classical constructions from analysis and undergraduate calculus .
although these analogies are compelling , one wonders if there is a deeper justification .
in the course of investigating the higher order chain rule for directional derivatives , we discovered that the properties of the directional derivative for abelian functor calculus correspond exactly to the axioms defining the cartesian differential categories of blute , cockett , and seely @xcite . this concept captures the notion of differentiation in a wide variety of contexts ( the category of real vector spaces and smooth maps equipped with the usual differential operator is a standard example of a cartesian differential category ) and provides an explanation for some of the formal resemblances .
we prove the homotopy category @xmath69 is a cartesian differential category . in this statement , @xmath70 is the homotopy category of @xmath68 obtained by inverting pointwise chain homotopy equivalences .
the directional derivative @xmath71 gives this category its cartesian differential category structure .
fitting calculus into this context led to additional insight stemming from the fact that a cartesian differential category is also an example of a tangent category , as defined by cockett and cruttwell @xcite .
tangent categories are characterized by the existence of an endofunctor @xmath72 satisfying essential properties of tangent bundles for manifolds .
our first proof of the chain rule of theorem [ t : higherchainrule ] did not make use of the tangent category structure on @xmath70 , but the proof presented here , which does use the endofunctor @xmath72 , provides a conceptual simplification . as the first - order chain rule can be seen to encode the functoriality of the derivative , the @xmath0th order chain rule can be extracted from the functoriality of the iterated endofunctor @xmath73 , which is an easily derived consequence of our corollary [ cor : cdc ] . proving that @xmath70 is a cartesian differential category required reworking many of the constructions and results for abelian functor calculus established in @xcite . driving these challenges
was the fact that chain rules in the abelian context do not come in the form of isomorphisms .
for example , applying the definition of @xmath71 ( see definition [ defn : dir - derivative ] or [ defn : dir - derivative - jm ] ) to @xmath1 produces a chain complex that is not isomorphic to @xmath74 in general . instead
, when @xmath3 is reduced ( @xmath75 ) , @xcite prove that there is a quasi - isomorphism between @xmath76 and @xmath74 .
the axioms of a cartesian differential category require that @xmath77 and @xmath74 be isomorphic .
this suggests that we attempt to realize @xmath68 as a cartesian differential category by inverting quasi - isomorphisms .
but because the functors we consider between abelian categories are far from being exact , when we invert quasi - isomorphisms in @xmath68 , the composition of morphisms is no longer well - defined .
however , the somewhat delicately defined composition in our kleisli category preserves chain homotopy equivalences , so we define @xmath70 to be the category that inverts these instead . because of this we needed to upgrade many definitions and results of @xcite from quasi - isomorphisms to chain homotopy equivalences . revising results of @xcite to work up to chain homotopy equivalence rather than quasi - isomorphism was a relatively straightforward process , with the notable exception of the chain rule itself . that required proving the following proposition , which refines lemma 5.7 of @xcite .
for any composable pair of functors @xmath78 and @xmath79 with @xmath3 reduced , there is a chain homotopy equivalence @xmath80 to prove this , we prove a technical general result that might be of independent interest ( or might already be in the literature ) : let @xmath81 be a first - quadrant bicomplex so that every row except the zeroth row @xmath82 is contractible
. then the natural inclusion @xmath83 is a chain homotopy equivalence .
we include a proof of this lemma in appendix a. the proof defines an explicit chain homotopy on the total complex that does not come from a chain homotopy of bicomplexes , as defined in @xcite .
the proof of proposition 5.7 appears is in appendix b. combining proposition 5.7 with some other results , we prove for any pair of functors @xmath78 and @xmath79 , there is a chain homotopy equivalence @xmath84 this improves on proposition 5.6 of @xcite in two ways by replacing the quasi - isomorphism with a chain homotopy equivalence , and by removing the condition that @xmath3 be a reduced functor . in section 2 ,
we define and review properties of the cross effect functors .
these are the building blocks for the taylor towers of abelian functor calculus , including the linearization and directional derivative functors .
section 3 is used to define the kleisli and homotopy categories in which the constructions and main results of the paper take place . in section 4 ,
we show how the taylor towers of abelian functor calculus are built and establish their fundamental properties .
this treatment differs from that of @xcite in two important ways . as discussed in section [ s : context ] , we need the fundamental properties of the tower to hold up to chain homotopy equivalence instead of quasi - isomorphism . in addition , we have streamlined the means by which terms in the abelian functor calculus tower are defined .
we do so by identifying a comonad @xmath85 on the category of all functors between a fixed pair of abelian categories , and defining @xmath20 directly as a resolution of @xmath2 by this comonad . in @xcite ,
this approach was only used for the category of reduced functors . an extra step involving a mapping cone
was used to extend this construction to all functors . in section 5 ,
we start working out the main ingredients for our chain rules we define the linearization functor @xmath86 and derive its essential properties .
we use these to define the directional derivative for abelian functor calculus in section 6 , and prove that it gives @xmath70 the structure of a cartesian differential category . in section 7 , we define the iterated partial directional derivatives and the higher order directional derivatives and prove that they are related by the fa di bruno formula of theorem [ thm : restatedgoal ] .
we prove the analog of the higher order chain rule of @xcite in section 8 , using the tangent category structure of @xmath70 .
finally , in section 9 , we discuss how the directional derivatives are related to a notion of ordinary derivatives for abelian functor calculus and use the main result of section 7 to prove the fa di bruno - style formula for these derivatives in corollary [ cor : faachainrule ] .
this paper introduces a number of constructions on functors ( e.g. , the higher order directional derivatives ) and investigates their behavior with respect to composition , product constructions , and so forth .
a convenient context to describe these operations involves a large category whose objects are abelian categories and whose morphisms are functors between them ; the precise construction of this category , which is somewhat delicate , is given in
[ sec : kleisli ] .
there likely exist more than a set s worth of functors between any fixed pair of non - small abelian categories , so if our constructions on functors are to be interpreted globally , they need to take place in some extension of the usual zfc axioms for set theory : e.g. , by assuming there exists a hierarchy of inaccessible cardinals .
see @xcite for a friendly discussion of the myriad possible choices .
the authors thank the banff international research station for hosting the second women in topology workshop , which brought us together for a week in which many of the results in this paper were proven , and the pacific institute for the mathematical sciences , which provided travel support for us to work together in calgary .
we also thank maria basterra and kathryn hess for their roles in organizing the women in topology workshop .
the second author is grateful for support from the union college faculty research fund .
the fourth author is grateful for support from the national science foundation through dms-1551129 .
tslil clingman suggested that chain complexes might define a monad on abelian categories and robin cockett pointed out that non - reduced functors may fail to preserve chain complexes .
this is why the monad of [ sec : kleisli ] is pseudo and not strict .
geoffrey cruttwell explained the benefits of thinking of a cartesian differential category as a tangent category , a perspective which led to a simplified proof of theorem [ t : higherchainrule ] .
we thank randy mccarthy for helpful conversations in starting this project , and for the ideas that inspired it .
classically , the term _ cross effect _ was used to describe the combined effects of two or more forces , or to describe the difference between the quantities @xmath87 and @xmath88 . for _
reduced _ functions satisfying @xmath89 , the condition that @xmath90 for all @xmath10 and @xmath91 is equivalent to linearity .
thus , the cross effect function , defined by @xmath92 measures the failure of a reduced function @xmath9 to be linear ; this failure is called the _ deviation _ by eilenberg and mac lane @xcite . in this section , we will study an analogous notion for functors from a pointed category to an abelian category .
the cross effects were first extended to functors of additive categories by eilenberg and mac lane @xcite , and these became the fundamental building blocks of abelian functor calculus as developed by johnson and mccarthy @xcite .
we recall the definition of the cross effect and summarize the properties that we will need for cross effects functors . for the duration of this section , let @xmath93 be a category with a _ basepoint _ ,
i.e. , an initial object @xmath94 which is also terminal , and finite coproducts , denoted by @xmath95 .
let @xmath60 be an abelian category with zero object @xmath96 and biproducts denoted @xmath97 .
[ def : cr ] the _ @xmath0th cross effect _ of a functor @xmath98 is the @xmath0-variable functor @xmath99 defined recursively by @xmath100 @xmath101 and in general , @xmath102 despite the asymmetry in this definition , the @xmath0th cross effect is symmetric in its @xmath0-variables ( * ? ? ?
* proposition 1.2 ) . in a category with a basepoint and finite coproducts ,
each coproduct inclusion @xmath103 is a split monomorphism ; the relevance of the retraction is that split monomorphisms are preserved by any functor . in an abelian category ,
split monomorphisms extend to split short exact sequences .
in particular , it makes sense to define the cross effects as the direct sum complements of definition [ def : cr ] .
eilenberg and mac lane defined cross effects as the images of certain homomorphisms , but the direct sum complement definition is more useful for the types of computations we wish to do .
[ ex : add - constant ] fix an object @xmath104 in an abelian category @xmath60 and consider the functor @xmath105 defined by @xmath106 . since @xmath107
, it follows that @xmath108 .
now @xmath109 implies that @xmath110 , so all higher cross effects must also vanish . definition [ def : cr ] is _ functorial _ in the sense that a natural transformation @xmath111 induces a naturally - defined map @xmath112 . to precisely describe this functoriality , write : * @xmath113 for the category of all functors from @xmath93 to @xmath60 ; * @xmath114 for the category of all _ strictly reduced _
functors from @xmath93 to @xmath60 , that is , functors for which @xmath115 ; and * @xmath116 for the category of all _ strictly multi - reduced _ functors from @xmath117 to @xmath60 , that is , functors for which @xmath118 if any @xmath119 . an easy exercise ( * ? ? ?
* proposition 1.2 ) shows that @xmath120 is an object in @xmath121 .
[ lem : cr - idempotency ] for any @xmath122 and any @xmath98 , @xmath123 .
the isomorphism @xmath124 implies that @xmath125 .
now @xmath126 proves that @xmath127 .
the general case now follows immediately from definition [ def : cr ] by induction .
importantly , the cross effect functors admit adjoints : [ prop : coreflective ] the inclusion of the category of strictly reduced functors into the category of not - necessarily reduced functors admits a right adjoint , namely the first cross effect functor : the component of the counit of this adjunction at a functor @xmath128 is the natural inclusion @xmath129 .
the universal property of the adjunctions asserts that if @xmath2 is reduced and @xmath3 is not necessarily reduced then any natural map @xmath130 factors uniquely through the inclusion @xmath131 .
let @xmath132 denote the natural transformation to the basepoint object in @xmath93 , @xmath133 , and consider the diagram : the right - hand square commutes by naturality of @xmath134 .
the kernels of the right - hand horizontal maps define the first cross effects . by commutativity of the right - hand square
, there exists a unique map @xmath135 so that the left - hand square commutes .
but because @xmath136 , the top left - hand map is an isomorphism , which gives us the desired unique factorization .
a full subcategory is _ coreflective _ when the inclusion admits a right adjoint .
proposition [ prop : coreflective ] asserts that @xmath114 defines a coreflective subcategory of @xmath113 . because the left adjoint is full and faithful the unit is necessarily an isomorphism . by abstract nonsense ,
the inclusion of a coreflective subcategory is necessarily _ comonadic _ , that is , @xmath114 is the category of coalgebras for the idempotent comonad @xmath137 acting on @xmath113 .
this formalism gives a characterization of the reduced functors : namely @xmath2 is reduced if and only if the natural map @xmath138 is an isomorphism .
johnson and mccarthy observe that when the construction of the @xmath0th cross effect is restricted to reduced functors , it is right adjoint to pre - composition with the diagonal functor ( * ? ? ?
* example 1.8 ) .
proposition [ prop : coreflective ] allows us to extend this adjunction to non - reduced functors : [ cor : comonad ] there is an adjunction between the @xmath0th cross effect functor and the functor given by precomposing with the diagonal @xmath139 , inducing a comonad @xmath85 on @xmath113 defined by @xmath140 by lemma [ lem : cr - idempotency ] @xmath141 for @xmath122 .
thus , the adjunction @xmath142 is the composite of the adjunctions of proposition [ prop : coreflective ] with the adjunction of ( * ? ? ?
* example 1.8 ) . the functor categories
@xmath113 and @xmath143 are abelian , with kernels and direct sums and so forth defined objectwise in @xmath60 . in particular , these categories have short exact sequences . a functor from one abelian category to another which preserves exact sequences
is called _ exact_. [ p : cr is exact ] for each @xmath144 , the functors @xmath145 and @xmath146 are exact .
let @xmath147 be a short exact sequence of functors .
consider the diagram below @xmath148\ar[d ] & { \mathrm{cr}}_1f\ar[r]\ar[d]&{\mathrm{cr}}_1g\ar[r]\ar[d]&{\mathrm{cr}}_1h\ar[r]\ar[d]&0\ar[d]\\ 0\ar[r]\ar[d]&f\ar[r]\ar[d]&g\ar[r]\ar[d]&h\ar[r]\ar[d]&0\ar[d]\\ 0\ar[r]&f(\star)\ar[r]&g(\star)\ar[r]&h(\star)\ar[r]&0.\\ } \ ] ] since the columns are exact , and the bottom two rows are exact , the @xmath149 lemma @xcite guarantees that the top row is exact . then assuming that @xmath150 is exact
, we can apply the same lemma to the diagram @xmath151\ar[d]&{\mathrm{cr}}_{n+1}g(a_1 , a_2 , b_2 , \dots , b_n)\ar[r]\ar[d]&{\mathrm{cr}}_{n+1}h(a_1 , a_2 , b_2 , \dots , b_n)\ar[d]\\ { \mathrm{cr}}_nf(a_1\vee a_2 , b_2 , \dots , b_n ) \ar[r]\ar[d]&{\mathrm{cr}}_ng(a_1\vee a_2 , b_2 , \dots , b_n)\ar[r]\ar[d]&{\mathrm{cr}}_nh(a_1\vee a_2 , b_2 , \dots , b_n)\ar[d]\\*+\txt{${\mathrm{cr}}_nf(a_1,b_2,\ldots , b_n)$\\$\oplus { \mathrm{cr}}_nf(a_2 , b_2,\ldots , b_n)$}\ar[r]&*+\txt{${\mathrm{cr}}_ng(a_1,b_2,\ldots , b_n)$\\$\oplus { \mathrm{cr}}_ng(a_2 , b_2,\ldots , b_n)$}\ar[r]&*+\txt{${\mathrm{cr}}_nh(a_1,b_2,\ldots , b_n)$\\$\oplus { \mathrm{cr}}_nh(a_2 , b_2,\ldots , b_n)$ } } \ ] ] to conclude that @xmath152 is exact .
the proof follows by induction .
the result for @xmath85 follows immediately .
_ abelian functor calculus _ also called _ additive _ or _ discrete functor calculus _ @xcite considers arbitrary functors valued in an abelian category .
the linear approximation defined in @xcite satisfies a universal property `` up to homotopy . ''
for this to make sense , the target abelian category must be replaced by some sort of homotopical category in which strict universal properties ( asserting that certain diagrams commute on the nose ) can be replaced by weak ones ( where the commutativity is up to some sort of homotopy relation ) . for an abelian category @xmath60 ,
let @xmath59 denote the category of chain complexes on @xmath60 concentrated in non - negative degrees .
in general , the linear approximation of @xmath98 defines a functor @xmath153 ; a precise definition of this is given in definition [ def : d1 ] .
more generally , @xcite define a linear approximation @xmath153 for any functor @xmath78 so that when this construction is applied to a functor concentrated in degree zero it recovers the construction for @xmath98 .
now consider a composable pair of functors @xmath154 and @xmath155 between abelian categories . , it suffices to assume that the domain of @xmath3 is a pointed category with finite coproducts , but when considering composite functors it is linguistically convenient to suppose that all of the categories are abelian .
] their linear approximations @xmath156 and @xmath153 are not obviously composable . the standard trick , which appears in ( * ? ? ?
* lemma 5.7 ) , is to prolong the second functor , applying @xmath157 degreewise to a chain complex in @xmath93 to produce a chain complex in chain complexes in @xmath60 , and then convert this double complex into a chain complex in @xmath60 by means of the totalization .
that is , the composite @xmath158 is defined to be the functor if @xmath3 is reduced , then ( * ? ? ?
* lemma 5.7 ) ( a stronger version of which appears as proposition [ lem:5.7 ] below ) proves that this composite is quasi - isomorphic to @xmath159 , i.e. , that @xmath86 is `` functorial up to quasi - isomorphism '' with respect to the composition structure just introduced . to keep track of
which functors should be composable and which are not ( and to avoid a proliferation of @xmath67 s ) it is convenient to regard both @xmath2 and @xmath157 as functors from @xmath93 to @xmath60 .
this sort of bookkeeping is effortlessly achieved by the categorical formalism of a kleisli category for a monad that we now introduce .
[ obs : ch - monad ] there is a pseudomonad @xmath160 acting on the ( large ) 2-category of abelian categories , arbitrary functors between them , and natural transformations .
here we are not interested in the 2-dimensional aspects , so we instead describe the quotient monad @xmath160 acting on the 1-category @xmath161 of abelian categories and isomorphism classes of functors : * the monad carries an abelian category @xmath60 to the category @xmath59 of non - negatively graded chain complexes in @xmath60 . * the monad carries a functor @xmath98 to its prolongation @xmath162 . because @xmath2 might not preserve zero maps , the definition of its prolongation is somewhat delicate . making use of the dold - kan equivalence between
non - negatively graded chain complexes and simplicial objects , the functor @xmath163 is defined to be the composite @xmath164^{\quad\simeq } & { \mathcal{b}}^{{\mathbbe{\delta}}^{\mathrm{op } } } \ar[r]^{f _ * } & { \mathcal{a}}^{{\mathbbe{\delta}}^{\mathrm{op } } } \ar[r]^{\simeq } & { { \textup{\textsf{ch}}}}{\mathcal{a}}.}\ ] ] where the action of @xmath2 on simplicial objects is by post - composition .
note this operation is not strictly functorial : if @xmath165 and @xmath166 are composable functors , then @xmath167 and @xmath168 are naturally isomorphic but not identical .
this is why we take the morphisms in @xmath161 to be functors up to natural isomorphism . *
the components of the unit of the monad are the functors + + that embed @xmath60 as the subcategory of chain complexes concentrated in degree zero . *
the components of the multiplication of the monad are the functors @xmath169^{\mathrm{tot } } & { { \textup{\textsf{ch}}}}{\mathcal{a}}}\ ] ] that convert a double complex in @xmath60 into a chain complex in @xmath60 by forming the total complex .
we leave to the reader the straightforward verification that these functors define the components of a monad @xmath160 on @xmath161 . for any category acted upon by a monad
there is an associated kleisli category ( see , e.g. , @xcite ) , which we describe explicitly in the special case of @xmath160 acting on @xmath161 .
there is a ( large ) category acting on the 2-category of abelian categories , functors , and natural transformations .
the category @xmath68 is then the quotient 1-category whose morphisms are isomorphism classes of parallel 1-cells . while it is somewhat inelegant to define the morphisms in a 1-category to be isomorphism class of functors , much of our work actually takes place in a further quotient of @xmath68 where naturally chain homotopically equivalent functors are identified ; see definition [ defn : hoabcat ] . ] @xmath68 whose : * objects are abelian categories ; * morphisms @xmath170 are natural isomorphism classes of functors @xmath171 ; * identity morphisms @xmath172 are the functors @xmath173 ; and in which * composition of morphisms @xmath174 and @xmath170 , corresponding to the pair of functors @xmath79 and @xmath78 , is defined by @xmath175^{\quad g } & { { \textup{\textsf{ch}}}}{\mathcal{b}}\ar[r]^{{{\textup{\textsf{ch}}}}f\ } & { { \textup{\textsf{ch}}}}{{\textup{\textsf{ch}}}}{\mathcal{a}}\ar[r]^{\mathrm{tot } } & { { \textup{\textsf{ch}}}}{\mathcal{a}}}\ ] ] note that @xmath161 defines a subcategory of @xmath68 , where a functor @xmath98 is identified with the morphism @xmath170 in @xmath68 represented by the functor @xmath176^f & { \mathcal{a}}\ar[r]^{{\textup{deg}}_0\ } & { { \textup{\textsf{ch}}}}{\mathcal{a}}}\ ] ] note that the results of section [ sec : cross - effects ] apply to the homs in the kleisli category @xmath68 by considering the functor categories @xmath177 , henceforth denoted simply by @xmath113 . henceforth , we work in the kleisli category @xmath68 without further comment .
practically , this means that we may simply write `` @xmath178 '' to denote what is really a functor @xmath78 .
there is no ambiguity in the special case where this functor is concentrated in degree zero , i.e. , is actually a functor valued in @xmath60 .
composition of functors is implemented by the kleisli construction : by prolonging the second functor and taking the totalization of the resulting double complex . in particular , all functors whose target is an abelian category are regarded as chain - complex valued , so for instance we can safely refer to chain homotopy equivalences classes of functors or ask whether a given functor is objectwise contractible . in everything that follows ,
it is always possible to relax the hypotheses on the domain category and assume only the existence of finite coproducts and a zero object , but we often refer simply to functors `` between abelian categories '' to avoid being overly pedantic with our language . accordingly , we now write `` @xmath97 '' and `` @xmath96 '' for the coproduct and basepoint in our domain categories . the kleisli category @xmath68 is the appropriate context to study composition relations involving the johnson - mccarthy polynomial functors @xmath179 and linearization functors @xmath180 .
however , there remains one additional technical issue that needs to be addressed .
many of the key properties of @xcite are proven up to pointwise quasi - isomorphism in the codomain @xmath59 .
in fact , @xcite introduce various models of the functors @xmath20 and @xmath157 that are only well - defined up to quasi - isomorphism .
unfortunately , quasi - isomorphisms are only preserved under composition with exact functors , while the functors we include as morphisms in @xmath68 are far from being exact .
thus , if we take the point of view that the functor @xmath181 is only defined up to quasi - isomorphism , the kleisli composite @xmath182 will not be well - defined for arbitrary functors @xmath2
. however , the composition operation in @xmath68 does respect natural chain homotopy equivalence of functors , and with some care we will show in sections [ sec : functor - calculus ] and [ sec : linearapproximations ] that the universal properties up to quasi - isomorphism established in @xcite in fact hold up to chain homotopy equivalence .
specifically , two functors @xmath183 are chain homotopic if the chain complexes @xmath184 and @xmath185 are chain homotopic in @xmath59 for every object @xmath186 .
the following lemma proves that composition defined in the kleisli category @xmath68 respects chain homotopy .
[ lem : che - comp ] suppose @xmath187 are chain homotopic functors .
then for any pair of functors @xmath178 and @xmath188 , the composite functors @xmath189 are chain homotopic .
it is obvious that if @xmath3 and @xmath190 are pointwise chain homotopic , then so are the restrictions @xmath191 and @xmath192 .
the non - trivial part is to show that @xmath193 and @xmath194 are chain homotopic .
we first argue that if @xmath195 are pointwise chain homotopic , then the composite functors are pointwise chain homotopic , where we define chain homotopies in @xmath196 as in any category of chain complexes valued in an abelian category ( which in this case happens to be @xmath59 ) .
this is a consequence of the dold - kan prolongation used to define the monad @xmath160 .
the dold - kan equivalence takes takes chain homotopies of chain complexes in @xmath197 to simplicial homotopies of simplicial objects in @xmath198 and vice - versa .
simplicial homotopies in @xmath198 are structurally defined , and so preserved by post - composition with @xmath78 .
thus , we conclude that a pointwise chain homotopy between @xmath3 and @xmath190 is carried to a pointwise chain homotopy between @xmath199 and @xmath200 .
to prove that the kleisli composites @xmath201 are pointwise chain homotopic , we need only argue that the totalization functor @xmath202 preserves chain homotopies .
the chain homotopies we are considering in @xmath196 are a special case of the more general notion of `` chain homotopy of bicomplexes , ''
so this follows from @xcite .
each hom - set in the category @xmath68 is equipped with a `` homotopy equivalence of functors '' relation defined as pointwise chain homotopy equivalence in the codomain of the functor .
lemma [ lem : che - comp ] implies that the pointwise chain homotopy equivalences classes in each @xmath203 are respected by the composition operation in @xmath68 .
we denote these equivalence classes by @xmath204 $ ] , and we let @xmath69 denote the category with the same objects as @xmath68 and with hom - sets @xmath204 $ ] .
this is the advertised categorical context for abelian functor calculus .
[ defn : hoabcat ] there is a ( large ) category @xmath69 whose : * objects are abelian categories ; * morphisms @xmath170 are chain homotopy equivalence classes of functors @xmath171 ; * identity morphisms @xmath172 are the functors @xmath173 ; and in which * composition of morphisms @xmath174 and @xmath170 , corresponding to the pair of functors @xmath79 and @xmath78 , is defined by @xmath205^{\quad g } & { { \textup{\textsf{ch}}}}{\mathcal{b}}\ar[r]^{{{\textup{\textsf{ch}}}}f\ } & { { \textup{\textsf{ch}}}}{{\textup{\textsf{ch}}}}{\mathcal{a}}\ar[r]^{{\mathrm{tot } } } & { { \textup{\textsf{ch}}}}{\mathcal{a}}}\ ] ] this category is an identity - on - objects quotient of @xmath68 with hom - sets denoted by @xmath206 $ ] defined by taking pointwise chain homotopy equivalence classes of functors in @xmath113 .
in this section we review the taylor tower of a functor constructed in section 2 of @xcite .
the constructions we will provide are essentially the same as those in @xcite , though our presentation differs in two ways that are relevant to our treatment of the chain rule . first , we have chosen to emphasize functors which are not necessarily reduced .
the results of section [ sec : cross - effects ] make it clear that both the reduced and non - reduced cases can be treated simultaneously .
second , we refine the notion of _ degree @xmath0 _ functors , replacing the requirement of acyclicity ( quasi - isomorphism ) in ( * ? ? ?
* definition 2.9 ) with a stronger contractibility ( chain homotopy equivalence ) condition .
proposition [ p : pnproperties ] proves that the up to quasi - isomorphism universal properties of the polynomial approximations proven in @xcite become up to chain homotopy equivalence universal properties in the present context .
the taylor tower will consist of a list of polynomial degree @xmath0 functors with natural transformations between functors of degree @xmath0 and degree @xmath207 .
we begin this section by explaining what it means for a functor to be _
degree @xmath0_. a functor @xmath178 is _
degree @xmath0 _ if @xmath208 is _ contractible _ , i.e. , pointwise chain homotopy equivalent to zero .
we write `` @xmath209 '' when there exists a pointwise chain homotopy equivalence between functors .
note that if @xmath210 , then @xmath211 for all @xmath212 .
in particular , degree @xmath213 functors are also degree @xmath214 for all @xmath212 .
following @xcite , we now define the universal polynomial degree @xmath0 approximations to a functor @xmath44 . from any comonad acting on an abelian category and object in that category one can extract a chain complex ( * ? ? ?
* definition 2.4 ) .
the following definition is a combination of the two definitions of @xmath20 for reduced and unreduced functors of ( * ? ? ?
* definition 2.8 ) into a single definition .
[ def : pn ] the _ @xmath0th polynomial approximation
_ @xmath215 of a functor @xmath178 is the functor that carries @xmath216 to the ( totalization is represented by a functor @xmath217 that is not concentrated in degree zero , we take @xmath20 to be the totalization of the bicomplex obtained from the comonad resolution of @xmath2 , as described in definition [ def : pn ] . ] of the ) chain complex in @xmath60 defined by @xmath218 for @xmath219 with differentials defined to be the alternating sums @xmath220 of the counit map .
[ rmk : p0 ] when @xmath221 , it is easy to give an explicit computation of the chain complex @xmath222 . by lemma [ lem : cr - idempotency ] ,
@xmath223 for all @xmath219 .
thus , we can explicitly compute this functor as the chain complex of functors @xmath224 ^ 0 & { \mathrm{cr}}_1f \ar[r]^{{\mathrm{id } } } & { \mathrm{cr}}_1f \ar[r]^0 & { \mathrm{cr}}_1f \ar[r]^\epsilon & f,}\ ] ] where the differentials continue to alternate between @xmath96 and @xmath225 . since @xmath226
, we can rewrite the chain complex @xmath222 as a direct sum of the two chain complexes @xmath224 ^ 0 & { \mathrm{cr}}_1f \ar[r]^{{\mathrm{id } } } & { \mathrm{cr}}_1f \ar[r]^0 & { \mathrm{cr}}_1f \ar[r]^{id } & { \mathrm{cr}}_1f}\ ] ] and @xmath227 & 0 \ar[r ] & 0 \ar[r ] & 0 \ar[r ] & f(0 ) . } \ ] ] here we have used the fact that the map @xmath228 is the identity on the component @xmath229 of @xmath226 .
the chain complex on the top line is contractible .
thus , we will use the chain complex in the second line as our model for @xmath222 , and accordingly we will write @xmath230 . recalling our convention of working in the kleisli category described in section [ sec : kleisli ] , the @xmath0th polynomial approximation construction defines a functor @xmath179 . proposition [ p : cr is exact ]
implies that this functor has good homotopical properties .
[ p : pnproperties ] for any @xmath231 , 1 .
@xmath232 is exact .
@xmath233 preserves preserves chain homotopies , chain homotopy equivalences , and contractibility . for ( i )
, let @xmath234 be an exact sequence of functors . by proposition [ p :
cr is exact ] , @xmath235 is a sequence of chain complexes that is exact in each degree .
hence , the sequence is exact .
the properties enumerated in ( ii ) are consequences of ( i ) : exact functors preserve chain homotopies and the zero chain complex .
the functor @xmath20 receives a natural transformation @xmath236 defined by inclusion into the degree zero part of the chain complex @xmath20 .
the basic properties of the @xmath0th polynomial approximation are summarized in the following proposition , an adaptation of ( * ? ? ?
* lemma 2.11 ) .
[ prop : poly - facts ] for @xmath178 , 1 .
[ itm : degree ] the functor @xmath20 is degree @xmath0 .
[ itm : q - iso ] if @xmath2 is degree @xmath0 , then the map @xmath237 is a chain homotopy equivalence .
[ itm : univ ] the pair @xmath238 is universal up to chain homotopy equivalence with respect to degree @xmath0 functors receiving natural transformations from @xmath2 .
we start with , writing @xmath239 for the @xmath213th degree component of @xmath78 . by definition
, @xmath20 is the total complex of the bicomplex whose @xmath213th column is @xmath240 .
we prove that @xmath20 is degree @xmath0 by showing that the bicomplex @xmath241 is contractible . using ( * ? ? ?
* definition 5.7.3 ) , a contraction for a bicomplex is given by horizontal contractions @xmath242 and vertical contractions @xmath243 which satisfy @xmath244 , @xmath245 , and @xmath246 .
in particular , when the bicomplex has naturally contractible columns we see that it is contractible by setting @xmath247 . the proof of ( * ? ? ? * proposition 2.7 ) implies that each of the columns @xmath248 is contractible with the contracting homomorphism given by @xmath249 .
since totalization preserves chain homotopy equivalence , we conclude that the chain complex @xmath241 is contractible . for
, recall that @xmath20 is the totalization of the first - quadrant bicomplex whose @xmath213th row is @xmath250 the map @xmath237 is the natural inclusion of the zeroth row into the totalization . by theorem [ thm : chain - htpy - equiv ] , to prove that this is a chain homotopy equivalence , it suffices to show that each chain complex @xmath251 is contractible for @xmath252 .
by exactness of @xmath253 , it suffices to prove that the chain complex @xmath254 is contractible .
but the hypothesis that @xmath2 is degree @xmath0 tells us immediately that @xmath255 is contractible .
part follows formally .
let @xmath256 be a natural transformation from @xmath2 to a degree @xmath0 functor @xmath3 .
the natural map @xmath257 provides a commutative diagram where the right hand map is a chain homotopy equivalence by part .
this shows that @xmath258 factors through @xmath237 up to pointwise chain homotopy equivalence . for uniqueness , consider another factorization the maps @xmath259 are not identical , but do agree up to a natural automorphism of @xmath260 . in particular , because @xmath20 is degree @xmath0 by , @xmath261 is a chain homotopy equivalence by , and thus so is @xmath262 . in this way
, we see that @xmath263 is determined up to chain homotopy equivalence by @xmath264 , and so is unique up to chain homotopy equivalence .
since @xmath265 is also degree @xmath0 , the universal property of proposition [ prop : poly - facts ] provides a factorization the resulting tower of functors : is called the _ algebraic taylor tower _ of @xmath2 .
[ rmk : rho]explicitly , the map @xmath266 is induced by a natural transformation of comonads @xmath267 given on components by the composite @xmath268 of the canonical direct summand inclusion followed by the image of the fold map @xmath269 @xcite .
[ ex : add - constant2 ] recall the functor @xmath105 defined by @xmath270 from example [ ex : add - constant ] . since @xmath271 and @xmath272 for @xmath273 , it follows immediately from definition [ def : pn ] that @xmath274 for all @xmath144 .
when @xmath221 , it follows immediately from remark [ rmk : p0 ] that @xmath275 .
indeed , since the chain homotopy @xmath276 is given by the chain map which projects @xmath184 onto @xmath277 in degree 0 , one sees that the map @xmath278 , or @xmath279 , is also this projection map .
in this section we consider the linear approximation to a functor , or the homotopy fiber of the map @xmath280 .
the linearization of a functor will be an important ingredient in defining the directional derivative and the two types of higher order derivatives considered in sections [ sec : faa ] and [ sec : chain - rule ] .
since the linearization of a functor @xmath2 with codomain @xmath93 is a functor @xmath157 with codomain @xmath197 , the kleisli category conventions of section [ sec : kleisli ] will be particularly useful here , especially when we consider linearization together with composition of functors . in order to maximize the benefits of this structure , we make two main changes in our presentation of the linear approximation from the presentation in @xcite .
first , we insist that the linear approximation @xmath157 be a morphism in @xmath68 . in particular , this means we insist that @xmath157 take values in _ non - negatively graded _ chain complexes . in @xcite , @xmath157 ( and indeed all of the homotopy fibers of the maps @xmath281 ) are defined using a particular model for the homotopy fiber given by taking the mapping cone of @xmath282 shifted down one degree .
this shift means that the resulting chain complex will have a non - zero object in degree @xmath283 . in definition [ def : d1 ]
we give an alternate model of @xmath157 which is chain homotopic to the model in @xcite , but which is concentrated in non - negative degrees .
second , since composition by prolongation does not always preserve quasi - isomorphisms , we have chosen to work up to the stronger notion of chain homotopy equivalence .
in particular , we have provided explicit chain homotopy equivalences for the @xmath86 chain rule in proposition [ lem:5.7 ] , which strengthens ( * ? ? ? * lemma 5.7 ) .
this chain rule is strengthened further in proposition [ prop : d_1chainrule ] , where we present the analogue of the @xmath86 chain rule for functors which are not necessarily reduced .
indeed , all of the properties of the linear approximation @xmath157 are developed up to chain homotopy equivalence , and for functors which need not be reduced .
these properties occupy the majority of this section .
[ def : d1 ] the _ linearization _ of @xmath284 is the functor @xmath285 given as the explicit chain complex @xmath286 where : @xmath287 the chain differential @xmath288 is given by the map @xmath289 of [ rmk : rho ] , and the chain differential @xmath290 is given by @xmath291 when @xmath219 .
[ rmk : d1 ] @xmath292 1 . using the model of @xmath222
described in remark [ rmk : p0 ] , the map @xmath293 is surjective , so the homotopy fiber @xmath157 is simply the kernel of @xmath294 .
there is an inclusion map from @xmath157 into the homotopy fiber of @xmath295 , and this inclusion is a chain homotopy equivalence .
thus , this model is chain homotopy equivalent to the one defined in @xcite .
2 . viewed another way , definition [ def : d1 ] is exactly the same chain complex as @xmath296 , which is the degree 1 approximation to the reduction of @xmath2 .
these two observations imply that , up to chain homotopy equivalence , the two processes of reducing a functor and taking its degree 1 approximation produce the same result when applied in either order .
[ ex : linearization - of - adding - constant ] returning to the functor @xmath297 from examples [ ex : add - constant ] and [ ex : add - constant2 ] , since @xmath298 and @xmath299 , the linearization @xmath157 is given by the chain complex which is the functor @xmath225 concentrated in degree 0 .
thus , the linearization of @xmath2 is @xmath225 .
we now turn our attention to the properties of @xmath86 .
first , we see that the linearization construction defines an exact functor : [ p : dnproperties ] @xmath292 1 .
@xmath300 is exact .
@xmath86 preserves chain homotopies , chain homotopy equivalences , and contractibility .
since @xmath157 is isomorphic to @xmath301 , ( i ) is a direct consequence of proposition [ p : pnproperties ] , and ( ii ) follows from exactness .
next , we would like to justify the use of the term _ linearization _ for @xmath86 .
we start by defining what it means for a functor to be linear : [ defn : linear ] a functor @xmath98 is _ linear _ if it is degree one and also reduced , meaning that @xmath277 is contractible .
equivalently , @xmath2 is linear if it preserves finite direct sums up to chain homotopy equivalence .
we will start with reduced functors .
if @xmath2 is reduced , @xmath157 is exactly the analog of the linear approximation of a function for functors .
that is , a function @xmath302 whose graph passes through the origin is linear if @xmath90 .
[ lem : d1-linearity ] @xmath292 1 . for any @xmath178 ,
the functor @xmath303 is strictly reduced , and for any @xmath304 , the natural map @xmath305 is a chain homotopy equivalence .
in particular , @xmath157 is linear .
2 . the functor @xmath306 \to [ { \mathcal{b}},{\mathcal{a}}]$ ] is linear in the sense that @xmath307 and for any pair of functors @xmath308 $ ] , @xmath309 first , note that @xmath157 is strictly reduced because the functors @xmath310 and @xmath229 are both strictly reduced functors .
the fact that @xmath157 is strictly reduced now follows immediately from the definition of @xmath157 . to prove the second part of ( ii ) , note that the map @xmath311 is the inclusion of a direct summand with complement @xmath312 .
but @xmath313 by remark [ rmk : d1 ] , and @xmath314 is chain contractible by ( * ? ? ?
* lemma 2.11 ) .
it follows that the map @xmath315 is a chain homotopy equivalence
. part ( ii ) is a consequence of proposition [ p : dnproperties ] : any exact functor preserves finite direct sums up to isomorphism .
johnson and mccarthy prove that @xmath86 is functorial up to quasi - isomorphism whenever the first of a composable pair of functors is reduced .
the proof of ( * ? ? ?
* lemma 5.7 ) proceeds by showing that both @xmath316 and @xmath317 are quasi - isomorphic to a third chain complex , coming from a tricomplex containing each of the other two .
the next proposition , whose technical proof is deferred to appendix [ a : appb ] , shows that in fact @xmath316 is chain homotopy equivalent to @xmath158 .
[ lem:5.7 ] for any composable pair of functors @xmath178 and @xmath318 with @xmath3 reduced , there is a chain homotopy equivalence @xmath80 our next task is to extend proposition [ lem:5.7 ] to not necessarily reduced functors @xmath318 and @xmath319 .
the proof follows easily from the following two lemmas .
the first of these indicates that the linearization of @xmath2 is the same as the linearization of its reduced component , @xmath229 .
this is analogous to the statement that the linearizations of the functions @xmath320 and @xmath321 have the same slope .
the second function , @xmath321 , is reduced in the sense that its graph goes through the origin .
this analogy explains precisely how to linearize unreduced functors .
[ lem : reduced - vs - unreduced ] for any @xmath178 , the natural map @xmath322 is an isomorphism .
by lemma [ lem : cr - idempotency ] and corollary [ cor : comonad ] , @xmath323 and @xmath324 .
the result now follows from the definition of @xmath157 .
[ lem : cr_1chainrule ] let @xmath44 and @xmath45 be composable functors . then @xmath325 note that @xmath326 on the other hand , @xmath327 .
taking complements we conclude the claimed isomorphism . [
prop : d_1chainrule ] if @xmath44 and @xmath45 are composable functors , then there is a chain homotopy equivalence @xmath328 note that when @xmath3 is reduced , @xmath329 , so the second term vanishes .
thus proposition [ prop : d_1chainrule ] reduces to proposition [ lem:5.7 ] in this case .
taking the linearization of @xmath1 , or equivalently , the linearization of @xmath330 , we have @xmath331 we finish this section with a few observations about linearizations of functors of more than one variable . when @xmath332 is a functor of @xmath0 variables , it is possible to linearization @xmath2 with respect to a subset of the variables .
[ lem : multilinearization ] given a functor @xmath332 , let @xmath333 be the functor defined by @xmath334 where @xmath335 are fixed objects of @xmath93 .
write @xmath336 for @xmath337 .
when linearizing @xmath2 with respect to two or more variables , we have more than one option for how to proceed .
* we can linearize @xmath2 with respect to two or more variables _
simultaneously_. for @xmath338 , the linearization of @xmath2 with respect to its @xmath315 and @xmath339 variables simultaneously is the linearization of the functor @xmath340 defined by @xmath341 where the @xmath342 s are fixed objects in @xmath93 .
we denote this simultaneous linearization by @xmath343 .
similarly , we write @xmath344 for the linearization with respect to @xmath213 variables simultaneously .
* we can linearize @xmath2 with respect to two or more variables _
sequentially_. for @xmath338 , the linearization of @xmath2 with respect to its @xmath315 and @xmath339 variables sequentially is the linearization of the functor @xmath345 with respect to the @xmath339th variables .
that is , @xmath346 where @xmath347 is the functor obtained from @xmath348 by holding all but the @xmath339th variable constant , as in [ lem : multilinearization ] .
the special case in which @xmath349 has been linearized with respect to each of its @xmath0 variables sequentially is denoted @xmath350 .
that is , @xmath351 since the variables of a given multilinear functor are not always ordered , it is possible for confusion to arise when we are multilinearizing a functor either simultaneously or sequentially . to disambiguate , we will use the name of the variable in this case .
for example , the notation @xmath352 indicates that we should simultaneously linearize the third cross effect of @xmath2 with respect to the variables @xmath29 and @xmath353 , which occur in the second and third slots of @xmath354 , respectively .
we could equally well use the notation @xmath355 to indicate the same multilinearization , and context will determine which one is more practical .
the simultaneous linearization of functors of several variables is often a trivialization of the functor , as in the next lemmas .
[ lem : kristine s ] suppose that @xmath356 is strictly multi - reduced .
then for any @xmath357 , @xmath358 is contractible .
for simplicity of notation , we prove the lemma for a functor @xmath359 of two variables .
since @xmath360 is linear simultaneously in the variables @xmath361 , the inclusion @xmath362 is a chain homotopy equivalence by lemma [ lem : d1-linearity ] .
note that @xmath363 , so as a special case @xmath364 is chain homotopic to @xmath365 , which is zero because @xmath190 is strictly multi - reduced .
thus @xmath360 must be contractible as well .
lemma [ lem : kristine s ] is often applied in the following form : [ cor : kristine s ] suppose @xmath178 factors as where @xmath366 is the diagonal functor and @xmath190 is strictly multi - reduced
. then @xmath157 is contractible .
the diagonal functor is strictly reduced so by proposition [ lem:5.7 ] , @xmath367 is chain homotopic to @xmath368 , but here @xmath369 is a simultaneous linearization to which lemma [ lem : kristine s ] applies , implying that @xmath370 is contractible .
hence we have a chain contraction @xmath371 .
the simultaneous linearization of the cross - effects functor @xmath372 is contractible by lemma [ lem : kristine s ] .
however , the sequential multilinearization @xmath373 is generally not acyclic , hence not contractible .
indeed , @xmath374 recovers the @xmath0th layers of the taylor tower by ( * ? ? ?
* proposition 3.9 ) .
so , for example , if @xmath2 is any ( non - zero ) homogenous degree @xmath0 functor , then @xmath375 .
when linearizing sequentially , it does not matter in which order we linearize the variables .
[ lem : d1-commutativity ] for any @xmath376 , there is an isomorphism @xmath377 by using the definition of the cross effects , is possible to show that @xmath378 , where @xmath379 denotes the functor defined in convention [ lem : multilinearization ] .
this is a somewhat lengthy , though straightforward , argument .
instead we simply note that both functors are reduced and linear in the @xmath315th and @xmath339th variables , respectively and indeed that @xmath380 and @xmath381 satisfy the same universal property
. finally , it will be useful to record how linearization behaves with respect to products and projections .
the remaining results in this section describe this behavior .
[ lem : product ] let @xmath382 be the functor @xmath383
. then @xmath384 .
the functor @xmath385 is strictly linear so is isomorphic to its linearization .
[ cor : product ] for any @xmath178 , let @xmath386 be the functor defined by @xmath387 then @xmath388 . by direct computation , @xmath389 from which we conclude that @xmath390 .
similarly , @xmath391 since @xmath392 , we conclude from the definition of @xmath393 that @xmath394 . by definition of @xmath86 , we now conclude that @xmath395 .
note that @xmath396 is a composition of the functor @xmath2 and the reduced functor @xmath397 ; thus it is possible to obtain corollary [ cor : product ] as an immediate consequence of lemma [ lem : product ] and proposition [ lem:5.7 ] .
however , this result would only allow us to conclude that @xmath398 and @xmath157 are chain homotopy equivalent .
the direct proof provides a stronger result .
[ lem : other - product ] for a functor @xmath399 with components @xmath400 and @xmath318 , @xmath401 in the product abelian category @xmath402 , direct sums are defined coordinate - wise : @xmath403 in particular , the cross effects , the polynomial functors , and the homogeneous layers are all defined coordinate - wise , and the result follows .
applying the results of the previous section , we will now define the directional derivative for a functor in analogy with the directional derivative in function calculus .
we will show that our definition , which is new , recovers the directional derivatives defined in @xcite .
we review , generalize , and prove new properties of this functor , in analogy with the directional derivative from functional calculus except that equations are replaced with chain homotopy equivalences .
these results can be summarized in a single theorem : the directional derivative satisfies the axioms for a categorical derivative in the sense of @xcite , equipping @xmath69 with the structure of a cartesian differential category . in particular , we show that the directional derivative satisfies the chain rule .
the properties which we need in order to establish that @xmath404 is a differential category are precisely the properties we will need in order to obtain a higher order chain rule analogous to ( * ? ? ?
* theorem 3 ) , which we accomplish in section [ sec : chain - rule ] .
the directional derivative of a differentiable function @xmath405 ( or more generally , a function between banach spaces ) at the point @xmath406 in the direction @xmath407 measures how the value of @xmath9 at @xmath10 changes while translating along the infinitesimal vector from @xmath10 in the direction @xmath11 .
one way to make this idea precise is to define @xmath408 to be the derivative of the composite function , substituting the affine linear function @xmath409 into the argument of @xmath9 , evaluated at @xmath410 : @xmath411 this construction motivates the following definition for functors of abelian categories .
[ defn : dir - derivative ] for @xmath178 and @xmath412 , define @xmath413 alternatively , @xcite define the directional derivative for a functor @xmath178 from a category with finite coproducts and basepoint to an abelian category by using a sequence of analogies with the formula @xmath414\ ] ] from the calculus of functions of a real variable .
let @xmath186 be the ` point ' , corresponding to the point @xmath406 at which we will evaluate our derivative .
let @xmath415 denote the ` direction ' .
sums in @xmath93 are coproducts , which we will denote @xmath97 .
the difference operation in the expression @xmath416 becomes a kernel operation in our analogy , @xmath417 where the map @xmath418 in @xmath60 is obtained by applying @xmath2 to the map @xmath419 in @xmath93 which sends the @xmath420 component to the basepoint .
the linearization @xmath86 ( in the variable @xmath420 ) produces the degree 1 , reduced component of the functor , and hence corresponds to taking the limit as @xmath421 goes to @xmath96 in the expression for the directional derivative of @xmath9 .
[ defn : dir - derivative - jm ] let @xmath44 be a functor of abelian categories , and let @xmath29 and @xmath420 be objects in @xmath93 .
the directional derivative of @xmath2 at @xmath29 in the direction @xmath420 is the bifunctor defined by @xmath422 the following lemma proves that our two definitions agree .
[ lem : directional ] for a functor @xmath178 between abelian categories and any pair of objects @xmath412 , there is an isomorphism @xmath423 and a chain homotopy equivalence @xmath424 for the first isomorphism , recall that by the definition of the cross - effects , @xmath425 the direct summand inclusion @xmath426 is a section to the projection @xmath418 , so the kernel @xmath427 is isomorphic to @xmath428 . by lemma [ lem : d1-linearity](ii )
@xmath429 is linear , so @xmath430 with the last simplification by lemma [ lem : reduced - vs - unreduced ] . for the chain homotopy equivalence ,
apply proposition [ prop : d_1chainrule ] to @xmath2 and @xmath431 to obtain a chain homotopy equivalence @xmath432 in the proofs that follow it will be useful to have both definitions [ defn : dir - derivative ] and [ defn : dir - derivative - jm ] and the third description of lemma [ lem : directional ] to characterize @xmath27 .
one important observation about the directional derivative is that it preserves chain homotopy equivalences between functors . if @xmath2 and @xmath3 are pointwise chain homotopic , then @xmath433 .
this observation follows immediately from the fact that @xmath434 , since @xmath435 whenever @xmath436 .
thus , @xmath71 is an operation on chain homotopy equivalences classes of functors from @xmath93 to @xmath60 .
the properties of the directional derivative of a functor are reminiscent of the properties of the directional derivative of a function in classical calculus of several real variables .
the analogies between the classical notion of differentiation and this more categorical approach has been axiomatized in the notion of a _
cartesian differential category _
, introduced in @xcite though presented here in the equivalent form of ( * ? ? ?
* proposition 4.2 ) .
the following theorem is a collection of seven lemmas which will be very useful to us in proving the higher order chain rule and analyzing the higher order directional derivatives in the next two sections .
collectively , these properties tell us that the directional derivative @xmath71 equips the category @xmath69 with a cartesian differential structure .
[ thm : diffcat ] the category @xmath68 , together with the directional derivative @xmath71 , satisfies the following seven properties : 1 . for parallel functors
@xmath437 , @xmath438 2 . for any @xmath178 ,
@xmath33 is linear in the direction variable : i.e. , there is a chain homotopy equivalence @xmath439 and an isomorphism @xmath440 3 .
the directional derivative of the identity functor @xmath441 is given by projection onto the direction variable : @xmath442 4 .
for a functor @xmath399 with components @xmath400 and @xmath318 , @xmath443 5 .
whenever @xmath318 and @xmath178 are composable , there is a chain homotopy equivalence @xmath84 6 .
for any @xmath178 , there is an isomorphism @xmath444 7 . for any @xmath178
, there is a chain homotopy equivalence @xmath445 before proving theorem [ thm : diffcat ] , we summarize its statements .
a _ cartesian differential category _ is a category satisfying the following axioms .
firstly , it is _ left additive _ , meaning that each hom - set is a commutative monoid and pre - composition with any map is a monoid homomorphism .
a morphism in a left - additive category is then called _ additive _ if post - composing with it is also a monoid homomorphism .
secondly , the category has finite products so that the diagonal and product - projection maps are additive , and moreover the product of two additive maps is again additive .
finally , a cartesian differential category is equipped with an operation @xmath71 that takes a morphism @xmath446 and produces a morphism @xmath447 satisfying the properties ( i)-(vii ) of theorem [ thm : diffcat ] . see @xcite , @xcite .
the usual cartesian product of two categories that are abelian is again abelian ( see , for example , ( * ? ? ?
* exercise viii.3.2 ) ) .
one can verify that this construction defines a product in the category @xmath69 .
the hom - sets @xmath206 $ ] in this category are ( large ) commutative monoids , with addition of chain homotopy equivalence classes of functors defined pointwise using the direct sum in @xmath60 .
pre - composition manifestly respects this structure , as does post - composition with diagonal and product - projection maps .
it is similarly easy to verify that the product of two additive maps is additive .
thus theorem [ thm : diffcat ] can be summarized by the following corollary : [ cor : cdc ] the homotopy category @xmath69 is a ( large ) cartesian differential category .
the proof of theorem [ thm : diffcat ] will occupy the remainder of this section . for parallel functors
@xmath437 , @xmath448 the isomorphism @xmath449 now follows from linearity of @xmath86 in the form of lemma [ lem : d1-linearity](ii ) .
the chain homotopy equivalence @xmath439 and the isomorphism @xmath450 are immediate from either definition of @xmath33 and lemma [ lem : d1-linearity](i ) .
evidently @xmath451 .
the linearization of the identity functor is the identity , so by definition [ defn : dir - derivative - jm ] @xmath452 consider a functor @xmath399 . by definition [ defn :
dir - derivative - jm ] @xmath453 whenever @xmath318 and @xmath178 are composable and @xmath3 is reduced , ( * ? ? ?
* proposition 5.6 ) proves a quasi - isomorphism @xmath84 we will upgrade this to a chain homotopy equivalence and drop the hypotheses on @xmath3 . to make use of lemma [ lem : directional ] for @xmath454
, we first compute the second cross effect of @xmath455 , which is defined by @xmath456 this in turn is isomorphic to @xmath457 taking complements to cancel these terms in our two formulas for @xmath458 , we conclude that @xmath459 now each of these terms is multi - reduced , so upon applying @xmath429 , corollary [ cor : kristine s ] implies that any term in which @xmath420 appears more than once will vanish .
hence there is a chain homotopy equivalence : @xmath460 by proposition [ prop : d_1chainrule ] and this computation , this proves that there is a chain homotopy equivalence @xmath461 now let s work out the other side : @xmath462 which agrees with the formula for @xmath463 above . to prove the remaining statements theorem [ thm : diffcat](vi ) and ( vii ) we will provide a general formula for @xmath464 in terms of the cross - effects of @xmath2 that will prove important for our investigation of higher order chain rules in the next section .
our formula will make use of the following characterization of @xmath465 .
[ lem : nabla - cr2 ] for any @xmath178 and objects @xmath466 @xmath467 note that @xmath468 , which implies that @xmath469 .
thus @xmath393 is defined as the direct sum complement : @xmath470 and the claim follows by cancelling the isomorphic complements . with lemma [ lem : nabla - cr2 ] in hand , we can now compute @xmath471 .
the next lemma shows that @xmath472 is chain homotopy equivalent to @xmath473 plus an error term , which will be interpreted in the next section ; see proposition [ prop : n=2nablas ] .
[ lem : nablanabla ] for any @xmath474 , there is a chain homotopy equivalence @xmath475 by lemma [ lem : directional ] @xmath476 the functors @xmath477 and @xmath478 are strictly multi - reduced in the @xmath479 and @xmath480 variables , so after distributing the linearization @xmath481 over the direct sum , lemma [ lem : kristine s ] implies that these two terms are contractible .
thus , the previous term is chain homotopy equivalent to : @xmath482 and by corollary [ cor : product ] this is isomorphic to @xmath483 the proofs of theorem [ thm : diffcat ] ( vi ) and ( vii ) follow immediately from lemma [ lem : nablanabla ] .
however , using lemma [ lem : nablanabla ] will only provide a chain homotopy equivalence between the desired term .
when @xmath484 , we can strengthen this to an isomorphism by reexamining the proof of lemma [ lem : nablanabla ] . from the proof of lemma [ lem : nablanabla ] , recall that @xmath485 when @xmath484 , lemma [ lem : d1-linearity](i ) implies that the second term and the last three terms are 0 , so @xmath486 the proof of [ thm : diffcat](vii ) is now an immediate consequence of lemma [ lem : nablanabla ] . by inspecting the formula of lemma [ lem : nablanabla ] , lemma [ lem
: d1-commutativity ] and the symmetry of the cross effects implies that there is a chain homotopy equivalence @xmath487
the directional derivative @xmath488 of a functor @xmath178 can be iterated to a higher order directional derivative in several different ways .
the nave iteration @xmath489 has some redundancy stemming from the fact that @xmath33 is linear in its first variable ; see theorem [ thm : diffcat](ii ) and lemma [ lem : nablanabla ] .
this problem compounds for higher iterations : in some sense @xmath490 has @xmath207 too many variables . in this section
, we ll consider two `` more efficient '' formulations of the higher directional derivative that bear a very interesting relationship to one another captured by theorem [ thm : restatedgoal ] .
the first of these is a functor @xmath491 introduced by johnson
mccarthy , which we refer to as the @xmath0th _ iterated partial directional derivative_. the second of these is a new functor @xmath492 modeled after the higher directional derivatives of huang , marcantognini , and young , which we call the @xmath0th _ higher order directional derivative_. [ def : tildenabla^n ] let @xmath44 be a functor .
the @xmath0th _ iterated partial directional derivative _ of @xmath2 is defined recursively by @xmath493 and for @xmath273 and objects @xmath494 and @xmath29 in @xmath93 , the @xmath0th iterated partial directional derivative of @xmath2 at @xmath29 in the directions @xmath495 is @xmath496 johnson
mccarthy prove that the @xmath0th iterated partial directional derivative can be understood as a multilinearization of the @xmath0th and @xmath14st cross effects . recall that for a functor of at @xmath0-variables , @xmath497 indicates the linearization in each of its first @xmath0 variables sequentially . by lemma [ lem : d1-commutativity ] ,
the order in which these linearizations are performed does not matter .
@xcite [ prop:5.9 ] for any @xmath178 and variables @xmath498 1 .
there is a natural isomorphism @xmath499 2 . for any permutation @xmath500
, there is a natural isomorphism @xmath501 the first statement is ( * ? ? ?
* proposition 5.9 ) .
the second statement , appearing as ( * ? ? ?
* corollary 5.10 ) , follows immediately from the symmetry of the cross effects and lemma [ lem : d1-commutativity ] . on the other hand ,
the @xmath0th higher order directional derivative of @xcite suggests that the following definition should be considered .
[ def : delta_n ] for a functor @xmath44 the @xmath0th _ higher order directional derivatives _ of @xmath2 are defined recursively by @xmath502 and for @xmath503 and objects @xmath495 and @xmath29 in @xmath93 , the @xmath0th higher order directional derivative of @xmath2 at @xmath29 in the directions @xmath495 is defined to be @xmath504 note that @xmath505 and @xmath506 , so the recursive formulas for @xmath507 and @xmath508 given in definitions [ def : tildenabla^n ] and [ def : delta_n ] hold for all @xmath509 as well .
[ r : diagrammatically ] both the iterated partial directional derivative @xmath510 and the higher order directional derivative @xmath508 can be thought of as an iteration of the directional derivative @xmath71 where a new direction is introduced at each instance of differentiation . the difference between these two is that in the first case , the @xmath0th iterated partial directional derivative is considered as a functor of a single variable with its first @xmath511 `` directions '' ( that is , @xmath512 ) fixed .
the new single variable functor of @xmath29 is then differentiated in a new direction @xmath513 . in the second case ,
the @xmath0-variable functor @xmath514 is differentiated in the `` direction '' @xmath515 and at the `` point '' @xmath516 .
this `` point '' should be thought of as having recorded the history of how the functor has been differentiated in previous iterations , that is , in the directions @xmath517 , respectively .
the effect of this is that the functor @xmath508 can be understood as a restriction of @xmath518 along a diagonal functor @xmath519 defined by @xmath520 .
thus , as an operation on functors from @xmath93 to @xmath60 , we have @xmath521 where @xmath522 is the functor which precomposes with @xmath48 .
we can write this diagrammatically as @xmath523\ar[r]^-{\delta_{n-1 } } & [ { \mathcal{b}}^{n } , { \mathcal{a}}]\ar[r]^-{\nabla } & [ { \mathcal{b}}^{n}\times { \mathcal{b}}^n , { \mathcal{a}}]\ar[r]^-{l_n^ * } & [ { \mathcal{b}}^{n+1 } , { \mathcal{a}}]}\ ] ] where @xmath204 $ ] denotes the hom - set in @xmath70 . by definition , @xmath524 and @xmath525 . however ,
the two possible ways of iterating the directional derivative diverge for @xmath526 .
their precise relationship for @xmath527 is explained in the following proposition .
[ prop : n=2nablas ] for a functor @xmath44 and objects @xmath528 in @xmath529 , there is a chain homotopy equivalence @xmath530 specializing to the object @xmath531 in the image of @xmath532 , we have a chain homotopy equivalence @xmath533 by lemma [ lem : nablanabla ] , we have a chain homotopy equivalence @xmath534 by the characterization of @xmath33 in lemma [ lem : directional ] , @xmath535 .
we need only show that @xmath536 but this is precisely the natural isomorphism of proposition [ prop:5.9](i ) .
proposition [ prop : n=2nablas ] can be generalized to explain the relationship between @xmath508 and @xmath507 for all @xmath537 .
this relationship is surprisingly rich . in particular ,
its formulation is reminiscent of the fa di bruno formula for the @xmath0th derivative of @xmath5 for composable _ functions _ @xmath9 and @xmath17 .
the relationship is stated precisely in the next theorem . for a partition @xmath538 of the set @xmath539
, we use the following notation : @xmath540 where @xmath541 denotes the cardinality of the set @xmath542 . note that by proposition [ prop:5.9](ii )
, the order in which we write the @xmath543 s does nt matter .
[ ex : base - case ] for the partition @xmath544 , @xmath545 , while for @xmath546 , @xmath547 .
not coincidentally , these are the two terms appearing in the formula for @xmath548 proven in proposition [ prop : n=2nablas ] .
we are now prepared to state the general result : [ thm : restatedgoal ] for any @xmath178 and objects @xmath498 , there is a chain homotopy equivalence @xmath549 where the sum is taken over all partitions @xmath397 of the set @xmath550 .
theorem [ thm : restatedgoal ] is a consequence of the following lemma .
[ l : nabla1nablatwiddle ] for a partition @xmath538 of @xmath539 , there is a chain homotopy equivalence @xmath551 this lemma permits an expeditious proof of theorem [ thm : restatedgoal ] . the right - hand side of can be interpreted as the sum of terms of the form @xmath552 taken over all partitions @xmath263 of @xmath553 that are obtained from the partition @xmath538 of @xmath539 by either adding @xmath554 as a separate set ( contributing the first summand ) or adding @xmath555 to one of the @xmath556 s ( contributing the remaining @xmath213 summands ) . since all partitions of @xmath557 can be obtained in this way from partitions of @xmath539 , theorem [ thm : restatedgoal ] is proven by a simple induction with proposition [ prop : n=2nablas ] as the base case , as explained by example [ ex : base - case ] , and lemma [ l : nabla1nablatwiddle ] as the inductive step .
next we prove the lemma .
let @xmath558 be a partition of @xmath539 . by proposition [ prop:5.9](i ) @xmath559 by definition [ defn : dir - derivative - jm ]
, @xmath560 is equal to @xmath561 where we have added new variables @xmath562 to keep track of where the linearizations are being applied . recalling convention [ lem : multilinearization ] , the notation @xmath563 indicates the linearization @xmath370 of a functor @xmath564 , where @xmath190 is the functor @xmath565 and the @xmath566 s and @xmath29 are fixed values of @xmath93 . because @xmath190 is a functor of several variables , this linearization involves the simultaneous linearization of the variables @xmath567 .
our strategy will be to use proposition 5.9 to rewrite in terms of multilinearizations of cross effects , use lemma [ lem : kristine s ] and corollary [ cor : kristine s ] to eliminate many of the summands after taking the kernel and applying @xmath568 , and use proposition [ prop:5.9](i ) to verify that what is left is equivalent to ( [ e : lemma ] ) . by proposition [ prop:5.9](i ) ,
is isomorphic to @xmath563 applied to the kernel of the map since the domain of this map is linear in its first @xmath213 variables , it is chain homotopy equivalent to an expansion of this expression as illustrated for instance by : @xmath569 but rather than carrying out these expansions completely , we note that because we are applying @xmath563 , corollary [ cor : kristine s ] guarantees that any summand that has more than one @xmath562 will be contractible .
hence we see that @xmath570 is chain homotopy equivalent to @xmath563 of @xmath571 that is , the only terms with a single variable of the form @xmath572 appear in the kernel . a term of the form @xmath573 can be expanded as @xmath574 but , again by corollary [ cor : kristine s ] , the first and third summands will be contractible after applying @xmath563 . hence , after computing the kernel and applying @xmath563 , we see that @xmath575 is chain homotopy equivalent to @xmath576 by proposition [ prop:5.9](i ) , this is isomorphic to : @xmath577
this brings us to the original motivation for this paper , the higher order directional derivative chain rule . having established a relationship between the @xmath0th higher directional derivative and the @xmath0th iterated partial directional derivative , our goal is to establish a higher order chain rule in the style of @xcite for the former of these two .
[ t : higherchainrule ] for a composable pair @xmath44 and @xmath45 of functors of abelian categories , the @xmath0th directional derivative satisfies @xmath578 for the case @xmath527 , the third and fifth authors carried out a very careful computation verifying this theorem using only properties of linearization and cross effects .
a short paper containing this computation is in preparation .
we will explain how this chain rule can be derived as a natural consequence of structures associated to cartesian differential categories .
cockett and cruttwell introduce a more general notion of a _ tangent category _ @xcite and show that any cartesian differential category gives an example .
the key component of a tangent category structure is an endofunctor @xmath72 . in our case , @xmath72 is defined as follows .
[ def : tangent - functor ] there is a functor @xmath579 defined on objects by @xmath580 and on morphisms @xmath178 by @xmath581 where @xmath582 denotes the projection onto the second , or right - hand , factor .
that is @xmath583 importantly , theorem [ thm : diffcat](v ) implies that @xmath72 is _ functorial _ up to chain homotopy equivalence , as asserted in definition [ def : tangent - functor ] .
[ lem : tangent - functor ] for a composable pair @xmath44 and @xmath45 of functors of abelian categories , @xmath584 is chain homotopy equivalent to @xmath585 .
we use theorem [ thm : diffcat](v ) to show that @xmath72 is a functor up to chain homotopy equivalence . by definition , @xmath586 is given by while @xmath584 is given by by theorem [ thm : diffcat](v ) these outputs are chain homotopy equivalent : @xmath587 lemma [ lem : tangent - functor ] shows that @xmath72 is an endofunctor of @xmath69 . as an immediate corollary , its iterates @xmath73
are again functorial : that is , @xmath588 . since chain homotopy equivalences are respected by pre- and post - composition in @xmath69 , the composite functors @xmath589 are again chain homotopy equivalent , where @xmath590 is the projection onto the leftmost component and @xmath591 is a diagonal functor to be described below .
we claim that the top composite of is the functor @xmath592 , while the bottom composite is the functor appearing on the right - hand side of the chain homotopy equivalence of theorem [ t : higherchainrule ] .
once we make these identifications , which will be achieved by the following series of combinatorial lemmas , we will be able to conclude that theorem [ t : higherchainrule ] is a consequence of the functoriality of @xmath72 .
the diagonal functor @xmath593 is a special case of a `` reindexing functor '' in the sense of the following definition .
[ defn : reindex ] write @xmath594 for the set with @xmath0 elements , and let @xmath595 be any function .
the product category @xmath596 can be thought of as the category of functors from the discrete category @xmath597 to @xmath93 , from which perspective pre - composition with @xmath598 defines a _ reindexing functor _ @xmath599 .
explicitly , @xmath600 is the functor @xmath601 whose components @xmath602 project onto the @xmath315th factor .
diagonal functors are reindexing functors which come from surjections of sets .
importantly , reindexing functors commute with the directional derivative @xmath71 in the sense of the following result .
[ lem : nabla - and - diagonals ] suppose @xmath603 factors as up to chain homotopy equivalence , where @xmath604 is the reindexing functor associated to some function @xmath605 . then @xmath606 factors as up to chain homotopy equivalence let @xmath607 . by theorem [ thm : diffcat](v ) , @xmath608 . by theorem [ thm : diffcat](iii ) and
( iv ) , @xmath609 , so @xmath610 .
this observation allows us to make the relationship between @xmath611 and @xmath612 explicit , which we describe in the next lemma by examining the coordinates of @xmath613 one at a time . for convenience , we identify the cardinal @xmath614 with the set @xmath615 , the power set of @xmath597 , which has cardinality @xmath614 . under this identification
, @xmath616 is a collection of functors @xmath617 , indexed by elements @xmath542 of @xmath615 , i.e. , by subsets @xmath618 .
[ lem : tn - component ] for any @xmath618 , there is a chain homotopy equivalence where @xmath619 is the inclusion .
we prove this by induction on @xmath0 .
the case @xmath620 is given by the definition of @xmath72 , but we include it in detail here in order to begin to examine the definition of components of @xmath72 in terms of subsets @xmath542 of @xmath621 .
when @xmath620 , the power set @xmath622 contains the two elements @xmath623 and @xmath624 .
this corresponds to the two diagrams this in turn corresponds directly to the definition of @xmath625 where the second coordinate is indexed by @xmath623 and the first coordinate is indexed by @xmath626 .
in particular , @xmath627 when @xmath542 contains the `` top '' element @xmath628 , and @xmath629 when @xmath542 does not contain @xmath96 . for the inductive step , first consider those @xmath618 that do not contain the top element .
for these components , @xmath630 , and here @xmath631 is the map defined by restricting along the map @xmath632 induced by the inclusion of the first @xmath207 elements . by the inductive hypothesis , @xmath633 and the claim follows .
now consider the components indexed by a subset @xmath618 that does contain the top element @xmath634 . here
by the inductive hypothesis , @xmath636 , so applying lemma [ lem : nabla - and - diagonals ] @xmath637 the proof is completed by the observation that the map @xmath638 coincides with the map @xmath639 .
we now construct the function @xmath640 inducing the diagonal functor of . the function @xmath640 is defined by @xmath641 , where @xmath541 denotes the cardinality of @xmath542 .
equivalently , if we think of the elements of @xmath614 as @xmath0-tuples of binary digits ( the vertices of a unit @xmath0-cube ) , there is a function @xmath642 that counts the number of 1s .
this can be thought of as an order - preserving projection onto the diagonal of the cube .
recall remark [ r : diagrammatically ] , which decomposes the functor @xmath643 \to [ { \mathcal{b}}^{n+1 } , { \mathcal{a}}]$ ] as a composite @xmath523\ar[r]^-{\delta_{n-1 } } & [ { \mathcal{b}}^{n } , { \mathcal{a}}]\ar[r]^-{\nabla } & [ { \mathcal{b}}^{n}\times { \mathcal{b}}^n , { \mathcal{a}}]\ar[r]^-{l_n^ * } & [ { \mathcal{b}}^{n+1 } , { \mathcal{a}}],}\ ] ] first applying the directional derivative @xmath71 to @xmath644 and then restricting the variables along the diagonal functor @xmath645 .
this diagonal functor is also defined by precomposing with a certain function @xmath646 .
writing elements in the domain as ordered pairs of elements in @xmath647 and @xmath648 , the function @xmath649 adds the two coordinates .
[ lem : commuting - projections ] when @xmath650 is decomposed into its left - most coordinate paired with its remaining coordinates , the following diagram commutes for a binary @xmath0-tuple @xmath651 , by definition @xmath652 .
the lower composite sends @xmath651 first to @xmath653 , then to @xmath654 then to @xmath655 .
[ lem : redefinedeltan ] the higher order directional derivative @xmath508 is chain homotopy equivalent to the composite which restricts @xmath656 along the diagonal @xmath657 .
we prove this by induction on @xmath0 .
for the base case @xmath620 , @xmath658 for @xmath659 , consider the diagram below : composition from @xmath206 $ ] to @xmath660 $ ] along the top is @xmath661 , whereas composition along the bottom is @xmath508 , since @xmath662 .
the right triangle commutes by lemma [ lem : commuting - projections ] .
the middle square commutes up to chain homotopy equivalence by lemma [ lem : nabla - and - diagonals ] . by the inductive hypothesis
, the left triangle commutes up to chain homotopy equivalence , so the result follows by induction .
write @xmath663 ( respectively @xmath590 ) for the projection from a product category onto the product formed by its @xmath213 rightmost ( respectively leftmost ) variables , where the correct arity @xmath213 is determined by the context .
for example , the diagonal functor @xmath664 can be understood as the pairing of functors @xmath665 and @xmath666 , i.e. , @xmath667 let @xmath668 \to [ { \mathcal{c}}^{n+1},{\mathcal{b}}^{n+1}]\ ] ] be the functor defined by @xmath669 note that this expression appears as the argument for the @xmath670 appearing on the right - hand side of the chain homotopy equivalence of theorem [ t : higherchainrule ] .
the next lemma tells us that @xmath671 and @xmath672 are related by restricting the domain of @xmath673 and the codomain of @xmath674 along the diagonal functor @xmath591 .
[ lem : knd ] for @xmath318 , @xmath675 as functors @xmath676 .
again it is convenient to identify @xmath614 with @xmath677 .
it suffices to prove that these functors @xmath678 have chain homotopy equivalent components indexed by each subset @xmath679 .
lemma [ lem : tn - component ] tells us that @xmath680 . as functors from @xmath681 to @xmath682 , @xmath683 .
hence , @xmath684 the last chain homotopy equivalence by lemma [ lem : redefinedeltan ] .
on the other hand , immediately from the definition of @xmath674 , @xmath685 consider the diagram below .
the left and right squares commute up to chain homotopy equivalence by lemmas [ lem : knd ] and [ lem : redefinedeltan ] , respectively . using this diagram and lemma [ lem : tangent - functor ]
we see that @xmath686 by lemma [ lem : redefinedeltan ] , we have @xmath687 which completes the proof .
so far , we have considered the functors @xmath688 as analogs of the directional derivatives . in classical analysis ,
the derivative of a single variable function @xmath689 is a special case of the directional derivative in which the canonical ` direction ' is the positive direction , or the direction @xmath690 .
indeed , we see that setting @xmath691 in the definition of the directional derivative yields the following formula : @xmath692\ ] ] which recovers the usual single - variable derivative @xmath693 . in @xcite , this idea is extended from functions @xmath689 to functors @xmath694 by using generating objects of @xmath695 in place of the unit @xmath696 .
the canonical example of a generating object is the object @xmath52 in the category of @xmath52-modules @xmath697 , where @xmath52 is a commutative ring with unit .
the object @xmath52 is a generating object " in the sense that all other @xmath52-modules can be obtained from finitely generated free @xmath52-modules using free resolutions and colimits . in this section , we consider functors @xmath698 and use the generating object @xmath52 in @xmath697 to define a derivative @xmath699 .
we use this derivative to obtain analogs of the chain rules for compositions of functions involving at least one function of a single variable .
we note that these results could be generalized to functors @xmath700 by following the definition of @xmath701 in ( * ? ? ?
* definition 2.13 ) , however we have chosen to work with the category @xmath697 for the sake of exposition .
[ defn : derivatives ] ( * ? ? ?
* definition 2.13 ) for a functor @xmath702 and @xmath122 , the @xmath0th _ derivative _ of @xmath2 at @xmath29 is @xmath703 the following lemma indicates another way to define @xmath704 .
[ lem : derivatives ] for a functor @xmath705 , there is a chain homotopy equivalence @xmath706 theorem [ thm : restatedgoal ] expresses @xmath707 as a direct sum , indexed by partitions @xmath558 of @xmath539 , of terms @xmath708 . because @xmath709 vanishes if any of its `` direction '' variables is zero , the only partition that contributes a non - zero summand when @xmath710 is @xmath711 , the partition into one - element subsets .
thus @xmath712 as claimed .
as a consequence of theorem [ t : higherchainrule ] we obtain an analog of ( * ? ?
* theorem 2 ) , which is stated in the introduction in equation .
[ thm : hmy2 ] for a composable pair @xmath44 and @xmath713 of functors of abelian categories there is a chain homotopy equivalence @xmath714 by theorem [ t : higherchainrule ] @xmath715 applying lemma [ lem : derivatives ] this becomes @xmath716 combining theorems [ thm : hmy2 ] and [ thm : restatedgoal ] gives the next corollary , a fa di bruno - style characterization of the derivatives .
we note that arone and ching have obtained similar results for functors of spaces and spectra , by expressing such characterizations of derivatives in terms of composition products of symmetric sequences of derivatives ( see ( * ? ? ?
* theorem 0.2 ) and ( * ? ? ? * theorem 1.15 ) ) .
[ cor : faachainrule ] for a composable pair @xmath44 and @xmath713 of functors of abelian categories there is a chain homotopy equivalence @xmath717 where the sum is taken over all partitions @xmath397 of @xmath718 .
the derivatives @xmath719 play an important role in functor calculus in that they classify homogeneous degree @xmath0 functors , that is , degree @xmath0 functors @xmath2 with the property that @xmath720 for @xmath721 . for a functor of abelian categories @xmath44 ,
johnson - mccarthy define the @xmath0th layer of its taylor tower to be the homogeneous degree @xmath0 functor @xmath722 , i.e. , the mapping cone shifted down one degree .
this defines @xmath22 up to quasi - isomorphism .
the next few results relate our work to their characterizations of this functor , which hold up to quasi - isomorphism and are denoted with the symbol @xmath723 .
there is a quasi - isomorphism @xmath724 by ( * ? ? ?
* proposition 3.9 , corollary 5.11 ) .
the @xmath25-action is the natural action that permutes the variables of @xmath725 . in terms of the derivative of definition [ defn : derivatives ] , this yields the following corollary . for @xmath705
there is a quasi isomorphism : @xmath726 the objects @xmath727 are classifying objects for homogeneous degree @xmath0 functors by ( * ? ?
* theorem 5.13 ) .
thus , combining these results with theorem [ thm : hmy2 ] and corollary [ cor : faachainrule ] , we have the following formula for the homogeneous layers of a composition .
[ cor : layers ] for functors @xmath728 and @xmath155 , there is a quasi - isomorphism @xmath729 furthermore , there is a quasi - isomorphism @xmath730 where the sum is taken over all partitions @xmath397 of @xmath718 . a more thorough investigation of the relationship between corollary [ cor : layers ] , the classification of functors of @xmath731 , and modules over these is warranted
. this will be the subject of future work .
throughout this paper , and especially in section 4 and 5 , we have replaced many quasi - isomorphisms related to properties of polynomial approximation and linearization by explicit chain homotopy equivalences . in several cases ( especially proposition [ lem:5.7 ] ) these chain homotopy equivalences involved a bicomplex which we wished to retract onto a single row or column . in this appendix we establish a general criterion for doing so . in appendix b , we provide a proof of proposition [ lem:5.7 ] as an application .
let @xmath732 be a first - quadrant bicomplex .
contrary to the conventions in use elsewhere , to simplify in this appendix we assume that bicomplexes have anti - commutative squares , as this convention will simplify the signs . our aim is to prove : [ thm : chain - htpy - equiv ] let @xmath81 be a first - quadrant bicomplex so that every row except the zeroth row @xmath82 is contractible
. then the natural inclusion @xmath83 is a chain homotopy equivalence .
[ rmk : commutative - bicomplexes ] given a bicomplex with commuting squares , multiply all of the maps in every odd row by @xmath283 to obtain a bicomplex with anti - commuting squares .
this operation does not change the totalizations
. so theorem [ thm : chain - htpy - equiv ] applies .
we prove theorem [ thm : chain - htpy - equiv ] by constructing an explicit retraction in proposition [ prop : retraction ] and chain homotopy in proposition [ prop : chain - homotopy ] .
we first establish our notation .
explicitly , the bicomplex @xmath733 has : * horizontal differentials @xmath734 so that @xmath735 , and * vertical differentials @xmath736 so that @xmath737 , * so that @xmath738 ( squares anti - commute ) .
we suppose also that each row @xmath739 with @xmath740 is contractible : i.e. , there are * horizontal splittings @xmath741 so that @xmath742 . in this last equation
it is convenient to allow two additional maps @xmath743 and @xmath744 , both equal to zero .
that is , @xmath745 is the identity map @xmath690 .
the following lemma records some commutativity relations that follow from the defining relations @xmath746 the proof is a straightforward exercise , left to the reader . [ lem : commutativities ] for any @xmath252 : 1 .
@xmath747 2 .
@xmath748 3 .
@xmath749 recall the _ total complex _
@xmath750 of @xmath732 is the chain complex with @xmath751 ( where `` @xmath752 '' is less - cluttered notation for `` @xmath97 '' ) and with differential given by the matrix see @xcite for an explanation of the matrix notation for a map between finite direct sums .
[ prop : retraction ] the evident inclusion @xmath753 of chain complexes admits a retraction @xmath754 defined in degree @xmath0 by here @xmath755 . to show that this is the chain map
, we must verify that for each @xmath756 i.e. , that @xmath757 commutativity is obvious for the remaining case @xmath758 .
this is lemma [ lem : commutativities](i ) .
[ prop : chain - homotopy ] the composite map @xmath759 is chain homotopic to the identity via the chain homotopy @xmath760 defined in degree @xmath0 by writing @xmath761 for the total complex differential , we must verify that @xmath762 is the matrix i.e. , that @xmath763 for each pair @xmath764 , we must verify that the component of @xmath765 from @xmath766 to @xmath767 agrees with the matrix entry in the @xmath213th column and @xmath339th row , counting from the right and from the bottom .
we do this in several cases . for each case
, we display a square diagram whose horizontal arrows are components of @xmath263 and vertical arrows are components of the total complex differential @xmath768 .
the upper - right composite is the component of the map @xmath769 and the lower - left composite is the component of the map @xmath770 whose sum we are computing .
we begin with the entries representing the maps @xmath771 along the bottom row . for @xmath772
, we must show that the sum of the two maps is @xmath773 , which is indeed the case .
( here and elsewhere , we allow ourselves to include the term @xmath774 . ) when @xmath758 , the components of the chain homotopy @xmath263 are zero , so clearly their sum is the desired zero map @xmath775 .
we next verify the diagonal entires @xmath776 for the cases @xmath756 .
here we must show that the sum of the maps is @xmath777 , which is so because @xmath778 .
we next verify the upper diagonal entries @xmath779 for @xmath780 .
here we must show that the sum of the maps is zero , which is obvious .
it remains to verify the lower diagonal entries above the bottom row .
here we wish to show that for @xmath781 the sum of the maps is the zero map @xmath782 . here
the sum is @xmath783@xmath784 which vanishes by lemma [ lem : commutativities](iii ) .
we now apply theorem [ thm : chain - htpy - equiv ] to prove : in the kleisli category , the composition of @xmath318 with @xmath319 is defined by prolonging the functor @xmath78 to a functor @xmath792 via the dold - kan equivalence and then applying the totalization @xmath202 .
when @xmath2 is strictly reduced , there is a functor @xmath793 defined by simply applying the functor directly to each term in the chain complex .
when @xmath2 is also linear , ( * ? ?
* lemma 5.4 ) show that the two prolongations @xmath794 and @xmath795 are quasi-isomorphic.the
following lemma shows that when @xmath2 is a linearization @xmath370 of some functor @xmath796 , in which case lemma [ lem : d1-linearity](i ) shows that @xmath2 is strictly reduced and linear in the sense of definition [ defn : linear ] ( preserving finite direct sums up to natural chain homotopy equivalence ) , these procedures are in fact chain homotopy equivalent .
we will use this in the proof of proposition [ lem:5.7 ] to construct the composite @xmath158 using the simpler form of the prolongation .
[ lem : lastlemma ] let @xmath797 be a functor and @xmath798 .
let @xmath799 be the prolongation of @xmath2 via the dold - kan equivalence .
let @xmath800 be the functor obtained by applying @xmath2 degreewise to a chain complex .
then @xmath801 and @xmath802 are chain homotopy equivalent .
we note that the definition of @xmath803 only makes sense when @xmath2 is strictly reduced .
otherwise , applying @xmath2 degreewise to the objects and maps in a chain complex may not produce a chain complex , as @xmath804 may no longer be @xmath96 .
write @xmath805 and @xmath806 for the functors in the dold - kan equivalence .
the functor @xmath807 is naturally chain homotopy equivalent to another functor @xmath808 with a simpler definition : for a simplicial object @xmath809 , @xmath810 with differentials defined to be the alternating sum of the face maps ; see @xcite and ( * ? ? ?
* iii.2.4 ) for definitions of these functors and the chain homotopy equivalence .
let @xmath811 be the functor defined by post - composition with @xmath78 . recall that @xmath812 defines the prolongation . for a chain complex @xmath813
, there is a canonical comparison morphism @xmath814 in @xmath815 that looks like @xmath816\twoheadrightarrow [ k]}f(x_k)\rightarrow f\left ( \bigoplus_{[n]\twoheadrightarrow [ k ] } x_k\right)\ ] ] in simplicial degree @xmath0 where the sums are over all surjections @xmath817\twoheadrightarrow[k]$ ] in @xmath818 .
since @xmath2 is linear and strictly reduced , this is a chain homotopy equivalence as in definition [ defn : linear ] . moreover , since @xmath798 , the chain homotopy constructed in the proof of proposition [ prop : poly - facts ] is natural , commuting with the simplicial maps .
composing with the functor @xmath819 , this defines a canonical map of bicomplexes @xmath820 that is a chain homotopy equivalence in each row .
naturality implies that these chain homotopy equivalences assemble into a chain homotopy equivalence of bicomplexes in the sense of @xcite , where the vertical components of the chain homotopy are equal to zero .
this is enough to ensure that @xmath821 is a chain homotopy equivalence .
since we are safe to assume that both @xmath2 and @xmath3 are reduced , the chain complexes @xmath157 , @xmath181 , and @xmath824 are defined as comonad resolutions for the second cross effect comonad @xmath825 . for readability , we simplify our notation and write `` @xmath819 '' for the comonad @xmath825 and `` @xmath826 '' for the adjoint functors of corollary [ cor : comonad ] defining the comonad @xmath827 .
recall @xmath828 is the diagonal functor , which has an important property : in particular , @xmath829 and @xmath830 .
the left adjoint @xmath828 also preserves direct sums of functors as does the comonad @xmath827 , by proposition [ p : cr is exact ] .
we are now prepared to embark upon the proof . our strategy , following the proof of ( * ?
* lemma 5.7 ) , is to show that the natural inclusions of @xmath159 and @xmath158 into the double totalization of a first octant tricomplex @xmath831 are chain homotopy equivalences . consider the tricomplex @xmath831 defined by @xmath832 and its levelwise totalization @xmath833 the @xmath221 column of @xmath834 is @xmath824 while the @xmath835 row is chain homotopy equivalent to @xmath158 by lemma [ lem : lastlemma ] .
the @xmath836th row is @xmath837 .
lemma [ lem : diagonal - commutativity ] and the fact that @xmath828 preserves sums implies that the columns indexed by each @xmath838 are each the comonad resolution chain complexes for @xmath839 applied to a functor of the form @xmath840 for some functor @xmath190 .
in particular , each column except for the 0th is contractible , using the unit @xmath842 of the adjoint @xmath843 .
applying theorem [ thm : chain - htpy - equiv ] it follows immediately that the canonical inclusion @xmath844 is a chain homotopy equivalence . by theorem [ thm : chain - htpy - equiv ] , to show that @xmath845 is a chain homotopy equivalence it suffice to prove that the rows @xmath846 of @xmath841 are also contractible for each @xmath847 . by exactness of @xmath819 , it suffices to prove that the first row @xmath848 is contractible . by proposition [ p : pnproperties ] ,
the chain complexes @xmath849 and @xmath850 are contractible .
it follows that the inclusions @xmath851 which admit @xmath849 and @xmath850 as direct sum complements , are chain homotopy equivalences . by lemma [ lem : che - comp ] ,
these chain homotopy equivalences are preserved by pre- and post - composition . in particular
, there is a composite chain homotopy equivalence @xmath852 it follows that the direct sum complement @xmath853 is contractible . for sake of completeness
, we record an explicit construction of the contracting homotopy as the following lemma .
so that @xmath858 the component of the map @xmath859 from @xmath860 to @xmath860 is the zero map .
thus , the component from @xmath860 to @xmath860 of the relation asserts that @xmath861 , where @xmath862 is the differential in @xmath855 .
thus , the component @xmath863 of @xmath864 provides the desired contracting homotopy .
m. basterra , k. bauer , a. beaudry , r. eldred , b. johnson , m. merling , s. yeakel , unbased calculus for functors to chain complexes , _ women in topology : collaborations in homotopy theory , contemp .
_ , * 641 * , 2015 , pp 2948 . | in this paper , we consider _ abelian functor calculus _ , the calculus of functors of abelian categories established by the second author and mccarthy .
we carefully construct a category of abelian categories and suitably homotopically defined functors , and show that this category , equipped with the directional derivative , is a _
cartesian differential category _ in the sense of blute , cockett , and seely .
this provides an abstract framework that makes certain analogies between classical and functor calculus explicit .
inspired by huang , marcantognini , and young s chain rule for higher order directional derivatives of functions , we define a higher order directional derivative for functors of abelian categories .
we show that our higher order directional derivative is related to the iterated partial directional derivatives of the second author and mccarthy by a fa di bruno style formula .
we obtain a higher order chain rule for our directional derivatives using a feature of the cartesian differential category structure , and with this provide a formulation for the @xmath0th layers of the taylor tower of a composition of functors @xmath1 in terms of the derivatives and directional derivatives of @xmath2 and @xmath3 , reminiscent of similar formulations for functors of spaces or spectra by arone and ching . throughout
, we provide explicit chain homotopy equivalences that tighten previously established quasi - isomorphisms for properties of abelian functor calculus . |
relativistic heavy - ion collisions are studied with the goal of creating hot and dense hadronic matter and to investigate the nuclear phase diagram at high temperatures and densities , including the expected phase transition to a color deconfined quark - gluon plasma .
but even if the energy deposited in the reaction zone is quickly randomized and the fireball constituents reach an approximate state of local thermal equilibrium , a simple connection between heavy - ion observables and the phase diagram is still not easy : the pressure generated by the thermalization process blows the fireball apart , causing a strong time dependence of its thermodynamic conditions which is difficult to unfold from the experimental observations .
there are therefore two fundamental issues to be solved before one can extract information on the nuclear phase diagram from heavy - ion experiments : ( 1 ) to what degree does the fireball approach local thermal equilibrium ? ( 2 ) which observables are sensitive to which stage(s ) of its dynamical evolution , and which is the most reliable procedure for extracting the corresponding thermodynamic information ? combining microscopic models for the dynamical fireball evolution with macroscopic thermal models for the analysis of heavy - ion data , significant progress has been recently made in answering both of these questions .
crucial for this achievement was the dramatically improved quantity and quality of hadron production data from the analysis of collisions between very heavy nuclei ( au+au , pb+pb ) from sis to sps .
1 shows a compilation by cleymans and redlich @xcite of hadronic freeze - out points in the nuclear phase diagram from various collision systems and beam energies .
the upper set of points , parametrized by a constant average energy per particle @xmath0=1 gev @xcite , is obtained from measured hadron yields .
they indicate the average thermodynamic conditions at _ chemical freeze - out _ when the hadron abundances stopped evolving .
the lower set of points , compared with lines of constant energy and particle density @xcite , is obtained from analyses of hadron momentum spectra and/or two - particle momentum correlations .
they indicate _ thermal freeze - out _ ,
i.e. the decoupling of the momentum distributions .
the chemical and thermal freeze - out points at the sps and ags , respectively , are connected by isentropic expansion trajectories with @xmath1 36 - 38 for the sps and 12 - 14 for the ags .
[ f1 ] my first goal is a critical discussion of how these freeze - out parameters were extracted from the data and how reliable fig .
following that is a more detailed study of the fireball properties at chemical freeze - out , taking into account additional information not contained in fig . 1 , and a discussion of a consistent dynamical picture which can explain fig . 1 .
my main conclusion , based on a chain of arguments developed and sharpened over the last few years @xcite , is given in the abstract ; similar conclusions were reached and recently publicized by r. stock @xcite and are also found in e. shuryak s talk @xcite .
let me begin with a discussion of the thermal freeze - out points . freeze - out marks the transition from a strongly coupled system , which evolves from one state of local thermal equilibrium to another , to a weakly coupled one of essentially free - streaming particles .
if this transition happens quickly enough , the thermal momentum distributions ( superimposed by collective expansion flow ) are frozen in , and the temperature and collective flow velocity at the transition `` point '' can be extracted from the measured momentum spectra . in high energy heavy - ion collisions
the freeze - out process is triggered dynamically by the accelerating transverse expansion and the very rapid growth of the mean free paths as a result of the fast dilution of the matter @xcite .
idealizing the kinetic freeze - out process by a single point in the phase diagram is therefore not an entirely unreasonable procedure . as in the big bang , the observed momentum spectra mix the thermal information with the collective dynamics of the system . in the big bang ,
the observed microwave background radiation has a bose - einstein energy spectrum with an `` effective temperature '' ( inverse slope ) which is _ redshifted _ by cosmological expansion down from the original freeze - out temperature of about 3000 k to an observed value of only 2.7 k. in the little bang , where we observe the _ thermal hadron radiation _ from the outside , the transverse momentum spectra are _ blueshifted _ by the collective transverse motion towards the observer .
simple approximate expressions which capture this effect are , where @xmath2 is the radial velocity at the rms radius @xmath3 .
the analogous formula in @xcite lacks the factor @xmath4 in the second term since it uses the radial velocity at @xmath5 .
[ fn1 ] ] @xmath6 ( which applies for @xmath7 ) and @xmath8 ( which is good for @xmath9 ) . for a given species ( fixed @xmath10 )
the measured slope of the spectrum is thus ambiguous : temperature and flow can not be separated .
the ambiguity can be lifted in two ways : ( i ) one performs a simultaneous fit of the @xmath11-spectra of hadrons with different rest masses , thereby exploiting the mass dependence in the first of these two expressions @xcite .
this makes the implicit assumption that thermal freeze - out happens simultaneously for all particle species . for au+au collisions at the ags this works well and gives @xmath12 mev and @xmath13 at midrapidity @xcite . or ( ii ) one concentrates on a single particle species and correlates ( as described in detail by u. wiedemann @xcite ) their spectra with their two - particle bose - einstein correlations .
the @xmath14-dependence of the transverse hbt radius parameter accounts for different transverse density and flow profiles ; @xmath15 for the case described in fn .
[ fn2 ] ] @xmath16 then provides an orthogonal correlation between temperature and flow , allowing for their separation . for pions from pb+pb collisions at the sps
this leads again to kinetic freeze - out temperatures of 90100 mev and average tranverse flow velocities of 0.50.55@xmath17 ( perhaps even somewhat higher at midrapidity ) @xcite .
this fixes the position of the freeze - out point along the @xmath18-axis in fig . 1 , but what about @xmath19 ? since chemical equilibrium is already broken earlier , at @xmath20 mev ( see below ) , @xmath19 is , strictly speaking , not well - defined . in order to still be able to associate with kinetic freeze - out a point in the phase - diagram
one commonly adjusts @xmath19 in such a way that the deviations between the observed particle ratios and their chemical equilibrium values at @xmath21 are minimized .
this is acceptable if the deviations are small ; in practice they can approach a factor 2 or so .
this clearly causes irreducible systematic uncertainties in the baryon chemical potential at thermal freeze - out which are usually not evaluated and are not included in the horizontal error bars in fig . 1 ( where given ) .
let s nonetheless accept these thermal freeze - out parameters and now ask the question : _ how did the system get there ?
_ does the implied picture of a rapidly expanding , locally thermalized fireball , _ the little bang _ , make sense ?
these questions can be studied by microscopic kinetic simulations of rqmd , urqmd and hsd @xcite type ; even if they do not include quark - gluon degrees of freedom during the very dense initial stages and thus may parametrize the initial hadron production incorrectly ( see below ) , they can be used to explore the effects of scattering among the hadrons before kinetic freeze - out and the evolution of collective flow .
a detailed study of thermalization by rescattering was recently performed by the urqmd group for au+au and pb+pb collisions from ags to sps energies @xcite ( fig . 2 ) .
after an initial non - equilibrium stage lasting for about 810 fm/@xmath17 these systems reach a state of approximate local thermal equilibrium which expands and cools at roughly constant entropy for another 10 fm/@xmath17 before decoupling . during the adiabatic expansion stage strong collective flow builds up .
thermalization is driven by intense elastic rescattering , dominated by resonances ( e.g. @xmath22 ) ; inelastic processes are much rarer and lead only to minor changes in the chemical composition of the fireball @xcite . as a result ,
significant deviations from chemical equilibrium occur which increase with time ; most importantly , at thermal freeze - out one sees a large pion excess which can only partially be accounted for by the initial string fragmentation process @xcite .
remarkably , these deviations from chemical equilibrium produce very little entropy @xcite .
the @xmath23 values extracted from the urqmd simulations agree with those from the thermal model analysis of the data ( cf . figs . 1 and 2 ) . [ f2 ] hadron momentum spectra and two - particle correlations thus provide strong evidence for the existence of the little bang : thermal hadron radiation with @xmath24 mev and strong 3-dimensional ( `` hubble - like '' ) expansion with transverse flow velocities @xmath25 ( and even larger longitudinal ones @xcite ) .
these two observations play a similar role here as the discovery of hubble expansion and the cosmic microwave radiation played for the big bang .
but is there also a heavy - ion analogy to primordial big bang nucleosynthesis ? in the following section
i will argue that we have indeed evidence for `` primordial hadrosynthesis '' in the little bang .
chemical reactions , which exploit small inelastic fractions of the total cross section , are typically much slower than the ( resonance dominated ) elastic processes .
one thus expects chemical freeze - out to occur _ before _ thermal freeze - out @xcite ( @xmath26 ) but on an expansion trajectory with roughly the same entropy per baryon @xmath23 .
1 suggests that this is indeed the case .
this is analogous to the big bang where nucleosynthesis happened after about 3 minutes at @xmath27 kev whereas the microwave background decoupled much later , after about 300000 years at @xmath28 ev .
the much smaller difference between the two decoupling temperatures in the little bang is mainly due to its much ( about 18 orders of magnitude ) faster expansion rate . before discussing implications of the chemical freeze - out points in fig .
1 i first explain how they were obtained .
can thermal models be used to analyze chemical freeze - out ?
i discussed this question in some detail last year in padova @xcite and thus will be short here .
the first difficulty arises from the collective expansion which strongly affects the shape of the @xmath11- and @xmath29-spectra , in a way which depends on the particle mass . however , many experiments measure the particle yields only in small windows of @xmath11 and @xmath29 .
a chemical analysis of particle ratios from such experiments depends very strongly on model assumptions about the fireball dynamics @xcite .
static fireball fits yield chemical freeze - out parameters which are quite sensitive to the rapidity window covered by the data @xcite .
flow effects drop out , however , from @xmath30-integrated particle ratios as long as freeze - out occurs at constant @xmath18 and @xmath31 .
@xmath30 yields thus minimize the sensitivity to the collective fireball dynamics and are preferable for thermal model analyses .
[ f3 ] 11.7 cm -0.8 cm of course , dynamic systems never freeze out at constant @xmath18 and @xmath31 . while the steep @xmath18-dependence of the particle densities ( which determine the local scattering rates and control freeze - out @xcite ) prohibits strong temperature variations across the freeze - out surface @xcite , incomplete baryon number stopping causes at higher energies significant longitudinal variations of @xmath19 @xcite .
a global thermal fit replaces the freeze - out distributions @xmath32 ,
@xmath33 by average values @xmath34 , @xmath35 .
a recent study by sollfrank @xcite , in which he performed a global thermal fit to particle yields from a hydrodynamic calculation , showed that after optimizing @xmath36 and @xmath37 the thermal model predicted yields which differed by up to 15% from the actual ones , although hydrodynamic simulations implement perfect ( local ) chemical equilibrium by construction . _
thermal model fits can thus never be expected to be perfect _ ; without detailed dynamical assumptions local variations of the thermal parameters in the real collision can never be fully absorbed by the model .
discrepancies between model and data at the 15 - 30% level are inside the systematic uncertainty band of the thermal model approach . while being impressed by how thermal models can reproduce the measured particle ratios at this level of accuracy over up to 3 orders of magnitude ( see fig .
3 ) , i am thus deeply suspicious of `` perfect '' thermal model fits .
a second lesson to be learned from the exercise in @xcite is that the @xmath38/d.o.f .
resulting from such a fit is not very useful as an absolute measure for the quality of the fit : since discrepancies between the real yields and the predictions from the global thermal model can not be avoided , @xmath38/d.o.f .
becomes larger and larger as the data become more and more accurate .
while @xmath38 minimization can still be used to identify the optimal model parameters within a given model , one should be very careful in using the absolute value of @xmath38/d.o.f . to judge the relative quality of different model fits @xcite .
let me note that the _ ideal system for thermal model fits of particle ratios _ will be provided by heavy - ion collisions at rhic and lhc : hadron formation will happen at the confinement transition , and near midrapidity the baryon density is so low that @xmath39 is nearly independent of @xmath19 .
replacing @xmath40 by @xmath41 will then be an excellent approximation .
due to longitudinal boost - invariance near midrapidity , knowledge of @xmath42 will be good enough for a reliable chemical analysis .
finally , transverse flow can only be stronger at rhic and lhc than at the sps , so freeze - out will happen even more quickly after hadronization , strengthening the primordial character of the observed particle ratios .
[ f4 ] these expectations are borne out by a recent analysis by bass and dumitru @xcite who combined a hydrodynamic description of the dense early stage with a urqmd simulation of the late hadronic stage . fig .
4 ( bottom ) shows that indeed chemical freeze - out occurs quickly after hadronization : the yields at hadronization ( squares ) and after the last elastic scattering ( circles ) differ by less than 30% , in spite of many collisions in between @xcite . the fit of pb+pb data in fig . 3 @xcite yields a chemical freeze - out temperature @xmath43 mev at full strangeness saturation ( @xmath44=1 ) . in @xcite
the same data are fit with @xmath45 mev and a strongly oversaturated strange phase - space ( @xmath46 ) .
the authors also allow for oversaturation of the light quarks and find @xmath47 which allows to absorb the large pion multiplicity at a low value of @xmath48 .
this `` chemical non - equilibrium '' @xcite fit underpredicts @xmath49 by 40% and @xmath50 by 60% , a problem which disappears at @xmath51 mev ( fig . 3 ) .
more importantly , since @xmath52 , the freeze - out parameters in @xcite imply a very large pion chemical potential @xmath53mev@xmath54 ; this invalidates the assumption @xcite that bose statistics for pions can be accounted for by considering only the first correction to the boltzmann term .
hence , while the authors of @xcite prefer their fit on the basis of a low @xmath38/d.o.f .
, it has systematic uncertainties which far exceed the statistical errors given in @xcite . having established the location in the phase diagram where chemical freeze - out occurs , we should again ask : _ how did the system get there ? _ since @xmath48 turns out to be very close to the predicted critical value for the hadronization phase transition , there is clearly no time between hadron formation and chemical freeze - out for kinetic equilibration of the hadron abundances by inelastic hadronic rescattering @xcite .
the observed hadronic chemical equilibrium at @xmath48 must therefore be _ pre - established _ : it reflects a statistical occupation of the hadronic phase - space , following the principle of maximum entropy , by the hadronization process @xcite .
hadrons form from a prehadronic stage by filling each available phase - space cell with equal probability , subject only to the constraints of energy , baryon number and strangeness conservation ( the latter includes a possible overall suppression of strangeness ) . afterwards , the chemical composition decouples essentially immediately , without major modifications by hadronic rescattering . the parameter @xmath48 is thus _ not a hadronic temperature in the usual sense _ , i.e. not a result of hadronic kinetic equilibration . in the maximum entropy spirit
it should be interpreted as a lagrange multiplier which regulates the hadron abundances in accordance with the conservation laws and is directly related to the critical energy density at which hadronization can proceed .
@xmath55 mev translates into @xmath56 gev/@xmath57 .
this also explains naturally becattini s observation @xcite of hadronic chemical equilibrium at the same value of @xmath48 in @xmath58 , @xmath59 and @xmath60 collisions at essentially all collision energies , although there hadronic final state interactions are completely absent .
whether the constituents of the prehadronic stage themselves thermalize before or during hadronization is a question which final hadron abundances can not answer ; as likely as their thermalization may appear , it is not necessary for an explanation of the observed phenomena .
the concept of statistical hadron formation from a pre - existing state of color - deconfined , completely uncorrelated quarks and antiquarks is supported by a recent analysis of bialas @xcite which follows similar earlier arguments by rafelski @xcite but formulates them more generally such that they do not require thermalization .
bialas points out that by considering baryon / antibaryon ( or generally particle / antiparticle ) ratios , the unknown effects on hadronization from the internal hadron structure drop out and one can check directly whether the observed hadron abundances can be fully understood by just counting their conserved quantum numbers ( carried by their valence quarks ) , or whether additional correlations among the quarks exist .
he finds @xcite that in s+s and pb+pb collisions at the sps the former is true while hadron production yields in p+pb collisions point to correlations among the quarks .
the one decisive feature which distinguishes heavy - ion from elementary particle collisions is the strangeness content in the hadronic final state : the global strangeness fraction of the produced quark - antiquark pairs , @xmath61 , is about 2 times higher in nuclear collisions @xcite .
this can not be reproduced by hadronic rescattering models @xcite and must thus be a feature of the prehadronic state . here
i would like to discuss in more detail the specific enhancement factors for @xmath62 , and @xmath63 reported recently and during this meeting @xcite .
[ f5 ] fig .
5 shows that the relative strangeness enhancement between pb+pb and p+be collisions is the stronger the more strange quarks the hadron contains .
this is perfectly consistent with the above picture of statistical hadronization : disregarding other phase - space constraints , an @xmath64 , for example , which contains 3 strange quarks , is expected to be enhanced by a factor @xmath65 if strange quarks are enhanced by a factor 2 . on the other hand , this pattern contradicts expectations from final state hadronic rescattering : since hadrons with more strange quarks are heavier and strangeness must be created in pairs , the production of stranger particles is suppressed by increasingly higher thresholds .
an interesting observation from fig .
5 is the apparent centrality independence of the specific strangeness enhancement factors : the enhancement appears to be already fully established in semiperipheral pb+pb collisions with about 100 participating nucleons .
in fact , the global enhancement by a factor 2 was already seen in s+s collisions by na35 @xcite .
since it must be a prehadronic feature , but the lifetime of the prehadronic stage is shorter for smaller collision systems , this points to a new fast strangeness production mechanism in the prehadronic stage .
exactly this was predicted for the quark - gluon plasma @xcite . at this meeting
we saw new data on the centrality dependence of hadron yields @xcite .
unfortunately , different centrality measures and prescriptions to determine @xmath66 have been used .
this needs clarification before the pattern in fig .
5 can be considered confirmed .
much recent effort went into trying to explain these observations within microscopic simulations based on string breaking followed by hadronic rescattering . _ all such attempts failed . _
venus and rqmd give more strangeness enhancement for more central collisions @xcite , however not from hadronic rescattering , but mostly from the non - linear rise of the formation probability for quark matter droplets and color ropes .
strangeness enhancement is thus put in as an initial condition ; unlike fig . 5 it rises monotonically with @xmath66 .
@xcite uses baryon junction loops to enhance strange baryon production near midrapidity .
again this puts the enhancement into the initial conditions .
the measured @xmath66-dependence is not reproduced .
the model also disagrees with the observed pattern @xmath68 ( which the statistical hadronization picture @xcite explains nicely ) .
finally , the `` improved dual parton model '' @xcite , which gets some fraction of the enhancement in the initial state from `` diquark - breaking collisions '' and claims to obtain an even larger additional enhancement from hadronic final state interactions , suffers from a severe violation of detailed balance : it only includes inelastic channels ( like @xmath69 ) which increase multistrange baryons but neglects the ( at least equally important @xcite ) strangeness exchange processes ( like @xmath70 ) which destroy them . consequently it also fails to reproduce the apparent saturation of the enhancement factors seen in fig .
the observed strangeness enhancement pattern thus can not be generated by hadronic final state interactions , but must be put in at the beginning of the hadronic stage .
no working model which does so in agreement with the data is known , except for conceptually simplest one , the statistical hadronization model .
the analysis of soft hadron production data at the sps indicates that hadron formation proceeds by statistical hadronization from a prehadronic state of uncorrelated ( color - deconfined ) quarks .
this leads to pre - established apparent chemical equilibrium among the formed hadrons at the confinement temperature @xmath39 ; it is not caused by kinetic equilibration through hadronic rescattering .
after hadronization the hadron abundances freeze out more or less immediately .
the chemical freeze - out temperature thus coincides with the critical temperature , @xmath71 - 180 mev , corresponding to a critical energy density @xmath72 gev/@xmath57 as predicted by lattice qcd .
the prehadronic state in @xmath73 ( @xmath74 ) collisions contains about twice more strangeness than in @xmath58 and @xmath59 collisions .
this strangeness enhancement appears to be already fully established in nuclear collisions with 60 or more participant nucleons and can not be generated by hadronic final state interactions .
this suggests a fast @xmath75 creation mechanism in the prehadronic stage , as predicted for a quark - gluon plasma .
a clear hierarchy between chemical ( @xmath76180 mev ) and thermal ( @xmath77100 mev ) freeze - out is observed in pb+pb collisions at the sps ; the gap is somewhat smaller ( @xmath78140 mev vs. @xmath79 mev ) at the ags . in both cases thermal decoupling
is accompanied by strong radial collective flow .
the smaller inverse slopes of the @xmath64 @xmath11-spectra suggest that a considerable fraction ( but probably not all ) of this flow is generated by strong elastic hadronic rescattering after hadronization @xcite .
_ acknowledgements : _ i thank s. bass , j. cleymans and r. lietava for providing me with figs . 1 , 4 , and 5 .
several fruitful discussions with r. stock are gratefully acknowledged .
a comment by zhangbu xu led to the clarification presented in footnote 1 .
99 j. cleymans , k. redlich , phys .
81 ( 1998 ) 5284 ; and nucl - th/9903063 .
f. becattini , u. heinz , z. phys .
c76 , ( 1997 ) 269 ; u. heinz , nucl .
a638 ( 1998 ) 357c ; j. phys .
g25 ( 1999 ) 263 ; and hep - ph/9902424 .
r. stock , prog . part .
42 ( 1999 ) 295 ; phys .
b456 ( 1999 ) 277 .
e. shuryak , this volume ( hep - ph/9906443 ) .
e. schnedermann , u. heinz , phys .
c50 ( 1994 ) 1675 .
t. csrg , b. lrstad , phys .
c54 ( 1996 ) 1390 ; r. scheibl , u. heinz , phys .
c59 ( 1999 ) 1585 .
n. xu et al . ( na44 coll . ) , nucl .
a610 ( 1999 ) 175c .
h. dobler , j. sollfrank , u. heinz , phys .
b457 ( 1999 ) 353 .
wiedemann , this volume , and references therein ( nucl - th/9907048 ) .
h. appelshuser et al .
( na49 coll . ) , eur .
j. c2 ( 1998 ) 661 .
see contributions by h. sorge , m. bleicher , and w. cassing in this volume .
l. bravina et al . , j. phys .
g25 ( 1999 ) 351 ; hep - ph/9906548 ; phys .
b , in press . s. bass , a. dumitru et al .
, nucl - th/9902062 . s. soff et al . , nucl - th/9907026 .
u. heinz , nucl .
a566 ( 1994 ) 563c ; nato asi series b346 ( 1995 ) 413 . c. slotta , j. sollfrank , u. heinz , aip conf .
( 1995 ) 462 .
j. sollfrank , u. heinz , h. sorge , n. xu , phys .
c59 ( 1999 ) 1637 .
j. bondorf , s.i.a .
garpman , j. zimnyi , nucl .
a296 ( 1978 ) 320 .
e. schnedermann , j. sollfrank , u. heinz , nato asi series b303 ( 1993 ) 175 ; u. mayer , u. heinz , phys .
c56 ( 1997 ) 439 .
j. sollfrank , eur .
j. c9 ( 1999 ) 159 .
p. braun - munzinger , i. heppe , j. stachel , nucl - th/9903010 .
j. rafelski , j. letessier , nucl - th/9903018 .
f. becattini , z. phys .
c69 ( 1996 ) 485 .
j. zimnyi , t. bir , p. lvai , hep - ph/9904501 .
a. bialas , phys .
lett . b442
( 1998 ) 449 .
j. rafelski , phys .
b262 ( 1991 ) 333 .
see discussion by g. odyniec , nucl .
a638 ( 1998 ) 135c , and references therein .
f. becattini , m. gadzicki , j. sollfrank , eur .
j. c5 ( 1998 ) 143 .
e. andersen et al .
( wa97 coll . ) , phys .
lett . b433
( 1998 ) 209 ; r. lietava et al .
( wa97 coll . ) , j. phys .
g25 ( 1999 ) 181 ( for the correct fig . 7
in the same volume ) ; h. appelshuser et al .
( na49 coll . ) , phys .
lett . b444
( 1998 ) 523 .
see f. antinori , d. elia ( wa97 coll . ) ; f. sikler , c. hhne ( na49 coll . ) ; and n. willis ( na50 coll . ) , this volume .
j. rafelski , b. mller , phys .
48 ( 1982 ) 1066 . f. antinori et al . , eur .
j. c , in press ; r. caliandro , private communication . s.e .
vance , m. gyulassy , nucl - th/9901009 .
a. capella , c.a .
salgado , hep - ph/9903414 .
p. koch , b. mller , j. rafelski , phys .
142 ( 1986 ) 167 , fig . 5.2 .
h. van hecke , h. sorge , n. xu , phys .
81 ( 1998 ) 5764 . | the present status of soft hadron production in high energy heavy - ion collisions is summarized . in spite of strong evidence for extensive dynamical evolution and collective expansion of the fireball before freeze - out
i argue that its chemical composition is hardly changed by hadronic final state interactions .
the measured hadron yields thus reflect the primordial conditions at hadronization .
the observed production pattern is consistent with statistical hadronization at the hagedorn temperature from a state of uncorrelated , color deconfined quarks and antiquarks , but requires non - trivial chemical evolution of the fireball in a prehadronic ( presumably qgp ) stage _ before _ hadron formation . |
with the progress in high energy physics , it is obvious that an automate calculation of the physical processes by the perturbative quantum field theory becomes very useful and important .
there are many projects for this purpose , such as the grace@xcite , comphep@xcite , feyart&feycal@xcite , diana @xcite and several others .
fdc(feynman diagram calculation ) is one of these projects .
it was 1992 when we started to write a program for manipulations of one - loop amplitudes @xcite .
now a more developed program can be used to automatically deal with any process at tree level , and since then many applications have been made .
the paper is organized as follows . in section 2
we outline the basic parts of the fdc .
automatic deduction of the feynman rules for first principle model is presented and especially the procedure for the super - symmetry model is introduced in section 3 .
section 4 is devoted to the rate evaluation of multi - final - particle processes and building the corresponding event generators . in section 5 , a few sub - projects for fdc applications are displayed .
finally a summary is given in section 6 .
the main parts of the fdc are shown in fig.1 .
its source programs are mainly written in the reduce and rlisp , and there is also a fortran library which supplies necessary basic functions .
the feynman diagram plotter is written in c++ .
the first two parts _ gmodel and diag _ @xcite were developed in 1993 , where _ gmodel _ is used to deduce feynman rules for the standard model and _ diag _ is a feynman diagram generator which generates feynman diagrams to any perturbative orders .
_ psdraw _ was constructed to draw the feynman diagrams from the output of _ diag _ into a ps file ( up to two loops ) automatically in 1994 .
_ amp _ was finished in 1995 to treat the amplitude square manipulation and related fortran source generation for tree processes , much improvement had been made and other two optimized methods were introduced later . to properly handle the cancellation among large numbers which is required by the gauge invariance , in the calculations at tree level ,
a method was proposed @xcite and realized in the fdc in 1996 .
it is a very powerful method to solve the problem in which a t - channel photon peak appears simultaneous with the gauge invariance breaking due to the finite widths of the involved unstable particles . in 1997 , _ kine _ was developed to treat the phase space integration for multi - final - particle processes @xcite automatically and improvements were made later .
_ gmprocess _ was added to generate multi - final - particle processes in 1998 , it was used to generate all the processes on @xmath0 , @xmath1 colliders up to four final particles@xcite .
to automatic deduce the feynman rules for the minimal super - symmetry model and the r - parity violation model , a modified _ gmodel _ was developed in 1999@xmath22000 @xcite . *
_ gen_html _ * was developed in 2002 to do the translation and presentation of the fdc package results into the _ http version_. therefore most of the results generated by the fdc can be easily put on the fdc - homepage : .... www.ihep.ac.cn/lunwen/wjx/public_html .... many other options , functions have been implemented in the fdc since 1993 .
to build up a physical model in the package is the first step .
it is done in most packages by inputting all the feynman rules , counterterms and constants directly .
this is not a hard task for the standard model , but a very lengthy for more complicate models , such as the super - symmetry model .
for the first principle model , there is a standard way to construct the lagrangian , to quantize it and give all the feynman rules and counterterms . in the fdc , we have developed _ gmodel _ to perform this standard task , i.e , to construct the lagrangian and deduce the feynman rules for the sm and the super - symmetry model .
indeed , it is pretty easy to build in any physical model to our established system , even though developing the program was a very hard job . the standard procedure of including a new model in the system is shown in fig.2 .
the contents of the input documents are very simple and easy to understand for users , _ gmodel _ can construct the lagrangian and then deduce all the mixing matrices , all the feynman rules from the description of the model in the input file , and then prepare the latex version of the result and an internal file for later usage in the fdc . a parameterization scheme , i.e , a set of independent input parameters , can be chosen under interface with users .
the fortran source is prepared to calculate the deduced parameters which are needed in following calculations .
it has been used to generate the feynman rules for the mssm and the r - parity - violated susy model .
it is very easy to change many things in the input file of the model , such as the gauge fix terms , notations , the contents of particles , more leptons , more higgs bosons , soft - breaking terms , forms of the super potential , one can also choose a model which breaks global symmetries such as the lepton number conservation , baryon number conservation , etc .
the integrated script _ gprocess _ is shown in fig .
3 . it is used to calculate a given tree process , i.e. to prepare all the fortran sources for the cross section calculation and event generator .
the generator supplies an interface in the leshouches accord format to transfer the generated event to any parton shower mc for the later hadronization processes .
the interface fortran program between the event generator and pythia@xcite is implemented .
based on the _ gprocess _ , _ gmprocess _ is developed to automatically generate all the processes for a certain topic , such as @xmath3 , @xmath4 and @xmath5 in the sm .
as an example , we calculate the all the decay rates of the super - symmetry partners in this way , and compare the results with isajet@xcite at different snowmass benchmark points@xcite of the parameter space . the event generators are generated for all the processes with @xmath6 final - state particles at different collider energies , such as the lhc , tevatron and lc , in the sm and mssm .
a http version is written in the fdc - homepage .
recently we have extended the application of the fdc to a few very special aspects . in these cases ,
many new programs as well as the modifications of the old ones are developed .
the following three subsections are devoted to them .
based on the fdc package , the sub - project fdc - pwa was completed in 1999 .
it was developed to generate a complete set of the fortran sources to do the partial wave analysis on experimental data of decay processes at low energy regions ( @xmath7 mass , etc . ) .
we have developed a new _ gmodel _ to construct effective lagrangians and deduce the corresponding feynman rules based on the basic requirements such as isospin invariance , c - parity invariance , p - parity invariance , g - parity invariance , lorentz invariance , strange number conservation , charm number conservation , etc .
the input is a list of mesons and baryons with values of their spins ( 0 , 1/2 , 1 , 3/2 , 2 , 5/2 , 3 , 7/2 , 4 , 9/2 ) , isospin , p - parity , c - parity , g - parity , charm number , strange number , masses and widths .
the expression of the effective interaction vertices and the propagators for the high spin states are quite lengthy , and the related amplitudes and amplitude squares are complicated .
there are many free parameters in the effective model and these parameters will be fixed when the generated program is used to do likelihood fitting of experimental data . to work with high spin states and generate the interface to likelihood fitting program
, there are some changes in the feynman diagram generator _ diag _ , the amplitude manipulation program _
amp _ and the kinematics treatment program _ kine_.
the fdc - pwa has been used in the partial wave analysis for the bes experiments since 2000 .
the details about the fdc - pwa will be introduced in a different paper@xcite a new part in fdc , fdc - emt was developed to treat the effective meson model @xcite during 2001 to 2002 .
part of the results can be found in the fdc - homepage .
we have implemented the non - relativistic qcd ( nrqcd ) formalism for calculating the decay and production rates of the s - wave , p - wave , color octet , color singlet heavy quark meson states such as @xmath7 , @xmath8 , @xmath9 in the fdc .
the results about @xmath7 and @xmath9 production at various colliders can be found in the 2003 , 2004 result part of the fdc - homepage .
this part of the fdc will be discussed in another paper@xcite
the fdc package realizes a completely automatic deduction from physical models to final numerical results for tree processes .
we have developed _ gmodel _ to deduce feynman rules for a fwe physical models , standard model and their extension , super - symmetry model , and a few phenomenological models as well , i.e , nrqcd - related heavy quark meson model and effective lagrangian model for partial wave analysis .
three methods are implemented to manipulate amplitudes and generate the the total squared amplitudes .
it can be chosen by using different options . we have built in an automatic treatment program of phase space integration for multi - final - particle processes .
many improvements have been made since 1997 .
_ gmprocess _ has been developed for generating multi - final - particle processes and serves as the event generator for such processes .
it was used to generate the event generators for the lhc , tevatron and lc .
gen_html _ was developed to automatically generate http pages for physical models and multi - final - particle processes .
most of the results are presented in the fdc - homepage .
i would like to thank prof .
y. shimizu , c. h. chang , t. ishikawa , j. fujimoto and t. kaneko for discussions on many related issues , thanks to dr .
y. kurihara for discussion on kinematic treatment for multi - final particles , thanks to prof .
m. kuroda and k. hagiwara for discussion on super symmetry model , thanks to prof .
b. s. zou and y. c. zhu for discussion on partial wave analysis method and effective lagrangian model , thanks to prof .
b. a. li for discussion on his effective meson model and thanks to prof . c. f. qiao for discussion on nrqdc
formalism and related issues , thanks to prof .
x. q. li for many discussion .
this work was supported in part by the national natural science foundation of china under grant nos.90103013 and by the chinese academy of sciences under project no .
kjcx2-sw - n02
. 9 t. ishikawa et al . ,
grace manual , kek report 92 - 19(1993 ) , f. yuasa et al .
phys . suppl .
138,18(2000 ) .
e. boose et al . , inp msu-94 - 36/358 , snutp 94 - 116 ( seoul , 1994 ) ; hep - ph/9503280 j. kublbeck , r. mertig , m. bohm , a. denner , comp .
. 60 165 ( 1990 )
; 64 345 ( 1991 ) . m. tentyukov , j. fleischer , comput .
commun 132 ( 2000 ) 124 - 141 t. sjostrand , comput .
82,74 ( 1994 ) .
h. baer , f. e. paige , s.d .
protopopescu and x. tata , hep - hp/0001086 j. x. wang .
86 ( 1993 ) 214 - 231
. j. x. wang , y. shimizu .
1993 , proceeding of ai93 , oberamergau , germany .
new computing techniques in physics research iii * 517 - 522 c. h. chang , y. b. ding , x. q. li , j. x. wang and j. j. zhu .
d54 , 6963 , 1996 , talk given by j. x. wang on ai96 workshop , sept .
1996 , lausane , switzland .
talk given at first acaf workshop on physics detector at the linear collider , in beijing , 1998 j. x. wang , prepared for 6th accelerator and particle physics institute ( appi 2001 ) , morioka , japan , 20 - 23 feb 2001 . published in * appi 2001 , accelerator and particle physics * 108 - 121 bin - an li , phys .
rev . d52,5165 ( 1995 ) , 5162 ( 1995 ) .
j. x. wang , in preparing .
j. x. wang , in preparing .
allanach et al . , eur .
j. c25 ; 113(2000 ) . | the fdc is a general - purpose program package for feynman diagram calculation .
we outline previous successes in calculations and focus on its recent progress about automatic deduction the feynman rules for first principle model , especially for the supersymmetric model , proper evaluation of the rates of multi - final - particle processes and the event generators in the sm and mssm .
a few special applications are presented .
the fdc - homepage shows an automatic translation of the fdc results into the http version . |
as was clear from the 2010 us _ astronomy and astrophysics decadal survey _ , which ranked lsst as the highest - priority project for ground - based astronomy , time domain astronomy ( tda ) has emerged as a key field of current astronomy and astrophysics .
the temporal domain offers new routes to astrophysical understanding of extreme phases of stellar and galaxy evolution through studies of nov , supernov , gamma - ray bursts and agn , to list only a few .
wide - field and/or temporal surveys of long duration can discover and study binaries in various phases of evolution , some with exotic stellar components such as black holes , or add to the increasingly rich harvest of exoplanets .
but while these extreme events or phenomena have short time - scales ( hours to months ) , astronomical time - scales are predominantly long , often measured in millenia or even 10@xmath1 times longer still .
time - scales appreciably longer than the duration of a given survey require the assimilation of large samples of objects at different evolutionary ( or binary ) phases in order to piece together temporal histories . now that we have the possibility to digitize archived plate collections of large and ( nearly ) continuous duration and totalling many images , it is possible to initiate and conduct studies in the emerging field of _ historical tda _ for very large samples of objects on time - scales that are at least an order of magnitude longer than was possible before .
that was the prime motivation for the _ dasch _ project ( digital access to a sky century @xmath0 harvard ) .
the harvard college observatory plate collection is the world s largest , containing some 450,000 direct plates . as described elsewhere in these proceedings see , it has approximately uniform full - sky coverage from 1890 to 1990 .
the _ dasch _ project ( http://hea-www.harvard.edu/dasch/ ) was initiated to digitize all the plates and make their digital images ( @xmath2400 tb ) and reduced photometric data available on line ( ( * ? ? ? * grindlay et al .
2009 ) ; grindlay et al .
, in preparation ) .
_ dasch _ incorporates the world s fastest and most astrometrically precise plate scanner ( ( * ? ? ?
* simcoe et al . 2006 ) ) and a powerful astrometric and photometric reduction pipeline ( ( * ? ? ?
* laycock et al . 2010 ) ) . over the 5-year development phase of the scanner and reduction software , both the astrometry ( ( * ? ? ?
* servillat et al . 2011 ) ) and the photometry ( tang et al . , in preparation ) were further optimized while scanning about 19,500 plates from five fields selected for having calibrated sequences and a range of stellar densities .
a semi - automated plate - cleaning machine is in the final stages of development ; it will clean the glass back - side of each plate more quickly than the 80 seconds that it takes for an operating sequence of loading
scanning unloading a standard - size plate of 30 cm @xmath3 25 cm (
they are scanned two at a time ) .
full `` production scanning '' and processing of some 400 standard plates per day can commence later this year when two such cleaning machines are ready .
scanning of the whole collection of direct plates , together with associated photometric reduction and population of the mysql database , will require approximately 4 years .
for each resolved object the database contains positions ( j2000 ) , magnitudes , and a list of processing flags .
full processing into the database for all 400 plates scanned each day can be accomplished overnight .
in about 2 years the northern high latitude ( @xmath4 10@xmath5 ) full sky from _ dasch _ will be on - line and available for access and analysis .
the 100-year temporal coverage , compared with @xmath610 years of coverage by ptf and crts and the several epochs of sdss , will enable new studies of long time - scale phenomena .
several examples are shown in fig .
[ fig1 ] , where variable classes are plotted by their approximate ranges of absolute magnitude ( m@xmath7 ) against _ recurrence _ time - scale .
recurrence time - scales are chosen instead of variability time - scales because they signal better which objects can be discovered and measured . recurrence ( or occurrence ) is unambiguous , whereas a variability time - scale can only be measured if there is nearly complete temporal coverage ; in practice , that is only available for a continuous - viewing space mission like kepler .
the ptf and ps1 surveys have typical observing cadences ( for a given sky region ) of at least 1 day , though more frequent sampling is being achieved for limited fields with ptf .
lsst will be comparable , whereas for _ dasch _ the average observation cadence is about 2 weeks , though can be as short as 30 minutes on plates with multiple ( offset ) exposures of a given field . that sets the minimum recurrence time - scale that can be measured ;
the maximum is limited by the total duration of the survey .
all the surveys can cover the same m@xmath7 range , but of course the corresponding distance range differs by the apparent magnitude limit of each survey .
we plot on the right of fig .
1 an approximate distance scale out to which each class of variable or recurrently variable object can be detected . for _ dasch _
we adopted an approximate limiting magnitude of m@xmath7 = 15 , which is appropriate for an assumed ( @xmath8 ) = 1 ; it corresponds to the mean _ dasch _ limiting photographic ( @xmath9 ) magnitude m@xmath10 @xmath214 measured on the 19,500 plates scanned thus far .
the recurrence times for various classes are taken from the literature , and are either uncertain by the ranges shown ( e.g. , the recurrence time - scale range for black - hole `` nov '' is based on just 3 objects ) or are an estimated range given by systematic effects ( such as the metallicity and and red - shift of snia hosts ) .
recurrence - time ranges ending at the right side of the plot ( e.g. , for bright blazar flares ) are simply lower limits .
the recurrence times for large flares ( @xmath11 m @xmath121 ) from flare stars or qsos are estimates based on variability studies against luminosity .
the overall conclusion is simply that by expanding tda surveys to time - scales that are 1 or 2 orders of magnitude longer than those reached by current or immediately projected modern surveys , a range of fundamental classes of objects can be studied as `` individual '' objects in well - defined ( complete ) samples .
those could include measuring the snia rate in the virgo cluster , or the outburst recurrence times for the full set of the 20 or so currently known black - hole transients in the galaxy . for each , recurrence time - scales can be measured or limited in the `` local '' universe and then tested for red - shift dependence using much more distant samples of differing objects in the modern tda surveys .
= 14 , which corresponds to @xmath13 = 15 for typical colours . for ptf , crts and lsst
, the distance - scale axis for any object is increased by a factor of 1030 owing to their deeper limiting magnitudes .
note that shorter variability time - scales may be measured by all those surveys , but the short time - scale limits shown for each object class or survey are approximately those for the variability to _ recur_.,width=422 ]
photometric analyses developed by ( * ? ? ?
* laycock et al . 2010 ) , and recently further improved by tang et al .
( in preparation ) , yield rms uncertainties of @xmath20.10mag over the full range of century of data from the 9 or so series of plates that contribute to a typical light curve , despite differences in plate scale , image quality and any systematic effects .
the basic approach employs sextractor as the object detection and isophotal photometry engine for instrumental magnitude determination by using the now global hubble guide star catalogue ( gsc2.3 ) for a large sample of calibrators present on every plate , thereby allowing both global and local calibrations .
calibration curves are first derived in annular bins to account for vignetting , by fitting instrumental magnitudes against gsc2.3 magnitudes ( @xmath9 ) for an initial photometric solution . that initial calibration is followed by local corrections to remove spatially - dependent plate effects ( usually in the emulsion ) or sky - related effects ( atmospheric extinction and clouds ) .
the gsc2.3 catalogue is not ideal since its photometric precision is only @xmath14 0.2 mag and it is predominantly in a single band ( photographic @xmath9 ) .
fortunately , the all - sky apass ccd survey ( www.aavso.org/apass ) described by has johnson @xmath9 and @xmath13 as well as sloan @xmath15 , @xmath16 and @xmath17 , and will improve significantly both the precision and , particularly , the colour corrections for _ dasch _ photometry .
apass is now partly available and should be full - sky by 2013 .
the five fields scanned in the _ dasch _ development phase are centred on m44 , 3c273 , baade s window , kepler field and the lmc . for the first three
, plates were simply selected if they contained the object ( or the centre of baade s window ) on the plate , and at least 1 cm from the plate edge , or approximately interior to `` bin 9 '' , the outermost annular bin or the outermost 5 mm of the plate where both astrometry and photometry can have large errors . the coverage obtained from that mode of plate selection was increasingly incomplete with radial distance from the target object .
for both the kepler field and lmc , which are each extended regions , a wider boundary of plate - centre coordinates was applied when selecting plates to ensure more complete coverage of the full object . only the m44 and kepler fields have been analyzed in detail , though exploratory results have been obtained for the other 3 fields and will be reported soon : for 3c273 ( grindlay et al .
, in preparation ) and for a classical nova discovered in baade s window ( tang et al . , in preparation ) .
the m44 study led to the discovery of long - term dimming in a population of k giants ( ( * ? ? ? * tang et al .
2010 ) ) . as a follow - up
, more slowly - variable k giants were found in the kepler field and studied with the higher time - coverage of kepler data ( tang et al .
, in preparation ) .
the variability study of the kepler field also led to the discovery of a dust - accretion event in the binary star ku cyg ( ( * ? ? ? * tang et al .
2011 ) ) .
a third , and perhaps most dramatic , example from _ dasch _ is a new type of stellar variability , also discovered in the m44 field ( tang et al .
2012 ) , and is illustrated in fig .
observations with the telescopes at the harvard - smithsonian flwo observatory in arizona for spectroscopic classification of _ dasch _ variables revealed that this star was an m0 iii giant .
comparison with the asas ccd photometric survey ( ( * ? ? ? * pojmanski 2002 ) ) revealed it to be a semi - detached binary with a 119.2-day period and an amplitude of 0.16 mag in the @xmath13 band .
as described in detail by tang et al .
( 2012 ) , the cool giant s companion is a white dwarf , and we surmise that the remarkable flare in 1942 and its subsequent 10-year decline was most probably due to nuclear h - shell burning .
the lack of emission lines from this symbiotic nova is new , but is consistent with the behaviour of some other old nov and may also imply little or no ejection of the envelope mass ( @xmath23 @xmath3 10@xmath18 ) that is required for ignition on what is probably a @xmath20.6-white dwarf .
we include a brief summary of the _ dasch _ light curve for the bright quasar / blazar 3c273 , which is discussed in detail by grindlay et al .
( in preparation ) .
its 100-year light curve ( fig.[fig3 ] ) demonstrates the long - term variability of this luminous agn .
the inset shows a `` flare '' in 1941 with characteristic rise and fall times of about 3 days ( and there are several other examples ) .
time - scales of 130 days for brightening are evident in the overall light curve , and are comparable to the optical and infra - red variability for 3c273 reported by ( * ? ? ?
* courvoisier et al . ( 1988 ) ) , who found that the fastest variations were of the order of 1 day but their observations were too short to measure the longer - term variations detected with _
dasch_. the dominant power in the _ dasch _ variability spectrum is over longer time - scales like 0.52 years ( such as the abrupt decline in flux from about 19271930 ) , and provides constraints on the size of the optical emission region .
the _ dasch _ pipeline ( ( * ? ? ? * los et al . 2010 ) ) and database software run on a high - speed computer cluster and raid disk system . in production mode it can process ( overnight ) the full pipeline for the nominal 400 plates scanned in a day , to populate a mysql database with photometric values and errors for each of the resolved stellar images ; typically there are @xmath250,000 on a standard plate but there are more on `` a '' plates .
light curves are generated very rapidly for any object by extracting from the database the magnitudes thus determined from all plates , or only those with magnitude measures meeting a set of user - selected criteria .
variability measures and tests of their validity can then be derived readily .
additional variability analysis tools are being developed , and will be made available when the full database becomes public .
the full _ dasch _ output database of @xmath2450,000 plate images and derived magnitudes for each resolved object ( @xmath21pb in total ! ) will be made available for world access as it is completed incrementally .
the present plans are to digitize the northern sky at galactic latitudes @xmath4 10@xmath5 first , to allow comparisons with existing surveys such as sdss , ptf , pan - starrs-1 and the crts .
that stage could be completed by mid-2013 .
given the difference in limiting magnitudes ( up to 1418 for _ dasch _ against 2022 for the modern surveys ) , such comparisons will be mainly for context or extreme transients .
the southern sky above / below the galactic plane will come next , with expected completion in early 2015 , followed by the galactic plane by mid-2016or still well before lsst .
the reason for doing the galactic plane and bulge last is to allow time to develop analysis of crowded - field photometry further by invoking point - spread function ( psf ) and image subtraction techniques in order to improve the present isophotal photometry used for sextractor .
experiments have recently been undertaken to optimise the use of psfex+sextractor for magnitude- and position - dependent fitting of the plate psf . | the large - scale surveys such as ptf , crts and pan - starrs-1 that have emerged within the past 5 years or so employ digital databases and modern analysis tools to accentuate research into time domain astronomy ( tda ) .
preparations are underway for lsst which , in another 6 years , will usher in the second decade of modern tda . by that time
the digital access to a sky century @xmath0 harvard ( _ dasch _ ) project will have made available to the community the full sky historical tda database and digitized images for a century ( 18901990 ) of coverage .
we describe the current _ dasch _ development and some initial results , and outline plans for the `` production scanning '' phase and data distribution which is to begin in 2012 . that will open a 100-year window into temporal astrophysics , revealing rare transients and ( especially ) astrophysical phenomena that vary on time - scales of a decade .
it will also provide context and archival comparisons for the deeper modern surveys .
astronomical data bases : catalogues , surveys ; stars : variables ; galaxies : active |
direct identification of x - ray , optical and radio counterparts of long duration , @xmath10 sec , gamma - ray bursts ( grb ) , and hence their host galaxies has recently resolved the grb distance scale controversy : the observed redshifts span a wide range , from @xmath11 to @xmath0 @xcite .
this suggests a number of uses of grbs as cosmologically distributed probes .
since they are believed to be associated with compact remnants of massive stars , it has been suggested they be used to trace star formation rate obscured by dust @xcite , star formation rate at very high redshifts @xcite , as probes of the metal enrichment of the interstellar medium @xcite , intergalactic medium @xcite , and galactic and intergalactic dust at high redshifts @xcite . here
, we use grbs as sources for weak lensing by the large scale structure ; our ultimate goal is to probe the mass distribution on @xmath12 mpc scales , at a typical redshift of 0.3 . unlike weak shear lensing , which is detected through shape distortion of lensed _ resolved _ galaxies @xcite ,
weak lensing of unresolved _ point sources _ , like grbs , would manifest itself through the angular ( anti-)correlation between sources and lenses , the latter being the intervening clumpy mass distribution , which is assumed to be traced by visible galaxies @xcite .
the sign of the correlations is determined by the slope of the source number counts in the appropriate redshift interval .
for example ,
bright qsos have steep number counts implying that positive correlations between these and intervening galaxies should be expected .
in fact , optically selected lbqs qsos , with @xmath13 , and radio selected 1jy qsos , with @xmath14 , are independently correlated with faint apm galaxies on angular scales of @xmath15 @xcite .
the linear extent of the structures is @xmath16 mpc at the redshift of typical lenses , @xmath17 .
radio selected 1jy sources are also correlated with iras galaxies @xcite , and zwicky clusters @xcite on angular scales of @xmath18 . in these studies ,
the redshifts of the sources and probable redshift distribution of the galaxies do not overlap , insuring that physical associations do not contaminate the lensing signal .
qualitative signature of these correlations are those of lensing , however , it is hard to explain the results quantitatively : the observed correlations persist to scales of @xmath19 with amplitude of @xmath20 , whereas @xmath21 is expected if we live in a universe with @xmath8 , and matter power spectrum not too different from the observed galaxy power spectrum on large scales @xcite .
if lensing induced correlations were indeed as small as predicted , they could not have been detected on @xmath15 scales given the available number of sources @xcite .
these results by themselves are puzzling .
however , they become problematic in view of the recent observations of cosmic _ weak shear _ lensing , which agree well with the currently accepted cosmological model , matter fluctuation spectrum , and biasing parameter close to unity .
there are now several independent determinations of weak shear lensing in fields of up to @xmath22 diameter @xcite , and the agreement between them is impressive .
furthermore , when cast in terms of the best estimate for @xmath23 , weak lensing results agree remarkably well with cluster normalization constraints @xcite . grbs provide us with a different set of cosmologically distributed sources , which should also be affected by weak lensing , just as qsos are .
however , as we explain below , unlike qsos , which are predicted and shown to exhibit correlations with galaxies , grbs , due to their distribution in fluence and angular positional error , are predicted to be anti - correlated with the foreground lensing mass distribution .
thus , grbs should provide a new and interesting test of weak lensing on large scales .
there are two potential difficulties in using grbs in a weak lensing study .
first , grb positions are not well localized , with errors ranging from a fraction of a degree to as high as @xmath24 for some events with low fluences .
obviously , correlation scales that can be reliably probed are limited by the grb position errors on the sky . to circumvent this problem we use grb subsets with upper limits on error .
second , since individual grb redshifts are largely unknown , some of the grbs at low redshifts will be physically associated with the galaxies which we use to trace the lensing mass , and will ` contaminate ' the lensing induced signal .
currently , only about 26 grbs , i.e. a very small fraction of all events have confirmed redshifts . of these , about 3 are at @xmath25 .
guided by the known redshifts , several workers have proposed empirical relations connecting observable properties of grbs , such as functions constructed out of time information , to their luminosities @xcite .
redshift distribution implied by these indicate that the true redshift distribution is very broad , extending to @xmath26 .
the fraction of low redshift events is expected to be small ; for example , using the model of @xcite and their fig .
3 histogram , the fraction of events below @xmath27 is roughly estimated to be 6% ( 13% ) . as tracers of dark matter on large scales we use apm galaxies and abell - corwin - olowin ( aco ) clusters , which peak at @xmath28 , and extend to a maximum @xmath29 .
so the fraction of grbs arising in these structures should be small , @xmath30 .
furthermore , and more importantly , we predict and find anti - correlations between grbs and intervening galaxies , and so physical associations , if any , would have diminished the amplitude of lensing induced anti - correlations
. thus our observed signal ( section [ apm ] ) is a lower limit .
as we show in section [ can_we ? ] , the observed amplitude of lensing induced anti - correlations is hard to explain within standard cosmological models , so at this point there is no motivation to carefully subtract the effect of physical associations from the observed anti - correlation signal ; we neglect the effect of physical associations in our analysis .
the plan of the paper is as follows .
after describing the data in section [ data ] we use the fluence and error information of current batse catalog grbs to predict the amplitude and sign of grb - galaxy correlations , i.e. compute magnification bias ( section [ underden ] ) . in sections
[ apm ] and [ aco ] we estimate the strength of observed correlations between grbs and apm galaxies and aco clusters , respectively .
anti - correlations are found in both cases , as predicted , but the amplitude of the effect is higher than expected . in sections [ we_can ]
we discuss the various possibilities for reconciling observations with theory , and propose a speculative scenario .
we summarize and discuss our findings in section [ sum+disc ] .
we use the current batse grb catalog , ending with trigger number 8121 , which occurred on may 26 , 2000 . from these events we select those that were not overwritten by a later more intense trigger , and
that have non - zero fluences in the 50 - 100 kev and 100 - 300 kev energy channels .
grb fluxes and fluences are recorded in 4 channels , which cover energy ranges 20 - 50 kev , 50 - 100 kev , 100 - 300 kev , and @xmath31 kev , respectively .
we use the fluences in the middle two channels because the corresponding energy range has the peak flux , and coincides with the energy range of the nominal batse on - board burst trigger ( http://www.batse.msfc.nasa.gov/batse/grb/catalog/4b/4br_flux.html ) .
these cuts leave us with 2038 events .
batse events are not well localized in the sky ; positional errors come in two flavors , statistical and systematic .
statistical errors , @xmath2 , are recorded in the batse catalog , and range from a fraction of a degree to @xmath32 degrees .
( all quoted errors are @xmath33 . )
the peak of the statistical error distribution is at about @xmath34 .
systematic errors , @xmath35 , were analyzed for the revised 4b catalog @xcite , and were found to have a modified gaussian distribution such that 78% of the events have errors of @xmath36 . for our purposes
an approximation used for the 3b catalog will suffice : we assume the systematic error distribution to be a gaussian such that 68% of the events have systematic errors of @xmath37 . assuming that statistical and systematic errors are independent
, the total error is @xmath38 .
the statistical errors and fluences of the 2038 bursts are plotted in fig .
[ ncounts ] .
solid squares represent 10 batse grbs with known fluences and redshifts .
batse catalogs have a declination dependent completeness level on the sky , with a 74% difference between the most ( @xmath39 ) and least ( @xmath40 ) completely covered declinations @xcite .
this incompleteness will not matter for our analysis in section [ apm ] , because we compare galaxy density around each grb with the expected average density on the same apm plate .
this incompleteness will be important in the cross - correlation with aco galaxy clusters , section [ aco ] , and will be accounted for .
apm galaxy catalog @xcite is digitized possi plate material , with magnitude limits 20.0 and 21.5 on ` red ' and ` blue ' possi plates in the northern hemisphere , and 21.0 and 22.5 on ` red ' and ` blue ' ukst plates in the southern hemisphere , with internal accuracy of @xmath41 mag for all but the faintest objects , and external , i.e. plate - to - plate accuracy of @xmath42 mag .
these are tolerable errors for our purposes : internal accuracy is sufficiently good to trace the light fluctuations on each plate , and external accuracy is less important to us because we confine analysis for each grb to a single apm plate .
similar to previous work @xcite we select galaxies as objects that are classified as extended on the ` red ' plates by the apm .
the magnitude range is set to probe the most distant regions reached by the apm : on possi , @xmath43 , and on ukst , @xmath44 . in the northern hemisphere apm plates
do not cover a band between @xmath45 and @xmath46 galactic latitude , and in the southern hemisphere the coverage is even less complete .
furthermore , even though each plate is @xmath47 , because of plate vignetting we do not use grbs located beyond @xmath48 from the plate center .
thus , we only use a subset of 732 grbs in the apm analysis in section [ apm ] .
the intervening mass distribution between us and grb events is an uneven lens that stretches some areas of the background sky and shrinks others .
the redshift distribution of the observed apm galaxies is such that they trace dark matter from @xmath49 to a maximum of @xmath50 .
( see fig . 1 of williams & irwin 1998 ) .
let the total average lensing optical depth of this slab of matter be @xmath51 : @xmath52 here , @xmath53 is the thickness of the lensing slice at redshift @xmath54 , @xmath55 is its mass density , and @xmath56 is the critical lensing surface mass density at @xmath54 for a source at @xmath57 . within this redshift range ,
the fluctuations in the projected surface mass density are @xmath58 , and are assumed to be small on large scales . for a specified cosmology and source and lens redshifts , these fluctuations are translated into ( de-)magnifications on the sky , with respect to the ` smooth universe ' case : @xmath59 , where @xmath60 . for sources with exactly known positions , like qsos , this magnification field @xmath61 , combined with the qso number counts for the relevant range of redshifts , can be translated into an observed distribution of number density on the sky of sources down to some specified flux limit .
if @xmath62 is the slope of the number counts near the survey flux limit , then the number density of qsos will be a factor of @xmath63 different from the ` smooth universe ' ( @xmath64 ) case ; @xmath65 is called source over- or underdensity , depending on whether it is greater or less than 1 .
the fact that it is different from 1 is called the magnification bias .
bright qsos , whose number counts are steep , would appear to be @xmath65 times more abundant in the directions of mass concentrations , i.e. they would appear to be correlated on the sky with the nearby structure .
faint qsos , on the other hand , have a shallow number counts slope , and so will be anti - correlated with the foreground lenses . with grbs ,
the calculation of @xmath65 is somewhat different , because in addition to limiting these in fluence @xmath1 , one also needs to place an upper limit on grb positional error @xmath2 , so that the grbs have a reasonable chance of being within the specified area of correlations .
even though positional errors are not directly affected by lensing , the observed grb errors are well correlated with their fluences , therefore , as a grb s fluence is changed through magnification , so is its error .
thus , in order to correctly predict @xmath65 one needs to consider grb distribution in @xmath2 and @xmath1 , fig .
[ ncounts ] .
let @xmath66 be the normalized probability distribution of grb sources in statistical positional error @xmath2 at a given fluence @xmath1 .
as @xmath1 we use the sum of fluences in channels 2 and 3 .
probability distribution @xmath66 is estimated from the batse data itself , by binning the distribution presented in fig .
[ ncounts ] by fluence and error , and then normalizing distributions in @xmath2 for each fluence separately .
if the sources are limited in error by @xmath67 and @xmath68 from below and above , and in fluence by @xmath69 from below , then the number of such sources seen behind a smooth patch of lens with magnification @xmath61 is given by , @xmath70~df^\prime , \label{nlensed}\ ] ] where @xmath71 is the distribution of sources in fluence one would see in a universe with completely smooth mass distribution .
this distribution is not observable ; however , given the small typical magnifications it is reasonable to assume that the observed sources give a fair representation of @xmath71 .
the ratio of the number of sources observed and the number that would be seen if the mass were smoothly distributed everywhere , is the over- or underdensity , @xmath72 figure [ overden ] shows the result , for a few combinations of @xmath67 , @xmath68 , and @xmath69 .
the upper two lines ( cross and plus symbols ) are for grb subsets limited in fluence only , with no restrictions imposed on error .
the lower three lines are for grb subsets limited in error , with no limits placed on fluence .
we used all 2038 grbs in the current batse catalog to construct the @xmath73 relation in fig .
[ overden ] .
had we used , say the 732 grbs that are found on apm plates the shape of the curves above would have been the same , within the noise . similarly , the plot is not very sensitive to the particular choice of fluence channel .
note that all lines go through @xmath74 , as they should .
some of the @xmath75 lines vary non - monotonically with magnification , and most lines vary erratically .
this is due to the noise associated with the finite number of grb points in fig .
[ ncounts ] .
shot noise is especially pronounced when grbs with small errors and/or large fluences are considered , and when @xmath76 : in that case the @xmath77 term in eq . [ nlensed ] refers to a very small number of grbs at highest fluences , and so shot noise fluctuations can make the resulting @xmath65 vary a lot .
most of the subsets that can be constructed from the total grb catalog using different fluence / error cuts are predicted to be anti - correlated with lenses .
in general , overdensities would be expected only if there is a large ` reservoir ' of sources just below the detection limit .
the fact that the predicted grb @xmath65 s tend to be @xmath78 can be seen directly from fig .
[ ncounts ] : for almost any @xmath2 , @xmath1 cut there is no large ` reservoir ' of sources just below the @xmath2 , @xmath1 limits ready to be magnified into the observed subset .
the only subsets of grbs which are predicted to be correlated with lenses are those with large fluences , @xmath79 erg/@xmath80 .
however , these subsets contain very small numbers of grbs , so the uncertainties in the @xmath73 relation for these are probably large .
dashed lines in fig .
[ overden ] are fits to the @xmath81=@xmath82 and @xmath83 grb subsets , which we will be considering in some detail in the next section .
we wanted fits of the form @xmath84 , to mimic the qso relation , @xmath63 . in the present case ,
@xmath85 and @xmath86 provide adequate fits for the two subsets respectively ; the corresponding @xmath87 s are @xmath88 , and @xmath89 .
the fits are only rough , but are completely adequate for our purposes .
having made predictions as to the amplitude and sign of the grb correlations with the foreground matter , we can now proceed to do the corresponding ` observations ' .
in principle , we want to determine the number density of grbs as a function of projected mass excess , @xmath58 .
however , since grbs are rare and galaxies are plentiful , we instead estimate galaxy density in circles around grbs .
our analysis later , section [ can_we ? ] will take this difference into account .
let the number of galaxies within @xmath90 degrees of a grb , and in the magnitude ranges specified in section [ gal+clus ] , be @xmath91 .
the latter should be normalized by average expected galaxy count within similarly sized random @xmath90-patches on the sky . in our analysis , because the apm plate - to - plate magnitude calibration can be rather uncertain , about 0.3 mag , we select control areas for normalization from the same plate as the corresponding grb . furthermore , because the object density on plates varies as a function of distance from center , the control patches are restricted to lie at the same distance from plate center as the grb , @xmath92 . fig .
[ schematic ] illustrates our selection of control patches . for each grb , @xmath93
such random positions are created .
we scale @xmath93 linearly with @xmath92 , such that at @xmath94 , @xmath95 .
( we tried using same @xmath93 for all grbs ; the results below did not change substantially . )
the number of galaxies in @xmath90 circles around these are also recorded , and the average is calculated for every grb , @xmath96 . the galaxy excess , @xmath97 - 1 $ ] , is then equal to @xmath98 , where @xmath99 is the biasing parameter of apm galaxies with respect to the underlying mass . using fig .
[ overden ] we select grb subsets that are most likely to show weak lensing induced signature .
we choose subsets with @xmath100 and @xmath101 .
how do we decide on the size of the @xmath90-patch around each grb ?
obviously , each grb should have a good chance of actually being located within these circles .
the limiting factor here is the systematic positional error : even if a grb has zero statistical error there is only a 25% chance that it will be actually within a @xmath102 radius drawn around it . on the other hand ,
the size of the apm plate limits the size of @xmath90 from above ; we settle on @xmath103 .
given that the grb and the @xmath93 control @xmath90-patches are equidistant from plate center , to avoid significant overlap between these we restrict our grb subset further by selecting only those at @xmath104 . thus grbs are limited to lie at @xmath105 .
the @xmath100 and @xmath101 subsets contain 46 grbs and 74 grbs respectively , and we will refer to them by quoting these numbers .
we now ask if the galaxy density around grbs in the two subsets is different from what one would expect if random points were used in place of grbs . instead of generating a set of random points for this purpose
, we use the whole set of grbs found at @xmath105 , regardless of positional error .
there are 448 of these . since most of these have rather large errors , their real positions are quite far from batse recorded positions , so in effect we have a set of randomly selected points . in fig . [ subsample_hist ]
the heavy solid histogram is the distribution of @xmath106 , or , equivalently , @xmath107 for the 448 ` random ' points .
the dashed and dotted histograms are the 46 and 74 grb subsets respectively .
the averages of the three distributions are 0.998 , 0.883 , and 0.915 , so grbs in the two subsets are found in the directions of foreground regions that are , on average , 12% and 8% underdense in galaxies . in other words , we detect anti - correlations between grbs and intervening galaxies , which is what was predicted in section [ underden ] and fig .
[ overden ] .
we leave quantitative comparison with predictions until section [ can_we ? ] .
let us now evaluate the statistical significance of this result .
for each grb we calculate @xmath108 , where @xmath109 is the number of random @xmath90-patches , out of total @xmath93 , that have less galaxies in them than the @xmath110patch around the real grb .
in other words , @xmath108 is the rank of the real grb patch among its ` random peers ' ; @xmath111 if grbs are randomly distributed with respect to the foreground galaxies , but if grbs have an excess of galaxies in their foregrounds then @xmath112 . if grbs occupy random positions with respect to the foreground galaxies , then the _ distribution _ of @xmath108 values is known the cumulative distribution should be linear , and in fact for the whole set of 448 grbs it is , see solid line in fig .
[ smkstest2_histo ] .
the dashed and dotted histograms are for the 46 and 74 grb subsets , respectively ; their @xmath108 values are 0.381 and 0.423 , implying that grbs have a deficit of galaxies in their foregrounds .
using the kolmogorov - smirnof ( ks ) test these two distributions differ from the whole set of 448 grbs at the 99.68% and 97.62% significance level .
note that the 46 subset [ @xmath100 ] was predicted to be more strongly anti - correlated with galaxies than the 74 subset [ @xmath100 ] , which is what is observed , in spite of the 46 subset having less grbs than the 74 subset .
since the 46 and 74 grbs in the two subsets are a part of the 448 ` random ' points , these results could be a conservative estimate of statistical significance .
however , if we use truly random points , the statistical significance of the results is not much higher .
a further test of the statistical significance of the results for the two subsets in the last section is provided by carrying out the same analysis on many grb subsets , selected using a range of error and fluence criteria .
subsets are defined by @xmath90 , the size of the patch around grbs , statistical error range , @xmath114 , and the range of grb s @xmath92 .
we use four different @xmath115 ranges : @xmath102 , @xmath116 , @xmath117 , and @xmath118 , and include grbs with errors up to @xmath119 . for each of these
we use five different @xmath92 ranges : @xmath120 , @xmath121 , .... , @xmath122 .
we try three values of @xmath90 : @xmath102 , @xmath116 , and @xmath117 .
the total number of grb subsets is 615 ; the number of grbs in these subsets varies from 1 to 269 .
of the 615 total cases considered , table 1 lists those that deviate from the parent grb set at more than 97% confidence level , as judged by the ks test .
the two cases considered in section [ one ] are marked with a star .
most of the other cases in table 1 are related to these two subsets . even though the subsets are not independent , one should still expect some small fraction of the 615 subsets to be significant at @xmath123 , so it is not surprising that we have these .
it is interesting , however , that the subsets that do show significant deviations form the parent grb population are the ones that we argued would be most likely to be affected by lensing .
furthermore , all cases reported in table 1 are _
anti_-correlations , consistent with weak lensing predictions of section [ underden ] . here
we attempt to account for the amplitude of anti - correlations of two grb subsets considered in section [ one ] .
as before , we assume that the whole set of 448 grbs , found at distances @xmath105 from plate centers is a collection of random points on the sky , since most of them have very large position errors .
hence , the solid line histogram of fig . [ subsample_hist ] is a fair representation of the distribution of average projected galaxy densities in @xmath103 circles .
this counts - in - cells distribution does not include fluctuation power on scales larger than the size of an apm plate .
because we are dealing with weak lensing regime , we can separate the effects of density fluctuations on different projected scales ; our ` observations ' and analysis do not deal with scales larger than @xmath48 .
fluctuation power below the scale of correlations , @xmath124 , for example due to galaxy cluster cores , does not affect the correlations significantly .
this is illustrated by @xcite and @xcite , who have shown that including the fluctuation power on non - linear scales changes the amplitude of lensing - induced correlations by @xmath125 , for scales @xmath126 .
we start with the assumption that without lensing grbs and foreground apm galaxies are randomly distributed with respect to one another , on the sky .
we now construct a _ synthetic lensed _
grb subset .
suppose a large number of grbs with the batse - observed distribution of fluence and error properties go off in the direction of a patch with some @xmath97 $ ] value , picked at random from the solid line distribution in fig .
[ subsample_hist ] . using the various relations in section [ underden ] and
the fit in fig .
[ overden ] , we can predict the overdensity of grbs in that direction : @xmath127\big)^{-2\beta } \approx \big(1 + 2\beta{\kappa_0\over b } \big[{{n_{gal , d}}\over{\langle n_{gal , r}\rangle}}-1\big]\big ) .
\label{q_full}\ ] ] the apm galaxies trace matter extending , at most , to @xmath29 .
the corresponding optical depth in a flat @xmath128 universe is 0.025 for @xmath129 , and 0.031 for @xmath130 ; we take @xmath131 , as an average .
we set @xmath132 , as estimated on large scales @xcite .
we use @xmath85 and @xmath86 , for the 46 and 74 synthetic grbs subsets , as determined in section [ underden ] ) . if the number of grbs going off is 1 , then @xmath65 is the probability that it will be observed . using the latter definition ,
a grb is accepted into the synthetic lensed subset with probability @xmath65 ; if @xmath133 , the grb is accepted into the subset , and an additional one is accepted with probability @xmath134 .
we repeat the process until we build up two separate subsets of 46 and 74 grbs each ; we repeat the procedure a 1000 times .
the average @xmath97 $ ] in the 46 and 74 synthetic subsets is close to 0.996 , barely below what one expects for a randomly selected set of points on the sky , and far from 0.883 and 0.915 which are found for the real subsets .
in fact , the fraction of synthetic subsets whose average @xmath97 $ ] is less than 0.883 and 0.915 , is 0.3% and 0.75% respectively .
hence , this model fails to reproduce observations .
how much do we have to change the parameters in eq .
[ q_full ] to reach agreement with observations ?
if @xmath51 is increased by a factor of 10 ( 20 ) , then the respective percentages for the 46 and 74 subsets become about 2% ( 8% ) , and 3% ( 10% ) , so , roughly speaking , the observed subsets can occur with @xmath135 probability .
we conclude that observations could be deemed to agree with expectations if @xmath136 .
how realistic is this value ?
the optical depth of the entire column of matter between the observer and a source at @xmath137 ( 3.0 ) is @xmath138 ( @xmath139 ) .
it is very hard to imagine how apm galaxies , with faint galaxy magnitudes of around 20 - 21 in typical optical bands can be faithful tracers of matter fluctuations at redshifts 1 - 3 .
an alternative is to keep @xmath51 at 0.028 , but require @xmath140 .
this is inconsistent with observations that estimate @xmath99 on large scales to be within @xmath141 of unity @xcite .
we will return to interpretation of our results in section [ we_can ] .
if nearby mass distribution is weakly lensing grbs , then all tracers of the nearby mass should be anti - correlated with grb events .
here we carry out cross - correlations between grbs and galaxy clusters from the abell - colwin - olowin catalog @xcite .
the present work is not the first to consider grb - aco correlations .
prior to 1997 , before the first spectroscopic redshifts to grbs were established , several workers have looked for correlations between grbs and aco clusters .
their motivation was that if grbs are formed in galaxies , then the degree of their association with known populations of galaxy clusters will place constraints on their redshift distribution .
all these studies tacitly assumed that weak lensing effect was negligibly small .
based on the current cosmological models , and our predictions in section [ can_we ? ] , their assumption was perfectly justified .
some studies detected correlations between subsets of clusters and subsets of grbs , others detected no correlations .
for example , @xcite claimed an association of 136 @xmath142 grbs from the 3b catalog and 3616 @xmath143 aco clusters at separations @xmath144 , at a significance level of 95% .
@xcite reported correlation of 71 @xmath145 3b grbs with 185 nearby aco clusters with @xmath146 and @xmath147 at @xmath148 level .
they found even stronger correlations between 27 @xmath149 3b grbs and all 5250 aco clusters , at @xmath150 level .
given these findings , they were surprised that 40 very well localized inter planetary network ( ipn ) grbs were not correlated with aco clusters .
@xcite also used ipn positions but for a much larger sample , 157 grbs from 4b catalog .
the average reduction in error area is @xmath151 for ipn compared to batse positions . despite the accurate positions @xcite did not find any correlations , for assumed aco cluster radii of @xmath152 or @xmath153 .
when grb redshifts became available a few years ago , grb - cluster cross - correlation work stopped . in view of our apm galaxy results in section [ apm ] , it becomes interesting to revisit grb
aco correlations , especially since the current batse catalog contains about twice as many grb events as were used in earlier studies , and error estimates for some grb events have been recently revised @xcite .
abell - colwin - olowin catalog clusters are classified into distance and richness classes according to the standard abell criteria @xcite .
since lensing should be sensitive to cluster distances we divide the clusters into three distance ranges : nearby , d=1 - 4 , which have an average redshift of 0.06 , medium distant , d=5 , with an average @xmath54 of 0.11 , and distant , d=6 , with an average @xmath154 .
distance class 7 is very incomplete ; it has 8 clusters vs. about 2000 d=6 clusters , so we leave out d=7 clusters from the analysis . we further divide the clusters into poor ones , r=0 - 2 , and rich ones , r@xmath1553 .
aco catalog is incomplete close to the galactic plane , so we mask out areas at @xmath156 .
this leaves us with 3608 clusters and 1021 grb events which we use to compute cross - correlations on a range of angular scales .
[ aco_corr_one ] shows grb - aco correlation functions between grbs with @xmath157 and six subsets of aco clusters .
we used @xcite estimator , which has improved variance compared to the standard estimator .
the two sets of solid lines in each panel are the 95% and 99% confidence limits derived from cross - correlations using 300 random realizations of the grb catalog ( compensated for the same declination dependent sky coverage as the real batse catalog ) .
the dotted lines are the same correlations , but binned into narrower angular bins .
( we do not plot confidence limits for these ) .
out of sixty points plotted in six panels of fig .
[ aco_corr_one ] only one is significant at @xmath158 : it states that grbs with @xmath157 are anti - correlated with rich distant clusters on scales @xmath159 .
of the six cluster subsets , the rich distant ones are most likely to act as weak lenses for background grb , so this result supports our earlier grb - apm findings .
whereas in the apm study the angular scales we could probe were limited by the plate size , here we are not restricted in that regard .
we find that the full scale of anti - correlations corresponds to about 150@xmath160mpc at the redshift of the clusters , which is the scale where cluster power spectrum is observed to peak : @xcite find peak @xmath161 mpc@xmath162 for aco clusters , while @xcite find peak @xmath163 mpc@xmath162 for apm clusters .
results with @xmath164 ( corresponding to the 46 grb subset of the apm analysis ) are similar to the ones in fig .
[ aco_corr_one ] , however , the significance of anti - correlations with rich distant clusters drops to @xmath165 .
correlations with grbs unrestricted in positional error show no signal .
here we explore ways of accommodating grb - apm anti - correlations , and earlier reports of qso - galaxy correlations @xcite .
there are three possible avenues towards a resolution of the high amplitude of ( anti-)correlations ; these concern the sources ( grbs or qsos ) , the lensing process , and the lenses ( the mass distribution ) , respectively .
we discuss these in turn .
we assume that in the weak lensing regime the effects of these three sets of possibilities add up ` linearly ' , and so they can be considered separately .
suppose our assumptions about grb number distribution in fluence and error ( fig .
[ ncounts ] ) are incorrect , i.e. our assumed conversion between @xmath61 and @xmath65exponent @xmath166 in eq .
[ q_full]is wrong .
this could arise , for example , due to shot noise given small number of grbs .
the extreme value that @xmath166 can take is @xmath167 , which is when there are no more sources beyond the limit of the survey , and area dilution of lensing reduces the sky density of sources by @xmath61 .
even this extreme case does not come close to reproducing grb - apm anti - correlations .
a similar solution to the problem was considered in @xcite , who observed positive correlations between bright , optically selected qsos and apm galaxies .
the qso number counts in the relevant redshift range gave @xmath168 , whereas @xmath169 would be required to reproduce observations .
that explanation was ruled out as unlikely .
is it possible that propagation of light through an inhomogeneous universe is not well modeled by the standard lensing equation ? in terms of eq .
[ q_full ] that would mean that the expression in round parentheses is incorrect . this was considered in @xcite , who explored the effect of the second order term in the lensing equation . including that term had interesting consequences : magnification was increased , but only by 10% , and _ only _ in the mass distribution scenarios which were far from gaussian random fields on large spatial scales .
so this does not appear to be a promising avenue for the resolution of the problem .
furthermore , propagation of light through a numerically simulated universe of popular cosmological models was done using the full geodesic equation of motion @xcite , and the results were found to be in good agreement with the predictions based on the standard lensing equation .
the last set of possibilities is that the mass fluctuations are quite different from the observed projected galaxy density fluctuations . to reproduce the grb - apm anti - correlations , @xmath170 in eq .
[ q_full ] , would have to be increased by a factor of 10 - 20 .
increasing @xmath51 by that large a factor is ruled out ( section [ can_we ? ] ) . on the other hand ,
a constant biasing of factor of 0.1 - 0.05 is ruled out by dynamical measurements on cluster and supercluster scales . here
we propose another variant of the third set of possibilities ; we relax the requirement that @xmath99 is constant as a function of @xmath58 .
this is a toy model only .
we keep @xmath131 , which is appropriate for the average optical depth of the apm galaxies . within this slab of apm galaxies ,
the projected galaxy density is assumed to be a monotonic , but not linear tracer of the projected mass density .
the shape of the distribution of apm galaxy counts - in - cells on @xmath103 scales is fixed by observations . to maximize lensing effects
we chose a skewed shape for @xmath171 : let the distribution of projected mass densities averaged over circles of @xmath103 have the shape of a half - gaussian , with the sharp cut - off coinciding with @xmath172 , and the tail extending to positive values of @xmath58 . in terms of the optical depth
the cutoff is at @xmath173 ( see insert in fig .
[ smmodel_histo ] ) ; the corresponding lines of sight are ` empty beams ' and produce maximum possible demagnification of the sources .
the width of the half - gaussian is adjusted such that the average @xmath58 , and hence the average @xmath174 are zero . since @xmath171 is highly asymmetric , while the observed @xmath175 - 1)$ ] is symmetric , we must adjust the biasing function to map the former distribution onto the latter . fig .
[ smmodel_bulk ] plots one possible version of such a biasing function , and the corresponding distribution of projected galaxy densities is shown as the dashed line in fig .
[ smmodel_histo ] .
it is symmetric , and nearly zero - centered , i.e. similar to the observed apm distribution ( the solid histogram in this figure is the same as the solid histogram in fig .
[ smkstest2_histo ] ) . in short
, the biasing function takes the very skewed projected mass distribution ( insert in fig .
[ smmodel_histo ] ) and maps it onto a symmetric projected galaxy distribution ( dashed histogram in fig .
[ smmodel_histo ] ) . within this toy model ,
the average galaxy density for the 46 and 74 synthetic grb subsets is about 0.97 , and densities @xmath176 ( @xmath177 ) for the 46 ( 74 ) subset occur in 9% ( 13% ) of the cases .
thus the model can account for the observations .
furthermore , this model also reproduces the observed correlations reported in @xcite : if the slope of qso number counts is @xmath178 ( as quoted in that study ) , our model gives 1.018 for the average normalized galaxy foreground density in circles @xmath103 around bright optically selected qsos , a value consistent with fig .
6 of that study ( @xmath179 ) .
our model @xmath180 distribution and the resulting biasing function are very different from what is expected in currently accepted universe models , with initial density perturbations specified by harrison - zeldovich - type spectrum , and matter and cosmological constant ( or the like ) contributing comparably to the net zero curvature . in such models
@xmath180 is gaussian on large scales , and the biasing factor is close to unity for all values of density excess .
for example , fig . 13 of @xcite show that the symmetric shape for @xmath180 is already attained at smoothing scale of @xmath181 , and is symmetric for all larger scales .
our model requires that there are lines of sight with projected radii @xmath182 mpc and extending from @xmath11 to @xmath183 , or @xmath184 mpc in terms of proper length , that are nearly devoid of mass . at the same time , these lines of sight are not nearly as empty in terms of galaxies . within standard theoretical framework , where primordial fluctuation power spectrum and subsequent gravitational instability are solely responsible for structure on large scales , a dramatic redistribution of baryonic and dark matter , such as implied by our model , is not possible
note however , that even though our model is astrophysically implausible , it does not violate the physical constraint that densities must remain positive everywhere : the sharp cut - off in @xmath180 corresponds to @xmath173 , the minimum possible lensing optical depth presented by mass between @xmath185 and @xmath183 to sources at @xmath186 , in a flat @xmath187 universe model .
since baryonic and dark matter are unlikely to be distributed differently on large scales , we propose an alternative , speculative scenario .
suppose @xmath188 of the present day closure density is contributed by some dark energy that , because of its equation of state became dynamically important only recently , say @xmath189 .
until @xmath190 structure formation proceeded according to the standard picture , with galaxies tracing the total mass , and with galaxy and mass distributions looking gaussian on large scales .
suppose also that the dark energy can cluster on sub - horizon scales , or , more specifically , on scales as small as @xmath191 mpc .
then at late epochs the dark energy will start contributing to the potential fluctuations that are already defined by dark and baryonic matter , on those scales . if these fluctuations develop quickly , then matter ( light and dark ) , hampered by inertia , would not be able to respond quickly .
thus , power spectra obtained from galaxy surveys would not betray anything unusual .
on the other hand , light rays from distant sources will traverse the fluctuations at @xmath192 , thereby ` capturing ' the total amplitude of the fluctuations .
so , lensing is able to probe the full extent of gravitational potential wells , and if these are deep , then significant source - lens ( anti-)correlations would result . in this picture
, @xmath180 distribution introduced earlier would refer not to the dark matter , but mainly to the projected clumping of dark energy .
assuming 3d clustering scale of @xmath191 mpc , lines of sight @xmath184 mpc long that are devoid of dark energy are possible .
given the speculative nature of the proposed scenario , we will not develop it any further . instead , we ask if the qualitative features of our model are compatible with the existing observations . by qualitative features we mean the relation between fractional galaxy excess and fractional dark energy excess ( fig .
[ smmodel_bulk ] ) , with biasing now redefined in terms of these two quantities .
specifically , do observations rule out strong anti - biasing for regions with small @xmath193 , biasing for regions with @xmath194few , and moderate anti - biasing or @xmath195 for all other regions . in 3d , this general behavior of the biasing function will be same as in projection . 0.05in_cosmic
microwave background .
_ according to the proposed scheme the dynamical evolution at very high redshifts is same as in the standard cosmological scenarios , so the primary cmb fluctuations would remain unaltered .
the dark energy fluctuations grow at @xmath189 , and are therefore expected to contribute to cmb as late integrated sachs - wolfe effect , at @xmath196 .
this location coincides with that of the first acoustic peak , and so the isw signature will be either masked by the peak , or enhance the peak s amplitude compared to that of the secondary peak .
this is not ruled out by cmb observations .
0.05in_galaxy redshift surveys .
_ because of the late emergence of dark energy fluctuations , and galaxies sluggishness in responding to these , the observed galaxy distribution at redshifts @xmath197 should be similar to what one would expect in standard cosmological models .
0.05in_cosmic velocity flows .
_ regions with @xmath198 would show significant anti - biasing according to our model .
such regions within a few hundred megaparsecs around us are studied by cosmic velocity flows , and in the linear regime yield values for @xmath199 @xcite .
there are three basic methods of reconstructing 3d mass density fields from data : using density - density comparison , velocity - velocity comparison , or redshift - space distortions .
these methods often produce discrepant results for @xmath199 , with density - density giving consistently higher values than velocity - velocity comparison @xcite .
for example , iras 1.2jy + mark iii data set yields @xmath200 using density - density technique @xcite , while @xmath201 is obtained using velocity - velocity comparison @xcite .
other data sets produce similar results .
@xcite argue that no reasonable biasing scheme would generate such different @xmath199 emerging from different methods , and ascribe the discrepancy to errors .
this may well be the case , however , the discrepancy could also arise if biasing is a strong function of total density excess , as proposed by our toy model . in that case ,
density - density comparison would yield most reliable results , because the method makes no assumptions about the underlying mass density field , but derives it from observed velocities .
density - density comparisons currently yield @xmath202 @xcite ; for @xmath8 this would imply @xmath203 , not too different from the average biasing factor of our proposed model , which is 0.62 .
thus our model is not necessarily in conflict with observed cosmic velocity flows .
0.05in_cosmic weak shear lensing .
_ another important observational test of our model is provided by the recent measurements of weak shear @xcite , which are consistent with @xmath128 flat cosmological models , @xmath195 . ] , and gaussian mass distribution on large scales . even though the total mass and dark energy distribution in our model is far from gaussian ,
our model and weak shear observations may not be discrepant ; we list three possibilities : _
( i ) _ the scales of weak shear and grb - apm anti - correlations are different by about a factor of 3 : the former apply to @xmath204 , while the latter are on scales @xmath205 , thus the two techniques probe different physical scales . _
( ii ) _ observational factors may play a role .
the redshifts of individual galaxies in the weak shear studies are not known , and so the lenses can not be cleanly separated from the sources . because of that , weak shear method may suffer from a ` signal dilution ' effect : a foreground overdense patch will produce coherent shearing in background galaxies , but will also dilute the shear signal because of the increased number of foreground unsheared galaxies . _
( iii ) _ in models with gaussian distributed matter rms values of convergence and shear are the same .
suppose the distribution of dark energy trapped in potential wells of large overdense regions is top - hat - like .
in that case the rms of convergence can be a few times larger than rms of shear , when averaged over large portions of the universe . thus the current weak shear observations may be underestimating the total amplitude of fluctuations .
we conclude that the qualitative features of our proposed model are not strongly ruled out by observations .
as discussed in the introduction , bright optically - selected and radio - selected qsos are known to be correlated with foreground galaxies on large angular scales .
the statistical significance of any one of these studies is generally @xmath206 , and can be dismissed as a chance occurrence , but collectively these studies imply that the correlations are real , in which case the most likely explanation is weak magnification lensing .
the purpose of this work was to see if cosmologically distributed sources other than qsos are affected by weak lensing .
we started by computing the expected magnification bias of grb events due to weak lensing , and found that most grb subsets limited by statistical positional error should show negative magnification bias , i.e. should be anti - correlated with the foreground mass .
using apm galaxies to trace the mass we then looked for any correlations on the sky between apm galaxies and grbs .
because the projected number density of grbs is low , about 1 per apm plate , we did not use cross - correlation analysis of counting pairs , but instead counted galaxies in circles around individual grbs .
we found that grbs with small positional errors are preferentially located in the directions where apm galaxies show @xmath125 deficits on degree angular scales .
in particular , a subset of 46 grbs with @xmath207 and a subset of 74 grbs with @xmath208 have 12% and 8% less galaxies in circles of @xmath117 radii around them ( or @xmath182 mpc at the redshift of the typical lenses ) , than expected on average .
these are significant at 99.7% and 97.6% confidence levels .
whereas these significance levels are less than overwhelming , what makes the finding especially interesting is that the grbs are _ anti - correlated _ with the apm galaxies , as expected from the magnification bias ( and opposite to what is expected if physical associations are significant ) .
this is the first reported case of weak lensing induced anti - correlations on degree angular scales . to test the observed anti - correlation of grbs with foreground mass
, we carried out a cross - correlation analysis between grb subsets classed by positional error and abell - corwin - olowin galaxy clusters of three distance ranges , d=1 - 4 , d=5 , and d=6 , and two richness subsets , r=0 - 2 , and r@xmath209 .
only one combination of grbs and clusters showed significant results : @xmath208 grbs and rich distant aco clusters are anti - correlated on @xmath159 scales , consistent with predictions in section [ underden ] , and observed anti - correlations with apm galaxies .
as was the case with earlier reports of qso - galaxy correlations , present anti - correlations with grbs can not be accounted for by weak lensing using a standard model of mass distribution and moderate biasing on large scales ; the discrepancy is rather severe , one needs to increase the lensing optical depth by 10 , or reduce the biasing factor to @xmath140 , which is not a viable option .
since both the grb - apm and grb - aco anti - correlation signals are of 2 - 3 @xmath210 significance , they could well be statistical flukes .
however , if they are not , and if the signal is due to weak lensing , then there are three types of resolutions : those that have to do with sources , lensing process , or lenses .
we argue that the first two are inadequate , leaving us with the third .
we propose a speculative scenario which assumes that @xmath188 of closure density is contributed by dark energy that can clump on significantly sub - horizon scales , @xmath211 mpc , and that this clumping developed recently , @xmath189 . then at present epochs
there should be significant fluctuations in the dark energy component .
however , the amplitude of baryonic and dark matter fluctuations would remain relatively unchanged , because it would take time for matter to be accelerated and displaced by significant amounts on large spatial scales . in this scenario ,
the galaxy distribution would trace the underlying potential wells but would severely underestimate the amplitude of fluctuations on @xmath212 mpc scales .
dynamical measurements would do better , since velocities ` respond ' quicker to acceleration than displacements .
lensing , which relies on light traversing the fluctuations at @xmath192 , is the best tool to probe the full extent of these fluctuations . in this scheme ,
weak lensing induced grb - apm anti - correlations would be strong because the total amplitude of gravitational potential fluctuations are substantially larger than galaxies would have us believe . in principle , weak shear lensing should be able to detect the true extent of these fluctuations as well .
however , weak shear observations support the standard cosmological model with cold dark matter and unclustered dark energy component , and seem to be in conflict with our proposed scenario .
current observations of microwave background anisotropy , galaxy redshift surveys and cosmic velocity flows are not strongly incompatible with our hypothetical scenario .
abell , g.o .
1958 , , 3 , 211 abell , g.o . ,
corwin , h.g .
& olowin , r.p .
1989 , , 70 , 1 bacon , d. , refregier , a. & ellis , r.s .
2000 , , 318 , 625 bartelmann , m. & schneider , p. 2001 ,
physics reports , 340 , 291 bartelmann , m. 1995 , a&a , 298 , 661 bartelmann , m. & schneider , p. 1994
, a&a , 284 , 1 bartelmann , m. & schneider , p. 1993 , a&a , 268 , 1 bartsch , a. , schneider , p. & bartelmann , m. 1997 , a&a , 319 , 375 berlind , a.a . ,
narayanan , v.k . &
weinberg , d.h .
2001 , apj , 549 , 688 bogart , r. s. & wagoner , r. v. 1973 , apj , 181 , 609 dekel , a. 1994 , ara&a , 32 , 371 djorgovski , s.g .
2001 , preprint , astro - ph/0107535 , to appear in proc .
`` gamma - ray bursts in the afterglow era : 2nd workshop '' , eds . n. masetti et al . ,
eso astrophysics symposia , berlin : springer verlag dolag , k. & bartelmann , m. 1997 , mnras , 291 , 446 fiore , f. 2001 , preprint , astro - ph/0107276 fiore , f. , nicastro , f. , savaglio , s. , stella , l. & vietri , m. 2000 , apj , 544 , l7 fugmann , w. 1990 , a & a , 240 , 11 gastanaga , e. & juszkiewicz , r. 2001 , apjl , 558 , l1 hoekstra , h. , yee , h.k.c . , gladders , m.d . ,
barrientos , l.f . ,
hall , p.b . &
infante , l. 2002 , preprint , astro - ph/0202285 hurley , k. , hartmann , d.h . ,
kouveliotou , c. , kippen , r. m. , laros , j. , cline , t. & boer , m. 1999 , apj , 515 , 497 irwin , m. , maddox , s.j . &
mcmahon , r.g .
1994 , spectrum , 2 , 14 + see also http://www.ast.cam.ac.uk/~mike/apmcat/ jain , b. , seljak , u. & white , s.d.m .
2000 , apj , 530 , 547 kaiser , n. , wilson , g. & luppino , g. 2000 , astro - ph/0003338 kolatt , t. & piran , t. 1996 , apjl , 467 , l41 lamb , d. q. & reichart , d. e. 2000 , apj , 536 , 1 landy , s.d . &
szalay , a.s .
1993 , apj , 412 , 64 maoli , r. , van waerbeke , mellier , y. , schneider , p. , jain , b. , bernardeau , f. , erben , t. & fort , b. 2001 , a&a , 368 , 766 marani , g.f . ,
nemiroff , r.j . ,
norris , j.p . &
bonnell , j.t .
1997 , apj , 474 , 576 mellier , y. 1999 , ara&a , 37 , 127 mellier , y. et al .
2001 , in eso proceedings `` deep fields '' , garching oct 9 - 12 , 2000 menard , b. & bartelmann , m. 2002 , a&a , 386 , 784 norman , d. & williams , l. l. r. 2000 , aj , 119 , 2060 norris , j. p. , marani , g. f. & bonnell , j. t. 2000 , apj , 534 , 248 paciesas , w.s .
et al . 1999 ,
apjs , 122 , 465 peacock , j.a . 2002 ,
preprint , astro - ph/0204239 peacock , j.a . &
dodds , s.j .
1994 , mnras , 267 , 1020 perna , r. & aguirre , a. 2000 , apj , 543 , 56 reichart , d.e . & lamb , d.q .
2001 , preprint , astro - ph/0103255 reichart , d.e . & lamb , d.q , fenimore , e. e. , ramirez - ruiz , e. , cline , t. l. & hurley , k. 2001 , apj , 552 , 57 retzlaff , j. , borgani , s. , gottlober , s. , klypin , a. & muller , v. 1998 , newa , 3 , 631 rodrigues - williams , l.l .
& hogan , c.j . 1994 , aj , 107 , 451 sanz , j. , martnez - gonzlez , e. , & bentez , n. 1997 , mnras , 291 , 418 seitz , s. , & schneider , p. 1995 , a & a , 302 , 9 sigad , y. , eldar , a. , dekel , a. , strauss , m. a. & yahil , a. 1998 , apj , 495 , 516 stern , b. , poutanen , j. & svensson , r. 1999 , apj , 510 , 312 tadros , h. , efstathiou , g. & dalton , g. 1998 , mnras , 296 , 995 tomita , k. , premadi , p. , & nakamura , t. t. 1999 , prog .
133 , 85 totani , t. 1999 , preprint , astro - ph/9805263 van waerbeke , l. , hamana , t. , scoccimarro , r. , colombi , s. & bernardeau , f. 2001 , mnras , 322 , 918 van waerbeke , l. et al .
2000 , a&a , 358 , 30 williams , l.l.r . 2000 ,
apj , 535 , 37 williams , l.l.r .
& irwin , m. 1998 , mnras , 298 , 378 willick , j.a . &
strauss , m.a .
1998 , apj , 507 , 64 wittman , d.m . ,
tyson , a.j . ,
kirkman , d. , dellantonio , i. & bernstein , g. 2000 , nature , 405 , 143 rccccc 0.5 & 1.52.5 & 0.51.0 & 46 & 448 & 98.01 + 1.0 & 0.02.5 & 0.51.0 & 88 & 732 & 97.53 + 1.0 & 0.52.5 & 0.51.0 & 81 & 689 & 97.15 + 1.0 & 1.52.5 & 0.51.0 & 46 & 448 & 99.63 + 1.0 & 1.52.5 & 9.5 - 10.0 & 6 & 448 & 97.44 + @xmath213 1.5 & 1.52.5 & 0.51.0 & 46 & 448 & 99.69 + @xmath213 1.5 & 1.52.5 & 0.01.0 & 74 & 448 & 97.62 + | we investigate whether gamma - ray bursts ( grbs ) from the current batse catalog have been affected by weak lensing by the nearby large scale structure .
the redshift distribution of grbs is believed to be broad , extending to @xmath0 , so most events can be assumed to be at large redshifts , and hence subject to weak lensing , which would betray itself as projected ( anti-)correlations between grb events and galaxies or clusters that trace the intervening mass . given the observed distribution of grbs in fluence @xmath1 , and statistical positional error @xmath2 , we predict that most subsets drawn from batse catalog will be anti - correlated with the foreground structure due to weak lensing , i.e. will show negative magnification bias .
we find that grbs are indeed anti - correlated with the apm galaxies ( @xmath3 ) in the sense that galaxy density in circles of radii @xmath4 ( @xmath5mpc at @xmath6 ) centered on @xmath7 grbs is about 10% lower than expected from a random distribution ; the significance of grb - apm anti - correlations reaches 99.7% .
cross - correlation between grbs and distant rich abell - corwin - olowin clusters is also negative .
standard cosmological models with @xmath8 , @xmath9 , and matter distribution on large scales following observed apm galaxy distribution with the biasing parameter of around 1 are not able to reproduce our grb - apm anti - correlations .
we propose a speculative model that does account for these anti - correlations as well as positive correlations found previously , between qsos and apm galaxies .
we briefly discuss if the proposed scheme is in conflict with observations of cosmic microwave background , galaxy surveys , cosmic velocity flows , and weak shear lensing . |
here we will consider a spin polarized ensemble of fermionic atoms ( e.g. @xmath180 ) with two accessible electronic degrees of freedom associated to the @xmath181 states .
we focus on the case where the atoms are trapped in an external potential @xmath182 that is the same for @xmath183 and @xmath184 ( i.e. at the magic wavelength " @xcite ) .
if the atoms are illuminated by a linearly polarized laser beam with bare rabi frequency @xmath29 they are governed by the following many - body hamiltonian @xcite @xmath185 here @xmath186 is a fermionic field operator at position @xmath187 for atoms with mass @xmath188 in electronic state @xmath189 ( @xmath11 ) or @xmath184 ( @xmath13 ) , while @xmath190 is the corresponding density operator . since nuclear spin polarized fermions are in a symmetric nuclear spin state ,
their @xmath5-wave interactions are characterized by only one scattering length @xmath63 , with the corresponding interaction parameter @xmath191 , describing collisions between two atoms in the antisymmetric electronic state .
the laser with frequency @xmath30 and wavevector @xmath192 is detuned from the atom transition frequency @xmath32 by @xmath31 .
we consider the situation in which a deep 2d lattice freezes the atomic motion in the transverse @xmath131 plane creating an array of one dimensional tubes . along the longitudinal @xmath193-direction
the net effect of the lattice is to induce a weak harmonic confinement with frequency @xmath144 . while the deep 2d lattice confines the atoms to the lowest vibrational mode , we allow mobility along the longitudinal direction .
it is then convenient to expand the field operator , @xmath194 , in a harmonic oscillator basis , @xmath195 , where @xmath196 and @xmath197 are , respectively , the transverse and the longitudinal harmonic oscillator eigenmodes and @xmath198 creates a fermion in mode @xmath27 and electronic level @xmath199 . following refs .
@xcite , we assume that the probe is slightly misaligned with a small component along the @xmath193-direction : @xmath200 with @xmath201 . defining @xmath202 ,
where @xmath203 are the lamb - dicke parameters , @xmath204 the corresponding harmonic oscillator lengths and @xmath205 are laguerre polynomials @xcite , laser induced sideband transitions can be neglected if @xmath206 and @xmath207 .
these conditions are satisfied in the experiment that operates in the resolved sideband regime , where the clock probe fourier - limited spectral resolution is 10 hz , @xmath208 ( 700 hz ) .
hence , @xmath209 can be rewritten in the rotating frame as where @xmath211 and @xmath212 @xmath213 , @xmath214 , and @xmath215 are single - particle energies . in the parameter regime where @xmath216 ( relevant for current lattice clock experiments performed at @xmath26k temperatures ) , to a very good approximation the leading interaction processes correspond to those ones in which vibrational
quantum numbers are exchanged during the collision , i.e. @xmath217 or @xmath218 . since these processes conserve the number of particles per mode , for an initial state with at most one atom per mode ( @xmath183-polarized state ) , it is possible to reduce @xmath209 to a spin-@xmath219 model . denoting @xmath220 the initially populated modes ,
the hamiltonian becomes @xmath221 here @xmath222 , where @xmath223 are pauli matrices in the @xmath55 basis , @xmath224 , and constant terms were dropped .
the quantity @xmath225 .
the rotational invariance of the interaction term in @xmath226 ( @xmath227 ) is key for understanding the basic features of the model . due to the rotational symmetry
the interaction term is diagonal in the collective angular momentum basis @xmath228 , satisfying @xmath229 and @xmath230 , with @xmath231 and @xmath232 . here
the extra label @xmath233 is required to uniquely specify each state .
the fully symmetric ( dicke ) @xmath79 states do not interact .
they are unique and the label @xmath233 can be omitted for them .
to proceed further , we use the fact that @xmath234 is a slowly varying function of @xmath235 . @xmath236 . in the regime ( @xmath237 )
where the occupied modes @xmath238 are sufficiently sparse for the behavior of @xmath234 to be dominated by its slowly varying part , we can approximate @xmath239 and @xmath240again constant terms have been dropped . under this approximation
it is possible to have an analytic treatment of the many - body dynamics . with this purpose in mind ,
it is convenient to go to a rotated basis and rewrite the hamiltonian as @xmath241 where @xmath242 , @xmath243 , @xmath244 , @xmath245 and @xmath246 the mean rabi frequency .
we will consider @xmath247 as our perturbative parameter . to zero order in it ,
the eigenstates of the hamiltonian are conveniently described in terms of angular momentum eigenstates in the rotated basis , @xmath248 ( the quantum number @xmath84 is conserved in rotations ) , satisfying @xmath249 and @xmath250 . at time @xmath251
all the atoms are in the @xmath183 state and thus initially @xmath79 .
non - zero @xmath252 induce transitions outside the @xmath79 manifold .
however , to first order in perturbation theory , the term proportional to @xmath252 can only induce transitions to states with @xmath90 due to its linear dependence on @xmath253 .
this implies that the knowledge of the eigenstates and eigenvalues within the @xmath254 manifolds is enough to characterize the perturbative dynamics .
the @xmath255 eigenstates are just the well known dicke states invariant under particle permutation .
they have energies given by @xmath256 .
since the initially prepared state is a fully polarized state in the old basis , in the rotated basis it corresponds to a superposition of @xmath255 states with amplitude probabilities determined by the wigner rotation matrices : the states with @xmath90 are the so called spin - wave states .
they can be written in terms of dicke states as : @xmath258 with @xmath259 .
these states have energy @xmath260 . from the energy it is clear that the population of these states will give rise to an interaction energy shift .
for the simple case @xmath76 described in the main text , there is a unique spin wave state which corresponds to the singlet state with energy @xmath261 . using those matrix elements one can show after some algebra that @xmath265 depends on @xmath266 only through the mean rabi frequency , @xmath267 , and the standard deviation of the rabi frequency , @xmath91 .
more explicitly @xmath268 where the superscript @xmath269 indicates a homogeneous excitation . @xmath270
note that since @xmath271 then @xmath272 does not have first order corrections in @xmath102 . in clock experiments based on rabi interrogation the clock frequency shift , cfs , @xmath96
is measured by first locking the spectroscopy laser at two points , @xmath95 , of equal height in the transition lineshape ( equal final excited state fraction under the initial condition of all atoms in state @xmath183 ) and then determining the change in the mean frequency as the interaction parameters or density are varied , @xmath97 . to calculate the shift ,
we taylor expand @xmath95 around the zero order values , @xmath273 , and to lowest nonvanishing order obtain @xmath274 with @xmath275}+t \sqrt{\delta ^2+\bar{\omega } ^2 } \sin{\left[t \sqrt{\delta ^2+\bar{\omega } ^2}\right]}\right)}{\left(\delta ^2+\bar{\omega }
^2\right)^2}\\ c&=\frac{2 \delta \omega ^2 \delta}{n^2 u^2 \left(\delta ^2+\bar{\omega } ^2\right)^3 \left(-n^2 u^2/4+\delta ^2+\bar{\omega } ^2\right)^2 } \notag \\ & \times \left ( 2 n u \bar{\omega } ^4 \left(\delta ^2+\bar{\omega } ^2\right)^2 - 2 n u \cos [ t n u/2 ] \bar{\omega } ^2 \left(-n u/2+\bar{\omega } \right ) \left(n u/2+\bar{\omega }
\right ) \left(\delta ^2+\bar{\omega } ^2\right)^2\right .
\notag\\ & + n^5 u^5 \bar{\omega } ^2 /32\left(-7
\delta ^2 + 2 \bar{\omega } ^2\right)+n^3 u^3/8 \cos\left[t \sqrt{\delta ^2+\bar{\omega } ^2}\right]^2 \bar{\omega } ^2 \left(-n^2 u^2/4+\delta ^2+\bar{\omega }
^2\right ) \left(\delta ^2 + 2 \bar{\omega } ^2\right ) \notag \\ & -n^3 u^3/8 \left(8 \delta ^6 + 5 \delta
^4 \bar{\omega } ^2 + 3 \delta ^2 \bar{\omega } ^4 + 6 \bar{\omega } ^6\right ) \notag \\ & + 2 n u \cos\left[t \sqrt{\delta ^2+\bar{\omega } ^2}\right ]
\left(-\cos [ t n u/2 ] \left(-n u/2+\bar{\omega } \right ) \left(n u/2+\bar{\omega } \right ) \left(\delta ^2+\bar{\omega } ^2\right)^2 \left(2 \delta ^2+\bar{\omega } ^2\right)\right.\notag \\ & + \left.\bar{\omega } ^2 \left(n^4 u^4 \delta ^2/32+\left(\delta ^2+\bar{\omega } ^2\right)^2 \left(2 \delta ^2+\bar{\omega } ^2\right)-n^2 u^2/4 \left(\delta ^2+\bar{\omega } ^2\right ) \left(5
\delta ^2+\bar{\omega } ^2\right)\right)\right)\notag\\ & + \sin\left[t \sqrt{\delta ^2+\bar{\omega } ^2}\right ] \left(- t n u \bar{\omega } ^2 \sqrt{\delta ^2+\bar{\omega } ^2 } \left(-n^2 u^2/4+\delta ^2+\bar{\omega } ^2\right ) \left(n^2 u^2/4 \left(\delta ^2 - 2 \bar{\omega } ^2\right)\right.\right . \notag\\ & \left.+2 \bar{\omega } ^2 \left(\delta ^2+\bar{\omega } ^2\right)\right)+4 \left(\delta ^2+\bar{\omega } ^2\right)^{5/2 } \left(n^4 u^4/16 + \bar{\omega } ^4+n^2 u^2/4 \left(\delta ^2 - 2 \bar{\omega } ^2\right)\right ) \sin [ t n u/2]\notag\\ & \left.\left .- n^3 u^3/8 \bar{\omega } ^2 \left(-n^2 u^2/4+\delta ^2+\bar{\omega } ^2\right ) \left(\delta ^2 + 2 \bar{\omega } ^2\right ) \sin\left[t
\sqrt{\delta ^2+\bar{\omega } ^2}\right]\right)\right).\end{aligned}\ ] ] so far we have assumed a fixed set of populated modes , @xmath277 . at finite temperature ,
expectation values need to be calculated by averaging over all possible combinations of modes @xmath278 weighted according to their boltzmann factor : @xmath279 , with @xmath280 .
the thermally averaged expression of the shift becomes grouping all temperature dependent terms in a temperature dependent coefficient , we obtain the following scaling behavior of the shift : @xmath282 in the weakly interacting regime consistent with prior mean field analysis @xcite and @xmath283 in the strongly interacting regime . [ [ estimation - of - the - tunneling - rates - under - current - experimental - conditions ] ] estimation of the tunneling rates under current experimental conditions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ here we discuss the tunneling rates in the weakest lattice potential which is along the @xmath284 direction .
the trap frequency is @xmath285 khz which corresponds to a lattice depth of @xmath286 with @xmath287 the photon recoil energy and @xmath288 the atomic mass .
the energy levels of a sinusoidal lattice potential , @xmath289 , admit analytic solutions in terms of mathieu functions and their characteristic values .
we use those to compute the band structure . in fig .
[ bandstr ] we plot the dispersion relation showing the five relevant bound states .
in addition we write the corresponding tunneling energy , @xmath290 for each band in units of @xmath291 .
we evaluate the thermal average of the tunneling energy using a boltzmann distribution and obtain @xmath292 for experimental temperatures of @xmath293k respectively which correspond to tunneling times of @xmath294 ms . in the experiment in addition to the lattice potential along @xmath284
, the atoms feel the dipole potential generated by the @xmath295 lattice as well as the gravitational force . the total external potential along the @xmath284 direction is approximately given by : @xmath296 with @xmath297 hz and @xmath298 m the beam waist .
the additional parabolic potential creates an energy offset between adjacent lattice sites which tends to suppress tunneling . for a mean
energy offset @xmath299 between two adjacent lattice sites the effective tunneling rate becomes @xmath300 . due to the harmonic nature of the confinement @xmath299 depends on the actual position in the lattice .
for two adjacent lattice sites , at @xmath301 and @xmath302 respectively , the mean energy offset is @xmath303 with @xmath304 the lattice spacing . as mentioned in the main text
the loading procedure of the atoms is such that 100 rows of tubes are approximately uniformly distributed along @xmath284 .
this is consistent with an average @xmath305 , i.e @xmath306 . using this value we estimate @xmath307 for experimental temperatures of @xmath293k respectively which corresponds to effective tunneling times of @xmath308 ms .
these results demonstrate that for a typical interrogation time of the order of @xmath309 ms , tunneling events can be neglected , especially for the @xmath310k situation which was the temperature at which most data points were taken ( see main text ) .
interactions can modify the effective tunneling rates .
however , in typical cold atomic systems interacting via @xmath5-wave collisions these effects are so weak that can be generally ignored .
the role of interaction assisted tunneling effects has been explicitly investigated in a recent experiment @xcite .
this experiment has been the only one capable of observing super - exchange interactions at energy scales as low as 5 hz .
nevertheless , regardless of the high degree of experimental control and precision , the experimental data was not able to observe any deviation of the single - particle tunneling due to interactions compatible with the uncertainties of the lattice depths .
furthermore , recent theoretical efforts have predicted that only close to a feshbach resonance , interaction assisted tunneling terms can start to play a role , see ref@xcite .
this happens when the mean interaction energy of the gas becomes larger than the vibrational energy spacing . in the experiment described above ,
the density shift was determined by measuring a differential shift of the sr clock transition as the number of atoms loaded into the 2d lattice was varied .
the sample density was alternated between a high and a low value every two cycles of the experiment , as two interrogations of the clock transition were required to obtain one estimate of the transition center frequency .
the sequential frequency data points are indexed by @xmath314 , which runs from @xmath315 to @xmath316 , where @xmath317 is the total number of frequency points in a data run .
the correlation between the clock transition frequency and the sample density was determined by analyzing consecutive groups ( strings " ) of frequency data , of length @xmath27 .
if for all odd @xmath314 the density was at the low value , and for all even @xmath314 the density was high , the correlated frequency shift is @xmath318 we then perform the same string analysis on the associated sample densities and normalize each string point to obtain the collisional frequency shift per unit density .
a similar analysis was introduced in @xcite , and this method is equivalent to removing a polynomial of order @xmath319 from each @xmath27-point sequence of data . for a string of length @xmath320 , this simply corresponds to isolating the differential shift between the high and low density data , while for a string of length @xmath321 , a linear drift is removed .
as mentioned in the main text , we used strings of length @xmath322 to analyze our data , thereby eliminating sensitivity to both linear and quadratic drifts .
this drift removal is necessary because of the residual drift in the ultrastable fabry - perot cavity that served as our frequency reference . for each data run , we report the mean and standard error of the string points . as the data strings overlap with one another
, they are not completely independent , and a correction factor @xmath323 must be applied to the standard error of the string points .
stated in other words , without this factor there are too many points " and dividing the standard deviation of the data set by the square root of the number of measurements will underestimate the error . a single data run consisted of a period of continuous lock to the clock transition , and
therefore there was a significant variation in the lengths of individual runs .
this variation in run length is partly responsible for the variation in the errors of the data runs .
we applied a cut at the start of each data run to allow sufficient time for the clock laser servo to acquire a stable lock to the atomic transition .
all data runs corresponding to a given set of of experimental conditions were combined to give an estimate of value and the uncertainty of the frequency shift ( see fig .
[ fig5 ] in the main text ) .
the shift value was obtained by taking the error - weighted mean of all the data runs , and the uncertainty was estimated from the variance of the weighted mean .
we calculated a @xmath324 per degree of freedom , or a reduced @xmath324 ( @xmath325 ) for each set of data runs , and scaled the final uncertainty by @xmath326 .
we analyzed the data using several different protocols , each of which give results that are consistent within the stated uncertainties .
below , we briefly describe each protocol and present the shift each method calculates for the data records presented in fig .
[ fig5 ] of the main text : * considering all of the data as one single unit , without regard to unlocks or different days on which the data was taken . in this case , the central value of the result is taken to be the simple mean of the data , and the uncertainty is estimated by the standard error scaled by a factor to account for the overlap of the string points . for this method
we find @xmath327 for the 3.5 @xmath26k data , and @xmath328 for the 7 @xmath26k data . * separating the data into bins of a set number of string points and calculating the simple mean of each bin .
the error bars for each bin are determined from the standard error of the bin and scaled by the correction factor .
this data is then used to calculate the weighted mean and error of the weighted mean for the entire data record .
a reduced chi - square statistic was also calculated and the bin size is adjusted to achieve a reduced chi - square of @xmath329 for each data record .
this method has the advantage that it weights data according to actual scatter of the measurement points so that data with greater signal to noise is weighted more heavily . with this method ,
we are also able to consistently compare our longer data records to data records where the data was acquired over very few continuous locks to atoms and even fewer days ( making the two following protocols inappropriate ) . in an effort to present all of our data using a consistent method of analysis , all data point in fig .
[ fig4 ] of the main text are calculated using this method . when analyzed with this method ,
the data records give @xmath330 for the 3.5 @xmath26k data , and @xmath331 for the 7 @xmath26k data ( both @xmath332 ) .
* breaking the data into data runs by day ( where each day of data - taking includes multiple segments of continuous lock ) .
error bars for each data run are determined from the standard error of the run measurements and scaled by the correction factor .
simple means of each data run were used to determine the central value of the shift for that run . the weighted mean and
the error of the weighted mean were determined from these data , and a reduced chi - square statistic was calculated .
we then scale the error by the square - root of the chi - square value . for this method
the data records give @xmath333 with a @xmath334 for the 3.5 @xmath26k data , and @xmath335 with a @xmath336 for the 7 @xmath26k data ( error unscaled ) . *
similar to the previous method , but data runs are taken to be periods of continuous lock to the atoms rather than days of data taking .
this is the method used in fig .
[ fig5 ] of the main body of the paper . for this method
we find @xmath337 with a @xmath338 for the 3.5 @xmath26k data , and @xmath339 with a @xmath340 for the 7 @xmath26k data ( error unscaled ) .
we used several techniques to characterize important experimental parameters such as the sample temperature along the strongly- and weakly - confined directions in the 2d lattice , the lamb - dicke parameter , and the trapping frequencies .
the temperature along the weakly - confined direction ( @xmath19 ) was established by fitting the width of the doppler broadened spectroscopic feature obtained when the clock laser propagated along @xmath19 .
temperatures along the strongly - confined directions were determined via sideband spectroscopy , as described in @xcite and the main text .
the trap frequencies along @xmath17 and @xmath18 both vary as the horizontal lattice power is varied .
this occurs because the two lattice beams are not orthogonal but instead cross at an angle of @xmath341 .
the eigenaxes of the resulting 2d harmonic potential therefore are a function of the power in both of the two lattice beams .
this effect was properly taken into account when the trap confinement and transverse lamb - dicke parameter were determined . | optical lattice clocks have the potential for extremely high frequency stability owing to the simultaneous interrogation of many atoms , but this precision may come at the cost of systematic inaccuracy due to atomic interactions .
density
dependent frequency shifts can occur even in a clock that uses fermionic atoms if they are subject to inhomogeneous optical excitation @xcite . here
we present a seemingly paradoxical solution to this problem . by dramatically increasing the strength of atomic interactions , we suppress collisional shifts in lattice sites containing @xmath0 atoms ; strong interactions introduce an energy splitting into the system , and evolution into a many - particle state in which collisions occur
is inhibited .
we demonstrate the effectiveness of this approach with the jila sr lattice clock by reducing both the collisional frequency shift and its uncertainty by more than a factor of ten @xcite , to the level of @xmath1 .
this result eliminates the compromise between precision and accuracy in a many - particle system , since both will continue to improve as the particle number increases . strongly interacting quantum systems
can exhibit counterintuitive behaviors , under both equilibrium and non - equilibrium conditions .
for example , in a multi - component fermi degenerate gas , a frequency shift of a microwave transition remains finite close to a feshbach resonance @xcite . in low dimensions , the effective strength of atomic interactions can be significantly enhanced and as a result , particles tend to avoid each other to minimize their total energy
. this tendency can lead to behavior that in many aspects resembles that of non - interacting systems .
one such example is the tonks girardeau regime of an ultracold bose gas , in which the strong repulsion between particles mimics the pauli exclusion principle , causing the bosons to behave like non - interacting fermions @xcite .
here we describe how the enhancement of atomic interactions in a strongly interacting , effectively one - dimensional ( 1d ) system suppresses collisional frequency shifts in an optical atomic clock , and we characterize the density - dependent shift to an unprecedented level of precision . the @xmath2sr optical lattice clock at jila has reached an overall fractional frequency uncertainty of @xmath3@xmath4 @xcite .
this uncertainty is dominated by two contributions : atomic collisions and frequency shifts due to room temperature blackbody radiation .
the density - dependent shift arises from collisions between fermionic atoms that are subject to slightly inhomogeneous optical excitation @xcite , and several theories of the underlying frequency shift mechanism have been proposed @xcite . by tightly confining atoms in an array of quasi-1d potentials formed by a two - dimensional ( 2d ) optical lattice , we significantly increase the strength of atomic interactions , to the point where the thermally averaged mean interaction energy per particle becomes the largest relevant energy scale . although our system does not meet the conventional definition of the strongly interacting regime ( generally , that the interaction energy exceeds the system temperature ) , the interaction energy per particle is nevertheless larger than any other relevant dynamical energy scale , and thus the system is effectively strongly interacting . in this regime
, collisions are suppressed because evolution into a many - particle state in which @xmath5-wave scattering can occur is energetically unfavorable .
collisional frequency shifts could also be suppressed by confining atoms in a three - dimensional ( 3d ) lattice with filling factor less than or equal to one per lattice site .
however , vector and tensor shifts of the optical clock transition are a serious concern with a three - dimensional lattice clock using fermions @xcite .
a 3d lattice clock using bosonic @xmath6sr has been demonstrated @xcite , and its collisional shift was characterized at the level of @xmath7 . in atoms without hyperfine structure ,
the @xmath8 clock transition is completely forbidden , and therefore additional state mixing techniques must be used to enable it in bosonic sr isotopes .
these techniques result in sizable systematic shifts of the clock frequency that must be carefully controlled .
the work presented here will allow operation of a fermionic lattice clock with a filling factor much greater than one and a greatly reduced sensitivity to collisional effects .
our experiment employs ultracold fermionic @xmath2sr atoms that are nuclear spin - polarized ( e.g. , @xmath9 , @xmath10 ) .
an ultranarrow optical clock transition , whose absolute frequency has been precisely measured @xcite , exists between the ground @xmath11 @xmath12 and excited metastable @xmath13 @xmath14 states .
atoms are trapped in a deep 2d optical lattice at the magic wavelength where the ac stark shifts of @xmath15 and @xmath16 are matched @xcite .
the 2d lattice provides strong confinement along two directions ( @xmath17 and @xmath18 ) , and relatively weak confinement along the remaining dimension ( @xmath19 ) .
atoms in the lattice are sufficiently cold ( @xmath20k ) that they primarily occupy the ground state of the potentials along the tightly confined directions , with trap frequencies @xmath21 @xmath375 - 100 khz and @xmath22 @xmath345 - 65 khz .
this creates a 2d array of isolated tube - shaped potentials oriented along @xmath23 , which have trap frequencies @xmath240.55 - 0.75 khz .
we estimate that approximately 20% of these lattice sites are occupied by more than one atom . at a typical axial temperature @xmath25 of a few @xmath26k , various axial vibrational modes @xmath27
are populated in each tube . in a clock experiment
the @xmath28 transition is interrogated using rabi spectroscopy .
the atom laser coupling is characterized by the bare rabi frequency @xmath29 , which is defined in the absence of any motional effects .
the optical frequency @xmath30 is detuned by an amount @xmath31 from the atomic resonance at @xmath32 . as described in refs .
@xcite , any small projection of the probe beam along @xmath23 leads to a slightly different rabi frequency @xmath33 for each mode @xmath34 , where @xmath35 . here
@xmath36 is the lamb - dicke parameter , @xmath37 is the harmonic oscillator length , @xmath38 is atomic mass , and @xmath39 represents a small component of the probe laser wave vector along @xmath23 , resulting in a typical @xmath40 0.05 . to gain insight into the origin of the collisional frequency shift and the interaction - induced suppression , we consider a model system : two fermionic atoms , each of whose electronic degrees of freedom form a two - level ,
pseudo - spin 1/2 system ( @xmath15 and @xmath16 ) , confined in a 1d harmonic oscillator potential ( a fully many - body treatment for an arbitrary number of atoms @xmath41 is discussed later in the text and developed in the supplementary information ) .
the internal degrees of freedom of these two identical fermions can be expressed using a collective - spin basis , comprised of three pseudospin - symmetric triplet states and an antisymmetric singlet state @xcite . because the atoms are initially prepared in the same internal state @xmath42 , with their internal degrees of freedom symmetric with respect to exchange
, the pauli exclusion principle requires that their spatial wave function be antisymmetric .
these atoms thus experience no @xmath5wave interactions . if the atoms are coherently driven with the same rabi frequency ( @xmath43 = @xmath44 ) , their electronic degrees of freedom remain symmetric under exchange .
consequently , these atoms will not experience any @xmath5wave interactions during the excitation of the clock transition .
however , if @xmath45 = ( @xmath46 - @xmath47)/2 is not zero , the optical excitation inhomogeneity can transfer atoms with a certain probability to the antisymmetric spin state ( singlet ) that is separated from the triplet states by an energy @xmath48 , since in this state atoms do interact .
this interaction energy is what gives rise to a clock frequency shift during rabi interrogation @xcite . ) .
a weak excitation inhomogeneity characterized by @xmath49 is capable of producing triplet - singlet mixtures , causing a collisional frequency shift proportional to the interaction strength @xmath48 .
( b ) in the 2d lattice , the interaction energy exceeds the atom - light rabi frequency , creating an energy gap between the spin triplet and singlet states .
a small excitation inhomogeneity can not overcome the energy cost required to drive the transition .
evolution into the singlet state is inhibited and the collisional frequency shift is suppressed .
( c ) quasi-1d tube - like optical potentials formed by two intersecting optical lattices .
the laser which interrogates the clock transition propagates along @xmath18 , the vertical axis . ] figure [ fig1 ] contrasts the current 2d lattice experiment with prior studies carried out in a 1d lattice @xcite . in a 1d lattice ,
the interaction energy @xmath48 is typically smaller than @xmath50 ( the energy spread of the driven triplet states at zero detuning ) .
consequently , any small excitation inhomogeneity @xmath45 can efficiently populate the singlet state . by tightly confining atoms in a 2d lattice
, one can reach the limit where @xmath51 . in the presence of this large energy splitting
, evolution into the singlet state is inhibited , and as a result the collisional frequency shift of the clock transition is suppressed . in this regime ,
the singlet state can only participate as a virtual " state in second - order excitation processes and the frequency shift scales as @xmath52 . in this limit
, the energy - carrying singlet resonance has been shifted so far from the triplet resonances that it is completely resolved from them , and the remaining line - pulling effect can be negligible . such behavior is reminiscent of the dipolar blockade mechanism in a rydberg atom gas @xcite , where the interaction energy between an excited atom and its neighbors prevents their subsequent excitation .
the two - particle behavior described above can be generalized to an @xmath41-particle system ( @xmath53 ) .
the many - body hamiltonian can be quantitatively modeled by defining a set of effective spin operators , @xmath54 , in the @xmath55 basis . here
the subscript @xmath56 specifies the vibrational mode .
the description of the system in terms of effective spin operators is valid provided those initially populated modes remain singly occupied by either a @xmath15 or an @xmath16 atom during the excitation process .
the latter condition is satisfied in the lamb - dicke regime , @xmath57 , where one can neglect laser - induced inter - mode transitions . to evaluate thermally averaged quantities , we restrict the calculations to a fixed set of initially populated modes @xmath58 and sum over all possible mode configurations , weighted by the corresponding boltzmann factors . under the rotating wave approximation ,
the hamiltonian of the system becomes ( see the supplementary information ) , @xmath59 @xmath60 are collective spin operators .
the quantity @xmath61 measures the strength of the interactions between two atoms in the antisymmetric electronic state .
the interaction parameter @xmath62 , where @xmath63 is the singlet @xmath64 scattering length . since the volume @xmath65 , @xmath66 , where @xmath67 is the mean transverse trapping frequency .
@xmath68 is a mode overlap coefficient and its thermal average over all @xmath56 yields an interaction parameter @xmath69 .
note that @xmath70 is directly proportional to @xmath71 and thus increases with the transverse confinement .
the spin rotational invariance of the interaction term in @xmath72 is key to understanding the basic physics @xcite . due to the rotational symmetry , the interaction term is diagonal in the collective spin basis @xmath73 , @xmath74 and @xmath75 . for @xmath76 ,
the spin basis is spanned by the triplet states @xmath77 and the singlet @xmath78 . among the collective states only the @xmath79 states are noninteracting . states with @xmath80 experience an interaction energy shift . for a homogeneous excitation ( @xmath81 , where @xmath82 is the mean rabi frequency ) , the hamiltonian commutes with @xmath83 and
thus @xmath84 is a conserved quantum number .
the eigenstates of the system are just the original collective spin states up to a rotation around the @xmath85 axis ( for @xmath76 the eigenvalues are @xmath86 for @xmath87 and @xmath88 for @xmath89 ) .
if the system is prepared in the @xmath79 manifold it will never experience any interaction effects , and there will be no collision - induced frequency shift . in the presence of excitation inhomogeneity
, @xmath84 is no longer conserved . during excitation of the clock transition
, atoms will be transferred mainly between @xmath79 and @xmath90 states and will consequently experience a collisional frequency shift . as demonstrated in the supplementary information ,
the collisional shift experienced by atoms in multiply occupied lattice sites ( @xmath53 ) remains suppressed as @xmath41 increases .
it also increases with @xmath91 .
we note that detuning inhomogeneity ( for instance , atoms occupying different vibrational modes may have different resonance frequencies , @xmath92 = @xmath93 ) has an effect similar to atom - laser coupling inhomogeneity , and the resulting shift can be modeled using eq .
[ many ] . in our clock experiments
, the frequency of the laser which interrogates the @xmath94 @xmath13 transition ( the clock laser ) is modulated to probe the atomic resonance at two points @xmath95 on opposite sides of the line center .
the mean frequency of the clock laser is steered to achieve equal population of the excited state at these two interrogation frequencies .
the collisional frequency shift @xmath96 is given by the change in the mean frequency as the particle density is varied .
we determine the transition lineshape by evaluating the thermal expectation value of the excited state population as a function of detuning ( calculated from eq .
[ many ] ) , and compute the shift for a given excitation fraction as @xmath97 .
a summary of these results is presented in fig .
[ fig2 ] , which demonstrates the suppression of the collisional frequency shift as the interaction energy is increased . .
as @xmath98 increases , a larger @xmath70 is required for clock shift suppression . here
the temperature along @xmath23 was set to @xmath99 = 6.5 @xmath26k , the axial lamb - dicke parameter to @xmath100 and the laser detuning was fixed to achieve @xmath101 fractional population in the excited state @xmath16 .
the subfigures are a ) a three - dimensional plot and b ) a contour plot of the same theoretical function . ]
a qualitative understanding of the suppression can be developed when we calculate the frequency shift by treating @xmath102 as a perturbative parameter ( see supplementary information for details ) .
the perturbative analysis predicts , in the weak interacting regime of @xmath103 , a shift that scales linearly with @xmath70 , @xmath104 . here
@xmath105 is a temperature - dependent coefficient with contributions from @xmath102 , @xmath106 , and the mode - overlap coefficient @xmath107 .
we also note that @xmath108 .
this behavior is consistent with the standard mean - field expression of the density shift @xcite .
however , in the strong interacting regime of @xmath109 , the shift is suppressed as @xmath110 .
here @xmath111 again includes the temperature - dependent effects .
the suppression is consistent with the idea that in this regime the shift arises from a process that involves virtual occupations of non - fully symmetric states .
the suppression becomes less effective if @xmath112 becomes comparable to @xmath70 , or when @xmath102 increases at larger temperatures .
these considerations imply that clock experiments based on ramsey interrogation will not easily satisfy the suppression conditions outlined here , since the short pulses applied in the ramsey scheme generally have a rabi frequency more than ten times larger than those used in rabi spectroscopy . to prepare the atomic system
, we laser cool @xmath2sr atoms to about @xmath113k inside a magneto - optic trap based on the weak @xmath114 transition ( linewidth 7.6 khz ) , and then load them into a 1d vertical lattice ( along @xmath115 ) which overlaps with the mot .
the spatial distribution of occupied 1d lattice sites is determined by the vertical extent of the mot cloud , which is approximately gaussian with a standard deviation @xmath116 m .
we then adiabatically ramp up the horizontal lattice ( along @xmath117 ) to load the atoms into a 2d lattice .
the frequencies of the two lattice beams are offset from one another by 200 mhz to eliminate interference effects . to remove any atoms trapped in the 1d vertical lattice outside of the 2d intersection region , we ramp the vertical lattice off and then back on again .
the number of horizontal lattice sites occupied is then determined by the radial temperature of the vertical lattice along @xmath18 .
100 rows " of tubes are approximately uniformly distributed along @xmath18 , while the columns " distributed along @xmath17 are loaded according to a gaussian distribution with standard deviation @xmath118 of 6 - 10 @xmath119 m . after forming the 2d lattice ,
we perform doppler and sideband cooling using the @xmath120 transition .
simultaneously , atoms are optically pumped to the @xmath10 ground state sublevel , using @xmath121-polarized light on the @xmath122 transition , directed along a bias magnetic field parallel to the @xmath19 axis .
we determine the nuclear spin purity of the atomic sample to be greater than @xmath123 by scanning the probe laser over the clock transition frequencies for other nuclear spin states .
we perform spectroscopy of the clock transition using a narrow linewidth laser propagating along @xmath115 .
the clock laser and both lattice beams are linearly polarized along @xmath23 . after cooling for about 30 ms , the sample temperature in the tightly - confined transverse dimension
is lowered to 2 2.5 @xmath26k .
we determine @xmath124 by performing doppler spectroscopy along @xmath23 and we vary @xmath124 between 3 and @xmath125k by applying additional doppler cooling .
trap frequencies along all three directions are determined via sideband spectroscopy and studies of parametric resonance .
tunneling between lattice sites is suppressed because the potential gradient due to the dipole force lifts the degeneracy between lattice sites along both the @xmath17 and @xmath18 directions . for the lattice with the lowest depth ( along @xmath18 ) we estimate that the mean tunneling time is 150 ms for @xmath126 and @xmath127 of @xmath128k ( details of this calculation
are presented in som ) .
therefore , for our typical spectroscopy time of 80 ms , site - to - site tunneling can be neglected .
we also note that we measure a frequency shift as a function of number of atoms loaded into the lattice , with all other lattice parameters kept constant , and thus our measurements are insensitive to effects that do not depend on @xmath41 .
we quantify the number of atoms loaded into the 2d lattice by detecting fluorescence on the strong @xmath129 transition at 461 nm . with a total of @xmath130 atoms loaded into the 2d lattice
, we estimate that 2030% of lattice sites are multiply occupied .
khz and the sidebands are appreciably excited .
the two distinct sets of sidebands reflect trap frequencies in the tightly confined @xmath131 plane .
the features at detunings of @xmath132 khz are the first - order blue and red sidebands of the horizontal lattice . due to experimental constraints ,
the horizonal lattice is not strictly perpendicular to @xmath18 , and hence the clock laser ( along @xmath18 ) also excites these sidebands . ]
we build upon the method of extracting temperature information from vibrational sideband spectra that was derived for a 1d lattice @xcite , and develop a model for the sideband features that we observe in a 2d lattice .
these features can be understood in a fairly detailed fashion and our model reasonably describes the observed spectral features , as shown in fig .
[ fig3 ] . in the 1d lattice ,
all lattice sites are approximately equivalent , as the lattice s rayleigh range is much larger than the spatial extent of the mot cloud from which the lattice is loaded ; in the 2d lattice this scenario no longer holds , as lattice sites near the wings of the lattice beams gaussian intensity profiles are significantly shallower than those near the center of the beam intersection region . only after taking the distribution of site depths into account
does the axial temperature extracted from the sideband model become comparable with that determined from doppler spectroscopy along @xmath23 .
the width of the sideband features is a result of the coupling between the longitudinal and transverse degrees of freedom in the lattice @xcite . in the 2d lattice ,
the sideband shape is also affected by the distribution of trap depths .
the spread of trap frequencies accounts for the broadened sideband lineshape without requiring an exaggerated temperature along @xmath23
. the sharper edges of the sidebands at the largest detunings from the carrier are due to atoms trapped near the center of the beam intersection region , where the trap depths are greatest .
spectroscopy of the clock transition is performed with an 80-ms pulse , resulting in a fourier - limited linewidth of @xmath133 hz .
the clock laser is locked to the atomic resonance . the high finesse fabry
perot cavity @xcite used to narrow the clock laser s linewidth is sufficiently stable over short time scales that it can be used as a frequency reference in a differential measurement scheme @xcite . a single experimental cycle ( e.g. , cooling and trapping atoms , preparing the 2d lattice , and interrogating the clock transition ) requires about 1.5 s , and we modulate the sample density every two cycles .
the corresponding modulation of the atomic resonance frequency relative to the cavity reference is a measurement of the density shift .
when the sample density is varied , the spatial distribution of atoms in the 2d lattice might change . as lattice sites near the center and at the edge of the beam intersection region have different trap depths , this could allow ac stark shifts to contaminate our measurements , unless the lattice beams are tuned to the magic wavelength @xcite . in the experiment
we stabilize the two lattice frequencies to be @xmath134 mhz and @xmath135 mhz from the experimentally determined magic wavelength @xcite to minimize this effect .
we set a limit on this potential systematic by introducing a large frequency offset between the two lattices ; with a 2 ghz frequency offset no difference in the density shift was observed with an uncertainty of @xmath136 , and we therefore estimate that with the 200 mhz offset any potential corruption of our collisional shift measurements is at or below the @xmath137 level .
the 200 mhz separation used for the data presented here was chosen for reasons of experimental convenience , and in a full implementation of a 2d lattice clock both beams could be tuned much closer to the magic wavelength to reduce the associated ac stark shift . .
for both the main figure and the inset , the x - axis is plotted under the assumption of @xmath138 .
we varied three important parameters : intensity of the horizontal lattice beam , axial temperature and bare rabi frequency .
in the inset we explicitly show the experimental data with @xmath124 indicated on top of each point .
the back and blue colored symbols were taken at @xmath29 and the green colored symbols at @xmath139 and twice the interrogation time yielding a constant rabi pulse area .
we also use squares and triangles to distinguish two different sets of data points measured under slightly different lattice configurations ( different beam waist ) . in the inset
the corresponding spin model predictions are also displayed ( red open circles ) using a singlet @xmath64 scattering length , @xmath140-@xmath141 , and @xmath76 .
each theory point is calculated using the @xmath124 and @xmath29 at which experimental data were taken , @xmath142 .
the value of @xmath70 was varied by changing the horizontal lattice intensity , @xmath143 , which also modified @xmath144 , @xmath145 , and the lamb - dicke parameter .
the theoretical curves were scaled by the fraction of the atomic population in doubly occupied lattice sites .
the variation of @xmath144 and @xmath70 with @xmath143 was explicitly taken into account in theory which used @xmath146 and @xmath147 khz at the point with the smallest collisional shift .
the data confirm three trends in the prediction : the collisional shift @xmath148 decreases with increasing @xmath70 at similar temperature and trapping conditions , @xmath148 increases with increasing @xmath29 at similar temperature and trapping conditions , and @xmath148 decreases with smaller @xmath124 . to overcome the difficulty of comparing the shift at different experimental conditions we rescaled the points by a factor of @xmath149 with @xmath150 khz and @xmath151 @xmath152 .
the main plot shows the good agreement between the theoretical curve of @xmath153 in red solid line and rescaled experimental data . ]
we perform measurements at several trap depths to directly observe the interaction - induced suppression of the collisional frequency shift . to access different interaction energies ,
we vary the intensity of the horizontal lattice beam ( @xmath143 ) , which results in the change of mainly @xmath154 but also @xmath145 and @xmath144 .
the change in @xmath145 arises from the fact that the laser beams that create the two lattices are not orthogonal but instead at an angle of @xmath155 .
the change in @xmath144 is due to the gaussian profile of the beams .
since @xmath156 , an increase of the horizontal beam power leads to a monotonic increase of @xmath70 .
we observe a significant decrease of the collisional shift with increasing horizontal lattice power , as shown by the data points ( filled black squares and blue triangles ) in fig .
[ fig4 ] ( inset ) .
squares and triangles indicate data taken with slightly different beam waists . we have also studied the dependence of the collisional shift on the rabi frequency used to drive the clock transition .
@xmath29 was increased by a factor of two , and the interrogation time was decreased by 2 , yielding a constant rabi pulse area . under these conditions , we observe that the collisional shift under similar temperature and trapping conditions increases sharply ( green open square and green open triangle in fig .
[ fig4 ] ( inset ) ) , confirming that the shift suppression mechanism will not operate effectively for short , higher rabi frequency pulses .
this behavior is inconsistent with a frequency shift due to tunneling between lattice sites since points at similar trapping conditions exhibit substantially different shifts .
also shown in the inset of fig .
[ fig4 ] is the shift predicted by the spin model , assuming two atoms per lattice site .
the theoretical points are scaled by the fraction of the atomic population in doubly occupied lattice sites .
the red open circles colors are the theory results , @xmath142 with @xmath157 , obtained at different temperatures and trapping frequencies corresponding to the actual experiment conditions under which the data were taken . the data are consistent with the modeled shift , assuming @xmath63 @xmath158@xmath141 ( where @xmath159 is the bohr radius ) , @xmath146 , and @xmath147 khz at the point with the smallest collisional shift .
these parameters correspond to a value of @xmath160 for this case .
however , we note that @xmath63 has not been measured independently , and therefore the interaction energy can not be accurately estimated from the known trapping frequencies . since the temperature and trapping conditions substantially varied for different experimental data points ,
some scaling is required to make direct comparisons between data in fig .
[ fig4 ] ( inset ) and the behavior predicted in fig .
[ fig2 ] . to help visualization of the experimental confirmation of the interaction suppression mechanism , we rescaled the measured experimental values of the shift by a factor extracted from the theoretical model @xmath149 .
fig .
[ fig4 ] shows that after rescaling all data points lie very close to the theoretical curve of fractional frequency shift _ vs. _ @xmath70 at constant @xmath150 khz and @xmath161 = 3.5 @xmath26k .
the sign of the observed shift is negative , i.e. , an increased sample density shifts the atomic resonance to lower frequencies.previous studies of the collisional shift in a 1d optical lattice @xcite are consistent with this observation .
the simple mean - field analysis used in @xcite indicated a negative scattering length , but a more sophisticated many - body treatment @xcite showed that the experimental data were also consistent with a positive scattering length . from the present data set we can unambiguously conclude that @xmath162 is negative .
sr atoms confined in a 2d optical lattice .
each point represents a data set collected from a continuous operation of the sr clock , with error bars determined from the standard error of that data set . the weighted mean and weighted error of all the data
are determined from the shift and error values of each data set , and the weighted error is scaled by the square - root of the reduced chi - square , @xmath163 .
these are shown as the solid and dashed horizontal lines in panels ( a ) and ( c ) .
panels ( b ) and ( d ) show the corresponding allan deviations ( ignoring dead time between data runs ) of the frequency shift records displayed in ( a ) and ( c ) , respectively .
each measurement represents a differential comparison between two density conditions . under typical clock operating conditions ( @xmath164 ) , the weighted mean and the weighted standard error of the fractional frequency shift are @xmath165 at @xmath124 = 7 @xmath26k ( ( a ) and ( b ) ) and @xmath166 at @xmath124 = 3.5 @xmath26k ( ( c ) and ( d ) ) . for the 7 @xmath26k data
the reduced chi - square was @xmath167 , and for the @xmath168k data @xmath169 . ]
we have made an extensive series of collisional shift measurements at the largest trap depths available to us .
the results of these measurements are displayed in fig .
[ fig5 ] .
the free - running clock laser has a stability of about @xmath170 at time scales of 110 seconds @xcite .
therefore , a substantial integration time is required to determine the collisional shift with an uncertainty of @xmath137 .
frequency drifts are minimized by measuring the long - term drift in the resonance frequency ( relative to the ultrastable reference cavity ) and applying a feed - forward correction to the clock laser .
the correlation between the atomic resonance frequencies and the density of trapped atoms was calculated by analyzing overlapping sequences of four consecutive measurements and eliminating frequency drifts of up to second order @xcite .
approximately 60 hours of data were acquired at @xmath124 = 7 @xmath26k over a @xmath171 month time period for the record shown in fig .
[ fig5 ] ( a ) .
each data point represents a period during which the clock was continuously locked , with error bars determined from the standard error of the measurements in that data set .
the error bars of each data point were scaled by a correction factor @xmath172 that accounts for the fact that , due to the overlapping string analysis , each individual four - measurement sequence is only partially independent of its neighbors ( see som ) . the weighted mean and weighted standard error of these data determine the value of the shift .
on longer time scales , the data are underscattered , and we therefore scale the final error value by the square root of the reduced chi - square of the data set . at an axial temperature @xmath173k ,
the collisional shift in our 2d lattice clock was measured to be @xmath165 in fractional units , with @xmath174 . at a lower @xmath124 of 3.5 @xmath26k ,
the collisional shift is reduced to @xmath166 , with @xmath175 ( the record in fig .
[ fig5 ] ( c ) ) .
the corresponding allan deviations of both data sets are shown in subpanels ( b ) and ( d ) of fig .
[ fig5 ] . the data in both fig .
[ fig4 ] and fig .
[ fig5 ] have been analyzed using four different methods , the details of which are given in the som .
the results of all analyses are consistent with each other to within the stated error bars .
figures [ fig4 ] and [ fig5 ] demonstrate the suppression of the clock frequency shift as the system approaches the strongly interacting regime .
we note that , relative to previous measurements of collisional shifts in a 1d optical lattice @xcite , the atomic density in a 2d lattice is much higher . after accounting for the increased temperature in our current experiments ,
we find that the local density in a lattice site occupied by two particles is @xmath176 , an order of magnitude larger than the average density in earlier 1d lattice experiments .
therefore , given a similar level of excitation inhomogeneity , if the collisional shift in the 2d lattice were not suppressed , we would expect a significantly larger shift than that in the 1d lattice experiments , even after assuming that only @xmath177 of lattice sites are contributing .
the results presented here demonstrate that detailed understanding of a quantum system can result in dramatic improvements in the areas of precision measurement and atomic clocks .
this advance removes an important obstacle to further increasing the precision and accuracy of neutral atom - based optical clocks .
increasing the number of atoms loaded into our 2d lattice system will enable us to improve the stability of our clock without imposing an onerous systematic effect .
as clock lasers become more stable , we will be able to increase the duration of the rabi interrogation pulse , thus decreasing the rabi frequency .
this will allow optical lattice clocks to operate in the regime where the density shift is fully suppressed ( @xmath178 ) and further reduce the collisional shift systematic well into the 10@xmath179 domain .
this , together with the fact that in the strongly - interacting regime the collisional shift will remain suppressed as more atoms are loaded into individual lattice sites , will enable neutral atom clocks to operate with the large sample sizes needed to achieve the highest possible stability .
g. k. campbell _
et al .
_ _ science _ * 324 * , 360 ( 2009 ) .
n. d. lemke _ et al .
_ _ phys .
rev .
lett . _ * 103 * , 063001 ( 2009 ) .
a. d. ludlow _ et al .
_ _ science _ * 319 * , 1805 ( 2008 ) .
s. gupta _
et al .
_ _ science _ * 300 * , 1723 ( 2003 ) .
m. punk and w. zwerger .
_ phys .
rev .
lett . _ * 99 * , 170404 ( 2007 ) .
g. baym , c. j. pethick , z. yu and m. w. zwierlein .
_ phys .
rev .
lett . _ * 99 * , 190407 ( 2007 ) .
m. girardeau .
_ j. math .
phys_. * 1 * , 516 ( 1960 ) .
t. kinoshita , t. wenger , d. s. weiss .
_ science _ * 305 * , 1125 ( 2004 ) .
b. paredes _ et al .
_ _ nature _ * 429 * , 277 ( 2004 ) .
e. haller _ et al .
_ _ science _ * 325 * , 1224 ( 2009 ) .
m. d. swallows _
et al . _
_ ieee trans .
ultra . ferroelectric .
. control _ * 57 * , 574 ( 2010 ) .
k. gibble .
_ phys .
rev .
lett . _ * 103 * , 113202 ( 2009 ) .
a. m. rey , a. v. gorshkov and c. rubbo .
_ phys .
rev .
lett . _ * 103 * , 260402 ( 2009 ) . z. h. yu , and c. j. pethick . _ phys . rev
. lett . _ * 104 * , 010801 ( 2010 ) . m. m. boyd _
et al . _
_ phys .
rev . a _ * 76 * , 022510 ( 2007 ) . t. akatsuka , m. takamoto and h. katori .
_ nat .
phys . _ * 4 * , 954 ( 2008 ) .
g. k. campbell _
et al .
_ _ metrologia _ * 45 * , 539 - 548 ( 2008 ) .
j. ye , h. j. kimble and h. katori .
_ science _ * 320 * , 1734 ( 2008 ) .
s. blatt _
et al . _
_ phys .
rev .
a _ * 80 * , 052703 ( 2009 ) . m. saffman , t. g. walker and k. molmer .
arxiv:0909.4777 ( 2009 ) .
m. w. zwierlein , z. hadzibabic , s. gupta and w. ketterle .
_ phys .
rev .
lett . _ * 91 * , 250404 ( 2003 ) .
m. . oktel and l. s. levitov .
_ phys .
rev .
lett . _ * 83 * , 6 ( 1999 ) .
m. . oktel , t. c. killian , d. kleppner and l. s. levitov .
_ phys .
rev . a _ * 65 * , 033617 ( 2002 ) .
a. d. ludlow _ et al .
_ _ opt .
lett . _ * 32 * , 641 ( 2007 ) .
m. m. boyd _
et al . _
_ phys .
rev .
lett . _ * 98 * , 083002 ( 2007 )
. w. b. dress _
et al . _
_ phys .
rev .
d _ * 15 * , 9 ( 1977 ) .
* acknowledgements * we thank c. benko for technical assistance and a. d. ludlow and a. gorshkov for useful discussions .
this work was supported by a grant from the army research office with funding from the darpa ole program , the us national institute of standards and technology , the us national science foundation ( nsf ) physics frontier center at jila , and the air force office for scientific research .
m.d.s . is supported by a national research council postdoctoral fellowship .
m.b .
is partially supported by an ndseg graduate fellowship . y.l.s
permanent address is national institute of metrology , beijing , china . |
we introduce and study problems that are generically in @xmath0 , i.e. , decision problems that have partial errorless nondeterministic decision algorithms that solve the problem in polynomial time on `` most '' inputs .
we define appropriate reductions in this class and show that there are some complete problems there , called _ strongly generically @xmath0-complete _ problems .
in particular , the randomized bounded version of the halting problem is one of them .
rigorous formulation of notions of generic algorithms and generic complexity appeared first in group theory @xcite as a response to several challenges that algorithmic algebra faced at that time .
first , it was well understood that many hard , even undecidable , algorithmic problems in groups can be easily solved on most instances ( see @xcite for a thorough discussion ) .
second , the study of random objects and generic properties of objects has become the mainstream of geometric group theory , following the lead of graph and number theory ( see @xcite ) .
it turned out that `` random '' , `` typical '' objects have many nice properties that lead to simple and efficient algorithms .
however a rigorous formalization of this approach was lagging behind .
algorithmic algebra was still focusing mostly on the worst - case complexity with minor inroads into average case complexity .
third , with the rapid development of algebraic cryptography the quest for natural algorithmic problems , which are hard on most inputs , became one of the main subjects in complexity theory ( see discussion in @xcite ) .
it was realized that the average case complexity does not fit well here .
indeed , by definition , one can not consider average case complexity of undecidable problems , which are in the majority in group theory ; the proofs of average case results are usually difficult and technical @xcite , and , most importantly , there are problems that are provably hard on average but easy on most inputs ( see @xcite for details ) .
in fact , gurevich showed in @xcite that the average case complexity is not about `` most '' or `` typical '' instances , but that it grasps the notion of `` trade - off '' between the time of computation on hard inputs and how many of such hard instances are there . nowadays ,
generic algorithms form an organic part of computational algebra and play an essential role in practical computations . in a surprising twist generic algorithms and ideas of generic complexity
were recently adopted in abstract computability ( recursion theory ) .
there is interesting and active research there concerning absolutely undecidable problems , generic turing degrees , coarse computability , etc . , relating generic computation with deep structural properties of turing degrees @xcite .
we decided to relativize these ideas to lower complexity classes . here
we consider the class @xmath0 .
motivation to study generically hardest problems in the class @xmath0 comes from several areas of mathematics and computer science .
first , as we have mentioned above , average case complexity , even when it is high , does not give information on the hardness of the problem at hand on the typical or generic inputs .
therefore , to study hardness of the problem on most inputs one needs to develop a theory of generically complete problems in the class @xmath0 .
this is interesting in its own right , especially when much of activity in modern mathematics focuses on generic properties of mathematical objects and how to deal with them . on the other hand , in modern crypotography
, there is a quest for cryptoprimitives which are computationally hard to break on most inputs
. it would be interesting to analyze which @xmath0-problems are hard on most inputs , i.e. , which of them are generically @xmath0-complete .
note , there are @xmath0-complete problems that are generically polynomial @xcite .
all this requires a robust theory of generic @xmath0-completeness . as the first attempt to develop such a theory we study here the class of all generically @xmath0-problems , their reductions , and the complete problems in the class .
most of the time , our exposition follows the seminal gurevich s paper @xcite on average complexity .
we conclude with several open problems that seem to be important for the theory . here
we briefly describe the structure of the paper and mention the main results . in section [ se : preliminaries ] , we recall some notions and introduce notation from the classical decision problems . in section [
se : generic ] , we discuss distributional decision problems ( when the set of instances of the problem comes equipped with some measure ) , then define the generic complexity and problems decidable generically ( strongly generically ) in polynomial time .
in section [ se : reductions ] , we define generic polynomial time reductions . in section [ se : disthp ] , we show that the distributional bounded halting problem for turing machines is strongly generically @xmath0-complete .
notice that though generic ptime randomized algorithms are usually much easy to come up with ( than say ptime on average algorithms ) , the reductions in the class of generic @xmath0-problems are still as technical as reductions in the class of @xmath0-problems on average .
in fact , the reductions in both classes are similar . essentially , these are reductions among general randomized problems and the main technical , as well as theoretical , difficulty concerns the transfer of the measure when reducing one randomized problem to another one .
it seems this difficulty is intrinsic to reductions in randomized computations and does not depend on whether we consider generic or average complexity . in section [
se : problems ] we discuss some open problems that seem to be important for the development of the theory of generic @xmath0-completeness .
in this section we introduce notation to follow throughout the paper . informally , a _ decision problem _ is an arbitrary yes - or - no question for an ( infinite ) set of _ inputs _ ( or _ instances _ ) @xmath1 , i.e. , an unary predicate @xmath2 on @xmath1 .
the problem is termed _ decidable _ if @xmath2 is computable , and the main classical question is whether a given problem is decidable or not . in complexity
theory the predicate @xmath2 usually is given by its true set @xmath3 , so the decision problem appears as a pair @xmath4 .
furthermore , it is assumed usually that every input @xmath5 admits a finite description in some finite alphabet @xmath6 in such a way that given a word @xmath7 one can effectively determine if @xmath8 or not .
this allows one , without loss of generality , to assume simply that @xmath9 .
some care is required when dealing with distributional problems and we discuss this issue in due course . from now on , unless said otherwise , we assume that decision problems are pairs @xmath10 , where @xmath11 . in this case
@xmath6 is the _ alphabet _ of the problem @xmath12 and we denote it sometimes by @xmath13 ; @xmath14 is the set of _ inputs _ or the _ domain _ of @xmath12 ; the set @xmath15 is the _ yes _ or _ positive _ part of @xmath12 , denoted sometimes by @xmath16 or @xmath17 . in section [ subsec : restrictions ]
we briefly consider problems of the type @xmath4 , where @xmath18 , not assuming that @xmath1 is a decidable subset of @xmath19 .
it is natural now to define the _ size _ of @xmath20 to be its word length @xmath21 . as usual , we define the sphere @xmath22 of radius @xmath23 as the set of all strings ( words ) in @xmath19 of size @xmath24 , and @xmath25 . for a symbol @xmath26 and @xmath27 put @xmath28 to be the string of @xmath24 symbols @xmath29 .
we assume that alphabet @xmath6 comes equipped with a fixed linear ordering .
this allows one to introduce a _ shortlex _ ordering @xmath30 on the set @xmath19 as follows .
we order , first , the words in @xmath19 with respect to their length ( size ) , and if two words have the same length then we compare them in the ( left ) lexicographical ordering .
the successor of a word @xmath20 is denoted by @xmath31 . in this section
we recall the definition of a turing machine in order to establish terminology .
a one - tape _ turing machine _ ( tm ) @xmath32 is a @xmath33-tuple @xmath34 where : * @xmath35 is a finite set of _ states _ ; * @xmath36 is a finite set called the _ tape alphabet _ which contains at least @xmath37 symbols ; * @xmath38 is the _ initial state _ ; * @xmath39 is the _ final state _ ; * @xmath40 is the _ transition relation_. additionally , @xmath32 uses a _ blank symbol _ @xmath41 different from the symbols @xmath6 to mark the parts of the infinite tape not in use .
this is the only symbol allowed to occur on the tape infinitely often at any step during the computation .
we say that a transition relation @xmath42 in the definition of a tm is _ deterministic _
if for every pair @xmath43 there is a unique five - tuple @xmath44 in @xmath42 , i.e. , @xmath42 defines a function @xmath45 .
we say that a tm @xmath32 is _ deterministic _ if its transition relation is .
otherwise we say that @xmath32 is a nondeterministic machine ( ntm ) .
each turing machine has a _ tape _ with @xmath46-symbols written on it , a _ head _ specifying a position on the tape , and a _ state register _ containing an element @xmath47 .
we say that the head observes a symbol @xmath48 , if @xmath29 is written on the tape at the position specified by the head .
if a tm @xmath32 is in the state @xmath49 and observes a symbol @xmath50 , then to perform a step of computations : * @xmath32 chooses any element @xmath51 ; * puts @xmath52 into the state register ; * writes @xmath53 on the tape to the head position ; * moves the head to left or to the right depending on @xmath54 .
if @xmath42 contains no tuple @xmath55 , then we say that @xmath32 _
breaks_. we can define the operation of a tm formally using the notion of a configuration that contains a complete description of the current state of computation .
a _ configuration _ of @xmath32 is a triple @xmath56 where @xmath57 are @xmath6-strings and @xmath58 .
* @xmath59 is a string to the left of the head ; * @xmath60 is the string to the right of the head , including the symbol scanned by the head ; * @xmath49 is the current state .
we say that a configuration @xmath56 _ yields _ a configuration @xmath61 in one step , denoted by @xmath62 if a step of a machine from configuration @xmath56 results in configuration @xmath61 .
note that if the machine is nondeterministic , then a configuration can yield more than one configuration . using the relation `` yields in one step ''
one can define relations `` yields in @xmath63 steps '' , denoted by @xmath64 and `` yields '' , denoted by @xmath65 we say that @xmath32 _ halts _ on @xmath20 if the configuration @xmath66 yields a configuration @xmath67 for some @xmath6-strings @xmath59 and @xmath60 .
the number of steps @xmath32 takes to stop on a @xmath6-string @xmath68 is denoted by @xmath69 .
if @xmath32 does not halt on @xmath20 then we put @xmath70 .
the _ halting problem _ for @xmath32 is an algorithmic question to determine whether @xmath32 halts or not on an input @xmath71 , i.e. , whether @xmath72 or not .
we say that a tm @xmath32 _ solves _ or _ decides _ a decision problem @xmath12 over an alphabet @xmath6 if @xmath32 stops on every input @xmath20 with an answer : * @xmath73 ( i.e. , at a configuration @xmath74 , where @xmath59 starts with @xmath75 ) if @xmath76 ; * @xmath77 ( i.e. , at configuration @xmath74 , where @xmath59 starts with @xmath78 ) otherwise .
we say that @xmath32 _ partially decides _
@xmath12 if it decides @xmath12 correctly on a subset @xmath79 of @xmath12 and on @xmath80 it either does not stop or stops with an answer @xmath81 ( i.e. , stops at configuration @xmath74 , where @xmath59 starts with @xmath82 ) . in the event when @xmath32 breaks or outputs @xmath81 the value of @xmath69 is @xmath83 . for a function @xmath84 define @xmath85 [ @xmath86 resp . ]
to be the class of all decision problems decidable by some deterministic [ nondeterministic resp .
] turing machine within time @xmath87 .
two of the most used classes of decision problems @xmath88 and @xmath89 are defined as follows : @xmath90 clearly @xmath91 . it is an old , open problem whether @xmath92 or not . the classical _ polynomial time many - to - one _ or _ karp reductions _ provide a crucial tool to deal with problems in @xmath0 .
we recall it in the following definition and refer to them simply as to ptime reductions .
let @xmath93 and @xmath94 be decision problems .
we say that a function @xmath95 is a _ ptime reduction _ , or @xmath96 _ ptime reduces _
@xmath93 to @xmath94 , and write @xmath97 , if * @xmath96 is polynomial time computable ; * @xmath98 if and only if @xmath99 .
we say that a ptime reduction @xmath96 is _ size - invariant _ if @xmath100 notice , that many classical ptime reductions are size - invariant ( see @xcite ) .
now , for a size - invariant reduction @xmath96 the function @xmath101 is well defined and strictly increasing .
we refer to @xmath102 as the _ size growth _ of @xmath96 . a problem @xmath103 is called _ @xmath0-complete _ if every problem @xmath104 is ptime reducible to @xmath12 .
the following is a classic result in complexity theory ( see @xcite ) .
[ th : np_compl ] the following holds . 1 .
if @xmath96 is a ptime reduction from @xmath93 to @xmath94 and @xmath32 is an turing machine solving @xmath94 in polynomial time then @xmath105 solves @xmath93 in polynomial time . 2 .
3sat is @xmath0-complete . here , and below
, by @xmath105 we denote the algorithm that is a composition of the tm @xmath32 and a tm that computes @xmath96 .
let us first recall some basic definitions of probability theory that will be used in this section .
a _ probability measure _ on @xmath19 is a function @xmath106 $ ] satisfying @xmath107 .
ensemble of probability measures _ on @xmath19 is a collection of sets @xmath108 of @xmath19 ( not necessarily disjoint ) and a collection of probability measures @xmath109 satisfying @xmath110 and @xmath111 .
spherical ensemble of probability measures _ on @xmath19 is an ensemble with @xmath112 . in particular ,
a spherical ensemble of probability measures on @xmath19 is uniquely defined by a collection of measures @xmath113 satisfying @xmath114 .
the average case complexity deals with `` expected '' running time of algorithms , while the generic case complexity deals with `` most typical '' or _
generic _ inputs of a given problem @xmath115 . these require to measure or compare various subsets of inputs from @xmath19 .
there are several standard ways to do so , for example , by introducing either a probability measure @xmath116 on @xmath19 ( as was done in @xcite ) , or an _ ensemble _ of probability measures defined on spheres or balls of @xmath19 ( see @xcite ) . in many cases ,
all three approaches are equivalent and lead to similar results . following the current tradition in computer science ,
we elect here to work with a _ spherical _ ensemble @xmath117 of probability measures @xmath118 defined on the spheres @xmath22 . in
what follows , we always assume that @xmath6 is a finite alphabet and every measure @xmath118 from the ensemble @xmath116 is atomic , i.e. , it is given by a probability function ( which we denote again by @xmath118 ) @xmath119 so that @xmath120 for every subset @xmath121 .
the pair @xmath122 is termed a _
distributional space_. whether @xmath118 is a probability measure or the corresponding probability function will be always clear from the context , so no confusion should arise .
we want to stress here that generic properties of a given decision problem depend on the chosen ensemble @xmath116 and @xmath116 is an essential part of the problem ( see @xcite for details ) .
a _ distributional decision problem _ is a triple @xmath123 , where @xmath124 is a decision problem and @xmath122 a distributional space .
usually we refer to a distributional problem @xmath123 as a pair @xmath125 , where @xmath124 .
there are two important constructions on distributional spaces , introduced in @xcite .
since we use here ensembles of distributions , unlike @xcite , where single measures were used , we give below precise definitions .
notice , that we always assume that @xmath126 if @xmath127 .
let @xmath6 and @xmath128 be finite alphabets , @xmath122 and @xmath129 distributional spaces , and @xmath130 a size - invariant function . then @xmath131 is the @xmath96-_transfer _ of @xmath116 ( or @xmath96 transfers @xmath116 to @xmath131 ) if for any @xmath132 the following equality holds @xmath133 [ induced ensembles ] let @xmath122 be a distributional space and @xmath134 .
then an ensemble @xmath135 on @xmath19 is called @xmath136-induced by @xmath116 if for any @xmath71 the following equality holds @xmath137 let @xmath122 be a distributional space and @xmath138 .
the function @xmath139 is called the _ density function _ of @xmath136 and its limit ( if exists ) @xmath140 is called the _ asymptotic density _ of @xmath136 in @xmath122 .
a subset @xmath134 is called * _ generic _ in @xmath19 if @xmath141 ; * _ strongly generic _ in @xmath19 if @xmath141 and @xmath142 converges to @xmath143 super polynomially fast , i.e. , @xmath144 for any @xmath145 ; * _ negligible _ in @xmath19 if @xmath146 ; * _ strongly negligible _ in @xmath19 if @xmath146 and @xmath142 converges to @xmath147 super polynomially fast , i.e. , @xmath148 for any @xmath145 .
notice that we use the term `` generic '' in the sense of `` typical '' .
the same term has also been used in complexity and set theory to refer to sets that are far from typical , that are constructed through cohen forcing . [ de : generic_poly ] let @xmath149 be a distributional decision problem . * we say that @xmath149 is _ decidable generically in polynomial time _ ( or _ gptime decidable _ ) if there exists a turing machine @xmath32 that partially decides @xmath12 within time @xmath69 and a polynomial @xmath150 such that @xmath151 in this case we say that @xmath32 is a generic polynomial time decision algorithm for @xmath12 and @xmath12 has _
generic time complexity _ at most @xmath152 .
* we say that @xmath149 is _ decidable strongly generically in polynomial time _ ( or _ sgptime decidable _ ) if there exists a turing machine @xmath32 that partially decides @xmath12 within time @xmath69 and a polynomial @xmath150 such that for any polynomial @xmath153 @xmath154 in this case , we say that @xmath32 is a strongly generic polynomial time decision algorithm for @xmath12 and @xmath12 has
_ strong generic time complexity _ at most @xmath152 .
we refer to the sequence @xmath155 as a _ control sequence _ of the algorithm @xmath32 relative to the complexity bound @xmath156 and denote it by @xmath157 . in other words ,
a problem @xmath149 is gptime ( sgptime ) decidable if there exists a polynomial time tm that partially decides @xmath12 and its halting set is generic ( strongly generic ) in @xmath158 . in this section
we recall the notion of a distributional @xmath0-problem , which is a distributional analog of the classical @xmath0-problems .
[ ptime computable real - valued function ] a function @xmath159 $ ] is _ computable in polynomial time _ if there exists a polynomial time algorithm that for every @xmath71 and @xmath160 computes a binary fraction @xmath161 satisfying @xmath162 [ ptime computable ensembles of probability measures ] we say that a spherical ensemble of measures @xmath163 on @xmath19 is ptime computable if the function @xmath164 $ ] defined by @xmath165 is ptime computable .
denote by @xmath166 the ensemble of probability _ distributions _ defined by @xmath167 as above , the ensemble @xmath168 is called ptime computable if the function @xmath169 is ptime computable .
[ le : s - induced ] let @xmath122 be a distributional space .
then the following hold : * if @xmath168 is ptime computable then @xmath116 is ptime computable .
* if @xmath136 is a subset of @xmath19 such that the function @xmath170 is ptime computable then the @xmath136-induced on @xmath19 ensemble of measures @xmath171 is ptime computable
. follows directly from definitions .
@xmath172 is a class of distributional decision problems @xmath149 such that * @xmath103 ; * @xmath173 is a ptime computable ensemble of probability distributions on @xmath14 .
@xmath174 is the class of gptime decidable distributional decision problems ( not necessarily from @xmath172 ) .
@xmath175 is the class of sgptime decidable distributional decision problems .
we want to point out that classes @xmath174 and @xmath175 contain some exotic problems , e.g. , some undecidable problems . for more information see @xcite .
in this section we introduce the notion of a _ generic polynomial reduction _ and describe two particular types of reductions , called size and measure reductions .
observe first that the classical karp reductions do not work for generic complexity . indeed , the following example shows that a ptime reduction @xmath176 and a generic polynomial time decision algorithm for @xmath177 do not immediately provide a generic polynomial time decision algorithm for @xmath12 .
[ ex : gen - red ] let @xmath178 be a binary alphabet and @xmath116 the spherical ensemble of uniform measures @xmath118 on @xmath22 . let @xmath179 be a monoid homomorphism defined by @xmath180 now , for a decision problem @xmath115 consider a decision problem @xmath181 .
it follows from the construction that @xmath182 is a ptime reduction and @xmath183 , provided @xmath103 .
furthermore , it is easy to check that the set @xmath184 , as well as @xmath185 , is strongly negligible in @xmath122 .
this implies that a partial algorithm @xmath186 that on an each input from @xmath187 says `` no '' and does halt on @xmath184 , is a strongly generic polynomial time decision algorithm for @xmath188 .
nevertheless , @xmath186 does not reveal any useful information on @xmath12 .
let @xmath189 and @xmath190 a ptime size - invariant reduction .
* we say that @xmath96 is a _ weak gptime reduction _ if there exists a tm @xmath32 , which gptime decides @xmath191 and @xmath192 gptime decides @xmath149 . *
we say that @xmath96 is an _ gptime reduction _ if for every tm @xmath32 , which gptime decides @xmath191 the composition @xmath192 gptime decides @xmath149 . *
we say that @xmath96 is an _ sgptime reduction _ if for every tm @xmath32 , which sgptime decides @xmath191 the composition @xmath192 sgptime decides @xmath149 .
we give examples of sgptime reductions in the next two sections .
one can introduce reductions @xmath190 of a more general type by allowing the function @xmath96 to be defined only on a generic ( strongly generic ) subset @xmath193 of @xmath14 with the polynomial time computable characteristic function @xmath194 .
the classes of all gptime and sgptime reductions are closed under composition
. follows directly from the definitions .
it is not known if the class of weak gptime reductions is transitive .
let @xmath149 be a distributional decision problem .
we say that * @xmath149 is _ sgptime hard _ for @xmath172 if every @xmath172 problem sgptime reduces to @xmath149 . * @xmath149 is _ sgptime complete _ for @xmath172 if @xmath195 and @xmath149 is _ sgptime hard _ for @xmath172 . in this section ,
we introduce _ change of size _ ( cs ) reductions .
[ de : cs ] let @xmath196 and @xmath190 a ptime size - invariant reduction of @xmath12 to @xmath177 . if @xmath131 is the @xmath96-transfer by @xmath116 ( see section [ subsec : dist ] ) then @xmath96 is called a _ cs - reduction_. in this example @xmath191 is obtained from @xmath149 by increasing the sizes of elements . ]
[ le : cs ] let @xmath196 and @xmath197 a cs - reduction . if @xmath102 is bounded by a polynomial , then @xmath96 is a gptime and sgptime reduction .
let @xmath186 be an algorithm that generically decides @xmath191 within a polynomial time upper bound @xmath198 .
then @xmath199 is a partial decision algorithm for @xmath149 .
since @xmath131 is induced by @xmath116 , one has : @xmath200 observe , that @xmath201 is polynomially bounded , since @xmath102 and @xmath202 are polynomially bounded . clearly , the control sequence @xmath203 is at most @xmath204 .
notice , that @xmath204 is an infinite subsequence of @xmath205 ( because @xmath102 is strictly increasing ) , hence it converges to @xmath147 , so @xmath206 is a generic upper bound for @xmath199 .
this proves the first statement of the theorem . to prove the second statement ,
assume that @xmath191 is sgptime decidable by @xmath186 within a polynomial time @xmath156 .
then for the control sequence @xmath205 one has @xmath207 for any positive integer @xmath63 . due to the inequalities above
, the control sequence for @xmath208 with respect to the polynomial bound @xmath206 satisfies the following inequality @xmath209 hence @xmath210 is sgptime decidable by @xmath208 , as claimed . by theorem [ le : cs ]
a cs - reduction generally increases time complexity and improves control sequence . in this section , we define _ change of measure _ ( cm ) reductions . [ de : cm - red ] let @xmath196 and @xmath190 a ptime reduction such that * @xmath211 for any @xmath212 ; * there exists a polynomial @xmath54 such that for each @xmath212 @xmath213 then
@xmath96 is called a _ cm - reduction_. figure [ fi : measure_reduction ] depicts the situation under consideration .
scheme of a cm - reduction .
a function @xmath96 maps a distributional decision problem @xmath149 into a distributional decision problem @xmath191 so that the @xmath214th sphere in @xmath12 is mapped exactly into the @xmath214th sphere in @xmath177 .
the grey part of @xmath177 depicts the image of @xmath12 . ]
[ le : cm ] let @xmath196 and @xmath215 a cm - reduction .
then the following holds . *
if @xmath191 is decidable by a tm @xmath186 within a generic polynomial time bound @xmath156 and @xmath216 ( where @xmath217 is the function from definition [ de : cm - red ] ) then @xmath199 gptime decides @xmath149 .
* @xmath96 is an sgptime reduction .
\(a ) let @xmath186 be an algorithm that generically decides @xmath191 within a polynomial time upper bound @xmath198 .
then @xmath199 is a partial decision algorithm for @xmath149 .
recall , that @xmath96 preserves the size .
therefore , @xmath199 decides @xmath12 within the polynomial time bound @xmath218 everywhere , except , maybe , a subset @xmath219 to prove the statement it suffices to show that the set above is generic in @xmath220 .
@xmath221 ( b ) if an algorithm @xmath186 sgptime decides @xmath191 then @xmath222 for any polynomial @xmath49 .
therefore , by part 1 ) , @xmath223 for every polynomial @xmath49 . in particular
, for @xmath224 one has @xmath225 for any polynomial @xmath226 , as required . in this section
we show that each @xmath172 problem over a finite alphabet @xmath6 can be reduced to a @xmath172 problem over a binary alphabet @xmath227 .
[ th : bin_alph ] let @xmath149 be an @xmath172 problem over a finite alphabet @xmath6 . then there exists a @xmath172 problem @xmath191 over the binary alphabet @xmath227 and a cs - reduction @xmath190 with linear size function @xmath102 .
suppose that @xmath228 is an one - letter alphabet .
let @xmath229 be a monoid homomorphism defined by @xmath230 .
put @xmath231 and @xmath232 .
define a spherical ensemble of measures @xmath131 on @xmath233 to be @xmath234 clearly , @xmath235 and @xmath96 is a cs - reduction with linear size - growth function @xmath102 . by theorem [ le : cs ] , @xmath96 is an sgptime reduction .
suppose that @xmath236 , where @xmath237 .
define a function @xmath96 as follows .
put @xmath238 and , if @xmath239 and @xmath68 is the @xmath63th element in @xmath22 ( in the lexicographical order ) , then @xmath96 maps @xmath68 into the @xmath63th element of @xmath240 .
as above , we put @xmath231 .
let @xmath131 be the @xmath96-transfer of @xmath116 .
the problem @xmath191 belongs to @xmath172 because @xmath195 and @xmath96 is a ptime reduction .
clearly , @xmath96 is a cs - reduction with a linear size - growth function @xmath241 . by theorem [ le : cs ]
, @xmath96 is an sgptime reduction .
let @xmath242 be a problem and @xmath243 . in this section
we consider the restriction @xmath244 of @xmath12 to the subset @xmath136 .
intuitively , @xmath244 is the same problem as @xmath12 , only the set of inputs is restricted to @xmath136 .
the most natural formalization of @xmath244 would be @xmath245 , allowing the domain @xmath136 not equal to @xmath14 , contrary to our assumption on algorithmic problems . in this case
one can stratify the domain @xmath1 as a union @xmath246 , where @xmath247 , and leave only those @xmath248 that are non - empty .
then , one can obtain an ensemble of measures @xmath249 on @xmath1 relative to the stratification above , where @xmath250 is the measure on @xmath248 induced by @xmath118 .
after that , the theory of distributional problems of this type can be developed similarly to the one already considered .
however , it is a bit awkward and heavier in notation .
we choose another way around this problem we change the ensemble of measures , but do not change the input space .
let @xmath149 be a distributional problem .
for a subset @xmath251 consider the ensemble of probability measures @xmath171 on @xmath14 @xmath136-induced by @xmath116 ( see section [ subsec : dist ] ) .
the distributional problem @xmath252 is called the _ restriction _ of the distributional problem @xmath149 to the subset @xmath136 . [
le : comp - induced ] let @xmath253 and @xmath243 .
if the function @xmath254 is ptime computable then @xmath255
. follows immediately from lemma [ le : s - induced ] .
[ le : gp - s - induced ] let @xmath195 , @xmath243 , and @xmath255 .
if an algorithm @xmath186 gptime decides @xmath149 with a control sequence @xmath256 such that the sequence @xmath257 converges to @xmath147 , then @xmath186 gptime decides @xmath252 with the control sequence bounded from above by @xmath258 .
let @xmath152 be a generic polynomial time upper bound of the algorithm @xmath186 and @xmath259 .
set @xmath260 .
then @xmath261 and for @xmath262 one has @xmath263 hence @xmath264 thus , the sequence @xmath265 converges to @xmath147 .
[ co : restriction ] let @xmath195 , @xmath243 , and @xmath255 .
if there exists a polynomial @xmath54 such that @xmath266 for @xmath267 then the identity function @xmath268 gives an sgptime reduction @xmath269 we would like to emphasize that the situation with restrictions of problems in @xmath174 is quite different from the `` average - case '' one , where almost any restriction preserves the property of being polynomial time computable on average .
in this section we , following @xcite , define the distributional bounded halting problem and prove that it is sgptime complete in @xmath172 .
let @xmath32 be a nondeterministic turing machine with the binary tape alphabet @xmath178 .
intuitively , the bounded halting problem for @xmath32 is the following algorithmic question : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for a positive integer @xmath24 and a binary string @xmath59 such that @xmath270 decide if there is a halting computation for @xmath32 on @xmath59 in at most @xmath24 steps . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ by our definitions ( see section [ se : prelim ] ) instances of algorithmic problems are words ( not pairs of words ) in some alphabet , so to this end we encode a pair @xmath271 by the binary string @xmath272 such that @xmath273 .
notice , that any binary string containing @xmath147 is the code @xmath274 for some @xmath275 .
denote by @xmath276 the subset of all binary strings @xmath274 , where @xmath277 , such that @xmath32 halts on @xmath59 within @xmath24 steps . from now on we refer to the problem @xmath278 as the _ bounded halting problem_. to turn @xmath279 into a distributional problem we introduce a spherical ensemble @xmath280 of probability measures as follows . for @xmath281 put @xmath282 the problem @xmath283 is the _ distributional _ halting problem for @xmath32 , we refer to it as @xmath284 .
probability space for halting problem ( a finite part ) .
dashed lines correspond to @xmath147 , solid lines correspond to @xmath143 .
each dot is associated with the label of the path from the root to itself .
grey dots have non - trivial measure . ] a positive integer @xmath24 is called _ longevous _ for an input @xmath59 of an ntm @xmath32 if every halting computation of @xmath32 on @xmath59 has at most @xmath24 steps .
a function @xmath285 is a _
longevity guard _ for @xmath32 if for every input @xmath59 the number @xmath286 is longevous for @xmath59 .
notice , that if @xmath287 is a longevity guard for @xmath32 , then any function @xmath288 is also a longevity guard for @xmath32 . in
what follows , we always assume that a longevity guard satisfies the following conditions : * @xmath289 ; * @xmath286 is strictly increasing . [ re : longevity ] for every problem @xmath103 , there is an ntm @xmath12 that decides @xmath12 and has a polynomial longevity guard @xmath285 , satisfying the conditions ( 1 ) , ( 2 ) above .
since @xmath32 halts on an input @xmath59 if and only if it halts on @xmath59 within @xmath286 steps , there is no much use to consider instances @xmath271 of the halting problem for @xmath32 with @xmath290 .
a rigorous formalization of this observation is to restrict the problem @xmath284 to the subset of instances @xmath291 more generally , for a computable function @xmath285 , satisfying conditions ( 1 ) and ( 2 ) , consider the set @xmath292 as above and denote by @xmath293 the ensemble of measures for @xmath19 which is @xmath292-induced by @xmath131 .
let @xmath294 be the restriction of the problem @xmath284 to @xmath292 .
[ pr : restriction2bhmg ] let @xmath32 be an ntm and @xmath295 a polynomial function
. then @xmath296 and the identity function @xmath297 gives an sgptime reduction @xmath298 .
observe first that the function @xmath299 is ptime computable . indeed ,
if @xmath300 , then @xmath301 , so @xmath302 is uniquely defined ( since @xmath287 is monotone ) . in this case , @xmath303 depends only on @xmath24 , hence @xmath304 therefore , @xmath305 since the function @xmath287 is polynomial it takes at most @xmath306 time to check if @xmath307 or not .
now , by lemma [ le : comp - induced ] @xmath296 .
equalities [ eq : nu - c(g ) ] and corollary [ co : restriction ] imply that @xmath298 is an sgptime reduction , as claimed .
examples of restricted problems @xmath308 , where a ) @xmath309 and b ) @xmath310 . only grey dots have non - trivial probability . ] in the proofs below , we use the following encoding of natural numbers and turing machines .
let a string @xmath311 be a binary expansion for @xmath27 , i.e. , @xmath312 , @xmath313 , and @xmath314 . denote by @xmath315 the binary string @xmath316 obtained from @xmath317 by inserting @xmath143 in front of each symbol in @xmath317 .
let @xmath318 be a polynomial time computable enumeration of ( nondeterministic ) turing machines , such that @xmath319 is a binary representation of a natural number and every natural number is equal to @xmath319 for some @xmath32 .
denote by @xmath320 the string @xmath321 .
[ th : red2bh ] for any @xmath195 there exists an ntm @xmath32 over the binary alphabet @xmath178 , a polynomial longevity guard @xmath287 for @xmath32 , and an sgptime reduction @xmath322 .
fix @xmath195 .
we divide the proof of the theorem into two parts .
first , we construct a turing machine @xmath32 , a longevity guard @xmath287 for @xmath32 , and a ptime reduction @xmath96 from the original problem @xmath149 to @xmath308 .
then we show that @xmath96 is a composition of cs and cm reductions defined in sections [ se : change_size ] and [ se : change_measure ] and , hence , @xmath96 is an sgptime reduction @xmath322 . *
part i. * by theorem [ th : bin_alph ] we may assume that the alphabet of @xmath12 is binary .
now , since @xmath323 there exists an ntm @xmath324 such that : * @xmath324 has a halting computation on an input @xmath59 if and only if @xmath325 ; * @xmath324 has a polynomially bounded longevity guard . recall that @xmath168 is the ensemble of probability distributions @xmath326 since @xmath195 the ensemble @xmath327 is ptime computable . for @xmath328
define a function @xmath329 where @xmath31 is the lexicographic successor of @xmath68 .
for @xmath328 such that @xmath330 define @xmath331 to be the smallest ( in shortlex ordering ) binary string such that @xmath332 where @xmath333 is the binary expansion of a real number in the interval @xmath334 $ ] .
one can describe @xmath335 as follows .
assume first that @xmath336 .
since @xmath330 the binary expansions of @xmath337 and @xmath338 differ within the first @xmath339 bits after `` . '' , i.e. , @xmath340 for some @xmath341 . in this case
the case @xmath343 is similar .
it follows that for every @xmath344 such that @xmath345 we have @xmath346 and @xmath347 hence , @xmath348 .
define @xmath349 notice that for every @xmath344 , @xmath350 and @xmath351 .
now we describe an ntm @xmath32 , a function @xmath287 , and a reduction @xmath352 . if @xmath32 is defined and @xmath287 is a polynomial longevity guard for @xmath32 , then the reduction @xmath96 is defined for @xmath20 by @xmath353 where @xmath354 .
it is left to define @xmath32 and @xmath287 . the machine @xmath32 on a binary input @xmath60 executes the following algorithm : 1 . if @xmath60 is not in the form @xmath355 , where @xmath356 and @xmath7 , then loop forever .
2 . if @xmath357 then decode @xmath24 .
3 . if @xmath358 : 1 .
if @xmath359 then loop forever ; 2 . otherwise simulate @xmath324 on @xmath59 .
4 . if @xmath360 : 1 .
find the lexicographic smallest @xmath361 satisfying @xmath362 using divide and conquer approach ; 2 .
if @xmath363 or @xmath364 then loop forever ; 3 .
otherwise simulate @xmath324 on @xmath68 . by construction
, @xmath32 has a halting computation on @xmath365 if and only if @xmath366 for some @xmath344 and @xmath367 .
we claim that @xmath32 has a polynomial longevity guard @xmath287 .
indeed , since @xmath323 it follows that an ntm @xmath324 has a polynomial longevity guard , and all steps in the algorithm above , except simulation of @xmath324 , can be performed by deterministic polynomial time algorithms .
therefore , @xmath32 has a polynomial longevity guard @xmath287 , as claimed .
in particular , @xmath368 is a ptime reduction , as claimed . *
* now we prove that @xmath96 is an sgptime reduction .
we start with the following lemma . [
le : measure_decrease ] for every @xmath369 and @xmath328 the following inequality holds : @xmath370 for every @xmath328 there are two possibilities .
if @xmath371 , then @xmath372 and its measure is : @xmath373 if @xmath330 , then @xmath374 and its measure is : @xmath375 since @xmath376 and @xmath348 . in each case
the inequality ( [ eq : factor_rh_reduction ] ) holds . by construction of @xmath287 , all elements of @xmath377 are mapped to elements of size @xmath285 , hence , @xmath96 is size - invariant .
it follows that a function @xmath96 is a composition of a cs - reduction with the polynomial size - growth function @xmath378 and a cm - reduction with a polynomial density function @xmath379 .
thus , @xmath96 is an sgptime reduction .
let @xmath380 be a universal ntm such that : * @xmath380 accepts inputs of the form @xmath381 , where @xmath320 is the encoding of an ntm @xmath32 over a binary alphabet and @xmath7 ; * @xmath380 simulates @xmath32 on @xmath59 , i.e. , @xmath32 halts on @xmath59 if and only @xmath380 halts on @xmath381 , in which case they both have the same answer ( the same final configurations ) ; * @xmath380 has a polynomial - time slow - down , i.e. , there exists a polynomial function @xmath382 such that @xmath383 .
see , for example , @xcite on how such a deterministic turing machine @xmath380 can be constructed , a nondeterministic one can be constructed in a similar way .
[ th : red2bhu ] for every ntm @xmath32 over a binary alphabet @xmath6 , there exists a ptime computable function @xmath384 and an sgptime reduction @xmath385 .
let @xmath32 be an ntm over @xmath6 and @xmath287 a polynomial longevity guard for @xmath32 .
define ( in the notation above ) a function @xmath386 by @xmath387 were @xmath388 is the polynomial from the description of the machine @xmath380 above .
put @xmath389 .
clearly , @xmath96 gives a ptime reduction @xmath390 . to show that @xmath96 is an sgptime reduction
, one can argue as in the proof of theorem [ th : red2bh ] .
to carry over the argument , one needs the following inequality for every @xmath369 and @xmath328 : @xmath391 which differs from the inequality ( [ eq : factor_rh_reduction ] ) by a polynomial factor @xmath392 in the denominator .
the proof of this is similar to the one in lemma [ le : measure_decrease ] and we omit it .
there exists an ntm @xmath380 such that @xmath393 is sgptime complete . by theorem
[ th : red2bh ] for any @xmath172 problem @xmath149 there exists an ntm @xmath32 over a binary alphabet @xmath6 , a polynomial longevity guard @xmath287 of @xmath32 , and an sgptime reduction of @xmath149 to @xmath394 . by proposition [ pr : restriction2bhmg ] , there is an sgptime reduction of @xmath394 to @xmath284 . by theorem
[ th : red2bhu ] , there exists a ptime computable function @xmath395 and an sgptime reduction @xmath385 .
now , again by proposition [ pr : restriction2bhmg ] , there is an sgptime reduction of @xmath396 to @xmath393 .
hence , @xmath149 is sgptime reducible to @xmath393 , as claimed .
in this section we discuss some open problems on generic complexity . some of the well - known @xmath0-complete problems are in @xmath174 , or in @xmath175 , see @xcite for examples .
however , there is no general approach to this problem at present .
if the answer to one of the questions above ( in particular , the second one ) is affirmative , then it will imply that for all practical reasons @xmath0-complete problems are rather easy .
otherwise , we will have an interesting partition of @xmath0-complete problems into several classes with respect to their generic behavior .
it was shown in @xcite that the halting problem for one - end tape turing machines is in @xmath174 .
it remains to be seen if a similar result holds for turing machines where the tape is infinite at both ends .
it is known ( see @xcite ) that the classes of functions that are polynomial on average and generically polynomial are incompatible , i.e. , none of them is a subclass of the other .
nonetheless , the relationship between @xmath175-complete and @xmath0-complete on average is still unclear . to this end , the following problem is of interest . | in this note we introduce a notion of a generically ( strongly generically ) @xmath0-complete problem and show that the randomized bounded version of the halting problem is strongly generically @xmath0-complete
. * keywords . *
generic - case complexity , completeness , randomized problems , bounded halting problem . * 2010 mathematics subject classification .
* 68q17 . |
it has long been known that transport properties of a superconducting weak link may be strongly affected by external time dependent ( ac ) fields . to give an example , just after the discovery of the josephson effect it was found that an alternating rf - field will rectify the ac josephson current that results from a dc voltage bias and produce voltage steps shapiro steps in the current - voltage characteristics @xcite .
this effect can be viewed as a synchronization of the josephson current oscillations of frequency @xmath0 , generated by a bias voltage @xmath1 , and the current oscillations induced by the ac field . if the josephson coupling energy @xmath2 is a simple harmonic function of the superconducting phase difference @xmath3 , @xmath4 , the positions @xmath5 of the shapiro steps are given by the expression @xmath6 ( @xmath7 ) , where @xmath8 is the frequency of the ac field .
deviations from a simple harmonic relation between @xmath2 and @xmath3 result in additional `` fractional '' steps at voltages @xmath9 .
however their heights are exponentially small when @xmath10 . across the system .
an important difference between these two examples of a mesoscopic weak link is that the discrete energy levels inside the gap @xmath11 are well separated from the continuum levels ( dark bands ) for an scpt but not for an spc ( see text ) .
, width=302 ] new opportunities to influence the transport properties of josephson junctions by external ac fields appear in mesoscopic systems , where only few quantum states are responsible for the superconducting charge transfer . for such weak links the josephson energy
no longer a macroscopic quantity ( being proportional to the area of the junction ) may be of the same order of magnitude as the energy quantum @xmath12 of the external ac field .
if so , the alternating field may significantly affect the current by inducing resonant inter - level transitions between charge carrying states in the junction and thereby changing their population .
the internal structure of the mesoscopic weak links determines the physical origin of the discrete energy spectrum and the relative positions of the energy levels inside the superconducting gap @xmath13 .
superconducting point contacts ( spcs ) and single - cooper - pair transistors ( scpts ) are two fundamentally different examples of mesoscopic weak links . in an spc such as
the one sketched in fig .
[ fig1]a , the rapid variation of the superconducting phase across the junction gives rise to quasiparticle bound states ( andreev states ) that are localized to the vicinity of the contact .
the discrete energy levels ( andreev levels ) associated with these states appear inside the superconducting gap as shown in fig .
their positions @xmath14 are defined by the single electron scattering matrix and in the case of low transparency contacts one finds that @xmath15 . in an scpt , which is composed of two low - transparency tunnel junctions in series forming a small island as sketched in fig .
[ fig1]b , the coulomb blockade phenomenon @xcite strongly affects the structure of the energy spectrum . if the superconducting gap @xmath11 is greater than the coulomb energy of a grain charged by a single electron , the ground state of the scpt with a completely isolated central dot , for certain gate voltages , is degenerate with respect to adding one more cooper pair to the dot @xcite .
weak cooper - pair tunnelling that removes this degeneracy gives rise to current carrying cooper - pair resonant states , separated as illustrated in fig .
[ fig1]b from the continuum spectrum by an energy of order @xmath11 @xcite .
there is a qualitative difference between an spc and an scpt in how the energy spectrum depends on the superconducting phase change across the system . in an spc
the subgap level ( as described above this is a discrete energy level in the gap between the ground state and the continuum excitation spectrum ) is mixed with the continuum spectrum at certain values of the phase difference , while the cooper - pair resonant levels in an scpt are always separated from the continuum , forming a well defined two - level system .
the microwave dynamics of a biased spc was explored in ref . , where it was shown that quantum interference among a sequence of temporally localized inter - level transitions results in nontrivial transport properties of a josephson weak link of this type . in an spc , however , coherence may be preserved only during a fixed length of time , since the discrete andreev states die away as they periodically merge with the continuum spectrum at certain values of the phase difference .
( we recall that a voltage bias makes the phase difference increase linearly with time ) .
in contrast , since the discrete energy levels are here always separated from the continuum , the microwave - induced transport properties of a biased scpt is to a great extent defined by the coherent dynamics of the two - level system and strongly affected by interference processes . in order to demonstrate this phenomenon , we explore below the microwave dynamics and transport properties of a biased scpt
. applied to the gate electrode , which is capacitively coupled to the island .
a bias voltage @xmath1 is applied across the scpt giving rise to a linear time dependence of the phase difference @xmath16 between the leads , width=302 ]
a sketch of the scpt structure to be considered here is presented in fig .
[ fig2 ] , where a nanoscale superconducting island capacitively coupled to a gate electrode is shown to serve as a bridge between two bulk superconducting leads .
the hamiltonian for the system is @xmath17 where the cooper - pair number operator@xcite @xmath18 and the phase operator @xmath19 refer to the superconducting condensate on the grain , @xmath20 = i$ ] ; @xmath21 is the josephson coupling energy between the grain and the left ( right ) lead , @xmath22 is the total mutual capacitance and @xmath23 is an induced charge controlled by the gate voltage @xmath24 .
the first term in the hamiltonian governs the electrostatic energy associated with the number of extra cooper pairs on the grain , while the second and third terms describe the josephson coupling between grain and leads .
below we consider the symmetric case for which @xmath25
. therefore , the josephson part of the hamiltonian simplifies to @xmath26 .
it is well known that if the characteristic coulomb energy @xmath27 associated with charge fluctuations due to cooper pair exchange between grain and leads , is significantly larger than the josephson energy @xmath28 , while at the same time the coulomb energy @xmath29 related to _ single _
electron fluctuations is smaller than the superconducting energy gap , the low - energy states of the grain are given by a superposition of only two charge states that differ in charge by 2e .
below we will assume that the gate voltage is chosen in such a way that @xmath30 where @xmath31 .
in that case the state of the grain is a superposition of the charge state @xmath32 , describing the neutral grain , and the charge state @xmath33 with one extra cooper pair .
if @xmath34 the coulomb energy in the states @xmath32 and @xmath33 are the same .
this degeneracy is , however , removed by the josephson coupling that induces cooper pair exchange between the mesoscopic grain and the bulk electrodes . as a result
the eigenstates of the hamiltonian ( [ hamiltonian ] ) with eigenvalues @xmath35 are given by symmetric and antisymmetric superpositions of the states @xmath32 and @xmath36 .
these states , which we will refer to as cooper - pair resonant states , will if occupied carry currents given by @xmath37 . in the single - cooper - pair transistor ( scpt ) of fig .
dashed lines represent microwave - induced inter - level transitions occurring at times @xmath38 , when the resonance condition @xmath39 is satisfied .
, width=302 ] a bias voltage makes the phase difference over the junction change with time , @xmath40 .
this time dependence generates a periodic variation of the relative positions of the josephson levels as well as a periodic change in the current that can be carried by the corresponding resonant cooper pair states ( see fig .
[ fig3 ] ) .
the total josephson current through the junction depends on the relative `` population '' @xmath41 of the levels and is given by the formula @xmath42 ( @xmath43 are probabilities to find the system in the states @xmath44 , @xmath45 ) .
one can find important differences between the ac josephson effect in such a system and in ordinary josephson junctions .
if the coulomb blockade is totally lifted , there are no inter - level transitions as the levels cross each other and consequently @xmath46 is controlled only by relaxation processes .
even if the temperature is much lower than the characteristic inter - level distance , a weak relaxation to the ground state with rate @xmath47 leads to an equalization of the level populations . as a result
the josephson current will be proportional to @xmath48 and in the limit @xmath49 not only the dc but also the ac current through the josephson junction vanishes .
the situation is different if an oscillatory voltage is applied to the gate .
a time dependent gate voltage makes the induced charge vary in time as well , @xmath50 .
this in turn gives rise to inter - level transitions which control the level populations and by this means one may stimulate the josephson current .
the most interesting case , which will be considered below , appears when the frequency of the gate voltage oscillations @xmath8 is of the same order as the frequency of charge fluctuations on the grain @xmath51 . in this case
the inter - level transitions have a resonant nature and significantly influence the cooper pair transport even for relatively small variations of the gate potential . in the present work we consider a situation where the energy quantum @xmath52 of gate voltage oscillations , is of the same order but smaller than the maximum inter - level energy distance @xmath53 . at the same time we take @xmath8 to be much larger than the rate of the gate - voltage induced resonant inter - level transitions , @xmath54 . under these conditions and if one considers the time evolution of the system at low bias voltages , @xmath55 , one can neglect inter - level transitions except during short time intervals @xmath56 near the times
@xmath38 , when the resonance condition@xcite @xmath57 is fulfilled ( @xmath58 ) .
during these time intervals landau - zener type inter - level transitions will occur ( see fig .
[ fig3 ] ) .
therefore the evolution of the system we are interested in is given by a sequence of scattering events separated by periods of `` free '' evolution , during which the level populations do not change .
despite the fact that the level positions vary periodically in time with period @xmath59 , the dynamics of the system , strictly speaking , is not periodic .
this is because the phases of the complex elements of the scattering matrix depend on the phase @xmath60 which defines the value of induced charge at time @xmath61 .
this fact is of great importance for the transport properties of the system . as we will show below , if the ratio @xmath62 is an irrational number the dynamics of the system is qusiperodic and the dc josephson current is equal to zero .
but if this ratio is a rational number , then a dc josephson current does flow through the junction .
the maximum value of the current appears when @xmath62 is an integer .
using the cooper - pair resonant states representation one can recast the hamiltonian ( [ hamiltonian ] ) in the form @xmath63 where @xmath64 and @xmath65 are the pauli matrices .
as already anticipated , we will assume that the following relations are fulfilled : @xmath66 here @xmath67 and @xmath68 are integer numbers .
the final result , other than details of calculations , does not depend in any essential way on the arithmetic properties of @xmath69 , and for definiteness we take @xmath68 to be an even number ( @xmath70 is odd ) . under these conditions
the hamiltonian ( [ ham ] ) is a periodic function of time with period @xmath71 .
the dc josephson current is given by expression @xmath72 where @xmath73 is the current operator , @xmath74 is the density matrix of the two - level system , while its time evolution is governed by the liouville - von neumann equation @xmath75- \imath \hbar \nu ( \hat \rho - \hat \rho_{eq})\,.\ ] ] here @xmath76 is the quasistatic density matrix for the unperturbed hamiltonian ( @xmath77 ) , @xmath78 is the inverse temperature , @xmath48 is the relaxation rate .
the formal solution of eq .
( [ 13 ] ) can be expressed as @xmath79 where @xmath80\ ] ] is an evolution operator possessing the symmetry properties @xmath81 which derives from the symmetry of the hamiltonian ( [ ham ] ) , @xmath82 it follows that the density matrix @xmath83 satisfies the relation @xmath84 which we used in eq .
( [ 12 ] ) . in the limit
@xmath85 this relation together with eq .
( [ 16 ] ) immediately results in the equation @xmath86 for the density matrix @xmath87 . here
@xmath88 and @xmath89 equation ( [ 18 ] ) and the spectral properties of the unitary operator @xmath90 will play a central role as we proceed . in the limit
@xmath88 the solution of eq.([18 ] ) has the form @xmath91 where @xmath92 are the eigenvectors of the operator @xmath90 with eigenvalues @xmath93 and @xmath94 from the symmetry of the hamiltonian ( [ ham ] ) it follows that @xmath95 , @xmath96 and @xmath97 .
the expression @xmath98 gives the quasienergies of the periodic hamiltonian , @xmath99 ; we will refer to @xmath100 as a quasienergy factor .
the average josephson current , eq.([12 ] ) , can now be cast in the form @xmath101 in order to calculate the population coefficient @xmath102 and the average current in the quasienergy states , @xmath103 , it is convenient to introduce an effective hamiltonian , @xmath104 , of the form @xmath105 one can show that eq .
( [ 20 ] ) for the josephson current can be expressed as @xmath106 where @xmath107 is the appropriate quasienergy factor of the effective hamiltonian ( [ 255 ] ) .
only resonant inter - level transitions are of importance if the microwave interaction energy , @xmath108 , is much smaller than the characteristic inter - level distance @xmath109 . ignoring the nonresonant part of the hamiltonian @xmath104 ( _ i.e. _ using the resonance approximation )
one finds that the evolution operator @xmath110 has the form @xmath111 here @xmath112 is the evolution operator generated by the hamiltonian @xmath113 , where @xmath114 from the symmetry of the hamiltonian @xmath113 it follows that the operator @xmath112 has the form @xmath115 the parameter @xmath116 is the resulting probability amplitude for an inter - level transition when the superconducting phase difference changes from @xmath117 to @xmath118 .
equation ( [ 122 ] ) permits us to represent the desired evolution operator @xmath90 of eq .
( [ 19 ] ) as @xmath119 e^{i\sigma_3(\theta-\varphi/2 ) } \ , , \nonumber\end{aligned}\ ] ] where @xmath120 and the sign ( @xmath121 ) depend on the arithmetic properties of the natural numbers @xmath122,and @xmath68 .
now one can see that the eigenvalues of the unitary operator @xmath123 can be found by solving the difference equation @xmath124 with the boundary condition @xmath125 , @xmath126 a `` quasiperiodic '' equation of this type may be analyzed by method developed in ref . .
the result for the quasienergies @xmath100 from such an analysis is @xmath127 where the constants @xmath128 are irrelevant .
taking into account eqs .
( [ 112 ] ) and ( [ 02 ] ) the average current can now be rewritten as @xmath129 in this work we restrict ourselves to the low temperature limit , where @xmath130 . in this limit @xmath131 and , as a consequence , @xmath132 . therefore , in order to calculate the current at low temperatures one can use the hamiltonian @xmath133 .
as already mentioned , only resonant inter - level transitions are important here . in the time interval
@xmath134 the resonant condition @xmath135 is fulfilled only at the two times @xmath136 and @xmath137 as these `` points '' in time are reached , inter - level landau - zener transitions @xcite ( see fig . [ fig3 ] ) occur with probability amplitude @xmath138 , where @xmath139^{- \frac{1}{2}}\right]\,.\ ] ] dividing time interval @xmath140 into periods of free evolution ( where interactions with the microwave field can be neglected ) and describing the passage through the resonant points by a landau - zener scattering matrix , one finds the probability for an inter - level transition to be @xmath141 and the final expression for the microwave - induced dc josephson current becomes @xmath142 equation ( [ final ] ) determines the current - voltage characteristics of the scpt studied , which depend crucially on the frequency of the microwave irradiation .
its validity is restricted to the special values of @xmath8 and @xmath1 determined by eq .
( [ 09 ] ) averaging the current over a long time @xmath143 ( corresponding to a large value of @xmath68 , since @xmath71 ) , and plotting it as a function of the parameter @xmath62 , results in many sharp features .
( notice that two values of @xmath62 which are arbitrarily close to each other might correspond to widely different values of @xmath68 in eq.([09 ] ) ) .
it is clear , however , that any upper limit @xmath144 for the allowed averaging time will tend to smooth out the sharp features .
the value of @xmath144 might be determined by either a relaxation time or a finite observation time . in either case currents calculated for two values of @xmath145
will be indistinguishable if @xmath146 .
this argument provides a justification for smoothening the current as given by eq .
( [ final ] ) in order to get a regular dependence @xmath147 .
an appropriate procedure is to evaluate eq .
( [ final ] ) at all rational points for which @xmath148 .
then a smooth dependence @xmath149 can be obtained by making use of an interpolation procedure .
the result of such an interpolation procedure is presented in fig .
[ fig4 ] . in units of @xmath150 , plotted as a function the microwave frequency @xmath8 normalized to @xmath151 .
the result was obtained with @xmath152 and @xmath153 ( see text).,width=302 ]
we have shown that the microwave dynamics of a single - cooper - pair transistor is strongly influenced by interference effects caused by the quantum dynamics of the superconducting dot , which as in fig .
[ fig2 ] forms the central island in the transistor structure .
the sharp peaks that were found in the current - voltage ( @xmath154 ) characteristics for `` fractional '' values of the voltage , as shown in fig .
[ fig4 ] , are a signature of a resonant interaction with the microwave field .
the peaks occur when @xmath155 , where @xmath156 , @xmath70 , and @xmath68 are integers ; they are finite if @xmath157 and do not vanish in the limit of weak microwave irradiation , the interaction with the microwave field results in multiple , coherent transitions between resonant cooper - pair levels .
this picture , which differs qualitatively from the shapiro effect @xcite , is a direct manifestation of the role of strong coulomb correlations in the nonequilibrium superconducting dynamics of mesoscopic weak links .
the coulomb blockade of cooper pair tunnelling reduces the number of relevant charge states to two , and is responsible for the ensuing quantum coherent two - level dynamics of cooper pair resonant states .
a single - cooper - pair transistor requires the coulomb blockade energy @xmath158 to be much larger than the josephson coupling energy @xmath28 . if instead @xmath159 , a large number of dot charge states with different number of cooper pairs are degenerate . as a result tunnel coupling to the leads form dot states with a well defined superconducting phase .
microwave properties in this limit are governed by the _
classical dynamics _ of the superconductiong phases , caused by the driving voltage and the microvawe field .
if we were to imagine that @xmath158 could be diminished from a value much larger than @xmath28 to a value much smaller than @xmath28 , we would witness a transition from a situation where the quantum dynamics of the superconducting phase was relevant to one where the classical dynamics would dominate . during such a transition the resonance microwave absorption would disappear and be replaced by standard shapiro signatures , known from macroscopic weak links .
this work has been supported financially by the royal swedish academy of sciences , kva , and by the swedish research council , vr , ( lyg , ris ) .
sik acknowledges the hospitality of the department of applied physics at chalmers university of technology and gteborg university .
0 s. shapiro , _ phys .
lett . _ * 11 * , 80 ( 1963 ) .
i. o. kulik and r. i. shekhter , _ sov .
jetp _ , * 41 * , 308 ( 1975 ) .
k. a. matveev , m. gisselflt , l. i. glazman m. jonson and r. i. shekhter , _ phys .
lett . _ * 70 * , 2940 ( 1993 ) .
f. w. j. hekking , l. i. glazman , k. a. matveev and r. i. shekhter , _ phys .
lett . _ * 70 * , 4138 ( 1993 ) .
y. nakamura , c. d. chen , j. s. tsai , _ phys .
lett . _ * 79 * , 2328 ( 1997 ) .
y. nakamura , yu . a. pashkin and j. s. tsai , _ nature _ * 398 * , 786 ( 1999 ) ; y. nakamura and j. s. tsai , _ j. low temp . phys .
_ * 118 * , 765 ( 2000 ) .
a. cottet , d. vion , a. aassime , p. joyez , d. esteve and m. h. devoret , _ physica c _ * 367 * , 197 ( 2002 ) . l. y. gorelik , a. isacsson , y. m. galperin , r. i. shekhter and m. jonson , _ nature _ * 411 * , 454 ( 2001 ) . j. r. friedman and d. v. averin , _ phys .
lett . _ * 88 * , 050403 ( 2002 ) .
l. y. gorelik , n. i. lundin , v. s. shumeiko , r. i. shekhter and m. jonson , _ phys .
lett . _ * 81 * , 2538 ( 1998 ) .
we count the excess cooper pairs relative to the neutral state .
hence the eigenstate of the operator @xmath18 with eigenvalue @xmath160 corresponds to a neutral grain . for the sake of definiteness we let @xmath161 .
l. y. gorelik , s. i. kulinich , y. m. galperin , r. i. shekhter and m. jonson , _ phys .
* 78 * , 2196 ( 1997 ) .
l. d. landau , _ phys .
* 2 * , 46 ( 1932 ) ; c. zener , _ proc .
( london ) a _ * 137 * , 696 ( 1932 ) ; k. mullen , e. ben - jacob , y. gefen and z. schuss , _ phys .
lett . _ * 62 * , 2543 ( 1989 ) . | we consider the microwave dynamics and transport properties of a voltage - biased single - cooper - pair transistor .
the dynamics is shown to be strongly affected by interference between multiple microwave - induced inter - level transitions . as a result
the magnitude and direction of the dc josephson current are extremely sensitive to small variations of the bias voltage and to changes in the frequency of the microwave field . |
this research was supported the epsrc under grant gr / n15665 .
we wish to thank a. csoto for a discussion of his work .
a. csoto , _ phys .
c _ * 48 * , 165 ( 1993 ) ; * 49 * , 2244 , 3035 ( 1994 ) ; _ nucl .
a _ * a631 * , 783c ( 1998 ) ; _ few - body systems suppl .
_ in press ( nucl - th/0010105 ) ; t. myo , s. aoyama , k. kato and k. ikeda , _ nucl . phys . a _
* a738 * , 298 ( 2004 ) ; k. kato , _ riken review _ * 39 * , 61 ( 2001 ) .
r. f. bishop _
_ , in _ condensed matter theories volume 6 _ , ed .
s. fantoni and s.rosati ( plenum press , new york , 1991 ) , p. 405 ; r. f. bishop _ et al . _ , in _ exploring many - body theories in light nuclei _ , ed . w. greiner and h.stcker ( plenum press , new york , 1993 ) , p. 605 . | we investigate resonances in light halo nuclei using a fully microscopic cluster model and the complex scaling method . we make use of the hermitian representation of the complex scaling method .
the general structure of the cluster model is that of a correlation operator acting on a starting function that describes a number of neutrons relative to an alpha - particle . the correlation operator is expanded in terms of a small non - orthogonal set of gaussian basis states and we make use of a simplified , central but state - dependent , interaction .
the many - body integrals required for the computation are evaluated by a variational monte - carlo algorithm .
we show how to obtain resonant states for both @xmath0he and @xmath1he .
the study of resonant structures in loosely bound systems , such as halo nuclei , is an interesting area of nuclear physics that promises to be a rich source of information on many - body dynamics .
there exists a variety of methods for analyzing such structures and in this work we consider an approach that belongs to the category of cluster models , which have proven to be powerful tools in analyzing the structure of such loosely bound systems . the most common way in which present cluster models describe the continuum structure of light nuclei is by treating only the most important degrees of freedom properly , and ignore the difficulties with the pauli principle that entails .
in contrast to such approaches , we employ a fully microscopic cluster model in order to provide a variational approximation to the schrdinger equation with , in this case , a slightly simplified nucleon - nucleon interaction .
a related microscopic cluster model that can deal with three - body resonances by complex - scaling in coordinate space was developed by csoto and applied to @xmath1he , @xmath1li and @xmath1be @xcite .
this model can account in many cases both for the correct nuclear physics and proper three - body dynamics . among other results
the authors predict the non - existence of the soft dipole resonance in @xmath1he .
another example of a similar study of halo nuclei by the complex scaling method through a microscopic model can be found in ref .
@xcite .
an alternative to the use of the complex scaling in coordinate space is to perform a contour deformation in momentum space @xcite .
this was applied at the level of a two - body problem and an extension to few - body borromean halo systems is in progress .
despite the success of the two- and three - body microscopic cluster models the existing fully microscopic ones have problems when considering resonances @xcite .
the ab - initio approaches , e.g. , the monte - carlo approach of ref .
@xcite , even though very well suited for the bound - state problem , have difficulties describing resonances . by using the complex scaling method @xcite in the ` direct approach ' , our model can handle resonances correctly in a straightforward way .
the wavefunction we employ is fully microscopic and is explicitly antisymmetrized so that the pauli exclusion principle is exactly satisfied .
the general structure is that of a correlation operator acting on a reference function that describes a number of neutrons relative to an alpha - particle .
we include a multiplicative jastrow factor for the short - range correlations and an additive configuration - interaction ( ci ) term for the long - range correlations .
this last term only introduces pair correlations , whereas the jastrow term also includes higher correlations .
this mechanism has proved to be efficient in describing nuclear structure in various applications to closed and open shell systems , where all particles are treated on an equal footing @xcite .
recently , this has also been applied @xcite to nuclei with @xmath2 .
one advantage of the jastrow - ci scheme is the explicit inclusion of translational invariance and the fact that the correlation operators can be sufficiently well approximated by a small non - orthogonal basis set that makes application to heavier nuclei possible . for the moment we confine ourselves to central v4 interactions and in particular we make use of the s3 interaction of afnan and tang @xcite . for the computation of the many - body matrix elements we make use of the variational monte carlo method , which we have analyzed in detail @xcite as applied to this cluster model . in this letter
we focus on the cases of @xmath0he and @xmath1he .
although experimentally @xmath0he is unbound while @xmath1he is lightly bound , the absence of spin - orbit force in our simplified force results in both nuclei being unbound .
nevertheless , both nuclei exhibit a rich resonance structure .
the current work thus provides an excellent testing ground for the study of resonances , even though the nuclear physics may not be wholly realistic .
there exists many different version of the complex scaling technique . in the original complex - coordinate method ( see , e.g. , ref .
@xcite ) the resonance position @xmath3 and width @xmath4 are calculated from the complex eigenvalue @xmath5 of a non - hermitian hamiltonian .
the non - hermitian hamiltonian is obtained from the original one by a transformation of the particle coordinates . here
we shall only use the most straightforward one , @xmath6 , where the angle @xmath7 is the scaling parameter .
such a transformation leaves the bound states unaffected and rotates the positive energy scattering states by an angle of @xmath8 into the lower part of the complex plane relative to the threshold @xcite .
resonant states @xmath9 appear as complex eigenvalues in the lower half complex plane that are invariant with respect to changes in the scaling parameter @xmath7 , as long as this angle is large enough to uncover the resonances .
[ a more detailed review of the complex scaling method can be found in @xcite . ] for a complex scaled hamiltonian , the complex variational principle , unlike the conventional one , is a stationary principle rather than a minimum principle for either the resonance position or width .
this is not very attractive , and moiseyev @xcite has formulated a representation of the complex - coordinate method in which the resonance position @xmath10 and width @xmath11 are variational parameters of a hermitian hamiltonian which gives additional stability , especially when working with small basis sets , as in this work .
it is also gives a simple algorithm to follow the path of a given eigenvalue as @xmath7 changes .
the variational method was reformulated by bylicki @xcite , into a simpler and more convenient form . in this method ,
we first rewrite the complex scaled schrdinger equation as @xmath12 where @xmath13 and @xmath14 are the real and imaginary parts of the scaled hamiltonian @xmath15 . in matrix form this becomes @xmath16 since the exact eigenvalue of @xmath15 is unknown we can consider the alternative equation @xmath17 where @xmath18 is an approximation to the exact wavefunction @xmath19 . for a particular value of the scaling parameter @xmath7 the _ best _ choice of @xmath18
is obtained by minimizing the size of @xmath20 , and @xmath21 if @xmath22 . within this approximation
the parameters @xmath10 and @xmath23 that appear in @xmath24 no longer represent the exact real and imaginary parts of the complex energy but become variational parameters .
thus in addition to the linear variational principle that is employed for the wavefunction we also have to vary @xmath10 and @xmath23 so as to minimize @xmath25 .
although @xmath20 has both positive and negative values one interesting property of eq .
( [ eq2 ] ) is that the spectrum of eigenvalues is symmetric around 0 , thus if @xmath25 is an eigenvalue so is @xmath26 .
furthermore , the equation @xmath27 is useful for obtaining bounds for the resonance position and width and it can be used in an iterative method for determining the values of @xmath10 and @xmath23 that minimize @xmath25 @xcite .
the many body hamiltonian is of the form @xmath28 in this letter , we use a semi - realistic central nucleon - nucleon interaction ( v4 ) interaction that has the general form @xmath29 here the @xmath30 are the spin / isospin - dependent part composed of spin / isospin - exchange operators .
the radial part of each channel potential is expanded as a sum of gaussian functions .
we use the s3 interaction of afnan and tang @xcite . for the trial wavefunction
we consider the linearized approximation of the many - body wavefunction in terms of correlation operators acting on an uncorrelated reference function @xcite .
this provides a translationally invariant description of the many - body problem and was applied to a number of closed - shell systems in terms of an alpha - cluster model @xcite .
the wavefunction @xmath19 is given by @xmath31 where @xmath32 is the reference function , that we take to be the product of the different cluster wavefunctions .
the correlation operator @xmath33 is of the form @xmath34 this type of wavefunction ( referred to as the j - tici(2 ) scheme ) has been extensively used for the alpha - particle @xcite where it was shown to provide an adequate description for the ground - state properties .
an important aspect of this approximation is the expansion of the correlation operator in terms of the non - orthogonal gaussian functions @xmath35 .
only a very small number of components is required for a reasonable convergence , something that makes calculations in the j - tici(2 ) scheme less expensive than other microscopic methods .
due to the complexity of the correlation operator we can make use of the variational monte carlo method for the evaluation of the matrix elements , where the probability density function @xmath36 is taken to be the square of one of the components in the expansion of the wavefunction , i.e. , @xmath37 this is a natural choice for the pdf since the wavefunction can always be written as the product of @xmath36 and a function @xmath38 .
this is discussed in detail in ref .
@xcite .
thus eq .
( [ eq1 ] ) is approximated in terms of a non - orthogonal expansion .
the non - orthogonality requires that in eq .
( [ eq2 ] ) the real and imaginary parts of the complex scaled hamiltonian are modified according to @xmath39 @xmath40 represents the overlap matrix and @xmath41 , where @xmath42 is the matrix with the eigenvectors of @xmath40 as its columns and @xmath43 is the diagonal matrix with the eigenvalues of @xmath40 as its elements .
@xmath7-trajectories for the spectrum of the alpha - particle for various values of the rotation parameter @xmath7 between 0 and 0.65 radians .
the value of the ground state corresponding to @xmath44 is -27.3 mev .
the threshold is not stationary but rotates into the upper quadrant of the complex plain , while the rest of the eigenvalues move clockwise into the lower complex plain.,width=377 ] we firstly examine the application of the formalism of moiseyev and bylicki to the alpha - particle , where we do not expect to find any resonances . for the alpha - particle reference function
we use the spherically symmetric harmonic oscillator ground state .
it is an easy numerical calculation both in terms of computation time and complexity .
we can go one step further by introducing one or two additional nucleons , corresponding to the cases of @xmath0he and @xmath1he .
experimentally the alpha - particle is a stable nucleus with 20 mev difference between the ground and first excited state .
searching for resonances is thus not extremely important , but the calculation will help to highlight the implications of the approximation scheme used .
the results are displayed in fig .
[ fig1 ] .
the threshold for this calculation is the ground state energy as obtained from the variational approximation ( @xmath45 -27.3 mev ) . as can be seen from fig .
[ fig1 ] the ground - state remains almost constant for small @xmath7 but when we increase @xmath7 further it eventually moves .
this is not unexpected since the variational ground state is only an approximation to the exact threshold and it has a non - zero overlap with continuum states .
it also suggests that we should optimize our basis to change with @xmath7in this context we are especially concerned about the jastrow factor . the rest of the eigenvalues move clockwise into the lower complex plane , eventually turning up as well .
the behaviour of the threshold is somewhat of a worry , and shows that our variational form is not too good for finite @xmath7 . in all likelihood
this is caused by the gaussian functions in the jastrow correlations ( i.e. , the short range behaviour of the wave function ) , but this requires confirmation .
clearly , we expect all calculations to fail for @xmath46 , where all gaussians become oscillatory .
experimentally @xmath0he is unbound by 0.798 mev and is observed as a @xmath47 resonance in the neutron scattering on @xmath48he . in our approximation
the case of @xmath0he is that where a neutron is added to the alpha particle wavefunction , where the additional coordinates are specified relative to the alpha - particle center - of - mass .
the additional neutron is placed within a spherical shell and the many - body wavefunction has the form @xmath49 where @xmath50 is a solid harmonic , while @xmath51 represents the spin and isospin degrees of freedom . in the function
@xmath52 the parameter @xmath53 represents the distance from the alpha - particle , while @xmath36 represents the width of the shell .
@xmath54 is the relative distance of the neutron from the alpha - particle .
the linear correlation operator is invariant under permutation of particle labels and can be taken outside the antisymmetrizer .
@xmath7-trajectories obtained for @xmath0he @xmath55 ( @xmath56 ) for various values of the width parameter @xmath36 . in each plot the rotation parameter @xmath7 is varied in equal steps and the @xmath7-trajectories evolve from right to left as @xmath7 is increased.,width=302 ] both @xmath53 and @xmath36 are variational parameters and in principle we can have a number of them , i.e. expand over a number of shells each placed at a different distance from the alpha - particle with a given width . however , as @xmath36 increases the value of @xmath53 becomes insignificant and we can consider a single shell provided @xmath36 is large enough .
the results obtained for @xmath55 ( @xmath56 ) are illustrated in fig .
[ fig2 ] , where we have plotted @xmath7-trajectories for various values of the variational parameter @xmath36 . for relatively small values of @xmath36 the trajectory is a smooth curve . as @xmath36
is increased this curve gains a turning point that eventually becomes a cusp ( left - hand part of fig .
[ fig2 ] ) . further increase of @xmath7 leads to loop - like trajectories ( right - hand part of the figure ) .
the smallest absolute difference @xmath57 between adjacent values of the @xmath7-trajectory for a particular value of @xmath36 .
the error bars are due to the monte carlo sampling .
the value of @xmath36 for which @xmath57 is minimum indicates the position of the resonance.,width=302 ] in order to find the best @xmath7-trajectory we refer to the derivative of the eigenvalue with respect to @xmath7 . at resonance
we expect that the energy is stable with respect to @xmath7 , i.e. @xmath58 . thus the best @xmath36 is that where @xmath59 is closest to zero .
this is illustrated in fig .
[ fig3 ] , where we can see the minimum value of @xmath57 of the @xmath7-trajectory ( corresponding to the absolute distance between two adjacent points ) for a particular value of @xmath36 .
the presence of a resonance is observed as the minimum value of @xmath60 that occurs in a range of @xmath36 between 5 and 6 . according to fig .
[ fig2 ] this corresponds to a cusp in the trajectory .
the threshold for the problem ( which is not shown in this figure ) follows the trajectory for the @xmath48he ground state , as shown fig .
[ fig1 ] , closely .
this suggests little influence from the halo structure , on this state , and little influence from this state on the continuous spectrum of @xmath0he . from the obtained @xmath7-trajectories we can infer a complex eigenvalue corresponding to a resonance in the region ( @xmath61 mev,@xmath62 mev ) , for @xmath63 . at this point
the real part of the threshold has moved , but it has not developped an appreciable imaginary part .
this movement means that the threshold subtraction is somewhat uncertain .
nevertheless , it seems most sensible to use the @xmath44 @xmath48he binding energy as threshold , @xmath64 mev .
we thus find that a resonance [email protected] mev above threshold with a width of @xmath66 mev .
the uncertainty arising from the anomalous movement of the threshold with @xmath7 is unknown , but as argued above probably small .
having ensured the applicability of the complex - scaling method to @xmath0he we proceed to the case of @xmath1he . in the case of @xmath1he
we experimentally have a borromean nucleus : while @xmath0he is unbound , the @xmath1he ground state is stable .
this state lies only 0.973 mev below the threshold for decay into an alpha - particle and two neutrons ( @xmath48he + @xmath67 ) and thus @xmath1he is very weakly bound .
the first resonance of @xmath1he ( @xmath68 ) lies 1.797 mev above the ground state and has a strong decay to the @xmath48he + @xmath67 channel .
we find no bound states in our calculation due to the lack of spin - orbit force in the simplified nucleon - nucleon interaction used .
however , we can examine the structure of low - energy resonances in this light nucleus . in our model @xmath1he
is described by two neutrons added to the alpha particle wavefunction , where the additional coordinates are specified relative to the alpha - particle center - of - mass .
similar to the case of @xmath0he the two - neutrons are placed within a spherical shell , each parameterized in terms of a shifted gaussian with an additional term describing the interaction of the two neutrons .
this leads to a reference function , @xmath32 , that has the form @xmath69 where @xmath70 is the relative distance of the @xmath71th neutron from the alpha - particle , while @xmath72 is the relative distance between the two neutrons .
@xmath73 is a solid harmonic that couples the two neutrons relative to the alpha particle to a total orbital momentum @xmath74 .
the results obtained for @xmath1he @xmath75 ( @xmath76 ) for various values of the width parameter @xmath36 . in this case @xmath36 is related to the separation of the two neutrons that are placed in a shell around the alpha - particle . in each plot
the rotation parameter @xmath7 is varied in equal steps and the @xmath7-trajectories evolve from right to left.,width=302 ] the calculation is considerably more difficult than that of @xmath0he , because of the increase both in configuration space and in the number of variational parameters .
however , it is sufficient to consider a restricted set of variational parameters for a qualitative understanding .
the results obtained look rather similar to those of @xmath0he . in fig .
[ fig4 ] , the separation between the alpha - particle and the dineutron center - of - mass is kept fixed , while the the separation of the two - neutrons , represented by @xmath36 , is varied . for small values of @xmath36
the @xmath7-trajectories are smooth curves , while a cusp is observed when @xmath36 is sufficiently large .
again , this eventually changes into a loop . despite the qualitative nature of the results
a clear resonant structure is obtained . in both @xmath0he and @xmath1he calculations
only 9 components are used for the expansion of the correlation operator , and this is enough for providing convergence .
the small basis set allows the calculations to be carried without any significant restrictions in the numerical method .
the only difficulty that arises in going to heavier systems will be due to the explicit antisymmetrization . as a result of the hermitian representation of
the complex variational principle is simple to obtain @xmath7-trajectories for the complex eigenvalues . in order to identify the resonance we use the condition that at resonance the complex energy should be stationary with respect to the scaling parameter . for the moment the results obtained are somewhat qualitative .
although the numerical error due to the monte - carlo sampling is trivial for a particular point in the @xmath7-trajectories , the optimal value of the variational parameters ( e.g. @xmath36 ) that gives the resonance could only be obtained within a range .
precise determination of that would require a much more elaborate calculation .
another point that needs further investigation is the fact that the basis set employed does not provide a fixed threshold for the alpha - particle .
it is expected that this will further influence the position of the resonance but is not clear at this stage in exactly what way .
nevertheless , the difference between the @xmath7-trajectories for the alpha - particle and those of @xmath0he and @xmath1he is quite profound .
the position where the resonance occurs for both @xmath0he and @xmath1he is at a point where the motion of the alpha - particle continuum due to the scaling is clockwise into the lower complex plain .
therefore , it is extremely unlikely that the resonance can be attributed to the anomalous behavior of the alpha - particle threshold .
we have examined the application of a fully microscopic cluster model to resonances in halo nuclei .
we have obtained low - lying resonances for both @xmath0he and @xmath1he by determining the point where the complex energy is stationary with respect to a scaling parameter .
this is encouraging since experimentally it is expected that both systems have low - lying resonances .
the central outcome of our investigation is that the complex - scaling method can be successfully applied to our fully microscopic model for weakly bound nuclei .
this was achieved with a relatively simple approximation , where the number of basis functions is quite small . within our framework of a non - orthogonal expansion it is convenient to make use of the variational eq .
( [ eq2 ] ) , rather than the standard complex variational principle .
although both methods give the same results , the hermitian representation provides a systematic way of obtaining individual @xmath7-trajectories . apart from resonances our method provides an interesting way to investigate the correlation mechanism used , which is widely applied ( jastrow - ci ) .
despite the fact that such a correlation mechanism gives a good description of the bound - state properties of the alpha - particle , the ground state is not stationary under complex scaling .
this suggests that the basis used to expand the correlation mechanism can be improved .
it is possible to obtain more accurate positions for these resonances and in the future we plan to improve and extend the calculation in order to get results of a more quantitative nature .
this would include both increasing the accuracy ( more computer time ) , and obtaining upper and lower bounds ( as suggested in @xcite ) .
in this paper we are only interested in a qualitative understanding of the method .
however , a more complete interaction can be used , which should lead to results comparable to the experimental data .
hence , the first natural step in extending the above work is to include a spin - orbit force in the nucleon - nucleon interaction .
it is expected that this will be adequate to produce a bound state for @xmath1he and at the same time allow the results for resonances to be of a more quantitative nature . |
delay differential equations have received much attention over the years because of the significant role of delayed feedback in the dynamics of many physical and biological systems @xcite . on the one hand
, delay - differential systems often exhibit multistability i.e .
, the coexistence of several attractors and multistability enables such systems to act as memory devices @xcite , an idea first suggested by ikeda and matsumoto @xcite . on the other ,
the study of delay systems is motivated by the fact that these systems exhibit high - dimensional chaos and therefore can be used in communication systems based on chaotic synchronization , to securely encrypt information into their chaotic outputs @xcite .
several authors have recently shown the existence of two different regimes of synchronization in nonidentical coupled time - delayed differential equations , in a master - slave configuration @xcite .
when the master and the slave systems have both the same amount of delayed intrinsic feedback and the delay time @xmath3 is the same for both systems , the slave system variables synchronize with the master system variables with a lag time that is equal to the delay of the coupling , @xmath4 . in other words , @xmath5 , where @xmath6 and @xmath7 represent the states of the master and slave systems respectively . on the other hand ,
when the master and the slave systems have the same amount of delayed intrinsic feedback and the feedback coefficient of the master system is equal to the sum of the feedback coefficient of the slave system and the coupling coefficient , the systems synchronize with a different lag - time . in this case @xmath8 .
two synchronization regimes are therefore possible for appropriate choices of the slave map parameters , as shown analytically in specific instances @xcite . in this paper
we show that these two synchronization regimes can also occur in time - discrete dynamical systems . in section 2
we consider generic nonidentical delayed maps , unidirectionally coupled , and define the two synchronization regimes . as an illustration , in section 3 we analyze logistic and hnon delayed maps .
finally , section 4 presents our conclusions .
we consider a generic master map of the form @xmath9 where @xmath10 and @xmath11 are parameters .
the slave map is given by @xmath12 if the parameters of the slave map are tuned in such a way that @xmath13 and @xmath14 ( case i ) , the slave map reduces to @xmath15 . \label{lag1}\ ] ] full synchronization can be expected for sufficiently large @xmath16 on the synchronization manifold @xmath17 . on the other hand ,
if the parameters of the slave map are tuned in such a way that @xmath18 and @xmath19 ( case ii ) , the slave map reads @xmath20 , \label{lag2}\ ] ] and full synchronization may occur on the synchronization manifold @xmath21 .
therefore , depending on the parameters of the slave map , @xmath22 and @xmath23 , full synchronization can take place with two different lag times , @xmath24 in the case of eq .
( [ lag1 ] ) and @xmath25 in the case of eq .
( [ lag2 ] ) . note that , in case i , one can have anticipated syncronization for @xmath26 [ refs .
de anticipated synchronization ] .
the actual possibility of observing full synchronization in either case is determined by the stability of the synchronized state .
linear stability analysis of eqs .
( [ lag1 ] ) and ( [ lag2 ] ) can be carried out by noticing first that the two equations can be written in a unified form as @xmath27 , \label{gen}\ ] ] with @xmath28 . in eq .
( [ lag1 ] ) we have @xmath29 , whereas in eq .
( [ lag2 ] ) we have @xmath30 .
the synchronization manifold is @xmath31 . applying a perturbation @xmath32 , replacing in eq .
( [ gen ] ) , and taking into account eq .
( [ master ] ) we get , to the first order in the perturbation , @xmath33 where primes indicate derivatives .
equation ( [ delta ] ) can be formally integrated by introducing a linear @xmath34-dimensional map , with @xmath35 , for a variable @xmath36 , with @xmath37 .
this equivalent map is given by @xmath38 where the elements of the matrix @xmath39 are given by the ( time - dependent ) coefficients in eq .
( [ delta ] ) @xcite .
the solution to eq .
( [ z ] ) reads @xmath40 so that the state of full synchronization is linearly stable if all the eigenvalues of the evolution matrix @xmath41 vanish for @xmath42 .
whether this condition holds or not for a given value of the coupling constant @xmath16 can be readily verified by numerical means .
note that all the elements of matrix @xmath39 , given by the coefficients in eq .
( [ delta ] ) , involve a delay @xmath1 which thus acts as a uniform time shift .
this fact implies that , in the limit @xmath43 , the eigenvalues of @xmath41 become independent of @xmath1 .
consequently , the value of @xmath1 is irrelevant to the stability of full synchronization ( cf .
@xcite ) . note carefully that , generally , full synchronization for the two cases considered above will be stable on two different ranges of the coupling constant @xmath16 .
if the master map is chaotic , we expect that for sufficiently small and large @xmath16 full synchronization is unstable and stable in both cases , respectively , while for intermediate values only one of the cases admits full synchronization .
we have seen above that full synchronization is possible at two points in the ( @xmath22,@xmath23 ) parameter space , with different lag times in each case . to encounter such synchronized states ,
the slave - map parameters must be exactly tuned on one of those synchronization points .
their location is schematically shown in fig .
though when the slave system is slightly detuned with respect to the synchronization points full synchronization will not occur , it is expected that the slave - map orbit follows approximately the master - map orbit with the same lag time , @xmath44 or @xmath45 . to quantitatively characterize the degree of synchronization between the two orbits and the respective lag time
, we may use the so - called similarity function @xmath46 , defined as @xcite @xmath47 ^ 2\rangle } \over { [ \langle x_n^2\rangle \langle y_n^2\rangle ] ^{1/2 } } } , \ ] ] where the brakets @xmath48 stand for time averages over asymptotically large times . if @xmath49 and @xmath50 are independent time series with similar mean value and dispersion we have @xmath51 .
if , on the other hand , there is full synchronization with lag time @xmath52 , @xmath53 .
the similarity function @xmath46 can be determined , at each point ( @xmath22,@xmath23 ) parameter space and for each lag time @xmath52 .
at @xmath54 we should have @xmath55 , while at @xmath56 we should have @xmath57 .
it is expected , moreover , that in a region around each synchronization point the similarity function attains a minimum , as a function of the lag time , for @xmath58 and @xmath59 , respectively .
these regions are qualitatively illustrated in fig .
[ f0 ] . in the remaining of the parameter space ,
as far as the slave - map orbits do not diverge , the similarity function can attain a minimum for any other value of @xmath52 without reaching , however , @xmath53 .
note that the boundaries of such regions will in general depend on the coupling constant @xmath16 . in the following sections , we study these aspects of synchronization in eqs .
( [ master ] ) and ( [ slave ] ) for logistic and hnon delay maps in their chaotic regime .
as a first illustration of the synchronization properties of eqs .
( [ master ] ) and ( [ slave ] ) in cases i and ii , we consider the choice @xmath60 and @xmath61 .
the master system becomes a delayed logistic map @xmath62 whose orbits are bounded to the interval @xmath63 for @xmath64 and @xmath65 . in different regions of the parameter space @xmath66 and depending on the delay @xmath0 ,
this system displays periodic , quasiperiodic and chaotic evolution .
the corresponding slave map is given by @xmath67 its orbits are nondivergent for @xmath68 and @xmath69 .
figure 2 displays the synchronization regions in the parameter space ( @xmath22 , @xmath23 ) and how they vary as the coupling coefficient @xmath16 increases .
the master map parameters are @xmath70 , @xmath71 and the delay times are @xmath72 , @xmath73 . for each pair ( @xmath22 , @xmath23 ) we have determined @xmath46 as a function of the lag - time @xmath52 , and detected the value of @xmath52 for which the similarity function attains its minimum @xmath74 .
the left column of fig .
2 displays this minimum for three values of @xmath16 .
light tones represent low values of @xmath75 , i.e. high master - slave correlation , while darker tones correspond to poor correlation [ @xmath76 . in the black upper - right region
the slave - map orbits diverge . the right column of fig .
2 displays the lag - time @xmath52 at which the similarity function attains its minimum .
the region where @xmath46 is minimal with lag - time @xmath77 is represented by the darker gray tone , while the region with lag - time @xmath78 is represented by the lighter gray tone .
white represents the parameter region where the minimum value of the similarity function occurs for a lag - time which is different from @xmath44 or @xmath79 .
black represents the parameter region where the trajectory of the slave map diverges . for low coupling intensity , @xmath80 ,
both synchronization regimes are unstable [ figs .
2 ( a , b ) ] .
the synchronization regions are not well defined and have fuzzy boundaries .
while the minimum of the similarity function at point i occurs at the expected lag - time @xmath44 , the minimum of the similarity function at point ii occurs at a different lag - time .
notice , in fact , that in fig .
2 ( b ) point ii is in the white region that represents a lag - time different from @xmath44 or @xmath45 .
the region corresponding to each regime is disconnected and quite complex in shape , with parts in distant zones of the parameter space .
note , for instance , the light - gray zones near @xmath81 where master - slave correlation is however rather poor . as the coupling intensity grows , zones i and ii become more uniform and increase in total extension . for @xmath82
only regime i is stable [ figs .
2 ( c , d ) ] . in this case , @xmath75 at points i and ii is equal to 0 and 0.2 respectively . for large enough @xmath16 , both regimes are stable [ figs .
2 ( e , f ) ; @xmath83 and , as expected , @xmath75 equals zero at points i and ii .
figure 3 illustrates the master - slave correlation at different points of parameter space for @xmath84 ( all other parameters are as in fig .
2 ) . in this case ,
type i synchronization is stable [ fig . 3 ( a ) ] but type ii is not [ fig .
3 ( b ) ] , and it is worth mentioning that the minimum value of the similarity function , @xmath85 , does not occur at point ii ( @xmath86 , @xmath87 ) but at a point close to it ( @xmath88 , @xmath87 ) . figure 3 ( c ) displays the correlation plot at a point where the lag - time at which @xmath46 attains its minimum is @xmath89 . as a second example , we study now a master - slave configuration where each element is a two - dimensional delay map , namely , a hnon - like map .
the evolution of the master coordinates @xmath90 is given by the functions @xmath91 and @xmath92 , so that the master system is @xmath93 cf . eq .
( [ master ] ) . in the following
we choose @xmath94 .
the slave system , with coordinates @xmath95 , is governed by the equations @xmath96 so that coupling acts on the first coordinate only .
the synchronization manifold is given by @xmath97 where , as before , @xmath29 in case i and @xmath30 in case ii .
next we study in which regions of parameters the different synchronization regimes occur .
we take parameters for the master map @xmath10 , @xmath11 , @xmath0 such that its dynamics is chaotic .
figure 4 displays the minimum of the similarity function and the lag - time for which the minimum occurs , in the parameter space ( @xmath98 ) .
the results are similar to those found with the logistic map . for weak coupling the synchronization regions
are not well defined , but as the coupling increases their size grows and the boundary between them becomes well defined . for large @xmath16 both synchronization regimes
are stable .
thus , it can be though that a small variation of the slave map parameters @xmath22 or @xmath23 near the boundary region might induce a transition from synchronization with lag - time @xmath44 to synchronization with lag - time @xmath45 or viceversa .
however , near the boundary region we find @xmath99 , which indicates bad synchronization .
therefore , while the lag - time at which the minimum value of @xmath46 occurs changes abruptly ( from @xmath44 to @xmath45 ) , there is no sharp transition between one regime of synchronization to the other .
if the slave map parameters are gradually modified from point i to point ii , synchronization with lag - time @xmath100 is gradually lost , and as we enter the region ii , synchronization with lag - time @xmath45 is gradually established .
we have studied two regimes of synchronization of delayed nonidentical maps .
we have shown analytically that , by suitably tuning the slave map parameters , two distinct synchronization regimes can occur . in one regime the lag time between the slave and the master maps is given by the delay of the coupling , @xmath1 , while in the other regime is given by the difference between the delays , @xmath2 .
we have also shown that these two regimes are actually two particular cases of synchronization with a slave map that is identical to the master map but that has two delayed feedback terms .
the two synchronization regimes has been exemplified by considering delayed logistic and hnon maps . in both cases ,
the synchronization regimes are simultaneously stable only for large values of the coupling @xmath16 , and therefore , they occur at parameters of the slave map , ( @xmath22,@xmath23 ) , which are far away from each other . in other words ,
our results show that in the case of delayed logistic and hnon maps , a small variation of a parameter of the slave map can not induce a transition from regime i to regime ii or viceversa , since they occur in distant regions of the parameter space . on the contrary , in the case of semiconductor lasers with optical feedback
, it has been shown numerically @xcite that close to the lasing threshold , by carefully tuning a parameter of the slave laser one can induce a transition from one regime of synchronization to the other .
it will be interesting to study a delayed map which shows this type of transition , and allows an analytical investigation of the phenomenon .
this work was supported by proyecto de desarrollo de ciencias bsicas ( pedeciba ) and by comisin sectorial de investigacin cientfica ( csic ) , uruguay . | we study the synchronization of time - delayed nonidentical maps subject to unidirectional ( master - slave ) coupling .
the individual dynamics of the maps have a delay @xmath0 , and the coupling acts with a delay @xmath1 .
we show analytically that , suitably tuning the slave map parameters , two distinct synchronization regimes can occur . in one regime the lag time between the slave and the master maps
is given by the delay of the coupling , @xmath1 , while in the other regime is given by the difference between the delays , @xmath2 .
we analyze the effect of the coupling strength on the different synchronization regimes in logistic and hnon maps . , chaos synchronization , time - delayed systems 05.45.xt , 05.65.+b |
the spin glass theory of infinite - ranged models @xcite has inspired a generation of physicists to study many theoretically challenging and practically important problems in physics and information processing @xcite .
these problems share a common feature , in that the disordered interactions among their elements cause frustration and non - ergodic behaviour .
the replica method @xcite has been useful in explaining their macroscopic behaviour . at the same time , based on the microscopic descriptions of the models , the cavity method @xcite resulted in many computationally efficient schemes .
these approaches have laid the foundation for the study of many problems in complex optimisation using statistical mechanics , such as graph partitioning @xcite , travelling salesman @xcite , @xmath0-satisfiability @xcite , and graph colouring @xcite .
not only the graph colouring problem @xcite is among the most basic np - complete problems @xcite , but it also has direct relevance to a variety of applications in scheduling , distributed storage , content distribution and distributed computing . in the original problem , one
is given a graph and a number of colours , and the task is to find a colouring solution such that any two connected vertices are assigned different colours .
this is equivalent to the potts glass with nearest neighbouring interactions in statistical physics .
the problem has been studied by physicists using the cavity method @xcite . for a given number of colours
, a phase transition takes place when the connectivity increases , changing from a colourable to an uncolourable phase .
one of the statistical physics approaches was based on the replica symmetric ( rs ) ansatz .
it gave an over - estimate of the threshold connectivity of this phase transition @xcite .
the one - step replica symmetry - breaking ( 1rsb ) approach takes into account the possibility that the solution space can be fragmented @xcite . besides giving an estimate of the threshold connectivity within the mathematical bounds
, it correctly predicts the existence of a clustering phase below the threshold , in which the solution space spontaneously divides into an exponential number of clusters .
this is called the hard colourable phase , in which local search algorithms are rendered ineffective , and is a feature shared by other constraint satisfaction problems @xcite .
the sequence of phase transitions in the graph colouring problem , and their algorithmic implications , were further refined recently @xcite .
these advances in the spin glass theory stimulated the development of efficient algorithms .
the cavity method gave rise to equations identical to those of belief propagation ( bp ) algorithm for graphical models @xcite .
inspired by the 1rsb solution , survey propagation ( sp ) algorithms were subsequently developed to cope with situations with fragmented solution space @xcite , and they work well even in the hard phase of the graph colouring problem @xcite .
in this paper , we study a variant of the graph colouring problem , namely , the colour diversity problem . in this problem
, the aim is to maximise the number of colours within one link distance of any node .
this is equivalent to the potts glass with second nearest neighbouring interactions in statistical physics , and hence is more complex than the original graph colouring problem in terms of the increased number of frustrated links .
indeed , this variant of the colouring problem has been shown to be np - complete @xcite .
this optimisation problem is directly related to various application areas and in particular to the problem of distributed data storage where files are divided to a number of segments , which are then distributed over a graph representing the network .
nodes requesting a particular file collect the required number of file segments from neighbouring nodes to retrieve the original information .
distributed storage is used in many real world applications such as oceanstore @xcite .
compared with the original graph colouring problem , work done on the colour diversity problem mainly focused on algorithms @xcite .
belief propagation ( bp ) and walksat algorithms for solving the problem have been presented in @xcite .
both algorithms revealed a transition from incomplete to complete colouring , and the possibility of a region of hard colouring immediately below the transition point .
approximate connectivity regimes for the solvable case have been found , given the number of colours @xcite .
however , since the algorithms are based on simplifying approximations ( bp ) and heuristics ( walksat ) , both algorithms provide only upper bounds to the true critical values .
the current study aims at providing a more principled approach to study the problem , a theoretical estimate of the transition point , and more insights on the nature of the transition itself .
the method employed is based on a tree approximation , which is equivalent to the rs ansatz of the replica method or the cavity method .
it results in a set of recursive equations which can be solved analytically .
the connectivity values for which the tree approximation is valid and the types of phases present at each value are also investigated at both zero and finite temperatures . in section [ sec : model ] we introduce the model , followed by section [ sec : macro ] that explains briefly the derivation and how the macroscopic behaviour can be studied . in section [ sec : results ] we present the results obtained via population dynamics .
discussions on the behaviour at finite temperatures are presented in section [ sec : finitetemp ] followed by a concluding section .
the appendices contain further mathematical details .
consider a sparsely connected graph with connectivity @xmath1 and colour @xmath2 for node @xmath3 .
the connectivities @xmath1 are drawn from a distribution @xmath4 with mean @xmath5 . in this paper
we consider the case of _ linear connectivity _ , that is , the nodes have connectivities @xmath6 or @xmath7 , with probabilities @xmath8 and @xmath9 respectively .
the colour @xmath2 can take the values @xmath10 .
the colour diversity problem is trivial for the case @xmath11 , in which colour schemes with complete sets of colours available to all nodes can be found easily .
hence we will focus on the more interesting case @xmath12 , in which a transition between complete and incomplete colouring exists , as shown in previous work @xcite .
the set of colours available at the node and its local neighbourhood is @xmath13 where @xmath14 is the set of nearest neighbours of node @xmath3 . to find a colour scheme that maximises the number of different colours in @xmath15 and averaged over all nodes @xmath3
, we consider minimising the energy ( cost function ) of the form @xmath16 since the objective is equivalent to minimising the number of identical colours in the set , an appropriate form of the function @xmath17 is @xmath18 where @xmath19 for @xmath20 , and 0 otherwise .
@xmath17 can be rewritten as @xmath21 } ^2.\ ] ] the quadratic nature of @xmath17 confirms that it is an appropriate cost function for diversifying the colours in the neighbourhood of each node . due to the convexity of its quadratic form , its minimum solution tends to equalise the numbers of all colours in the neighbourhood of a node .
thus , besides maximising colour diversity , our choice of the cost function has an additional advantage for the distributed storage optimisation task , which has motivated the current study , where an even distribution of segments ( colours ) in a neighbourhood is also a secondary objective , offering greater resilience .
the need for an even distribution of colours is especially important when the total number of colours is less than the connectivity of a node .
consider the contribution from the function @xmath17 centred on a node in such a case .
some colours can appear more than once .
then the exact form of the function @xmath17 determines the selection of these extra colours . in general ,
two types of selection can be made . in the first type
, one may still use all colours , but they may be less evenly distributed than in the ground state . in the second type , one may use fewer colours .
the former maximises the number of available colours , but the latter does not . in this case
, an inappropriate choice of the cost function will mix these two cases assigning the same energies , rendering it impossible to distinguish optimal and suboptimal colour choices . on the other hand , eq .
( [ eq : phi ] ) does not suffer from this shortcoming in the topology considered here .
a geometric interpretation is able to illustrate this point .
let @xmath22 be the number of times colour @xmath23 appears in @xmath15 .
then the minimisation of eq .
( [ eq : phi ] ) reduces to the minimisation of @xmath24 subject to the constraint that @xmath25 .
note that the constraint defines a hyperplane in the @xmath26-dimensional space of @xmath22 , and the problem is equivalent to finding the point with integer coordinates on the hyperplane such that its distance from the origin is minimised .
the optimal solution is the point on the hyperplane closest to the normal , and no components should be zero when @xmath27 .
in fact , the optimal solution is @xmath28 $ ] for @xmath29 , and @xmath28 + 1 $ ] otherwise ( or its permutations ) .
we have also considered a worst case analysis of the change in the total cost due to colour changes in neighbouring nodes when the function @xmath17 , centred on a node @xmath3 , is minimised .
it shows that for networks with linear connectivities and @xmath12 , the ground states consist of all satisfied nodes only , if they exist .
we note that second nearest neighbour interactions are present in this cost function .
this is different from that of the original graph colouring problem , where the cost function involves only nearest neighbour interactions .
as we shall see , the messages in the resultant message - passing algorithm will be characterised by two components , instead of the single components in the case of the original graph colouring problem @xcite .
analysis of the problem is done by writing the free energy of the system at a temperature @xmath30 , given by @xmath31 where @xmath32 is the partition function given by @xmath33,\ ] ] @xmath34 being the inverse temperature . in the zero temperature limit , the free energy approaches the minimum cost function .
several methods exist for deriving the free energy based on the replica and tree - based approximations . here
, the analysis adopts a tree - based approximation , which is valid for sparse graphs .
when the connectivity of the graph is low , the probability of finding a loop of finite length on the graph is low , and the tree approximation well describes the local environment of a node . in the approximation , node @xmath3 is connected to @xmath1 branches in a tree structure , and the correlations among the branches of the tree are neglected . in each branch , nodes are arranged in generations .
node @xmath3 is connected to an ancestor node of the previous generation , and another @xmath35 descendent nodes of the next generation .
consider the free energy @xmath36 of the tree terminated at node @xmath37 with colour @xmath38 , given its ancestor node @xmath3 of colour @xmath39 . in the tree approximation ,
one notes that this free energy can be written as @xmath40 , where @xmath41 is the number of nodes in the tree terminated at node @xmath37 , and @xmath42 is referred to as the _ vertex free energy _ @xcite .
that is , the vertex free energy represents the contribution of the free energy extra to the average free energy due to the presence of the vertex . in the language of the cavity method ,
@xmath42 are equivalent to the _ cavity fields _ , since they describe the state of the system when node @xmath3 is absent .
the recursion relation of the vertex free energy of a node can be obtained by considering the contributions due to its descendent trees and the energy centred at itself .
using notations described in fig .
[ fig : vertex ] , the vertex free energy obeys the recursion relation @xmath43-f_{\rm av } \ . \nonumber\end{aligned}\ ] ] in the above expression , the subtraction of @xmath44 is due to the incorporation of node @xmath37 with the descendent trees to form the tree terminated at node @xmath37 . for brevity , we will use the alternative simplified notation @xmath45-f_{\rm av } \ , \ ] ] where the vector @xmath46 refers to the colours of all descendants in fig .
[ fig : vertex ] . to find the average free energy @xmath44 ,
one considers the contribution to a node @xmath37 due to all its @xmath47 neighbours , that is , @xmath48\right\rangle_{\rm node } \ , \label{eq : fav}\ ] ] where the average @xmath49 denotes sampling of nodes with connectivity @xmath50 being drawn with probability @xmath51 .
however , since the probability of finding a descendant node connecting to it is proportional to the number of links the descendant has , descendants are drawn with the _ excess probability _ @xmath52 .
equations ( [ eq : vertexfreerecursion_al ] ) and ( [ eq : fav ] ) can also be derived using the replica method as presented in appendix a. we remark that both the derivation and the results are very similar to those in the problem of resource allocation on sparse networks @xcite , where the dynamical variables are the real - valued currents on the links of the networks .
the parallelism between resource allocation and colour diversity is apparent when one notes that the currents in resource allocation can be expressed as the differences between current potentials defined on the nodes of the networks .
hence the vertex free energies in both problems can be considered as functions of two variables .
another useful relation can be obtained by substituting eq .
( [ eq : vertexfreerecursion_al ] ) into eq .
( [ eq : fav ] ) , @xmath53\right\rangle_{\rm link}=0 \ , \label{eq : link}\ ] ] where the average @xmath54 denotes sampling of link vertices with connectivity @xmath50 with the excess probability .
this relation can be interpreted by considering the free energy of forming a link between vertices @xmath3 and @xmath37 . since
no extra nodes are added in this process , the extra free energy should average to zero .
the average of a function @xmath55 is given by @xmath56\ca(\cl_i ) } { \mbox{tr}_{\{\cl_i\ } } \exp\left[-\beta\sum\limits_{j\in n_i } { f^{v}_{ij}(q_i , q_j ) } -\beta\phi\left(\cl_i\right)\right ] } \right\rangle_{\rm node}.\ ] ] hence the average energy is given by @xmath57 the edwards - anderson order parameter @xmath58 @xcite , whose nonzero value characterises the potts glass phase , is given by @xmath59 the performance measure of interest is the _ incomplete fraction _
@xmath60 , which is defined as the average fraction of nodes with an incomplete set of colours available at the node and its nearest neighbours , @xmath61 \right\rangle_{\rm node } , \end{aligned}\ ] ] where @xmath62 for @xmath63 , and 0 otherwise .
this performance measure is similar to the one used in @xcite , which we refer to as the _ unsatisfied fraction _
@xmath64 , and is defined as the average fraction of colours unavailable at the node and its nearest neighbours ( for the case that @xmath26 is not greater than the number of nearest neighbours plus 1 ) , @xmath65 \right\rangle_{\rm node}.\ ] ] one might consider using eq .
( [ eq : incomplete ] ) or ( [ eq : unsat ] ) to define the cost function to be minimised , instead of eq .
( [ eq : phi ] ) .
this is indeed possible and we expect that zero - energy ground states can be obtained when the condition of full colour diversity for each node is satisfiable . in the unsatisfiable case , no zero - energy ground states can be found , but one might still be interested in finding states that minimise the average number of colours unavailable to a node . in this case , @xmath60 might not be an appropriate choice , since it mixes up the energies of selecting more ( but unevenly distributed ) colours , and fewer colours .
the second measure @xmath64 favours those states with higher colour diversity , but for the same number of available colours , it does not distinguish states with different homogeneity of colour distribution . by comparison , the cost function in eq .
( [ eq : phi ] ) has the additional advantage of favouring homogeneous colour distributions in the neighbourhood of the nodes .
solutions to the recursive equation ( [ eq : vertexfreerecursion ] ) are obtained by population dynamics @xcite .
we start with samples of @xmath66 nodes , each with one of @xmath26 colours randomly assigned as the initial condition . at each time step of the population dynamics ,
all the @xmath66 nodes are updated once in random order . at the instant we update node @xmath37
, we select @xmath67 nodes to be its descendants , where @xmath47 is drawn from the distribution @xmath68 .
descendants with connectivities @xmath69 are randomly selected with excess probabilities @xmath70 .
the vertex free energy is then updated for all pairs @xmath71 _ before _ another node is updated .
we have also computed the solutions using _ layered _ dynamics . at each time step of the layered dynamics ,
the new vertex free energies of all the @xmath66 nodes are calculated , but are temporarily reserved until the end of the time step . hence at the instant we renew node @xmath37 , we select @xmath67 nodes to be its descendants , whose vertex free energies were computed in the _ previous _ time step . descendants with connectivities @xmath69 are randomly selected with excess probabilities @xmath70 . after the new vertex free energies of all the @xmath66 nodes have been computed , they are then updated synchronously and ready for the computation in the next time step .
we observe that a _ modulation instability _ is present in layered dynamics @xcite .
this means that after sufficient layers of computation , the colour distribution no longer remains uniform .
rather , each layer is dominated by a particular colour , and the dominant colour alternates from layer to layer .
this modulation is expected to be suppressed in random graphs due to the presence of loops of incommensurate lengths . furthermore .
the average free energy computed by the layered dynamics has variances increasing rapidly with layers .
hence the layered dynamics is not adopted in our studies . to avoid growing fluctuations of the vertex free energies in the population dynamics ,
their constant components are subtracted off immediately after each update , = 3pc @xmath72 where @xmath73 is a constant bias independent of colours @xmath39 and @xmath38 .
the recursion relation of the vertex free energy then becomes @xmath74+\mbox{constant}\ ] ] after every time step , we measure the average free energy .
this is done by repeatedly creating a test node @xmath37 and randomly selecting @xmath47 nodes to connect with the test node .
the average free energy is then given by @xmath75\right\rangle _
{ \rm node } \!\!\!\!\!\!+\langle c\rangle \langle g\rangle_{\rm link } \ .\ ] ] note that @xmath76 is averaged over links , since the descendants are drawn with excess probabilities . to calculate @xmath77
we employ the consistency condition ( [ eq : link ] ) for the average free energy of a link , which requires @xmath78 } \right\rangle _ { \rm link } \!\ !
+ 2\left\langle g \right\rangle _ { \rm link } = 0.\ ] ] the node and link samplings are identical for graphs with uniform connectivity .
this allows us to eliminate @xmath79 in eqs .
( [ eq : fav1 ] ) and ( [ eq : link_normalization ] ) , and thus obtain @xmath44 . to tackle the case of non - uniform connectivities , we need to generalise the consistency condition ( [ eq : link_normalization ] ) .
this can be done by restricting our consideration to links with vertices of given connectivities @xmath80 and @xmath81 , and consider the free energy due to the link connecting the trees on both sides of such links = 3pc @xmath82}\right\rangle _ { c_i = a , c_j = b } = 0 \ .\ ] ] the derivation is analogous to that of eq .
( [ eq : link_normalization ] ) , resulting in @xmath83 } \right\rangle _ { c_i = a , c_j = b } + \left\langle g \right\rangle _ a + \left\langle g \right\rangle _ b = 0,\ ] ] which facilitates the elimination of the biases @xmath76 in eq .
( [ eq : fav1 ] ) , resulting in an expression for the average free energy = 1 pc @xmath84\right\rangle _ { \rm node } \\ & & + \frac{\langle c\rangle}{2}\sum_{a , b } \frac{ap(a)}{\langle c\rangle}\frac{bp(b)}{\langle c\rangle } \left\langle t\ln \mbox{tr}_{a , b } \exp \left [ -\beta f_{ij}^v ( a , b)-\beta f_{ji}^v ( b , a ) \right ] \right\rangle _ { c_i = a , c_j = b}.\nonumber\end{aligned}\ ] ] = 6pc to evaluate @xmath44 one first performs the node average in the first term of eq .
( [ eq : fav ] ) , keeping a record of the number of times each node @xmath85 is sampled .
then one performs the average in the second term , randomly drawing the vertices @xmath3 and @xmath37 of the links from nodes @xmath85 with exactly the same number of times they appear in the first term .
hence in this procedure , the descendants in both terms are drawn from the excess distribution .
furthermore , it ensures that the @xmath86 s appearing in the first term are exactly cancelled by those appearing in the second term , thus eliminating a source of possible fluctuations .
we also note that there can be a variety of choices of @xmath86 s to be subtracted from the vertex free energies in eq .
( [ eq : gdef ] ) .
for example , one may choose @xmath86 to be @xmath87 and arrive at the same result eq .
( [ eq : fav ] ) .
in fact , this computationally simple choice is adopted in our computation .
expressions for the energy and entropy follow immediately using the identity @xmath88 and the averaging of eq .
( [ eq : avfree ] ) , = 1pc @xmath89 \left[\sum_{j\in n_i } e_{ij}^v ( q_i , q_j ) + \phi(\cl_i)\right ] } { { \rm tr}_{\{\cl_i\ } } \exp \left[-\beta \sum_{j\in n_i } f_{ij}^v ( q_i , q_j ) -\beta\phi(\cl_i)\right ] } \right\rangle_{\rm node } , \nonumber\\\end{aligned}\ ] ] where @xmath90 is the vertex energy with the recursion relation @xmath91 \left[\sum_{k=1}^{c_j-1 } e_{jk}^v ( b , q_k ) + \phi(a , b,{\mathbf q})\right ] } { \mbox{tr}_{\mathbf q } \exp \left[-\beta \sum_{k=1}^{c_j-1 } f_{jk}^v(b , q_k ) -\beta\phi(a , b,{\mathbf q})\right ] } \nonumber\\ & & -e_{\rm av } \ , \end{aligned}\ ] ] and = 6pc @xmath92 compared with the previous equation ( [ eq : locale ] ) for the average energy , eq .
( [ eq : eav ] ) includes the vertex energies of the descendants .
these vertex energies transmit the energy deviations from the average energy , from the descendants to the ancestors .
hence eq .
( [ eq : eav ] ) can be regarded as a global estimate of the average energy , and eq .
( [ eq : locale ] ) is a local estimate .
theoretically , one expects that both estimates should yield the same result .
numerically , however , we found that this is only valid in the paramagnetic phase . in the potts glass phase
, the discrepancy between the two estimates can be very significant .
this shows that in the paramagnetic phase , memories about the initial conditions are lost easily .
in contrast , in the potts glass phase , memories about the initial conditions can propagate for a long time through the vertex energies . to avoid propagating fluctuations in the computation of the average energy
, we subtract @xmath93 from all components @xmath94 immediately after each update , and find @xmath95 using = 1 pc @xmath96 \left[\sum_{j\in n_i } e_{ij}^v ( q_i , q_j ) + \phi(\cl_i)\right ] } { \mbox{tr}_{\{\cl_i\ } } \exp \left[-\beta \sum_{j\in n_i } f_{ij}^v ( q_i , q_j ) -\beta\phi(\cl_i)\right ] } \right\rangle_{\rm node } \nonumber\\ & & -\frac{\langle c\rangle}{2}\sum_{a , b } \frac{ap(a)}{\langle c\rangle}\frac{bp(b)}{\langle c\rangle } \nonumber\\ & & \times\left\langle\frac { \mbox{tr}_{a , b } \exp \left[-\beta f_{ij}^v(a , b)-\beta f_{ji}^v(b , a)\right ] \left [ e_{ij}^v(a , b)+e_{ji}^v(b , a)\right ] } { \mbox{tr}_{a , b } \exp \left[-\beta f_{ij}^v(a , b)-\beta f_{ji}^v(b , a)\right ] } \right\rangle_{c_i = a , c_j = b}. \label{eq : globale}\end{aligned}\ ] ] the derivation at zero temperature should be carried out with extra care due to possible degeneracy in the solutions . in the zero temperature limit ,
( [ eq : vertexfreerecursion_al ] ) reduces to = 6 pc @xmath97 -f_{\rm av } \ .\ ] ] the expression of the entropy at zero temperature can be computed directly from the _ vertex entropies_. differentiating eq .
( [ eq : vertexfreerecursion_al ] ) with respect to @xmath30 , and taking the zero temperature limit , one obtains @xmath98-s_{\rm av},\ ] ] where @xmath99 is the set of colours minimising the free energy @xmath100 at node @xmath37 .
similarly , differentiating eq .
( [ eq : fav ] ) with respect to @xmath30 and taking the zero temperature limit , one obtains = 1 pc @xmath101 \right\rangle _ { \rm node } \\ & -&\frac{\langle c\rangle}{2}\sum_{a , b } \frac{ap(a)}{\langle c\rangle}\frac{bp(b)}{\langle c\rangle } \left\langle { \ln \left [ { \sum\limits_{\{a^ * , b^ * \ } } { \exp\left ( { s_{ij}^v ( a^ * , b^ * ) + s_{ji}^v ( b^ * , a^ * ) } \right ) } } \right ] } \right\rangle _ { c_i = a , c_j = b } \nonumber \ , \end{aligned}\ ] ] where @xmath102 are the set of colours minimising the free energy @xmath103 at node @xmath3 , and @xmath104 are the set of the pair of colours minimising the free energy @xmath105 at link _
the performance measures are now weighted by the entropies , and eq .
( [ eq : avfree ] ) is replaced by the expression = 6 pc @xmath106\ca(\cl_i^ * ) } { \mbox{tr}_{\{\cl_i^*\ } } \exp\left[\sum\limits_{j\in n_i } { s_{ij}(q_i^*,q_j^*)}\right ] } \right\rangle_{\rm node}.\ ] ] in the paramagnetic state , the vertex free energies are symmetric with respect to permutation of colours at each node .
hence there are only two distinct values of the vertex free energy for each node , corresponding to the cases that the colours of the node and its ancestor are the same or different .
hence , we can derive the recursion relation for the single variable @xmath107 $ ] , where @xmath108 .
this is a significant simplification of the original recursion relation for @xmath42 , which involves @xmath109 components . specifically , we consider graphs with linear connectivity @xmath110 .
we first consider the vertex free energy of a node @xmath37 with @xmath111 , whose descendants are labelled 1 and 2 .
the recursion relations are given by = 1 pc @xmath112-f_{\rm av } , \nonumber\\ f_{ij}^v(a , b ) & = & -t\ln \mbox{tr}_{q_1 , q_2 } \exp \left [ { -\beta f_{j1}^v ( a , q_1 ) -\beta f_{j2}^v ( b , q_2 ) -\beta \phi ( a , b , q_1 , q_2 ) } \right]-f_{\rm av}.\end{aligned}\ ] ] by explicitly tabulating the different colour configurations and introducing the notations @xmath113 and @xmath114 , one can rewrite eq .
( [ eq : para_rec ] ) as @xmath115 \nonumber \\ & & + \sum_k f_{jk}^v(a , b)-f_{\rm av } , \nonumber \\ f_{ij}^v ( a , b)&= & -t\ln \left [ z^{10}z_{j1 } z_{j2}+(z^8+q_2z^6)(z_{j1 } + z_{j2 } ) + z^{10}+3q_2z^6+q_2q_3z^4\right ] \nonumber\\ & & + \sum_k f_{jk}^v(a , b)-f_{\rm av}.\end{aligned}\ ] ] these give rise to the recursion relation for @xmath116 , @xmath117 similarly , for node @xmath37 with @xmath118 , @xmath119 where @xmath120 expressions of the average free energy and average energy can be found in appendix b. in the zero temperature limit for @xmath121 , eqs .
( [ eq : recursion_z ] ) and ( [ eq : recursion_z4 ] ) reduce to @xmath122 for @xmath118 , the range of values of @xmath116 is @xmath123 .
hence the distribution of the vertex partition function is given for @xmath118 by @xmath124 \delta\left(z-\frac{q_1}{6+\sum_{r=1}^k z_r}\right ) , \label{eq : pz}\ ] ] where @xmath125 and @xmath126 are the excess probabilities , and are distinctive from the connectivity probabilities @xmath127 and @xmath128 in subsequent expressions .
for the average free energy , eq .
( [ eq : para4fav ] ) becomes @xmath129 \nonumber\\ & & \times\ln\left[q q_1 q_2 q_3 \left(6+\sum_{r=1}^k z_r\right)\right ] , \nonumber\\ \left.f_{\rm link}\right|_{c_1c_2 } & = & -(1-\delta_{c_1,4}\delta_{c_2,4})t(1-f_4 ^ 2)\ln q q_1 -\delta_{c_1,4}\delta_{c_2,4 } tf_4 ^ 2\int_{2q_1/(q_1 + 12)}^{q_1/6}dz_1 p(z_1 ) \nonumber\\ & & \times\int_{2q_1/(q_1 + 12)}^{q_1/6}dz_2 p(z_2 ) \ln\left[q(q_1+z_1z_2)\right].\end{aligned}\ ] ] hence in the zero temperature limit , @xmath130 this value of the average free energy interpolates between 4 and 7 at @xmath131 and 4 respectively .
this means that in the paramagnetic phase , there is a freedom in assigning the colours of the nodes so that all local energies are minimised . for a node with 3 neighbours and @xmath132
, the state of local energy minimum has one of each colour among itself and its neighbours .
hence the energy is 4 .
similarly , for a node with 4 neighbours and @xmath132 , the state of local energy minimum has , among itself and its neighbours , two nodes of the same colour and three nodes of mutually different colours .
hence the energy is 7 .
the result of @xmath133 is the average of 4 and 7 , weighted by the fraction of nodes with 3 and 4 neighbours respectively .
this is the lowest possible energy of the system .
the average entropy of the paramagnetic state is given by @xmath134 \nonumber\\ & & -p_4\int_{2q_1/(q_1 + 12)}^{q_1/6}dz p(z)\ln z -\left(\frac{\langle c\rangle}{2}f_4-p_4\right)f_4 \int_{2q_1/(q_1 + 12)}^{q_1/6}dz_1 p(z_1 ) \nonumber\\ & & \times\int_{2q_1/(q_1 + 12)}^{q_1/6}dz_2 p(z_2 ) \ln[q(q_1+z_1z_2 ) ] .
\label{eq : sav}\end{aligned}\ ] ] consider the case @xmath132 . when @xmath131 , @xmath135
. for general values of @xmath26 , we have @xmath136 .
hence the entropy becomes negative for @xmath137 , although the entropy remains positive for @xmath138 . on the other hand ,
when @xmath139 , the vertex partition function becomes node independent , implying @xmath140 , and @xmath141=0.13 $ ] .
hence at an intermediate value of @xmath5 , the entropy changes sign . thus there is a range of negative entropy for @xmath5 below 4 where the rs ansatz is unstable .
numerical solutions to the equations are obtained using population dynamics in the manner explained in subsection [ subsec : popdyn ] .
results are obtained for @xmath132 and ensembles of graphs with linear connectivity @xmath110 , mixing nodes with connectivities 3 and 4 in varying proportions . after every time step ,
we measure the following measures : the local estimate of the average energy , the incomplete fraction , and the edwards - anderson order parameter .
this is done by creating a test node @xmath3 and randomly selecting @xmath1 nodes to connect with the test node .
the node contributions to the average free energy , the global estimate of the average energy , and ( for zero temperature ) the entropy are also computed .
the computed measures are repeated for @xmath142 nodes for each sample .
the set of descendant nodes of these @xmath66 test nodes is recorded .
then , pairs of nodes are randomly drawn this set to form links , and the link contributions to the average free energy , the global estimate of the average energy , and ( for zero temperature ) the entropy are computed .
figure [ fig : qea ] shows the edwards - anderson order parameter as a function of @xmath5 .
it can be seen that the value of @xmath58 is 0 in the paramagnetic phase , which spans the region @xmath143 . in this phase ,
all nodes have free choices of colours .
the potts glass phase spans the region @xmath144 , where @xmath58 remains at a value around 0.7 , and its transition to the paramagnetic phase is of the first order .
figure [ fig : fincom ] shows incomplete fraction obtained from the steady state solution of the population dynamics at fixed @xmath145 values .
it remains nonzero in the potts glass phase , and vanishes discontinuously above @xmath146 in the paramagnetic phase . to find the stable as well as the unstable solutions of the population dynamics , which correspond to multiple solutions at fixed @xmath5
, we may run the population dynamics at fixed nonzero @xmath60 .
this can be done by monitoring @xmath60 conditionally averaged on the nodes with @xmath147 and @xmath148 at each step , and adjusting the value of @xmath5 to approach its targeted value , which is related to the targeted value of @xmath60 estimated at each time step by @xmath149 .
the population dynamics at fixed @xmath60 yields both stable and unstable solutions of the potts glass state below @xmath146 , confirming that the transition to the paramagnetic phase is discontinuous , and that @xmath146 corresponds to the spinodal point .
the edwards - anderson order parameter for both stable and unstable potts glass states are also shown in fig .
[ fig : qea ] , bearing features similar to those in fig .
[ fig : fincom ] .
figure [ fig : fav ] shows the average free energy .
the paramagnetic free energy of 3@xmath150 provides a baseline for comparing the energy and free energy of the different phases . below the spinodal point @xmath146
, the paramagnetic state continues to exist .
it is not accessible by the population dynamics , but one can find the paramagnetic free energy by first finding a paramagnetic state at @xmath151 , and then gradually reducing the connectivity to the desired value .
the resultant paramagnetic free energy is identical to that found directly in subsection [ subsec : para ] . as shown in fig .
[ fig : fav ] , the potts glass free energy becomes lower than the paramagnetic free energy near the spinodal point @xmath146 .
a first order transition appears to take place at @xmath152 , where the free energies of the two states cross each other .
the subscript zic refers to the zero initial condition used here , as distinguished from the random initial condition ( subscript ric ) to be discussed in the next subsection .
however , since the potts glass energy equals the free energy at zero temperature , this implies that the average energy is below the lowest possible energy of @xmath133 in the range @xmath153 ! similar observations of contradictory results have been observed in the rs ansatz of the original graph colouring problem @xcite and the 3-sat problem @xcite , this indicates that the rs ansatz in the present analysis is insufficient , and has to be improved by including further steps of replica symmetry - breaking .
furthermore , the solution of the population dynamics is insensitive to this transition point in the large @xmath66 limit .
instead , it yields the potts glass state above this transition point right up to the spinodal point @xmath146 .
( for smaller values of @xmath66 , say , @xmath154 , the discontinuous transition takes place below the spinodal point . )
thus , the transition at @xmath146 looks like a zeroth order one , with a discontinuous jump of the average free energy from the potts glass phase below @xmath146 to the paramagnetic phase above @xmath146 .
as mentioned in subsection [ subsec : esfinite ] , the local and global estimates of the average energy are different and are given by eqs .
( [ eq : locale ] ) and ( [ eq : globale ] ) respectively .
the global estimate yields results identical to the average free energy , showing that memories about initial conditions in both variables have been compensated .
however , we observe that the global average energy is numerically unstable in the potts glass phase . for @xmath154
, it diverges from the average free energy after about 100 steps in the population dynamics . as shown in fig .
[ fig : fav ] , the local estimate of the average energy is indistinguishable from the global estimate in the paramagnetic phase .
however , the local estimate is significantly higher than the global estimate in the potts glass phase .
unlike the global estimate which contradicts the lowest possible energy , the local estimate remains above it .
next , we consider the entropy .
the entropy of the paramagnetic state obtained from the theoretical prediction of eq .
( [ eq : sav ] ) agrees well with the results of population dynamics . as shown in fig .
[ fig : ent ] , the entropy of the paramagnetic state becomes negative for @xmath155 , while the entropy of the potts glass state is negative throughout . at the spinodal point @xmath146
, the entropy exhibits a small discontinuous jump .
clearly , results for @xmath156 should be investigated using a replica symmetry - breaking ansatz to identify the exact transition point , which is beyond the scope of this paper .
one puzzle of our results is that the edwards - anderson order parameter remains at a level around 0.7 in the entire potts glass phase .
this implies that a considerable fraction of nodes have free choices of colours even in the potts glass phase .
this is illustrated by the distribution of colour moments @xmath157 in fig .
[ fig : ric](a ) , which consists of a continuous background with peaks at simple rational numbers ( 1/5 , 1/4 , 1/3 , 2/5 etc . ) .
in fact , the existence of free spins at zero temperature has been considered an indication of broken replica symmetry @xcite .
however , this is apparently inconsistent with extrapolations from finite temperatures , which will be discussed in the next section . as will be seen
, @xmath58 approaches 1 in the limit of low but finite temperature , implying that all nodes lose the freedom of choosing more than one colour . to resolve this inconsistency
, we consider the effects of introducing a small randomness in the initial condition , that is , a small random bias is added to the initial values of the vertex free energies , which take integer values otherwise .
such randomness were known to cause significant changes in the optimal solution in the graph bipartitioning problem , where the field distribution is initialised to a rectangular distribution @xcite .
figure [ fig : ric](b ) shows that when a very small randomness is introduced in the initial condition , the final values of the edwards - anderson order parameter @xmath58 remain around 1 in both the paramagnetic and potts glass phase .
this means that effectively all spins are frozen due to the randomness in the initial condition .
the distribution of colour moments consists of two delta function peaks , located at @xmath158 and @xmath159 respectively .
this is consistent with the extrapolation of finite temperature results .
the difference between zero temperature and low but finite temperature distributions was also observed in the rs approximation of the original graph colouring problem @xcite .
randomness in the initial condition causes a significant change in the transition point between the potts glass and paramagnetic states .
figure [ fig : ric](c ) shows that the average free energy of the potts glass state crosses that of the paramagnetic state at @xmath160 and @xmath161 for the zero and random initial conditions , respectively .
as far as we can tell from our numerical precision , @xmath161 is effectively the same as the spinodal point @xmath162 .
as will be seen in the next section , the transition point @xmath163 is consistent with the phase transition line at finite temperatures .
the effects of randomness in the initial condition on the performance are shown in fig .
[ fig : ric](d ) . for the random initial condition ,
the incomplete fraction in the potts glass phase vanishes effectively continuously to 0 at @xmath146 .
this is in contrast with the incomplete fraction for the zero initial condition , which is much higher , and vanishes discontinuously at the spinodal point .
the entropy is effectively zero in both the potts glass phase and the paramagnetic phase in the case of random initial conditions .
this is different from the case of zero initial conditions shown in fig .
[ fig : ent ] , in which the entropy is negative in the entire potts glass phase and part of the paramagnetic phase .
( 380,360 ) ( 12,170 ) ( 210,170 ) ( 0,0 ) ( 210,0 ) to illustrate the difference between the paramagnetic and potts glass phases , we consider the evolution of damages for different average connectivities @xmath145 . the damaged configuration , with colours @xmath164 , is initialised identically to @xmath165 , except that the colours of the descendants of one randomly chosen node @xmath37 have been inverted , that is , @xmath166 where @xmath85 are the descendants of node @xmath37 .
we define the _ distance measure _ between \{@xmath167 } and \{@xmath167 } as the distance between the colour moments @xmath168 we monitor the population dynamics of the colour configuration \{@xmath167 } and its _ damaged _ configuration \{@xmath167}. they evolve with the same sequence of updates and choice of descendants . as shown in fig .
[ fig : damage ] , the distance is nonzero in the potts glass phase , but vanishes in the paramagnetic phase .
this shows that multiple solutions of the saddle point equation exist in the potts glass phase , but the solution is unique in the paramagnetic phase .
the spread of damage is consistent with the instability of the replica symmetric solution in the potts glass phase .
further insights about the thermodynamic behaviour can be obtained by considering the finite temperature behaviour .
let us first study the example of @xmath169 .
figure [ fig : qea3](a ) shows that @xmath58 of the thermodynamic state vanishes at temperatures above 0.575 . to verify that this phase transition is discontinuous
, we look for solutions of the population dynamics with variable @xmath30 for given values of @xmath58 , which yield the potts glass state . as shown in fig .
[ fig : qea3](b ) , the potts glass phase with positive @xmath58 does not vanish continuously into the paramagnetic phase .
rather , its stable and unstable branches merge at the temperature 0.575 , which is therefore identified to be the spinodal temperature .
( 380,180 ) ( 0,0 ) ( 200,0 ) figure [ fig : fav3](a ) shows the free energies of the paramagnetic state and the results of the population dynamics .
the free energy at the paramagnetic state reaches a maximum at @xmath170 .
below this temperature , the entropy becomes negative .
the population dynamics is in good agreement with the paramagnetic state down to the spinodal temperature , below which the population dynamics deviates from the paramagnetic state .
figure [ fig : fav3](b ) shows the free energies in the neighbourhood of the spinodal temperature , including the stable and unstable branches of the potts glass state .
the free energies of the potts glass and paramagnetic states become equal at @xmath171 .
while this can be interpreted as the thermodynamic transition temperature , we observe that it is not relevant to the population dynamics , in which the jump of @xmath58 , as shown in figs .
[ fig : qea3](a ) and ( b ) , takes place at the spinodal temperature instead .
this behaviour is consistent with the irrelevance of the first order transition point @xmath160 at zero temperature , as described in subsection [ subsec : para - glass ] .
( 380,180 ) ( 0,0 ) ( 200,0 ) the behaviour of the entropy is shown in fig .
[ fig : ent3](a ) .
the entropy of the paramagnetic state becomes negative below @xmath170 .
the stable and unstable branches of the potts glass state are shown in fig .
[ fig : ent3](b ) , and the population dynamics yields results jumping discontinuously from the stable branch of the potts glass state to the paramagnetic state at the spinodal temperature .
( 380,180 ) ( 0,0 ) ( 200,0 ) regions of negative entropy are often found in spin glasses .
they usually signal that the rs ansatz is unstable .
however , in the original sherrington - kirkpatrick model , the region of negative entropy is restricted to the low temperature regime deep inside the spin glass phase @xcite .
in contrast , the region of negative entropy at @xmath169 spans the entire potts glass phase and even covers part of the paramagnetic phase .
this indicates that frustration effects in the present model is unusually strong .
we propose that this increased frustration effect is a consequence of the second nearest neighbouring interactions present in the colour diversity problem , and does not exist in most models investigated so far . to verify this
, we consider the model @xmath172.\ ] ] the cases @xmath173 and 1 correspond to the graph colouring and colour diversity problems respectively , we will consider the range @xmath174 . in the paramagnetic phase ,
expressions for the entropy can be derived analogously to appendix b. as shown in fig .
[ fig : ent_pm ] , the region of negative entropy of the paramagnetic state shrinks when the second nearest neighbouring interaction is reduced .
thus , in the absence of second nearest neighbouring interaction , the region of paramagnetic phase with negative entropy is preempted by the potts glass phase . for general values of @xmath5 we will consider three transition lines in the space of @xmath5 and @xmath30 : the zero entropy line in the paramagnetic phase , the spinodal line of the glassy state , and the paramagnetic - glass transition line .
the transition lines are plotted in fig .
[ fig : transition ] .
when extrapolated to @xmath175 , the zero entropy , spinodal and free - energy crossing lines pass through the points @xmath176 , @xmath177 and @xmath177 , respectively , in full agreement with the results obtained for the zero temperature case . in summary ,
the system has a paramagnetic phase at high temperature or high connectivity . inferring from the studies of the graph colouring problem @xcite
, we expect that a phase transition to replica symmetry - breaking states takes place at the high temperature ( and high connectivity ) side of the zero entropy line , even when the system is still in the paramagnetic state .
however , the location of this transition can not be found in the present framework of replica symmetry .
nevertheless , the replica symmetric solution has provided us insights on the full solution , suggesting the following picture .
one expects the existence of the spinodal line , where the potts glass state with a nonzero edwards - anderson order parameter exists in its low temperature ( and low connectivity ) side .
the potts glass state exists as a metastable state in the vicinity of the spinodal line .
then , at the low temperature ( and low connectivity ) side of the paramagnetic - glass transition line , the potts glass state becomes thermodynamically stable .
we have studied the macroscopic behaviour in the colour diversity problem , a variant of the graph colouring problem of significant practical relevance , especially in the area of distributed storage and content distribution . to cope with the presence of second nearest neighbouring interactions , the analysis makes use of vertex free energies of two arguments , which enable us to study the behaviour in the rs analysis , and lays the foundation for future analyses incorporating replica symmetry - breaking effects .
the analysis is successfully applied to graphs with mixed connectivities . for @xmath132 and graphs with linear connectivity @xmath110 ,
the rs analysis identifies three transition lines according to : ( 1 ) when the entropy becomes negative ( ending at @xmath178 when @xmath179 ) , signalling the breakdown of the rs ansatz ; ( 2 ) when @xmath58 becomes multiple - valued function of @xmath30 the spinodal point ( ending at @xmath180 when @xmath179 ) ; and ( 3 ) the free - energy crossing point between the paramagnetic and potts glass state ( ending at @xmath181 when @xmath30 approaches 0 ) . the regime of negative entropy is so extensive that it covers the entire potts glass phase as well as part of the paramagnetic phase , and can be attributed to the increased frustration due to the presence of second nearest neighbouring interactions .
the picture that emerges is that the system is in a paramagnetic state at high temperature or high connectivity ; the rs ansatz breaks down prior to the temperature that identifies the zero entropy transition point .
the potts glass state exists first as a metastable state but becomes dominant at a lower temperature ( connectivity ) .
evidence from the population dynamics shows that the discontinuous transition takes place at the spinodal point rather than the crossing point .
however , the rs analysis results in the average energy falling below the lowest possible energy for @xmath153 , and a region of negative entropy .
since the entropy remains positive at the colourable - uncolourable transition @xcite , we conjecture that if replica symmetry - breaking is taken into account , the potts glass - paramagnetic transition should take place at the higher temperature ( and high connectivity ) side of the zero entropy line . for the optimisation of the colour diversity
, one should consider @xmath179 , implying that the incomplete - complete transition should take place at @xmath5 beyond @xmath182 .
this estimate of the transition point seems to be supported by simulation results using the walksat and bp algorithms @xcite . in summary
, we have demonstrated the value of different analytical approaches and the use of population dynamics in elucidating the system behaviour of the colour diversity problem on a sparse graph .
they provide insights on the estimates of the transition points , the existence of metastable states , and the nature of phase transitions .
we thank lenka zdeborov , david sherrington , bill yeung , edmund chiang for meaningful discussions , and stephan mertens for drawing our attention to @xcite .
this work is partially supported by research grants dag04/05.sc25 , dag05/06.sc36 , hkust603606 and hkust603607 of the research grant council of hong kong , by evergrow , ip no .
1935 in the complex systems initiative of the fet directorate of the ist priority , eu fp6 and epsrc grant ep / e049516/1 .
consider the minimisation of the energy ( cost function ) on a graph of connectivity @xmath50 : @xmath183 where @xmath17 is symmetric with respect to the permutation of the neighbours , @xmath184 , and @xmath185 if nodes @xmath3 and @xmath37 are connected on the graph , and 0 otherwise .
since there are @xmath186 values of the function @xmath17 , one can write @xmath187 the partition function is @xmath188.\ ] ] the replicated partition function , averaged over all graph configurations with connectivity @xmath50 , is given by @xmath189,\end{aligned}\ ] ] where @xmath190 is the total number of graph representations with connectivity @xmath50 .
it is convenient to express the exponential argument as an unrestricted sum over the nodes @xmath191 , @xmath192 where @xmath193 are integers accounting for the over - counting in rewriting the summations in terms of equal indices .
their precise values are not required in our final result .
this allows us to factorise the expression into @xmath194\cdots \left[\sum_{j_c}a_{ij_c}(q_{j_c}^\alpha)^{m_c}\right]\right .
\nonumber\\ & & -b_2\left[\sum_{j_1}a_{ij_1}(q_{j_1}^\alpha)^{m_1+m_2}\right ] \left[\sum_{j_c}a_{ij_3}(q_{j_3}^\alpha)^{m_3}\right ] \cdots \left[\sum_{j_c}a_{ij_c}(q_{j_c}^\alpha)^{m_c}\right ] \nonumber\\ & & \left.+\cdots+(-)^{c-1}b_c \left[\sum_{j_1}a_{ij_1}(q_{j_1}^\alpha)^{m_1+\cdots+m_c}\right ] \right\}. \label{eq : factorsum}\end{aligned}\ ] ] following steps similar to those in @xcite , one gets = 1 pc @xmath195\right)\right .
\nonumber\\ & & \times\left[\sum_{r_m^\alpha , s_m^\alpha } \hat q_{{\mathbf r},{\mathbf s}}\prod_{m,\alpha } ( -i\hat h_m^\alpha)^{r_m^\alpha}(q^\alpha)^{ms_m^\alpha } + \frac{1}{2}\sum_{r_m^\alpha , s_m^\alpha}\prod_{m,\alpha } \frac{(-i\hat h_m^\alpha)^{s_m^\alpha}}{r_m^\alpha!s_m^\alpha ! } ( q^\alpha)^{mr_m^\alpha}\right]^c \nonumber\\ & & \times\exp\left\{-\frac{\beta}{c!}\sum_{{\mathbf m},\alpha } \phi_{\mathbf m}(q^\alpha)^{m_0}\left [ h_{m_1}^\alpha\cdots h_{m_c}^\alpha -b_2 h_{m_1+m_2}^\alpha h_{m_3}^\alpha\cdots h_{m_c}^\alpha + \cdots\right.\right . \nonumber\\ & & \left.+(-)^{c-1}b_c h_{m_1+\cdots+m_c}^\alpha \right]\biggr\}\biggr\ } , \label{eq : partn}\end{aligned}\ ] ] where @xmath196 and @xmath197 are given by the saddle point equations of eq .
( [ eq : partn ] ) .
consider the generating function @xmath198 in the replica symmetric ansatz , we consider functions of the form @xmath199 substituting the saddle point equation for @xmath196 into eq .
( [ eq : gen ] ) , one finds @xmath200 where @xmath201 \prod_m(q^\alpha)^{ms_m^\alpha } \nonumber\\ & & \times\exp\biggl[-\frac{\beta}{c!}\sum_{{\mathbf m},\alpha } \phi_{\mathbf m}(q^\alpha)^{m_0}\biggl ( h_{m_1}^\alpha\cdots h_{m_c}^\alpha -b_2 h_{m_1+m_2}^\alpha h_{m_3}^\alpha\cdots h_{m_c}^\alpha+\cdots \nonumber\\ & & + ( -)^{c-1}b_c h_{m_1+\cdots+m_c}^\alpha\biggr ) \biggr|_{h_m^\alpha=(z_\alpha)^m + \sum_{k=1}^{c-1}(\mu_k^\alpha)^m } \biggr]\biggr\}\biggr\rangle,\end{aligned}\ ] ] and @xmath202 is a constant having the same expression as that of @xmath203 , except that @xmath85 runs from 1 to @xmath50 and @xmath204 are set to 0 .
the expression in the exponential argument of @xmath203 can be further simplified . rewriting @xmath17 as unrestricted sums over the neighbours analogously to eq .
( [ eq : factorsum ] ) , @xmath205\cdots \left[\sum_{k=1}^c(\mu_k^\alpha)^{m_c}\right]\right . \nonumber\\ & & -b_2\left[\sum_{k=1}^c(\mu_k^\alpha)^{m_1+m_2}\right ] \left[\sum_{k=1}^c(\mu_k^\alpha)^{m_3}\right ] \cdots \left[\sum_{k=1}^c(\mu_k^\alpha)^{m_c}\right]+\cdots \nonumber\\ & & \left.+(-)^{c-1}b_c \left[\sum_{k=1}^c(\mu_k^\alpha)^{m_1+\cdots+m_c}\right ] \right\}.\end{aligned}\ ] ] identifying each term in the square bracket as @xmath206 , we recognise the exponential argument as @xmath207
. we can now identify a recursion relation for the function @xmath208 which does not involve replica indices , @xmath209 \exp[-\beta\phi(q , z,\mu_1,\cdots,\mu_{c-1})].\ ] ] the denominator is given , in the limit @xmath210 approaching 0 , @xmath211 \exp[-\beta\phi(q,\mu_1,\cdots,\mu_c)]\right\}\right\rangle.\ ] ] letting the vertex free energy be defined by @xmath212 , we arrive at the recursion relation ( [ eq : vertexfreerecursion_al ] ) and the average free energy ( [ eq : fav ] ) .
the average free energy is given by @xmath213 where @xmath214(z_{j1 } + z_{j2}+z_{j3 } ) \nonumber\\ & & \left.\left.+q_1z^6(z_{j1}z_{j2}+z_{j2}z_{j3}+z_{j1}z_{j3 } ) + z^{12}z_{j1}z_{j2}z_{j3}\right\}\right\rangle \ , \nonumber \\
\left . { f_{\rm av } } \right|_{c=4 } & = & 5-\left\langle t\ln q\left\ { { q_1q_2q_3q_4 } + 6q_1q_2q_3z^2 + 3q_1q_2z^4 + 4q_1q_2z^6+q_1z^{12}\right.\right .
\nonumber\\ & & + [ q_1q_2q_3z^2 + 3q_1q_2z^4 + q_1z^8](z_{j1 } + z_{j2 } + z_{j3 } + z_{j4 } ) \nonumber \\ & & + [ q_1q_2z^6+q_1z^8](z_{j1}z_{j2}+z_{j1}z_{j3}+z_{j1}z_{j4 } + z_{j2}z_{j3}+z_{j2}z_{j4}+z_{j3}z_{j4 } ) \nonumber \\ & & + \left.\left . q_1z^{12}(z_{j1 } z_{j2 } z_{j3 } + z_{j1 } z_{j2 } z_{j4 } + z_{j1 } z_{j3 } z_{j4 } + z_{j2 } z_{j3 } z_{j4 } ) + z^{20}z_{j1 } z_{j2 } z_{j3 } z_{j4 } \right\ } \right\rangle , \nonumber
\\ \left .
{ f_{\rm link } } \right|_{c_i c_j } & = & \left .
{ \left\langle { -t\ln q\left [ { q-1+z_{ij } z_{ji } } \right ] } \right\rangle } \right|_{c_i c_j } \ .
\label{eq : para4fav}\end{aligned}\ ] ] the average energy is given by @xmath215 the components of which take the form @xmath216 where @xmath217(z_{j1 } + z_{j2 } + z_{j3 } ) \nonumber \\ & & + q_1z^6(z_{j1 } z_{j2 } + z_{j2 } z_{j3 } + z_{j1 } z_{j3 } ) + z^{12}z_{j1 } z_{j2 } z_{j3 } , \nonumber\\ e^{(3)}_n & = & 4q_1q_2q_3 + 18q_1q_2z^2 + 10q_1z^6 + [ 6q_1q_2z^2 + 8q_1z^4](z_{j1 } + z_{j2 } + z_{j3 } ) \nonumber \\ & & + 10q_1z^6(z_{j1 } z_{j2 } + z_{j2 } z_{j3 } + z_{j1 } z_{j3 } ) + 16z^{12}z_{j1 } z_{j2 } z_{j3 } \ , \nonumber \\ e^{(4)}_d&= & q_1q_2q_3q_4 + 6q_1q_2q_3z^2 + q_1q_2z^4 + 4q_1q_2z^6+q_1z^{12 } \nonumber \\ & & + [ q_1q_2q_3z^2 + 3q_1q_2z^4 + q_1z^8](z_{j1 } + z_{j2 } + z_{j3 } + z_{j4 } ) \nonumber \\ & & + [ q_1q_2z^6 + q_1z^8](z_{j1 } z_{j2 } + z_{j1 } z_{j3 } + z_{j1 } z_{j4 } + z_{j2 } z_{j3 } + z_{j2 } z_{j4 } + z_{j3 } z_{j4 } ) \nonumber \\ & & + q_1z^{12}(z_{j1 } z_{j2 } z_{j3 } + z_{j1 } z_{j2 } z_{j4 } + z_{j1 } z_{j3 } z_{j4 } + z_{j2 } z_{j3 } z_{j4 } ) + z^{20}z_{j1 } z_{j2 } z_{j3 } z_{j4 } , \nonumber \\ e^{(4)}_n & = & 5q_1q_2q_3q_4 + 42q_1q_2q_3z^2 + 27q_1q_2z^4 + 44q_1q_2z^6 + 17q_1z^{12 } \nonumber \\ & & + [ 7q_1q_2q_3z^2 + 27q_1q_2z^4 + 13q_1z^8](z_{j1 } + z_{j2 } + z_{j3 } + z_{j4 } ) \nonumber\\ & & + [ 11q_1q_2z^6 + 13q_1z^8](z_{j1 } z_{j2 } + z_{j1 } z_{j3 } + z_{j1 } z_{j4 } + z_{j2 } z_{j3 } + z_{j2 }
z_{j4 } + z_{j3 } z_{j4 } ) \nonumber \\ & & + 17q_1z^{12}(z_{j1 } z_{j2 } z_{j3 } + z_{j1 } z_{j2 } z_{j4 } + z_{j1 } z_{j3 } z_{j4 } + z_{j2 } z_{j3 } z_{j4 } ) + 25z^{20}z_{j1 } z_{j2 } z_{j3 } z_{j4 } \ .
\nonumber\end{aligned}\ ] ]
99 j. kubiatowicz , d. bindel , y. chen , p. eaton , d. geels , r. gummadi , s. rhea , h. weatherspoon , w. weimer , c. wells and b. zhao , oceanstore : an extremely wide - area storage system , u. c. berkeley technical report no .
ucb//csd-00 - 1102 ( 1999 ) . | colouring sparse graphs under various restrictions is a theoretical problem of significant practical relevance . here
we consider the problem of maximising the number of different colours available at the nodes and their neighbourhoods , given a predetermined number of colours . in the analytical framework of a tree approximation , carried out at both zero and finite temperatures , solutions obtained by population dynamics
give rise to estimates of the threshold connectivity for the incomplete to complete transition , which are consistent with those of existing algorithms .
the nature of the transition as well as the validity of the tree approximation are investigated . |
differential equations are extensively used in modeling dynamical systems in science and engineering . when dynamical systems are under random influences , stochastic differential equations ( sdes ) may be more appropriate for modeling .
the solutions of sdes are interpreted in terms of stochastic integrals @xcite .
dynamical systems subject to gaussian white noise are often modeled by sdes with brownian motion , and the solutions are in terms of the ito integral @xcite .
although the ito integral is self - consistent mathematically , it is not the only type of stochastic integrals that can be constructed to interpret an sde .
other stochastic integrals , such as the stratonovich integral @xcite , have also been used to interpret an sde as a stochastic integral equation .
there is no right or wrong choice when choosing either ito or stratonovich integrals in interpreting sdes mathematically , since the two integrals are equivalent and can be converted into each other , provided that the integrand satisfies certain smoothness conditions @xcite .
however , these stochastic integrals have different definitions , and one may be more directly related to a practical situation than the other .
while ito integral is a reasonable choice in many applications including finance and biology @xcite , stratonovich integral is believed to be more appropriate in physical and engineering applications @xcite .
stratonovich integral has an extra term comparing with the corresponding ito integral : the so - called correction term @xcite .
some authors @xcite attribute this correction term to the conversion from physical white noise to ideal white noise .
this explanation is not necessarily convincing @xcite .
dynamical systems driven by non - gaussian white noise , especially poisson white noise , have attracted a lot of attention recently .
correction terms for converting ito sdes to stratonovich sdes with poisson white noise are presented in @xcite .
although these correction terms have been accepted widely , there are some confusions @xcite . in this paper , we consider nonlinear random vibration under excitations of either gaussian or poisson white noises , modeled by appropriate stochastic differential equations .
the main objective of this paper is to explain the correction terms in both gaussian and poisson white noise cases , from a physical perspective .
we will show that the correction terms are natural consequences of fundamental physical laws satisfied by the vibration system .
note that conventional spectral analysis methods @xcite , which have found extensive applications in random vibration analysis , are not applicable in this case due to the nonlinearity of the system . to this end
, we consider a vibration system as a mass - spring - damping oscillator with random excitation @xmath0 where @xmath1 represents the mass , @xmath2 is the stiffness coefficient of the spring , @xmath3 is the displacement depending on time @xmath4 , and @xmath5 is the velocity . @xmath6 and @xmath7 represent the generalized force terms , which may originate from external or parametric excitations .
@xmath8 is a noise term defined as the formal derivative of some stochastic process @xmath9 where @xmath10 and @xmath11 are constants , @xmath12 is a gaussian process , and @xmath13 is some compound poisson process , which is expressed as @xmath14 in eq .
( [ section1_tmpp1 ] ) , @xmath15 is a poisson process with intensity parameter @xmath16 , @xmath17 is a unit step function ( a heaviside function ) at @xmath18 , @xmath19 is a random variable representing the @xmath20-th impulse .
it follows from ( [ impulsivemodel ] ) that @xmath21 where @xmath22 is the gaussian white noise , and @xmath23 is the poisson white noise expressed as @xmath24 note that ( [ impulsivemodel_1 ] ) expresses a general noise model including the gaussian white noise ( @xmath25 ) , the poisson white noise ( @xmath26 ) , and the combined gaussian and poisson white noise ( @xmath27 and @xmath28 ) .
the second - order equation ( [ no ] ) can be rewritten as a system of sdes @xmath29 since @xmath30 is non - differentiable almost everywhere , ( [ solutionofito_1 ] ) can not be interpreted in the framework of classical calculus .
thus the solution of ( [ solutionofito_1 ] ) is interpreted with a stochastic integral , @xmath31 defining @xmath32 , @xmath33 and using the variation of parameters formula , the solution to eq .
( [ no ] ) can also be rewritten as @xcite , @xmath34 where @xmath35 is the initial condition .
it can be shown that ( [ section1_tmp1 ] ) and ( [ solutionofito ] ) are equivalent @xcite .
it is straightforward to verify that @xmath36 where @xmath37 . substituting ( [ matrixa ] ) into ( [ solutionofito ] ) , we get @xmath38 note that the stochastic integrals in eqs .
( [ section1_tmp1 ] ) and ( [ solutionstobe ] ) are yet to be defined .
as stated earlier , the stochastic integrals which can be used to interpreted sdes may not be unique .
the question is that which stochastic integral will lead to the solution that is consistent with the physics of the system .
one possible answer is to compare solutions to the sdes with the corresponding experimental results .
however , this method may be impractical in many cases due to the high cost of performing the experiments , as one needs highly accurate data from sufficiently large number of samples in order to resolve the subtle difference in the theory .
in this paper , to construct a sde model that is physically relevant to the real system , we propose to apply a stochastic integral such that the fundamental physical law ( e.g. , energy conservation ) is satisfied .
this paper is organized as follows . in sec .
[ sec.2 ] , starting from the energy conservation law , we define the stochastic integral that is suitable for the sde model of the nonlinear random oscillators .
the relationship between the proposed models and the existing models is discussed in sec .
[ sec.rel ] .
numerical methods with an illustrative example are presented in sec .
[ sec.nm ] .
since there are multiple forms of the stochastic integrals that can be constructed from the sde , we define the stochastic integral such that the fundamental physical laws are satisfied . as for the nonlinear oscillators described by the sde ( [ no ] )
, we expect the energy - work conservation be satisfied @xmath39- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k x^2(0 ) \right]\nonumber\\ & \quad = \int_0^t \left [ g(x(s ) , \dot x(s ) ) + f(x(s ) , \dot x(s))\dot l(s)\right ] \,{\rm d}x(s),\end{aligned}\ ] ] where @xmath40 represents the total mechanical energy of the system at time @xmath4 , and the integrand in the right hand side is the forcing term of ( [ no ] ) . equation ( [ energyprincipal_0 ] ) expresses that the change in the total mechanical energy is equal to the work done by the external forces .
writing ( [ energyprincipal_0 ] ) in the form of stochastic integral , we have @xmath41\nonumber- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k x^2(0 ) \right]\nonumber\\ & \quad = \int_0^t g(x(s ) , \dot x(s))\dot x(s ) \,{\rm d}s + \int_0^t f(x(s ) , \dot x(s ) ) \dot l ( s ) \dot x(s ) \,{\rm d}s\nonumber\\ & \quad = \int_0^t g(x(s ) , \dot x(s))\dot x(s ) \,{\rm d}s + \int_0^t f(x(s ) , \dot x(s ) ) \dot x(s ) \,{\rm d}l(s).\end{aligned}\ ] ] as stated before , the stochastic integral with respect to @xmath30 should be defined such that the solution of ( [ no ] ) satisfies the energy conservation law ( [ energyprincipal_1 ] ) .
it follows from ( [ impulsivemodel ] ) that a stochastic integral with respect to @xmath30 can be decomposed into two terms : stochastic integral with respect to @xmath12 and stochastic integral with respect to @xmath13 .
we define the two terms in the next two subsections .
assume @xmath42 and @xmath43 , then it follows from ( [ impulsivemodel ] ) that the stochastic process @xmath30 reduces to a brownian motion , and ( [ solutionstobe ] ) and ( [ energyprincipal_1 ] ) become @xmath44 and @xmath45- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k x^2(0 ) \right]\nonumber\\ & = \int_0^t g(x(s ) , \dot x(s ) ) \dot x(s ) \,{\rm
d}s + \int_0^t f(x(s ) , \dot x(s ) ) \dot x(s ) \,{\rm d}b(s),\end{aligned}\ ] ] respectively .
there are two types of stochastic integral extensively used for sdes driven by brownian motions : ito integral and stratonovich integral . throughout this paper
, we use @xmath46 to denote ito calculus , and @xmath47 for stratonovich calculus . in the sense of ito ,
( [ solutions_ttmmpp ] ) and ( [ energy_tmp1 ] ) can be written as @xmath48 and @xmath49- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k
x^2(0 ) \right]\nonumber\\ & = \int_0^t g(x(s ) , \dot x(s ) ) \dot x(s ) \,{\rm d}s + \int_0^t f(x(s ) , \dot x(s ) ) \dot x(s)\star \,{\rm d}b(s),\end{aligned}\ ] ] respectively . in the sense of stratonovich ,
( [ solutions_ttmmpp ] ) and ( [ energy_tmp1 ] ) can be written as @xmath50 and @xmath51- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k x^2(0 ) \right]\nonumber\\ & = \int_0^t g(x(s ) , \dot x(s ) ) \dot x(s ) \,{\rm
d}s + \int_0^t f(x(s ) , \dot x(s ) ) \dot x(s)\circ \,{\rm d}b(s),\end{aligned}\ ] ] respectively .
provided that the function @xmath52 is sufficient smooth , the solutions in stratonovich integrals , ( [ solutions_stratonovich ] ) and ( [ energy_stratonovich ] ) , can be converted into the following forms with ito integrals @xcite @xmath53 \,{\rm d}s \\ & + \bigints_0^t \dfrac{\sin \omega(t - s)}{m\omega } f(x(s ) , \dot x(s))\star { \rm d}b(s),\\ \dot x(t ) & = -\omega \sin(\omega t ) x_0 + \cos(\omega t ) \dot
x_0 + \bigints_0^t \dfrac{\cos \omega ( t - s)}{m } \left[g(x(s),\dot x(s ) ) + \dfrac{1}{2 m } f(x(s ) , \dot x(s ) ) f_{\dot x } ( x(s ) , \dot x(s))\right ] \,{\rm d}s\\ & + \bigints_0^t \dfrac{\cos \omega ( t - s)}{m } f(x(s ) , \dot x(s ) ) \star { \rm d}b(s ) , \end{cases}\end{aligned}\ ] ] and @xmath54- \left[\frac{1}{2}m\dot x^2(0 ) + \frac{1}{2}k x^2(0 ) \right]\nonumber\\ & = \int_0^t \left[\left(g(x(s),\dot x(s ) ) + \frac{1}{2m}f(x(s ) , \dot x(s))f_{\dot x } ( x(s ) , \dot x(s ) ) \right)\dot x(s)+\frac{1}{2m}f^2(x(s ) , \dot x(s ) ) \right ] \,{\rm d}s\nonumber\\ & \quad + \int_0^t f(x(s ) , \dot x(s ) ) \dot x(s)\star { \rm d}b(s).\end{aligned}\ ] ]
as shown in the appendix , the solution ( [ solutions_stratonovich_2 ] ) satisfies the energy - work relation , suggesting that when the randomness is modeled in sense of stratonovich , the energy - work conservation law is satisfied . on the other hand , in a similar procedure as in the appendix , it can be shown that the energy - work law ( [ energyprincipal_ito ] ) contradicts with the solution ( [ solutions_ito ] ) .
therefore , stratonovich integral instead of ito integral should be used so that this nonlinear random oscillator model satisfies the energy conservation law .
this implies that when gaussian noise is present in this nonlinear vibration system , the sde model should be interpreted in the sense of stratonovich stochastic integral , but not in the sense of ito stochastic integral .
when @xmath55 and @xmath56 , the stochastic process as expressed in ( [ impulsivemodel ] ) reduces to a compound poisson process .
note that the jump size of @xmath57 at time @xmath58 can be expressed as @xmath59 , where @xmath60 is the left limit of @xmath57 at @xmath58 .
suppose @xmath57 jumps at times @xmath18 ( @xmath61 ) , then the solution ( [ section1_tmp1 ] ) can be written as @xmath62 where @xmath15 , as shown in ( [ section1_tmpp1 ] ) , represents the number of jumps upto time @xmath4 . in the following
, we shall derive the stochastic integral with respect to jumps such that the energy conservation law is satisfied .
first , let s examine the changes in the system at @xmath20-th jump occured at time @xmath18 ( @xmath63 ) . from ( [ section22_tmp2 ] ) ,
the displacement @xmath64 is continuous while the velocity @xmath65 undergoes an jump given by @xmath66 the change in the total energy ( [ energyprincipal_1 ] ) due to the @xmath20-th jump is that in the kinetic energy given by @xmath67 due to the continuity of the displacement @xmath64 across an jump .
if the integrals with respect to jumps are defined in sense of ito , then ( [ section22_tt2 ] ) and ( [ section22_tmp3 ] ) becomes @xmath68 and @xmath69 respectively .
since @xmath70 , it is obvious that ( [ 1-may152012 ] ) contradict with ( [ 2-may152012 ] ) , which indicates that the energy conservation law can not be satisfied when the integrals with respect to jumps are interpreted in sense of ito . in the following , we shall show that the integrals should be interpreted as some kind of riemann integral on the imaginary path along the jump to satisfy the energy conservation law .
let @xmath71 be the value of @xmath72 at time @xmath18 if @xmath57 jumped from @xmath73 to @xmath74
. then @xmath75 and @xmath76 . with the integrals being interpreted as the riemann integral on the imaginary path along the jump
, the energy - work law ( [ section22_tmp3 ] ) can be written as @xmath77 and the solution ( [ section22_tt2 ] ) becomes @xmath78 since the jump size can be any value , it follows from ( [ section22_tt1 ] ) and ( [ 3-may152012])that for any @xmath79 , it is true that @xmath80 and @xmath81 taking derivatives of both sides of ( [ section22_tmp4 ] ) and ( [ 4-may152012 ] ) with respect to @xmath16 , respectively , we get the identical ordinary differential equation(ode ) @xmath82 therefore , the energy conservation law is satisfied .
using the fact that @xmath75 and @xmath83 , it follows from ( [ section22_tmp7 ] ) that @xmath84 where @xmath85 is determined by the initial or terminal value problem of the ode @xmath86 note that in ( [ section22_tt3 ] ) , @xmath87 for @xmath88 or @xmath89 for @xmath90 . comparing the original solution expression ( [ section22_tt2 ] ) with the new formula ( [ section22_tmp5 ] )
, it can be seen that the last term in ( [ section22_tt2 ] ) should be defined as @xmath85 , and hence ( [ section22_tmp2 ] ) should be interpreted as @xmath91 where @xmath92 is the solution to the ode ( [ section22_tt3 ] ) . as it will become clear in sec .
[ sec.rel ] , this implies that when pure jump noise , such as poisson white noise , is present in this nonlinear vibration system , the sde model should be interpreted in the sense of di paola - falsone stochastic integral @xcite . when both @xmath27 and @xmath93 , the excitation is a combined gaussian and poisson white noise . combining the results in the subsections [ sec.gwn ] and [ sec.pwn ] , we find that , in order to satisfy the energy - work conservtion law , one has to interpret the stochastic integrals with respect to brownian motions as stratonovich integrals , and the integrals with respect to jumps as dipaola - falsone integrals .
therefore , the solution to is given by the expression , where the stochastic integral is defined as @xmath94 where @xmath95=c \delta c(t_i)}$ ] .
recall that , in ( [ section23_tmp2b ] ) , @xmath47 denote integrals in the stratonovich sense , @xmath15 represents the number of jumps up to time @xmath4 , and @xmath96 is the solution to the ode ( [ section22_tt3 ] ) , where @xmath16 takes value of @xmath97 for @xmath98 or @xmath99 for @xmath100 .
in this section , we shall show that the correction term @xmath101 , as given by the solution to the ode ( [ section22_tt3 ] ) , is consistent with the one proposed in the work by di paola and falsone @xcite . when @xmath52 is lipschitz continuous , it is easy to check that the solution of ( [ section22_tt3 ] ) exists and is unique .
if @xmath52 is assumed to be smooth , then @xmath102 is analytic with respect to @xmath103 . using taylor expansion
, @xmath104 it follows from ( [ section22_tt3 ] ) that for any @xmath105 , @xmath106 substituting ( [ section3_3ttp ] ) into ( [ section3_3 ] ) , and using the fact that @xmath107 , we get @xmath108 where @xmath109 thus the correction term given by ( [ section3_7 ] ) is exactly the same as the one proposed in @xcite .
we have shown that the correction term @xmath102 can be obtained in two ways : solving the initial value problem to the ode ( [ section22_tt3 ] ) or computing the expansion ( [ section3_7 ] ) .
note that the former approach of solving ( [ section22_tt3 ] ) is applicable under much more general condition than the latter one of evaluating the infinite series ( [ section3_7 ] ) , because the existence of the solution to the ode only requires @xmath110 to be integrable but ( [ section3_7 ] ) demands @xmath110 to be infinitely differentiable .
solutions of the sde for a nonlinear oscillator ( [ no ] ) , defined by ( [ section1_tmp1 ] ) and ( [ section23_tmp2b ] ) , can hardly be obtained with analytical methods . in this section ,
the sde ( [ no ] ) is numerically solved to verify the conclusion obtained in section 2 .
consider the case with both gaussian and poisson white noises , @xmath111 , with the compound poisson process given by .
the numerical procedure of the sde for a nonlinear oscillator ( [ no ] ) defined by ( [ section1_tmp1 ] ) and ( [ section23_tmp2b ] ) is as follows . on each time
subinterval @xmath112 , i.e. when no jumps occur , ( [ no ] ) becomes @xmath113 the above equation can be converted into ito sde and then computed by conventional algorithms for ito sdes , such as euler method , milstein method , or other algorithm of high - order accuracy based on stochastic taylor expansion @xcite . at the time @xmath18
when an jump occurs , it follows from ( [ section1_tmp1 ] ) and ( [ section23_tmp2b ] ) that @xmath114 where @xmath102 is obtained by solving the deterministic ode ( [ section22_tt3 ] ) using runge - kutta or multistep methods .
consider the following stochastic duffing - van der pol equation @xmath115 with the initial condition @xmath116 and @xmath117 .
the sde ( [ example_e1 ] ) can be written in form of ( [ no ] ) with @xmath118 , @xmath119 , @xmath120 and @xmath121 . in the simulation , we take @xmath30 as in with @xmath122 , i.e. @xmath123 , where @xmath13 is a pure jump process given by with @xmath15 being a poisson process with intensity parameter as @xmath124 and @xmath19 ( @xmath125 ) being random numbers of the standard normal distribution .
* case 1 * in this case , ( [ example_e1 ] ) or ( [ no ] ) is interpreted by ( [ section1_tmp1 ] ) and ( [ section23_tmp2b ] ) . in the simulation , to integrate ( [ section1_tmp1 ] ) and ( [ section23_tmp2b ] ) numerically , we use euler s method to advance ( [ section3_tmp1 ] ) when no jumps occur , while evaluate ( [ section3_tmp2 ] ) by solving ( [ section22_tt3 ] ) with euler s method when jumps arrive . note that to apply euler s methods , the stochastic integral in ( [ section3_tmp1 ] ) need to convert into ito integral
. the step size of euler s method for solving both ( [ section3_tmp1 ] ) and ( [ section22_tt3 ] ) is @xmath126 .
figure [ fig0 ] shows a sample path of the driven process @xmath30 , and figs .
[ fig1 ] and [ fig2 ] show the numerical solution of the displacement @xmath3 and the velocity @xmath5 respectively , corresponding to the path shown in fig .
[ fig0 ] . based on the numerical solutions shown in figs . [ fig1 ] and [ fig2 ]
, we compare in fig .
[ fig3 ] the energy increment @xmath127 , which is defined by @xmath128 , \end{aligned}\ ] ] with the work done @xmath129 defined by @xmath130 @xmath131 where the stochastic integral is taken in the sense similar to , i.e. , @xmath132 where @xmath95=c \delta c(t_i)}$ ] , @xmath47 denote integrals in the stratonovich sense , @xmath15 is the number of jumps upto time @xmath4 , and @xmath133 is the solution to the following ode , @xmath134 with @xmath16 taking value of @xmath97 for @xmath98 or @xmath99 for @xmath100 .
it can be seen clearly from fig .
[ fig3 ] that the energy increment of the system agrees with the work very well , indicating that the energy conservation law is satisfied .
* case 2 * in this case , ( [ example_e1 ] ) ( or ( [ no ] ) is interpreted by using ito stochastic integrals .
now ( [ section3_tmp1 ] ) and ( [ section3_tmp2 ] ) become @xmath135 and @xmath136 respectively . in the simulation , the driving process @xmath30 and all the simulation parameters
are taken the same as in case 1 .
figures [ fig5 ] and [ fig6 ] present the corresponding numerical solutions of the displacement and velocity , respectively .
comparison of the energy increment , defined by ( [ may222012 - 1 ] ) , and the work done , now defined by @xmath137 is presented in fig .
we can see clearly from fig .
[ fig7 ] that there is significant difference between the energy increment and the work done by the force .
note that all the curves in fig .
[ fig7 ] tend to have a very small variance in the time span @xmath138 .
this is the consequence of the fact that the velocity is very small for @xmath138 , as shown in fig .
[ fig6 ] . since @xmath139
, it follows from ( [ may222012 - 1 ] ) , ( [ may222012 - 4 ] ) and ( [ may222012 - 6 ] ) that both the energy increment and the work done change slowly for very small velocity @xmath5 . by comparing fig .
[ fig7 ] with fig .
[ fig3 ] , we can see that stratonovich integral and di paola - falsone integral should be used for excitations of gaussian and poisson white noises , respectively , in order for the model to satisfy the underlining physical laws .
for cosmetic purpose , we introduce the following simplified notations : @xmath140 , @xmath141 , @xmath142 . moreover , in this appendix
, all the stochastic integrals with respect to brownian motions are in sense of ito ( we have dropped @xmath46 notation ) . then the energy - work law ( [ energy_stratonovich_2 ] ) is equivalent to @xmath143- \frac{1}{2 } \left[\ \dot x^2(0 ) + \omega^2 x^2(0 ) \right]\nonumber \\ & = \int_0^t \left[\left(g(s ) + \dfrac{1}{2 } f(s)f_{\dot x } ( s ) \right)\dot x(s)+\frac{1}{2 } f^2(s ) \right ] \,{\rm d}s + \int_0^t f(s ) \dot x(s)\ , { \rm d}b(s).\end{aligned}\ ] ] next , we show the solution given in ( [ solutions_stratonovich_2 ] ) satisfies the energy - work law ( [ 6-may152012 ] ) .
denote the right and left hand sides of ( [ 6-may152012 ] ) as @xmath144 and @xmath145 , respectively .
substitute ( [ solutions_stratonovich_2 ] ) into the left hand side of ( [ 6-may152012 ] ) , we get @xmath146 \,{\rm d}s \right)^2+\frac{1}{2 } \left ( \int_0^t \sin(\omega ( t - s ) )
f(s ) \,{\rm d}b(s ) \right)^2\nonumber \\ & \quad + \frac{1}{2 } \left ( \int_0^t \cos \omega ( t - s)\left[g(s ) + \frac{1}{2}f(s ) f_{\dot x } ( s ) \right ] \,{\rm d}s \right)^2+\frac{1}{2}\left ( \int_0^t \cos(\omega ( t - s ) ) f(s ) \,{\rm d}b(s ) \right)^2 \nonumber\\ & \quad+ ( \omega \cos(\omega t ) x_0 + \sin(\omega t ) \dot x_0)\int_0^t \sin \omega ( t - s)\left[g(s ) + \frac{1}{2}f(s ) f_{\dot x } ( s ) \right ] \,{\rm d}s\nonumber\\ & \quad + ( \omega \cos(\omega t ) x_0 + \sin(\omega t ) \dot
x_0)\int_0^t \sin(\omega ( t - s ) ) f(s)\,{\rm d}b(s)\nonumber\\ & \quad+ \left(\int_0^t \sin \omega ( t - s ) \left[g(s ) + \frac{1}{2}f(s ) f_{\dot x } ( s ) \right ] \,{\rm d}s \right ) \left ( \int_0^t \sin(\omega(t - s ) ) f(s ) \,{\rm d}b(s)\right ) \nonumber\\ & \quad+ ( -\omega \sin(\omega t ) x_0 + \cos(\omega t ) \dot
x_0)\int_0^t \cos \omega ( t - s)\left[g(s ) + \frac{1}{2}f(s ) f_{\dot x } ( s ) \right ] \,{\rm d}s\nonumber\\ & \quad + ( -\omega \sin(\omega t ) x_0 + \cos(\omega t ) \dot
x_0)\int_0^t \cos(\omega ( t - s ) ) f(s ) \,{\rm d}b(s)\nonumber\\ & \quad+ \left ( \int_0^t \cos(\omega ( t - s ) \left[g(s ) + \frac{1}{2}f(s ) f_{\dot x } ( s ) \right ] \,{\rm d}s\right ) \left(\int_0^t \cos(\omega(t - s ) ) f(s ) \,{\rm d}b(s)\right).\end{aligned}\ ] ] substituting ( [ solutions_stratonovich_2 ] ) into the right - hand side of ( [ 6-may152012 ] ) , we get @xmath147\,{\rm d}p\,{\rm d}s\notag\\ & \quad+ \int_0^t\int_0^s \left[\cos(\omega(s - p ) ) \left(g(s)+ \frac{f(s)f_{\dot x}(s)}{2}\right ) f(p)\right]\,{\rm d}b(p ) \,{\rm d}s + \frac{1}{2}\int_0^t f^2(s ) \,{\rm d}s\nonumber\\ & \quad + \int_0^t f(s ) \left(-\omega \sin(\omega
s ) x_0 + \cos(\omega s ) \dot x_0 \right)\,{\rm d}b(s)\nonumber\\ &
\quad + \int_0^t \int_0^s \cos(\omega(s - p ) ) f(s)\left(g(p)+ \frac{f(p)f_{\dot x}(p)}{2}\right ) \,{\rm d}p \,{\rm d}b(s)\notag\\ & \quad + \int_0^t \int_0^s \cos(\omega ( s - p ) ) f(s)f(p ) \,{\rm d}b(p)\,{\rm d}b(s).\end{aligned}\ ] ] to prove @xmath145 in is equal to @xmath144 in , we claim the following facts @xmath148 ^ 2 \notag\\ & \quad + \frac{1}{2 } \left [ \int_0^t \cos(\omega ( t - s ) )
\left(g(s)+ \frac{f(s)f_{\dot x}(s)}{2}\right)\,{\rm d}s \right]^2,\end{aligned}\ ] ] @xmath149 @xmath150 f(p ) \,{\rm d}b(p ) \,{\rm
d}s \nonumber\\ & \quad + \int_0^t\int_0^s \left[\cos(\omega(s - p)\left(g(p)+ \frac{f(p)f_{\dot x}(p)}{2}\right ) \right ] f(s ) \,{\rm d}p\,{\rm d}b(s)\notag\\ & = \int_0^t \cos(\omega ( t - s ) ) \left(g(s)+ \frac{f(s)f_{\dot x}(s)}{2}\right)\,{\rm d}s \int_0^t \cos(\omega(t - s ) ) f(s ) \,{\rm d}b(s ) \notag\\ & \quad + \int_0^t \sin(\omega ( t - s ) ) \left(g(s)+ \frac{f(s)f_{\dot x}(s)}{2}\right)\,{\rm d}s \int_0^t \sin(\omega(t - s))f(s ) \,{\rm d}b(s)\,\end{aligned}\ ] ] @xmath151 and @xmath152 one can easily see that ( [ a1_6 ] ) and ( [ a1_7 ] ) are true by using the trignometric identities @xmath153 and @xmath154 in the following , we give the proofs for ( [ a1_3 ] ) and ( [ a1_4 ] ) . the proof of ( [ a1_5 ] ) is similar to those for ( [ a1_3 ] ) and ( [ a1_4 ] ) and is not given here . to prove ( [ a1_3 ] ) is true , we rewrite the right - hand side of ( [ a1_3 ] ) as double integrals @xmath155 similarly , to prove ( [ a1_4 ] ) , we rewrite the right - hand side ( [ a1_4 ] ) as @xmath156 the integral domain for the right - hand side of ( [ a1_8 ] ) is a square given by @xmath157 , p\in [ 0 , t]\}$ ] . decompose the square into three parts : @xmath158 , @xmath159 , and @xmath160 , then the right - hand side of ( [ a1_8 ] ) becomes @xmath161 note that @xmath162 and @xmath163 it follows from ( [ a1_9 ] ) and ( [ a1_10 ] ) that ( [ a1_4 ] ) is true . | nonlinear random vibration under excitations of both gaussian and poisson white noises is considered .
the model is based on stochastic differential equations , and the corresponding stochastic integrals are defined in such a way that the energy conservation law is satisfied .
it is shown that stratonovich integral and di paola - falsone integral should be used for excitations of gaussian and poisson white noises , respectively , in order for the model to satisfy the underlining physical laws ( e.g. , energy conservation ) .
numerical examples are presented to illustrate the theoretical results . * keywords * : random vibration , nonlinear systems , poisson noise , gaussian noise , stochastic differential equations , stochastic integrals . |
let @xmath0 be a function that is analytic in the open unit disc @xmath1 in the complex plane @xmath2 .
we are interested in obtaining upper bounds of the euclidean length @xmath3 of the @xmath4-image of the ray @xmath5 $ ] , and in understanding the geometry that lies behind these bounds .
when we wish to emphasize the role of @xmath0 in this expression ( for example , when we are discussing several functions at once ) we shall use @xmath6 in the natural way .
the dirichlet space @xmath7 is the space of analytic functions @xmath8 with @xmath9 the quantity @xmath10 is the area of the image @xmath11 , counting multiplicity .
it is easy to see that if @xmath12 then @xmath13 is finite for almost all @xmath14 . indeed , by the cauchy - schwarz inequality ,
@xmath15 so that @xmath16 it follows that if @xmath12 , then @xmath17 is finite for almost all @xmath14 . beurling ( @xcite , and @xcite ) has proved the stronger result that if @xmath12 then @xmath17 is finite except when @xmath18 lies in some subset of @xmath19 of logarithmic capacity zero ; thus @xmath20 as @xmath21 for only a small set of @xmath14 .
the following result gives an upper bound on @xmath22 for all @xmath14 , and it is the starting point of the work in this paper . [
a ] suppose that @xmath0 is in @xmath7 .
then , for each @xmath14 , @xmath23^{1/2}\right)\ ] ] as @xmath21 .
the exponent @xmath24 is best possible even for the subclass of functions that are bounded and univalent in @xmath1 .
theorem a is due to keogh @xcite ( who attributes it to j.e .
littlewood ) and also to rosenblatt ( see @xcite ) .
both keogh and rosenblatt state the result for bounded univalent functions but , as remarked by jenkins @xcite , their proof is valid without change for functions in @xmath7 .
theorem a has been extended in @xcite where the following two local versions are proved ( see @xcite , pp.492 - 493 and theorem 1 ) .
[ b ] suppose that @xmath0 is analytic in @xmath1 , and that for some stolz region @xmath25 at @xmath18 , @xmath26 has finite area .
then ( 1.2 ) holds for this @xmath14 .
[ c ] there is a constant @xmath27 such that , if @xmath0 is analytic and univalent in @xmath1 with @xmath28 , then @xmath29 for each @xmath14 and each @xmath30 .
further , if @xmath0 is bounded on @xmath31 then ( 1.2 ) holds for this @xmath14 .
theorem b appears to be stronger than theorem a , and its proof in @xcite is substantially harder than a proof of theorem a. we shall show that theorems a and b are equivalent up to a very simple argument in hyperbolic geometry ( which involves no function theory at all ) .
indeed , our first objective is to understand theorems a , b and c from a geometric point of view , and we shall show that all three results are more transparent when placed in the context of the hyperbolic geometry of a plane domain .
the class of universal covering maps includes the class of univalent maps , and we know of no bounds in the literature on @xmath32 for universal covering maps .
the following simple example shows what can happen for covering maps , and also illustrates the use of hyperbolic geometry in this topic .
the map @xmath33 is a conformal map of the upper half - plane @xmath34 onto @xmath1 , and @xmath35 is a universal covering map of @xmath34 onto the annulus @xmath27 given by @xmath36 . now let @xmath37 . then @xmath0 is a universal covering map of @xmath1 onto @xmath27 , so that the hyperbolic length of @xmath38)$ ] is @xmath39 .
however , the hyperbolic metric in @xmath27 is @xmath40 , where @xmath41 and a calculation shows that @xmath42 when @xmath43 .
it follows that the euclidean length of @xmath38)$ ] is the same as its hyperbolic length ; thus @xmath44 as @xmath21 .
the striking feature of this example is that we have ` lost ' the exponent @xmath24 from the estimate in ( 1.2 ) .
the exponent @xmath24 in theorem a is best possible , and the work in this paper was partly motivated by the desire to understand why this exponent appears in the estimates for functions in the dirichlet class but not for universal covering maps .
the explanation of this will be given in terms of hyperbolic geometry in section 9 .
we shall also discuss why , for geometric reasons , other exponents arise in other circumstances .
we regard this as the most interesting part of the paper , and our geometric treatment of theorems a , b and c should be regarded as preparation for this work .
we come now to the idea that underpins most of the work in this paper . as the hyperbolic metric ( and not the euclidean metric )
is the natural metric on @xmath1 , when we consider an analytic map @xmath8 we should primarily be concerned with _ the change of scale from the hyperbolic metric to the euclidean metric _ , and not with the euclidean change of scale @xmath45 .
now the hyperbolic metric @xmath46 on @xmath1 is given by @xmath47 , where @xmath48 .
it follows , then , that _ the primary role in this discussion should be given to the function _
@xmath49 and that we should consider @xmath22 to be the integral of this function with respect to the hyperbolic length on @xmath50 ; that is , @xmath51 further , as the geodesic segment @xmath5 $ ] has hyperbolic length @xmath52 , where @xmath53 it seems clear that ( 1.2 ) should be written as @xmath54 the use of the apparently simpler term @xmath55 in ( 1.2 ) only serves to conceal the geometry behind these estimates .
of even greater importance is the fact that the hyperbolic metric is conformally invariant whereas the logarithmic term in ( 1.2 ) is not . to summarize these ideas , our view is that @xmath56 to develop this idea further , the class @xmath7 should now be considered to be the class of analytic maps @xmath8 for which the basic scaling function @xmath57 is square - integrable over the hyperbolic plane . of course
, one can also consider the other @xmath58-spaces of functions @xmath0 for which @xmath59 these spaces occur in the theory of automorphic functions ( they are the @xmath60 spaces , with @xmath61 , introduced by bers ) , and we shall consider them briefly in this context in section 5 . to illustrate these ideas , we pause to show that _ if @xmath62 is square - integrable over the hyperbolic plane ( that is , if @xmath12 ) , then @xmath8 is a lipschitz map with respect to the natural metrics on @xmath1 and @xmath2_. this is our next result , and we give the best lipschitz constant .
suppose that @xmath12 . then , for any @xmath63 in @xmath1 , @xmath64 further , for each @xmath63 , equality occurs when @xmath65 .
this theorem shows that @xmath66 , so @xmath0 is in the bloch class .
the proof , together with much more information about the bloch class , is given by anderson , clunie and pommerenke in @xcite .
suppose that @xmath67 , where @xmath68 .
the cauchy - schwarz inequality gives @xmath69 since @xmath70 and @xmath71 this proves ( 1.6 ) .
the statement about equality is easily verified ( note that for these functions , @xmath72 ) . this simple application of the cauchy - schwarz inequality underlies all of the results in this paper .
the argument used in the proof above can be strengthened slightly to show that @xmath73 this will be clear from the proof of theorem a in section 3 ( and it shows that any @xmath0 in @xmath7 is in the little bloch class ) .
the plan of the paper is as follows . in section 2
we briefly outline the main ideas concerning the hyperbolic metric of a simple connected domain .
section 3 contains our discussion of theorem a. this contains two examples to show that the exponent @xmath24 in theorem a is best possible , and both are based on hyperbolic geometry .
although one involves a technical geometric estimate ( which is of value in its own right ) , these examples are completely transparent .
we also give a geometric proof of a result which includes theorem a for univalent functions as a special case . in section 4
we show how theorem b follows from theorem a in an elementary way using only hyperbolic geometry .
this proof is considerably shorter than the published proof of theorem b ; it is based on a conformally invariant form of theorem a , and it applies without change to many other subdomains of @xmath1 .
section 5 is concerned with theorem c and it contains a geometric proof of this and some results about @xmath58-spaces .
the proof of theorem c in @xcite depends on an integral inequality due to marcinkiewicz and zygmund , and in section 6 we show that the published proof of this inequality is a euclidean version of a standard argument in hyperbolic geometry .
we have already noted that the exponent @xmath24 in theorem a is best possible , but kennedy and twomey @xcite have shown how we can obtain more detailed information on the rate of growth of @xmath22 .
see also balogh and bonk @xcite .
we complete our discussion of the known results in sections 7 and 8 where we discuss their result , a generalization of it to @xmath58-spaces , and some known estimates on the basic scaling function @xmath74 . in the remaining sections of the paper we raise and discuss the analogous issues for universal cover maps instead of functions in the dirichlet class . finally ,
in the appendix , we give the proofs of some of the geometric results that we have used in the earlier sections .
the authors are grateful for helpful comments by t. carroll and j.b .
twomey on an earlier draft of this paper .
each simply connected proper subdomain @xmath75 of @xmath2 supports a hyperbolic metric @xmath76 with density @xmath77 , where @xmath78 and where @xmath79 is any conformal map of @xmath1 onto @xmath75 .
it is easy to see that @xmath77 and @xmath76 are independent of the choice of @xmath79 , and that @xmath79 is an isometry from @xmath80 to @xmath81 .
as @xmath75 is simply connected , @xmath82 } \ \leqslant\ \lambda_d ( w ) \
\leqslant\ { 2\over { \mathop{\rm dist}\nolimits}[w,\partial d]},\ ] ] where @xmath83 $ ] denotes the euclidean distance from @xmath84 to the boundary @xmath85 of @xmath75 ( these are standard estimates ; see @xcite or @xcite ) .
next , @xmath77 is continuous and positive on @xmath75 . if @xmath75 is bounded , then @xmath86 as @xmath63 approaches @xmath85 , and so @xmath77 has a positive lower bound , say @xmath87 , on @xmath75 .
thus if @xmath0 is bounded and univalent in @xmath1 , and @xmath88 , then @xmath89 and @xmath90 these conclusions explain some of the earlier results in geometric terms ( albeit in the simpler case of bounded univalent maps ) .
they show , for example , that if @xmath0 is bounded and univalent on @xmath1 then the map @xmath8 is lipschitz with respect to the natural metrics on @xmath1 and @xmath2 .
theorem 1.2 gives a stronger result than this ; however , this argument goes beyond theorem 1.2 .
a domain @xmath75 is a _ bloch domain _
if @xmath75 does not contains arbitrary large euclidean discs , and it is clear from ( 2.2 ) that a simply connected domain @xmath75 is a bloch domain if and only if @xmath77 has a positive lower bound on @xmath75 . thus _
if @xmath91 is a univalent map of @xmath1 onto a bloch domain @xmath75 , then @xmath8 is lipschitz with respect to the natural metrics on @xmath1 and @xmath2_. of course , in this case @xmath0 need not be in @xmath7 .
finally , we note that ( 2.4 ) goes some way to explaining why @xmath92 may tend to zero as @xmath21 . as @xmath22 is the integral of @xmath57 over a hyperbolic segment of length @xmath52
, we certainly see that if ( 2.4 ) holds , then @xmath93 .
more generally , if the convergence in ( 2.4 ) is sufficiently rapid , we might expect to get some result of the form @xmath94 as @xmath21 .
as @xmath95 may tend to @xmath96 as @xmath21 , it is perhaps harder to see why the estimates of @xmath22 might hold if we take the euclidean point of view and integrate @xmath97 with respect to euclidean length on @xmath31 .
keogh s proof of theorem a in @xcite is short and elementary , and for completeness we include it here .
after this we give two examples to show that the exponent @xmath24 is best possible , and we end with a geometric proof of a similar result .
it suffices to prove the result for @xmath98 , and we write @xmath99 .
then , for each positive integer @xmath100 , @xmath101 as @xmath102 converges , ( 1.2 ) follows from this .
we continue our discussion of theorem a by providing the following conformally invariant version of it .
it is essential that we use the conformally invariant hyperbolic length here rather than the classical bound @xmath55 .
suppose that @xmath103 is a domain that is conformally equivalent to @xmath1 , and that @xmath0 is analytic in @xmath103 with @xmath104 of finite area @xmath10 .
then , for any hyperbolic geodesic segment @xmath105 in @xmath103 , @xmath106 where @xmath107 is the euclidean length of @xmath108 , and @xmath109 is the hyperbolic length of @xmath105 . in particular ,
if @xmath103 has finite area @xmath27 , then @xmath110 .
this is easy .
we can find a conformal map @xmath79 of @xmath103 onto @xmath1 in such a way that @xmath79 maps @xmath105 onto some real segment @xmath111 $ ] of @xmath1 , and then apply keogh s result to @xmath112 defined by @xmath113 .
if we take @xmath114 ( and @xmath115 ) in the proof given above , we obtain @xmath116 and this gives the stated inequality because @xmath117 , @xmath118 and @xmath119 the estimate @xmath120 is universally valid over all geodesic segments @xmath105 .
however , if we take @xmath105 to be the initial segment of any geodesic ray starting from a fixed point of @xmath103 then , exactly as in keogh s result , we can show that @xmath121 .
this will be considered further in section 4 .
theorem a applies to all functions in the dirichlet space @xmath7 .
however in @xcite , keogh showed that the exponent @xmath24 in theorem a is best possible when the result is restricted to the smaller class of functions that are bounded and univalent in @xmath1 .
further examples were given by jenkins @xcite , kennedy @xcite , and carroll and twomey @xcite , to show that not only is the exponent @xmath24 in theorem a best possible , but so is the function on the right hand side of ( 1.2 ) , and we shall look at these results in section 7 .
keogh s example in @xcite is based on the function @xmath122 , where @xmath79 is a conformal map of @xmath1 onto a half - infinite strip , and @xmath123 is a suitable constant , and his argument is technical , and not self - contained . in our examples ( examples 3.2 and 3.3 ) we study the analytic properties of a univalent map @xmath124 indirectly through the hyperbolic geometry of @xmath125 .
example 3.2 is extremely simple , but the function in this example is unbounded .
example 3.3 is of a bounded univalent function in @xmath126 .
suppose that @xmath127 is a constant such that for all @xmath0 in @xmath7 , and all real @xmath14 , we have @xmath128 as @xmath129 ; we shall show that @xmath130 .
take any positive @xmath131 , and let @xmath132 .
it is clear that if @xmath133 , then the open rectangle @xmath134 lies in @xmath75 , and from this we see that @xmath135 } \
\leqslant\ 2(t+1)^{1+\varepsilon}.\ ] ] now let @xmath0 be the conformal map of @xmath1 onto @xmath75 that maps @xmath50 onto @xmath136 . clearly , @xmath12 .
next , as @xmath91 is a hyperbolic isometry , @xmath137^{2+\varepsilon}.\ ] ] finally , as @xmath0 is increasing on @xmath138 it maps @xmath111 $ ] onto @xmath139 $ ] so that @xmath140 , and hence @xmath141
. as @xmath128 , it follows that @xmath142 .
as @xmath131 is arbitrary we have @xmath130 .
the reader will notice that , in this example , the conclusion follows once we know that @xmath143 as @xmath21 for a single value of @xmath14 .
this example is similar to that given in @xcite to show that there is a function @xmath0 that is univalent , bounded , close - to - convex and in @xmath126 such that the image curve @xmath144 may be of infinite length .
here we let @xmath0 be a conformal map of @xmath1 onto a particular bounded simply connected domain @xmath75 which we shall now construct .
* figure 1 . * ( the geodesic @xmath145 is shown as a dashed line . )
suppose that @xmath146 and for @xmath147 , let @xmath148 now consider the rectangle @xmath149 and form a simply connected domain @xmath75 by deleting from @xmath150 the vertical slits @xmath151 this is illustrated in figure 1 . for @xmath147 ,
let @xmath152 be the segment of the @xmath153-axis that joins the slits @xmath154 and @xmath155 ; let @xmath156 be the mid - point of @xmath152 for @xmath157 ; and @xmath158 .
now let @xmath0 be the unique univalent map of @xmath1 onto @xmath75 with @xmath159 and @xmath160 .
then @xmath161 , and there is some real @xmath14 such that the image curve @xmath162 accumulates at , and only at , the right hand edge of @xmath150 ( see @xcite and @xcite ) . using this value of @xmath14 we shall construct a sequence of values @xmath163 with @xmath164 , @xmath165,@xmath166 and @xmath167 let us assume for the moment that we have done this .
now suppose that @xmath127 is such that @xmath168 for every univalent bounded function in @xmath7 and every choice of @xmath14 ; then this holds for the function @xmath0 constructed above , and with @xmath169 .
now ( 3.1 ) and ( 3.2 ) imply that @xmath170 , and as this is true for every positive @xmath131 we again conclude that @xmath130 as required .
it remains only to construct the sequence @xmath163 satisfying ( 3.1 ) for the function @xmath0 and the value @xmath14 described above .
we put @xmath171 and , for @xmath172 , we let @xmath163 be the smallest value of @xmath173 in @xmath138 at which the point @xmath174 lies on @xmath152 . it is evident that @xmath175 , that @xmath165 , and that @xmath176 ; this last inequality is the first inequality in ( 3.1 ) .
it remains to establish the estimate for @xmath177 in ( 3.1 ) .
as @xmath0 is a conformal map of @xmath1 onto @xmath75 , it is a hyperbolic isometry , so that @xmath177 is the hyperbolic length of the curve @xmath178 , @xmath179 .
as this curve is a geodesic with end - points @xmath180 and @xmath181 , its length is at most the hyperbolic length of any curve , say @xmath182 , in @xmath75 that passes through the points @xmath183 in this order .
it suffices , then , to show that there is a choice of @xmath182 whose length is @xmath184 .
we take @xmath182 to be a polygonal curve constructed as follows . for @xmath185 ,
we join @xmath186 to @xmath187 by a curve @xmath188 in @xmath75 comprising a vertical segment of euclidean length one , followed by a horizontal segment ( from left to right ) , and then by another vertical segment of euclidean length one .
we now let @xmath182 be the curve traced out by @xmath189 , followed by the horizontal segment @xmath190 of @xmath152 that joins @xmath156 to @xmath181 . thus
if we denote the hyperbolic length of a curve @xmath191 in @xmath75 by @xmath192 ) , we have @xmath193 we now estimate the terms on the right hand side of this inequality . as each point of @xmath188
is at least a distance @xmath194 away from @xmath85 , where @xmath195 we see from ( 2.2 ) that @xmath196 . thus @xmath197 our discussion of this example will be complete if we can show that for some positive number @xmath198 , and all @xmath199 , @xmath200 in fact , @xmath201 as @xmath202 , because if a domain @xmath75 contains a long thin rectangle whose long sides lie in @xmath85 , then any hyperbolic geodesic in @xmath75 that travels the length of the rectangle must pass very close to the centre of the rectangle .
( see @xcite and @xcite . )
we do not need as much as this , and the following result ( whose proof is given in the appendix ) yields ( 3.3 ) immediately , and thereby completes our discussion of example 3.3 .
note that the hyperbolic metric is unchanged by ( euclidean ) translations and scalings of a domain .
so we may assume that the rectangle @xmath150 is centred at the origin and of a fixed width .
there is a positive number @xmath198 with the following property .
let @xmath150 be the square @xmath203 and let @xmath75 be a simply - connected , hyperbolic domain such that @xmath204 , and the two sides of @xmath150 parallel to the @xmath153-axis lie in @xmath85 .
if @xmath105 is any hyperbolic geodesic in @xmath75 that does not have an endpoint on the sides of @xmath150 parallel to the @xmath153-axis , then it can only meet the crosscut @xmath205 of @xmath75 at a hyperbolic distance of at most @xmath198 from the origin .
we complete this discussion of theorem a by returning to the conformally invariant version of it , namely theorem 3.1 .
this says that if @xmath105 is a hyperbolic geodesic segment in a simply connected domain @xmath103 of finite area @xmath27 , then @xmath206 .
the following result shows that a similar inequality also holds for euclidean geodesic segments ; note that in this result _ the domain need not be simply - connected_. .
let @xmath75 be a domain in @xmath2 with finite area , and let @xmath207 , @xmath208 , be a euclidean ray in @xmath75 .
then the euclidean and hyperbolic lengths @xmath209 and @xmath210 , respectively , of the segment @xmath211 $ ] satisfy @xmath212 as @xmath213 .
we may assume that @xmath105 is the positive real axis , and that @xmath214 for @xmath215 . in @xcite solynin
shows that if @xmath75 is a domain in @xmath2 that meets the real axis , and if @xmath216 is the domain obtained from @xmath75 by symmetrising it about @xmath217 , then the two hyperbolic densities satisfy @xmath218 for @xmath219 .
we let @xmath220 be the euclidean length of the intersection of @xmath75 with the vertical line @xmath221 ; then @xmath216 is given by @xmath222 , and if @xmath215 then @xmath223 } \ \geqslant\ { 1 \over 2(\frac{1}{2 } m(t ) ) } = { 1\over m(t)}.\ ] ] this implies that the hyperbolic length @xmath224 of @xmath111 $ ] satisfies @xmath225 next , the euclidean length @xmath226 of @xmath111 $ ] is @xmath173 so , using the cauchy - schwarz inequality , we have @xmath227 more generally , for any positive @xmath228 , and any @xmath173 with @xmath229 , we have @xmath230 as the integral @xmath231 is the euclidean area of the domain @xmath75 , and this is finite , we see that @xmath232 thus @xmath233 as required .
theorems 3.1 and 3.5 raise the question of characterising those curves @xmath105 in a given domain whose euclidean and hyperbolic lengths satisfy some inequality of the form @xmath234 , or some variant of this .
we can show that this conclusion also holds for curves that are in some sense sufficiently close to euclidean rays .
let @xmath235 be a smooth curve in @xmath2 parametrised by euclidean arc length , and suppose that @xmath105 is sufficiently close to an euclidean ray so that the map @xmath105 extends to a quasi - conformal map @xmath79 that maps a neighbourhood @xmath236 of @xmath237 of finite area onto a neighbourhood @xmath75 of @xmath105 .
now theorem 3.5 applies to the ray @xmath237 in the domain @xmath236 .
however , @xmath79 does not change euclidean length along @xmath237 and , since @xmath79 is quasi - conformal , it only alters hyperbolic length by at most a fixed factor ; thus the same conclusion holds for @xmath105 in @xmath75 .
we end this section by showing that theorem 3.5 contains theorem a in the special case of conformal maps .
suppose that @xmath0 is a conformal map of @xmath1 onto a domain @xmath75 of finite area .
we write @xmath238 , and define @xmath239 by @xmath240 .
note that @xmath79 maps @xmath50 into the positive real axis , @xmath241 and the euclidean lengths of @xmath38)$ ] and @xmath242)$ ] satisfy @xmath243 if @xmath244 is finite there is nothing to prove . if not , then @xmath125 contains the interval @xmath245 and we can apply theorem 3.5 to @xmath79 and obtain @xmath246 where @xmath247 is the hyperbolic length of @xmath248 $ ] in @xmath125 .
the schwarz - pick lemma applied to @xmath249 implies that @xmath250 so we obtain the conclusion of theorem a for @xmath0 , namely that @xmath251 as @xmath21 .
we shall now show how the apparently stronger theorem b can be deduced directly from theorem a by a geometric argument that is more transparent than the analytic proof of theorem b given in @xcite .
the idea is to apply theorem a to the restriction of @xmath0 to the stolz region @xmath25 contained in @xmath1 .
on the radial line @xmath252 , the hyperbolic metric @xmath253 for @xmath25 and the hyperbolic metric for @xmath1 are comparable , so theorem a for @xmath254 implies theorem b. the details of this proof rely on the hyperbolic geometry of stolz regions , so we write it out fully . .
let @xmath255 be any analytic function .
for any two points @xmath256 in @xmath1 , let @xmath257 $ ] be the hyperbolic geodesic segment from @xmath258 to @xmath259 and let @xmath260 ) < d \ } $ ] for a fixed constant @xmath261
. then @xmath262 ) ) \
\leqslant\ \left({{a(f(\omega))}\over{\pi d_e}}\right)^{1/2 } \rho_\mathbb{d}(z_0 , z_1)^{1/2}\ ] ] where @xmath263 .
first observe that @xmath103 is a simply connected subdomain of @xmath1 . also , @xmath103 is hyperbolically symmetric about the geodesic through @xmath258 and @xmath259 , so @xmath257 $ ] is also a geodesic segment for the hyperbolic metric @xmath264 on @xmath103 .
we wish to compare the two hyperbolic metrics @xmath265 and @xmath266 .
consider the ratio @xmath267 for @xmath268 $ ] .
we can apply a mbius isometry to move @xmath63 to the origin .
then @xmath103 contains the disc @xmath269 with hyperbolic radius @xmath270 and euclidean radius @xmath263 .
consequently , @xmath271 and so @xmath272 ( cf .
( 2.2 ) ) . integrating this along the curve @xmath257 $ ]
relative to the hyperbolic arc length @xmath273 in @xmath1 shows that @xmath274 } \lambda_\omega(z)/\lambda_\mathbb{d}(z ) \ ;
ds_\mathbb{d}(z ) \
\leqslant\ { { \rho_\mathbb{d}(z_0 , z_1)}\over{d_e}}\ .\ ] ] theorem 3.1 is an invariant form of theorem a , so we can apply it to the restriction .
this gives @xmath262 ) ) \
\leqslant\ \left({{a(f(\omega))}\over\pi}\right)^{1/2 } \rho_\omega(z_0 , z_1)^{1/2}\ .\ ] ] therefore we see that @xmath275 ) ) \
\leqslant\ \left({{a(f(\omega))}\over{\pi d_e}}\right)^{1/2 } \rho_\mathbb{d}(z_0 , z_1)^{1/2}\ .
\esinglebox\ ] ] this theorem is enough to show that @xmath276 but we want the slightly stronger result @xmath277
. we could deduce this by applying theorem a rather than theorem 3.1 .
however , we give an alternative argument below .
fix @xmath258 at the origin and let @xmath259 be a point @xmath278 .
we will write @xmath279 for @xmath280 .
let @xmath105 be the half - line @xmath281 and let @xmath282 be the region @xmath283 .
( this is actually a stolz region with vertex at @xmath284 . ) more generally , let @xmath285 .
if the area @xmath286 is finite , then @xmath287 for any fixed @xmath288 we can apply theorem 4.1 to show that @xmath289 ) ) & ~\leqslant~ & { \cal e}(f([0 , r_0 ] ) ) + { \cal e}(f([r_0 , r ] ) ) \\ & ~\leqslant~ & { \cal e}(f([0 , r_0 ] ) ) + \left({{a(f(\sigma(r_o)))}\over{\pi d}}\right)^{1/2 } \rho_\mathbb{d}(r_0 , r)^{1/2 } \\ & ~\leqslant~ & { \cal e}(f([0 , r_0 ] ) ) + \left({{a(f(\sigma(r_o)))}\over{\pi d}}\right)^{1/2 } \rho^{1/2 } \ .\\\end{aligned}\ ] ] this shows that @xmath290 ) ) = o(\rho^{1/2})$ ] and so proves theorem b when the stolz region is @xmath282 .
it remains to show that any stolz region @xmath25 with vertex at @xmath284 contains @xmath282 provided that @xmath270 is chosen small enough .
let @xmath25 be a stolz region with vertex at @xmath284 and @xmath291 finite .
since @xmath0 is bounded on any finite disc @xmath292 of radius @xmath293 , we may assume that @xmath25 contains the entire half - line @xmath50 .
the fact that @xmath25 is a stolz region means that there is a constant @xmath294 with @xmath295 \geqslant c ( 1-r ) \qquad \hbox { for each } \qquad r\in [ 0,1 ) .\ ] ] for each @xmath63 in the disc @xmath296 we have @xmath297 so @xmath298 this means that the hyperbolic distance @xmath299 from @xmath173 to the boundary of @xmath25 satisfies @xmath300 hence @xmath25 contains the region @xmath301 provided that @xmath302 . since @xmath291 is finite , so is @xmath286 .
we have already proved that , in this case , @xmath303
so the proof of theorem b is complete .
note that we also have a version of these inequalities for functions that are not in the dirichlet class .
write @xmath304 for the area of the image under @xmath0 of the disc @xmath305 with hyperbolic radius @xmath279 .
for the geodesic @xmath306 $ ] , the region @xmath103 is a subset of the disc @xmath307 .
hence theorem 4.1 gives @xmath308 ) ) \
\leqslant\ \left({{a(\rho + d)}\over{\pi d}}\right)^{1/2 } \rho^{1/2 } .\ ] ]
in this section we give a simple geometric argument to show how theorem c follows directly from theorem b and the distortion theorem for univalent functions .
the proof in @xcite of the main inequality in theorem c depends crucially on theorem d , which is stated and proved geometrically in the next section .
we will first prove the following slightly modified form of theorem c. there is a constant @xmath309 such that , if @xmath0 is analytic and univalent in @xmath1 with @xmath310 , then @xmath311 for each @xmath14 and each @xmath278 .
this certainly implies that @xmath312 when @xmath0 is bounded on the half - line @xmath31 .
we begin with a simple geometric result .
this is a version of the distortion theorem for univalent functions .
let @xmath75 be a simply connected subdomain of @xmath2 , and suppose that @xmath313 .
then for any two points @xmath314 and @xmath315 in @xmath75 with @xmath316 , we have @xmath317 as @xmath75 is simply connected and @xmath313 , if @xmath318 then @xmath319 } \ \geqslant\ { 1\over 2|z|}.\ ] ] now let @xmath191 be the hyperbolic geodesic in @xmath75 joining @xmath314 to @xmath314
. then @xmath320\right| \ = \ \log { |w_2|\over |w_1|}\\\end{aligned}\ ] ] and
the proof is complete .
let @xmath321 ) < d \}$ ] be the neighbourhood of the geodesic segment @xmath322 $ ] as in the previous section .
if we show that @xmath104 lies within a disc of radius @xmath323 , then @xmath324 so theorem 4.1 will show that @xmath325 and complete the proof .
for each point @xmath326 , there is a @xmath327 $ ] with @xmath328 .
the schwarz - pick lemma shows that @xmath329 decreases hyperbolic lengths .
so @xmath330 .
lemma 5.2 now shows that @xmath331 .
consequently , @xmath332\ } .\ ] ] this means that @xmath104 lies within a disc centred on @xmath84 and with radius @xmath333 , as required .
theorem 5.1 is certainly strong enough to show that @xmath334 when @xmath0 is bounded on the half - line @xmath31 .
however , for completeness , we will show that theorem 5.1 implies theorem c. suppose that @xmath255 is analytic and univalent with @xmath28 .
choose @xmath84 to be a point of @xmath335 with minimal modulus , and let @xmath336 .
we will show that @xmath337 .
we may suppose that @xmath338 as , otherwise , this is obviously true .
then , as @xmath339 is a hyperbolic contraction of @xmath1 into @xmath11 , we have @xmath340 so that @xmath341 in this case too . since @xmath342 , we now have @xmath343 provided that @xmath344 , the right side is certainly bounded by @xmath345 and so theorem c follows from 5.1 .
we conclude this discussion of theorem c with the following variant of it .
first , we introduce the spaces @xmath346 of measurable functions on @xmath50 whose @xmath347-th power is integrable with respect to the hyperbolic length @xmath348 . in the notation
above , if @xmath349 then @xmath350 as @xmath21 , where @xmath351 . in particular , if @xmath352 is square - integrable with respect to hyperbolic length along the ray ending at @xmath18 , then @xmath353 as @xmath21 .
we shall only consider the radius @xmath138 but a similar argument will hold for any radius .
suppose that @xmath0 is analytic in @xmath1 , and take any @xmath354 and @xmath355 in @xmath356 $ ] with @xmath357 .
hlder s inequality gives @xmath358 so that if the function @xmath359 is in @xmath346 then ( 5.1 ) holds with a @xmath360-estimate instead of a @xmath361-estimate .
the conversion from the @xmath360-term to the @xmath361-term is carried out as in the proof of theorem a in section 3 .
the proof of theorem c given in @xcite depends on the following inequality due to marcinkiewicz and zygmund ( theorem 2 , @xcite ) . for each stolz region @xmath25 in @xmath1 at the point
@xmath18 there is a number @xmath362 such that if @xmath0 is analytic in @xmath1 then @xmath363 there is no discussion in @xcite of why one might expect the factor @xmath364 to appear in the integrand in ( 6.1 ) , but a close examination of the proof there shows that it is an euclidean version of what is a completely standard geometric construction in hyperbolic geometry .
for this reason one should not be surprised at the appearance of such a factor .
consistent with the general point of view expressed in this article , we suggest that theorem d ( and its proof ) should be rewritten in the following form . for each stolz region @xmath25 in @xmath1 at the point
@xmath18 there is a number @xmath362 such that if @xmath0 is analytic in @xmath1 then @xmath365 note that this implies that , when @xmath26 has finite area , then @xmath366 is square integrable with respect to hyperbolic length on the hyperbolic ray @xmath50 .
our proof of theorem d is essentially that given in @xcite but with the considerable benefits gained by using hyperbolic geometry .
first , we clarify what we mean by a stolz region .
it is customary to consider a stolz region at the point @xmath284 to be a jordan domain whose closure lies in @xmath367 , and whose boundary contains two euclidean straight line segments which have a common endpoint at @xmath284 .
this is a euclidean stolz region and it is not conformally invariant . for most purposes it is better to use a hyperbolic stolz region which we now describe .
let @xmath368 be any geodesic ray in @xmath1 ending at , say @xmath369 , and let @xmath105 be the entire geodesic that contains @xmath368 .
for each positive number @xmath270 the _ lens region _
@xmath370 is the set of points in @xmath1 whose hyperbolic distance from @xmath105 is strictly less than @xmath270 ; this is clearly the union of all open discs of radius @xmath270 whose centres lie in @xmath105 .
hyperbolic stolz region _ at @xmath369 is one end of a lens region ; for example , the union of all open discs of radius @xmath270 whose centres lie in the half - ray @xmath368 . if @xmath105 is the real diameter @xmath371 of @xmath1 , then @xmath370 is bounded by two arcs of circles , each passing through @xmath372 , and it follows from this that in almost every case in complex analysis one can use either a euclidean or a hyperbolic stolz region .
the advantage of hyperbolic stolz regions ( and lens regions ) is that they are defined in any domain that supports a hyperbolic metric , and they are conformally invariant .
we are now ready to give our version of the proof of theorem d and in this we shall use a hyperbolic stolz region .
we begin by stating two standard results in hyperbolic geometry ( which , for completeness , we prove in the appendix ) .
the first result is the hyperbolic version of the well known fact that if @xmath0 is analytic in a domain containing a closed euclidean disc @xmath373 with centre @xmath258 , then @xmath374 is the average value of @xmath0 over the disc @xmath373 .
it is perhaps not so well known that if @xmath0 is analytic in @xmath1 , the same is true when @xmath373 is a closed hyperbolic disc .
the second result is a type of harnack inequality for the hyperbolic density in @xmath1 .
suppose that @xmath0 is analytic in @xmath1 , and that @xmath373 is a closed hyperbolic disc in @xmath1 with hyperbolic centre @xmath258 .
then @xmath374 is the non - euclidean average of @xmath0 taken over @xmath373 . for all @xmath63 and @xmath84 in @xmath1 , @xmath375 we may assume that @xmath376 .
let @xmath25 be the hyperbolic stolz region consisting of points whose hyperbolic distance from @xmath138 is strictly less than @xmath270 .
for any point @xmath377 on the diameter @xmath371 , let @xmath378 be the open hyperbolic disc with centre @xmath377 and radius @xmath270 , and let the hyperbolic area of @xmath378 ( which is independent of @xmath377 ) be @xmath379 . then , using lemma 6.1 followed by the cauchy - schwarz inequality , we have @xmath380 we deduce that @xmath381 the last step being by lemma 6.2 .
now define a sequence @xmath382 by @xmath383 and @xmath384 , where @xmath385 .
next , let @xmath386 be the point in @xmath387 $ ] that maximises @xmath388 .
then , using ( 6.2 ) with @xmath389 we obtain @xmath390 now each @xmath391 has radius @xmath270 , and as @xmath392 we see that any point in @xmath25 can lie in at most five of the @xmath391 .
this implies that @xmath393 and this completes our proof of theorem d. notice that this proof also gives an explicit estimate for the size of @xmath394 , namely @xmath395 as @xmath396 .
in order to apply theorem d we need to know when the image of a stolz region has finite area , and for this we quote the following result ( theorem 4 @xcite ) .
let @xmath0 be analytic in @xmath1 , and let @xmath397 be the set of points on @xmath19 where @xmath0 has an angular limit ( from within a stolz region ) .
then , for almost all @xmath369 in @xmath397 , and any stolz region @xmath25 at @xmath369 , we have @xmath398 .
although the exponent @xmath24 in theorem a is the best possible , we can obtain more detailed information on the rate of growth of @xmath22 by using integrals .
kennedy and twomey ( * ? ? ?
* theorem 1 ) showed that if @xmath10 is finite then @xmath399 ^ 2}}\,dr \
\leqslant\ { a(f)\over\pi}\ ] ] for all @xmath14 . as usual , it is sufficient to consider the case @xmath400 and we write @xmath226 for @xmath98 . as @xmath401
this inequality can be rewritten as @xmath402 or better still as @xmath403 for some constant @xmath309 .
we shall now state and prove an analogous result for @xmath58-spaces .
let @xmath404 be analytic and recall that @xmath366 is the factor by which @xmath0 changes scale from the hyperbolic metric at @xmath405 to the euclidean metric in @xmath2 .
the hyperbolic area measure on @xmath1 is @xmath406 .
let @xmath407 be the set of analytic functions @xmath408 for which @xmath409 is finite .
we will assume that @xmath410 and write @xmath355 for the conjugate index so that @xmath411 .
note that , for @xmath412 , we have @xmath413 thus @xmath414 consists of those analytic functions with finite area @xmath10 and so is the dirichlet class @xmath7 .
we now generalize ( 7.1 ) . for a function @xmath0 in @xmath407
, @xmath415 consequently , @xmath416 poisson s formula shows that if @xmath417 , then @xmath418 as @xmath419 we may apply jensen s inequality to this and obtain @xmath420 now let @xmath421 and @xmath422 ; thus @xmath423 .
then @xmath424 therefore , @xmath425 now consider the integral @xmath426 integration by parts gives @xmath427 so hlder s inequality gives @xmath428 this , together with inequality ( 7.2 ) , shows that @xmath429 as required .
this completes the proof of theorem 7.1 .
the inequalities in theorem 7.1 can be shown to be optimal by using the idea in example 3.2 , and we give a brief informal discussion of this .
first , the following argument shows that the integral inequality of kennedy and twomey is stronger than theorem a. when the integral @xmath430 converges we see that , for any @xmath431 , there is an @xmath432 with @xmath433 the distance @xmath434 is an increasing function of @xmath435 , so @xmath436 and hence @xmath437 . next , we consider the inequalities in theorem 7.1 . consider a continuous function @xmath438 $ ] and the corresponding simply connected domain @xmath439 there is a conformal map @xmath440 from @xmath1 onto @xmath441 with @xmath442 and @xmath443 .
this @xmath0 maps the radius @xmath50 onto the positive real axis in @xmath441 .
provided that @xmath444 varies slowly , say @xmath445 , it is easy to estimate the hyperbolic density @xmath446 at any point @xmath63 .
if @xmath447 , the domain @xmath441 lies between the square @xmath448 and the slit plane @xmath449 , so the density lies between the densities for these two domains .
hence , @xmath450 $ ] .
this is enough to show that @xmath451 if and only if @xmath452 . in this situation ,
@xmath453 thus the integral in ( 7.2 ) is @xmath454 it is now clear that ( 7.2 ) is optimal
. for the inequality in theorem 7.1 we need to do a little more .
the same argument gives @xmath455 provided we look at functions @xmath444 that decay very slowly ( so they are essentially constant over longer and longer intervals as @xmath456 ) , we can ensure that @xmath457 this shows that the inequality in theorem 7.1 is optimal .
finally , we remark that inequalities between @xmath45 and @xmath458 were studied extensively in @xcite and although the results in @xcite are not expressed in hyperbolic terms , we shall nevertheless write them in this form here . there is an example @xcite which shows that there exists a bounded analytic @xmath0 in @xmath1 , and a sequence @xmath459 in @xmath1 , such that @xmath460 and @xmath461 and there is a proof ( on p.146 ) that ( 8.1 ) is best possible in the following sense .
let @xmath462 be defined and positive for @xmath463 , and suppose that @xmath464 as @xmath21 . then there exists a function @xmath0 , analytic and univalent in @xmath1 , and continuous in @xmath465 , and a sequence @xmath382 satisfying @xmath466 , such that as @xmath467 , @xmath468 next , ( * ? ? ? *
theorem 6 , p.141 , corollary 8 , p.143 ) , and @xcite contain the following result ( which may be considered as a local version of beurling s theorem although the exceptional set here has measure zero ) .
suppose that @xmath0 is analytic and univalent in @xmath1 .
then for almost all @xmath14 , @xmath469 as @xmath470 uniformly in any stolz region at @xmath18
. in particular , for almost all points @xmath369 on @xmath471 , the curve @xmath472\big)$ ] is rectifiable .
a slightly sharper result is also given in ( * ? ? ?
* lemma 4 , p.143 ) .
suppose that @xmath0 is analytic and univalent in @xmath1 .
then for almost all @xmath14 , as @xmath21 , @xmath473
if the complement of a domain @xmath75 in @xmath474 has at least three points then @xmath1 is the universal covering space of @xmath75 and the hyperbolic metric on @xmath1 projects under any universal covering map @xmath475 to the hyperbolic metric @xmath76 of @xmath75 which is defined by the line element @xmath476 , where @xmath477 it follows from this that @xmath478 is a local isometry relative to the two hyperbolic metrics .
when @xmath75 is multiply connected @xmath478 is not injective and so it is not a global isometry .
let us illustrate these ideas in the particular case when @xmath75 is an annulus .
let @xmath479 then @xmath355 maps @xmath1 conformally onto the upper half - plane @xmath34 , @xmath79 maps @xmath34 conformally onto the strip @xmath480 , and @xmath481 maps @xmath25 ( conformally , but not bijectively ) onto the annulus @xmath482 we write @xmath483 ; then @xmath0 is a universal cover map of @xmath27 , and the hyperbolic metric of @xmath27 has density @xmath484 , where @xmath485 a straightforward calculation shows that @xmath486 if and only if @xmath487 , and that for @xmath488 , we have @xmath489 .
we deduce that in this example , @xmath490 we would like to know whether or not there is a _ geometric _ explanation for the fact that we have an exponent @xmath24 in theorem a but not in ( 9.2 ) .
there is an obvious and significant geometric difference between the case of a univalent map @xmath0 of @xmath1 onto some domain @xmath75 , and the case of a covering map , say @xmath491 , of the annulus : in the former case the @xmath4-images of geodesics in @xmath1 accumulate only on the boundary of @xmath75 , whereas in the case of the annulus described above , we have constructed a geodesic in @xmath1 whose image under the covering map does not accumulate anywhere on the boundary of the annulus .
it is reasonable , then , to ask whether this geometric difference accounts for the different exponents in ( 3.1 ) and ( 9.2 ) .
we note in passing that some type of boundedness condition on @xmath0 is necessary in order to ensure that @xmath226 grows no faster than some power of @xmath492 . indeed , for the function @xmath493 , which is a conformal map of @xmath1 onto the upper half - plane @xmath34 , we have @xmath494 $ ] as @xmath21 .
we shall restrict ourselves here to a description of a simple case of covering maps and this will be sufficient to substantiate the ideas described above in many cases . the following discussion is simply a generalization of the case of the annulus described above .
let @xmath75 be a bounded domain of finite connectivity bounded by a finite number , say @xmath495 of ( sufficiently ) smooth closed curves ( none of which degenerate to a point ) .
in such circumstances there is a universal covering map @xmath91 which is invariant under a fuchsian group @xmath496 acting on @xmath1 and which is such that @xmath75 is conformally equivalent to the quotient surface @xmath497 .
these assumptions on @xmath75 imply the following facts ; here , we assume familiarity with the basic theory of fuchsian groups . 1
. the group @xmath496 is finitely generated , and its limit set @xmath432 on @xmath471 has hausdorff dimension strictly between @xmath498 and @xmath284 .
in particular , @xmath432 has positive capacity and zero linear measure
each ray from @xmath498 in @xmath1 that ends at a point @xmath18 in @xmath499 has a rectifiable @xmath4-image that ends at a point on @xmath471 . for such rays , @xmath500 as @xmath21 .
each ray from @xmath498 in @xmath1 that ends at a point @xmath18 in @xmath432 has an @xmath4-image that returns infinitely often to some compact subset of @xmath1 .
we shall now obtain estimates of @xmath22 in this situation .
roughly speaking , our result show that the calculations given above for the annulus are typical of this situation .
let @xmath0 be a universal cover map of a bounded domain @xmath75 as described above .
then , for every @xmath14 , @xmath501 as @xmath21
. moreover , 1 . if @xmath502 then @xmath500 as @xmath21 , and 2 .
if @xmath18 is fixed by some hyperbolic element in @xmath496 ( and such points are dense in @xmath432 ) , then there is some positive number @xmath503 ( that depends on @xmath14 ) such that for all @xmath173 , @xmath504 we suggest that the way that @xmath22 varies in comparison with @xmath52 might be governed entirely by nature of the limit point @xmath18 as exhibited by the geometric behaviour of the @xmath4-image of the geodesic ray , say @xmath105 , from @xmath498 to @xmath18 .
explicitly , we make the following 1
. if @xmath18 is not in @xmath432 , then @xmath108 ends at a single point on the boundary of @xmath75 ( or , more generally , if @xmath85 is not smooth , at a prime end on @xmath85 ; see example 3.3 ) , and in this case we have @xmath505 .
2 . if @xmath506 , then @xmath108 returns infinitely often to a compact subset of @xmath75 ( this is a known fact ) , and this probably implies that @xmath22 grows rather rapidly with @xmath173 . in this case we might expect that @xmath507 , where @xmath127 is in @xmath508 $ ] and depends on the geometric position of @xmath108 .
3 . in the extreme case when @xmath18 is a hyperbolic fixed point , @xmath108 lies entirely within a compact subset @xmath503 , say , of @xmath75 . in this case , the euclidean and hyperbolic lengths along @xmath108 are comparable , and we find that @xmath509 is bounded above and below by positive finite constants as is given by ( 9.3 ) and ( 9.4 ) .
this is the case of the annulus discussed above .
as @xmath75 is bounded it lies in some disc @xmath510 .
if we now apply the schwarz - pick lemma to the function @xmath511 ( which maps @xmath1 into itself ) we obtain @xmath512 this gives @xmath513 , and integration of both sides gives a sharper form of ( 9.3 ) , namely @xmath514 . to establish ( 9.4 )
suppose that @xmath18 is a fixed point of some hyperbolic element in @xmath496 ; then the @xmath4-image of the ray to @xmath18 lies in a compact subset of @xmath75 so that @xmath77 is bounded above , say by a positive number @xmath515 , on this image . as the covering map
is a local isometry , we find from ( 9.1 ) that @xmath516 this completes the proof .
theorem a shows that in a simply connected domain with finite area we have @xmath517 .
however , we have seen that @xmath434 can grow like the first power @xmath435 when we measure lengths along a circular path in an annulus .
it seems likely that , given any @xmath123 in @xmath138 , one could construct a multiply connected domain with finite area in which @xmath518 for a suitable geodesic .
we give a brief sketch of an idea which we believe will work ; however , the estimates needed would be very delicate and we have not completed the details . consider a domain @xmath75 obtained by abutting a sequence of annuli @xmath519 , as in the diagram . here
@xmath519 is an annulus of inner radius @xmath520 and outer radius @xmath521 .
the total area is approximately @xmath522 , so this sum must converge . for any sequence of natural numbers @xmath523 , there is a hyperbolic geodesic @xmath105 in @xmath75 that starts at a fixed point @xmath524 and winds @xmath525 times around @xmath519 before moving to @xmath526 .
( there is such a geodesic since every homotopy class contains a geodesic . ) the euclidean length of the geodesic from @xmath527 until it leaves @xmath528 is asymptotically : @xmath529 * figure 2 . *
( the geodesic is shown as a dashed line . )
the part of @xmath105 within @xmath519 lies at a distance of approximately @xmath520 from the boundary , so its hyperbolic length is roughly @xmath530 thus the hyperbolic length of the geodesic from @xmath527 until it leaves @xmath528 is asymptotically : @xmath531 for any @xmath532 , set @xmath533 and @xmath534 .
then @xmath535 converges , so @xmath75 will have finite area .
moreover , @xmath536 and @xmath537 so @xmath538 .
in this section we shall consider functions @xmath0 that are analytic in @xmath1 , and such that the image domain @xmath11 supports a hyperbolic metric , and we shall say that @xmath0 is _ hyperbolic _ in @xmath1 whenever this is so .
not every analytic map in @xmath1 is hyperbolic ; for example , if @xmath79 maps @xmath1 conformally onto the strip given by @xmath539<3\pi$ ] , then @xmath540 is not hyperbolic .
however , many functions analytic in @xmath1 are hyperbolic ; for example , all bounded functions , and all univalent functions , are .
suppose now that @xmath0 is analytic in @xmath1 and hyperbolic .
then , instead of comparing @xmath226 with the hyperbolic length @xmath52 of @xmath111 $ ] , we can attempt to compare it with the hyperbolic length of the image curve @xmath38)$ ] .
let us denote this length by @xmath224 ; thus @xmath226 and @xmath224 denote the euclidean and hyperbolic lengths , respectively , of @xmath38)$ ] , and we are asking for a comparison of the euclidean and hyperbolic lengths of the curve @xmath38)$ ] .
we emphasize that the problem of estimating @xmath226 in terms of @xmath224 is more difficult than the original problem of obtaining an estimate in terms of @xmath492 because the schwarz - pick lemma shows that for any analytic hyperbolic @xmath0 we have @xmath541 .
of course , if @xmath0 is univalent in @xmath1 ( or if @xmath0 is a covering map of @xmath1 onto @xmath542 ) , then @xmath543 so the two problems are the same . we shall now prove the following result .
suppose that @xmath0 is analytic and bounded in @xmath1 , and that @xmath0 is hyperbolic .
then as @xmath21 .
@xmath544 we remark first that the example of the universal covering map of an annulus ( or indeed , any covering map ) shows that one can not do better than the estimate ( 10.1 ) for in these cases , @xmath545 .
the proof follows the same lines as the proof of theorem a in @xcite , except that we use integrals instead of power series . using the cauchy - schwarz inequality we obtain @xmath546 now as @xmath75 is bounded
, @xmath77 has a positive lower bound , say @xmath547 , in @xmath75 and we obtain the inequality @xmath548 , or @xmath549 .
this completes the proof of theorem 10.1 . finally , we observe that the proof of theorem 10.1 also provides a proof of the following result .
suppose that @xmath0 is analytic and hyperbolic in @xmath1 , and that the function @xmath77 is bounded below on the ray @xmath31 .
then ( 10.1 ) holds as @xmath21 .
we end with the remark that in the general case of a map @xmath0 analytic in @xmath1 , if @xmath0 is smooth enough on , say , an arc of @xmath19 centred at @xmath18 , then we might expect that for a sufficiently small positive @xmath550 , the @xmath4-image of @xmath551 has finite area .
if this is so then , from theorem b , we have @xmath552 as @xmath21 .
thus , in some general sense , if the @xmath4-image of a geodesic ray ends ( smoothly ) on the boundary of @xmath542 , or , more generally , on the boundary of the riemann surface for the function @xmath553 , then we expect that ( 11.1 ) holds .
by contrast , if the @xmath4-image of the geodesic ray circulates infinitely often around one of the ` holes ' in @xmath542 , then @xmath22 and @xmath52 are of a comparable size as @xmath21 .
it may be that more delicate estimates relate the comparative sizes of @xmath22 and @xmath52 to the finer detail of the geometry of the images of the geodesic rays .
we complete this paper by giving the proofs of theorem 3.4 , and of lemmas 6.1 and 6.2 .
we begin with a proof of theorem 3.4 , this will follow from a result showing how hyperbolic geodesics move when we expand a domain and we will use the observation that a geodesic is the level set for harmonic measure . let @xmath123 be a closed arc of the unit circle and @xmath127 the complementary arc . then the harmonic measure for @xmath123 in @xmath554
is the unique harmonic function @xmath555 with boundary values @xmath498 on @xmath127 and @xmath284 on the interior of @xmath123 .
there are no boundary values at the endpoints @xmath556 of @xmath123 but the harmonic measure is bounded there .
it is simple to calculate the harmonic measure explicitly and see that the level set @xmath557 is the hyperbolic geodesic from @xmath558 to @xmath559 .
a similar result holds for domains with a jordan boundary .
let @xmath397 be a simply - connected , hyperbolic , jordan domain with @xmath123 as a closed arc of @xmath560 and @xmath127 as the complementary arc of @xmath560 .
then the riemann map @xmath561 extends continuously to the boundary @xmath560 and maps it homeomorphically onto the unit circle .
if @xmath562 is the harmonic measure of the arc @xmath563 in @xmath554 , then @xmath564 is the harmonic measure of @xmath123 in @xmath397 .
this function @xmath565 is harmonic with boundary values @xmath498 on @xmath127 and @xmath284 on the interior of @xmath123 .
the hyperbolic geodesic between the endpoints @xmath556 of @xmath123 is then the level set @xmath566 since @xmath0 is a hyperbolic isometry . 1 .
@xmath397 is a jordan domain whose boundary @xmath560 consists of a closed arc @xmath123 , joining the endpoints @xmath558 and @xmath559 , and the complementary arc @xmath127 ; 2 .
@xmath397 is a subset of @xmath75 with @xmath568 and @xmath569 .
then a geodesic @xmath550 from @xmath558 to @xmath559 in @xmath75 and the geodesic @xmath570 from @xmath558 to @xmath559 in @xmath397 do not meet in @xmath397 .
indeed , @xmath570 separates @xmath550 and @xmath123 .
we first show that it is sufficient to prove the result when @xmath75 is the unit disc .
let @xmath571 be a riemann map for @xmath75 . since @xmath397 is a jordan domain , each boundary point @xmath369 of @xmath123 is accessible . also , since @xmath572 , the point @xmath369 is also an accessible boundary point of @xmath75 defined by an approach to @xmath369 within @xmath397 .
each such point @xmath573 corresponds to a unique prime end for @xmath75 defined by an approach to @xmath369 within @xmath397 .
these prime ends fill out a closed arc @xmath574 of @xmath575 with endpoints @xmath576 .
the image @xmath577 is a subdomain of the unit disc @xmath554 and is bounded by the closed arc @xmath574 and the arc @xmath578 that lies in the interior of @xmath579 .
the conformal map @xmath0 preserves geodesics , so it is sufficient to prove the proposition when @xmath75 is the unit disc .
in this situation , both @xmath75 and @xmath397 are jordan domains and so the harmonic measures extend continuously to the boundary except at endpoints . let @xmath565 be the harmonic measure of @xmath123 in @xmath397 and let @xmath580 be the harmonic measure of @xmath123 in the unit disc @xmath75 .
then @xmath14 has boundary values @xmath498 on @xmath127 and @xmath284 on the interior of @xmath123 .
the restriction of @xmath562 to @xmath397 has boundary values @xmath581 at each @xmath582 and @xmath284 at each interior point of @xmath123 .
hence , @xmath583 in particular , the geodesics @xmath584 and @xmath585 are disjoint .
furthermore , @xmath586 for each @xmath587 , so the geodesic @xmath570 lies in the region @xmath588 bounded by @xmath123 and @xmath550 .
we now complete the proof of theorem 3.4 .
it will suffice to show that the geodesic @xmath105 in @xmath75 does not cross the segments @xmath589 \times \{0\}$ ] and @xmath590 \times \{0\}$ ] for some constant @xmath515 with @xmath591 .
let @xmath397 be the square @xmath592 , then @xmath397 is a jordan domain lying within @xmath75 and the line segment @xmath593 $ ] lies in the boundary of both @xmath397 and @xmath75 .
label the endpoints of this line segment @xmath594 and @xmath595 .
let @xmath550 be the hyperbolic geodesic in @xmath75 joining @xmath558 to @xmath559 and let @xmath570 be the hyperbolic geodesic in @xmath397 joining @xmath558 to @xmath559 .
the reader is urged to draw a diagram . the arcs @xmath570 and @xmath123 bound a jordan region that intersects the @xmath377-axis in some segment @xmath590 $ ] .
the last proposition shows that the geodesic @xmath550 in @xmath75 can not meet this segment . by hypothesis
, the geodesic @xmath105 in @xmath75 does not have an endpoint in @xmath123 . by using the riemann map from @xmath75 to the unit disc
, we see that @xmath105 can not meet the region bounded by @xmath550 and @xmath123 , or it would have at least one endpoint in @xmath123
. therefore it certainly can not meet the segment @xmath590 $ ] .
an entirely similar argument shows that @xmath105 can not meet @xmath589 $ ] and completes the proof of theorem 3.4 .
first we prove this when @xmath598 . in this case
@xmath373 is of the form @xmath599 , where @xmath293 , and the result follows from the euclidean form of the mean value property .
explicitly , we have @xmath600 we now integrate with respect to @xmath14 first , and using the fact that @xmath601 is the mean value of @xmath0 over the circle @xmath602 , we obtain @xmath603 as this holds when @xmath0 is the constant function @xmath284 , the integral on the right must be @xmath604 and this gives ( a.1 ) when @xmath598 . now suppose that @xmath373 is any hyperbolic disc with centre @xmath258 , say .
there is a mbius map @xmath105 of @xmath1 onto itself such that @xmath605 , say , and @xmath606 . now
the function @xmath112 given by @xmath607 is analytic in @xmath608 and @xmath609 so that @xmath374 is the average value of @xmath112 over the disc @xmath608 . as @xmath610 ( this is just ( 3.2 ) again ) , this shows that @xmath611 as required .
this is easy .
take @xmath63 and @xmath84 in @xmath1 , and without loss of generality we may suppose that @xmath612
. then @xmath613 \\ & ~=~ & \exp \left [ \log\left({1+|w|\over 1-|w|}\right ) -\log\left({1+|z|\over 1-|z|}\right)\right]\\ & ~=~ & \left({1+|w|\over 1+|z|}\right)^2{1-|z|^2\over 1-|w|^2}\\ & ~\geqslant~ & { \lambda_\mathbb{d } ( w ) \over 4\lambda_\mathbb{d } ( z)}.\end{aligned}\ ] ] the other inequality follows by symmetry . | let @xmath0 be a function that is analytic in the unit disc .
we give new estimates , and new proofs of existing estimates , of the euclidean length of the image under @xmath0 of a radial segment in the unit disc .
our methods are based on the hyperbolic geometry of plane domains , and we address some new questions that follow naturally from this approach . |
dyadic data are central in social science applications ranging from international relations to `` speed dating . ''
a challenge with dyadic data is to account for the complex dependency structure that arises due to the connections between dyad members .
for instance , in a study of international conflict , a change in leadership in one country may affect relations with all countries with which that country is paired in the data , thereby inducing a correlation between these dyadic observations .
this generates a cluster of dependent observations associated with that country .
as leadership changes occur in multiple countries , the correlations emanating from each of these countries will overlap into a web of interwoven clusters .
we refer to such interwoven dependency in dyadic data as `` dyadic clustering . '' by ignoring the dyadic clustering , the analysis would take the dyad - level changes emanating from a single leadership change as independently informative events , rather than a single , clustered event .
an analysis that only accounts for correlations in repeated observations of dyads ( whether by clustering standard errors or using dyad fixed effects ) would fail to account for such inter - dyad correlation . statistical analysis of dyads typically estimates how dyad - level outcomes ( e.g. , whether there is open conflict between countries or the decision for one person to ask another on a date ) relate to characteristics of the individual units as well as to the dyad as a whole ( e.g. , measures of proximity between units ) .
the usual approach is to regress the dyad - level outcome on unit- and dyad - level predictors . due to dyadic clustering ,
the observations contributing to such an analysis are not independent .
failure to account for dyadic clustering may result in significance tests or confidence intervals with sizes that are far too small relative to the true distribution of the parameters of interest .
we establish sufficient conditions for the consistency of a non - parametric sandwich estimator @xcite for the variance of regression coefficients under dyadic clustering .
cluster - robust sandwich estimators are common for addressing dependent data ( ( * ? ? ?
8) ; @xcite ) .
@xcite provide a sandwich estimator for `` multi - way '' clustering , accounting , for example , for clustering between people by geographic location and age category .
we extend these methods to dyadic clustering , accounting for the fact that dyadic clustering does not decompose neatly into a few crosscutting and disjoint groups of units ; rather , each _ unit _ is the basis of its own cluster that intersects with other units clusters .
2.5 ) propose a sandwich estimator for dyadic clustering that is very similar to what we propose below .
their derivation is constructed through analogy to the results of @xcite .
however , neither paper establishes conditions for consistency under dyadic clustering .
we establish such consistency conditions .
we also provide extensions to the longitudinal case and generalized linear models such as logistic regression . we evaluate performance with simulations and a reanalysis of a classic study on interstate disputes @xcite .
the appendix generalizes to weighted data and generalized linear models , and the supporting information provides another illustration with a speed dating experiment @xcite .
current statistical approaches to handling dyadic clustering include the use of parametric restrictions in mixed - effects models @xcite , the spatial error - lag model @xcite , or permutation inference for testing against sharp null hypotheses @xcite .
these approaches have important limitations that our approach overcomes .
first , mixed effects models and spatial lag models impose a parametric structure to address the clustering problem .
this makes them sensitive to misspecification of the conditional mean ( that is , deviations between the data and the linearity assumption ) .
our approach is robust in that it provides asymptotically valid standard errors even under such misspecification .
this is valuable in itself to the extent that models are typically approximations @xcite , although readers should not take this to mean that they are free from the obligation to fit as good an approximation as possible .
it is also valuable in providing a reliable benchmark to use in evaluating model specification @xcite .
second , @xcite s solution of non - parametric randomization inference does not provide a procedure for obtaining valid confidence intervals , while our procedure does .
third , all three of these alternatives require considerable computational power , possibly exceeding available computational resources even for data analyses that are common in international relations or network analysis .
we demonstrate below that a mixed - effects or spatial error - lag approach is infeasible in a typical international relations example .
in contrast , our proposed estimator is easy to compute .
the variance estimation methods we develop are a natural complement to non- and semi - parametric approaches to regression with dyadic data @xcite .
we work within the agnostic " @xcite or `` assumption lean '' @xcite regression framework developed by @xcite , @xcite , and @xcite .
we begin with a cross - section of undirected dyads and derive the basic convergence results for this case .
below we extend these results to repeated dyads , which covers directed dyads and longitudinal data .
proofs are in the appendix .
begin with a large population from which we take a random ( i.i.d . )
sample of units , with the sample members indexed by @xmath0 and grouped into @xmath1 dyads .
pairs of unit indices within each dyad map to dyadic indices , @xmath2 , through the index function @xmath3 , with an inverse correspondence @xmath4 , and we assume no @xmath5 type dyads . consider a linear regression of @xmath6 on a @xmath7-length column vector of regressors @xmath8 ( which includes the constant ) : @xmath9 where @xmath10 is the slope that we would obtain if we could fit this model to the entire population , allowing for possible non - linearity in the true relationship between @xmath11 $ ] and @xmath8 , and @xmath12 is the corresponding population residual.$ ] is nonlinear in @xmath8 , the population residual will itself vary in @xmath8 , which undermines inference that assumes that @xmath12 is independent of @xmath8 .
the agnostic regression approach avoids such an assumption . ] to lighten notation , for the remainder of the discussion we use @xmath13 .
define the sample data as @xmath14 and @xmath15 .
the ordinary least squares ( ols ) estimator is @xmath16 with residual @xmath17 . since the values of @xmath18 and @xmath19
are determined by the characteristics of the units @xmath20 and @xmath21 , @xmath22 and @xmath23 for dyads containing either unit @xmath20 or @xmath21 are allowed to be correlated by construction . however , by random sampling of units , @xmath24 for all dyads that do not share a member . the number of pairwise dependencies between @xmath20 and @xmath21 for which @xmath25 is @xmath26 .
let the support for @xmath27 and @xmath28 be bounded in a finite interval and assume the usual rank conditions on @xmath29 @xcite .
[ lemma - var ] suppose data and @xmath30 as defined above . as @xmath31 , the asymptotic distribution of @xmath32 has mean zero and variance @xmath33^{-1}\var\left[\sum_{d=1}^d x_d\epsilon_d\right]\e[x_dx_d']^{-1 } , \label{varb}\ ] ] with , @xmath34 = \sum_{d=1}^d \left ( \underbrace{\e\{x_{d}x'_d \var[\epsilon_d|{\mathbf{x}}]\}}_{a } + \underbrace{\sum_{d ' \in \mathcal{s}(d ) } \e\{x_{d}x'_{d ' } \cov[\epsilon_d,\epsilon_{d'}|{\mathbf{x}}]\}}_{b } \right ) , \label{varxe}\ ] ] where @xmath35 , the set of dyads other than @xmath36 that share a member from @xmath36 .
@xmath37 is the dyad - specific contribution to the variance , and @xmath38 is the contribution due to inclusion of common units in multiple dyads .
note that these features of the distribution of @xmath32 establish the consistency of @xmath30 as well , which is not surprising given standard results for the consistency of ols coefficients on dependent data @xcite .
given data as defined above , we examine the properties of a plug - in variance estimator that is analogous to the sandwich estimators defined for heteroskedastic or clustered data @xcite .
we establish sufficient conditions for consistency of the plug - in variance estimator .
we consider the cross - sectional case and repeated observations case .
the appendix also contains extensions to weighted observations and generalized linear models .
define the variance estimator @xmath39 under the conditions of lemma [ lemma - var ] and if @xmath27 and @xmath40 have bounded support with non - zero second moments and cross - moments , then as @xmath31 , @xmath41 [ prop1 ] the proposition indicates that @xmath42 provides a consistent estimator to characterize the true asymptotic sampling distribution of the regression coefficients , @xmath30 .
standard error estimates for @xmath30 are obtained from the square roots of the diagonals of @xmath42 .
the assumption of bounded support for @xmath27 and @xmath40 merely rules out situations that are unlikely to arise in real world data anyway ( e.g. , where the mass of the data pushes out toward infinity as @xmath43 grows ) .
repeated dyad observations are common in applied settings . for example , the data may include multiple observations for dyads over time .
dyadic panels are very common in studies of international relations . or ,
if the dyadic information is directional , then the data will contain two observations for each dyad , with one observation capturing outcomes that go in the @xmath20 to @xmath21 direction , and the other capturing outcomes that go in the @xmath21 to @xmath20 direction .
this is conceptually distinct from repeated observations over time .
but if there are dyadic dependences for both senders and receivers of a directed dyad , then the dependence structure for a pair of directed @xmath20-@xmath21 dyads will be the same as if we had repeated observations of an undirected @xmath20-@xmath21 dyad .
the results above translate straightforwardly to the repeated dyads setting . formally
, suppose that for each dyad @xmath36 observations are indexed by @xmath44 , where the @xmath45 values are fixed and finite .
let @xmath46 represent the data for observation @xmath47 from dyad @xmath36 , @xmath48 and @xmath49 and @xmath50 , that is , the stacked @xmath28 vectors and @xmath27 matrices .
let @xmath51 and @xmath52 denote , respectively , the population slope and ols estimator as defined above but now applied to the repeated dyads data .
for the repeated dyads case , assume the same conditions as in proposition [ prop1 ] and consider the following variance estimator @xmath53 then as @xmath31 , @xmath54 - v_r \overset{p}{\rightarrow } 0,\ ] ] and @xmath55 where @xmath56 ^ 2 } \e[x_dx_d']^{-1 } \e\left ( \e\left[x_d'\cov[\epsilon_d|{\mathbf{x}}]x_d\right ] + \sum_{d ' \in \mathcal{s}(d)}\e\left [ x_d'\cov[\epsilon_d , \epsilon_{d'}|{\mathbf{x}}]x_{d'}\right]\right ) \e[x_dx_d']^{-1}.\ ] ] some remarks are in order when it comes to the repeated observations case .
first , our analysis of the repeated observations case takes the @xmath45 values to be fixed and finite and demonstrates consistency as @xmath43 grows . in a cross - national study
, this would mean that one should pay most attention to the number of countries , rather than the amount of time , that are in the data . under fixed @xmath43 and growing @xmath45 further assumptions would be required for consistency , such as serial correlation of fixed order or , more generally , strong mixing over time .
second , with repeated observations , a common strategy for identifying effects is to use dyad - specific fixed effects @xcite . with fixed and finite @xmath45
, this does not introduce any new complications .
as in @xcite for the case of serial correlation with fixed effects , the results for dyadic clustering translate directly to centered data , and dyadic fixed - effects regression amounts to centering the data on dyad - specific means . with fixed @xmath43 and growing @xmath45 ,
however , additional assumptions are needed for consistency @xcite . for efficient computation ,
we follow @xcite to perform a `` multi - way decomposition '' of the dyadic clustering structure : we have the following algebraic equivalence , @xmath57 where @xmath58 @xmath59 @xmath60 and @xmath61 @xmath62 [ prop - decomp ] @xmath63 is the usual asymptotically consistent cluster - robust variance estimator ( with no degress - of - freedom adjustment ) that clusters all dyads containing unit @xmath20 and assumes all other observations to be independent .
@xmath64 is the same cluster robust estimator but clustering all repeated dyad observations .
@xmath65 is the usual asymptotically consistent heteroskedasticity robust ( hc ) variance estimator that assumes all observations ( even within repeated dyad groupings ) are independent . to understand this decomposition , note that dyadic clustering involves clustering on each of @xmath43 units .
but in summing the contributions from unit - specific clusters ( the @xmath63s ) , we double count the dyad contributions ( @xmath64 ) and add in the independent contributions ( @xmath66 ) @xmath43 times .
we can correct for this by subtracting @xmath64 , which also removes the @xmath66 component , and then subtracting @xmath67 of the @xmath66 components .
( the cross section cases simply sets @xmath68 ) .
this decomposition shows that one can compute the dyadic cluster robust estimator using readily available robust standard error commands .
our dyadic cluster robust variance estimator allows one to perform valid inference under less restrictive dependence assumptions that are used to identify random effects or spatial error lag models .
our multi - way decomposition also shows that it is computationally much simpler than those alternatives .
the latter point is relevant when one has many units or time periods in which case random effects and spatial error lag models , despite their restrictions , may be infeasible with current computational resources given the challenges of evaluating a likelihood with as many dependencies as emerge in a dyadic analysis .
we use monte carlo simulations to evaluate the finite sample properties of the proposed estimators under the cross - sectional and repeated dyads settings . we suppose that population values obey the following , @xmath69 where @xmath70 is the @xmath71 observed outcome for the dyad that includes units @xmath20 and @xmath21 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , and @xmath76 are independent draws from standard normal distributions , and the compound error , @xmath77 , is unobserved . in the cross - sectional case ,
we only observe one outcome per dyad , so @xmath78 for all observations ( that is , the @xmath47 subscript is extraneous for the cross - sectional case ) . in the repeated observations case , we have two observations per dyad , so @xmath79 for all dyads .
we fix @xmath80 and @xmath81 .
we use ols to estimate @xmath82 and @xmath83 . in the supplemental information
we show results for mixed effects models as well .
the fact that @xmath72 and @xmath74 are constant across dyads that include unit @xmath20 ( same for @xmath21 ) implies non - zero intra - class correlation in both @xmath84 and @xmath85 among sets of dependent dyads , in which case ignoring the dependence structure will tend to understate the variability in @xmath82 and @xmath83 .
this is the dyadic version of @xcite s problem .
results from 500 simulation runs are shown in figure [ fig : sim1 ] .
the @xmath86-axis shows the sample size , where we show results for samples with 20 , 50 , 100 , and 150 units ( implying 190 , 1,225 , 4,950 , and 11,175 dyads , respectively ) .
the @xmath87-axis is on the scale of the standard error of the coefficients .
the black diamonds plot the empirical standard standard errors for each of the sample sizes ( that is , the standard deviation of the simulation replicates of the coefficient estimates , @xmath82 and @xmath83 ) .
the black box plots show the distribution of our proposed standard error estimator .
the gray box plots show , in the top figures , the distribution of the `` hc2 '' heteroskedasticity robust variance estimator , which does not account for either dyadic or repeat observation clustering @xcite , and for the bottom figures , the `` cluster robust '' analog of @xcite s estimator accounting for dependence in repeated dyad observations @xcite
. goes from 20 to 150 ( implying number of dyads goes from 190 to 11,175 ) for a cross section of dyads ( top ) and repeated dyad observations ( t=2 for all dyads ) .
black box plots show the distribution of standard error estimates using the proposed dyadic cluster robust estimator . in the top two figures ,
gray box plots show standard error estimates from a heteroskedasticity robust estimator while in the bottom two figures , gray box plots show standard error estimates from a `` naive '' cluster - robust estimator that clusters by dyad over repeated observations ( but does not cluster by units across dyads ) .
, title="fig:",scaledwidth=100.0% ] goes from 20 to 150 ( implying number of dyads goes from 190 to 11,175 ) for a cross section of dyads ( top ) and repeated dyad observations ( t=2 for all dyads ) .
black box plots show the distribution of standard error estimates using the proposed dyadic cluster robust estimator . in the top two figures ,
gray box plots show standard error estimates from a heteroskedasticity robust estimator while in the bottom two figures , gray box plots show standard error estimates from a `` naive '' cluster - robust estimator that clusters by dyad over repeated observations ( but does not cluster by units across dyads ) .
, title="fig:",scaledwidth=100.0% ] convergence of our proposed standard errors based on @xmath42 and @xmath88 ( black box plots ) to the true standard error ( black diamonds ) is quick .
the alternative estimators , which do not account for dyadic clustering , grossly understate the variability in the coefficient estimates .
these results suggest that in finite samples from data generating processes that resemble our simulations , the proposed estimator is quite reliable for inference on ols coefficients so long as the sample size is on the order of 50 to 100 units .
most applications in political science operate with samples of at least this size .
changes to the shape of the error distributions ( e.g. , uniform or bimodal ) yield the same results . in the supporting information , we demonstrate comparisons between mixed effects models and our `` robust '' estimator .
we sought also to include a comparison to spatial error lag models , but the models failed to converge with @xmath89 .
we note that in the original @xcite paper , they limited their analysis of dyadic data to small cross sections .
our experience suggests that it may be infeasible to fit such models with substantially larger datasets . ] specifically , we compare ols with dyadic cluster robust standard errors to a two - way random effects model that incorporates normal random effects for each of the @xmath43 units @xcite . the data generating process captured above ( that is , expression [ eq : dgp ] with normal data ) corresponds precisely to the assumptions of a two - way normal random effects model , and so it is no surprise that we find this model to be more efficient than ols and to produce consistent standard errors .
however , as discussed above , such models are sensitive to misspecification .
for example , in fitting a linear approximation model to mildly non - linear data , the dyadic cluster robust standard errors continue to be consistent for the coefficients of the linear approximation .
the degree of misspecification in this case is not extreme ( as illustrated in the appendix ) and resembles the kind of approximations that is common in applied work @xcite .
nonetheless , the random effects standard errors are inconsistent and substantially anti - conservative . such sensitivity
to misspecification is one problem for the random effects model .
another problem arises when there is unobserved country - level heterogeneity that could confound estimates of the coefficients of interest .
in such situations , our dyadic cluster robust estimator would be a natural complement to a fixed effects analysis , analogous to what @xcite proposes for non - dyadic repeated observation settings .
the third problem arises in computational feasibility , an issue that arises in our application to which we now turn .
our application is based on the classic study by @xcite on the determinants of international militarized conflicts .
they study how the likelihood of a militarized dispute between states in a dyad relates to various dyad - level attributes , including an indicator of whether the two states are formal allies , the log of the ratio of an index of military capabilities , the lowest score in the dyad on a democracy index , the ratio of the value of trade and the larger gdp of the two states , an indicator of whether both states are members of a common international organization , an indicator of whether the states are geographically noncontiguous , the log of the geographic ( euclidean ) distance between the two states capitals , and an indicator of whether both states are `` minor powers '' in the international system .
dyadic clustering could arise in many ways with these data , for example if a country entered into an alliance , thereby changing the joint alliance indicators , or if the military capabilities of a country changed , thereby changing the power ratios .
we replicate russett and oneal s primary analysis as reported in their table a5.1 .
they use annual data on 146 states in the international system paired into 1,178 dyads ( out of 10,585 possible ) and observed for as few as one and as many as ninety years between 1885 and 1991 , for a total of 39,988 observations . in their original analysis ,
russett and oneal fit a gee model assuming ar(1 ) errors within dyads over time @xcite .
table [ tab : russett ] shows the results of our reanalysis .
columns ( 1 ) and ( 2 ) replicate the original published results . columns ( 3)-(6 ) show coefficients and various standard error estimates for a simple ( pooled ) logistic regression .
there is little difference in the coefficient estimates from the gee - ar(1 ) model as compared to the simple logistic regression , so we focus on the standard error estimates .
column ( 4 ) contains estimates that account only for dyad - year heteroskedasticity , column ( 5 ) also accounts for arbitrary dependence over time for each dyad , and then column ( 6 ) also accounts for dyadic clustering .
accounting for the dyadic and repeated - observation clustering results in standard error estimates that are sometimes an order of magnitude larger than what we obtain when we ignore all clustering and also considerably larger than what one would estimate were one to account for repeated dyads clustering but ignore the inter - dyadic clustering .
the latter result is also relevant when comparing the standard errors from the original gee - ar(1 ) model , as they resemble the estimates in column ( 5 ) . in their original analysis , russett and
oneal found that all of the predictors had a statistically significant relationship to the likelihood of militarized conflict .
but when one takes into account dyadic clustering , the coefficient for international organizations would fail to pass a conventional significance test ( e.g. , @xmath90 ) .
& & + & & & & het . & naive cluster & dyadic cluster + predictor of militarized conflict & coef . & s.e . & coef . & robust s.e . &
robust s.e . & robust s.e .
+ alliances & -0.539 & 0.159 & -0.595 & 0.069 & 0.175 & 0.265 + power ratio & -0.318 & 0.043 & -0.328 & 0.020 & 0.047 & 0.070 + lower democracy & -0.061 & 0.009 & -0.064 & 0.005 & 0.010 & 0.015 + lower dependence & -52.924 & 13.405 & -67.668 & 10.734 & 17.560 & 24.749 + international organizations & -0.013 & 0.004 & -0.011 & 0.002 & 0.005 & 0.008 + noncontiguity & -0.989 & 0.168 & -1.048 & 0.074 & 0.181 & 0.185 + log distance & -0.376 & 0.065 & -0.375 & 0.026 & 0.068 & 0.102 + only minor powers & -0.647 & 0.178 & -0.618 & 0.078 & 0.188 & 0.344 + constant & -0.128 & 0.536 & -0.181 & 0.211 & 0.562 & 0.840 + + [ tab : russett ] we attempted also to study how inferences might change with a random effects approach and a spatial error lag approach .
the appropriate random effects model would be a three - way random effects model to account for each state in a dyad as well as the dyad as a whole over time .
we attempted to fit three - way random effects logit models using the lmer package in r and the melogit function in stata .
we posted the jobs to a university cluster with 24 intel xeon cores and 48 gigabytes of ram per node . in neither case
did the random effects models converge .
in fact , in neither case could we obtain estimates of any form after days on the cluster .
for the spatial error lag model , we sought to implement both the spatial error lag model in the spdep package in r and the two - step estimator of @xcite in stata .
the latter take as an input dyadic dependence information that can be produced using the spundir package in stata .
the spatial error lag model in the spdep failed to produce estimates and spundir package failed to resolve within reasonable time ( a week ) while running on the same cluster .. the computational complexity is due to the fact that the model attempts to account for dependence across all countries in the dataset ( rather than small subsets , as would be the case in a typical spatial analysis ) and the fact that in this application the set of cross - section units changes over time and therefore does not allow for a stable adjacency matrix over time . ]
this brings to the foreground the issue of computational complexity , by which our cluster robust estimator is much less demanding .
we have established convergence properties for a non - parametric variance estimator for regression coefficients that accounts for dyadic clustering .
the estimator applies no restrictions on the dependency structure beyond the dyadic clustering assumption .
the estimator is robust to the regression model being misspecified for the conditional mean .
such robustness is important because regression analysis typically relies on linear ( in coefficients ) approximations of unknown conditional mean functions @xcite .
of course our analysis in no way excuses analysts from their responsibility to obtain as good an approximation as possible nor does a robust fix for standard errors also solve the problem of finding a good approximation . given a reasonable approximation for the conditional mean , the methods we have developed here allow for more accurate asymptotic statistical significance tests and confidence intervals .
the estimator is consistent in the number of units that form the basis of dyadic pairs .
simulations show that the estimator approaches the true standard error with modestly - sized samples from a reasonable data generating process .
applications show that inferences can be seriously anti - conservative if one fails to account for the dyadic clustering .
this estimator is a natural complement to the non - parametric and semi - parametric regression analyses that are increasingly common in the social sciences @xcite .
given that we can express the estimator as the sum of simpler and easy - to - compute robust variance estimators , it could be applied to any estimator for which a cluster - robust option is available .
a cost of the robust approach is efficiency .
our simulations show that a two - way ( for cross - sectional data ) random effects model can be considerably more efficient and provide reliable inference when the conditional mean is correctly specified .
our robust estimator could provide the basis of a test for misspecification @xcite .
if there is little evidence of misspecification , the random effects estimator would be a reasonable choice given its efficiency .
however , problems of computational non - convergence may make the random effects estimator infeasible we encountered this in our application when we tried to use a three - way random effects model for dyadic panel data .
of course , accounting for dynamic dyadic clustering may fail to fully account for all relevant dependencies in the data .
for example , units may exhibit higher - order network effects : for example a shock to unit @xmath37 may affect unit @xmath91 through connections that run via a third unit , @xmath38 . in such cases , the methods developed here will likely yield standard errors that are too small , although they should still outperform methods that fail even to account for dyadic clustering .
we have @xmath92^{-1 } \frac{\sqrt{n}}{d } \sum_{d=1}^d x_d \epsilon_d.\ ] ] take @xmath93^{-1}$ ] first . by continuity of the inverse , the finite support of @xmath27 and @xmath40 , @xmath29 full rank , and the fact that the sum of dependent terms in this sum is @xmath94 ( * ? ? ?
2.8.4 ) , we have that @xmath95^{-1 } \overset{p}{\rightarrow } \e[x_d x_d']^{-1}.\ ] ] by slutsky s theorem , @xmath32 has the same asymptotic distribution as @xmath96^{-1}\frac{\sqrt{n}}{d}\sum_{d=1}^d x_d\epsilon_d$ ] , which has mean zero and variance , @xmath33^{-1}\var\left[\sum_{d=1}^d x_d\epsilon_d\right]\e[x_dx_d']^{-1}. \label{varb}\ ] ] then , @xmath97 & = \sum_{d=1}^d \sum_{d'=1}^d \cov[x_d\epsilon_d , x_{d'}\epsilon_{d'}]\nonumber \\ & = \sum_{d=1}^d\sum_{d'=1}^d \e\{x_{d}x'_{d ' } \cov[\epsilon_d,\epsilon_{d'}|{\mathbf{x}}]\}\nonumber \\ & = \sum_{d=1}^d \left ( \underbrace{\e\{x_{d}x'_d \var[\epsilon_d|{\mathbf{x}}]\}}_{a } + \underbrace{\sum_{d ' \in \mathcal{s}(d ) } \e\{x_{d}x'_{d ' } \cov[\epsilon_d,\epsilon_{d'}|{\mathbf{x}}]\}}_{b } \right ) \label{varxe}\end{aligned}\ ] ] where @xmath35 , the set of dyads other than @xmath36 that share a member from @xmath36 .
by chebychev s inequality , two conditions are sufficient for @xmath98 as @xmath31 : ( i ) @xmath99 \rightarrow 0 $ ] , and ( ii ) @xmath100 = \var[n\hat v ] \rightarrow 0 $ ] ( * ? ? ?
. for ( i ) , consider the interior ( `` meat '' ) of @xmath42 : @xmath101 and the corresponding term for @xmath102 : @xmath103\ } + \sum_{d ' \in \mathcal{s}(d ) } \e\{x_{d}x'_{d ' } \cov[\epsilon_d,\epsilon_{d'}|{\mathbf{x}}]\ } \right).\ ] ] we have @xmath104 = \sum_{d=1}^d \left ( \e\{x_{d}x'_d \e [ e_d^2|{\mathbf{x}}]\ } + \sum_{d ' \in \mathcal{s}(d ) } \e \{x_{d}x'_{d ' } \e[e_d e_{d'}|{\mathbf{x}}]\}\right ) \rightarrow \sigma\ ] ] by consistency of @xmath30 , an implication of lemma [ lemma - var ] .
lemma [ lemma - var ] also established that @xmath105^{-1}$ ] , while boundedness of @xmath27 implies uniform boundedness of @xmath96^{-1}$ ] . taking these elements together
, uniform continuity implies @xmath99 \rightarrow 0 $ ] ( cf .
@xcite , thm .
1 ) . for ( ii )
, we follow the strategy of @xcite , which requires showing that growth of covariance contributions due to dependent units is bounded in @xmath43 .
begin by rewriting @xmath106 as @xmath107 where @xmath108 is an indicator function denoting a pair of dyads @xmath36 and @xmath109 that share a member , as defined above .
we have established that @xmath110 converges in probability to a finite limit , @xmath111^{-1}$ ] , so we can treat this as @xmath112 . given bounded support , @xmath113 $ ] is of the same order as @xmath114 \nonumber \\ & = n^2 d^{-4 } \sum_{d } \sum_{d ' } \sum_{d '' } \sum_{d ' '' } \cov(i_{d , d'}x_{d}x'_{d ' } e_de_{d ' } , i_{d'',d'''}x_{d''}x'_{d ' '' } e_{d''}e_{d ' '' } ) .
\nonumber\end{aligned}\ ] ] to show @xmath115 \rightarrow 0 $ ] , it is sufficient to establish that @xmath116 is at most @xmath117 . consider quadruples of dyads indexed by @xmath118 such that @xmath119 .
suppose there are @xmath120 quadruples and define @xmath121 .
the covariance terms take positive values when @xmath122 , which occurs at rate @xmath123 ( that is , when we have pairs of terms that are actually summed into @xmath42 ) , and we have dependence across @xmath124 and @xmath125 .
then , the six distinct ways that such dependence can occur and the associated proportion of quadruples for which these occur for a sample of size @xmath43 are as follows : @xmath126 therefore the proportion of quadruples yielding a positive covariance is the proportion of cases contributing non - zero values to the sum @xmath116 .
this proportion is bounded by the proportion of cases where the indicator functions are both one and the order of the proportion of dependent quadruples from across the six cases characterized above , that is @xmath127 .
then @xmath115 $ ] is at most @xmath128 .
consider the interior ( `` meat '' ) of @xmath88 : @xmath129 taking the second component , @xmath130 define @xmath131 then working with @xmath132 , @xmath133 , and @xmath134 as defined in the proposition , @xmath135 then @xmath136 \nonumber \\ & = \sum_{i=1}^n \hat \sigma_{c , i } - \hat \sigma_d - ( n-2 ) \hat \sigma_0 , \nonumber \end{aligned}\ ] ] and by linear distributivity for @xmath137 @xmath138 with @xmath63 , @xmath64 , and @xmath65 as defined in the proposition .
weighting is a common way to adjust for unequal probability sampling of dyadic interactions , among other applications .
the extension to the weighted case is straightforward .
assume weighted directed dyad observations with weights finite and fixed ; denote the weight for dyad @xmath36 as @xmath139 .
define the sample data as @xmath140 and @xmath15 and the matrix of weights as @xmath141 .
then the weighted least squares estimator is @xmath142 for the weighted dyads case , assume the same conditions as in proposition [ prop1 ] and consider the following variance estimator @xmath143 then as @xmath31 , @xmath144 - v_w \overset{p}{\rightarrow } 0,\ ] ] and @xmath145 where @xmath146^{-1 } \e\left ( \e\left[x_dx'_d w_{d}^{2}\var[\epsilon_d|{\mathbf{x}}]\right ] + \sum_{d ' \in \mathcal{s}(d)}\e\left[x_dx'_{d'}w_{d}w_{d'}\cov[\epsilon_d , \epsilon_{d'}|{\mathbf{x}}]\right]\right ) \e[w_{d}x_dx_d']^{-1}.\ ] ]
an implementation for generalized linear models follows the usual @xmath147-estimation results ( @xcite ; @xcite , ch .
12 ) . given an estimating equation , @xmath148 , on data @xmath149 , and with parameters @xmath150 and parameter estimates
@xmath151 , the sandwich approximation for the variance is @xmath152 , where @xmath153 \text{\hspace{1em } and \hspace{1em}}{\mathbf{b}}= \e \psi(d;\hat \theta)\psi(d_i;\hat \theta)'.\ ] ] for logistic regression the estimating equation is , @xmath154 where @xmath155 , the predicted probability for dyad @xmath36 .
the plug - in variance estimator for logistic regression coefficients is given by @xmath156 where @xmath157 the extensions to repeated and weighted observations follow analogously .
xx angrist , joshua d. guido w. imbens .
`` comment on ` covariance adjustment in randomized experiments and observational studies ' by paul r. rosenbaum . ''
_ statistical science _ 17(3):304307 .
angrist , joshua d. jorn - steffen pischke .
princeton , nj : princeton university press .
arellano , manuel .
1987 . `` computing robust standard errors for within - group estimators . '' _ oxford bulletin of economics and statistics _ 49(4):431434 .
beck , nathanial , kristian skrede gleditsch kyle beardsley . 2006 .
`` space is more than geography : using spatial ecometrics in the study of political economy . '' _ international studies quarterly _ 50:2744 .
buja , andreas , richard berk , lawrence brown , edward george , emil pitkin , mikhail traskin , linda zhao kai zhang .
`` models as approximations : a conspiracy of random predictors and model violations against classical inference in regression . ''
manuscript , the wharton school , university of pennsylvania , philadelphia .
cameron , a. colin , jonah b. gelbach douglas l. miller .
`` robust inference with multi - way clustering . '' _ journal of business and economic statistics _ 29(2):238249 .
chamberlain , gary .
`` multivariate regression models for panel data . ''
_ journal of econometrics _ 18(1):546 .
conley , timothy g. 1999 .
`` gmm estimation with cross sectional dependence . '' _ journal of econometrics _ 92:145 .
davidson , russell james g. mackinnon .
oxford : oxford university press .
erikson , r. s. , p. m. pinto k. t. rader .
2014 . `` dyadic analysis in international relations : a cautionary tale . ''
_ political analysis _ 22(4):457463 .
fafchamps , marcel flore gubert .
`` the formation of risk sharing networks . '' _ journal of development economics _ 83:326350 .
fisman , raymond , sheena s iyengar , emik kamenica itamar simonson .
`` gender differences in mate selection : evidence from a speed dating experiment . '' _ quarterly journal of economics _ 121:673697 .
gelman , andrew jennifer hill .
cambridge : cambridge university press .
goldberger , arthur s. 1991 . .
cambridge , ma : harvard university press .
green , donald p. , soo yeon kim david h. yoon .
2001 . `` dirty pool . '' _ international organization _ 55(2):441468 .
greene , william h. 2008 . .
upper saddle river , nj : pearson .
hansen , christian b. 2007 .
`` asymptotic properties of a robust variance matrix estimator for panel data when _
t _ is large . ''
_ journal of econometrics _ 141:597620 .
hoff , peter d. 2005 . ``
bilinear mixed - effects models for dyadic data . ''
_ journal of the american statistical association _ 100(469):286295 .
hubbard , alan e. , jennifer ahern , nancy l. fliescher , mark van der laan , sheri a. lippman , nicholas jewell , tim bruckner william a. satariano .
`` to gee or not to gee : comparing population average and mixed models for estimating the associations between neighborhood risk factors and health . '' _ epidemiology _ 21(4):467474 .
huber , peter j. 1967 .
the behavior of maximum likelihood estimates under nonstandard conditions . in _ proceedings of the fifth berkeley symposium on mathematical statistics and probability_. vol . 1 pp . 221233
. kenny , david a. , deborah a. kashy william l. cook .
new york , ny : guilford press .
king , gary margaret e. roberts .
`` how robust standard errors expose methodological problems they do not fix , and what to do about it . ''
_ political analysis _
( online early view):112 .
lehmann , erich l. 1999 . .
new york : springer - verlag .
liang , kung - yee scott l. zeger .
`` longitudinal data analysis using generalized linear models . '' _ biometrika _ 73(1):1322 .
lin , winston .
`` agnostic notes on regression adjustments to experimental data : reexamining freedman s critique . '' _ annals of applied statistics _ 7(1):295318 . , james g. halbert white .
`` some heteroskedasticity - consistent covariance matrix estimators with improved finite sample properties . ''
_ journal of econometrics _ 29(3):305325 .
moulton , brent r. 1986 .
`` random group effects and the precision of regression estimates . '' _ journal of econometrics _ 32:385397 .
neumayer , eric thomas pluemper . 2010 .
`` spatial effects in dyadic data . '' _ international organization _ 64(1):145165 .
russett , bruce m. john r. oneal .
new york , ny : norton .
stefanski , leonard a. dennis d. boos .
2002 . `` the calculus of m - estimation . ''
_ the american statistician _ 56(1):2938 .
stock , james h. mark w. watson .
`` heteroskedasticity - robust standard errors for fixed effects panel data regression . '' _ econometrica _ 76(1):155174 .
white , halbert .
1980_a_. `` a heteroskedasticity - consistent covariance matrix estimator and a direct test for heteroskedasticity . '' _ econometrica _ 48(4):817838 .
white , halbert .
1980_b_. `` using least squares to approximate unknown regression functions . '' _ international economic review _
21(1):149170 .
white , halbert .
consequences and detection of misspecified nonlinear regression models . ''
_ journal of the american statistical association _ 76(374):419433 .
white , halbert .
`` maximum likelihood estimation of misspecified models . '' _ econometrica _ 50(1 - 25 ) .
white , halbert . 1984 . .
new york , ny : academic press .
wooldridge , jeffrey m. 2010 . .
cambridge , ma : mit press .
zorn , christopher .
`` generalized estimating equation models for correlated data : a review with applications . ''
_ american journal of political science _ 45:470490
. cluster robust variance estimation for dyadic data + supporting information not for publication
figure [ fig : norm ] and [ fig : miss ] show additional simulation results where we make comparisons to mixed effects models specifically , to a two - way random effects model that incorporates random effects for each of @xmath43 units in the data .
all random effects models were fit using the lmer package in r. in both sets of graphs , black box plots show the distribution of standard error estimates for ols coefficients using the proposed dyadic cluster robust estimator .
gray box plots show standard error estimates for ols coefficients from a heteroskedasticity robust estimator .
blue box plots show the distribution of standard error estimators from a two - way random effects model with a random effect for each of @xmath43 units .
black diamonds show the true standard error of the ols coefficients and the blue diamonds show the true standard error of the random effects coefficients .
figure [ fig : norm ] uses the same data generating process as was described in the main text , and presents the same results for ols with either the dyadic cluster robust standard error estimator or the heteroskedastic robust estimator . in addition , we have added random effects estimates . given that the data generating process conforms exactly to the assumptions of a two - way random effects model , we obtain the straightforward result that the random effects estimates are both more efficient ( the blue diamonds tend to be lower than the black ones ) and the random effects standard errors are consistent . figure [ fig : miss ] illustrates the robustness of the dyadic cluster robust estimator relative to a random effects estimator .
in this case , we induce mild misspecification by assuming that the true data generating process is @xmath158 where all parameters are set as in the main text but we also have @xmath159 .
we nonetheless fit a model that includes only @xmath160 , thereby mildly misspecifying the model such that we do not account for the nonlinearity induced by the @xmath161 term .
figure [ fig : miss - scatter ] shows the scatter plot of @xmath70 against @xmath160 , the linear approximation ( in red ) for one of the simulation replicates ( @xmath162 ) , and the true conditional mean ( in blue ) .
such linear approximations are the norm in social science research @xcite . in this case
, we see that , despite the mild misspecification , the dyadic cluster robust standard error correctly characterizes the standard error of the linear approximation coefficients ( the black box plots still zero in on the black diamonds ) .
the random effects model is still more efficient ( the blue diamonds still tend to be below the black diamonds ) .
but the random effects standard error estimator ( as computed in the lmer package for r ) is inconsistent and anti - conservative : the blue box plots converge to a limit that is substantially below the true standard error ( the blue diamonds ) .
our second application is based on @xcite s seminal study of the determinants of mate selection in a `` speed - dating '' experiment .
we replicate their primary analysis for female participants ( as reported in column 1 of their table iii ) using data from 21 dating sessions conducted by the experimenters . ]
these data include 278 women paired into 3,457 female - male dyads .
@xcite regress a binary indicator for whether the female subject desired contact information for a male partner on the subject s ratings of the partner s ambition , attractiveness , and intelligence based on a 10-point likert scale .
the regression controls for female - subject fixed effects and weights observations contributed by each female subject by the inverse of the number of partners with which she was paired .
table [ tab : fisman ] shows the results of our re - analysis .
, our estimates differ slightly from those that appear in the original paper . ] again , we see that accounting for the dyadic clustering ( column 4 ) yields standard error estimates that are larger than what one would get if one ignored all clustering ( column 2 ) , although in this case there are no pronounced differences with what one gets when one only accounts for within - subject clustering ( column 3 ) .
this is to be expected given that the amount of dyadic dependence in this dataset is limited : dyads were formed only _ within _ sessions that included between 5 and 22 male partners for each female subject . in addition , the female subjects are never paired with each other .
the only clustering that occurs is therefore within female subjects and then for male partners that appear across multiple female subjects .
the former is addressed in part by the female subject - specific fixed effects .
the latter is limited by the within - session pairings . & & het . & naive cluster & dyadic cluster + predictor of mate selection & coef . & robust s.e . & robust s.e . & robust s.e .
+ ambition & 0.0192 & 0.0052 & 0.0057 & 0.0061 + attractiveness & 0.1157 & 0.0041 & 0.0051 & 0.0054 + intelligence & 0.0465 & 0.0062 & 0.0073 & 0.0074 + + | dyadic data are common in the social sciences , although inference for such settings involves accounting for a complex clustering structure .
many analyses in the social sciences fail to account for the fact that multiple dyads share a member , and that errors are thus likely correlated across these dyads .
we propose a nonparametric , sandwich - type robust variance estimator for linear regression to account for such clustering in dyadic data .
we enumerate conditions for estimator consistency .
we also extend our results to repeated and weighted observations , including directed dyads and longitudinal data , and provide an implementation for generalized linear models such as logistic regression .
we examine empirical performance with simulations and an application to interstate disputes .
cluster - robust variance estimation for dyadic data keywords : cluster robust variance estimation , dyadic data , agnostic regression + + word count : 5,022 ( including main text , appendix for print , captions , and references ) . |
the inelastic hard spheres model @xcite without energy input , initially prepared in a homogeneous state , exhibits , after a rapid transient , a regime characterized by homogeneous density and a probability distribution of velocities that depends on time only through the total kinetic energy ( global granular temperature @xmath0 ) , i.e. a scaling velocity distribution .
this is the so called _ homogeneous cooling state _ ( hcs ) or haff regime @xcite .
it has been shown by several authors @xcite that this state is unstable with respect to shear and clustering instabilities : structures can form that seem to minimize dissipation , mainly in the form of velocity vortices and high density clusters .
these instabilities grow on different space and time scales , so that one can investigate them separately .
several theories have been proposed to take into account the emergence of structures in granular gases .
some of these are more fundamental , because derive the correlation functions directly from the kinetic equations @xcite ; others , that assume the validity of the hydrodynamic description , deserve the name of mesoscopic theories @xcite ; others are phenomenological theories that suggest analogies with burgers equation @xcite or spinodal decomposition models @xcite or mode coupling theories @xcite
. some of these theories can describe the behavior of the cooling granular gas far deeply into the correlated regime , giving predictions for the asymptotic decay of energy . in this paper , after a brief summary of the results of these theories , we show how the departure from the hcs can be well modeled by a class of models obtained placing on a lattice the original homogeneous inelastic maxwell model .
these models have the disadvantages of being conceived under the assumption of negligible density fluctuations and therefore can be useful only in the description of the first instability , i.e. the growth of velocity fluctuations . in section 2 we review the hcs , its instabilities and the existing theories . in section 3
we briefly discuss the inelastic maxwell model @xcite , which is a starting point for the introduction of the inelastic lattice maxwell models ( ilmm ) . in section 4 and in section 5 the analysis and results of the ilmm in one and two dimensions
are reviewed @xcite . finally in section 6 conclusions are drawn .
a cooling granular gas in @xmath1 dimensions is defined as an ensemble of @xmath2 grains , i.e. hard objects ( rods if @xmath3 , disks if @xmath4 , spheres if @xmath5 ) of linear size ( diameter ) @xmath6 , placed in a volume @xmath7 with periodic boundary conditions .
the grains evolve freely and interact with each other through instantaneous binary inelastic collisions . the rule that gives the velocities after the collision as functions of the velocities before the collision is the definition of the particular granular gas model . in this case
we use the model with constant restitution coefficient without rotational degrees of freedom and set the mass of the grains @xmath8 .
other models have been considered in the literature @xcite .
the collision rule between a particle with velocity @xmath9 and one with velocity @xmath10 for this model is : @xmath11\hat{\mathbf{n}}\ ] ] where the primed velocity is the post - collisional one , @xmath12 is the unit vector in the direction joining the centers of colliding particles , and @xmath13 is the restitution coefficient and takes values between @xmath14 and @xmath15 .
when @xmath16 the collision is elastic .
usually ( in numerical or real experiments ) granular gases are prepared in a homogeneous situation : uniform - random positions of grains , gaussian or uniform - random initial velocity with no preferred direction .
it happens that , if the system is large enough or the inelasticity large enough , the imposed homogeneity is broken after a certain time .
the more accepted scenario is a two time homogeneity breaking : at a time @xmath17 the velocity field becomes unstable to the formation of shear bands , then at a time @xmath18 the density field becomes unstable toward the formation of high density clusters . a granular gas evolving from a homogeneously random state loses memory of its initial condition after a time of the order of one collision per particle and rapidly enters the homogeneous cooling regime .
this regime is expected to be well described by the granular boltzmann equation ( see for example @xcite ) .
this means that in this regime the total probability density function ( p.d.f . ) at collision is factorisable in one - particle p.d.f.s : @xmath19 .
the kinetic definition of hcs is given by the homogeneity ansatz plus the scaling ansatz for the one particle distribution function : @xmath20 where @xmath21 and @xmath22 is the thermal velocity defined by @xmath23 with @xmath24 the granular temperature ; here we have assumed that @xmath25 and also that @xmath26 . if the eq .
is inserted in the boltzmann equation an equation for the temperature is obtained : @xmath27 where @xmath28 is the time dependent collision frequency , while @xmath29 is the time independent cooling rate .
these two functions can be approximated , using the maxwellian approximation @xmath30 , by @xmath31 and @xmath32 : @xmath33 in this case the solution of the temperature equation reads : @xmath34 where @xmath35 is the mean free time at the initial temperature @xmath36 and @xmath37 is the cumulated collision number obtained from the definition @xmath38 .
eq . is known as haff s law @xcite .
corrections to the constants appearing in eq .
stem from a more careful consideration of the hcs : when the volume fraction is non negligible the enskog - boltzmann equation should be employed instead of boltzmann equation .
this is identical to the boltzmann equation but for a multiplicative constant in the collision integral that takes into account static density correlations due to the fact that the gas is not perfectly dilute .
very recently it has been shown @xcite that there are also small velocity correlations which must be considered and that modify the molecular chaos hypothesis : with this correlations the law @xmath39 is still valid but more precise corrections to the constant @xmath40 appear .
the haff law can be also derived in the framework of granular hydrodynamics @xcite : in the hcs hydrodynamics can be considered valid as a consequence of homogeneity of the density and velocity field and of the slow temperature decay .
the evolution of the ( space uniform ) temperature field is exactly the same as in eq . .
kinetic arguments , linear stability analysis of hydrodynamic equations and numerical simulations lead to the evidence that the hcs is unstable .
in particular , it has been shown that large scale fluctuations of velocity and density can grow exponentially breaking the homogeneity of the system . the fact that the fluctuations arise at large scale means that , if the system is small enough , the hcs lasts forever .
moreover the minimum scale of formation of fluctuations depends upon inelasticities and is smaller at larger inelasticities : this means that for a quasi - elastic gas it is necessary to take a very large volume to observe the breaking of homogeneity . for shear ( @xmath41 ) , heat ( @xmath42 ) and sound ( @xmath43 ) modes versus @xmath44 for inelastic hard disks with @xmath45 at a packing fraction @xmath46 .
the dashed line indicate the imaginary parts of the sound modes . here
@xmath6 is the diameter of a particle .
( from orza et al .
@xcite)[apfig_disp_rel ] ] the linear stability analysis of hydrodynamics gives a good description of the instability of fluctuations .
the procedure is applied to the fields rescaled with respect to the homogeneous cooling state , i.e. @xmath47 ( being @xmath48 the average density ) , @xmath49 and @xmath50 , and therefore any instability or emerging structure is not absolute but relative to this decaying reference state .
moreover the instabilities are studied in @xmath51 space ( the space of fourier modes ) and , when @xmath52 , the velocity field is decomposed in a parallel component @xmath53 and in @xmath54 orthogonal components @xmath55 with respect to the vector @xmath51 . in figure [ apfig_disp_rel ]
we report the results of such an analysis , taken from the literature @xcite .
this analysis shows that the evolution of fluctuations of normal velocity components ( shear modes , @xmath56 ) are not coupled with any other fluctuating component . at the same time , the remaining components are coupled together .
the rate of decay / growth for the shear mode reads @xmath57 where @xmath58 depends on the transport coefficients appearing in the hydrodynamics .
we refer to @xcite for detailed calculations of this correlation length . at low values of @xmath59 ( in the so - called `` dissipative range '' )
also the heat mode is `` pure '' , as it is given by the longitudinal velocity mode @xmath60 only , with eigenvalue @xmath61 ; in this range the sound modes are combination of density and temperature fluctuations .
the most important result of this analysis is that @xmath62 and @xmath63 are _ positive _ below the threshold values @xmath64 and @xmath65 respectively , indicating two linearly unstable modes with exponential ( in @xmath66 , i.e. the time measured by the cumulated number of collisions per particle ) growth rates .
here we have used the notation @xmath67 .
the shear and heat instabilities are well separated at low inelasticity ( i.e. low @xmath68 ) , as @xmath69 while @xmath70 , so that @xmath71 .
it is also important to note that the linear total size @xmath72 of the system can suppress the various instability , because the minimum wave number @xmath73 can be larger than @xmath74 or even than @xmath75 .
when fluctuations grow , structures emerge . in molecular dynamics simulations
these structure appear as shocks in @xmath3 , or vortices and clusters in @xmath4 . in @xmath76 , a detailed analysis of structure factors ( from fluctuating hydrodynamics @xcite or ring kinetic theory @xcite )
has established that all the main correlation lengths ( which indicate the typical size of structures _ in the rescaled fields _ ) should grow as @xmath77 . in particular this
is expected for the size of vortices @xmath78 and the size of clusters @xmath79 : this again shows that , at small inelasticities @xmath68 , the size of clusters grows more slowly than that of vortices . after a brief transient during which the instabilities grow exponentially in @xmath66 with the rates given above , the description suggested by linear stability analysis ceases to be meaningful .
molecular dynamics ( md ) , in the absence of experiments ( the cooling granular gas being a reference model and not a real system ) , are the more reliable tools , but they are very cpu - time consuming : to obtain a decent statistics of the asymptotic dynamics of a granular gas , large enough to observe instabilities , md simulations must be carried on for days . to our knowledge a definitive clear picture of the asymptotics of this model is still lacking ( with month - lasting simulations isobe @xcite has obtained what he called the `` final state '' , which resembles 2d turbulence , but this was done in a quasi - elastic limit ) . on the other side
there are effective models that can grasp the essence of the phenomena with particular assumptions . here
we rapidly review some of these : * brito and ernst have shown by means of a mode - coupling @xcite theory that the asymptotic energy decay should follow a diffusive form @xmath80 . *
wakou et al .
@xcite have demonstrated that the evolution of the flow field of a granular fluid , neglecting the convective term @xmath81 , can be cast in the form of a time dependent ginzburg - landau equation for a non - conserved order parameter .
the energy functional has a _ continuous _ set of degenerate minima , having the shape of a mexican hat . when the convective term is added , only subsets of admissible solutions are selected out of this infinite set of minima . in two dimensions
only _ two _ distinct minima survive : therefore the @xmath4 cooling granular gas , during the formation of instabilities , greatly resembles the spinodal decomposition for a non - conserved order parameter ( model a universality class , in the hohenberg - halperin classification @xcite ) . *
ben - naim et al .
@xcite have studied the cooling granular gas in @xmath3 , discovering that , after the homogeneous ( haff ) phase , it asymptotically becomes independent of the value of inelasticity @xmath68 and maps onto the sticky gas model .
for such a sticky gas the asymptotic temperature decay is proportional to @xmath82 , when density clusters and velocity shocks form .
the velocity field of the sticky gas @xcite is described by the inviscid limit of the burgers equation @xcite .
the relation to the burgers equation is useful to establish an estimate of the tails of the asymptotic velocity distribution @xmath83 ( independent of @xmath13 ) , which reads @xmath84 .
the md simulations have revealed only slight deviations from the gaussian , and the authors have imputed the discrepancy from the expected behavior to the smallness of the constant in front of @xmath85 .
more recently it has been conjectured @xcite that the burgers equation describes the flow velocity @xmath86 of a cooling granular gas for arbitrary values of the inelasticity @xmath68 in generic @xmath1 dimensions .
this conjecture implies that the asymptotic behavior of the cooling inelastic gas is independent of @xmath68 ( as it always falls in the universality class of the sticky gas ) and that the upper critical dimension for the disappearance of the inelastic collapse is @xmath87 .
however md simulations in @xmath88 seem to show that the shear instability and the inelastic collapse do not disappear @xcite .
to better illustrate the analysis of the inelastic lattice maxwell models ( ilmm ) we start with a brief description of the homogeneous inelastic maxwell model ( imm ) .
the definition of the model is given in terms of its boltzmann equation , which , for scalar velocities ( @xmath3 ) , reads : @xmath89 with @xmath90 .
the imm can be introduced in different ways : ( a ) its boltzmann equation can be regarded as the master equation of a simple stochastic process inspired to a model put forward by s.ulam
@xcite : the process consists in the evolution of an ensemble of @xmath2 velocities , at each time step ( corresponding to @xmath91 , so that @xmath66 increases of @xmath15 after averagely one collision per particle ) two velocities are chosen at random with uniform distribution and are changed according to the inelastic collision rule ; ( b ) the associated boltzmann equation can be considered the inelastic generalization of the equation for elastic maxwell molecules @xcite ; ( c ) its boltzmann equation can be derived from the ordinary boltzmann equation for inelastic hard spheres assuming that the term @xmath92 in the collision integral can be approximated by @xmath93 , as suggested by bobylev et al . @xcite .
the recent surge of interest for this model has been triggered by the discovery of an exact solution @xcite for the @xmath3 case ( eq . ) and the appearance of computable power law tails in the solutions of @xmath52 cases .
@xcite . in a previous paper
ben - naim and krapivsky @xcite showed that eq .
could not have a scaling solution with finite moments : in fact , a scaling form @xmath94 with @xmath95 , imposes a temporal dependence of the higher moments ( if finite ) : [ apscaling - moments ] @xmath96 does not depend on time .
but they found that @xmath97 this result does not exclude the existence of a scaling solution but requires a solution with diverging moments of order @xmath98 , i.e. very large algebraic tails . in this case standard methods @xcite based on fourier transform fails due to the singularity in @xmath99 corresponding to such algebraic tails . versus @xmath100 for different values of @xmath13 from the simulation of the inelastic maxwell model in 1d ( left ) and 2d ( right ) .
in 1d the asymptotic distribution is independent of @xmath13 and collapse to the eq . .
in 1d , the chosen initial distribution ( exponential ) is drawn ( same result with uniform and gaussian initial distribution ) . in 2d the distributions still present power - law tails , but the power depends upon @xmath13 : for @xmath101 the pdf tends to a maxwell distribution .
data refers to more than @xmath102 particles.[apfig_maxwell ] ] remarkably , a direct inspection shows that the following velocity distribution @xmath103 ^ 2}\ ] ] is indeed a solution of the non - linear boltzmann equation , for every value of @xmath13 @xcite .
the solution has the property that the moments of order @xmath104 diverge , and its fourier transform has a singularity of the type @xmath105 in @xmath99 .
we believe that eq
. represents the asymptotic solution for a large class of initial distributions , for the following arguments : * as shown by ben naim and krapivsky @xcite , the dynamics of the moments for a generic starting distribution can be computed , giving the following limit @xmath106 for @xmath107 ; * secondly we performed numerical simulations of the bk model , collecting evidence of the convergence to the solution for several starting velocity distributions , namely uniform , exponential ( see figure [ apfig_maxwell ] , left frame ) or gaussian . for higher dimensions ,
the problem of a scaling solution has been recently addressed in @xcite .
assuming a solution with large algebraic tails , it is possible to compute the asymptotic exponent as a solution of a transcendental equation .
such equation depends explicitly on the restitution coefficient , indicating that the tail exponent varies with @xmath13 ( and correctly diverges for @xmath101 , corresponding to the recovering of the maxwell distribution for the elastic case ) .
a numerical observation of such algebraic tails in @xmath108 dimensions are shown in right frame of figure [ apfig_maxwell ] , where a dependence on @xmath13 is put in evidence .
a more precise comparison of the measured exponent and the theoretical predictions are shown in figure [ apfig_tails ] . in this figure
we show the results of extensive numerical simulations for the completely inelastic case ( @xmath109 ) of several models , in two and three dimensions . in fact ,
as noted in @xcite , the tail exponent is in general dependent on the details of the collisions , as for instance on the distribution of the impact parameter .
two different choices have been studied in @xmath110 dimensions , corresponding to the models denoted imm - a and imm - b in reference @xcite : in the first model grazing collisions are favored with respect to the second one .
it is interesting to notice that the analytical and numerical study confirm the existence of a scaling solution with algebraic tails , irrespectively of the collision details , but that the precise value of the exponent is not universal .
the relevance of such a scaling solution for @xmath52 has been addressed by bobylev _
et al . _
@xcite , who have proved the existence and the uniqueness of such an asymptotic solution ( giving details on the basin of attraction ) . versus @xmath111 for @xmath109 from the simulation of the inelastic maxwell model in 2d ( left ) and 3d ( right ) . in 2d
the asymptotic distribution tail converges to the exactly computed exponent ( model imm - a ) . in 3d
the convergence is slower , but the difference of the predicted exponent for model imm - a and imm - b can be appreciated . in the bottom panels
the convergences of the local logarithmic slope is put in evidence .
( for the precise definition of the two models see . )
data refers to more than @xmath102 particles.[apfig_tails ] ] interesting features emerge when the inelastic maxwell model is extended to treat grains having different physical properties , such as unequal masses , different restitution coefficients , radii , etc .
the study of a binary maxwell mixture with scalar velocities showed few unexpected properties @xcite : 1 .
the kinetic temperatures of the two species are different , i.e. there is no energy equipartition , but the the ratio @xmath112 reaches asymptotically a constant value .
such a feature agrees with the results obtained in the framework of the boltzmann - enskog transport equation by garz and dufty @xcite .
2 . the velocity distributions in general differ in shape .
the velocity distributions have power law tails .
these exponents can be calculated analytically , by a suitable generalization of a technique devised by ernst and brito @xcite .
the power law decay of cooling mixtures has been shown to vary between @xmath108 and @xmath113 .
the solution of the imm given in has nothing to do with the real dynamics of a cooling granular gas in the homogeneous regime , where no algebraic tails appear . instead , a good agreement ( in the hcs )
is found between the solution of boltzmann equation @xcite and molecular dynamics @xcite .
however we will show in the following that the inelastic maxwell model with the addiction of a special topology ( the 1d lattice ) is able to reproduce the homogeneous regime and some essential features of the first instability ( enhancement of velocity fluctuations ) .
an md simulation of @xmath2 hard rods on a ring of length @xmath72 shows that , besides hydrodynamic instabilities discussed above , also a kinetic singularity appears that prevents long simulation runs , the so - called `` inelastic collapse '' @xcite .
the inelastic collapse is the divergence of the collision rate of a small set of grains , analogous to the divergence of the collision rate of a ball left bouncing on the floor .
this means that a group of particles will experience an infinite number of collisions in a finite time . to avoid such a problem ,
many authors @xcite have proposed the introduction of a velocity cut - off @xmath114 such that , whenever two particles collide with a relative velocity smaller , in absolute value , than @xmath114 , the collision is elastic .
taking @xmath115 , being @xmath116 the minimum granular temperature that one expects to observe , one can be sure that the choice of @xmath114 will not influence the main results of the simulation . performing an efficient event driven simulation scheme ( based on a tree ordering of the event times )
it is possible to study easily very large systems , with @xmath117 for thousands of collisions per particle .
is the physical time , defined only in md simulations , while @xmath66 is the cumulated number of collisions per particle ) .
the haff law @xmath118 is verified for both systems .
the correlated regime presents a behavior @xmath82 independent of @xmath13 for the hard rods , while appears diffusive ( in collision units ) @xmath119 and @xmath13-dependent for the lattice model .
data refer to @xmath102 particles ( both models).[apfig_energia_1d ] ] particles ( both models ) with @xmath120 and @xmath121 ( for the lattice model in the inhomogeneous regime ) .
[ apfig_dist_1d ] ] in the left panel of figure [ apfig_energia_1d ] the decay of kinetic energy @xmath122 is shown .
it displays , for different values of @xmath13 , the same behavior with two regimes : in the hcs @xmath123 ( haff s law ) , then it decays differently , i.e. @xmath124 .
noticeably all the curves seem to collapse in the second regime , i.e. the asymptotic regime is equivalent ( not only in the exponent , but also in the coefficients ) for every value of @xmath13 .
the velocity pdf s are shown in figure [ apfig_dist_1d ] : in the hcs it appears the peculiar `` two - peaks '' form , which is also predicted by the study of the boltzmann equation in the quasi - elastic limit @xcite . in the second regime ( which can not be described by the boltzmann equation ,
because of the presence of correlations ) the pdf becomes a gaussian .
, middle @xmath125 ) and for the 1d lattice gas model ( bottom , @xmath126 ) . in the middle frame
we display the md profile against the particle label in order to compare the shocks and preshocks structures with the lattice gas model ( the dotted lines show how shocks and preshocks transform in the two representations for the md ) .
data refers to @xmath127 particles , @xmath120 ( both models ) .
[ apfig_shocks ] ] against @xmath128 for the 1d md and against @xmath129 for the 1d lattice gas model , in the inhomogeneous phase .
times are chosen so that the two systems have the same energies .
data refers to system with more than @xmath130 particles , @xmath121 ( both models)[apfig_sf_1d ] ] what can be said about correlations in the second regime ?
an inspection of the velocity profile @xmath131 vs. @xmath132 ( see top panel of figure [ apfig_shocks ] ) suggests the presence of shocks , i.e. asymmetric `` discontinuities '' of the velocity profile in correspondence with peaks of density ( clusters ) . a more quantitative information about the velocity field is given by the study of the structure function @xmath133 where @xmath134 is the fourier transform of the field @xmath125 , presented in the left frame of figure [ apfig_sf_1d ] .
a good collapse is obtained if @xmath135 is plotted at different times @xmath136 versus @xmath137 .
the presence of short scale defects ( shocks ) is put in evidence by the form of structure factor at high values of @xmath59 , where @xmath138 : this tail of the structure function is expected , in @xmath3 , in the coarsening of a two - phases system , due to the non - analyticity ( at short distances ) of the spatial correlation function .
this is also known as porod law in the theory of phase ordering kinetics @xcite .
the inelastic lattice maxwell model in @xmath3 is defined as follows .
a set of @xmath2 scalar velocities are placed on @xmath2 sites on a ring .
after a time @xmath91 a couple of _ neighbors _ is chosen and is updated with the usual inelastic collision rule .
the only constraint in the choice of the couple is that @xmath139 : this is called the _ kinematic constraint _ and represents the physical condition necessary for a collision .
it must be noted that the choice of colliding particles does not depend on the modulus of the relative velocity : this is why we consider this a `` inelastic maxwell model '' embedded in a @xmath140 lattice . in this model
the only measure of time is @xmath66 , i.e. the cumulated number of collisions per particle .
this makes many measures difficult to be compared with analogous md measurements : in fact in md simulations @xmath141 depends upon the choice of @xmath114 ( the elastic cut - off ) , while all observables depend on @xmath136 independently of @xmath114 .
the analysis of the ilmm follows closely that of md simulations and the results @xcite are presented , in parallel , in the same figures [ apfig_energia_1d ] , [ apfig_dist_1d ] , [ apfig_shocks ] , [ apfig_sf_1d ] .
the study of the decay of energy ( figure [ apfig_energia_1d ] ) indicates that the model well reproduces the haff regime ( @xmath142 ) and displays a decay in the second regime of the form @xmath143 .
the velocity pdf s reproduce exactly the ones measured in md simulations ( see fig .
[ apfig_dist_1d ] ) , showing the `` two - peaks '' form in the haff stage and the gaussian in the late stage .
it must be stressed again that these features are obtained simply applying to the inelastic maxwell model ( which displays tails @xmath144 ) a two - neighbors ( instead of @xmath2-neighbors ) topology .
the inspection of velocity profiles must be carefully carried : in this model the particles can not move and therefore a comparison with md should be made considering @xmath131 vs. @xmath145 instead of @xmath132 . the good agreement between md and ilmm
can be appreciated in the two bottom panels of figure [ apfig_shocks ] : with this choice of abscissa , the shocks observed in md simulations appear reversed and smoothed , very similar to the ones observed in the ilmm .
analogously the structure factor presents similar features with md , mainly a @xmath146 behavior at high values of @xmath59 ( short scales ) which is a signature of topological defects . in the lattice case
the collapse can be obtained plotting @xmath147 against @xmath129 , i.e. @xmath148 as in the md case .
we have also verified that , removing the kinematic constraint , this structure disappears and the model behaves as governed by a simple diffusion equation .
the number of interesting and realistic results obtained with the one dimensional inelastic lattice maxwell model , in spite of its extreme simplicity , convinced us to extend our study to two dimensions .
however , a smaller number of results is known from molecular dynamics in @xmath4 , especially in the correlated regime , which requires very long simulations .
therefore the comparison between md and the ilmm to the case @xmath4 is more difficult . only recently md simulations start to give reliable measurements even for long times , and , where possible , the comparison with our ilmm is promising . more speculatively , we tried a series of less standard measurements with two different aims : the first is to try to understand what kind of effective equation could describe the velocity field of the ilmm ; the second is to discuss some basic hypothesis necessary to a hydrodynamic description of this granular fluid . after a brief review of the known results from md , we present the results obtained with our 2d ilmm .
first the same analysis of energy decay , velocity pdf and structure factors is carried on , as in the 1d case .
moreover , to better capture the nature of the topological defects generated by the dynamics , we consider the pdf of the velocity increments
. then we present an analysis of temporal correlations , mainly aimed to identify an effective equation for the velocity field .
finally we suggest a measurement of a hydrodynamic temperature and we sketch a discussion about the very presence of scale separation . to our knowledge few md simulations
have inspected the behavior of inelastic hard disks in the late stage , i.e. after the end of haff regime and the growth of instabilities .
the departure from the homogeneity has been studied by ernst and co - workers @xcite and their md simulations have shown results compatible with their predicted @xmath149 scaling law for the linear size of structures ( vortices and clusters ) . a figure in a work of huthmann et al .
@xcite was the first suggestion that the global velocity pdf asymptotically returns to a gaussian ( while it has fatter tails in the hcs ) and recently different works on large md simulations have confirmed this @xcite .
recently this `` return - to - the - gaussian '' scenario was put in strict relation with the presence of a burgers - like dynamics @xcite .
the debate on the asymptotic energy seems to be still open , even if there are strong indications of a universal ( independent of @xmath13 ) decay of the form @xmath150
a recent work of isobe @xcite demonstrates that many features of the `` last '' stage resemble the dynamics of 2d turbulence .
the ilmm in @xmath4 is identical to that in @xmath3 except for the fact that the particles have a @xmath108-components vector velocity @xmath151 and that they are placed on a triangular @xmath152 lattice , i.e. every particle has @xmath153 neighbors .
the dynamics is the same as in @xmath3 ; the kinematic constraint here reads @xmath154 . and @xmath155 ( @xmath156 sites ) .
the bold dashed line @xmath157 is a guide to the eye for the asymptotic energy decay , while the light dashed line is the exponential fit corresponding to the haff law @xmath158 .
the haff regime is too short to be observed in the system with @xmath155 .
the two indicated times @xmath159 and @xmath160 correspond to the plots of fig .
[ apfig_tempint_2d ] .
[ apfig_energia_2d ] ] the decay of energy @xmath161 is shown in figure [ apfig_energia_2d ] for two different values of @xmath13
. the haff regime @xmath162 is well reproduced ( when @xmath155 it terminates too early to be appreciated ) .
the asymptotic decay reads @xmath163 .
both this measurement and the decay observed in the @xmath3 case , are compatible with a universal law @xmath164 as for a purely diffusive field . and
the distributions are rescaled in order to have unit variance .
the initial distribution is a gaussian . the distribution becomes broader in the uncorrelated phase ( first regime ) , and then turns back toward a gaussian .
[ apfig_vdist_2d ] ] the rescaled velocity pdf s ( see figure [ apfig_vdist_2d ] ) , which are initially gaussian , become non - gaussian in the haff regime , with larger tails , but return gaussian in the late stage .
the analysis that follows shows that , in the late stage , the velocity field is far from being homogeneous : the global velocity pdf is strongly influenced by this homogeneity and its gaussian form could be simply the signature of the presence of many independent domains ( i.e. it is just the distribution of the average velocities of these domains ) . in the figures [ apfig_vorticia_2d ] and [ apfig_vorticib_2d ] we show the presence of vortices in the velocity fields .
the second figure represents the same system at a later time , demonstrating the coarsening of vortices ( their number reduces and their size increases ) . ) and longitudinal ( @xmath166 ) structure functions for @xmath167 and @xmath45 ( system size @xmath156 sites , times ranging from @xmath168 to @xmath169 ) .
the wave - number @xmath59 has been multiplied by @xmath170 .
notice the presence of the plateaux for the more elastic system . for comparison
we have drawn the laws @xmath171 and @xmath172 .
[ apfig_sf_2d ] ] in @xmath4 the structure function is defined as @xmath173 where the superscripts @xmath174 indicate the transverse and longitudinal components of the field with respect to the wave vector @xmath175 and the sum @xmath176 is over a circular shell of radius @xmath59 .
the analysis of the structure factor , shown in figure [ apfig_sf_2d ] , is performed for two systems at different inelasticities .
the common feature is a good collapse of different time curves if plotted against @xmath129 which is the signature of a domain growth ( diameter of the vortices ) @xmath177 .
the elasticity however changes dramatically the other features : the more elastic system has a gaussian structure factor at large scale ( low values of @xmath59 ) and a plateau at small scales , i.e. quasi - elastic collisions randomize neighbour velocities while the effect of inelasticity induces a structure at large scale .
the less elastic system does not present such a plateau , but a decay @xmath178 at small scales which is analogous to the porod law @xcite expected from the presence of defects in a phase ordering process .
the difference between transversal and longitudinal component ( more appreciable in the quasi - elastic system ) is coherent with the theoretical analysis discussed in the introduction , which predicts an earlier and stronger correlation for the transversal compoenent . .
the inset shows the gaussian shape measured for @xmath179 ( larger than @xmath180 for this simulation : @xmath155 , @xmath181 , system size @xmath182 ) .
[ apfig_vgrad_2d ] ] in the @xmath4 ilmm we have also observed shock - like phenomena .
the inspection of the distribution of longitudinal and transversal velocity differences , defined as @xmath183 is contained in figure [ apfig_vgrad_2d ] .
the main graph shows the case @xmath184 , i.e. the distributions of gradients .
both the longitudinal and transverval components are distributed with non - gaussian tails , but the longitudinal component presents also a strong asymmetry : this means that preferentially there are strong velocity differences in the direction of growing longitudinal velocity .
this is compatible with the analysis of shocks in the bottom panels of figure [ apfig_shocks ] in the @xmath3 case .
when @xmath185 the distribution of velocity differences is a gaussian . for different values of the waiting time @xmath186 and @xmath45 ( @xmath156 sites ) .
the graph on the left shows the convergence to the @xmath187 diffusive scaling regime , for large @xmath186 . for small @xmath186 ,
a local minimum is visible ( for such a quasi elastic dynamics ) . in the graph on the right
the same data are plotted vs @xmath188 : note that the small @xmath186 curves tend to collapse . for higher
@xmath186 the position of the local minimum does not move sensibly , but its value grows and goes to @xmath15 for large @xmath186 [ apfig_ac_2d ] ] to better study the dynamics of the system we have calculated the two - times self - correlation of the velocity components @xmath189 during a short time transient , the self - correlation function of our model depends on @xmath190 , i.e. it is time translational invariant ( _ tti _ ) .
later , @xmath191 reaches an `` aging '' regime and depends only on the ratio @xmath192 .
this _ tti _
transient regime is similar to what occurs during the coarsening process of a quenched magnetic system : the self - correlation of the local magnetization @xmath193 for @xmath194 shows a _ tti _ decay toward a constant value @xmath195 that is the square of the equilibrium magnetization , indicating that the local magnetization is evolving in an ergodic - like fashion .
thereafter the self - correlation decays with the _ aging _ scaling law indicated above .
obviously , when @xmath196 , the _ tti _ transient regime disappears . in our model the behavior of the self - correlation is even more subtle , as the cooling process imposes a ( slowly ) decreasing `` equilibrium '' temperature @xmath196 : this progressively erodes the _ tti _ regime and better resembles a finite rate quench .
the same dependence on the _ tti _ manifests itself in the angular auto - correlation , shown in figure [ apfig_ac_2d ] : @xmath197 the non - monotonic behavior of @xmath198 suggests that the initial direction of the velocity induces a change in the velocities of the surrounding particles , which in turn generates , through a sequence of correlated collisions , a kind of retarded field oriented as the initial velocity .
as @xmath199 increases the maximum is less and less pronounced . the behavior of the energy decay ( @xmath164 ) , the law of growth of domains ( @xmath177 ) and the `` aging '' form of the two - times self - correlations , as well as the gaussian velocity pdf s and the gaussian form at large scale in the structure factors , are strong evidence that the model behaves similarly to a system governed by the diffusion equation : @xmath200 where @xmath201 is a suitable field depending on the velocity field ( it could be a scalar , for example the @xmath202-component of the vorticity field @xmath203 ) . , for two different values of the restitution coefficient .
the power - law regime corresponds to the correlated regime .
the measured exponent @xmath204 is , with this precision , equal to the persistence exponent numerically measured in the diffusive dynamics.[apfig_pers_2d ] ] to perform a further check of this conjecture , we have measured the persistence exponent @xcite of the model , which is very sensitive to the class of universality .
we have counted the number @xmath205 of sites where the @xmath206 velocity component never changed sign from the starting time of the dynamics up to time @xmath66 . in the correlated regime
we observe @xmath207 with @xmath204 which is equal , up to this precision , to the persistence exponent numerically measured for the diffusive dynamics . ,
defined as function of the coarse graining size @xmath6 for @xmath159 ( incoherent regime ) and @xmath160 ( correlated regime for both choices of @xmath13 ) : the system is the same of figure [ apfig_energia_2d ] . in the correlated regime
the more elastic case presents a plateau at intermediate wavelengths , indicating a well defined _
mesoscopic _ temperature , and therefore a clear separation between the microscopic and the macroscopic scales.[apfig_tempint_2d ] ] in the spirit of verifying the possibility of a hydrodynamic ( mesoscopic ) description of this model , we have also analysed the average local granular temperature @xmath208 , defined as : @xmath209 where @xmath210 means an average on a region of linear size @xmath6 . if we call @xmath180 a characteristic correlation length of the system , since when @xmath211 the local average tends to the global ( zero ) momentum , then @xmath212 .
for @xmath213 , instead , @xmath214 .
the behavior of @xmath215 in the uncorrelated ( haff ) regime and in the correlated ( asymptotic ) regime for two different values of @xmath13 is presented in figure [ apfig_tempint_2d ] . for quasi elastic systems
@xmath215 exhibits a plateau for @xmath216 that identifies the strength of the internal noise ( see also the plateau in the structure factor , figure [ apfig_sf_2d ] ) and indicates the mesoscopic scale necessary for a hydrodynamics description . the local temperature ceases to be well defined for smaller @xmath13 , revealing an absence of scale separation between microscopic and macroscopic fluctuations in the strongly inelastic regime @xcite .
we have analysed essentially three different problems with a strong interconnection , in the framework of the physics of cooling granular gases .
our first step ( imm ) was to demonstrate that a simplified 1d version of the boltzmann equation for inelastic gases can be exactly solved by a @xmath217 with power law tails .
however , realistic granular systems never show such large tails : the key feature that must be added to this oversimplified model is a realistic topology .
therefore our second step ( ilmm , @xmath3 ) was to embed the imm on a @xmath218 lattice .
the @xmath217 changes dramatically and becomes identical to those measured in md simulations , also in the velocity - correlated stage .
moreover , the ilmm allows for a study of spatial correlations , resulting in shock - like structures very similar to the one observed in md and strong analogies with the md structure functions .
our last step ( ilmm , @xmath4 ) has been the passage from @xmath3 to @xmath4 , which mainly results in the appearance of vortices .
vortices are ubiquitous in granular gases experiments and simulations .
the ilmm simply produces vortices as a result of the competion between the parallelizing effect of inelastic collisions and the constraint given by the conservation of total momentum .
moreover , a large set of evidences ( energy decay , structure factors , gaussian velocity pdf s , aging self - correlations , persistence exponent ) indicate that the dynamical behavior of the system is compatible with a diffusion equation , even if short scale defects ( appearing as shocks , internal noise and tails _ la _ porod in the structure functions ) make this model richer . | we study the formation and the dynamics of correlations in the velocity field for 1d and 2d cooling granular gases with the assumption of negligible density fluctuations ( `` homogeneous velocity - correlated cooling state '' , hvcs ) .
it is shown that the predictions of mean field models fail when velocity fluctuations become important .
the study of correlations is done by means of molecular dynamics and introducing an inelastic lattice maxwell models .
this lattice model is able to reproduce all the properties of the homogeneous cooling state and several features of the hvcs .
moreover it allows very precise measurements of structure functions and other crucial statistical indicators .
the study suggests that both the 1d and the 2d dynamics of the velocity field are compatible with a diffusive dynamics at large scale with a more complex behavior at small scale . in 2d the issue of scale separation , which is of interest in the context of kinetic theories , is addressed . |
in condensed matter physics non - equilibrium problems have attracted increasing interest in recent years . despite many advances , describing those problems accurately in the presence of strong correlations remains a formidable challenge even today .
the physics is often involved and the number of available non - equilibrium many - body methods allowing for a controlled access beyond plain perturbation theory is limited . furthermore , each of those has its own advantages and shortcomings .
an alternative way to gain insights into the interplay of non - equilibrium and correlations is to identify non - equilibrium problems which can largely be understood in terms of their equilibrium physics ; compared to non - equilibrium , the latter is frequently rather well understood .
generally , scrutinizing under which conditions such a reduction in complexity can be applied , is of importance to draw a more complete picture .
here we conduct such a study for the non - equilibrium physics of so called quantum quenches in quantum impurity problems . as they provide a basic probe of the non - equilibrium dynamics of many - body quantum systems , quantum quenches have been of great interest recently .
this development is motivated by the progress of experiments with ultra - cold atoms in a tunable potential .
@xcite at time @xmath2 , the system is suddenly brought far from equilibrium by abruptly changing a control parameter , and is subsequently left to evolve unitarily .
_ global _ quenches , for which the control parameter is quenched throughout the entire system , correspond to injecting an extensive amount of energy , thereby allowing one to address the intriguing issues of thermalization and non - equilibrium steady - states in closed many - body quantum systems .
@xcite _ local _ quantum quenches , for which the sudden change is restricted in space , @xcite allow one to investigate the energy propagation .
one might expect that for large times only the low - energy excitations matter and the non - equilibrium dynamics shows universality , provided the system in equilibrium has this property .
we here provide evidence that for several quantum impurity problems this is indeed the case . in this work ,
we study a particular class of local quantum quenches : at time @xmath2 , two independent metallic reservoirs are either directly tunnel - coupled or coupled through a single - level quantum dot and left to evolve unitarily following the schrdinger equation .
in addition , we investigate the quench dynamics after abruptly tunnel coupling a single lead to a quantum dot .
a quantum quench of this type has recently been realized experimentally exploiting the optical absorption of a semiconductor quantum dot .
@xcite in equilibrium , the considered impurity setups are characterized by a low - energy scale @xmath0 , the kondo temperature . from a theory perspective
, it was argued @xcite based on general considerations and explicit computations for a non - interacting impurity problem , that the energy scale @xmath0 characterizes a crossover in the post - quench dynamics , with the long - time behavior being essentially controlled by the low - energy properties of the hamiltonian after the quench . for the considered type of local quench , one can expect that the same holds also in the presence of two - particle interactions .
we focus on parameter regimes in which the sub - system coupling ( weak link or dot ) is a relevant perturbation in the renormalization group ( rg ) sense in equilibrium , and therefore drastically alters the low - energy properties of the system . the explicit non - interacting result of ref . and the expectation for interacting systems can thus be interpreted as follows : time essentially acts as the inverse of an energy scale , and the post - quench dynamics effectively follows the rg flow . at large times , the system ` heals ' itself so that the quantum impurity becomes strongly hybridized with the lead(s ) . via different contacts .
we focus on two different of such geometries : ( a ) coupling the reservoirs indirectly by a single level quantum dot and ( c ) connecting the two chains directly by a weak link , respectively .
furthermore , we consider the setup ( b ) in which a single lead is coupled to a single level quantum dot . the subsequent out - of - equilibrium dynamics is analyzed.,scaledwidth=48.0% ] because of the energy scale @xmath0 that breaks conformal invariance , many of the analytical results on local quenches from conformal field theory @xcite can not be used .
it is in general very difficult to provide closed analytical expressions for the time evolution of observables showing the full crossover , even for formally exactly solvable ( ` integrable ' ) or non - interacting systems .
however , if the above scenario holds , one can expect that the post - quench dynamics is universal in the sense that every observable only depends on @xmath1 ( in the absence of any other energy scale ) and that the long - time regime @xmath3 can be described using boundary conformal field theory ( bcft ) . in this manuscript
, we investigate numerically whether this is indeed the case for various interacting quantum impurity setups , using a combination of density matrix renormalization group ( dmrg ) @xcite and functional renormalization group ( frg ) @xcite methods .
we focus on two key observables : the entanglement entropy between sub - systems , whose long - time behavior reduces to the cft prediction of refs .
( see ref . for a similar observation ) , and the time dependent fidelity or loschmidt echo , whose long - time behavior carries signatures of the effective boundary condition felt by the lead(s ) .
the latter is induced by the screened impurity .
the loschmidt echo is related to the equilibrium anderson orthogonality catastrophe @xcite and is of particular interest since it can be linked to the fourier transform of the absorption spectrum measured in a recent experiment .
@xcite in all the examples considered , we find that the field theory predictions describe reasonably well the dynamics of the microscopic models under scrutiny . the remainder of this paper is organized as follows . in sect .
2 we introduce the different models and observables studied , along with the analytical and numerical methods used in our analysis .
section 3 contains a detailed discussion of the so - called interacting resonant level model ( irlm ) , the arguably simplest interacting impurity model with spinless non - interacting fermi liquid leads . in sect .
4 we generalize this impurity setup to the richer case of interacting luttinger liquid leads , with various impurity configurations ( one or two leads tunnel - coupled to a quantum dot , two leads connected through a point contact weak link ) .
finally , sect .
5 provides a summary of the main results as well as a discussion of perspectives for future work .
we mainly consider setups in which two decoupled fermionic reservoirs prepared in their respective ground states are coupled to each other at time @xmath2 .
thus the combined system is abruptly brought out of equilibrium at this time .
this protocol constitutes a specific local quantum quench .
we analyze different microscopic lattice models corresponding to various settings , with the reservoirs being either ( single - channel ) non - interacting fermi liquids or one - dimensional ( 1d ) interacting luttinger liquids , and the junction being either a single link or a quantum dot .
these setups are depicted schematically in figs .
[ fig : setup ] ( a ) and ( c ) .
furthermore , we investigate the case of a single luttinger liquid reservoir being coupled abruptly to a single level quantum impurity as shown in fig .
[ fig : setup ] ( b ) .
_ irlm _ first , we investigate a microscopic realization of the irlm ( see e.g. refs . as well as references therein ) , which describes a single localized charge level ( quantum dot ) of energy @xmath4 coupled to a left ( index @xmath5 ) and a right ( index @xmath6 ) non - interacting 1d reservoir ; see fig .
[ fig : setup ] ( a ) . in our non - equilibrium setup the tunneling couplings
@xmath7 between the quantum dot level and the reservoirs are abruptly turned on at time @xmath2 .
we consider a situation in which the two - particle interactions @xmath8 between dot and lead fermions are active for all times .
the microscopic hamiltonian reads @xmath9,\\ h_{\rm{dot}}= & \epsilon \hat n_0,\\ h_{\rm{coup},\alpha}(t)=&\theta(t ) \gamma_\alpha^\prime \left(c_{1,\alpha}^\dagger c_0+c_0^\dagger c_{1,\alpha } \right ) \notag \\+&u_\alpha \left(\hat n_0- \frac12 \right ) \left(\hat n_{1,\alpha } - \frac12 \right),\\ h_{\rm{res},\alpha}= & -\gamma_\alpha \sum \limits_{j=1}^{l-1 } \left(c_{j+1,\alpha}^\dagger c_{j,\alpha}+c_{j,\alpha}^\dagger c_{j+1,\alpha}\right).\end{aligned}\ ] ] we use standard second quantization notation by introducing the fermionic annihilation ( creation ) operator @xmath10 ( @xmath11 ) of a spinless particle at site @xmath12 .
furthermore , @xmath13 and @xmath14 denote the occupancy operators of the dot and the first site of reservoir @xmath15 , respectively .
we have chosen a tight - binding description of the reservoirs of length @xmath16 with hopping @xmath17 and open boundary conditions .
the total number of sites in such a setup is @xmath18 .
_ interacting reservoirs _
we also consider models with interacting reservoirs ( rather than non - interacting fermi liquids as in the irlm ) .
this is achieved by including nearest - neighbor interactions of strength @xmath19 in the 1d leads .
the hamiltonian is given by @xmath20 with the chain part ( nearest - neighbor hopping @xmath21 ) @xmath22 .
\label{eqxxzlead}\end{aligned}\ ] ] by subtracting @xmath23 from the occupancy operator @xmath24 the hamiltonian is particle - hole symmetric .
employing a jordan - wigner transformation ( see e.g. ref . ) , it can alternatively be written in terms of spin degrees of freedom ( xxz heisenberg model ) .
two different realizations of the contact between the reservoirs are considered .
the first one of these is referred to as _ point contact _ and described by @xmath25 .\end{aligned}\ ] ] in this a hopping @xmath26 as well as a nearest - neighbor interaction @xmath27 across the link connecting the left and right chains is turned on .
it is schematically shown in fig .
[ fig : setup ] ( c )
. the second one is called _ dot contact _ and the corresponding part of the hamiltonian reads @xmath28 .\end{aligned}\ ] ] in this contact , the two reservoirs are tunnel coupled by a single dot site of energy @xmath4 located at @xmath29 ; see fig .
[ fig : setup ] ( a ) .
we note that these point and dot contact impurity problems are described by lattice systems with total number of sites @xmath30 and @xmath18 , respectively . additionally , we study the case of a single interacting reservoir [ @xmath31 of eq . ]
tunnel coupled by a term of the form of eq . , but only considering the @xmath5 part , to a single level quantum dot at @xmath2 ; see fig . [
fig : setup ] ( b ) .
this system is described by a lattice of in total @xmath32 sites and referred to as the _ single - lead case _ in the following . _
initial preparation _ the initial density matrix is prepared in each case as a product @xmath33 with @xmath34 given by the ground states @xmath35 of the reservoirs at half filling . in the examples involving a quantum dot , @xmath36 is given by the vacuum ( a spin - down in the spin representation ) .
it is well established that the _ equilibrium _ low - energy physics of both our non - interacting as well as our interacting 1d reservoirs can be described by a gapless continuum field theory the tomonaga - luttinger model as long as @xmath37 ( at half - filling ) .
@xcite for @xmath38 our spinless fermion model falls into the luttinger liquid universality class .
all terms not captured by the field theory are rg irrelevant and flow to zero .
however , the bare amplitude of the leading irrelevant bulk couplings ( e.g. the umklapp scattering for @xmath39 ) grows if one approaches the transitions to the gapped phases @xmath40 .
this leads to a decreasing energy scale below which luttinger liquid physics can be observed .
this energy scale vanishes at the phase transition .
@xcite thus , if the range of accessible energies is bounded from below , e.g. by finite size effects , it might be impossible to observe the physics of the field theory .
@xcite we here avoid this problem by restricting the interaction strength to @xmath41 .
it was shown that at low energies the equilibrium or steady - state impurity physics established when the above local couplings between the reservoirs are active for all times is captured by field theory .
@xcite we here focus on quench setups where the coupling between the reservoirs , be it via a structureless point contact or via a quantum dot , is a relevant perturbation in the rg sense leading to a flow to strong coupling .
this means that the _ effective _ coupling between the reservoirs should be thought of as scale dependent ; it is growing as the energy scale is lowered across the typical energy @xmath0 .
the dependence of @xmath0 on the parameters of the hamiltonian depends on the model considered and is given below .
this flow can even be found if the bare couplings are infinitesimally small . the rg flow from weak to strong coupling
is sometimes called ` healing flow ' for obvious reasons . in the presence of a single additional energy scale besides @xmath0 , e.g. the temperature @xmath42 , it implies single parameter scaling of observables with variable @xmath43 and the physics can be regarded as _
_ this equilibrium physics becomes particularly transparent if we further restrict the parameters of our models .
to avoid any complications due to the interplay of several energy scales in our dot models we assume @xmath44 ; the dot fulfills the resonance condition .
for the irlm it was shown that for left - right asymmetric setups the relation between the flowing level - reservoir hoppings and observables becomes involved .
@xcite as a consequence the latter no longer shows the simple scaling behavior found for restored left - right symmetry . in luttinger liquids
the sharp crossover between two half chains with different bulk interactions effectively induces single - particle scattering at the contact and destroys the resonance condition .
@xcite although such asymmetry effects are of genuine interest and of relevance in most experimental realizations , in our first attempt to understand the non - equilibrium dynamics we suppress them by considering left - right symmetric setups with @xmath45 and @xmath46 for the irlm as well as @xmath47 and @xmath48 for the models with luttinger liquid leads .
we furthermore fix our unit of energy by setting @xmath49 .
, such that the time evolution follows the rg flow from weak to strong coupling .
in particular , the large time limit @xmath50 corresponds to the low energy , strong coupling regime . ,
scaledwidth=45.0% ] the question whether the same type of universality is realized in the non - equilibrium post - quench dynamics is largely open ; it lies at the heart of our present work .
more specifically we investigate numerically whether the entanglement entropy and the loschmidt echo introduced below are scaling functions with @xmath1 being the relevant variable indicating a ` dynamical healing ' .
we thus ask whether ` time follows the rg flow ' , with the large time limit @xmath3 corresponding to the low energy fixed point ( see fig .
[ fig : figrg ] for a sketch ) .
this is what one expects based on a field - theoretical approach to the type of local quenches considered here .
@xcite in this approach , the 1d non - equilibrium problem is formally ` folded ' into a two - dimensional ( 2d ) boundary statistical mechanics problem in equilibrium using a wick rotation .
however , this mapping is based on _ assumptions _ on the analytic properties of the appearing partition functions . by providing indications of one - parameter scaling with @xmath1
we thus not only establish evidence that the dynamics of the non - equilibrium problems considered here is ( 1 ) universal and that ( 2 ) field theory can be applied but also that ( 3 ) the assumptions underlying the field theory approach itself are justified .
we furthermore compare the long time dynamics of our observables to the corresponding field - theoretical predictions .
the alleged simple replacement of equilibrium energy scales such as temperature by inverse time in standard rg arguments is obviously very specific to the considered type of _ local _ quenches . within field theory on a technical level it is linked to the possibility of ` folding ' and the subsequent use of bcft .
we already emphasize at this point that universal dynamics can only be expected for times @xmath51 with @xmath52 being the reservoirs band width and @xmath53 their fermi velocity .
this holds in close analogy to the equilibrium situation in which scaling is cut off at large energies by non - universal band effects and at low ones by finite size effects .
there are many different observables that could be considered to monitor the dynamics of the system after the quantum quench .
a very instructive quantity is the entanglement of the sub - systems as a function of time that we measure by the entanglement entropy @xmath54 .
it characterizes the exchange of information between the two reservoirs .
more precisely , the entanglement entropy is calculated between the left reservoir ( @xmath55 ) and the rest of the system ( @xmath56 which in our dot setups contains the dot site ) and is defined as @xmath57\right\},\ ] ] with the reduced density matrix of the left subsystem @xmath58 $ ] with respect to the full system @xmath59 and @xmath60 .
initially , we have @xmath61 as the left reservoir is decoupled from the rest of the system .
the evolution of the entanglement entropy across a marginal defect in non - interacting fermionic systems was computed in refs . .
the generalization to interacting problems in the case of a weakly modified link was carried out recently in ref .
( see ref . for the corresponding analysis in equilibrium ) . in the following ,
we investigate whether the results of ref .
hold for a microscopic model .
in addition , we study @xmath54 for dot setups .
another quantity that we study is the so - called loschmidt echo @xmath62 which is the fourier transform of the work distribution @xcite @xmath63\right)\left|\big\langle \psi^{(n ) } \big| \psi_0\big\rangle\right|^2 , $ ] with the eigenvalues @xmath64 and eigenstates @xmath65 of the hamiltonian after the quench , and @xmath66 the ground state of the hamiltonian before the quench ( decoupled reservoirs ) with corresponding ground - state energy @xmath67 .
the square modulus of the loschmidt echo is the probability to find the system at time @xmath68 in its initial state .
it can be thought of as a quantum return probability which , if decaying , characterizes the irreversibility of the time evolution .
the loschmidt echo has attracted a lot of attention recently , in the context of ` dynamical ' anderson orthogonality catastrophes , see e.g. refs . , but also in studies of ` dynamical phase transitions ' .
@xcite we note that the work distribution can be argued to be directly proportional to the absorption spectrum in quantum dot setups where the absorption of a photon effectively triggers a quantum quench .
@xcite a general field theoretical formalism to compute @xmath69 for ( ` integrable ' ) impurity problems after a local quench was developed in ref . but only applied to the non - interacting resonant level model ( rlm ) . _ boundary conformal field theory _ in general , it is difficult to provide closed analytical expressions for the time evolution of observables after a local quench . within field theory
this can be linked to the emergence of an energy scale @xmath0 that breaks conformal invariance , and to the presence of interactions . however , in the large time limit @xmath50 , one can obtain asymptotic results using the above described mapping to a 2d inhomogeneous statistical mechanics problem as well as bcft . for the entanglement
entropy one finds @xcite @xmath70 where we have inserted the central charge @xmath71 of the luttinger liquid ( or dirac fermions ) reservoirs ( see e.g. ref . ) . here
@xmath72 is a non - universal constant .
this relation should hold in all setups with two reservoirs in the single - lead case , the entanglement entropy is bounded from above by @xmath73 ( see below ) .
computing the large time asymptotics of the loschmidt echo is slightly more involved . for a _ point contact junction _ between two luttinger liquids
, we can use the cft results of refs . and obtain @xmath74 independent of @xmath75 as long as @xmath76 so that the local tunnel coupling is relevant . note that we have once again inserted the central charge @xmath71 of the gapless reservoirs .
the _ irlm _ and _ dot setups _ are more subtle because of the dynamical nature of the impurity . in these cases
, it is convenient to use the general framework introduced in ref . and consider the imaginary time loschmidt echo as a correlation function ( or modified partition function ) in a semi - infinite critical 2d statistical mechanics problem , with the impurity corresponding to a boundary condition ; see ref . for a similar calculation in the context of the fermi edge singularity . in that language , a quantum quench at @xmath2 can be considered as a sudden change of boundary condition at imaginary time @xmath77 , effectively creating infinitely many massless excitations in the bulk . following ref .
, one can then interpret the loschmidt echo @xmath78 for @xmath79 as the two - point function of a boundary condition changing operator , @xcite whose scaling dimension is fixed by conformal invariance .
assuming analyticity to rotate back to real time , we expect @xcite @xmath80 with the scaling dimension @xmath81 which is model dependent and given below .
we note that the vanishing of the loschmidt echo at large time is the real - time analog of the well - known anderson orthogonality catastrophe .
@xcite as a consequence the work distribution is characterized by an edge singularity at low energy @xcite which can be observed in optical absorption experiments ( see e.g. refs . in the context of the kondo effect ) . _ density matrix renormalization group _
dmrg has proven to be an invaluable tool to numerically study the equilibrium and non - equilibrium many - body physics of interacting one - dimensional systems .
@xcite in dmrg the numerical cost depends in an exponential fashion on the entanglement in the system . for typical real - time evolutions , entanglement grows linearly with time and thus the numerical resources available are exhausted exponentially fast until no further progress in time can be made .
the time scales reachable within dmrg thus critically depend on the entanglement of the system under scrutiny . in this work
, we apply dmrg in a very natural representation via matrix product states .
@xcite we use the implementation outlined in ref . to tackle the problems introduced in the previous section .
we checked that preparing the ground states in the decoupled reservoirs either via a simple imaginary time evolution starting from an initial random product state ( see sect .
7 of ref . ) or a more sophisticated iterative procedure ( see sect . 6 of ref . ) yield coinciding results .
the iterative ground state search is performed with a single - site algorithm . to protect this algorithm from getting stuck in non - global minima when optimizing the energy we use the ideas introduced in ref . .
for the imaginary time evolution , we apply a second - order trotter decomposition with @xmath82 , and we gradually increase the bond dimension @xmath83 during the convergence process . once the ground states have been prepared , we employ a real time evolution algorithm ( see again sect . 7 of ref . ) with the full hamiltonian including the tunneling terms .
we use a fourth - order suzuki - trotter decomposition with @xmath84 chosen small enough to give converged results on the scale of every plot presented in the following .
additionally , we can exploit the trivial rewriting of the loschmidt echo @xcite @xmath85 this allows us to reach times twice as large as in the original form exploiting the same numerical resources .
the bond dimension is dynamically increased during the real time evolution so that the discarded weight @xmath4 always remains below @xmath86 .
_ functional renormalization group _ for the irlm we also analyze the quench dynamics using the frg .
frg is a versatile method to tackle quantum many - body problems .
@xcite recently , it was extended to time evolution in non - equilibrium @xcite including quench dynamics .
@xcite in the present case it allows to study our lattice realization of the irlm at larger system sizes as well as to reach larger times compared to dmrg
. in frg one sets up an infinite hierarchy of flow equations for the many - particle vertex functions .
we here employ the lowest order truncation scheme for this hierarchy .
this means that the two - particle vertex is set constant ( remaining the bare two - particle interaction throughout the entire flow ) , while in contrast the self - energy acquires a rg flow .
this truncation poses the only relevant approximation to the many - body problem at hand .
it is known that this procedure is well suited to describe the impurity physics of the irlm .
it leads to the correct resumation of logarithmic terms in form of power laws , with exponents agreeing to the exact ones to leading order in @xmath87 .
@xcite we supplement the scheme of ref . by a suzuki - trotter decomposition as used in dmrg to describe the propagation in time .
this significantly boosts the performance .
we always choose a symmetric fourth order decomposition , with @xmath88 taken small enough such that it corresponds to a negligible approximation .
since the frg scheme outlined in ref
. aims at the single particle green functions we need to explain how the loschmidt echo can be deduced . in our truncation order interactions
are incorporated effectively in a non - interacting , but time - dependent , renormalized hamiltonian ( in form of the self - energy ) .
since the time steps @xmath88 used are small enough , such that the hamiltonian can be approximated as a constant @xmath89 during each of such time steps we can straightforwardly perform the time evolution of the initial wave function @xmath90 ( ground state of the decoupled reservoirs ) with the renormalized hamiltonian .
more explicitly , we can write the ground state of the non - interacting reservoirs as a product state @xcite @xmath91 where @xmath92 denotes the total number of fermions and @xmath93 is the vacuum . the wave function can be propagated in time with a stepwise ( in time ) constant hamiltonian @xmath94 by @xmath95 where @xmath96 and @xmath97 are the matrices with entries @xmath98 and @xmath99 , respectively . the loschmidt echo in turn
can then be calculated as @xmath100 . \end{split}\ ] ] while in the outlined truncated frg treatment single - particle green functions are approximated correctly ( at least ) to leading order in @xmath87 this is less clear for a quantity like the loschmidt echo .
we have therefore carefully checked our frg results against numerically exact dmrg data for shorter times and smaller systems ( see fig .
[ fig : irlm_com_frg_dmrg ] ) .
the agreement at small @xmath87 is very convincing such that we can trust the frg results for @xmath69 at larger @xmath16 and @xmath68 as well . in a similar , but more complicated fashion one could derive an expression for @xmath54 accessible to frg ;
we do not pursue this here for the sake of brevity .
in the field theory description of the semi - infinite chain - like reservoirs , we first linearize the single - particle dispersion near the fermi energy and then conveniently `` unfold '' these half - infinite wires to obtain infinite ones with only right moving fermions .
this leads to @xmath101 , with the fermi velocity @xmath102 and the field operators @xmath103 .
the coupling between the dot and the reservoirs can then be expressed as @xmath104 , where @xmath105 and @xmath106 depend on the microscopic parameters @xmath107 and @xmath87 in a non - universal ( and in general unknown ) way . here
@xmath108 denotes normal ordering .
we then bosonize the fermionic fields @xmath109 , @xcite with bosonic fields @xmath110 , and introduce an effective spin operator through @xmath111 and @xmath112 with @xmath113 a majorana fermion .
after several canonical transformations , the irlm hamiltonian can be mapped onto an effective anisotropic kondo problem , @xcite described by a single chiral boson @xmath114 that depends on @xmath115 in a complicated non - linear fashion ( see e.g. ref . for details of the mapping ) .
the resulting hamiltonian reads @xmath116 ^ 2 + \theta(t ) \frac{\tilde{\gamma}^\prime}{\sqrt{\pi } } \left[\mathrm{e}^{i \beta \phi(0 ) } s^+ + { \rm h.c . }
\right ] , \label{eqirlmft}\ ] ] where the boundary perturbation ( second term ) has dimension @xmath117 ( assumed to be smaller than one ) .
this means that the boundary perturbation flows under a rg procedure as @xmath118 with @xmath119 , and @xmath120 is the infrared cutoff . solving this equation leads to the length dependent coupling @xmath121 , where @xmath122 is the bare coupling , and @xmath16 a typical length scale .
the kondo temperature @xmath123 is defined as the scale at which @xmath124 .
it thus scales as @xmath125 the rlm has @xmath126 such that @xmath127 and @xmath128 . for small @xmath87
, one finds @xmath129 ( see e.g. ref . ) . a particularly interesting value of the coulomb interaction is given by the so - called self - dual point for which @xmath130 in our regularization scheme , such that @xmath131 and @xmath132 .
this self - dual point corresponds to @xmath133 on the lattice .
@xcite without loss of generality we fix the prefactor of @xmath0 to be 4 .
why this choice is useful will become clear in the next section .
at low energy , the boundary interaction flows to the conformally invariant boundary condition @xmath134 , where @xmath135 corresponds to the phase shift felt by the effective fermion @xmath136 .
we stress that this fermion @xmath137 has a very complicated expression in terms of the original fermionic fields @xmath138 . as discussed in ref .
the large time behavior of the entropy is given by eq .
, whereas the loschmidt echo behaves as eq . with @xmath139 using eq .
we therefore expect @xmath140 for the rlm and @xmath141 for the irlm at the self dual point .
this provides an interesting example where the interactions strongly renormalizes the large time exponent of the loschmidt echo . for the irlm ( including the rlm with @xmath142 ) .
we show the ratio of the kondo temperature @xmath143 obtained by collapsing our numerical dmrg data by hand , the @xmath144 from eq . , and the field theory prediction @xmath145.,scaledwidth=44.0% ]
when comparing predictions from field theory to results obtained numerically for the microscopic models obtained by dmrg and frg , one needs to be careful about the range of validity of the field theory description .
two conditions have to be fulfilled simultaneously : ( a ) the large @xmath16 limit @xmath146 and ( b ) the scaling limit @xmath147 . at a given large @xmath16
they impose an upper as well as lower bound on @xmath148 . as a first step
let us assume correspondence between the microscopic model and the field theory in the proper limit and let us choose @xmath149 with @xmath150 as a point of reference .
the corresponding reference curve for the loschmidt echo @xmath151 obtained for @xmath152 using dmrg is scaled with the field theory kondo temperature @xmath153 .
we then determine the kondo temperature @xmath154 for the microscopic irlm by collapsing all curves ( with different @xmath148 ) on top of this scaled reference curve .
the ratio between @xmath154 extracted by this procedure and the field theory prediction @xmath155 is shown in the main panel of fig .
[ fig : tkmicro ] . for @xmath156 , we leave the scaling limit regime and deviations between the field theory @xmath155 and the @xmath154 of the microscopic model become prominent . for very small
@xmath148 , one does no longer fulfill @xmath146 and finite size effects yield deviations in the kondo temperatures .
this is confirmed by the fact that reducing the size of the system ( from @xmath150 to @xmath157 ) enhances this effect .
we remark that for the irlm it has been shown @xcite that yet another reasonable definition of @xmath0 at small values of @xmath158 is the charge susceptibility @xmath159 this choice of @xmath0 for the microscopic irlm again results in a very convincing collapse of the curves at small @xmath158 . with this definition of @xmath0 we can fix the prefactor of the field theory @xmath0 to 4 , by demanding that the two definitions agree at @xmath142 .
since @xmath160 does not rely on calculating the loschmidt echo , we can now for @xmath142 compare @xmath160 with @xmath155 in the limit @xmath161 .
the condition @xmath162 is thus guaranteed to hold and the finite size deviations for small @xmath148 disappear , while effects arising from the second inequality @xmath163 being no longer fulfilled remain roughly the same as for @xmath152 ( see fig .
[ fig : tkmicro ] ) . since overall the kondo temperatures @xmath0 defined from field theory , or via the fit procedure as well as the charge susceptibility for the microscopic model agree reasonably well , in the following we will only use the field theory definition @xmath164 to collapse the curves obtained numerically for the microscopic model .
nevertheless , it is important to keep in mind that simultaneously fulfilling both inequalities in numerical computations for finite chains infers restrictions on @xmath148 . in the single lead case at fixed @xmath165 . gradually increasing
the interaction @xmath166 between the dot and the single lead ( @xmath150 ) reveals that the marginal term dropped in our field theory analysis does not alter the large time exponent .
the time scales for which asymptotic behavior can be observed , however , increases with increasing @xmath166 .
the field theory prediction @xmath167 is included as a dashed red line.,scaledwidth=44.0% ] let us now discuss the numerical data obtained for the time evolution of the loschmidt echo as well as the entanglement entropy .
the main plot of fig .
[ fig : irlm_com_frg_dmrg ] ( a ) shows a comparison of the loschmidt echo obtained by dmrg and frg at @xmath168 and for different @xmath148 .
the agreement between the two methods is very convincing for this small value of the two - particle interaction .
frg can be used to tackle larger system sizes and times , which is the reason why data for @xmath169 is shown compared to @xmath150 for dmrg .
the inset of this panel shows the collapse of the same frg curves when the kondo temperature @xmath0 is used as the proper energy scale .
in addition , a similar collapse of dmrg data for a larger value of the interaction strength @xmath152 corresponding to the self - dual point is presented .
the large time bcft predictions eq . of sect .
[ subsecftirlm ] with @xmath170 and @xmath171 ( shown as dashed lines ) are consistent with our numerical data .
similarly , fig .
[ fig : irlm_com_frg_dmrg ] ( b ) shows the one - parameter scaling of the entanglement entropy .
the main plot depicts @xmath172 while the inset covers @xmath152 .
the curves collapse well when rescaled by @xmath0 , and the long time behavior is consistent with the universal cft prediction for the entropy eq . .
considerations about the limits of applicability of field theory scaling in lattice models similar to the equilibrium ones outlined in sect .
[ subsecscalinglimit ] are also crucial for the dynamics . in particular , we expect that for ` large ' values of @xmath148 the scaling limit condition is not fulfilled perfectly .
this is the reason why for the largest considered @xmath148 at small times deviations from scaling and even oscillations in the loschmidt echo are apparent in fig .
[ fig : irlm_com_frg_dmrg ] .
the amplitude of the latter vanishes for @xmath173 and their frequency is set by the bandwidth .
we here will not further investigate this non - universal piece of physics.@xcite nonetheless , the curves collapse nicely at large times and allow us to access larger rescaled times @xmath1 . in particular , only for sizable @xmath148 sufficiently large @xmath1 for analyzing the asymptotic behavior can be reached . in general , we find that ` small ' values of @xmath148 describe well the universal scaling curve for small @xmath174 , while larger values of @xmath148 leave the scaling limit regime for small @xmath1 but collapse very well for larger values of @xmath1 . it is known from equilibrium that sampling the entire scaling function for a finite size lattice model requires the use of different values of the impurity strength .
@xcite after rescaling by @xmath0 the different curves overlap and in combination form the scaling function .
we will use this procedure also for our other impurity models .
although the non - equilibrium dynamics of the irlm studied above involved _ two non - interacting reservoirs _ connected through a single - level dot quenching the tunneling between a _ single interacting _ luttinger liquid lead and a quantum dot results in very similar physics . this motivates why this case is discussed next .
we consider a reservoir given by eq .
( @xmath5 only ) whose low - energy physics in the gapless regime ( @xmath175 ) falls into the luttinger liquid universality class with luttinger parameter @xmath176 .
@xcite in the bosonized language , @xcite the reservoir can be described in terms of a massless compactified boson @xmath177 whose right- and left - moving components are scattering off the quantum dot .
the chiral fields @xmath178 and @xmath179 live on the interval @xmath180 , with the impurity at @xmath181 .
it is again convenient to unfold the semi - infinite wires to define a chiral boson @xmath182 ( say , right - moving ) on the real line : @xmath183 for @xmath184 , and @xmath185 for @xmath186 .
one then expects that the low energy sector of the hamiltonian eq .
is given by @xmath187 ^ 2 + \theta(t ) \tilde j'\ !
\left ( \mathrm{e}^ { - i \phi(0 ) } s_0^{+ } + { \rm h.c . }
\right)+ \dots , \label{eqsingleleadbosonized}\ ] ] with @xmath188 for small @xmath26 and @xmath189 is the auxiliary spin of the impurity as in the irlm .
the boundary term has dimension @xmath190 and is thus relevant for all values of @xmath191 .
remarkably , this field theoretical hamiltonian is formally equivalent to that of the two lead irlm eq .
, so all the formulas derived for the irlm also apply here . in particular , the system again flows to a ` healed ' fixed point at low energy and the corresponding kondo energy scale reads @xcite @xmath192 in analogy to the irlm we choose the prefactor to be 4 . further pursuing the analogy , the large time behavior of the loschmidt echo is given by eq . with @xmath193 .
the entanglement entropy in this case is expected to scale exponentially to @xmath194 , which signals the onset of the singlet formation between reservoir and the dot level . however , when ` deriving ' eq . , we dropped the term @xmath195 for @xmath196 of the microscopic lattice model which is _ marginal _ and could therefore change the exponents in a non - universal way . within field theory one
considers the limit @xmath197 .
therefore , since this marginal term scales with @xmath26 as well , we do not expect it to modify our results in the field theory limit .
in practice however , we also consider sizable @xmath26 to cover the full crossover , so one needs to verify numerically the importance of this marginal coupling .
gradually increasing the interaction term @xmath198 starting at 0 , we find numerically that the long - time exponent of the loschmidt echo seems to be independent of this marginal contribution as shown in fig .
[ fig : singlesite_infdl ] .
note however , that the time scale for which asymptotic behavior is reached evolves to larger times with increasing @xmath198 . as an alternative protocol one could consider a setup in which the interaction term @xmath199 is present already initially ( only the reservoir - dot hopping is switched on ) .
we expect that in this case the long - time exponent of the loschmidt echo acquires an extra contribution depending on @xmath198 ; see ref . for a related calculation .
our dmrg results are presented in fig .
[ fig : singlesite ] . the data for the loschmidt echo , fig .
[ fig : singlesite ] ( c ) for @xmath200 , collapse rather nicely if rescaled by @xmath0 for both positive and negative @xmath75 .
moreover , the long time prediction @xmath201 shown as black dashed lines appears to be consistent with our data , despite the presence of the marginal term @xmath202 .
we find that the data for @xmath203 collapse using the same @xmath0 albeit with stronger oscillations ; see fig .
[ fig : singlesite ] ( a ) . in fig .
[ fig : singlesite ] ( b ) and ( d ) we depict the entropy .
scaling is again very convincing .
the inset of fig .
[ fig : singlesite ] ( d ) shows that for @xmath204 , @xmath54 indeed approaches @xmath194 in an exponential fashion . for the dot contact at fixed @xmath165 .
gradually turning on the interaction @xmath198 between the dot and the leads ( @xmath150 ) reveals that the marginal term dropped in our field theory analysis of the dot case does not alter the large time exponent found ( see fig .
[ fig : singlesite_infdl ] for the single - lead setup ) .
the time scales for which asymptotic behavior can be observed increases with increasing @xmath166 . the field theory prediction @xmath205 is included as a dashed red line.,scaledwidth=48.0% ] the low energy limit of the dot contact corresponds to two luttinger liquid reservoirs connected through an effective spin-@xmath206 impurity @xmath207 .
it can be described in the bosonized language starting from the single - lead setup discussed above by introducing two bosons ( one per channel ) @xmath208 the contact term eq .
is then given by @xmath209 with @xmath188 for small @xmath26 .
as in the single - lead case , this term has dimension @xmath210 ( @xmath211 for @xmath212 on which we focus ) and is thus relevant .
the kondo temperature is given by eq . .
note that we have again dropped the term @xmath213 s_0^z$ ] which is also marginal and could change the scaling exponent .
in analogy to the single lead case we analyze the influence of this marginal contribution on the asymptotic exponent of @xmath69 numerically by increasing @xmath198 starting at 0 . as shown in fig .
[ fig : dot_infdl ] , the exponent again appears to be independent of the marginal contribution , although the asymptotic time regime is shifted to larger times with increasing @xmath198 .
the low - energy fixed point of the two - lead model is slightly more complex than its single - lead analog .
it is convenient to introduce the new basis of bosons @xmath214 , so that the boundary perturbation now reads @xmath215 . at low energy ,
the value of the bosonic field @xmath216 is pinned down at @xmath181 , while @xmath217 satisfies a kondo - like boundary condition @xmath218 , where @xmath219 ( see related discussions in refs . ) .
the long - time exponent of the loschmidt echo eq .
is thus given by @xmath220 , where the @xmath221 contribution corresponds to changing the boundary condition at @xmath181 for @xmath222 from neumann to dirichlet , while the other piece @xmath223 can be associated with the phase shift @xmath224 for @xmath217 .
for non - interacting leads , the problem reduces to the rlm , and one finds @xmath225 as expected .
when @xmath226 ( isotropic limit @xmath227 of the lattice xxz chain ) , the loschmidt echo decays as @xmath228 , where the value @xmath229 is consistent with the two - channel kondo orthogonality exponent reported in ref . .
we summarize our numerical dmrg results for the dot contact in fig .
[ fig : dot ] . both the loschmidt echo [ figs .
[ fig : dot ] ( a ) and ( c ) ] and the entanglement entropy [ figs . [
fig : dot ] ( b ) and ( d ) ] collapse well using the kondo temperature eq . , although the loschmidt echo obtained for @xmath203 shows stronger oscillations again .
the entanglement entropy at large times is consistent with eq . .
we find that the large time behavior of the loschmidt echo is in agreement with the power law prediction eq . , with an exponent consistent with the bcft prediction @xmath230 for @xmath203 as well as @xmath200 .
as a last example , let us consider a point contact eq . between two luttinger liquid reservoirs .
this case does not involve a dynamical impurity ( quantum dot ) , and the bosonized version of the tunneling term reads @xmath231 + \dots,\ ] ] where the dots stand for terms being rg irrelevant in equilibrium , and @xmath188 for small @xmath26 .
we emphasize that eq
. corresponds only to the tunneling part @xmath232 of eq . .
we have dropped @xmath233 as it has scaling dimension 2 and is therefore rg irrelevant ( in equilibrium ) .
forming odd and even combinations , one finds that the even boson decouples while the non - trivial part of the dynamics is encoded in a boundary sine - gordon hamiltonian for the odd field .
the dimension of the perturbation is @xmath234 such that we shall focus on the attractive regime ( @xmath235 , @xmath236 ) where it is relevant .
@xcite to leading order in @xmath237 the rg equation for the amplitude of the tunneling term is @xmath238 . for @xmath239 the kondo scale in that case is thus given by @xmath240 in the following , we will set the prefactor to 4 by convention ( to match the definition of @xmath0 in the irlm ) .
the large time behavior of the entanglement entropy and the loschmidt echo in the field theory are given by eqs . and , respectively .
this corresponds to having @xmath241 in eq .
, which is known to be the scaling dimension of the operator changing boundary conditions from neumann ( @xmath242 ) to dirichlet ( @xmath243 ) in the free boson theory . for the point contact and different @xmath75 .
we show the kondo temperature @xmath143 obtained from collapsing our numerical dmrg data by hand ( red symbols ) as well as the small @xmath165 field theory prediction @xmath244 ( black dashed lines ) on a log - log scale .
the lines are shifted such that @xmath143 and @xmath245 coincide at the smallest @xmath165 available .
the power - law behavior of @xmath143 found for small @xmath165 is consistent with the prediction from field theory.,scaledwidth=48.0% ] to sample the entire scaling function for a lattice with @xmath150 sites we can not restrict our considerations to @xmath246 .
it is known that for @xmath247 the dual picture of a weak impurity , instead of a weak link , is the appropriate one .
@xcite the dependence of @xmath0 on the impurity strength is modified @xcite and eq .
does no longer apply .
we thus proceed as follows .
we consider different @xmath165 and scale the dmrg post - quench time - evolution data obtained for the loschmidt echo by hand until they collapse ; see fig .
[ fig : pointcontact ] ( a ) .
this gives us an estimate of @xmath0 which we then also use to rescale @xmath54 .
approaching small @xmath165 we expect to find the scaling eq . of @xmath0 .
the corresponding analysis is shown in fig .
[ fig : tkmicro_point ] .
the field theory prediction works well for small @xmath165 compare the symbols to the dashed lines representing the power law eq .
while rather large deviations can be found for @xmath248 ( note the y - axis log - scale ) .
this shows that in contrast to the above dot junction cases , in which it was possible to take the small hopping field theory expression for @xmath0 ( for a detailed analysis indicating this for the ( i)rlm , see fig .
[ fig : tkmicro ] ) , determining the kondo temperature by hand is vital for the point contact . in fig .
[ fig : pointcontact ] ( a ) we show the collapsed dmrg data for @xmath69 and different @xmath236 .
the collapse works particularly well for larger @xmath249 and due to finite size effects deteriorates for @xmath250 .
remind that we can not further increase @xmath249 as the leading irrelevant _ bulk terms _ will increase and spoil scaling .
independent of @xmath75 the asymptotic behavior is consistent with @xmath251 ( dashed lines ) providing indications for the field theory prediction @xmath241 .
we remark that the _ static _ orthogonality exponent @xmath81 for the same lattice model was studied in refs . .
in the latter work the authors employed a very stringent numerical analysis based on a logarithmic derivative . in this context the precise numerical evaluation of @xmath81 remains an open question due to the difficulty to reach very large system sizes ( see also ref . ) . in our work taking the logarithmic derivative to determine @xmath81 is impractical due to small oscillations prevailing even at large times .
therefore , in the light of our analysis , we can only state that the large time exponent appears to be consistent with the prediction @xmath241 .
we certainly can not rule out small corrections to this exponent , let alone show or disproof their existence .
analogous to the problem of reaching very large system sizes ( as encountered in ref . )
, we would need to analyze much larger times , which is impractical .
our rescaled dmrg results of @xmath54 [ using the @xmath0 extracted from the by hand scaling of @xmath69 ] for the point contact are shown in fig .
[ fig : pointcontact ] ( b ) .
again , the collapse is reasonable . at large times , the data follow the field theory prediction eq . .
in this paper , we studied the non - equilibrium dynamics of the loschmidt echo and of the entanglement entropy resulting out of abruptly coupling two reservoirs which are non - interacting fermi liquids or interacting luttinger liquids .
the coupling is either realized directly by a weak link between the two systems ( point contact ) , or indirectly by an additional single site dot in between them ( dot contact ) .
in addition , we considered the dynamics when coupling a single luttinger liquid lead to a single site dot ( single - lead case ) .
microscopic lattice models were used to describe these setups .
the observables were accessed using both dmrg and frg .
we checked numerically the scaling expected from field theory and investigated whether the large time behavior can be successfully captured by ( boundary ) conformal field theory . simultaneously fulfilling the conditions @xmath146 , @xmath147 , and @xmath51 under which field theory is expected to describe the physics of lattice models at fixed @xmath16 sets bounds on the strength of the sub - system hoppings entering in @xmath0 as well as on the time @xmath68 .
furthermore , the times reachable are bounded from above by the methods used . taking this into account
, we gathered evidence that for a microscopic realization of impurity systems with fermi or luttinger liquid reservoirs , the dynamics is universal and described by field theory .
it would be interesting to generalize our work to finite temperatures , where the work distribution satisfies an out - of - equilibrium fluctuation - dissipation relation , @xcite which might bear interesting consequences .
the finite temperature generalization of the loschmidt echo would then be @xmath253 , where @xmath254 refers to thermal average with the density matrix @xmath255 , corresponding to the system being held at temperature @xmath42 before the quench . because of this new energy scale , we expect both the loschmidt echo and the entanglement entropy to have a more complicated scaling behavior as functions of @xmath1 and @xmath256
. for instance , at small temperature @xmath257 , one can argue from bcft that the loschmidt echo for @xmath3 should scale as @xmath258^{4 \delta_{\rm bcc}}$ ] , with @xmath0 playing the role of a ultraviolet cutoff . in the large time regime @xmath259
, this implies an exponential decay of the loschmidt echo , contrasting with the power - law behavior for @xmath260 .
we leave the detailed analysis of finite temperature quenches for future work .
in addition , we expect that the dynamics resulting out of local quenches for spinfull impurity systems ( spin kondo effect ) should lead to a richer crossover physics .
in particular , the anderson orthogonality exponent dominating the long time behavior of the loschmidt echo will have contributions coming from both spin channels . other interesting generalizations of our work include the study of quantum quenches in impurity systems governed by more complicated rg flows , with for instance intermediate fixed points .
we are grateful to the mpipks dresden for hosting the workshop ` quantum many body systems out of equilibrium ' where this work was initiated .
this work was supported by the quantum materials program of lbnl ( rv ) and the forschergruppe 723 of the dfg ( dmk and vm ) .
dmk thanks the university of california , berkeley for hospitality during his visit in summer 2013 .
rv also wishes to thank h. saleur and j.e .
moore for discussions , and the university of southern california for hospitality and support through the us department of energy ( grant number de - fg03 - 01er45908 ) .
s. andergassen , m. pletyukhov , d. schuricht , h. schoeller , and l. borda , phys .
b * 83 * , 205103 ( 2011 ) ; ibid .
s. andergassen , m. pletyukhov , d. schuricht , h. schoeller , and l. borda , phys .
b * 84 * , 039905(e ) ( 2011 ) . c. karrasch and d. schuricht , phys .
b * 87 * , 195104 ( 2013 ) . f. andraschko and j. sirker , phys . rev .
b * 89 * , 125120 ( 2014 ) .
s. vajna and b. dra , phys .
b * 89 * , 161105 ( 2014 ) . b. pozsgay , j. stat .
( 2013 ) p10028 .
m. heyl , arxiv:1403.4570 ( 2014 ) . | the _ equilibrium _ physics of quantum impurities frequently involves a universal crossover from weak to strong reservoir - impurity coupling , characterized by single - parameter scaling and an energy scale @xmath0 ( kondo temperature ) that breaks scale invariance .
for the _ non - interacting _ resonant level model , the _ non - equilibrium _ time evolution of the loschmidt echo after a local quantum quench was recently computed explicitly [ r. vasseur , k. trinh , s. haas , and h. saleur , phys .
rev .
lett . * 110 * , 240601 ( 2013 ) ] .
it shows single - parameter scaling with variable @xmath1 . here
, we scrutinize whether similar universal dynamics can be observed in various _ interacting _ quantum impurity systems . using density matrix and functional renormalization group
approaches , we analyze the time evolution resulting from abruptly coupling two non - interacting fermi or interacting luttinger liquid leads via a quantum dot or a direct link .
we also consider the case of a single luttinger liquid lead suddenly coupled to a quantum dot .
we investigate whether the _ field theory _ predictions for the universal scaling as well as for
the large time behavior successfully describe the time evolution of the loschmidt echo and the entanglement entropy of _ microscopic models .
_ our study shows that for the considered local quench protocols the above quantum impurity models fall into a class of problems for which the non - equilibrium dynamics can largely be understood based on the knowledge of the corresponding equilibrium physics . |
it is well known that , with few exceptions , the presence of non - mesomorphic solutes in nematics depress the normal nematic - isotropic transition temperature ( @xmath5 ) @xcite .
moreover , experiments @xcite over last few decades have confirmed that the presence of such impurities leads to the formation of a two phase region .
this is consistent with the laws of thermodynamics and the first order nature of the @xmath6 transition .
two types of models have proved to be generally useful for investigating @xmath6 transitions in dilute solutions .
the first is a lattice model @xcite which takes into account the role of intermolecular repulsion or the steric effect .
the other @xcite is the mean field theory of the maier - saupe type which takes into account the anisotropic intermolecular attraction or dispersion forces .
more recently mukherjee @xcite has explained the behavior of the two phase region using the phenomenological landau de gennes theory .
the role of non mesogenic impurities on a mesogen which exhibits a smectic phase as well , has not received a great deal of attention of either experimentalists or theorists . here too , in favourable circumstances , remains a possibility of the existence of a two phase region near the smectic transition @xcite . in this paper
we report the results of measurements carried out on a mixture of biphenyl and @xmath1-n - octyl - cyano biphenyl ( 8cb ) .
the latter which has a strong polar group and a long alkyl chain has received considerable attention of experimentalists over last few decades @xcite and is known to exhibit a first order @xmath7 transition .
the biphenyl+8cb system to our knowledge , has not been studied before and is a natural and interesting choice in its own merit since the solute , which is chemically @xmath0 , is a non - mesogenic , non - polar , rigid molecule and is nothing but 8cb deprived of its cyano group and the flexible alkyl chain .
we have observed that biphenyl is easily miscible with 8cb and lowers both @xmath8 and @xmath5 by an amount which depends on the impurity concentration .
the range of the nematic phase i.e. , @xmath9 however is not significantly altered .
the effect of this rod - like non - polar impurity on the dielectric anisotropy , the frank elastic constants ( splay and bend ) and on the effective rotational viscosity coefficient of pure 8cb has been reported here . to our knowledge
, no measurement has yet been reported on the effect of impurities on the elastic constants and the viscosity coefficient of a nematic .
the experiments we have carried out involve electric field induced freedericksz transition and consist of two parts .
the dielectric anisotropy and the frank elastic constants are extracted from a set of measurements of the voltage dependence of capacitance of a sample cell filled with the 8cb and biphenyl mixture . for the determination of the rotational viscosity coefficient a method based on dynamic freedericksz transition is used .
interesting features have emerged from these measurements and the splay elastic constant is seen to exhibit anomalous behavior in the vicinity of the @xmath7 transitions for the impurity concentration exceeding a certain value .
the splay elastic constant @xmath2 is related to the freedericksz threshold voltage @xmath10 by the relation @xmath11 where @xmath12@xmath13 is the permittivity of free space and @xmath14@xmath15 , the dielectric anisotropy of the sample .
the method of determination of @xmath3 is described below .
the exact relationship between the cell capacitance @xmath16 and the voltage @xmath17 applied across the cell was obtained by gruler @xcite @xmath18 ^{1/2}d\phi\ ] ] and @xmath19 ^{1/2}d\phi } { \int _ { 0}^{\phi _ { m}}\left [ \frac{(1+\kappa \sin ^{2}\phi ) } { ( 1+\gamma \sin ^{2}\phi ) ( \sin ^{2}\phi _ { m}-\sin ^{2}\phi ) } \right ] ^{1/2}d\phi } \ ] ] where @xmath20 ( @xmath21 being the bend elastic constant ) , @xmath22 , @xmath23 is the tilt angle made by the director with a direction parallel to the cell walls , @xmath23@xmath24 is the tilt angle at the center of the cell and @xmath25 is the capacitance of the cell when the liquid crystal molecules are homogeneously aligned , that is , before the onset of freedericksz transition when the voltage applied is lower than the threshold voltage .
these equations can be combined to yield @xmath26 ^{1/2}dx\ ] ] when the applied voltage is much higher that the threshold voltage , the director at the center of the cell becomes perpendicular to the cell walls and @xmath27 .
then the above equation reduces to @xmath28 ^{1/2}dx\ ] ] or dividing by @xmath29 @xmath30 ^{1/2}dx\ ] ] where @xmath31 may be called the reduced capacitance .
@xmath16@xmath32 is the capacitance of the cell when the nematic is homeotropically oriented , i.e. , the value of @xmath16 in the limit @xmath33 @xmath34 .
@xmath35 and @xmath36 may be obtained by dividing @xmath37 and @xmath25 by the empty cell capacitance @xmath38 respectively . in the above equation use
has been made of the relation @xmath39 .
thus equation ( [ crat2 ] ) predicts that a plot of @xmath40@xmath41 against @xmath33 for @xmath17@xmath42 should be linear and the extrapolated value of the ordinate for @xmath33 @xmath34 should directly provide the value of @xmath29=@xmath43 .
this procedure for obtaining @xmath29 was first suggested by meyerhofer @xcite .
the variation of capacitance with applied voltage was fitted against equations ( [ vrat ] ) and ( [ crat ] ) to obtain @xmath10 and @xmath44 .
we then used ( [ k11eq ] ) to calculate @xmath2 and hence @xmath21 .
this method for obtaining @xmath10 and @xmath44 was suggested by morris et al @xcite .
the viscous behavior of a nematic can be obtained by studying the response to a sudden change of an applied magnetic or electric field which is normally stronger than the freedericksz - threshold value . in case of deformations involving a pure twist
there is no hydrodynamic flow .
the molecules merely rotate without any translational motion and the analysis is rather simple @xcite .
the situation is far more complicated in case of a splay geometry which we have studied since the director reorientation is now accompanied by a hydrodynamic flow .
the gradient of the angular velocity of the director produces a backflow motion , first demonstrated by pieranski @xcite giving rise to a frictional torque .
the sudden removal of an external field in this case would result in a director relaxation time @xmath45 where @xmath46 is the effective rotational viscosity coefficient , @xmath47 is the cell thickness and @xmath2 is the splay elastic constant . in practice
@xmath46 is about @xmath48 less than the rotational viscosity coefficient @xmath49 .
we have determined the relaxation time @xmath50 by measuring the decay of capacitance in a nematic cell .
the details are given in the next section .
we have used a hewlett - packard lcr meter hp 4274a to measure the capacitance of the sample cell .
the instrument employs the so called `` auto balancing bridge method '' to measure both the real and complex parts of impedance simultaneously .
the voltage was varied from 0.1 v to 5 v at a frequency of 1khz .
the sample cells , which were of very high precision , were obtained from displaytech , usa . the cells consisting of two ito coated glass plates with a spacing of 4 @xmath51 m , and an active area of 0.26 @xmath52 , had brushed polyimide treatment to ensure planar orientation . they also had a guard ring incorporated in them to minimize fringe electric fields .
the temperature of the cells were controlled to within @xmath53 by placing the cells within a mettler hot stage fp 82 .
the texture of the samples were viewed through a leitz polarizing microscope and photographs were taken with the help of photoautomat .
both 8cb and biphenyl were obtained from merck and were used without further purification . the samples which we have used were mixtures of various concentrations of biphenyl , ranging from 0.4% to 6.4% , in pure 8cb .
the smectic to nematic transition temperature , @xmath8 and the nematic to isotropic transition temperature @xmath5 were determined from texture studies of the samples .
various photographs were taken at different stages of transition . in the capacitance measurement experiments the probe voltage of the lcr meter hp 4274a provided the aligning a.c . electric field .
the voltage was varied from 0.1 v to 5 v at a frequency of 1khz in the @xmath54 mode of the instrument . in the neighborhood of freedericksz transition readings
were taken at intervals of 10mv .
the samples were filled in the cells and the c - v variation was recorded at a large number of temperatures ranging from the smectic to nematic transition temperature , @xmath8 , to a few degrees beyond the nematic to isotropic transition temperature , @xmath5 . it must be noted that all measurements were carried out as the temperatures of the sample were increased gradually starting from @xmath55 and none of the measurements were taken during cooling .
fig.[c - v variation ] shows a few capacitance - voltage variations of the mixture with biphenyl concentration @xmath56 at temperatures near the two transitions @xmath7 and @xmath6 and at an intermediate temperature .
there is always a need to have a refinement in the value of @xmath10 as the c - v curves obtained in a polyimide cell usually do not exhibit a very sharp threshold of freedericksz transition so as to yield a precise value of @xmath10 from the data directly . from the c - v data @xmath36 and @xmath35
were calculated and subsequently @xmath10 and @xmath44 were determined as described below .
@xmath36 was determined directly from the value of capacitance at voltages lower than the threshold voltage , ( before the onset of the freedericksz transition ) by dividing the capacitance , @xmath25 by the empty cell capacitance @xmath38 .
@xmath35 was obtained by plotting @xmath16 against @xmath33 for the higher voltage range @xmath57 .
the extrapolated value of capacitance in the limit as @xmath58 gives @xmath37 which when divided by @xmath38 gives @xmath35 . a direct measurement of @xmath37
was also carried out to check the reliability of the value of @xmath29 obtained by the above method .
a homeotropic alignment was obtained in the ito coated glass cells using ctab ( cetyl trimethyl ammonium bromide ) and the capacitance of the cell was measured while a @xmath59 magnetic field was applied across the cell .
the values of @xmath29 thus obtained never differed from the @xmath29 s obtained from the @xmath33 -extrapolation method by more than @xmath60 . a two parameter non - linear least square fit for finding @xmath10 and
@xmath44 worked in the following manner .
starting values of @xmath10 and @xmath29 were taken directly from the experiment .
( [ vrat ] ) was then used to obtain the @xmath61 s for all values of v for which measurements of capacitance were carried out .
the @xmath61 s thus obtained were used in ( [ crat ] ) to obtain c. the error which was minimized in the least square programme is @xmath62 where @xmath63 is the measured value of the capacitance at v=@xmath64 and @xmath65 is the value of capacitance obtained by solving ( [ crat ] ) , n being the number of data points for a particular temperature .
the final values of @xmath10 never differed a great deal from the input values . for the rotational viscosity coefficient measurements ,
we have used a hewlett - packard impedance gain - phase analyser hp @xmath66a in the programmable mode to record the decay of capacitance with time .
an integration time of @xmath67 @xmath51sec was chosen which resulted in an interval of @xmath68 msec between the successive readings at the operating frequency which was fixed at @xmath69 khz . with our instrument a delay time , in multiples of @xmath70 msec , could be introduced between the readings but in this case we did not introduce any delay since the decay of capacitance was found to be pretty fast .
probe voltage across the sample cell was held fixed at @xmath71 v. a low frequency a.c .
voltage about @xmath72 volts peak to peak was applied across the cell for about @xmath73 secs .
it was switched off and the transient ( decaying ) capacitance of the cell , was captured as a function of time by the hp @xmath66a .
a plot of this decay of capacitance with time which was always found to be exponential was then dumped into a hp @xmath74a plotter which was connected to the impedance analyser over a hpib bus . from the plots thus obtained for different concentration of biphenyl mixtures , at different temperatures ,
the time constant , @xmath50 was calculated and using the values of @xmath2 obtained from the static capacitance measurements , the effective rotational viscosity coefficient , @xmath46 was evaluated .
the transition temperatures @xmath5 ( upper ) , @xmath5 ( lower ) and @xmath8 were determined from optical texture studies and the results are presented in table [ transition temp(table ) ] and fig.[phase diagram ] . the @xmath5 ( upper )
is the temperature at which the field of view of the polarizing microscope ( with crossed polarizer and analyzer ) becomes completely dark .
all real nematics contain impurities and rosenblatt @xcite has observed in 8cb a two phase @xmath75 region of temperature of width @xmath76 in absence of an electric field .
needless to say , with our temperature resolution of @xmath77 , we were unable to confirm this .
the width of the two phase region was seen to increase with the concentration of biphenyl and for the maximum concentration of @xmath78 this was @xmath79 .
we were unable to work with higher impurity concentration because of non - availability of an experimental arrangement for producing sub - ambient temperatures .
all the transition temperatures reported in table [ transition temp(table ) ] were determined without the presence of any external field .
however , we confirmed that with the highest electric field we applied across the sample cells , namely @xmath80 @xmath81 , there was no noticeable change in any of the transition temperatures .
the temperature dependence of the dielectric anisotropy for the concentration ( c ) of biphenyl ranging from @xmath82 to @xmath56 is shown in fig.[deltaeps ] .
in pure 8cb and for all mixtures we have studied @xmath3 , in the neighborhood of @xmath5 , shows the same feature in that it disappears at @xmath5 ( upper ) . at this particular temperature
the field of view in the microscope with crossed polars is completely dark and hence the phase is isotropic .
values of the dielectric constant @xmath83 obtained well within the isotropic phase show a gradual increase with temperature for pure 8cb as well as for all mixtures we have studied .
focussing on the behavior of @xmath3 as @xmath8 is approached from the higher temperature side we find that for biphenyl concentration , c upto 2.00% @xmath3 in mixtures behaves the same way as that in pure 8cb , namely that it shows a sharp increase . however , for impurity concentrations of @xmath84 and higher , @xmath3 goes down sharply , while always remaining positive ( at least in the temperature range we have explored ) . we have noted that the decrease in @xmath3 results from a reduction of @xmath35 , ( @xmath36 remaining fairly constant ) as @xmath8 is approached .
a similar decrease in @xmath35 as @xmath8 is approached has been observed in pure @xmath85 - diheptylazoxybenzene where the @xmath4 transition is almost second order .
this has been attributed to the presence of pretransitional effects within the nematic phase @xcite .
the temperature and concentration dependence of the splay elastic constant @xmath2 is shown in fig .
the ratio @xmath44 was seen to approach the value 0 and stay around that value at and around transition but then it was seen to increase again with increase in temperature . the plot of bend elastic constant @xmath21 with @xmath86 is shown in fig . [ k33 ] .
the bend elastic constant @xmath21 is seen to diverge as the smectic phase is approached from the nematic side .
the behavior was of the type @xmath87 , where @xmath88 is the difference @xcite between @xmath21 and its nematic part , @xmath89 and @xmath90 is the critical exponent .
we found that for all the samples studied @xmath90 is @xmath91 which is in agreement with the exponent obtained by morris et al @xcite in pure 8cb .
it must be recalled , however , that anisotropic scaling laws , predict that @xmath88 should vary as the correlation length @xmath92 both above and below @xmath8 and this should result in an exponent @xmath93 , which from x - ray scattering experiments in many samples turn out to be @xmath94 @xcite .
but the temperature @xmath95 at which @xmath21 diverges is slightly higher by @xmath96 than @xmath8 for the mixtures and for ease of eye , indicated by the vertical lines in fig.[k33 ] .
we are however unable to say if the weakly first order @xmath7 transition in pure 8cb approaches a second order transition in the mixtures and this needs more detail studies .
on plotting @xmath2 against @xmath29 we find that the variation for all the concentrations lie nearly on a universal curve , as shown in fig.[k11vsgamma ] .
the deviations which are seen in the figure are mainly the points for the higher concentration mixtures which lie in the vicinity of the two transitions .
the abrupt decrease in @xmath35 as the smectic phase is approached , seems to occur at relatively higher concentrations of biphenyl and is totally absent in pure 8cb and in the mixtures of low impurity concentration .
the optical texture studies we have performed fail to reveal the existence of any smectic phase other than smectic a in the mixtures at low temperatures .
it may however be conjectured that the increase in the proportion of biphenyl in the mixture , though small , is leading to a type of ordering , cybotactic or otherwise , of the 8cb molecules where the contribution of the permanent dipole moment of the @xmath97 group to @xmath35 is greatly reduced .
the decrease in @xmath35 which suggests the existence of pre - transitional smectic effects within the nematic phase occurs only for biphenyl concentrations greater than a limiting value .
it may be noted that the ratio @xmath98 changes from @xmath99 in pure 8cb to @xmath100 in the @xmath56 biphenyl-8cb mixture .
it may be pointed out that the decrease in this ratio , though marginal , is believed to favor a second order @xmath7 transition .
the temperature dependence of the rotational viscosity coefficient for the concentration of biphenyl ranging from @xmath82 to @xmath56 is shown in fig.[rotviscvst ] . the @xmath5 ( upper )
is the temperature at which the field of view of the polarizing microscope ( with crossed polarizer and analyser ) becomes completely dark . in pure 8cb and for all the mixtures @xmath46 in the neighborhood of @xmath5 , shows the same feature in that it disappears at or before @xmath5 ( upper ) .
however , as we approach @xmath8 from the higher temperature side we find that for impurity concentrations upto @xmath101 @xmath46 shows a sharp increase as in pure 8cb while for impurity concentration of and higher than @xmath84 , @xmath46 is seen to decrease sharply .
we are not aware of any systematic study involving te effect of non - mesogenic solutes on the rotational viscosity coefficient of the nematics in the neighborhood of nematic - smectic transition .
the miesowicz viscosities of a mixture of 8ocb and 4tpb ( exhibiting the nematic - smectic transition ) were recently investigated by janik et al @xcite using a miesowicz viscometer .
[ orderparm ] shows the plot of @xmath102 vs. @xmath103 .
the dielectric anisotropy @xmath3 is roughly proportional to the long range nematic order parameter .
for pure as well as all mixtures the plot of @xmath102 vs. @xmath103 gives a straight line .
the same sort of dependence has been obtained by gasparoux and prost @xcite , heppke and schneider @xcite for mbba and by prost et al @xcite , for a mixture of two isomers of @xmath104 .
close to @xmath5 and @xmath8 there are deviations and those points have not been shown in the figure . from the slope of these plots
the activation energy for diffusion @xcite was calculated and has been shown in table [ activation energy ] .
to summarize it can be stated that the presence of the rigid , non - polar and non - mesogenic solute biphenyl in 8cb results in a change in the dielectric anisotropy @xmath3 , the splay elastic constant @xmath2 and the effective rotational viscosity coefficient @xmath46 . in the neighborhood of the nematic - smectic a transition for the concentration of biphenyl @xmath105
the changes in these quantities have been found to be anomalous in that , instead of the usual rapid increase observed on cooling , @xmath3 , @xmath2 and @xmath46 sharply decrease .
this may perhaps be attributed to the presence of pretransitional effects within the nematic phase . from optical texture studies we could not identify any new smectic phase in the mixtures other than what is found in 8cb .
the bend elastic constant @xmath21 in all the mixtures was found to diverge as the smectic phase was approached and the critical exponent was found to be the same as that observed for pure 8cb , although the temperatures at which the divergences occur were slightly higher than the respective values of @xmath8 .
more elaborate studies involving dsc , x - ray diffraction and refractive index measurements ( to obtain the long range nematic order parameter ) may perhaps reveal more about what is actually happening in the mixtures near the smectic a - nematic transition particularly for those containing relatively higher biphenyl concentration .
.[transition temp(table)]variation of phase transition temperatures @xmath106 and @xmath107 with concentration of biphenyl in 8cb [ cols="^,^,^,^",options="header " , ]
the authors acknowledge valuable discussions with dr .
a. dasgupta , dr . p.k . mukherjee and dr .
k. mukhopadhyay .
the research was supported by dst ( grant no .
sp / s2/m-20/95 ) .
sdg acknowledges the award of a fellowship . | the effect of a rigid nonpolar non - mesogenic solute , `` biphenyl '' which is @xmath0 , on the splay and bend elastic constants and on the rotational viscosity coefficient of @xmath1-n - octyl - cyano biphenyl ( 8cb ) is reported .
the experiments involve the measurement of voltage dependence of capacitance of a cell filled with the mixture .
anomalous behavior of both @xmath2 and @xmath3 near the @xmath4 transition have been observed .
department of physics , jadavpur university , calcutta-700 032 , india |
the field of circumstellar silicate research started in the 1960s when @xcite first theoretically investigated the condensation of dust grains in the circumstellar envelopes around m type long period variables and argued that _ silica _
( sio@xmath2 ) should be the most abundant condensed species .
@xcite showed several years later that grains around oxygen - rich cool giants are mainly _ silicates _ such as al@xmath2sio@xmath3 and mg@xmath2sio@xmath4 .
the presence of silicate dust in evolved stars was first revealed through the detection of the 9.7@xmath0 si o stretching emission band in m stars @xcite .
this was later further supported by the detection of the 18@xmath0 o si o bending emission band , also in m stars @xcite .
the 9.7 and 18@xmath0 silicate bands have also been seen in _ absorption _ in evolved stars with an extended , heavily obscured circumstellar dust shell ( see and references therein ) .
all the earlier observations showed that the 9.7 and 18@xmath0 circumstellar silicate bands were broad , smooth and featureless , suggesting that they were predominantly amorphous @xcite .
indeed , they can be well fitted with laboratory data for amorphous olivine silicates ( e.g. see ) .
this picture has dramatically changed in the 1990s , thanks to the _ infrared space observatory _ ( iso ) and the _ spitzer space telescope_. using the _ short wavelength spectrometer _ ( sws ) on board _ iso _ , obtained the 1245@xmath0 emission spectra of six oxygen - rich evolved stars .
they for the first time reported the detection of crystalline silicates in evolved stars , as revealed by the narrow emission features characteristic of crystalline silicate minerals at wavelengths between 30 and 45@xmath0 .
crystalline silicates have now been observed in all evolutionary stages of evolved stars : red giants and supergiants , asymptotic giant branch ( agb ) stars , post - agb stars , planetary nebulae ( pne ) and luminous blue variable ( lbv ) stars , a mass up to @xmath5150@xmath6 , and a lifetime of only a few million years .
lbvs are a stage in the evolution of very massive stars .
they may evolve to wolf - rayet stars before exploding into supernovae . ] .
extragalactic crystalline silicates have also been found in evolved stars by _ spitzer _ in the large magellanic cloud ( e.g. @xcite ) and in r71 , a lbv star by _
we note that like amorphous silicates , crystalline silicates are usually found in _ oxygen - rich _
environments @xcite.1 ) , all oxygen atoms will be bound in co and therefore no silicates or oxides will be able to form ; instead , one expects to form carbonaceous dust and sic dust @xcite .
] however , the presence of crystalline silicates in _ carbon - rich _ evolved stars has also been reported . in these cases the crystalline silicate grains
were likely produced in previous oxygen - rich mass - loss episodes .
the identification of circumstellar crystalline silicates has been made based on the characteristic mid - ir bands of crystalline silicates .
the _ iso _
sws and lws ( _ long wavelength spectrometer _ ) spectra of evolved stars in the 2.4 to 195@xmath0 wavelength region allow one to identify at least 49 narrow bands which are attributed to crystalline silicates , with distinct emission bands at approximately 10 , 18 , 23 , 28 , 33 , 40 and 60@xmath0 .
the widths and peak wavelengths of these bands and their relative strengths contain useful information about the crystalline silicate composition ( e.g. the fe / mg ratio ) and stoichiometry ( e.g. olivine @xmath7 or pyroxene @xmath8 where @xmath9 ; see ) .
the mid - ir spectra of evolved stars obtained by _
iso _ and _ spitzer _ so far indicate that circumstellar crystalline silicates all appear to be extremely mg - rich and fe - poor or even fe - free ( i.e. forsterite @xmath10 or enstatite @xmath11 ; see ) .
however , there are several outstanding problems about crystalline silicates remain to be addressed : first , the exact composition and stoichiometry of crystalline silicate dust in evolved stars are not precisely known .
second , the mechanism for making circumstellar crystalline silicates remains debated .
third , it is unclear whether ( and how ) the fraction of silicate dust in crystalline form depends on the mass loss rate , the temperature of the circumstellar dust shell , and the evolutionary stage of an evolved star .
it has also been suggested that high crystallinity ( @xmath110% ) is typically associated with the presence of a disk , rather than being correlated with the mass - loss rate .
further , it is widely believed that interstellar dust originates from stardust produced by evolved stars .
it is puzzling that interstellar silicates are predominantly amorphous @xcite while an appreciable fraction of circumstellar silicates the primary source of interstellar dust are crystalline . to improve our understanding of these questions , we plan to systematically explore the spectroscopic properties of astronomical crystalline silicates and their formation , evolution , and dependence on the properties of the parent stars as well as their relevance to interstellar silicates . in this work we present the _ spitzer_/irs spectra of three evolved objects ( iras16456 - 3542 , 18354 - 0638 , and 23239 + 5754 ) all of which display rich sets of crystalline silicate emission features .
this is a part of our ongoing efforts for a systematic investigation of crystalline silicates in evolved stars .
the data were obtained through the _ spitzer _ cycle 3 go proposal ( _ spitzer _ program id30403 ; pi : biwei jiang ) .
sixteen evolved objects ( post - agb stars and pne ) have been observed through this program .
they were selected from the _ infrared astronomical satellite _ ( iras ) _ point source catalog _ ( psc ) , with their iras color indices satisfying @xmath12 and @xmath13 .
this is to exclude young stellar objects ( ysos ) or galaxies and limit us to evolved stars . in this work we focus on the three sources ( iras16456 - 3542 ,
18354 - 0638 , and 23239 + 5754 ) whose mid - ir spectra are characterized by rich sets of crystalline silicate emission features .
the other ( 13 ) objects will be discussed elsewhere . in table
[ tab : infrared ] we list their ir brightness ( and thus colors ) and their _ simbad _ classification .
these sources were observed using the _ infrared spectrograph _ ( irs ) on board _ spitzer _ with the _ short - high _ ( sh ) and _ long - high _ ( lh ) modules during cycle 3 from june 2006 to may 2007 .
the data were taken in ` staring mode ' at two nod positions along each irs slit .
thanks to the anticipated high brightness of the sources , no specific off - set observations for measuring the background were needed .
the journal of observation is shown in table[tab : observation ] .
we start the data processing from the _ spitzer science center s _ pipeline data set ( version s16.1.0 ) .
the @xmath14 program is used to combine unflatfielded image ( droopres ) and the rogue pixels image from the same campaign for a compound mask , afterwards this combined mask is applied to the flatfielded image ( bcd ) to remove rogue and flagged pixels as much as possible .
the spectra are extracted from the cleaned image in the _ spice _ software package and further analysis is done in the _ smart _ ( version 6.2.6 ) package . since the lh aperture ( [email protected]@xmath16 ) is larger than the sh aperture ( [email protected]@xmath16 ) and it is likely that some sources are so extended that unlike the lh slit , the sh slit could not cover the entire emission area , one sometimes need to multiply a scaling factor in order for the sh flux to agree with the flux of the lh part .
the consequence of the aperture difference shows up clearly in iras23239 + 5754 , for which the sh spectrum must be multiplied by a factor of @xmath55.2 to agree with the lh flux at the overlap part and for the whole spectrum to be consistent with the previous _ msx _ ( midcourse space experiment ) and _ iras _ photometry .
as shown in figure [ fig : aor ] , for iras23239 + 5754 the sh slit covers less than half of the emission range while the lh slit covers most of the emission area , confirming a factor of @xmath175.2 difference .
the flux matching between the sh part and the lh part is much better for iras16456 - 3542 , and 18354 - 0638 .
we just need a scale factor of @xmath51.21.3 .
the reduced spectra of these three sources are displayed in figure [ fig : spectra ] , where the objects are ordered in terms of their evolutionary classes : post - agb stars and pne . based on their ionized nebular emission lines ,
we classify iras16456 - 3542 and 23239 + 5754 as pne , consistent with the previous classifications in the literature . for the sources lack of nebular lines ,
they are commonly considered either as post - agb stars or as extremely evolved agb stars . in this context
we classify iras18354 - 0638 as a post - agb star ( see 3.1 ) .
the irs wavelength range between 9.9@xmath0 and 37.2@xmath0 is full of spectral features arising from crystalline silicates . in order to clearly illustrate the crystalline silicate features
, we try to subtract the underlying continuum . instead of adopting a single - temperature blackbody spectrum
, we simply fit the global continuum by a mathematical spline function .
this approach was taken by in a comprehensive analysis of the iso spectra of 17 oxygen - rich circumstellar dust shells surrounding evolved stars . for an overview of the ir spectral energy distribution ( sed ) , previous photometric data , in the _
2mass _ j band , _ msx _ a band ( msxb1 ) , b@xmath2 ( msxb2 ) , c ( msxc ) , d ( msxd ) , and e ( msxe ) bands respectively peak at 8.28 , 4.29 , 4.35 , 12.13 , 14.65 , and 21.34@xmath0 @xcite . ] and _ iras _
60@xmath0 band , are supplemented .
the continuum is then maximized and subtracted . in principle , the strength of the features might be underestimated since there is a possibility that very broad features are mistaken as continuum . after this global continuum subtraction ,
one more step is taken to remove the `` local continuum '' by a third - order polynomial .
the so - called `` local continuum '' refers to the part that is above the global continuum while below the minimum of the features .
this step would possibly further underestimate the strengths of the features .
figure [ fig : continuum ] is an example for subtracting the continuum underneath the spectral features .
figure [ fig : features ] displays the continuum - subtracted spectral features of these three sources ( iras16456 - 3542 , 18354 - 0638 , and 23239 + 5754 ) .
table[tab : feature ] lists the peak wavelengths and fwhm ( full width half maximum ) of the crystalline silicate features derived from a gaussian fitting .
the spectral features are compared with the iso spectrum of ngc6302 , a pn .
the features previously identified in the molster sample are labeled by thin vertical lines . probably due to the high resolution and sensitivity of _ spitzer_/irs , several `` new features '' ( labeled by thick vertical lines in figure [ fig : features ] )
appear to be present in the irs spectra of these sources : ( 1 ) the 11.6@xmath0 feature seen in iras16456 - 3542 , ( 2 ) the 14.4@xmath0 feature ( or the 14.6@xmath0 feature ) seen in iras16456 - 3542 and iras23239 + 5754 , ( 3 ) the 15.4@xmath0 feature seen in iras23239 + 5754 and in iras16456 - 3542 ,
feature reported in . ] and ( 4 ) the 25.47@xmath0 feature in iras23239 + 5754 ( or the 25.85@xmath0 feature in iras16456 - 3542 ) .
these `` new features '' have not previously been reported in the literature .
a more detailed study of a large sample of both astronomical and laboratory crystalline silicate spectra is necessary for confirming these `` new '' features .
the three sources which show abundant features of crystalline silicates are iras16456 - 3542 , 18354 - 0638 , and 23239 + 5754 .
we discuss these objects below , in the order of their evolutionary stages . 1 . _
iras18354 - 0638_. the _
spitzer_/irs spectrum of this source exhibits one nebular line ( i.e. [ siii ] at 33.5@xmath0 ) which may suggest the start of a nebular phase .
we hesitate to classify it as a pn because pne usually exhibit @xmath13 nebular lines ( while iras18354 - 0638 displays only one emission line ) .
we classify it as a post - agb star . in comparison with the other two objects ( iras23239 + 5754 and 16456 - 3542 ; see below ) which display abundant crystalline silicate features and are classified as pne , this object seems to be at an earlier stage of evolution ( i.e. extreme agb or post - agb ) .
iras23239 + 5754_. this object is a very young bipolar pn with a complex inner structure inside the bipolar lobes .
the effective temperature of the central star is about @xmath18 @xcite .
the mean temperature of the bulk dust in the circumstellar shell is @xmath17200k , as estimated from the overall ir spectral energy distribution .
several nebular lines ( e.g. , [ siv ] at 10.5@xmath0 , [ neii ] at 12.8@xmath0 , [ siii ] at 18.7@xmath0 , [ siii ] at 33.48@xmath0 , and [ neiii ] at 36.0@xmath0 ) clearly show up in its irs spectrum .
this source is well known for its pure fluorescent h@xmath2 emission in the near ir wavelength region @xcite .
it has long been suspected to be a binary .
+ also known as hubble 12 at a distance of @xmath173kpc @xcite , this pn , in particular its multiple coaxial rings in the whole - scale bipolar structure , has been well studied in the literature ( see e.g. @xcite and references therein ) .
the shape of this object reveals that it is probably at a very early pn phase , with a kinematic age of @xmath5300 years .
the multiple rings extend about 4@xmath19 8@xmath16 .
this size is comparable to the aperture of the irs sh slit ( 3.7@[email protected]@xmath16 ) and much smaller than the irs lh slit ( 11.1@[email protected]@xmath16 ) . as discussed in 2
, this explains the discrepancy between the sh and lh fluxes .
iras16456 - 3542_. this object is also known as pn g347.4 + 05.8 and h1 - 2 .
its central star is a `` weak emission - line star '' ( wels ; ) . with a dynamical age of @xmath51400 years
, it is a young pn ( but more evolved than iras23239 + 5754 ) .
@xcite derived a distance of @xmath52.47kpc , but larger distances have also been suggested ( e.g. , @xmath56.96kpc [ @xcite ] , @xmath59.01kpc [ @xcite ] ) .
in addition to the crystalline silicate bands , this object also emits at the 11.3 , 12.7 and 16.4@xmath0 pah bands ( see figure [ fig : spectra ] ) . to derive the mass percentage of crystalline silicates
, we follow to apply a simple dust emission model to fit the observed irs spectra ( including the continuum underlying the crystalline silicate features ) .
this model assumes that the dust shell is optically thin in the ir .
the sed is then fitted with mass absorption coefficients @xmath21 multiplied by blackbody functions @xmath22 : @xmath23 ~~.\ ] ] the dust species considered here include : ( 1 ) amorphous silicate , ] ( 2 ) forsterite , and ( 3 ) enstatite . for each species , we consider both a cool component and a warm component respectively at temperature @xmath24 and @xmath25 .. we start with a single - temperature component and then add a component at another temperature to see if the reduced @xmath26 significantly drops . ]
the mass of dust species @xmath27 , @xmath28 , is a multiplication factor related to the total mass and only represents the relative mass ratio .
the absorption coefficient of amorphous silicate is calculated from the optical constants of with mie theory , assuming the dust to be spherical with radii @xmath29=0.1@xmath0 , the typical size of interstellar dust .
amorphous silicate band in our sources is better fitted in terms of small - sized grains than large ones . ] for the minerals ( forsterite and enstatite ) , the optical properties measured in laboratory depend on the size , shape , structure and chemical composition of the dust analog .
we adopt those of @xcite because their coefficients closely resemble the irs spectra of our sources .
the relative mass of each component ( amorphous silicate , forsterite and enstatite ) is treated as a free parameter .
for enstatite , we consider two sub - types : clino - enstatite and ortho - enstatite . following , we assume equal mass of ortho- and clino - enstatite . and 21.7@xmath0 .
the relative strengths of these two features show a notable difference between clino - enstatite and ortho - enstatite : for clino - enstatite they are about equally narrow , while for ortho - enstatite the 21.7@xmath0 feature is stronger and broader than the 20.65@xmath0 feature .
the 20.65@xmath0 and 21.7@xmath0 features are comparably narrow in the three sources considered here and in afgl4106 , the source in which these two features were first identified by . ]
we take the levenberg - marquart minimization algorithm to obtain the best - fit .
the results of the sed fitting are shown in figure [ fig : modelfit ] and the parameters are listed in table[tab : model ] .
the mass percentage of crystalline silicates is @xmath5@xmath30 , @xmath31 and @xmath32 respectively for iras18354 - 0638 ( see footnote [ ftnt : crystfrac ] ) , 23239 + 5754 and 16456 - 3542 . except for iras18354 - 0638
, the fraction of crystalline silicate is much higher than the average of @xmath510% in evolved stars , and the amount of crystalline silicate even exceeds amorphous silicate . in iras16456 - 3542 , the fraction of crystalline silicate reaches @xmath597% , which exceeds the highest crystalline fraction measured so far from the crystalline silicate _ emission _
spectra ( @xmath575% in iras09425 - 6040 ) [email protected] toward pks1830 - 211 , a gravitationally lensed quasar @xcite .
the crystalline fraction derived from the _ absorption _ spectrum of this object may be @xmath3495% .
] even when we take into account the uncertainty ( @xmath355% ) in the modeling , the fraction of crystalline silicate in this source is one of the highest .
we conclude that there must be some evolved stars in which the silicates are even completely crystallized . under the assumption that the crystalline silicate in the evolved stars has only two forms : enstatite and forsterite ,
the mass ratio of the two forms @xmath36 is approximately 4 , 54 and 24 respectively in these three sources , implying that the main form of crystalline silicate is enstatite i.e. mgsio@xmath37 .
enstatite accounts for more than two - thirds of the crystalline silicate dust .
this result is consistent with who found the enstatite to forsterite mass ratio is between @xmath51 and @xmath511.4 for 12 evolved objects .
the model fit presented in [ sec : model ] shows that although most of the crystalline silicate features can be fitted reasonably well in terms of two silicate mineral types ( i.e. forsterite and enstatite ) , some features are not well fitted ( see figure [ fig : modelfit ] ) .
most noticeably , the feature at 32.7@xmath0 , prominent in all three sources and particularly strong in iras18354 - 0638 ( in which the 32.7@xmath0 feature is nearly as strong as the 33.5@xmath0 feature ) , is not accounted for by the current model consisting of forsterite and enstatite minerals .
similarly , the model fails in predicting the features at 29.6 , 30.6 and 31.1@xmath0 seen in all three sources ( see figure [ fig : modelfit ] ) .
we note that the 29.6 , 30.6 and 31.1@xmath0 features seem to comprise a group as they have similar relative intensities in all three sources , with the 29.6@xmath0 feature being the strongest and the 31.1@xmath0 feature being the weakest .
such a phenomenon was also seen in the larger sample of .
one may speculate that these features arise from a dust mineral other than enstatite and forsterite .
it is also interesting to note that the overall spectral features at 1035@xmath0 of iras23239 + 5754 and iras16456 - 3542 are similar to each other ( probably because they are both very young pne ) , except that the 29.6@xmath0 feature is much stronger in the older pn iras16456 - 3542 than that in the younger pn iras23239 + 5754 . in summary ,
the modeling approach discussed in [ sec : model ] assumes three grain types : ( 1 ) spherical amorphous silicates of radii @xmath29=0.1@xmath0 with the indices of refraction taken from ; ( 2 ) crystalline forsterite and ( 3 ) crystalline enstatite for which the mass absorption coefficients are taken from @xcite . for each grain type ,
two temperatures are assumed ( i.e. a cold component and a warm component ) unless a single temperature model results in more or less the same @xmath26 . to examine whether the silicate crystallinity derived in
[ sec : model ] is dependent on the modeling technique , we take a different approach : we follow the approach documented in detail by @xcite and incorporate more sizes and more silicate mineral compositions . the detailed mineralogy of the dust is determined by comparing the observed _
spitzer_/irs spectrum to a wide range of plausible silicate minerals with high - quality mid - ir transmission and thermal emission laboratory spectra .
the calculated model flux from each mineralogical component is the product of a blackbody of characteristic temperature times the experimentally determined absorption efficiency for each sized particle , summed over all particles in the dust population . for illustration , in figure
[ fig : lissemodel ] we show the _
spitzer_/irs spectrum of iras16456 - 3542 and the model spectrum as well as the appropriately scaled contributions from each grain species .
the overall fit is reasonably good , although not as good as that described in
[ sec : model ] ( see figure [ fig : modelfit ] ) .
the derived silicate crystallinity is @xmath590% , consistent with the silicate crystallinity of @xmath597% derived in
[ sec : model ] . finally , we note that various techniques have been proposed in the literature to fit the crystalline silicate emission spectra ( e.g. , ) .
all these approaches treat dust size , composition and temperature as free parameters .
this is physically _ incorrect _ ( see @xcite ) .
the equilibrium temperature of dust is sensitive to its size and composition : smaller grains have higher temperatures , and silicates with a higher fe / mg ratio have higher temperatures .
the dust temperature is also sensitive to the distance from the dust to the illuminating source ( see @xcite ) .
therefore , one can not treat the dust temperature as an adjustable parameter independent of the dust composition and size .
the dust mass is very sensitive to the dust temperature .
the crystalline fraction derived from all the approaches in which the dust temperature is treated as a free parameter should be taken with caution .
the correct approach should physically calculate the dust temperature for a given dust size of a given composition at a given distance from the illuminating source ( e.g. , see ) .
we report the 9.937.2@xmath0 spectra of three evolved objects ( post - agb stars and pne ) using the high - resolution mode of the _ irs _ instrument on board _
spitzer_. these objects exhibit rich sets of crystalline silicate emission features .
these features are fitted by a mixture of three dust species : amorphous silicate , crystalline forsterite , and crystalline enstatite , with each species consisting of a warm component and a cold component .
it is found that the crystalline silicates are dominated by crystalline enstatite .
the crystalline silicate spectral modeling shows that iras16456 - 3542 , has a degree of silicate crystallinity ( i.e. mass fraction of silicate dust in crystalline form ) of @xmath597% , the highest to date in crystalline silicate sources . , l. b. f. m. , cami , j. , de jong , t. , molster , f. j. , van loon , j. t. , bouwman , j. , de koter , a. , waelkens , c. , van winckel , h. , morris , p. w. , cami , j. , bouwman , j. , de koter , a. , de jong , t. , & de graauw , t. 1998 , , 391 , 868 , l. b. f. m. , molster , f. j. , de jong , t. , beintema , d. a. , waelkens , c. , boogert , a. c. a. , boxhoorn , d. r. , de graauw , t. , drapatz , s. , feuchtgruber , h. , genzel , r. , helmich , f. p. , heras , a. m. , huygen , r. , izumiura , h. , justtanont , k. , kester , d. j. m. , kunze , d. , lahuis , f. , lamers , h. j. g. l. m. , leech , k. j. , loup , c. , lutz , d. , morris , p. w. , price , s. d. , roelfsema , p. r. , salama , a. , schaeidt , s. g. , tielens , a. g. g. m. , trams , n. r. , valentijn , e. a. , vandenbussche , b. , van den ancker , m. e. , van dishoeck , e. f. , van winckel , h. , wesselius , p. r. , & young , e. t. 1996 , , 315 , l361 , l. b. f. m. , voors , r. h. m. , morris , p. w. , trams , n. r. , de koter , a. , & lamers , h. j. g. l. m. 1999 , in lecture notes in physics , berlin springer verlag , vol .
523 , iau colloq . 169 : variable and non - spherical stellar winds in luminous hot stars , ed
. b. wolf , o. stahl , & a. w. fullerton , 381 , a. a. , matsuura , m. , wood , p. r. , sloan , g. c. , lagadec , e. , van loon , j. t. , groenewegen , m. a. t. , feast , m. w. , menzies , j. w. , whitelock , p. a. , blommaert , j. a. d. l. , cioni , m .-
l. , habing , h. j. , hony , s. , loup , c. , & waters , l. b. f. m. 2006 , , 370 , 1961 18354@xmath380638 & post - agb & 8.55 & 5.85 & 4.51 & 9.10 & 19.60 & 24.40 & 49.10 & 18.90 & 68.00 & 39.20 + 23239 + 5754 & young pn & 10.23 & 9.83 & 8.81 & 10.99 & 16.95 & 18.53 & 50.81 & 20.86 & 71.22 & 35.79 + 16456@xmath383542 & pn & 11.83 & 11.48 & 10.53 & 2.26 & 3.84 & 7.06 & 22.19 & 4.88 & 29.61 & 10.96 + [ tab : infrared ] 1 & 18354@xmath380638 & [email protected] & 04/20/07 & s16.1.0 & + 2 & 23239 + 5754 & [email protected] & 09/09/06 & partial s16.1.0 & poor lh nod2 + 3 & 16456@xmath383542 & [email protected] & 03/25/07 & partial s16.1.0 & + [ tab : observation ] & & 10.63 & 0.11 & 10.60&0.24 + & & 11.26 & 0.39 & 11.22&0.17 + & & & & * 11.61*&*0.27 * + & & 13.54 & 0.22 & 13.55&0.25 + & & 13.75 & 0.25 & 13.75&0.23 + & & 14.18 & 0.12 & & + & & * 14.40 * & * 0.13 * & * 14.39*&*0.26 * + & & * 14.63 * & * 0.08 * & * 14.61*&*0.19 * + & & _ 15.45 _ & _ 0.33 _ & 1__5.39__&_0.31 _ + & & 16.10 & 0.60 & 16.10&0.30 + & & 17.47 & 0.99 & 17.43&0.39 + & & 18.00 & 0.53 & 18.96&0.42 + 19.45 & 0.83 & 19.59 & 0.54 & 19.61&0.50 + 20.65 & 0.34 & 20.70 & 0.31 & 20.65&0.29 + 21.66 & 0.47 & 21.60 & 0.45 & 21.60&0.53 + & & 23.09 & 0.54 & 23.07&0.50 + 23.21 & 1.21 & 23.74 & 0.49 & 23.68&0.53 + 24.49 & 0.23 & 24.58 & 0.39 & 24.52&0.24 + & & * 25.47 * & * 0.55 * & * 25.85*&*0.30 * + 27.73 & 0.85 & 27.72 & 1.30 & 27.88&1.14 + 29.19 & 0.85 & 29.53 & 0.86 & 29.58&0.66 + 30.54 & 0.29 & 30.59 & 0.36 & 30.57&0.27 + 31.11 & 0.19 & 31.12 & 0.14 & 31.13&0.14 + 32.78 & 0.95 & 32.81 & 0.50 & 32.80&0.42 + 33.57 & 0.88 & 33.80 & 0.93 & 33.72&0.62 + 35.41 & 1.36 & 35.64 & 1.55 & 35.55&1.14 + ccccccccc 18354@xmath380638 & 79 & @xmath39 & 87 & @xmath39 & 85 & 213 & 4 & @xmath40 + 23239@xmath415754 & @xmath39 & 129 & 81 & @xmath39 & @xmath39 & 150 & 54 & @xmath42 + 16456@xmath383542 & 67 & 125 & 72 & 144 & @xmath39 & 158 & 24 & @xmath43 + rich crystalline silicate features in iras16456 - 3542 , 18354 - 0638 , and 23239 + 5754 . the features already listed in are labeled with thin solid black line . the `` new features '' not reported in the literature are labeled with thick solid red line .
those with noticeable shifts of peak wavelength ( compared to that of ) are labeled with dotted red line . ]
comparison of the model spectra ( red solid line ) with the _ spitzer_/irs spectra ( gray solid line ) .
three types of silicates are considered : forsterite ( blue dashed line ) , enstatite ( ortho - enstatite and clino - enstatite ; green dot - dashed line ) , and amorphous silicate ( red dotted line ) . for iras18354 - 0638
we consider amorphous carbon instead of amorphous silicate .
[ fig : model ] ] modeling the _ spitzer_/irs spectrum of iras16456 - 3542 with the approach of @xcite which incorporates a power - law size distribution of @xmath5@xmath46 with 0.1@xmath0@xmath35@xmath29@xmath3510@xmath0 for the grain size and various crystalline and amorphous olivine and pyroxene grain materials .
the solid lines beneath the irs spectrum show three components : the upper red and green lines for amorphous carbon and carbonates , the middle cyan lines for pyroxenes and sulfides , the lower blue lines for olivines . ] | we report the _ spitzer _ infrared spectrograph ( irs ) observations of three evolved stars : iras16456 - 3542 , 18354 - 0638 , and 23239 + 5754 . the 9.937.2@xmath0 _
spitzer_/irs high resolution spectra of these three sources exhibit rich sets of enstatite - dominated crystalline silicate emission features .
iras16456 - 3542 is extremely rich in crystalline silicates , with @xmath190% of its silicate mass in crystalline form , the highest to date ever reported for crystalline silicate sources . |
most resonance lines of important ions in the local interstellar medium ( lism ) lie in the ultraviolet ( uv ) spectral range .
this paper continues our inventory of lism absorption line observations observed by the two high resolution spectrographs onboard the _ hubble space telescope _ ( _ hst _ ) , the goddard high resolution spectrograph ( ghrs ) and the space telescope imaging spectrograph ( stis ) . in this paper , we extend the work of @xcite , who analyzed all available lism and absorption lines observed by _ hst _ and lines observed from the ground , along sightlines @xmath1100pc , to include six more ions : , , , , , and . these ions , which all fall in the 1200 - 1800 spectral range , extend the resonance line database to include a range of ion masses from atomic weight 2.01 , , to atomic weight 55.85 , .
this collection of lism absorption line observations will form the basis for a comprehensive analysis of many of the physical properties of the lism , which will be published in subsequent papers of this series .
the format of this work is similar to that of @xcite , and we refer the reader to that paper for a thorough introduction to our procedure for the analysis of interstellar absorption lines .
the most important difference between these two papers is our inclusion of moderate resolution spectra for the analysis of low mass ions .
the velocity structure along a line of sight is best measured by resonance lines of the heaviest ions observed with high spectral resolution , as discussed by @xcite .
we use the kinematic information obtained from the high resolution spectra of the high mass ions when fitting the moderate resolution spectra of the lighter ions . because of the cosmological and galactic chemical evolution implications of the abundance of deuterium in the lism @xcite
, measurements have already been published for many nearby stars , and 11 more measurements are presented in this paper .
the other light ions : , , , , and have not been analyzed for the majority of sightlines with available spectra .
although a few sightline investigations have already been presented @xcite , this paper provides a valuable new database of lism measurements of these ions which is complete to date . by coupling the datasets presented here with those of and @xcite
, we can investigate fundamental physical properties of the lism , including : the morphology @xcite , small scale structure @xcite , ionization structure @xcite , and the temperature and turbulent velocity structure @xcite .
we list in table [ sw_table1 ] all stars located within 100 pc that have moderate ( @xmath2 ) or high ( @xmath3100,000 ) resolution spectra of lism absorption lines in the 1200 - 1800 spectral range .
we include in this sample only those sightlines that also have high resolution observations of and/or to ensure that we have an accurate measurement of the kinematic structure along the line of sight @xcite
. is not used alone , as a kinematic constraint on other uv lines , because low column density clouds not detected in are observed in stronger uv resonance lines ( e.g. the @xmath4 cen line of sight ) .
resonance lines that show lism absorption in the 1200 - 1800 spectral range include : the fine structure doublet at 1215.3376 and 1215.3430 , at 1334.532 , the triplet at 1199.5496 , 1200.2233 , and 1200.7098 , at 1302.1685 , at 1670.7874 , and five resonance lines at 1190.4158 , 1193.2897 , 1260.4221 , 1304.3702 , and 1526.7066 .
because some sightlines with light ion absorption spectra included recent high resolution observations of and , we have included the analysis of new heavy ion absorption lines in appendix [ app ] .
these measurements complement the and survey published by @xcite . for all of the stars , the stellar spectral type , visible magnitude , stellar radial velocity , and galactic coordinates
are taken from the simbad database unless otherwise noted .
the _ hipparcos _
distances @xcite are listed without errors , but the errors in trigonometric parallax measurements for stars within 100pc are small enough to have no influence on the interpretation of our results . in general , at 1 pc the @xmath5 error in the distance is @xmath6 pc , at 30 pc the @xmath5 error is @xmath7 pc , and at 100 pc the @xmath5 error is @xmath8 pc .
also listed are the predicted absorption velocities for the local interstellar cloud ( lic ) and galactic ( g ) clouds computed using the vectors proposed by @xcite and @xcite .
all observations were taken with the high resolution spectrographs onboard the _ hst _ : the ghrs and the stis instruments .
the ghrs instrument is described by @xcite and @xcite .
@xcite describe improvements to the ghrs instrument for observations taken after the installation of the corrective optics space telescope axial replacement ( costar ) in december 1993 .
the stis instrument is described by @xcite and by @xcite .
many _ hst _ observations of the stars listed in table [ sw_table1 ] were taken for the express purpose of studying the structure of the lism along these sight lines . however , more than a third of the observations along lines of sight listed in table [ sw_table1 ] were taken for other purposes .
we list in table [ sw_table2 ] the _ hst _ observations of those stars with lism absorption data that have not been published .
all of these data were taken from the _ hst _ data archive and are publicly available .
the data reduction procedure for the observations listed in table [ sw_table2 ] is described in section [ sw_datareduc ] , and the spectral analysis is discussed in section [ sw_specanal ] .
we reduced the ghrs data acquired from the _ hst _ data archive with the calhrs software package using the image reduction and analysis facility ( iraf ) and the space telescope science data analysis system ( stsdas ) software .
the most recent reference files were used in the reduction . in many cases
, the current reference files produced an improvement in data quality , compared to data reduced with reference files contemporaneous with the observations .
many of the echelle observations were obtained in the fp - split mode to reduce fixed - pattern noise .
the individual readouts of the fp - split spectra were combined using a cross - correlation procedure called hrs_merge @xcite .
the reduction included assignment of wavelengths using calibration spectra obtained during the course of the observations .
the calibration spectra include either a wavecal , a direct pt - ne lamp spectrum from which a dispersion relation can be obtained , or a spybal ( spectrum y - balance ) , from which only a zero - point offset can be obtained .
any significant errors involved in the wavelength calibration are included in our central velocity determinations .
we reduced the stis data acquired from the _ hst _ data archive using the stis team s calstis software package written in idl @xcite .
the reduction included assignment of wavelengths using calibration spectra obtained during the course of the observations .
the echelle_scat routine in the calstis software package was used to remove scattered light .
figures [ sw_fig1]-[sw_fig6 ] show our fits to the lism absorption lines that were observed in the 1200 - 1800 spectral range and have not yet been published ( see table [ sw_table2 ] for a list of the observational parameters ) .
all of the spectra are shown with a heliocentric velocity scale , together with our estimate of the stellar continuum ( _ thin solid lines _ ) , the best fits to the absorption by each interstellar component ( _ dashed lines _ ) , and the total interstellar absorption convolved with the instrumental profile ( _ thick solid lines _ ) .
table [ sw_table2_5 ] presents an inventory of detected lism absorption toward all 41 stars within 100pc that have both a high resolution heavy ion detection and at least one moderate to high resolution detection of a light ion .
we differentiate the new measurements presented in this work , with those already analyzed in the literature . of the 41 sightlines listed in table [ sw_table2_5 ] ,
33 have new lism absorption line observations .
the lism absorption lines are fit using standard techniques @xcite .
this involves estimating the stellar continuum , determining the fewest number of absorption components required to obtain a satisfactory fit , convolving the absorption feature with an instrumental line spread function , and fitting all lines of the same ion simultaneously . determining the stellar continuum is usually straightforward when dealing with high resolution data , and systematic errors in this procedure
are further reduced when several resonance lines of the same ion are fit simultaneously . because the central wavelength of the interstellar absorption is different for each line of sight
, the placement of the unobserved stellar flux for some cases can be both difficult and the dominant source of error . in most cases ,
the `` continuum '' against which the absorption is measured can be determined by fitting a simple polynomial to spectral regions just blueward and redward of the interstellar absorption .
of the 206 interstellar absorption spectra shown in figures [ sw_fig1]-[sw_fig6 ] ( as well as those shown in appendix [ app ] ) , 68% of the continua were determined in this straightforward way .
occasionally , a simple polynomial is not sufficient to accurately estimate the missing stellar continuum .
we employ three additional techniques to accurately estimate the continuum : mirroring , similarity , and simultaneous fitting .
the mirroring technique for estimating the unobserved stellar flux involves reflecting the emission line about the stellar photospheric radial velocity .
if the interstellar absorption is far from line center , typically because of the high radial velocity of the star , this technique can be very useful in estimating the stellar flux . since these circumstances are not very common , we used this method for estimating the continuum for only 5% of the sample . fitting the line of 36 oph a and 70 oph
( see figure [ sw_fig1 ] ) are examples of this technique .
a stellar emission line with lism absorption is often a member of a multiplet .
because the other members of the multiplet are not resonance transitions , no lism absorption is seen against these emission lines .
the shape of this line can be used to help reconstruct the continuum of the absorbed resonance line .
this is particularly useful with lines that may or may not exhibit self - reversals at line center or multiple gaussian profiles , such as , , and , as well as lines that are severely blended with additional lines or a rising stellar continuum , such as .
although the line width and integrated flux vary among multiplet members , the shapes of the line profiles are often very similar .
examples of this technique are the lines of @xmath9 eri ( see figure [ sw_fig1 ] ) , and the lines of @xmath10 gem , @xmath10 cet , and @xmath11 cap ( see figures [ sw_fig2 ] , [ sw_fig3 ] , and [ sw_fig5 ] , respectively ) .
this method is used for estimating 13% of the continua in this sample .
one of the strengths of this sample , is that there are often several absorption lines for the same ion , particularly , , , and . because the oscillator strengths are known , the measured absorption in one line , predicts the absorption for all of the other resonance lines of the same ion . in this way
, we can estimate continua for low signal - to - noise ( s / n ) absorption features in moderate resolution data for an ion , as long as we have at least one other reasonably simple absorption feature of the same ion . by simultaneously fitting all of the absorption features of the same ion
, we can make a reasonable estimate of the continuum that is consistent with each absorption line .
although this technique can cause some curious continua estimates , it is the best method for reducing the systematic errors involved in continua placement . in only a few cases , approximately 14% ,
does this method produce results that deviate from a simple polynomial continuum estimation .
some examples include the lines of gd 153 ( see figure [ sw_fig5 ] ) and the lines of hd 184499 . once the continuum has been determined , we apply the minimum number of absorption components for a satisfactory fit to the spectra . adding more absorption components
invariably improves the goodness - of - fit metric ( @xmath12 ) .
we use the f - test to determine whether the improvement in the fit is statistically significant @xcite , and allows us to determine the minimum number of necessary absorption components in an objective way .
the instrumental line spread functions assumed in our fits for ghrs spectra are taken from @xcite and for stis spectra from @xcite .
the rest wavelengths and oscillator strengths of the lines used in our fits are taken from @xcite with updates from d.c .
morton ( 2003 , private communication ) .
we fit all lines of each ion simultaneously in order to obtain the most robust parameters for the interstellar absorption .
the difference in oscillator strengths between the various lines can be very useful in constraining the interstellar parameters , the stellar continuum level , and the number of absorption components . by providing an independent measure of the same absorbing column using absorption lines with different optical depths and at different locations relative to the stellar continuum
, our analysis of several lines of the same ion reduces the effect of systematic errors , in particular from saturation and line blends , that result from the measurement of only one line .
occasionally , absorption features due to terrestrial material are detected in spectra of and ( see hz43 , gd153 , and dkuma , in figures [ sw_fig4]-[sw_fig6 ] ) .
the absorption is easily identified because it is centered at the mean velocity of the earth during the time of observation , and because it is often also seen in the other members of the multiplet , where no lism absorption is detected .
the airglow absorption features are easily modeled and do not adversely affect the fit to the lism absorption lines . for each absorption component
there are three fit parameters : the central velocity ( @xmath13 [ km s@xmath14 ) , the doppler width ( @xmath15 [ km s@xmath14 ) , and the column density ( @xmath16 [ @xmath17 ) .
the central velocity corresponds to the mean projected velocity of the absorbing material along the line of sight to the star .
if , as a first approximation , we assume that the warm partially ionized material in the solar neighborhood exists in small , homogeneous cloudlets each moving with a single bulk velocity , then each absorption component will correspond to a single cloudlet , and its velocity is the projection of its three - dimensional velocity vector along the line of sight .
the doppler width parameter is related to the temperature ( @xmath18 [ k ] ) and nonthermal velocity ( @xmath19 [ km s@xmath14 ) of the interstellar material by the following equation : @xmath20 where @xmath21 is boltzmann s constant , @xmath22 is the mass of the ion observed , and @xmath23 is the atomic weight of the element in atomic mass units ( where @xmath23 can range from @xmath24 to @xmath25 ) .
the doppler width parameter , or line width , of an ion with a large atomic weight is more sensitive to turbulent broadening than thermal broadening .
the column density measures the amount of material along the line of sight to the star .
if we assume homogeneous cloudlets with constant density , then the column density will be directly proportional to the cloud thickness . the compilation of all measurements for stars within 100 pc provides an important database for the analysis of the structure of the lism .
the velocity structure is discussed in section [ sw_veldist ] , the doppler width parameter structure is discussed in section [ sw_doppdist ] , and the column density structure is discussed in section [ sw_coldist ] . in order to extract as much information as possible from the uv observations of lism absorption lines , we apply knowledge of the velocity structure along the line of sight derived from high resolution spectra of intrinsically narrow absorption lines of and to the fitting of the moderate resolution spectra of the lower mass , and thus , intrinsically broader absorption lines . for example , an absorption line formed by interstellar material at the canonical lism temperature , @xmath26k , and with no turbulent broadening will have a doppler line width of @xmath27kms@xmath28 , whereas the absorption line for the same conditions will have @xmath29kms@xmath28 . to fully resolve the interstellar absorption feature requires a spectral resolution of @xmath30200,000 , whereas the feature requires a resolution of @xmath3140,000 .
if the heavy ions accurately trace the same collection of gas as the light ions , then the required resolving power can decrease , as the atomic weight of the ion decreases , provided that the kinematic structure along the line of sight has already been determined from the analysis of intrinsically narrow lines observed at high resolution . in this paper
, we take advantage of such high resolution observations of and lines presented by @xcite to fit moderate resolution data of the same sightlines . in particular
, we fit the same number of velocity components to the light ion spectra that were necessary in the high resolution spectra . in practice , the vast majority of light ion absorption spectra have enough spectral resolution and signal - to - noise to independently constrain the number of components and their central velocities .
however , in cases where the absorption components are severely blended , or the s / n is particularly poor , we use information from the high resolution fits to and . in only two sightlines , 70 oph ( , , and ) and hd 28568 ( ) ,
the absorption features are so blended that we use the velocity information directly from the data to fit these spectra . in a handful of additional cases ,
the absolute velocity structure is not constrained , only the differences in central velocities are taken from the high resolution fits to constrain the light ion spectra .
tables [ sw_table3 ] through [ sw_table8 ] list the interstellar absorption parameters and @xmath5 errors for all targets in table [ sw_table1 ] that show interstellar absorption in , , , , , and , respectively .
the data in tables [ sw_table3]-[sw_table8 ] include our fits to the lines of sight shown in figures [ sw_fig1]-[sw_fig6 ] .
all errors quoted in the text , tables , and figures are @xmath5 error bars . in table
[ sw_table3 ] we list 1@xmath32 errors in @xmath33 that for the best cases are as small as 0.01 . for those lines of sight with only one velocity component and high s
/ n spectra , we can reliably interpolate the line profile in the neighborhood of the line to achieve this precision .
observations with moderate s / n , and/or several velocity components detected along the line of sight , typically have uncertainties in @xmath33 on the order of 0.1 .
the following discussion will be primarily a statistical analysis of the lism absorption fit parameters discussed above . in future papers we will present more detailed analyses of the database , including the morphology , kinematics , and physical properties of identifiable structures ( i.e. clouds ) in the lism .
the location of all stars listed in table [ sw_table1 ] are shown in galactic coordinates in figure [ sw_fig9 ] .
although spatial coverage overall is fairly good , there are areas of the sky that are poorly sampled , while others are densely sampled .
because the focus of this paper is the fitting of relatively broad absorption lines of low mass ions with moderate resolution data , the measurement of velocity centroids is unlikely to improve the accuracy obtained from fits to the high resolution spectra of narrow absorption lines .
however , our analysis of the low mass ion profiles provides an opportunity to measure the consistency of the observed velocity centroids for a particular absorber . by comparing the measured absorption velocities of many resonance lines along the same sightline
, we can test : ( 1 ) the implicit assumption that all of the observed ions sample the same collection of gas , and will therefore have the same observed projected velocity , and ( 2 ) the magnitude of systematic errors involved in comparing observations obtained with different spectral resolutions and observed with different instruments ( e.g. ghrs or stis ) .
ionization models of the lism by @xcite indicate that all of the measured ions discussed in this paper are expected to be the dominate ion of the element , and therefore we expect them all to sample the same collection of gas .
each individual sightline now includes multiple measurements with different ions of the projected velocity of each interstellar absorber ( cloud ) along the line of sight .
figure [ sw_fig10 ] shows the distribution of the standard deviation about the weighted mean of these velocity measurements for each particular absorber .
if all ions sample the same collection of gas and all of our instruments were infinitely precise and perfectly calibrated , then the standard deviation should be @xmath34 .
although there is a distribution of standard deviations , it is strongly peaked at 0 .
therefore , all of the measured ions for each cloud appear to sample the same collection of gas .
the dashed line indicates a gaussian fit to the distribution , which well characterizes the data .
the gaussian has a standard deviation of @xmath35kms@xmath28 .
we believe that this is a good estimate of the typical systematic error involved in measuring lism absorption lines in moderate and high resolution spectra observed with the ghrs and stis instruments . because we are comparing a large number of individual observations , each with its own systematic errors particular to each absorption feature ( i.e. continuum placement , degree of blending with other lines , etc ... ) , the distribution of systematic errors approaches a random distribution .
therefore , the range of standard deviations for all of the central velocity measurements of a particular absorber ( cloud ) should and does approximate a gaussian distribution with a rather small half width . the observed value of the doppler width parameter ( @xmath15 ) is a measure of the temperature ( @xmath18 ) and turbulent velocity ( @xmath19 ) .
the relationship is provided by equation [ sw_eq1 ] . with increasing atomic mass ,
the doppler width parameter is influenced less by thermal broadening and more by turbulence or unresolved clouds along the line of sight . in figure
[ sw_fig11 ] , the doppler width parameter distribution is plotted for all ions , from the heaviest ion , , at the top , to the lightest ion , , at the bottom . the increase in line width for ions with decreasing atomic weight demonstrates that thermal broadening becomes increasingly important relative to turbulent broadening for the lightest ions .
the doppler width parameter distributions of the heaviest ions look very similar , probably because turbulent broadening completely dominates over the thermal broadening , thereby making the doppler width parameter independent of atomic mass .
the weighted mean and standard deviation of the doppler width parameter for all of the ions is given in table [ sw_table11 ] .
the high doppler width parameter tails for all of the distributions likely results from unresolved blends @xcite .
we attempt to minimize the occurrence of unresolved blends by , ( 1 ) observing many absorbing ions along the same line of sight , ( 2 ) observing optically thin absorption lines where individual cloud absorbers are more easily identified , ( 3 ) using high resolution observations , and ( 4 ) observing the shortest ( @xmath36100pc ) lines of sight to keep the absorption profile relatively simple . however , despite our efforts , we may not be resolving all the absorbers . observations with higher spectral resolution would help to disentangle the component structure and to fully resolve the smallest line widths , allowing us to investigate more accurately the thermal and nonthermal characteristics of warm clouds in the lism .
comparing the line widths of various atomic species with a range of atomic masses , is a powerful technique for measuring the temperature and turbulent velocity of clouds in the lism . in figure
[ sw_fig11_5 ] , we plot the mean doppler width parameters , inversely weighted by the variance of each measurement , of the entire dataset , as a function of atomic weight .
we fit the weighted mean values to equation [ sw_eq1 ] to calculate a typical temperature and turbulent velocity for the lism in the aggregate .
the variations in temperature and turbulent velocity among the various warm clouds in the lism will be discussed in detail in @xcite . as a whole , the lism can be roughly characterized by a temperature of @xmath37k and a turbulent velocity of @xmath38kms@xmath28 , where the error bars likely reflect the intrinsic range in these parameters rather than measurement uncertainties @xcite .
the observed column densities give the number of ions along the line of sight to the background source . if we assume a homogeneous medium with constant gas density , the column density is directly proportional to the distance that the line of sight traverses through the cloud . if the gas density is known , the column density of an individual absorption component is a direct measurement of the cloud thickness , which is vital to understanding the morphology of the lism .
the measured abundances also provide information on the degree of ionization and depletion of ions in the lism .
these characteristics can vary from cloud to cloud and even within a single cloud structure .
the cloud column density distribution function for each ion is given in figure [ sw_fig18 ] .
the distribution functions vary due to differences in cosmic abundances , depletion patterns , and ionization structure .
the weighted mean and standard deviation of the individual cloud column density for all of the ions are given in table [ sw_table11 ] .
the relatively small dispersion in the column densities , particularly in , suggests that nearby clouds in the lism may have similar sizes .
if we assume a constant d / h ratio of @xmath39 in the lism @xcite , and a constant hi number density of 0.1@xmath40 @xcite , then we can use the weighted mean and standard deviation of the column density distribution to estimate a typical length scale for clouds in the lism .
this calculation results in a mean length of @xmath41pc , and a range from 0.1pc to 11pc . in figures
[ sw_fig12]-[sw_fig17 ] , the total column density for each ion is shown in galactic coordinates .
the size of a symbol is inversely proportional to the target s distance , and the shading of a symbol indicates the total column density , as shown by the scale at the bottom of each plot .
the total column densities are clearly spatially correlated . as was found by @xcite ,
the highest column densities tend to be for lines of sight south of the galactic center , and the lowest column densities tend to be in the northern anti - galactic center direction .
however , exceptions to this generalization exist .
good spatial coverage of the sky and of objects at various distances is crucial to understanding the morphology of the lism .
we present a compilation of lism absorption line measurements of ions in the spectral range from 1200 - 1800 , which includes lines of , , , , , and .
this work is a companion paper to the research presented by @xcite on the high resolution lism absorption lines of the heavy ions , and .
these two papers provide a complete inventory of available lism measurements in the uv that can be used to determine the physical properties of the lism and to test models of the structure of the gas in our local environment .
the results of this work can be summarized as follows : 1 .
we take advantage of the known kinematic structure along a line of sight , obtained from high resolution observations of intrinsically narrow ( high resolution ) absorption lines @xcite , to accurately measure the lism absorption features of intrinsically broad ( low mass ion ) lines .
we successfully fit the absorption lines with this technique , even with moderate resolution spectra .
this greatly enlarges the number of sightlines with useful lism absorption line measurements .
the central velocity of a particular absorbing cloud is now measured using many resonance lines of different ions . the central velocity measurement for a given velocity component made with two different instruments ( ghrs and stis ) and with a range of spectral resolutions have a standard deviation of only 0.59kms@xmath28 .
3 . the distribution of doppler widths shows a systematic increase as the atomic weight of the absorbing ion decreases .
this demonstrates that thermal broadening becomes more important compared to turbulent broadening with decreasing ion mass . for the lism in aggregate
, we measure a temperature of @xmath37k and a turbulent velocity of @xmath38kms@xmath28 .
variations in temperature and turbulent velocity are significant in the lism as is discussed in detail in @xcite .
support for program # 9525 is provided to the university of colorado at boulder by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas ar-09525.01a .
this work is also supported by a nasa gsrp student fellowship grant ngt 5 - 50242 .
we would also like to thank tom ayres and mike shull for their helpful comments and suggestions .
a survey of all available high resolution observations of and absorption along sightlines within 100pc was presented by @xcite .
we have expanded the survey in this work to include other ions , such as , , , , , , and . for three lines of sight with new light ion measurements presented here
, we also have new heavy ion observations that we analyzed before fitting the moderate to high resolution light ion spectra .
we include in this appendix the new high resolution line ( 2600.1729 , 2586.6500 , 2382.7652 , 2374.4612 , 2344.2139 ) and h and k line ( 2803.5305 and 2796.5318
) observations toward stars within 100pc , since the publication of @xcite .
these sightlines are listed in table [ sw_table_a1 ] , while the particular observational details are given in table [ sw_table_a2 ] .
figures [ sw_fig7]-[sw_fig8b ] show our best fits to the new and observations .
tables [ sw_table9 ] and [ sw_table10 ] give the absorption line fit parameters for and , respectively , for all targets shown in figures [ sw_fig7]-[sw_fig8b ] .
in addition , some errors in the tables of @xcite are corrected here .
tables [ sw_table9 ] and [ sw_table10 ] are meant to be continuations of tables 3 and 4 in @xcite .
based on the number of observed lism absorbers for each ion , given in table [ sw_table11 ] , the best sampling of the distribution of lism clouds will come from and . by comparing the number of velocity components ( clouds ) along a given sightline with the distance of the background star
, we can investigate the rough distribution of warm clouds within 100pc . in figure [ sw_fig26 ] ,
the average number of clouds along the line of sight toward stars located in 10pc bins is compared for the heavy ions , , , and , based on the analysis of only high - resolution spectra presented here and in @xcite .
it is remarkable that the average number of clouds remains nearly constant out to 100pc .
only a very slight increasewith stellar distance is detected , indicating that the distribution of clouds in the lism is not uniform , but instead is concentrated near the sun .
the specific morphology of the lism can induce biases into these estimates , particularly if only one star is used in a particular distance bin , as is the case for and at distances @xmath42pc .
these particular sightlines happen to be at high latitudes , where little lism material is detected ( see figures [ sw_fig12 ] to [ sw_fig17 ] ) .
the specific morphology of the lism will be the subject of a future paper in this series .
previous estimates of the number density of clouds in the ism have been between 8 and 10 cloud components per kpc @xcite .
these , relatively low resolution observations , were of more distant sightlines that traversed a more varied ism environment and more clouds .
our estimate , as shown in figure [ sw_fig26 ] is @xmath43 clouds per kpc .
although we only sample a relatively simple ism environment , namely warm partially ionized clouds within the local bubble , our data are not severely blended because of the short lines of sight and high spectral resolution .
the cloud density estimated from more distant sightlines may suffer from such severe blending that many clouds are undetected , leading to an underestimate of the number of clouds along the line of sight .
bevington , p.r . , & robinson , d.k .
1992 , data reduction and error analysis for the physical sciences ( new york : mcgraw - hill ) brandt , j.c .
, et al . 1994 , pasp , 106 , 890 brandt , j.c . ,
2001 , , 121 , 2173 dring , a.r . ,
linsky , j.l . ,
murthy , j. , henry , r.c . , moos , w. , vidal - madjar , a. , audouze , j. , & landsman , w. 1997 , , 488 , 760 duflot , m. , figon , p. , & meyssonnier , n. 1995 , a&as , 114 , 269 gnova , r. , molaro , p. , vladilo , g. , & beckman , j.e .
1990 , , 355 , 150 gilliland , r.l .
1994 , in ghrs instrument science report 063 ( baltimore : stsci ) gnaciski , p. 2000 , acta astronomica , 50 , 133 heap , s.r . , et al . 1995 , pasp , 107 , 871 hbrard , g. , mallouris , c. , ferlet , r. , koester , d. , lemoine , m. , vidal - madjar , a. , & york , d. 1999 , , 350 , 643 kimble , r.a .
, et al . 1998 , , 492 , l83 lallement , r. , & bertin , p. 1992
, , 266 , 479 lallement , r. , & ferlet , r. , lagrange , a.m. , lemoine , m. , & vidal - madjar , a. 1995 , , 304 , 461 lindler , d. 1999 , in calstis reference guide ( greenbelt : nasa / lasp ) linsky , j.l .
1998 , space sci .
, 84 , 285 linsky , j.l . , & wood , b.e .
1996 , , 463 , 254 moos , h.w . ,
2002 , , 140 , 3 morton , d.c .
1991 , , 77 , 119 perryman , m.a.c .
, et al . 1997 , , 323 , l49 piskunov , n. , wood , b.e . , linsky , j.l . ,
dempsey , r.c .
, & ayres , t.r .
1997 , , 474 , 315 redfield , s. , & linsky , j.l .
2000 , , 534 , 825 redfield , s. , & linsky , j.l .
2001 , , 551 , 413 redfield , s. , & linsky , j.l .
2002 , , 139 , 439 redfield , s. , & linsky , j.l .
2003 , , submitted reid , n. , & wegner , g. 1988 , , 335 , 953 robinson , r.d . , blackwell , j. , feggans , k. , lindler , d. , norman , d. , & shore , s.n .
1992 , in a user s guide to the ghrs software , version 2.0 ( greenbelt : goddard space flight center ) sahu , k.c .
, et al . 1999 , in stis instrument handbook ( baltimore : stsci ) slavin , j.d .
, & frisch , p.c .
2002 , , 565 , 364 soderblom , d.r . , hulbert , s.j . , leitherer , c. , & sherbert , l.e .
1994 , in _ hst _ goddard high resolution spectrograph instrument handbook , version 5.0 ( baltimore : stsci ) spitzer , l. 1978 , physical processes in the interstellar medium ( new york : john wiley & sons , inc . )
vidal - madjar , a. , des etangs , a.l . , & ferlet , r. 1998 , planet .
space sci .
, 46 , 629 vennes , s. , & thorstensen , j.r .
1994 , , 108 , 1881 welty , d.e . ,
morton , d.c . , & hobbs , l.m . 1996 , , 106 , 533 wood , b.e . , & linsky , j.l
1997 , , 474 , l39 wood , b.e .
, redfield , s. , linsky , j.l .
, & sahu , m.s .
2002 , , 581 , 1168 woodgate , b.e . ,
1998 , pasp , 110 , 1183 llcccccccc 128620 & @xmath4 cen a&g2v&0.0&24.6&315.7&0.7 & 1.3 & 15.6 & 18.0 + 128621 & @xmath4 cen b&k1v&1.3&20.7&315.7 & 0.7 & 1.3 & 15.6 & 18.0 + 48915 & @xmath4 cma a&a1v&1.5&7.6 & 227.2 & 8.9 & 2.6 & 19.5 & 21.6 + 48915b & @xmath4 cma b&da&8.4 & & 227.2 & 8.9 & 2.6 & 19.5 & 21.6 + 22049 & @xmath9 eri&k2v&3.7&15.5 & 195.8 & 48.1 & 3.2 & 21.6 & 25.7 + 201091 & 61 cyg a & k5v & 5.2 & 64.3&82.3 & 5.8 & 3.5 & 5.1 & 4.7 + 61421 & @xmath4 cmi&f5v & 0.3 & 3.2 & 213.7 & 13.0 & 3.5 & 19.7 & 21.1 + 26965 & 40 eri a & k1v & 4.4 & 42.2&200.8 & 38.1 & 5.0 & 23.2 & 27.2 + 165341 & 70 oph & k0v & 4.0 & 6.9 & 29.9 & 11.4 & 5.1 & 23.5 & 26.4 + 187642 & @xmath4 aql&a7v & 0.8 & 26.1&47.7 & 8.9 & 5.1 & 17.1 & 18.2 + 155886 & 36 oph a & k0v & 5.3 & 0.6 & 358.3 & 6.9 & 5.5 & 25.2 & 28.4 + 131156a&@xmath19 boo a&g8v & 4.7 & 3.0 & 23.1 & 61.4 & 6.7 & 17.7 & 21.5 + 216956 & @xmath4 psa&a3v & 1.2 & 6.5 & 20.5 & 64.9 & 7.7 & 3.6 & 1.9 + 172167 & @xmath4 lyr&a0v & 0.0 & 13.9&67.5 & 19.2 & 7.8 & 13.5 & 15.2 + 39587 & @xmath44 ori&g0v&4.4 & 13.5&188.5 & 2.7 & 8.7 & 24.9 & 27.9 + 20630 & @xmath45 cet&g5v&4.8 & 19.9 & 178.2 & 43.1 & 9.2 & 22.8 & 27.0 + 197481 & au mic & m0 & 8.6 & 1.2 & 12.7 & 36.8 & 9.9 & 15.3 & 15.7 + 62509 & @xmath10 gem&k0iii&1.2 & 3.3 & 192.2 & 23.4 & 10.3 & 19.6 & 21.0 + 102647 & @xmath10 leo&a3v & 2.1 & 0.2 & 250.6 & 70.8 & 11.1 & 3.4 & 6.1 + 124897 & @xmath4 boo&k1.5iii&0.0&5.2&15.1&69.1 & 11.3 & 15.5 & 19.3 + 33262 & @xmath46 dor&f7v & 4.7 & 2.0 & 266.0 & 36.7 & 11.7 & 7.8 & 9.4 + 34029 & @xmath4 aur&g5iii&0.1 & 30.2 & 162.6 & 4.6 & 12.9 & 22.0 & 24.6 + 36705 & ab dor & k1iii & 6.9 & 33 & 275.3 & 33.1 & 14.9 & 4.3 & 5.3 + 432 & @xmath10 cas&f2iv&2.3 & 11.3 & 117.5 & 3.3 & 16.7 & 9.4 & 11.3 + 11443 & @xmath4 tri&f6iv&3.4 & 12.6&138.6 & 31.4 & 19.7 & 18.0 & 21.7 + 29139 & @xmath4 tau&k5iii&0.9 & 54.3 & 181.0 & 20.3 & 20.0 & 25.5 & 29.4 + 22468 & hr 1099 & g9v & 5.9 & 23 & 184.9 & 41.6 & 29.0 & 23.3 & 27.4 + 4128 & @xmath10 cet&k0iii&2.0 & 13.0 & 111.3 & 80.7 & 29.4 & 8.2 & 11.4 + 120315 & @xmath47 uma & b3v & 1.9 & 10.9&100.7 & 65.3 & 30.9 & 5.8 & 8.1 + 209952 & @xmath4 gru&b7iv&1.7 & 11.8 & 350.0 & 52.5 & 31.1 & 8.7 & 8.1 + & hz 43 & daw & 12.7 & 62.0 & 54.1 & 84.2 & 32.0 & 8.9 & 12.0 + 82210 & dk uma & g4iv & 4.6 & 27.2&142.6 & 38.9 & 32.4 & 9.4 & 9.5 + 80007 & @xmath10 car&a2iv&1.7 & 5 & 286.0 & 14.4 & 34.1 & 2.3 & 2.8 + 62044 & @xmath32 gem&k1iii&4.3 & 45.8 & 191.2 & 23.3 & 37.5 & 19.7 & 21.0 + 28568 & sao 93981&f5v & 6.5 & 40.9 & 180.5 & 21.4 & 41.2 & 25.5 & 29.3 + 220657 & @xmath48 peg&f8iv&4.4&11.1&98.6 & 35.4 & 53.1 & 5.1 & 7.6 + 203387 & @xmath11 cap&g8iii&4.3 & 11.5 & 33.6 & 40.8 & 66.1 & 11.8 & 11.5 + & g191-b2b & daw & 11.6 & 5 & 156.0 & 7.1 & 68.8 & 20.3 & 22.7 + & gd 153 & daw & 13.4 & & 317.3 & 84.8 & 70.5 & 8.7 & 12.0 + 111812 & 31 com & g0iii & 4.9 & 1.4 & 115.0 & 89.6 & 94.2 & 7.2 & 10.2 + 45348 & @xmath4 car & f0ii & 0.7 & 20.5 & 261.2 & 25.3 & 95.9 & 8.8 & 10.1 + llllccclcl 128620 & @xmath4 cen a & stis & e140h & 1140 - 1335 & 114000 & 4695.0 & linsky & 7263 & 1999 feb 12 + & & stis & e140h & 1315 - 1517 & 114000 & 4695.0 & linsky & 7263 & 1999 feb 12 + & & stis & e140h & 1497 - 1699 & 114000 & 4695.0 & linsky & 7263 & 1999 feb 12 + 48915 & @xmath4 cma a & ghrs & g160 m & 1645 - 1679 & 20000 & 1740.8 & lallement & 2461 & 1992 aug 15 + & & ghrs & g160 m & 1651 - 1686 & 20000 & 108.8 & wahlgren & 3496 & 1993 mar 16 + & & ghrs & g160 m & 1651 - 1686 & 20000 & 108.8 & wahlgren & 3496 & 1993 may 8 + 22049 & @xmath9 eri & stis & e140 m & 1140 - 1735 & 45800 & 5798.0 &
jordan & 7479 & 2000 mar 17 + 165341 & 70 oph & stis & e140 m & 1140 - 1735 & 45800 & 7809.0 & ayres & 8280 & 2000 aug 23 + 187642 & @xmath4 aql & ghrs & g160 m & 1315 - 1351 & 20000 & 544.0 & walter & 3950 & 1993 apr 7 + & & ghrs & g160 m & 1317 - 1352 & 20000 & 435.2 & simon & 3737 & 1993 apr 7 + & & ghrs & g140 m & 1201 - 1229 & 20000 & 761.6 & simon & 6083 & 1996 mar 5 + 155886 & 36 oph a & stis & e140h & 1170 - 1372 & 114000 & 8769.0 & linsky & 7262 & 1999 oct 10 + 131156a&@xmath19 boo a & stis & e140 m & 1140 - 1735 & 45800 & 2450.0 & timothy & 7585 & 1998 dec 16 + 216956 & @xmath4 psa&ghrs & g160 m & 1524 - 1559 & 20000 & 3628.8 & ferlet & 2537 & 1993 oct 31 + & & ghrs & g160 m & 1644 - 1679 & 20000 & 576.0 & ferlet & 2537 & 1993 oct 31 + 172167 & @xmath4 lyr&ghrs & g140 m & 1189 - 1217 & 20000 & 435.2 & vidal - madjar & 6828 & 1996 aug 28 + & & ghrs & g160 m & 1185 - 1222 & 20000 & 435.2 & vidal - madjar & 6828 & 1996 dec 23 + 39587 & @xmath44 ori&stis & e140 m & 1140 - 1735 & 45800 & 6770.0 & ayres & 8280 & 2000 oct 3 + 20630 & @xmath45 cet&stis & e140 m & 1140 - 1735 & 45800 & 7805.0 & ayres & 8280 & 2000 sep 19 + 197481 & au mic & stis & e140 m & 1150 - 1735 & 45800 & 10105.0 & linsky & 7556 & 1998 sep 6 + 62509 & @xmath10 gem & ghrs & ech - a & 1300 - 1307 & 100000 & 761.6 & linsky & 5647 & 1995 sep 20 + & & ghrs & ech - a & 1300 - 1307 & 100000 & 761.6 & linsky & 5647 & 1995 sep 20 + 102647 & @xmath10
leo & ghrs & g160 m & 1524 - 1559 & 20000 & 6912.0 & ferlet & 2537 & 1992 jan 23 + & & ghrs & g160 m & 1645 - 1679 & 20000 & 1209.6 & ferlet & 2537 & 1992 jan 24 + 124897 & @xmath4 boo & stis & e140 m & 1150 - 1735 & 45800 & 5208.0 & ayres & 7733 & 1998 aug 24 + 33262 & @xmath46 dor&stis&e140 m & 1140 - 1735 & 45800 & 6000.0 & ayres & 8280 & 1999 may 1 + 34029 & @xmath4 aur & stis & e140 m & 1140 - 1735 & 45800 & 8113.0 & drake & 8546 & 1999 sep 12 + 36705 & ab dor & ghrs & g160 m & 1318 - 1354 & 20000 & 1305.6 & walter & 5181 & 1994 nov 14 - 15 + 29139 & @xmath4 tau&stis & e140 m & 1140 - 1735 & 114000 & 10512.0 & ayres & 9273 & 2002 jan 3 + 22468 & hr1099 & stis & e140 m & 1140 - 1735 & 45800 & 12225.0 & ayres & 8280 & 1999 sep 15 + 4128 & @xmath10 cet&stis & e140 m & 1140 - 1735 & 100000 & 4737.0 & linsky & 8039 & 1999 sep 13 + 120315 & @xmath47 uma & ghrs & ech - a & 1212 - 1218 & 100000 & 652.8 & frisch & 6886 & 1997 jan 1 + & & ghrs & ech - a & 1298 - 1306 & 100000 & 217.6 & frisch & 6886 & 1997 jan 1 + & & ghrs & ech - a & 1332 - 1339 & 100000 & 1171.4 & frisch & 6886 & 1997 jan 1 + 209952 & @xmath4 gru&ghrs&g160 m & 1174 - 1211 & 20000 & 460.8 & smith & 3941 & 1993 jul 24 + & & ghrs & g160 m & 1277 - 1313 & 20000 & 230.4 & smith & 3941 & 1993 jul 24 + & & ghrs & g160 m & 1309 - 1345 & 20000 & 230.4 & smith & 3941 & 1993 jul 24 + & & ghrs & g160 m & 1642 - 1677 & 20000 & 230.4 & smith & 3941 & 1993 jul 24 + & hz 43 & stis & e140h & 1140 - 1335 & 114000 & 11656.0 & sahu & 8595 & 2001 may 3 + 82210 & dk uma & stis & e140 m & 1140 - 1735 & 45800 & 9300.0 & ayres & 8280 & 2000 feb 24 + 80007 & @xmath10 car&ghrs & g160 m & 1524 - 1559 & 20000 & 748.8 & ferlet & 2537 & 1993 mar 16 + & & ghrs & g160 m & 1645 - 1679 & 20000 & 345.6 & ferlet & 2537 & 1993 mar 16 + 28568 & sao 93981&stis & e140 m & 1140 - 1735 & 45800 & 1065.0 & bohm - vitense & 7389 & 1998 nov 8 + 220657 & @xmath48 peg&stis & e140m&1140 - 1735 & 45800 & 8226.0 & ayres & 8280 & 2000 jul 11 + 203387 & @xmath11 cap&stis & e140 m & 1140 - 1735 & 45800 & 11651.0 & ayres & 8280 & 2000 apr 15 + & g191-b2b & stis & e140h & 1140 - 1335 & 114000 & 2789.0 & leitherer & 8067 & 1998 dec 17 + & & stis & e140h & 1497 - 1699 & 114000 & 2703.0 & leitherer & 8067 & 1998 dec 17 + & & stis & e140h & 1140 - 1335 & 114000 & 8279.0 & leitherer & 8421 & 2000 mar 16 - 17 + & & stis & e140h & 1497 - 1699 & 114000 & 8284.0 & leitherer & 8421 & 2000 mar 17 + & & stis & e140h & 1140 - 1335 & 114000 & 1734.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1170 - 1372 & 114000 & 696.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1206 - 1408 & 114000 & 654.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1242 - 1444 & 114000 & 686.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1279 - 1481 & 114000 & 752.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1315 - 1517 & 114000 & 851.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1352 - 1554 & 114000 & 1038.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1388 - 1590 & 114000 & 1263.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1425 - 1627 & 114000 & 1636.0 & valenti & 8915 & 2001 sep 17 + & & stis & e140h & 1461 - 1663 & 114000 & 1996.0 & valenti & 8915 & 2001 sep 17 + & & stis & e230h & 1624 - 1896 & 114000 & 1314.0 & valenti & 8915 & 2001 sep 18 + & gd 153 & stis & e140h & 1140 - 1335 & 114000 & 11697.0 & sahu & 8595 & 2001 mar 14 + & & stis & e140h & 1140 - 1335 & 114000 & 1851.0 & sahu & 8595 & 2001 mar 22 + 111812 & 31 com & stis & e140 m & 1140 - 1735 & 45800 & 7733.0 & ayres & 8889 & 2001 mar 13 + 45348 & @xmath4 car&stis&e140 m & 1140 - 1735 & 45800 & 2000.0 & brown & 7764 & 2002 jun 11 + & & stis & e140h & 1497 - 1699 & 114000 & 1200.0 & brown & 7764 & 2002 jun 11 + llcccccccc 128620 & @xmath4 cen a & ( @xmath49 ) & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 128621 & @xmath4 cen b & ( @xmath49 ) & & & & & & ( @xmath49 ) & ( @xmath49 ) + 48915 & @xmath4 cma a & ( @xmath49 ) & ( @xmath49 ) & & & @xmath1 & & ( @xmath49 ) & ( @xmath49 ) + 48915b & @xmath4 cma b & ( @xmath49 ) & ( @xmath49 ) & ( @xmath49 ) & ( @xmath49 ) & & ( @xmath49 ) & ( @xmath49 ) & ( @xmath49 ) + 22049 & @xmath9 eri & ( @xmath49 ) & @xmath49 & & @xmath49 & @xmath1 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 201091 & 61 cyg a & ( @xmath49 ) & & & & & & ( @xmath49 ) & + 61421 & @xmath4 cmi & ( @xmath49 ) & & & & & & ( @xmath49 ) & ( @xmath49 ) + 26965 & 40 eri a & ( @xmath49 ) & & & & & & ( @xmath49 ) & + 165341 & 70 oph & @xmath49 & @xmath49 & & @xmath49 & @xmath1 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 187642 & @xmath4 aql & @xmath49 & @xmath49 & & & & & ( @xmath49 ) & ( @xmath49 ) + 155886 & 36 oph a & ( @xmath49 ) & @xmath49 & & & & & ( @xmath49 ) & ( @xmath49 ) + 131156a&@xmath19 boo a & @xmath49 & @xmath49 & & @xmath49 & @xmath1 & @xmath49 & ( @xmath49 ) & + 216956 & @xmath4 psa & & & & & @xmath1 & @xmath49 & & ( @xmath49 ) + 172167 & @xmath4 lyr & & & @xmath49 & & & @xmath49 & & ( @xmath49 ) + 39587 & @xmath44 ori & @xmath49 & @xmath49 & & & @xmath1 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 20630 & @xmath45 cet & @xmath49 & @xmath49 & & @xmath49 & @xmath1 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 197481 & au mic & @xmath49 & & & & & & ( @xmath49 ) & + 62509 & @xmath10 gem & ( @xmath49 ) & & & @xmath49 & & & ( @xmath49 ) & ( @xmath49 ) + 102647 & @xmath10 leo & & & & & @xmath1 & @xmath49 & & ( @xmath49 ) + 124897 & @xmath4 boo & & @xmath49 & & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 33262 & @xmath46 dor & @xmath49 & @xmath49 & & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 34029 & @xmath4 aur & ( @xmath49 ) & ( @xmath49 ) & @xmath49 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 36705 & ab dor & & @xmath49 & & & & & ( @xmath49 ) & + 432 & @xmath10 cas & ( @xmath49 ) & & & & & & ( @xmath49 ) & ( @xmath49 ) + 11443 & @xmath4 tri & ( @xmath49 ) & & & & & & ( @xmath49 ) & ( @xmath49 ) + 29139 & @xmath4 tau & @xmath49 & & & & & & ( @xmath49 ) & + 22468 & hr 1099 & ( @xmath49 ) & @xmath49 & & @xmath49 & @xmath1 & @xmath1 & ( @xmath49 ) & + 4128 & @xmath10 cet & ( @xmath49 ) & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & + 120315 & @xmath47 uma & @xmath49 & @xmath49 & & @xmath49 & & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 209952 & @xmath4 gru & & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + & hz 43 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 82210 & dk uma & @xmath49 & @xmath49 & @xmath1 & @xmath49 & @xmath49 & @xmath1 & ( @xmath49 ) & ( @xmath49 ) + 80007 & @xmath10 car & & & & & @xmath49 & @xmath49 & & ( @xmath49 ) + 62044 & @xmath32 gem & ( @xmath49 ) & & & & & & ( @xmath49 ) & + 28568 & sao 93981 & & @xmath49 & & & & & ( @xmath49 ) & + 220657 & @xmath48 peg & @xmath49 & @xmath49 & @xmath1 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + 203387 & @xmath11 cap & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & ( @xmath49 ) & ( @xmath49 ) + & g191-b2b & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & @xmath49 + & gd 153 & @xmath49 & @xmath49 & @xmath49 & @xmath49 & & @xmath49 & & @xmath49 + 111812 & 31 com & ( @xmath49 ) & @xmath49 & & @xmath49 & @xmath1 & @xmath1 & ( @xmath49 ) & ( @xmath49 ) + 45348 & @xmath4 car & & & & & @xmath49 & @xmath49 & @xmath49 & @xmath49 + llccccl 128620 & @xmath4 cen a & 1 & 18.2 @xmath50 0.1 & 6.55 @xmath50 0.27 & 12.799 @xmath50 0.015 & 1 + 128621 & @xmath4 cen b & 1 & 18.2 @xmath50 0.1 & 6.88 @xmath50 0.17 & 12.775
@xmath50 0.009 & 1 + 48915 & @xmath4 cma a & 1 & 19.5 & & 12.43@xmath51 & 2 + & & 2 & 13.7 & & 12.43@xmath51 & 2 + 48915b & @xmath4 cma b & 1 & 17.6 & 8.14 & 12.81@xmath52 & 3 + & & 2 & 11.7 & 5.66 & @xmath1 12.48 & 3 + 22049 & @xmath9 eri & 1 & 20.5 @xmath50 0.7 & 8.1 @xmath50 0.4 & 13.03 @xmath50 0.06 & 4 + 201091 & 61 cyg a & 1 & 3.0 & 7.8 & 13.01 & 5 + & & 2 & 9.0 & 7.8 & 13.01 & 5 + 61421 & @xmath4 cmi & 1 & 24.0 @xmath50 0.10 & 7.59 @xmath50 0.10 & 12.81 @xmath50 0.03 & 6 + & & 2 & 20.5 @xmath50 0.10 & 7.59 @xmath50 0.10 & 13.08 @xmath50 0.04 & 6 + 26965 & 40 eri a & 1 & 21.4 & 8.2 & 13.03 & 5 + 165341 & 70 oph & 1 & 26.2 @xmath50 0.9 & 6.22 @xmath50 0.54 & 12.95 @xmath50 0.07 & 7 + & & 2 & 32.4 @xmath50 0.9 & 4.95 @xmath50 1.51 & 12.64 @xmath50 0.24 & 7 + & & 3 & 42.9 @xmath50 0.9 & 5.73 @xmath50 1.40 & 12.27 @xmath50 0.11 & 7 + 187642 & @xmath4 aql & 1 & 18.4 @xmath50 1.5 & 10.3 @xmath50 1.0 & 13.09 @xmath50 0.27 & 7 ( see 8) + & & 2 & 22.0 @xmath50 1.5 & 10.3 @xmath50 1.0 & 13.11 @xmath50 0.54 & 7 ( see 8) + & & 3 & 25.8 @xmath50 1.5 & 10.3 @xmath50 1.0 & 12.65 @xmath50 0.22 & 7 ( see 8) + 155886 & 36 oph a & 1 & 29.0 @xmath50 0.2 & 7.33 @xmath50 0.17 & 13.00 @xmath50 0.01 & 9 + 131156a&@xmath19 boo a & 1 & 17.9 @xmath50 0.3 & 6.83 @xmath50 0.46 & 13.06 @xmath50 0.02 & 7 + 39587 & @xmath44 ori & 1 & 21.6 @xmath50 0.2 & 7.94 @xmath50 0.22 & 12.98 @xmath50 0.01 & 7 + 20630 & @xmath45 cet & 1 & 22.8 @xmath50 1.4 & 6.96 @xmath50 0.98 & 12.68 @xmath50 0.13 & 7 + & & 2 & 15.3 @xmath50 1.4 & 6.04 @xmath50 1.67 & 12.66 @xmath50 0.17 & 7 + & & 3 & 9.2 @xmath50 1.4 & 7.06 @xmath50 1.42 & 12.59 @xmath50 0.21 & 7 + 197481 & au mic & 1 & 21.7 @xmath50 0.2 & 9.50 @xmath50 0.26 & 13.41 @xmath50 0.01 & 7 + 62509 & @xmath10 gem & 1 & 33.0 @xmath50 1.0 & 7.8 @xmath50 0.7 & 13.0 @xmath50 0.1 & 4 + & & 2 & 21.9 @xmath50 1.0 & 8.7 @xmath50 0.7 & 13.2 @xmath50 0.1 & 4 + 33262 & @xmath46 dor & 1 & 10.8 @xmath50 1.3 & 9.48 @xmath50 1.35 & 13.32 @xmath50 0.29 & 7 + & & 2 & 5.4 @xmath50 1.3 & 8.25 @xmath50 1.00 & 12.94 @xmath50 0.24 & 7 + 34029 & @xmath4 aur & 1 & 22.04 & 7.81 & 13.47 & 10 + & & 1 & 21.8 @xmath50 0.4 & 7.8 @xmath50 0.3 & 13.44 @xmath50 0.02 & 6 + 432 & @xmath10 cas & 1 & 10 @xmath50 1 & & 13.4 @xmath50 0.1 & 11 + & & 1 & 8.3 @xmath50 0.3 & 9.0 @xmath50 0.4 & 13.36 @xmath50 0.06 & 4 + 11443 & @xmath4 tri & 1 & 16.9 @xmath50 0.7 & 8.4 @xmath50 1.0 & 13.3 @xmath50 0.1 & 4 + & & 2 & 12.5 @xmath50 0.7 & 8.5 @xmath50 1.0 & 13.0 @xmath50 0.1 & 4 + 29139 & @xmath4 tau & 1 & 20.62 @xmath50 0.15 & 8.55 @xmath50 0.20 & 12.990 @xmath50 0.008 & 7 + 22468 & hr 1099 & 1 & 22 @xmath50 2 & & 13.06 @xmath50 0.03 & 11 + & & 2 & 15 @xmath50 2 & & 12.76 & 11 + & & 3 & 8 @xmath50 2 & & 12.36 & 11 + 4128 & @xmath10 cet & 1 & 8 @xmath50 2 & & 13.7@xmath53 & 11 + & & 2 & 1 @xmath50 2 & & 12.1 & 11 + 120315 & @xmath47 uma & 1 & 0.76 @xmath50 0.34 & 4.73 @xmath50 1.82 & 11.74 @xmath50 0.03 & 7 + & & 2 & -4.77 @xmath50 0.34 & 8.64 @xmath50 0.63 & 13.08 @xmath50 0.03 & 7 + & hz 43 & 1 & 6.15 @xmath50 0.27 & 8.09 @xmath50 0.40 & 13.15 @xmath50 0.02 & 7 + 82210 & dk uma & 1 & 7.8 @xmath50 0.1 & 7.59 @xmath50 0.12 & 13.14 @xmath50 0.01 & 7 + 62044 & @xmath32 gem & 1 & 32.0 @xmath50 1.0 & 7.8 @xmath50 0.7 & 12.9 @xmath50 0.1 & 4 + & & 2 & 21.4 @xmath50 1.0 & 8.8 @xmath50 0.7 & 13.1 @xmath50 0.1 & 4 + 220657 & @xmath48 peg & 1 & 7.0 @xmath50 1.2 & 4.46 @xmath50 1.37 & 12.61 @xmath50 0.24 & 7 + & & 2 & 0.0 @xmath50 1.2 & 5.39 @xmath50 0.71 & 13.09 @xmath50 0.05 & 7 + & & 3 & 8.6 @xmath50 1.2 & 4.84 @xmath50 1.00 & 12.58 @xmath50 0.18 & 7 + 203387 & @xmath11 cap & 1 & 5.3 @xmath50 2.3 & 10.5 @xmath50 1.4 & 13.55 @xmath50 0.37 & 7 + & & 2 & 15.2 @xmath50 2.3 & 10.5 @xmath50 1.4 & 13.11 @xmath50 0.33 & 7 + & & 3 & 22.9 @xmath50 2.3 & 10.5 @xmath50 1.4 & 13.30 @xmath50 0.18 & 7 + & g191-b2b & 1 & 18.87 @xmath50 0.66 & 7.42 @xmath50 0.28 & 13.36 @xmath50 0.03 & 7 + & & 2 & 7.78 @xmath50 1.54 & 6.96 @xmath50 0.60 & 12.62 @xmath50 0.10 & 7 + & gd 153 & 1 & 5.04 @xmath50 0.47 & 7.56 @xmath50 0.71 & 13.13 @xmath50 0.03 & 7 + 111812 & 31 com & 1 & 4
@xmath50 1 & & 13.1 @xmath50 0.1 & 11 + & & 1 & 3.3 @xmath50 0.6 & 8.5 @xmath50 0.5 & 13.19 @xmath50 0.05 & 4 + llccccl 128620 & @xmath4 cen a & 1 & 18.70 @xmath50 0.11 & 2.93 @xmath50 0.15 & 14.30 @xmath50 0.12 & 1 + 48915 & @xmath4 cma a & 1 & 17.6 @xmath50 1.5 & & 14.62@xmath54 & 2 + & & 2 & 11.7 @xmath50 1.5 & & 13.78@xmath55 & 2 + 48915b & @xmath4 cma b & 1 & 17.6 @xmath50 1.5 & & 14.62@xmath54 & 2 + & & 2 & 11.7 @xmath50 1.5 & & 13.78@xmath55 & 2 + 22049 & @xmath9 eri & 1 & 18.50 @xmath50 0.06 & 4.30 @xmath50 1.06 & 15.08 @xmath50 0.63 & 1 + 165341 & 70 oph & 1 & 26.5 & 3.93 @xmath50 0.92 & 14.35 @xmath50 0.31 & 1 + & & 2 & 32.9 & 3.40 @xmath50 1.26 & 13.92 @xmath50 0.26 & 1 + & & 3 & 43.4 & 4.86 @xmath50 3.44 & 14.96 @xmath50 0.34 & 1 + 187642 & @xmath4 aql & 1 & 15.46 @xmath50 1.21 & 5.87 @xmath50 2.46 & 13.40 @xmath50 0.11 & 1 + & & 2 & 18.22 @xmath50 2.09 & 6.56 @xmath50 3.13 & 13.17 @xmath50 0.28 & 1 + & & 3 & 23.94 @xmath50 3.21 & 6.01 @xmath50 3.32 & 13.50 @xmath50 0.26 & 1 + 155886 & 36 oph a & 1 & 26.63 @xmath50 0.86 & 4.19 @xmath50 0.82 & 14.44 @xmath50 0.16 & 1 + 131156a&@xmath19 boo a & 1 & 19.00 @xmath50 0.49 & 2.92 @xmath50 1.96 & 16.21 @xmath50 0.39 & 1 + 39587 & @xmath44 ori & 1 & 23.70 @xmath50 0.37 & 4.34 @xmath50 1.39 & 14.64 @xmath50 0.32 & 1 + 20630 & @xmath45 cet & 1 & 20.82 @xmath50 0.55 & 3.90 @xmath50 0.60 & 13.89 @xmath50 0.08 & 1 + & & 2 & 14.29 @xmath50 0.98 & 3.55 @xmath50 0.95 & 13.82 @xmath50 0.17 & 1 + & & 3 & 8.21 @xmath50 1.49 & 4.15 @xmath50 2.02 & 13.29 @xmath50 0.38 & 1 + 124897 & @xmath4 boo & 1 & 13.11 @xmath50 0.41 & 3.90 @xmath50 0.78 & 14.03 @xmath50 0.20 & 1 + 33262 & @xmath46 dor & 1 & 18.41 @xmath50 1.43 & 5.63 @xmath50 1.59 & 13.82 @xmath50 0.18 & 1 + & & 2 & 12.15 @xmath50 1.37 & 4.04 @xmath50 0.83 & 15.57 @xmath50 0.97 & 1 + 34029 & @xmath4 aur & 1 & 20.2 @xmath50 1.5 & 3.5 @xmath50 0.4 & 14.8 @xmath50 0.3 & 3 + 36705 & ab dor & 1 & 17.58 @xmath50 2.00 & 4.08 @xmath50 1.87 & 13.82 @xmath50 0.55 & 1 + & & 2 & 12.48 @xmath50 2.00 & 4.98 @xmath50 2.01 & 13.41 @xmath50 1.20 & 1 + & & 3 & 3.18 @xmath50 2.00 & 6.90 @xmath50 1.78 & 13.76 @xmath50 0.48 & 1 + 22468 & hr 1099 & 1 & 21.38 @xmath50 0.82 & 3.42 @xmath50 0.50 & 14.44 @xmath50 0.19 & 1 + & & 2 & 14.80 @xmath50 0.82 & 3.27 @xmath50 0.71 & 13.80 @xmath50 0.18 & 1 + & & 3 & 8.11 @xmath50 0.62 & 3.39 @xmath50 0.73 & 13.43 @xmath50 0.09 & 1 + 4128 & @xmath10 cet & 1 & 7.52 @xmath50 1.52 & 3.21 @xmath50 0.43 & 15.30 @xmath50 0.33 & 1 + & & 2 & 2.85 @xmath50 0.97 & 4.97 @xmath50 0.49 & 14.98 @xmath50 0.43 & 1 + 120315 & @xmath47 uma & & & & 14.21 & 4 + & & 1 & 9.19 @xmath50 0.64 & 6.16 @xmath50 0.71 & 13.18 @xmath50 0.05 & 1 + & & 2 & 2.04@xmath50 0.16 & 3.79 @xmath50 0.28 & 14.40 @xmath50 0.13 & 1 + 209952 & @xmath4 gru & 1 & 3.81 @xmath50 1.19 & 4.45 @xmath50 1.04 & 13.81 @xmath50 0.11 & 1 + & & 2 & 4.30@xmath50 1.39 & 4.34 @xmath50 0.49 & 15.12 @xmath50 0.55 & 1 + & & 3 & [email protected] & 4.73 @xmath50 0.85 & 13.14 @xmath50 0.18 & 1 + & hz 43 & 1 & 6.83 @xmath50 0.11 & 3.59 @xmath50 0.29 & 14.27 @xmath50 0.18 & 1 + 82210 & dk uma & 1 & 9.71 @xmath50 0.16 & 4.53 @xmath50 1.65 & 14.41 @xmath50 0.36 & 1 + 28568 & sao 93981 & 1 & 23.9 & 5.10 @xmath50 1.32 & 14.27 @xmath50 0.60 & 1 + & & 2 & 16.5 & 5.81 @xmath50 1.81 & 13.48 @xmath50 0.72 & 1 + 220657 & @xmath48 peg & 1 & 8.83 @xmath50 2.67 & 5.19 @xmath50 1.76 & 13.72 @xmath50 0.32 & 1 + & & 2 & 0.57 @xmath50 1.37 & 5.36 @xmath50 1.80 & 14.35 @xmath50 0.33 & 1 + & & 3 & 7.28@xmath50 1.96 & 3.06 @xmath50 1.57 & 13.58 @xmath50 0.34 & 1 + 203387 & @xmath11 cap & 1 & 1.86 @xmath50 0.68 & 5.31 @xmath50 0.54 & 14.25 @xmath50 0.14 & 1 + & & 2 & 14.19@xmath50 1.92 & 4.27 @xmath50 1.74 & 16.57 @xmath50 0.31 & 1 + & & 3 & 19.45@xmath50 1.04 & 5.76 @xmath50 0.58 & 14.75 @xmath50 0.33 & 1 + & g191-b2b & 1 & 18.12 @xmath50 1.45 & 2.98 @xmath50 0.54 & 15.70 @xmath50 0.36 & 1 + & & 2 & 9.02 @xmath50 0.49 & 4.43 @xmath50 0.62 & 16.12 @xmath50 0.33 & 1 + & gd 153 & 1 & 4.91 @xmath50 0.12 & 3.51 @xmath50 0.65 & 14.27 @xmath50 0.26 & 1 + 111812 & 31 com & 1 & 3.84 @xmath50 0.41 & 3.48 @xmath50 0.44 & 14.33 @xmath50 0.12 & 1 + llccccl 128620 & @xmath4 cen a & 1 & 19.58 @xmath50 0.64 & 3.44 @xmath50 1.17 & 13.89 @xmath50 0.20 & 1 + 48915b & @xmath4 cma b & 1 & 17.6 @xmath50 1.5 & & 13.11@xmath56 & 2 + & & 2 & 11.7 @xmath50 1.5 & & 12.96@xmath57 & 2 + 172167 & @xmath4 lyr & 1 & 12.06 @xmath50 2.16 & 3.10 @xmath50 1.87 & 13.80 @xmath50 0.13 & 1 + & & 2 & 18.52 @xmath50 2.77 & 3.88 @xmath50 1.95 & 13.64 @xmath50 0.11 & 1 + & & 3 & 20.87 @xmath50 2.05 & 3.11 @xmath50 1.30 & 14.45 @xmath50 0.58 & 1 + 34029 & @xmath4 aur & 1 & 19.31 @xmath50 0.35 & 2.73 @xmath50 0.64 & 13.94 @xmath50 0.20 & 1 and 3 + 4128 & @xmath10 cet & 1 & 10.20 @xmath50 1.73 & 3.49 @xmath50 1.29 & 13.37 @xmath50 0.20 & 1 + & & 2 & 1.95 @xmath50 1.56 & 4.58 @xmath50 1.65 & 13.60 @xmath50 0.13 & 1 + 209952 & @xmath4 gru & 1 & 7.27 @xmath50 1.80 & 4.72 @xmath50 1.36 & 14.76 @xmath50 0.62 & 1 + & & 2 & 12.48 @xmath50 2.07 & 3.25 @xmath50 0.94 & 14.17 @xmath50 0.29 & 1 + & & 3 & 18.82 @xmath50 3.46 & 3.28 @xmath50 1.56 & 14.05 @xmath50 0.62 & 1 + & hz 43 & 1 & 6.46 @xmath50 0.24 & 3.39 @xmath50 0.31 & 13.53 @xmath50 0.03 & 1 + 82210 & dk uma & & & & @xmath1 13.8 & 1 + 220657 & @xmath48 peg & & & & @xmath1 13.7 & 1 + 203387 & @xmath11 cap & 1 & 0.22 @xmath50 2.48 & 5.34 @xmath50 1.38 & 13.41 @xmath50 0.16 & 1 + & & 2 & 9.34@xmath50 3.19 & 5.22 @xmath50 2.80 & 13.98 @xmath50 0.36 & 1 + & & 3 & [email protected] & 5.56 @xmath50 1.44 & 14.37 @xmath50 0.32 & 1 + & g191-b2b & 1 & 19.23 @xmath50 0.08 & 3.22 @xmath50 0.10 & 13.76 @xmath50 0.02 & 1 + & & 2 & 8.62 @xmath50 0.19 & 4.34 @xmath50 0.63 & 13.16 @xmath50 0.03 & 1 + & & 1 & 20.3 @xmath50 0.8 & & 13.83@xmath58 & 4 + & & 2 & 13.2 @xmath50 0.8 & & 12.54@xmath59 & 4 + & & 3 & 8.2 @xmath50 0.8 & & 13.08@xmath60 & 4 + & gd 153 & 1 & 5.10 @xmath50 0.22 & 3.15 @xmath50 0.38 & 13.46 @xmath50 0.02 & 1 + llccccl 128620 & @xmath4 cen a & 1 & 18.63 @xmath50 0.03 & 2.64 @xmath50 0.19 & 14.43 @xmath50 0.12 & 1 + 48915b & @xmath4 cma b & 1 & 17.6 @xmath50 1.5 & & 14.53@xmath61 & 2 + & & 2 & 11.7 @xmath50 1.5 & & 13.70@xmath62 & 2 + 22049 & @xmath9 eri & 1 & 18.75 @xmath50 0.15 & 3.87 @xmath50 0.91 & 14.66 @xmath50 0.52 & 1 + 165341 & 70 oph & 1 & 26.5 & 3.32 @xmath50 0.75 & 14.02 @xmath50 0.08 & 1 + & & 2 & 32.9 & 2.87 @xmath50 1.51 & 13.95 @xmath50 0.16 & 1 + & & 3 & 43.3 & 4.76 @xmath50 3.65 & 14.86 @xmath50 0.31 & 1 + 131156a&@xmath19 boo a & 1 & 18.00 @xmath50 0.61 & 3.12 @xmath50 0.73 & 14.36 @xmath50 0.14 & 1 + 20630 & @xmath45 cet & 1 & 21.20 @xmath50 1.24 & 3.02 @xmath50 0.81 & 13.88 @xmath50 0.10 & 1 + & & 2 & 14.46 @xmath50 0.83 & 3.52 @xmath50 0.94 & 14.07 @xmath50 0.09 & 1 + & & 3 & 7.26 @xmath50 1.24 & 2.49 @xmath50 0.37 & 12.90 @xmath50 0.31 & 1 + 62509 & @xmath10 gem & 1 & 32.67 @xmath50 0.38 & 4.05 @xmath50 1.31 & 14.25 @xmath50 0.26 & 1 + & & 2 & 21.39 @xmath50 0.35 & 4.07 @xmath50 0.89 & 14.56 @xmath50 0.26 & 1 + 124897 & @xmath4 boo & 1 & 13.97 @xmath50 0.31 & 3.72 @xmath50 0.77 & 14.32 @xmath50 0.19 & 1 + 33262 & @xmath46 dor & 1 & 13.44 @xmath50 0.81 & 6.12 @xmath50 1.05 & 14.48 @xmath50 0.14 & 1 + & & 2 & 8.25 @xmath50 1.05 & 5.31 @xmath50 2.30 & 14.30 @xmath50 0.26 & 1 + 34029 & @xmath4 aur & 1 & 21.74 @xmath50 0.05 & 3.09 @xmath50 0.29 & 15.04 @xmath50 0.26 & 1 and 3 + 22468 & hr 1099 & 1 & 22.05 @xmath50 0.70 & 3.36 @xmath50 0.90 & 13.93 @xmath50 0.11 & 1 + & & 2 & 14.41 @xmath50 1.35 & 2.16 @xmath50 0.76 & 14.13 @xmath50 0.37 & 1 + & & 3 & 8.11 @xmath50 1.21 & 2.89 @xmath50 1.13 & 13.52 @xmath50 0.13 & 1 + 4128 & @xmath10 cet & 1 & 9.84 @xmath50 0.47 & 2.69 @xmath50 0.78 & 14.12 @xmath50 0.25 & 1 + & & 2 & 3.53 @xmath50 1.31 & 4.77 @xmath50 0.86 & 15.14 @xmath50 0.34 & 1 + 120315 & @xmath47 uma & 1 & & & @xmath1 12.5 & 1 + & & 2 & 3.02 @xmath50 0.15 & 4.00 @xmath50 0.26 & 14.23 @xmath50 0.08 & 1 + 209952 & @xmath4 gru & 1 & 1.90 @xmath50 0.95 & 4.14 @xmath50 1.05 & 15.20 @xmath50 0.47 & 1 + & & 2 & 5.85@xmath50 1.99 & 3.86 @xmath50 1.20 & 14.57 @xmath50 0.17 & 1 + & & 3 & [email protected] & 3.14 @xmath50 1.24 & 13.38 @xmath50 0.05 & 1 + & hz 43 & 1 & 6.40 @xmath50 0.14 & 3.22 @xmath50 0.34 & 14.40 @xmath50 0.14 & 1 + 82210 & dk uma & 1 & 10.47 @xmath50 0.87 & 4.79 @xmath50 2.52 & 15.36 @xmath50 0.70 & 1 + 220657 & @xmath48 peg & 1 & 8.56 @xmath50 1.01 & 3.93 @xmath50 0.82 & 13.75 @xmath50 0.20 & 1 + & & 2 & 1.43 @xmath50 0.85 & 5.19 @xmath50 1.34 & 14.27 @xmath50 0.10 & 1 + & & 3 & 6.67@xmath50 2.26 & 3.18 @xmath50 1.11 & 14.06 @xmath50 0.37 & 1 + 203387 & @xmath11 cap & 1 & 3.18 @xmath50 2.30 & 6.15 @xmath50 1.56 & 14.32 @xmath50 0.18 & 1 + & & 2 & 10.40@xmath50 3.44 & 6.63 @xmath50 2.99 & 14.33 @xmath50 0.21 & 1 + & & 3 & 19.91@xmath50 2.21 & 5.55 @xmath50 0.93 & 15.09 @xmath50 0.37 & 1 + & g191-b2b & 1 & 19.12 @xmath50 0.29 & 2.88 @xmath50 0.36 & 14.71 @xmath50 0.28 & 1 + & & 2 & 8.79 @xmath50 0.16 & 3.69 @xmath50 0.34 & 14.64 @xmath50 0.17 & 1 + & & 1 & 20.3 @xmath50 0.8 & & 14.49@xmath63 & 4 + & & 2 & 13.2 @xmath50 0.8 & & 14.18@xmath64 & 4 + & & 3 & 8.2 @xmath50 0.8 & & 14.45@xmath65 & 4
+ & gd 153 & 1 & 5.37 @xmath50 0.21 & 3.00 @xmath50 0.43 & 14.36 @xmath50 0.24 & 1 + 111812 & 31 com & 1 & 2.62 @xmath50 0.38 & 4.07 @xmath50 1.53 & 15.04 @xmath50 0.49 & 1 + llccccl 128620 & @xmath4 cen a & 1 & 17.81 @xmath50 0.21 & 2.08 @xmath50 0.34 & 11.59 @xmath50 0.09 & 1 + 48915 & @xmath4 cma a & & & & @xmath1 11.8 & 1 + 22049 & @xmath9 eri & & & & @xmath1 11.3 & 1 + 165341 & 70 oph & & & & @xmath1 12.0 & 1 + 131156a & @xmath19 boo a & & & & @xmath1 12.1 & 1 + 216956 & @xmath4 psa & & & & @xmath1 12.6 & 1 + 39587 & @xmath44 ori & & & & @xmath1 11.8 & 1 + 20630 & @xmath45 cet & & & & @xmath1 11.7 & 1 + 102647 & @xmath10 leo & & & & @xmath1 11.9 & 1 + 124897 & @xmath4 boo & 1 & 19.28 @xmath50 3.06 & 2.61 @xmath50 1.76 & 12.06 @xmath50 0.35 & 1 + 33262 & @xmath46 dor & 1 & 13.21 @xmath50 2.82 & 4.34 @xmath50 2.12 & 11.63 @xmath50 0.11 & 1 + & & 2 & 7.88 @xmath50 3.02 & 3.57 @xmath50 1.43 & 11.49 @xmath50 0.15 & 1 + 34029 & @xmath4 aur & 1 & 21.37 @xmath50 0.70 & 2.33 @xmath50 1.07 & 11.45 @xmath50 0.08 & 1 and 2 + 22468 & hr 1099 & & & & @xmath1 12.0 & 1 + 4128 & @xmath10 cet & 1 & 9.83 @xmath50 0.94 & 2.90 @xmath50 1.21 & 11.91 @xmath50 0.13 & 1 + & & 2 & 1.89 @xmath50 0.89 & 4.19 @xmath50 1.59 & 12.02 @xmath50 0.08 & 1 + 209952 & @xmath4 gru & 1 & 2.91 @xmath50 0.61 & 3.28 @xmath50 1.03 & 11.77 @xmath50 0.06 & 1 + & & 2 & 8.94 @xmath50 0.21 & 2.48 @xmath50 0.22 & 11.83 @xmath50 0.02 & 1 + & & 3 & 14.78@xmath50 2.24 & 3.93 @xmath50 1.92 & 10.69 @xmath50 0.11 & 1 + 82210 & dk uma & 1 & 11.19 @xmath50 0.97 & 3.60 @xmath50 1.62 & 12.43 @xmath50 0.20 & 1 + 80007 & @xmath10 car & 1 & 5.27 @xmath50 1.77 & 4.49 @xmath50 0.69 & 12.08 @xmath50 0.21 & 1 + & & 2 & 2.02@xmath50 2.43 & 3.64 @xmath50 0.95 & 11.68 @xmath50 0.31 & 1 + 220657 & @xmath48 peg & 2 & 2.15 @xmath50 0.48 & 4.37 @xmath50 1.26 & 12.14 @xmath50 0.08 & 1 + 203387 & @xmath11 cap & 1 & 2.86 @xmath50 0.77 & 4.64 @xmath50 1.46 & 12.10 @xmath50 0.07 & 1 + & & 2 & 12.92@xmath50 1.88 & 3.55 @xmath50 2.02 & 11.88 @xmath50 0.11 & 1 + & & 3 & & & @xmath1 11.3 & 1 + & g191-b2b & 1 & 20.41 @xmath50 0.38 & 2.49 @xmath50 0.25 & 11.50 @xmath50 0.03 & 1 + & & 2 & 10.56 @xmath50 0.20 & 3.84 @xmath50 0.56 & 11.99 @xmath50 0.01 & 1 + 111812 & 31 com & & & & @xmath1 12.2 & 1 + 45348 & @xmath4 car & 1 & 8.37 @xmath50 0.16 & 3.18 @xmath50 0.25 & 12.05 @xmath50 0.02 & 1 + llccccl 128620 & @xmath4 cen a & 1 & 17.92 @xmath50 1.28 & 2.15 @xmath50 0.60 & 12.83 @xmath50 0.15 & 1 + 48915b & @xmath4 cma b & 1 & 17.6 @xmath50 1.5 & & 12.48@xmath66 & 2 + & & 2 & 11.7 @xmath50 1.5 & & 12.43@xmath67 & 2 + 22049 & @xmath9 eri & 1 & 18.89 @xmath50 0.62 & 3.07 @xmath50 0.46 & 12.76 @xmath50 0.14 & 1 + 165341 & 70 oph & 1 & 26.8 & 3.01 @xmath50 0.48 & 13.16 @xmath50 0.13 & 1 + & & 2 & 33.2 & 2.80 @xmath50 1.31 & 13.88 @xmath50 0.57 & 1 + & & 3 & 43.6 & 2.65 @xmath50 0.56 & 14.42 @xmath50 0.37 & 1 + 131156a&@xmath19 boo a & 1 & 17.62 @xmath50 0.98 & 3.97 @xmath50 2.39 & 14.21 @xmath50 0.32 & 1 + 216956 & @xmath4 psa & 1 & -7.64 @xmath50 1.22 & 2.34 @xmath50 1.13 & 13.69 @xmath50 0.41 & 1 + & & 2 & -11.45 @xmath50 1.40 & 2.82 @xmath50 0.97 & 13.60 @xmath50 0.87 & 1 + 172167 & @xmath4 lyr & 1 & 13.63 @xmath50 1.64 & 1.73 @xmath50 0.84 & 14.76 @xmath50 0.33 & 1 + & & 2 & 17.49 @xmath50 1.95 & 1.83 @xmath50 0.72 & 14.69 @xmath50 0.23 & 1 + & & 3 & 19.99 @xmath50 2.08 & 2.26 @xmath50 0.91 & 14.64 @xmath50 0.25 & 1 + 39587 & @xmath44 ori & 1 & 23.67 @xmath50 4.70 & 3.46 @xmath50 1.95 & 13.59 @xmath50 0.57 & 1 + 20630 & @xmath45 cet & 1 & 21.20 @xmath50 0.59 & 3.05 @xmath50 0.98 & 12.60 @xmath50 0.16 & 1 + & & 2 & 14.17 @xmath50 0.53 & 2.50 @xmath50 1.44 & 12.66 @xmath50 0.16 & 1 + & & 3 & 6.89 @xmath50 1.21 & 2.02 @xmath50 2.29 & 12.46 @xmath50 0.28 & 1 + 102647 & @xmath10 leo & 1 & 6.50 @xmath50 7.49 & 4.15 @xmath50 1.03 & 12.17 @xmath50 0.13 & 1 + & & 2 & 1.50 @xmath50 1.41 & 4.45 @xmath50 0.80 & 13.06 @xmath50 0.04 & 1 + 124897 & @xmath4 boo & 1 & 12.27 @xmath50 0.93 & 2.18 @xmath50 0.62 & 13.71 @xmath50 0.49 & 1 + 33262 & @xmath46 dor & 1 & 13.05 @xmath50 1.81 & 5.79 @xmath50 1.19 & 13.13 @xmath50 0.12 & 1 + & & 2 & 10.48 @xmath50 2.22 & 3.96 @xmath50 3.09 & 12.85 @xmath50 0.20 & 1 + 34029 & @xmath4 aur & 1 & 21.10 @xmath50 0.29 & 2.58 @xmath50 0.56 & 13.00 @xmath50 0.05 & 1 and 3 + 22468 & hr 1099 & & & & @xmath1 13.0 & 1 + 4128 & @xmath10 cet & 1 & 10.02 @xmath50 2.01 & 2.60 @xmath50 1.28 & 13.45 @xmath50 0.50 & 1 + & & 2 & 3.33 @xmath50 1.56 & 3.67 @xmath50 1.44 & 14.08 @xmath50 0.40 & 1 + 120315 & @xmath47 uma & 1 & 2.15 @xmath50 1.64 & 4.25 @xmath50 1.31 & 12.01 @xmath50 0.17 & 1 + & & 2 & 3.24@xmath50 0.05 & 2.61 @xmath50 0.06 & 12.90 @xmath50 0.02 & 1 + 209952 & @xmath4 gru & 1 & 7.95 @xmath50 0.85 & 2.70 @xmath50 0.90 & 13.54 @xmath50 0.08 & 1 + & & 2 & 13.64@xmath50 1.04 & 2.63 @xmath50 0.33 & 14.07 @xmath50 0.48 & 1 + & & 3 & 19.99@xmath50 2.96 & 2.82 @xmath50 1.06 & 11.97 @xmath50 0.31 & 1 + & hz 43 & 1 & 6.54 @xmath50 0.20 & 2.80 @xmath50 0.37 & 12.69 @xmath50 0.06 & 1 + 82210 & dk uma & & & & @xmath1 13.3 & 1 + 80007 & @xmath10 car & 1 & 3.56 @xmath50 0.12 & 3.07 @xmath50 0.11 & 13.62 @xmath50 0.03 & 1 + & & 2 & 4.58@xmath50 0.69 & 2.67 @xmath50 1.94 & 12.89 @xmath50 0.03 & 1 + 220657 & @xmath48 peg & 1 & 10.48 @xmath50 1.79 & 3.95 @xmath50 1.07 & 12.79 @xmath50 0.17 & 1 + & & 2 & 3.17 @xmath50 0.62 & 3.61 @xmath50 0.72 & 13.39 @xmath50 0.09 & 1 + & & 3 & 6.01 @xmath50 0.86 & 2.43 @xmath50 1.35 & 12.84 @xmath50 0.11 & 1 + 203387 & @xmath11 cap & 1 & 0.73 @xmath50 4.35 & 3.57 @xmath50 1.25 & 13.94 @xmath50 0.49 & 1 + & & 2 & 10.81@xmath50 3.22 & 5.00 @xmath50 2.80 & 13.32 @xmath50 0.20 & 1 + & & 3 & 20.18@xmath50 1.80 & 4.54 @xmath50 1.06 & 13.30 @xmath50 0.23 & 1 + & g191-b2b & 1 & 19.61 @xmath50 0.32 & 2.78 @xmath50 0.41 & 12.97 @xmath50 0.05 & 1 + & & 2 & 9.15 @xmath50 0.36 & 3.53 @xmath50 0.37 & 13.42 @xmath50 0.18 & 1 + & & 1 & 20.3 @xmath50 0.8 & & 13.04@xmath68 & 4 + & & 2 & 13.2 @xmath50 0.8 & & 12.92@xmath69 & 4 + & & 3 & 8.2 @xmath50 0.8 & & 13.28@xmath70 & 4 + & gd 153 & 1 & 4.98 @xmath50 0.27 & 2.43 @xmath50 0.52 & 12.64 @xmath50 0.10 & 1 + 111812 & 31 com & & & & @xmath1 13.1 & 1 + 45348 & @xmath4 car & 1 & 8.10 @xmath50 0.08 & 2.30 @xmath50 0.14 & 13.37 @xmath50 0.03 & 1 + lcccccc di & 7.65 & 0.67 & 48 & 13.01 & 0.25 & 54 + cii & 3.64 & 0.81 & 45 & 13.82 & 0.60 & 49 + ni & 3.28 & 0.30 & 17 & 13.51 & 0.24 & 19 + oi & 3.24 & 0.70 & 36 & 14.00 & 0.44 & 38 + mgii & 2.79 & 0.55 & 107 & 12.45 & 0.33 & 107 + alii & 2.80 & 0.60 & 19 & 11.93 & 0.17 & 19 + siii & 2.70 & 0.34 & 41 & 13.08 & 0.34 & 43 + feii & 2.13 & 0.25 & 85 & 12.45 & 0.24 & 91 + llcccccccc 4614 & @xmath47 cas a&g0v & 3.5 & & 122.6 & 5.1 & 6.0 & 11.6 & 13.8 + 103095 & cf uma & g8vp & 6.5 & 98.0&168.5 & 73.8 & 9.2 & 0.4 & 2.5 + 142806 & @xmath71 ser&f6iv&3.9 & 6.7 & 27.7 & 45.7 & 11.1 & 21.2 & 25.0 + 16895 & 13 per & f7v & 4.1 & 25.0 & 141.2 & 9.6 & 11.2 & 18.4 & 21.5 + 157214 & 72 her & g0v & 5.4 & 78.5&55.9 & 32.3 & 14.4 & 17.3&20.0 + 165908 & 99 her & f7v & 5.1 & 1.0 & 57.0 & 22.3 & 15.7 & 17.1 & 19.4 + 99028 & @xmath11 leo&f4iv & 4.0 & 10.3&247.6 & 63.6 & 24.2 & 1.3&3.7 + 15798 & @xmath32 cet&f5v&4.7 & 29.4&191.1 & 63.8 & 25.8 & 17.4 & 21.3 + 17948 & hr860 & f4v & 5.6 & 29.2 & 137.2 & 2.2 & 26.5 & 15.9 & 18.3 + 198084 & sao 32862 & f8iv - v&4.5 & 31.4&94.4 & 9.1 & 27.1 & 1.9 & 1.7 + 157466 & sao 85045&f8v & 6.9 & 28 & 47.5 & 29.8 & 29.8 & 19.7 & 22.6 + 13555 & @xmath47 ari&f5v & 5.2 & 6.0 & 147.2 & 37.8 & 30.1 & 19.6 & 23.6 + 184499 & sao 68491 & g0v & 6.6 & 163.3 & 67.0 & 6.5 & 32.0 & 12.7 & 13.8 + 109799 & hr4803 & f1iv & 5.4 & 0.9&299.2 & 35.6 & 34.6 & 12.1 & 15.4 + 43905 & 45 aur & m6iii & 5.3 & 0.5&161.2 & 17.3 & 57.0 & 19.2 & 21.1 + 140283 & sao 159459 & f3 & 7.2 & 171.4 & 356.3 & 33.6 & 57.3 & 24.3 & 28.4 + & g191-b2b & daw & 11.6 & 5 & 156.0 & 7.1 & 68.8 & 20.3 & 22.7 + & gd 153 & daw & 13.4 & & 317.3 & 84.8 & 70.5 & 8.7 & 12.0 + 45348 & @xmath4 car & f0ii & 0.7 & 20.5 & 261.2 & 25.3 & 95.9 & 8.8 & 10.1 + llllccclcl 4614 & @xmath47 cas a&stis&e230h&2374 - 2646 & 114000 & 2060.0 & boesgaard & 8770 & 2001 feb 14 + 103095 & cf uma & stis & e230h & 2624 - 2895 & 114000 & 2904.0 & peterson & 9455 & 2003 jul 19 + 142860 & @xmath71 ser&stis&e230h&2374 - 2646 & 114000 & 1822.0 & boesgaard & 8770 & 2001 aug 31 + 16895 & 13 per & stis & e230h & 2374 - 2646 & 114000 & 1979.0 & boesgaard & 8770 & 2000 dec 12 + 157214 & 72 her & stis & e230h & 2374 - 2646 & 114000 & 1870.0 & boesgaard & 8770 & 2001 mar 1 + 165908 & 99 her & stis & e230h & 2374 - 2646 & 114000 & 1870.0 & boesgaard & 8770 & 2000 nov 3 + 99028 & @xmath11 leo&stis&e230h & 2374 - 2646 & 114000 & 1824.0 & boesgaard & 8770 & 2001 feb 8 + 15798 & @xmath32 cet&stis&e230h & 2374 - 2646 & 114000 & 1822.0 & boesgaard & 8770 & 2000 dec 18 + 17948 & hr860 & stis & e230h & 2374 - 2646 & 114000 & 3690.0 & deliyannis & 9048 & 2002 aug 24 + 198084 & sao 32862 & stis&e230h & 2374 - 2646 & 114000 & 3960.0 & deliyannis & 9048 & 2001 nov 28 + 157466 & sao 85045 & stis & e230h & 2374 - 2646 & 114000 & 5812.0 & peterson & 9455 & 2002 sep 18 + & & stis & e230h & 2624 - 2895 & 114000 & 2906.0 & peterson & 9455 & 2002 sep 18 + 13555 & @xmath47 ari & stis&e230h & 2374 - 2646 & 114000 & 1831.0 & boesgaard & 8770 & 2001 sep 26 + 184499 & sao 68491 & stis & e230h & 2374 - 2646 & 114000 & 2945.0 & peterson & 9455 & 2002 sep 11 + & & stis & e230h & 2624 - 2895 & 114000 & 2945.0 & peterson & 9455 & 2002 sep 11 + 109799 & hr 4803 & stis & e230h & 2374 - 2646 & 114000 & 1956.0 & deliyannis & 9048 & 2002 jul 19 + 43905 & 45 aur & stis & e230h & 2374 - 2646 & 114000 & 2144.0 & deliyannis & 9048 & 2001 nov 16 + 140283 & sao 159459&stis & e230h & 2374 - 2646 & 114000 & 2899.0 & peterson & 9455 & 2002 aug 21 + & g191-b2b & stis & e230h & 2374 - 2646 & 114000 & 8304.0 & leitherer & 8421 & 2000 mar 18 + & & stis & e230h & 2274 - 2546 & 114000 & 774.0 & valenti & 8915 & 2001 sep 12 + & & stis & e230h & 2074 - 2346 & 114000 & 620.0 & valenti & 8915 & 2001 sep 18 + & & stis & e230h & 2124 - 2396 & 114000 & 710.6 & valenti & 8915 & 2001 sep 18
+ & & stis & e230h & 2174 - 2446 & 114000 & 734.3 & valenti & 8915 & 2001 sep 18 + & & stis & e230h & 2224 - 2496 & 114000 & 748.7 & valenti & 8915 & 2001 sep 18 + & & stis & e230h & 2324 - 2596 & 114000 & 668.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2374 - 2646 & 114000 & 696.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2424 - 2696 & 114000 & 731.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2474 - 2746 & 114000 & 769.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2524 - 2796 & 114000 & 992.3 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2574 - 2846 & 114000 & 900.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2624 - 2895 & 114000 & 978.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2674 - 2945 & 114000 & 1097.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2724 - 2995 & 114000 & 1232.0 & valenti & 8915 & 2001 sep 19 + & & stis & e230h & 2774 - 3045 & 114000 & 1422.0 & valenti & 8915 & 2001 sep 19 + & gd 153 & stis & e230h & 2374 - 2646 & 114000 & 10117.0 & sahu & 8595 & 2001 mar 22 + 45348 & @xmath4 car&stis&e230h & 2374 - 2646 & 114000 & 600.0 & brown & 7764 & 2002 jun 11 + & & stis & e230h & 2624 - 2895 & 114000 & 600.0 & brown & 7764 & 2002 jun 11 + llccccl 48915 & @xmath4 cma a & 1 & 17.6 @xmath50 1.5 & & 12.23@xmath58 & 1 + & & 2 & 11.7 @xmath50 1.5 & & 12.00@xmath72 & 1 + 103095 & cf uma & 1 & 2.05 @xmath50 0.07 & 2.08 @xmath50 0.20 & 12.58 @xmath50 0.14 & 2 + 157466 & sao 85045 & 1 & 19.01 @xmath50 0.08 & 2.40 @xmath50 0.27 & 14.15 @xmath50 0.59 & 2 + 184499 & sao 68491 & 1 & 14.00 @xmath50 0.58 & 3.22 @xmath50 0.38 & 13.68 @xmath50 0.35 & 2 + & & 2 & 19.80 @xmath50 0.82 & 3.71 @xmath50 0.60 & 12.94 @xmath50 0.27 & 2 + & & 3 & 27.85 @xmath50 0.35 & 1.93 @xmath50 0.48 & 11.97 @xmath50 0.09 & 2 + 30738 & sao 94162 & 1 & 20.3 @xmath50 0.4 & 3.8 @xmath50 0.5 & 12.48 @xmath50 0.10 & 3 + & g191-b2b & 1 & 19.15 @xmath50 0.05 & 2.74 @xmath50 0.10 & 12.73 @xmath50 0.04 & 2 + & & 2 & 8.51 @xmath50 0.04 & 3.22 @xmath50 0.38 & 13.83 @xmath50 0.52 & 2 + 45348 & @xmath4 car & 1 & 8.56 @xmath50 0.23 & 3.39 @xmath50 0.59 & 13.23 @xmath50 0.13 & 2 + & & & 20 to 35 & & & 2 + llccccl 48915 & @xmath4 cma a & 1 & 17.6 @xmath50 1.5 & & 11.94@xmath73 & 1 + & & 2 & 11.7 @xmath50 1.5 & & 11.74 @xmath50 0.03 & 1 + 61421 & @xmath4 cmi & 1 & 22.0 @xmath50 0.10 & 2.1 @xmath50 0.10 & 11.94 @xmath50 0.03 & 2 + & & 2 & 19.0 @xmath50 0.10 & 2.1 @xmath50 0.10 & 12.05 @xmath50 0.02 & 2 + 4614 & @xmath47 cas a & 1 & 11.18 @xmath50 0.40 & 3.49 @xmath50 0.43 & 12.43 @xmath50 0.09 & 3 + & & 2 & 5.80 @xmath50 0.87 & 2.00 @xmath50 0.89 & 11.67 @xmath50 0.24 & 3 + 216956 & @xmath4 psa & 1 & 3.2 @xmath50 1.5 & 2.7 & 12.58 & 4 + & & 2 & 9.7 @xmath50 1.5 & 2.8 & 12.26 & 4 + 172167 & @xmath4 lyr & 1 & 12.7 @xmath50 1.5 & 2.25 & 13.03 & 5 + & & 2 & 16.0 @xmath50 1.5 & 1.53 & 12.84 & 5 + & & 3 & 18.3 @xmath50 1.5 & 1.83 & 12.81 & 5 + 142806 & @xmath71 ser & 1 & 18.19@xmath50 0.51 & 2.15 @xmath50 0.62 & 12.33 @xmath50 0.10 & 3 + & & 2 & 22.12@xmath50 0.19 & 1.55 @xmath50 0.27 & 12.62 @xmath50 0.08 & 3 + 16895 & 13 per & 1 & 16.45 @xmath50 0.21 & 2.15 @xmath50 0.13 & 12.61 @xmath50 0.02 & 3 + 157214 & 72 her & 1 & 15.51@xmath50 0.31 & 2.82 @xmath50 0.70 & 14.32 @xmath50 0.44 & 3 + & & 2 & 25.95@xmath50 0.20 & 2.46 @xmath50 0.16 & 12.85 @xmath50 0.07 & 3 + 165908 & 99 her & 1 & 17.43@xmath50 0.30 & 2.28 @xmath50 0.29 & 13.27 @xmath50 0.14 & 3 + & & 2 & 22.90@xmath50 0.56 & 2.51 @xmath50 0.29 & 12.92 @xmath50 0.06 & 3 + 99028 & @xmath11 leo & 1 & 4.99 @xmath50 0.56 & 2.22 @xmath50 0.91 & 12.05 @xmath50 0.09 & 3 + & & 2 & 1.97 @xmath50 0.66 & 2.10 @xmath50 0.89 & 12.16 @xmath50 0.09 & 3 + 15798 & @xmath32 cet & 1 & 21.85 @xmath50 0.71 & 2.56 @xmath50 0.22 & 13.03 @xmath50 0.25 & 3 + & & 2 & 15.99 @xmath50 0.57 & 2.87 @xmath50 1.02 & 13.19 @xmath50 0.10 & 3 + & & 3 & 9.74 @xmath50 0.23 & 2.07 @xmath50 0.60 & 12.48 @xmath50 0.16 & 3 + 17948 & hr860 & 1 & 15.10 @xmath50 0.40 & 1.67 @xmath50 0.24 & 12.60 @xmath50 0.03 & 3 + 198084 & sao 32862 & 1 & 2.60 @xmath50 0.15 & 2.06 @xmath50 0.08 & 12.80 @xmath50 0.02 & 3 + & & 2 & 12.88@xmath50 0.20 & 2.03 @xmath50 0.30 & 11.96 @xmath50 0.04 & 3 + 157466 & sao 85045 & 1 & 19.42 @xmath50 0.45 & 2.40 @xmath50 0.44 & 13.70 @xmath50 0.36 & 3 + 13555 & @xmath47 ari & 1 & 16.99 @xmath50 0.36 & 2.66 @xmath50 0.36 & 12.38 @xmath50 0.05 & 3 + 184499 & sao 68491 & 1 & 14.35 @xmath50 0.30 & 1.96 @xmath50 0.23 & 14.36 @xmath50 0.82 & 3 + & & 2 & 19.55 @xmath50 1.28 & 2.82 @xmath50 1.23 & 12.51 @xmath50 0.25 & 3 + & & 3 & 27.87 @xmath50 0.63 & 2.23 @xmath50 1.46 & 11.78 @xmath50 0.08 & 3 + 109799 & hr 4803 & 1 & 0.37 @xmath50 0.29 & 1.69 @xmath50 0.27 & 12.61 @xmath50 0.07 & 3 + & & 2 & 7.52 @xmath50 0.30 & 1.89 @xmath50 0.45 & 13.43 @xmath50 0.41 & 3 + & & 3 & 11.41 @xmath50 0.34 & 1.57 @xmath50 0.50 & 12.77 @xmath50 0.15 & 3 + 43905 & 45 aur & 1 & 18.47 @xmath50 0.29 & 1.53 @xmath50 0.24 & 12.47 @xmath50 0.03 & 3 + & & 2 & 11.87 @xmath50 0.63 & 2.21 @xmath50 0.85 & 12.37 @xmath50 0.14 & 3 + & & 3 & 8.14 @xmath50 0.73 & 2.61 @xmath50 0.66 & 12.56 @xmath50 0.12 & 3 + 140283 & sao 159459 & 1 & 22.48 @xmath50 1.33 & 2.43 @xmath50 0.83 & 13.34 @xmath50 0.31 & 3 + & & 2 & 28.55 @xmath50 1.04 & 3.62 @xmath50 0.92 & 14.50 @xmath50 0.40 & 3 + & g191-b2b & 1 & 19.09 @xmath50 0.80 & 2.44 @xmath50 0.35 & 12.44 @xmath50 0.07 & 3 + & & 2 & 9.17 @xmath50 0.75 & 3.55 @xmath50 0.14 & 12.93 @xmath50 0.01 & 3 + & gd 153 & 1 & 5.21 @xmath50 0.39 & 1.59 @xmath50 0.72 & 12.21 @xmath50 0.12 & 3 + 45348 & @xmath4 car & 1 & 8.25 @xmath50 0.23 & 2.40 @xmath50 0.23 & 13.08 @xmath50 0.06 & 3 + & & & 20 to 35 & & & 3 + | moderate and high - resolution measurements ( @xmath040,000 ) of interstellar resonance lines of , , , , , and ( hereafter called light ions ) are presented for all available observed targets located within 100 pc which also have high - resolution observations of interstellar or ( heavy ions ) lines .
all spectra were obtained with the goddard high resolution spectrograph ( ghrs ) or the space telescope imaging spectrograph ( stis ) instruments aboard the hubble space telescope ( hst ) . currently , there are 41 sightlines to targets within 100pc with observations that include a heavy ion at high resolution and at least one light ion at moderate or high resolution .
we present new measurements of light ions along 33 of these sightlines , and collect from the literature results for the remaining sightlines that have already been analyzed .
for all of the new observations we provide measurements of the central velocity , doppler width parameter , and column density for each absorption component .
we greatly increase the number of sightlines with useful lism absorption line measurements of light ions by using knowledge of the kinematic structure along a line of sight obtained from high resolution observations of intrinsically narrow absorption lines , such as and .
we successfully fit the absorption lines with this technique , even with moderate resolution spectra . because high resolution observations of heavy ions are critical for understanding the kinematic structure of local absorbers along the line of sight
, we include 18 new measurements of and in an appendix .
we present a statistical analysis of the lism absorption measurements , which provides an overview of some physical characteristics of warm clouds in the lism , including temperature and turbulent velocity . this complete collection and reduction of all available
lism absorption measurements provides an important database for studying the structure of nearby warm clouds , including ionization , abundances , and depletions .
subsequent papers will present models for the morphology and physical properties of individual structures ( clouds ) in the lism . |
a central problem in quantum information theory @xcite is the formulation of appropriate measures that quantify the degree of entanglement in composite systems .
particularly important entanglement measures are the concurrence @xcite and the entanglement of formation @xcite .
these quantities have been widely used in many applications .
examples include studies on the role of entanglement in quantum phase transitions @xcite , on the emergence of long - distance entanglement in spin systems @xcite , and on additivity properties of the holevo capacity of quantum channels @xcite .
the explicit determination of most of the proposed entanglement measures for a generic state @xmath1 is an extremely demanding task that requires the solution of a high - dimensional optimization problem .
the development of analytical lower bounds for the various entanglement measures is therefore of great interest .
recently , chen , albeverio , and fei @xcite have derived such bounds for the concurrence @xmath2 and for the entanglement of formation @xmath3 .
they achieved this by relating @xmath2 and @xmath3 to two important and strong separability criteria , namely to the peres criterion of positive partial transposition ( ppt ) @xcite and to the realignment criterion @xcite . according to these criteria
a given state @xmath1 is entangled ( inseparable ) if the trace norms @xmath4 or @xmath5 are strictly larger than 1 , where @xmath6 denotes the partial transposition and @xmath7 the realignment transformation . in refs .
@xcite tight lower bounds for @xmath2 and @xmath3 have been formulated in terms of these trace norms . here ,
we extend the connection between separability criteria and entanglement measures to a new criterion which has been developed recently @xcite .
this criterion is based on a universal nondecomposable positive map which leads to a class of optimal entanglement witnesses . employing these witnesses we derive analytical lower bounds for the concurrence and for the entanglement of formation that can be expressed in terms of a simple linear functional of the given state @xmath1 .
the entanglement witnesses constructed here have the special feature of being nondecomposable optimal .
this notion has been introduced in refs .
@xcite to characterize optimality properties of entanglement witnesses .
it means that the witnesses are able to identify entangled ppt states and that no other witnesses exist which can detect more such states .
it follows that the bounds developed here can be sharper than those obtained from the ppt criterion and that they are particularly efficient near the border that separates the ppt entangled states from the separable states .
in addition , we will demonstrate that they can also be stronger than the bounds given by the realignment criterion .
hence , the new bounds complement and considerably improve the existing bounds .
the paper is organized as follows . in sec .
[ maps ] we introduce a new separability criterion which is based on a nondecomposable positive map that operates on the states of hilbert spaces with even dimension @xmath8 .
we formulate and prove the most important properties of this map , and derive the associated class of optimal entanglement witnesses .
analytical lower bounds for the concurrence are developed in sec .
[ sec - concurrence ] . in sec .
[ sec - example ] we discuss an example of a certain family of states in arbitrary dimensions .
it will be demonstrated explicitly with the help of this example that the new bounds can be much sharper than the bounds of the ppt and of the realignment criterion . the new class of entanglement witnesses is used in sec .
[ sec - eof ] to develop corresponding lower bounds for the entanglement of formation . finally , some conclusions are drawn in sec .
[ conclu ] .
we consider a quantum system with finite - dimensional hilbert space @xmath9 . without loss of generality
one can regard @xmath9 as the state space of a particle with a certain spin @xmath10 , where @xmath11 . as usual , the corresponding basis states are denoted by @xmath12 , where the quantum number @xmath13 takes on the values @xmath14 .
we will develop a necessary condition for the separability of mixed quantum states which employs the symmetry transformation of the time reversal @xcite .
in quantum mechanics the time reversal is to be described by an antiunitary operator @xmath15 . as for any antiunitary operator ,
we can write @xmath16 , where @xmath17 denotes the complex conjugation in the chosen basis @xmath12 , and @xmath18 is a unitary operator . in the spin representation introduced above the matrix elements of @xmath18
are given by @xmath19 .
for even @xmath20 , i. e. , for half - integer spins @xmath10 , this matrix is not only unitary but also skew - symmetric , which means that @xmath21 , where @xmath22 denotes the transposition .
it follows that @xmath23 which leads to @xmath24 this relation expresses a well - known property of the time reversal transformation @xmath15 which will play a crucial role in the following : for any state vector @xmath25 the time - reversed state @xmath26 is orthogonal to @xmath25 .
this is a distinguished feature of even - dimensional state spaces , because unitary and skew - symmetric matrices do not exist in state spaces with odd dimension . the action of the time reversal transformation on an operator @xmath27 on @xmath9 can be expressed by @xmath28 this defines a linear map @xmath29 which transforms any operator @xmath27 to its time reversed operator @xmath30 .
for example , if we take the spin operator @xmath31 of the spin-@xmath10 particle this gives the spin flip transformation @xmath32 .
a positive map is a linear transformation @xmath33 which takes any positive operator @xmath27 on some state space @xmath34 to a positive operator @xmath35 , i. e. , @xmath36 implies that @xmath37 .
a positive map @xmath33 is said to be completely positive if it has the additional property that the map @xmath38 operating on any composite system with state space @xmath39 is again positive , where @xmath40 denotes the unit map .
the physical significance of positive maps in entanglement theory is provided by a fundamental theorem established in ref .
@xcite . according to this theorem a necessary and sufficient condition for a state @xmath1 to be separable
is that the operator @xmath41 is positive for any positive map @xmath33 .
hence , maps which are positive but not completely positive can be used as indicators for the entanglement of certain sets of states .
an important example for a positive but not completely positive map is given by the transposition map @xmath22 .
the inequality @xmath42 represents a strong necessary condition for separability known as the peres criterion of positive partial transposition ( ppt criterion ) .
the second relation of eq .
( [ def - theta ] ) shows that the time reversal transformation @xmath29 is unitarily equivalent to the transposition map @xmath22 .
hence , the ppt criterion is equivalent to the condition that the partial time reversal @xmath43 is positive : @xmath44 we define a linear map @xmath45 which acts on operators @xmath27 on @xmath9 as follows @xcite @xmath46 where @xmath47 denotes the trace of @xmath27 and @xmath40 is the unit operator .
this map has first been introduced in ref .
@xcite for the special case @xmath48 , in order to study the entanglement structure of su(2)-invariant spin systems @xcite . for any even @xmath8 the map
@xmath45 defined by eq .
( [ phi ] ) has the following features : ( a ) : : @xmath45 is a positive but not completely positive map .
( b ) : : the map @xmath45 is nondecomposable .
( c ) : : the entanglement witnesses corresponding to @xmath45 are nondecomposable optimal . in the following we
briefly explain and prove these statements . * ( a ) * we first demonstrate that @xmath45 is a positive map . to this end
, we have to show that the operator @xmath49 is positive for any normalized state vector @xmath25 . using definition ( [ phi ] ) we find : @xmath50 because of eq .
( [ ortho ] ) the operator @xmath51 introduced here represents an orthogonal projection operator which projects onto the subspace spanned by @xmath25 and @xmath26 .
it follows that also @xmath49 is a projection operator and , hence , that it is positive for any normalized state vector @xmath25 .
this proves that @xmath45 is a positive map .
we remark that for @xmath52 the projection @xmath51 is identical to the unit operator such that @xmath45 is equal to the zero map in this case . for this reason
we restrict ourselves to the cases of even @xmath0 .
it should be emphasized that @xmath45 would not be positive if we had used the transposition @xmath22 instead of the time reversal @xmath29 in the definition ( [ phi ] ) .
the positivity of @xmath45 implies that the inequality @xmath53 provides a necessary condition for separability : any state @xmath1 which violates this condition must be entangled . to show that @xmath45 is not completely positive , i. e. , that the condition ( [ trc ] ) is nontrivial
, we consider the tensor product space @xmath54 of two spin-@xmath10 particles .
the total spin of the composite system will be denoted by @xmath55 . according to the triangular inequality @xmath55 takes on the values @xmath56 .
the projection operator which projects onto the manifold of states corresponding to a definite value of @xmath55 will be denoted by @xmath57 . in particular
, @xmath58 represents the one - dimensional projection onto the maximally entangled singlet state @xmath59 .
we define a hermitian operator @xmath60 by applying @xmath61 to the singlet state : @xmath62 where the factor @xmath20 is introduced for convenience .
more explicit expressions for @xmath60 can be obtained as follows .
first , we observe that @xmath63 since @xmath58 is a maximally entangled state ( @xmath64 denotes the partial trace taken over subsystem 2 ) .
second , we note that the partial time reversal of the singlet state is given by the formula @xcite @xmath65 where @xmath66 denotes the swap operator defined by @xmath67 using then definition ( [ phi ] ) we get @xmath68 another useful representation is obtained by use of the fact that the sum of the @xmath57 is equal to the unit operator . expressing @xmath66 as shown in eq .
( [ theta - p0 ] ) we then find : @xmath69 we infer from eq .
( [ def - w-3 ] ) that @xmath60 has the negative eigenvalue @xmath70 corresponding to the singlet state @xmath59 . therefore , the operator @xmath60 is not positive and , hence , the map @xmath45 is not completely positive . *
( b ) * since @xmath45 is positive but not completely positive the operator @xmath60 defined in eq .
( [ def - w ] ) is an entanglement witness @xcite .
we recall that an entanglement witness is an observable which satisfies @xmath71 for all separable states @xmath72 , and @xmath73 for at least one inseparable state @xmath1 , in which case we say that @xmath60 detects @xmath1 .
an entanglement witness @xmath60 is called nondecomposable if it can detect entangled ppt states @xcite , i. e. , if there exist ppt states @xmath1 that satisfy @xmath74
. we will demonstrate in sec .
[ sec - example ] by means of an explicit example that there are always such states for the witness defined by eq .
( [ def - w ] ) .
it follows that our witness @xmath60 is nondecomposable .
this implies that also the map @xmath45 is nondecomposable @xcite , and that the criterion ( [ trc ] ) is able to detect entangled ppt states . *
( c ) * the observable @xmath60 introduced above has a further remarkable optimality property . to explain this property we introduce the following notation @xcite .
we denote by @xmath75 the set of all entangled ppt states of the total state space which are detected by some given nondecomposable witness @xmath60 . a witness @xmath76
is said to be finer than a witness @xmath77 if @xmath78 is a subset of @xmath79 , i. e. , if all entangled ppt states which are detected by @xmath77 are also detected by @xmath76 .
a given witness is said to be nondecomposable optimal if there is no other witness which is finer , i. e. , if there is no other witness which is able to detect more entangled ppt states .
it can be shown that the witness @xmath60 defined by ( [ def - w ] ) is always optimal in this sense .
the proof can be carried out by showing that the set of product vectors @xmath80 satisfying @xmath81 spans the total hilbert space .
the details of the proof are given in ref .
the generalized concurrence of a pure state @xmath82 is defined by @xcite @xmath83 where @xmath84 represents the reduced density matrix of subsystem 1 , given by the partial trace taken over subsystem 2 .
we consider the schmidt decomposition @xmath85 where @xmath86 and @xmath87 are orthonormal bases in @xmath88 and @xmath34 , respectively , and the @xmath89 are the schmidt coefficients satisfying @xmath90 and the normalization condition @xmath91 the concurrence can then be expressed in terms of the schmidt coefficients : @xmath92 for a mixed state @xmath1 the concurrence is defined to be @xmath93 where the minimum is taken over all possible convex decompositions of @xmath1 into an ensemble @xmath94 of pure states with probability distribution @xmath95 .
let @xmath96 be an optimal decomposition of @xmath1 for which the minimum of eq .
( [ concurrence ] ) is attained . denoting the schmidt coefficients of @xmath97 by @xmath98
we then have : @xmath99 in the second line we have used eq .
( [ concu - pure ] ) , and the third line is obtained with the help of the inequality @xmath100 which holds for any set of @xmath20 numbers @xmath89 @xcite .
consider now any real - valued and convex functional @xmath101 on the total state space with the following property .
for all state vectors @xmath102 with schmidt decomposition ( [ schmidt ] ) we have : @xmath103 given such a functional we can continue inequality ( [ concu-1 ] ) as follows : @xmath104 in the second line we have used inequality ( [ prop - f ] ) , and in the third line the convexity of @xmath101 .
we conclude that any convex functional @xmath101 with the property ( [ prop - f ] ) leads to a lower bound for the concurrence : @xmath105 in ref .
@xcite two example for such a functional @xmath101 have been constructed which are based on the ppt criterion and on the realignment criterion : @xmath106 where @xmath6 denotes the partial transposition and @xmath7 the realignment transformation .
these functionals are convex because of the convexity of the trace norm which is defined by @xmath107 . moreover , for both functionals the equality sign of eq .
( [ prop - f ] ) holds : @xmath108 consider the functional @xmath109 where @xmath60 is the entanglement witness introduced in eq .
( [ def - w ] ) .
this functional is linear and of course convex .
we claim that @xmath110 also satisfies the bound ( [ prop - f ] ) , i. e. , for any state vector @xmath102 with schmidt decomposition ( [ schmidt ] ) we have @xmath111 to show this we first determine the expectation value of @xmath60 . from eq .
( [ theta - p0 ] ) we have @xmath112 and , hence , the expression ( [ def - w-2 ] ) can be written as @xmath113 .
this gives @xmath114 with the help of the definitions of the swap operator @xmath66 [ eq . ( [ swap ] ) ] and of the time reversal transformation [ eq .
( [ def - theta ] ) ] it is easy to verify the formulae @xmath115 this leads to @xmath116 where @xmath117 hence , we have @xmath118 it is shown in appendix [ app - a ] that @xmath119 this leads immediately to the desired inequality : @xmath120 where we have used the normalization condition ( [ normalization ] ) .
summarizing we have obtained the following lower bound for the concurrence : @xmath121 of course , this bound is only nontrivial if @xmath1 is detected by the entanglement witness @xmath60 , i. e. , if @xmath122 .
it will be demonstrated in sec .
[ sec - example ] that this bound can be much stronger than the bounds given by @xmath123 and @xmath124 , which is due to the fact @xmath60 identifies many entangled states that are neither detected by the ppt criterion nor by the realignment criterion .
we illustrate the application of the inequality ( [ newbound - concu ] ) with the help of a certain family of states .
this family contains a separable state , entangled ppt states , as well as entangled states whose partial transposition is not positive .
the example will also lead to a proof of the claim that the map @xmath45 and the witness @xmath60 are nondecomposable .
consider the following one - parameter family of states : @xmath125 these normalized states are mixtures of the singlet state @xmath58 and of the state @xmath126 where @xmath127 denotes the projection onto the symmetric subspace under the swap operation @xmath66 .
we note that @xmath128 is a separable state which belongs to the class of the werner states @xcite .
since @xmath127 can be written as a sum over the projections @xmath57 with odd @xmath55 , we immediately get with the help of eq .
( [ def - w ] ) : @xmath129 hence , we find that @xmath130 for @xmath131 .
it follows that all states of the family ( [ family ] ) corresponding to @xmath131 are entangled , and that @xmath128 is the only separable state of this family . employing eqs .
( [ w - rho ] ) and ( [ newbound - concu ] ) we get the following lower bound for the concurrence : @xmath132 to compare this bound with those obtained from the ppt and the realignment criterion we have to determine the trace norms @xmath133 and @xmath134 .
the details of the calculation are presented in appendix [ app - b ] .
one finds that the ppt criterion gives the bounds : @xmath135 , & \frac{1}{n+2 } \leq \lambda \leq \frac{1}{2 } \\ \sqrt{\frac{2(n-1)}{n } } \frac{n\lambda-1}{n-1 } , & \frac{1}{2 } \leq \lambda \end{array } \right . \nonumber\end{aligned}\ ] ] while the realignment criterion yields : @xmath136 the relations ( [ new - bound])-([realign - bound ] ) lead to a number of important conclusions .
first of all , we observe from eq .
( [ ppt - bound ] ) that the states within the range @xmath137 have positive partial transposition ( in this range @xmath133 is equal to @xmath138 , see appendix [ app - b ] ) .
but from eq .
( [ w - rho ] ) we know that all states with @xmath139 must be entangled .
it follows that all states in the range @xmath140 are entangled ppt states which are detected by the witness @xmath60 .
this proves , as claimed in sec .
[ maps ] , that the witness @xmath60 and , hence , also the map @xmath45 are nondecomposable . according to eq .
( [ ppt - bound ] ) the ppt criterion only detects the entanglement of the states with @xmath141 .
it is thus weaker than the criterion based on the witness @xmath60 . as can be seen from eq .
( [ realign - bound ] ) the realignment criterion is even weaker because it only recognizes the entanglement of the states with @xmath142 ( the trace norm of @xmath143 is larger than @xmath138 if and only if @xmath142 , see appendix [ app - b ] ) . a plot of the various lower bounds for the example @xmath48 is shown in fig .
[ figure1 ] .
we see that the new bound ( [ new - bound ] ) is the best one within the range @xmath144 .
the bounds given by the ppt and the realignment criterion coincide in the range @xmath145 . in this range
they are better than the new bound .
note that these features hold true for all @xmath20 .
we remark that for large @xmath20 the concurrence approaches the limit @xmath146 . ) for @xmath48 .
solid line : the new lower bound given by eq .
( [ new - bound ] ) . dashed line :
lower bound given by the ppt criterion [ eq .
( [ ppt - bound ] ) ] .
dashed - dotted line : lower bound given by the realignment criterion [ eq . ( [ realign - bound ] ) ] .
dotted line : upper bound given by @xmath147.[figure1 ] ]
for a pure state @xmath102 with schmidt decomposition ( [ schmidt ] ) one defines the entanglement of formation by @xmath148 where @xmath149 denotes the vector of the schmidt coefficients , and @xmath150 is the base @xmath151 logarithm .
the quantity @xmath152 is the shannon entropy of the distribution @xmath153 , which is equal to the von neumann entropy of the reduced density matrices .
this definition is extended to mixed states through the convex hull construction : @xmath154 where the minimum is again taken over all possible convex decompositions of @xmath1 .
an analytical lower bound for the entanglement of formation has been constructed in ref .
@xcite , which may be described as follows .
first , for @xmath155 one defines the function @xmath156 the minimum is taken over all schmidt vectors @xmath149 , i. e. , @xmath157 is the minimal value of the entropy @xmath152 under the constraint @xmath158 .
the solution of this minimization problem has been derived by terhal and vollbrecht @xcite : @xmath159 \log ( n-1),\ ] ] where @xmath160 ^ 2,\ ] ] and @xmath161 is the binary entropy .
second , one introduces the convex hull @xmath162 $ ] of @xmath157 .
this is the largest convex function which is bounded from above by @xmath157 .
one then gets the following lower bound for the entanglement of formation : @xmath163 , \nonumber \\
\lambda_0 & \equiv & { \mathrm{max } } \left\ { ||t_2\rho|| , ||{\mathcal{r}}\rho|| \right\}.\end{aligned}\ ] ] the decisive point of the construction given in ref .
@xcite is the fact that [ see eq .
( [ ppt - realign ] ) ] @xmath164 this means that the function @xmath157 yields the minimal entropy @xmath152 under the constraint of a fixed value @xmath33 for the trace norm @xmath165 or @xmath166 .
but from eq . ( [ ineq - fw ] ) we also have @xmath167 by use of this inequality one can immediately repeat the proof of ref .
@xcite , replacing the trace norm @xmath4 or @xmath5 by the quantity @xmath168 .
hence , we are led to a sharper bound for the entanglement of formation : @xmath169 , \nonumber
\\ \lambda_0 & \equiv & { \mathrm{max } } \left\ { ||t_2\rho|| , ||{\mathcal{r}}\rho||,1-{\mathrm{tr}}(w\rho ) \right\}.\end{aligned}\ ] ] ) for @xmath48 .
solid line : the new lower bound given by eqs .
( [ bounds - eof-2])-([eof - bound2 ] ) . dashed line :
lower bound given by eq .
( [ bounds - eof-1 ] ) .
dotted line : upper bound given by @xmath170.[figure2 ] ] let us apply this result to the family of states given in eq .
( [ family ] ) . in this case
we have : @xmath171 by use of the terhal - vollbrecht conjecture @xcite on the form of the function @xmath162 $ ] ( see also ref .
@xcite ) we get : @xmath172 \log ( n-1)\ ] ] for @xmath173 , and @xmath174 for @xmath175 .
the general features of this result are similar to those discussed within the context of the concurrence . the special case @xmath48 is plotted in fig . [ figure2 ] .
we finally note that @xmath176 represents the asymptotic limit of the entanglement of formation for large @xmath20 .
by use of a universal positive map @xmath45 we have obtained a class of nondecomposable optimal entanglement witnesses @xmath60 . employing
these witnesses analytical bounds for the concurrence and for the entanglement of formation have been developed .
similar bounds can be derived for other measures , e. g. , for the entanglement measure which is known as tangle @xcite . due to the fact that @xmath60 is a nondecomposable optimal entanglement witness , the bounds obtained here are particularly good near the boundary which separates the region of classically correlated states from the region of entangled states with positive partial transposition .
it should be clear from the general considerations in secs .
[ sec - concurrence ] and [ sec - eof ] and from the example of sec .
[ sec - example ] that the bounds derived here are not intended to _ replace _ other known bounds , but rather to _
complement _ these .
in fact , the bounds based on the witness @xmath60 can be weaker than those given by ppt or the realignment criterion , in particular in those cases in which the optimal decomposition of @xmath1 consists entirely of maximally entangled states . to give an example
, we consider the family of states which are invariant under all unitary product transformations of the form @xmath177 , were @xmath178 denotes the complex conjugation of @xmath179 .
these states , known as isotropic states @xcite , can be parameterized by a single parameter , namely by their fidelity @xmath180 $ ] .
the isotropic states for @xmath181 are separable and for @xmath182 their concurrence is given by @xcite @xmath183 the right - hand side of this equation coincides with the bound given by the ppt criterion , i. e. , the application of the latter already yields the exact expression for the concurrence @xcite . on the other hand ,
the bound given by eq .
( [ newbound - concu ] ) yields @xmath184 since the right - hand side is strictly larger than zero for @xmath185 we can conclude that , like the ppt criterion , also the new criterion ( [ trc ] ) provides a necessary and sufficient condition for the separability of the isotropic states .
but if we compare eqs .
( [ c - f-1 ] ) and ( [ c - f-2 ] ) we see that the bound obtained by the witness @xmath60 is always weaker than the bound of the ppt criterion .
note however that for large @xmath20 the difference between these bounds is only of order @xmath186 .
we finally indicate some generalizations of the present approach .
an obvious extension concerns the definition of the entanglement witness @xmath60 given by eq .
( [ def - w ] ) . according to this definition
the witness @xmath60 depends of course on the chosen basis and is not invariant under local unitary transformations .
however , for any product transformation @xmath187 with unitary operators @xmath188 and @xmath189 the observable @xmath190 is again an entanglement witness .
it is clear from the proof given in appendix [ app - a ] that for arbitrary @xmath188 and @xmath189 the witness @xmath191 also satisfies the inequality ( [ ineq - fw ] ) .
it follows that we may perform the replacement @xmath192 in the bounds for the concurrence [ eq .
( [ newbound - concu ] ) ] and for the entanglement of formation [ eq . ( [ bounds - eof-2 ] ) ] , where the minimum is taken over all product unitaries @xmath193 .
this replacement sharpens the lower bounds and ensures that they are invariant under local unitary operations . for simplicity
we have restricted ourselves to the case of bipartite systems with state space @xmath194 .
the definition @xmath195 for the witness @xmath60 can be extended to state spaces @xmath196 with arbitrary @xmath197 in an obvious way , replacing the singlet state @xmath58 by any maximally entangled pure state in @xmath198 .
moreover , an extension of the present approach to state spaces with odd dimension @xmath20 seems to be possible . to this end
one has to drop the condition that the operator @xmath18 introduced in sec .
[ time - reversal ] is unitary , i. e. , one only requires that @xmath18 is skew - symmetric and that @xmath199 .
it is worth investigating applications of these constructions to bipartite and multipartite quantum systems .
to prove inequality ( [ ineq - aij ] ) we consider a fixed pair @xmath200 of indices and decompose @xmath201 into a component which is parallel to @xmath202 and a component which is perpendicular to @xmath202 : @xmath203 where @xmath204 and @xmath205 . applying the time reversal transformation to this equation
we get : @xmath206 inserting ( [ app - a-1 ] ) and ( [ app - a-2 ] ) into the expression ( [ def - aij ] ) one obtains : @xmath207.\ ] ] since @xmath208 this leads to @xmath209 where we have introduced the quantities : @xmath210 since @xmath202 is perpendicular to @xmath211 by construction , we have @xmath212 and , therefore , @xmath213 . in a similar manner
we get @xmath214 .
hence , we obtain : @xmath215 it is easy to see that the right - hand side of this inequality is smaller than or equal to 1 for all @xmath216 $ ] , which yields the desired inequality ( [ ineq - aij ] ) .
since @xmath217 and @xmath218 are unitarily equivalent we have @xmath219 . using eqs .
( [ family ] ) and ( [ theta - p0 ] ) , and the representation @xmath220 we get @xmath221 p_j.\ ] ] hence , the trace norm of @xmath222 is found to be : @xmath223 carrying out this sum one gets : @xmath224 , & \frac{1}{n+2 } \leq
\lambda \leq \frac{1}{2 } \\ n\lambda-1 , & \frac{1}{2 } \leq \lambda \end{array } \right.\ ] ] which yields the lower bounds of eq . ( [ ppt - bound ] ) . to determine @xmath134 we note that the realignment transformation may be written as @xmath225 . using @xmath226 and
@xmath227 one easily deduces that @xmath228 p_j,\ ] ] which yields : @xmath229 the evaluation of this sum leads to : @xmath230 which gives the lower bounds of eq .
( [ realign - bound ] ) .
99 g. alber , t. beth , m. horodecki , p. horodecki , r. horodecki , m. rtteler , h. weinfurter , r. werner , and a. zeilinger , _ quantum information _ ( springer - verlag , berlin , 2001 ) .
k. eckert , o. ghne , f. hulpke , p. hyllus , j. korbicz , j. mompart , d. bru , m. lewenstein , and a. sanpera , in _ quantum information processing _ , edited by g. leuchs and t. beth ( wiley - vch , berlin , 2005 ) . w. k. wootters , phys .
lett . * 80 * , 2245 ( 1998 ) .
w. k. wootters , quant .
inf . comp . * 1 * , 27 ( 2001 ) .
p. rungta , v. buek , c. m. caves , m. hillery , and g. j. milburn , phys .
a * 64 * , 042315 ( 2001 ) . c. h. bennett , d. p. divincenzo , j. a. smolin , and w. k. wootters , phys
a * 54 * , 3824 ( 1996 ) .
g. vidal , j. mod . opt . * 47 * , 355 ( 2000 ) .
a. osterloh , l. amico , g. falci , and r. fazio , nature * 416 * , 608 ( 2002 ) .
t. j. osborne and m. a. nielsen , phys .
a * 66 * , 032110 ( 2002 ) .
wu , m. s. sarandy , and d. a. lidar , phys .
lett . * 93 * , 250404 ( 2004 ) .
l. campos venuti , c. degli esposti boschi , and m. roncaglia , .
p. w. shor , commun .
phys . * 246 * , 453 ( 2004 ) .
k. chen , s. albeverio , and s .- m .
fei , phys .
lett . * 95 * , 040504 ( 2005 ) .
k. chen , s. albeverio , and s .- m .
fei , phys .
lett . * 95 * , 210501 ( 2005 ) .
a. peres , phys .
lett . * 77 * , 1413 ( 1996 ) .
m. horodecki , p. horodecki , and r. horodecki , phys .
a * 223 * , 1 ( 1996 ) .
k. chen and l .- a .
wu , quant .
comp . * 3 * , 193 ( 2003 ) .
o. rudolph , phys .
a * 67 * , 032312 ( 2003 ) .
h. p. breuer , .
m. lewenstein , b. kraus , j. i. cirac , and p. horodecki , phys .
a * 62 * , 052310 ( 2000 ) .
m. lewenstein , b. kraus , p. horodecki , and j. i. cirac , phys .
a * 63 * , 044304 ( 2001 ) .
a. galindo and p. pascual , _ quantum mechanics i _ ( springer - verlag , berlin , 1990 ) .
h. p. breuer , phys .
a * 71 * , 062330 ( 2005 ) .
h. p. breuer , j. phys .
a * 38 * , 9019 ( 2005 ) .
j. schliemann , phys .
a * 72 * , 012307 ( 2005 ) .
b. m. terhal , phys .
a * 271 * , 319 ( 2000 ) .
s. l. woronowicz , rep .
* 10 * , 165 ( 1976 ) .
r. f. werner , phys .
a * 40 * , 4277 ( 1989 ) .
b. m. terhal and k. g. h. vollbrecht , phys .
85 , 2625 ( 2000 ) . s .-
fei and x. li - jost , phys .
a * 73 * , 024302 ( 2006 ) .
p. rungta and c. m. caves , phys .
a * 67 * , 012307 ( 2003 ) .
m. horodecki and p. horodecki , phys .
a * 59 * , 4206 ( 1999 ) . | employing a recently proposed separability criterion we develop analytical lower bounds for the concurrence and for the entanglement of formation of bipartite quantum systems .
the separability criterion is based on a nondecomposable positive map which operates on state spaces with even dimension @xmath0 , and leads to a class of nondecomposable optimal entanglement witnesses .
it is shown that the bounds derived here complement and improve the existing bounds obtained from the criterion of positive partial transposition and from the realignment criterion . |
dynamical and chemical properties of globular clusters ( gcs ) in the galaxy have been discussed in variously different contexts , such as the galaxy formation by accretion of low - mass dwarfs ( e.g. , searle & zinn 1978 ) , self - pollution by agb stars within proto - gc clouds ( e.g. , cottrell & da costa 1981 ) , and the formation process of gcs at high redshifts ( e.g. , djorgovski 1993 ) .
the observed correlations and anticorrelations in light elements ( e.g. , c , n , o , na , mg , and al ) of stars within gcs have long been considered to provide vital clues for the early formation histories of gcs ( e.g. , gratton et al . 2004 ) .
recent observational studies , which have revealed star - to - star abundance inhomogeneity among light elements of stars _ on the main sequence _ in the galactic gcs , have suggested that the observed abundance inhomogeneity is due to the second generation of stars formed from ejecta of the first generation of stars ( e.g. , agb and ob stars ) within gcs ( e.g. , da costa et al .
2004 ) : proto - gc clouds could be chemically polluted by earlier generations of stars within forming gcs ( which is often referred to as `` self - pollution '' ) .
it however remains unclear ( 1 ) how such `` self - pollution '' was possible within proto - gc clouds ( e.g. , smith & norris 1982 ) and ( 2 ) whether massive stars or agb ones are responsible for the self - pollution processes ( e.g. , prantzos & charbonnel 2006 ; smith 2006 ) .
previous studies discusses whether chemical evolution models with _
canonical imfs _ and the self - pollution processes by agb stars could have some problems in explaining the observed large fraction of cn - strong stars ( smith & norris 1982 ; bekki 2006 ) .
decressin et al .
( 2006 ) have recently suggested that self - pollution by stellar winds of massive stars with masses of @xmath7 and fast rotation can possibly explain the observed abundance patterns of gcs , such as o - na and mg - al anticorrelations .
it is however unclear how stellar winds from massive stars influence chemical evolution of forming gcs owing to the lack of extensive chemodynamical simulations of forming gcs .
the purpose of this paper is thus to discuss , for the first time , whether stellar winds from massive stars can be vital in chemical evolution of forming gcs based on chemodynamical simulations of forming gcs within gmcs .
we assume that gcs can be formed in gmcs within low - mass galaxies embedded in dark matter halos at high redshifts ( @xmath8 ) and thereby investigate the formation processes of gcs within the galaxies .
we investigate how stellar winds from massive stars , in particular , those with the masses ( @xmath9 ) larger than @xmath10 influence chemical evolution of forming gcs within gmcs . by comparing the observed abundance ratios of [ n / fe ] and [ c / fe ] and the corresponding simulation results , we raise important questions on imfs and star formation histories of forming gcs .
since we focus on chemical abundances of gcs , we do not intend to discuss structural properties and scaling relations ( e.g. , the fundamental plane ) of gcs in the present paper .
dynamical properties of gcs with different masses will be discussed in our future papers ( bekki & chiba 2007 , bc07 , in preparation ) .
we do not intend to discuss the roles of agb stars in the chemical evolution of forming gcs in this paper ( see bekki 2006 and bekki et al .
2007 for the detailed discussions on this matter ) .
recent numerical simulations have suggested that high - density gaseous regions within low - mass dwarf galaxies embedded in dark matter halos at high redshifts ( @xmath11 ) can be the formation sites of gcs ( e.g. , bromm & clarke 2002 ) .
theoretical studies based on numerical simulations of gmcs with internal turbulent flows suggested that bound star clusters can be formed in gmcs with high star formation efficiencies ( e.g. , klessen et al . 2000 ) .
guided by these previous studies , we consider that gcs can be formed in turbulent gmcs within low - mass dwarfs at high @xmath11 . since the details of numerical methods and techniques on chemodynamical simulations using grape5-sph codes for gc formation are given in bc07 , we briefly summarize these in the following .
gmcs with the sizes ( @xmath12 ) and masses ( @xmath13 ) are assumed to have homogeneous spherical distributions for most models .
we set up the initial velocity fields due to turbulent flows within gmcs in the same way as mac low et al ( 1998 ) did .
we therefore assume that a turbulent velocity field within a gmc is a gaussian random field with power spectrum @xmath14 , where @xmath15 is set to be 2.0 for most models and @xmath16 is a parameter controlling the total kinematical energy due to the turbulent flow in the gmc .
the virial ratio ( @xmath17 ) of @xmath18 , where @xmath19 and @xmath20 are the total kinematical energy and the absolute magnitude of the total potential energy for a gmc , respectively , is determined by @xmath16 and set to be 0.5 in the present study . since an isothermal equation of state
is suggested to be appropriate for star - forming interstellar clouds of molecular gas ( e.g. , mac low et al .
1998 ; klessen et al .
2000 ) , we adopt the equation with the initial temperature of 10 k. a gmc represented by @xmath21 sph particles is assumed to be within a low - mass dark matter halo with a mass of @xmath22 and the universal `` nfw '' radial density profile ( navarro et al .
1996 ) with the scale length ( @xmath23 ) of 496 pc and the concentration parameter ( @xmath24 ) of 10 .
a gmc is assumed to have a circular orbit and the initial distance from the center of its host s dark matter halo is a free parameter described by @xmath25 . since gmcs in a halo are influenced strongly by tidal fields of the halo , the formation processes of gcs within gmcs can be different in different locations within the halo .
gas particles are assumed to be converted into new stellar particles , if ( i ) local dynamical time scales are shorter than local sound crossing time and ( ii ) local gas densities exceed a threshold gas density ( @xmath26 ) of star formation ( e.g. , nakasato et al .
recent high - resolution numerical simulations on the effects of stellar winds from massive stars on interstellar medium have demonstrated that about 30% of the total energy ( @xmath27 erg ) from a massive star can be converted into kinematical energy of gas around the star ( e.g. , freyer et al .
guided by these simulations , we consider that one massive star can give its surrounding gas kinematical energy of @xmath28 erg in the present study .
stellar particles are assumed to lose gradually their masses ( @xmath9 ) owing to mass loss through stellar winds from massive stars . since chemical abundances of stellar winds from massive stars
are time - dependent and quite different in light elements from those of gas from where the stars are born ( e.g. , schaller et al .
1992 ) , stars later formed from gas chemically polluted by stellar winds from the first generation of stars in a gmc can show abundances different from those in the first generation .
we use the numerical tables for time - dependent mass - loss rates of non - rotating massive stars and for the chemical yields of the stars shown in schaller et al .
( 1992 ) for a metallicity of @xmath29 in order to calculate masses and abundances of stellar winds from stellar particles at each time step for a given imf .
this is because these tables are only the ones that show explicitly the time evolution of abundances relevant to the present study ( h , he , c , n , and o ) for different massive stars . since chemical yield tables
( including na , mg , and al ) for rotating massive stars by decressin et al .
( 2006 ) are only for those with a given mass ( @xmath30 ) and do not show the time - dependent evolution of mass - loss rates and abundances of massive stars , we do not use these tables in the present study .
we adopt power - law imfs ( @xmath31 ) with the slopes of @xmath32 and an upper- and lower - mass cut off ( @xmath33 and @xmath34 ) .
@xmath34 is set to be @xmath35 for all models whereas @xmath33 is regarded as a free parameter controlling self - pollution processes by massive stars in the present study .
initial abundances of gas are set to be the same as those used by schaller et al .
( 1992 ) for consistency .
we focus mainly on the observed c - n anticorrelation ( e.g. , smith & norris 1993 ) that stars with higher [ n / fe ] have smaller [ c / fe ] in gcs , firstly because these observations have been extensively discussed in other theoretical papers ( e.g. , bekki et al .
2007 ) and secondly because [ n / fe ] and [ c / fe ] can be investigated by the present models based on stellar yield table by schaller et al .
( 1992 ) _ which does not provide tables for na , mg , and al_. we will discuss o - na and mg - al anticorrelations in our future papers , if the tables for _ time - dependent evolution of na , ma , and al for massive stars with variously different masses and metallicities _ are available : for this reason , we do not use the latest important work by maeder & meynet ( 2006 ) for rotating and non - rotating massive stars .
it should be stressed here that the observed c - n , o - na , and mg - al anticorrelations can be possibly reproduced by chemical evolution models using stellar yield tables for fast rotating massive stars ( decressin et al .
2006 ) .
although we have investigated models with different parameter values , we mainly show the `` standard model '' with @xmath36 , @xmath37pc , @xmath38 , @xmath39 , and @xmath40pc .
this is mainly because this model clearly shows how a star - to - star abundance inhomogeneity within a forming gc is achieved during chemodynamical evolution of the gmc with star formation : the details of the the parameter dependences of abundance patterns of simulated gcs will be given in bc07 .
the mass density of the initial gmc in the standard model is much higher than that adopted in the model with @xmath41 and @xmath42pc by geyer & burkert ( 2002 ) in which gc formation does not occur owing to low efficiencies of star formation .
the mass densities ( @xmath43 ) of the preset models are however assumed to roughly follow the observed relation of @xmath44 ( atom @xmath45 ) by larson ( 1981 ) .
we describe only five representative and important models ( e.g. , standard model ) in this paper and the parameters values and some important results in these models are given in table 1 : @xmath13 ( column 1 ) , @xmath25 ( 2 ) , @xmath32 ( 3 ) , @xmath33 ( 4 ) , @xmath46y ( 5 ) , @xmath46[c / fe ] ( 6 ) , and @xmath46[n / fe ] ( 7 ) .
@xmath46y ( @xmath46[c / fe ] and @xmath46[n / fe ] ) is the difference between the minimum and maximum y ( [ c / fe ] and [ n / fe ] ) at @xmath47 myr .
for example , the values shown in the table for the standard model mean that the maximum y , [ c / fe ] , and [ n / fe ] are 0.275 , 0.168 , and 0.324 , respectively , at @xmath47 myr
. [ o / fe ] inhomogeneity is not the main focus of this paper so that it is not so extensively discussed .
we however found that ( i ) @xmath46[o / fe ] is 0.030 for the standard model and
( ii ) it is larger for models with larger @xmath46[c / fe ] .
we do not intend to discuss extensively the roles of type ii supernovae ( sneii ) in gc formation in the present study , mainly because we confirm that sneii are not so important for dynamical evolution and star formation in gmcs .
previous simulations by nakasato et al .
( 2000 ) also showed that self - enrichment of heavy elements due to sneii does not occur so that simulated gcs show homogeneities in fe - peak elements ( which is consistent with most gcs in the galaxy ) .
we find that if the time delay between star formation and the onset of supernovae explosion is similar to a typical lifetime of massive stars ( @xmath48 yr , which corresponds to an average lifetime for stars with @xmath49 in models with @xmath38 ) , sneii can be important only for expelling the remaining gas of gc - forming gmcs into interstellar regions of low - mass halos .
we thus show the results of models including feedback effects of massive stars only for clarity : the roles of sneii in gc formation will be given in bc07 .
thus , the present study adopts a fundamentally different approach from previous one - zone chemical evolution models ( e.g. , bekki et al .
2007 ) in the sense that not only mixing of stellar ejecta ( from massive stars ) and the surrounding gas but also star formation from the mixed gas are self - consistency investigated through dynamical simulations with jeans instability for star formation , kinematical feedback effects of ob stars , and non - instantaneous recycling of stellar ejecta .
the present simulations thus enable us to discuss whether the simulated gcs can explain _ both _ dynamical properties and chemical ones , although the present models do not allow us to discuss the observed anti - correlations ( e.g. , c - n , o - na , mg - al relations ) in a self - consistent manner .
figure 1 describes how the star formation rate within a gmc evolves with time in the standard model .
first stars are formed within a gmc at @xmath50 myr when the turbulence within the gmc has decayed significantly to form local high - density gaseous regions ( @xmath51 atoms@xmath45 ) .
the star formation reaches its maximum rate ( @xmath52 ) at @xmath53 myr and suddenly declines and finally stops completely till @xmath54 myr owing to ( i ) feedback effects of massive stars ( i.e. , the momentum input ) on the remaining gas and ( ii ) stripping of the gas by the background tidal field of the dark matter halo .
almost 90% of initial gas is converted into new stars within a time scale of @xmath55 myr to form a number of star clusters ( scs ) .
figure 2 shows the final spatial distribution of scs at @xmath56 myr which corresponds to one orbital rotation after most stars are formed within the gmc .
the largest sc shows a size of 10.6 pc , an appreciably flattened shape , two companion scs , and a total stellar mass ( including field stars and the companions ) of @xmath57 within 50 pc from the sc s center .
the two small , low - mass , and bound scs survived from tidal destruction by their host halo may well be identified as low - mass open clusters .
field stars that are diffusely distributed within the halo were initially weakly bound or unbound star - forming complexes .
these results imply that not only a gc ( and open clusters ) but also field stars can be formed from a single high - density gmc with a mass of @xmath58 .
figure 3 shows that ( i ) the standard model can reproduce a c - n anticorrelation that stars with smaller [ c / fe ] have larger [ n / fe ] in the simulated gc at @xmath47 myr and ( ii ) the star - to - star abundance spread in [ c / fe ] and [ n / fe ] is about 0.03 and 0.22 , respectively .
however it can not explain the existence of very n - rich stars with [ n / fe]@xmath59 observed in ngc 6752 ( smith & norris 1993 ) .
it should be noted here that the gaseous components show high [ n / fe ] ( @xmath59 ) and low [ c / fe ] ( @xmath60 ) .
these results imply that although chemical pollution of gas by massive stars earlier formed does occur , stars can not form efficiently from the polluted gas later because the gas is dispersed into the interstellar space to have a lower density due both to feedback effects of massive stars and to tidal stripping .
figure 4 shows that the simulated number fraction of n - rich ( [ n / fe]@xmath61 ) stars is small ( @xmath62 ) .
this result is inconsistent with observations which show the number fraction of cn - strong stars , which are considered to be n - rich stars , is as large as 0.5 in some gcs ( e.g. , ngc 6752 ; smith & norris 1993 ) .
this inconsistency results from the fact that only a very small fraction of chemically polluted gas is converted into new stars after the final burst of star formation ( @xmath63 myr ) .
it should be noted here that the number fraction of n - rich components ( [ n / fe ] @xmath61 ) in the gas is relatively high ( @xmath64 ) .
these results imply that if a gc is formed from a gmc _ which evolves in isolation ( i.e. , without interacting with other gmcs in the halo ) and has a canonical imf _ , self - pollution by stellar winds of massive stars can hardly explain the large fraction ( @xmath65 ) of cn - strong ( i.e. , n - rich , c - depleted ) stars observed in some galactic gcs .
it should be however noted that the observed fraction of n - rich stars is quite diverse ( e.g. , norris 1988 ) : some gcs with very low fraction of n - rich stars could be consistent with some of the present models .
figure 4 shows that although stars show the he abundance spread , the number fraction of moderately high he - rich ( y@xmath4 ) stars is very small ( @xmath5 ) : this model is reasonable for gcs with `` normal '' y. accordingly this model can not reproduce the high fraction ( @xmath66% ) of he - rich ( @xmath67 ) stars observed for ngc 2808 ( dantona et al . 2005 ) and for ngc 6441 ( caloi & dantona 2007 ) , which implies that models with canonical imfs are not reasonable for some gcs with high fraction of he - rich stars . although y of gaseous particles range from 0.24 to 0.57 , new stars can not form from he - rich gas with y @xmath68 : the origin of stars with possibly very high y ( @xmath68 ) in ngc 2808 ( dantona et al . 2005 ) can not be explained by self - pollution processes of gc - forming gmcs in the standard model with a canonical imf .
although he - rich gaseous components can be finally formed , the local densities of these components become too low when they are formed in the standard model .
this is why the number fraction of he - rich stars is very small in this model .
figure 5 shows that abundance inhomogeneity in [ o / fe ] is much less significant in comparison with [ n / fe ] .
the simulated results are not so consistent with the observations of gcs stars , where a great reduction of the surface oxygen abundance is observed in the most polluted objects .
dependences of the results on model parameters are described as follows .
although all four models with different @xmath13 and @xmath25 and a canonical imf with @xmath38 ( i.e. , the salpeter imf ) clearly show c - n anticorrelations and star - to - star abundance spread in he , c , and n , they can not show very n - rich stars with [ n / fe]@xmath59 observed for cn - strong agb stars in ngc6752 ( smith & norris 1993 ) : @xmath46[n / fe ] ranges from 0.19 to 0.33 ( corresponding to @xmath69 } \le 0.43 $ ] ) in these models ( see the table 1 ) .
the `` nucleus '' model shows a slightly larger degree of abundance inhomogeneity in the simulated gc , because gas ejected from massive stars can not be brown away so effectively owing to the deep gravitational potential of the host halo : the chemically polluted gas can be used for further star formation .
the number fractions of stars with y @xmath4 ( or with [ n / fe ] @xmath61 ) are at most @xmath5 in these models with @xmath38 .
the `` top - heavy imf '' model shows the presence of stars with very high y ( @xmath70 ) , because stellar winds of stars with @xmath71 and thus with y=0.72 in their later wind phases can more effectively pollute the gc - forming gmc .
figure 6 however shows that the fraction of y - rich stars with y @xmath72 is very small ( @xmath5 ) .
figure 6 also shows that the number fraction of n - rich stars with [ n / fe ] @xmath61 is 0.06 , which is more consistent with the observed fraction of cn - strong stars in ngc 6752 ( smith & norris 1993 ) than the standard model .
we confirm that the fraction of n - rich stars can be as high as 0.4 , if we adopt the top - heavy imf models with initial isothermal radial density profiles ( @xmath73 ) of gmcs for which stellar winds of massive stars can more effectively pollute gc - forming gmcs in the very early stage of star formation .
thus the present models with smaller @xmath32 can better explain stellar abundances of gcs with larger degrees of abundance inhomogeneities in y , [ c / fe ] , and [ n / fe ] .
although the above five models with @xmath39 clearly show abundance spread among stars within the simulated gcs , the `` bottom - heavy imf model '' with @xmath74 shows a very small degree of the abundance inhomogeneity ; @xmath46y , @xmath46[c / fe ] , and @xmath46[n / fe ] are all less than @xmath75 .
we confirm that ( i ) this result does not depend on @xmath13 and ( ii ) @xmath76 is required for the simulated gcs to show star - to - star abundance inhomogeneities .
these results imply that total numbers of massive stars with @xmath77 , which are determined by @xmath33 for a given @xmath32 ( and @xmath34 ) , are important determinants controlling whether gcs can show observable star - to - star abundance inhomogeneities .
recent observations have reported that star - to - star abundance inhomogeneity can be seen in less evolved stars on the main sequence and subgiant - branch ( e.g. , cannon et al .
1998 for 47 tuc ) , for which deep mixing is highly unlikely to occur : the self - pollution scenario might well be more promising than the mixing one .
the present numerical simulations have first demonstrated that star - to - star abundance inhomogeneity is possible within _ gc - forming gmcs _ being polluted by stellar winds of massive stars ( from now on this scenario is referred to as the stellar wind scenario just for convenience ) .
however the simulated fractions ( @xmath78 ) of n - rich ( [ n / fe ] @xmath61 ) stars for the adopted canonical imf are too small to be consistent with the observed one for some gcs ( e.g. , ngc 6752 ; smith & norris 1993 ) .
this inconsistency implies either that stellar populations other than massive stars ( e.g. agb stars ) could be responsible for the observed abundance patterns of gcs or that the present models did not include important ingredients of gc formation processes ( and thus failed to explain the observations ) . since abundance inhomogeneity due to agb stars
were already discussed in previous papers ( cottrell & da costa 1981 ; bekki et al .
2007 ) , we focus on the later implication below . if imfs of forming gcs are significantly top - heavy ( i.e. , much smaller @xmath32 ) , the observed large fraction of n - rich stars can be reproduced ( smith & norris 1982 ; see also prantzos & charbonnel 2006 in the context of the observed o - na anticorrelation ) .
although our top - heavy imf model with @xmath79 can show a large fraction of n - rich stars , the stellar wind scenario with a top - heavy imf has serious problems as described below .
if we adopt the salpeter imf with @xmath34 = @xmath35 and @xmath33 = @xmath80 , the mass fraction of massive stars with masses ranging from @xmath10 to @xmath80 is about 0.84 .
since the vast majority ( @xmath81% ) of gas of these massive stars can be expelled from gc - forming gmcs via sne ii ( owing to very shallow gravitational potentials of gmcs ) , the total masses of gcs can dramatically decrease after their formation .
previous theoretical works demonstrated that if self - gravitating star clusters like gcs lose more than 50% of their orignal masses , they become soon disintegrated ( e.g. , hills 1980 ) .
furthermore dynamical models of gcs with top - heavy imfs demonstrated that gcs losing a significant fraction of their masses can be disintegrated during their dynamical evolution around the galaxy ( e.g. , chernoff & weinberg 1990 ) .
thus we think that top - heavy imfs in forming gcs can not be plausible .
we can provide the following two possible solutions for the above inconsistency in the stellar wind scenario .
firstly , most massive stars in gmcs are first formed before the major epoch of low - mass star formation so that their stellar winds can chemically pollute a large fraction of remaining gas from which low - mass stars are later formed . this time delay between the major epoch of massive star formation and that of low - mass one was not modeled properly in the present simulations for which stellar particles ( `` stars '' ) have almost identical masses . in this second solution , a large fraction of low - mass stars can be formed from n - rich gas well polluted by massive stars earlier formed : a larger degree of abundance inhomogeneity can be seen in gcs .
it is however currently unclear in what physical conditions of gmcs massive stars can be formed earlier than low - mass ones : more quantitative discussions on this solution are given in 4.2 just for convenience .
secondly , gmcs forming gcs were chemically polluted not only by stars within the gmcs but also by those outside the gmcs ( e.g. , field massive stars ) : `` external pollution '' by field massive stars is required to explain the large fraction of n - rich stars .
owing to the lack of extensive simulations on this external pollution processes , it is unclear whether and how this process can occur in gmcs within low - mass galaxies at high redshifts .
thus , we can not determine whether the first or the second solution is more reasonable and realistic until we perform new sets of simulations on time delay between low- and high - mass star formation and on the above external pollution processes .
recent photometric observations of stars in @xmath6 cen have discovered a double main sequence ( dms ) in the color magnitude diagrams ( cmds ) of its stellar populations ( e.g. , bedin et al .
one of the most promising interpretation is that stars on the bluer main sequence ( bms ) of the dms represents a very helium - rich ( @xmath82 ) population ( e.g. , norris 2004 ) .
although the present chemodynamical simulations have first demonstrated that the formation of stars with moderately high y ( @xmath4 ) is possible in gmcs chemically polluted by stellar winds from massive stars , the derived number fraction of he - rich stars ( @xmath83 ) is too small ( @xmath5 ) to be consistent with the observed one ( e.g. , 0.3 for the bms in @xmath6 cen ) .
bekki & norris ( 2006 ) suggested that @xmath6 cen was a nucleus of an ancient dwarf galaxy where he - rich gas could be efficiently transferred into its nucleus and consequently used for star formation to form the bms .
we have suggested above that if the vast majority of massive stars ( `` polluters '' of star - forming gmcs ) are formed well before the formation of low - mass ones , the fraction of n - rich stars can be significantly increased to be consistent with the observed one . in order to discuss this point in a quantitative way
, we here assume that ( i ) a gc - forming gmc has two major epochs of star formation , ( ii ) the total gas mass consumed for the formation of the first ( second ) generation of stars is @xmath84 ( @xmath85 ) , ( iii ) the slope of the power - law imf of the 1st ( 2nd ) generation is @xmath86 ( @xmath87 ) , ( iv ) the initial mass fraction of n in the gmc is @xmath88 ( from the table by schaller et al .
1992 for @xmath71 , @xmath29 , and @xmath89 yr ) , and ( v ) the mass fraction of n in the winds of stars in the 1st generation is @xmath90 ( from the table by schaller et al .
1992 for @xmath71 , @xmath29 , and @xmath91 yr ) .
for a given @xmath92 , @xmath86 , and @xmath87 ( fixed at 2.35 ) , we search for parameter values of @xmath93 for which @xmath46[n / fe ] between the 1st and the 2nd generations is as large as 0.8 observed for ngc 6752 ( smith & norris 1993 ) . in these models , the vast majority of massive stars ( `` polluters '' ) are formed in the first epoch of star formation owing to the adopted top - heavy imf ( @xmath94 ) whereas most low - mass stars are formed in the first and the second epochs .
figure 7 shows ( i ) the @xmath86 dependence of @xmath95 required for @xmath46[n / fe]@xmath59 in ten models with different @xmath86 and ( ii ) that of @xmath96 , where @xmath97 and @xmath98 are the number of low - mass stars with @xmath99 in the first generation of stars and that in the 2nd , respectively .
figure 7 shows that ( i ) @xmath92 needs to be significantly larger than @xmath93 and ( ii ) the required @xmath95 is larger for smaller @xmath86 ( i.e. , more top - heavy imf ) .
these results imply that if stellar winds of massive stars are responsible for the star - to - star abundance inhomogeneity in a gc , most gas ( 96 - 99% ) of the gc - forming gmc needs to be consumed for the 1st generation massive stars .
figure 7 also shows that @xmath96 can be as large as 1.0 ( i.e. , as observed in ngc 6752 ) for the models with very top - heavy imfs of @xmath100 .
this result implies that top - heavy imfs in the first epochs of massive star formation are essential for explaining the observed large ( @xmath101 ) fraction of cn - strong stars in some gcs .
one of possibly serious problems in this `` bimodal star formation '' scenario is that the 2nd generation of stars in a gc - forming gmc needs to be formed well before the supernovae explosion of the 1st : otherwise the gc shows abundance spreads in fe - peak elements , which are not observed for typical gcs .
this problem would not be so serious _ if stellar winds from the 1st can trigger star formation of the 2nd_. it is however theoretically unclear in what physical conditions star formation of the 2nd can be triggered by stellar winds of the 1st owing to the lack of numerical simulations on these processes .
another problem is that a significant amount of mass loss ( more than 90% ) due to ( i ) top - heavy imfs and ( ii ) @xmath92 much larger than @xmath93 can lead rapid disintegration of forming gcs .
we thus suggest that the proposed bimodal star formation scenario would not be so convincing without solving the above disintegration problem due to the assumed top - heavy imfs . then is the above imf problem serious also for models in which chemical yield tables for rotating massive stars are used ? in order to answer this question , we investigate dependences of @xmath96 and @xmath95 on @xmath86 by using ( i ) final mass fraction of stellar winds for rotating massive stars with different masses and ( ii ) n abundances for these stars shown in decressin et al .
figure 8 shows that for slightly flatter imfs ( i.e. , @xmath102 ) , @xmath96 can be as large as @xmath103 ( e.g. , @xmath104 for @xmath105 ) and the required @xmath95 is significantly larger ( @xmath68 ) , which suggests that the above disintegration problem is much less serious in the bimodal star formation scenario with stellar yields from rotating massive stars .
we thus suggest that future chemodynamical models based on the bimodal star formation scenario are promising in terms of explaining the observed spread in n abundances of gcs . we have shown that the models with @xmath74 show a very small degree of the abundance inhomogeneity ; @xmath46y , @xmath46[c / fe ] , and @xmath46[n / fe ] are all less than @xmath75 .
these results imply that ( i ) it depends on @xmath33 of gc - forming gmcs whether the gcs show abundance inhomogeneities and ( ii ) the observed differences in the degree of abundance inhomogeneities in the galactic gcs can be due also to the differences in the number fraction of very massive stars formed in gc - forming gmcs . the fraction of n - rich stars in a gc can be observationally quantified by an `` @xmath106 '' parameter by counting the number fraction of cn - strong stars vs cn - weak stars ( norris 1988 ) , and the values of these `` @xmath106 '' parameters
are observed to be different between the galactic gcs ( norris 1988 ) .
we suggest that the observed difference of @xmath106 is due to the difference in @xmath33 and the number fraction of very massive stars between gc - forming gmcs .
although the galactic gcs show abundance inhomogeneities , the old galactic open clusters ( ocs ) do not show significant abundance inhomogeneities ( e.g. , de silva et al .
one of the possible explanations for ocs without abundance inhomogeneities is that massive stars with @xmath77 could not be formed in gmcs of ocs for some physical reasons . for the salpeter imf ,
the total number of massive stars with @xmath77 is just 0.3 ( i.e. , less than 1 ) for low - mass ocs with masses ( @xmath107 ) of @xmath108 .
this means that the probability of low - mass ocs having very massive stars with @xmath77 is very low ( owing to small - number statistics ) .
we thus suggest that the origin of the observed lack of abundance inhomogeneities in the galactic old ocs can be understood in term of incapability of their host gmcs to form massive stars with @xmath77 .
the vast majority ( 90% ) of stars are observed to form in scs embedded within gmcs ( lada & lada 2003 ) : strongly bound scs can finally evolve into gcs or ocs whereas weakly bound , low - mass ones can be disintegrated into field stars in galaxies .
previous simulations ( bekki & chiba 2000 , 2001 ) showed that the galactic stellar halo can be formed from by both dissipative and dissipationless merging of subgalactic clumps and their resultant tidal disruption in the course of gravitational contraction of the galaxy at high redshift ( @xmath109 ) : the halo field stars were initially field stars originating from low - mass , unbound ( or weakly bound ) star clusters formed from gmcs not polluted by massive stars in high-@xmath11 building blocks of the galaxy . these previous results combined with the present ones therefore imply that there can be significant differences in abundance patterns of light elements between the galactic halo field stars and stars within the gcs .
observational studies have extensively discussed similarities and differences in abundances between the galactic halo field stars and the gcs ( e.g. , frogel 1993 ; sneden 2005 ) .
for example , suntzeff ( 1993 ) showed that ( i ) the galactic halo field stars show an apparently no clear o - na anticorrelation , ( ii ) oxygen abundance is relatively constant at [ o / fe ] @xmath110 for the halo field stars with [ fe / h ] @xmath111 , and ( iii ) these two properties of the halo field stars are thus in a striking contrast with those of the gcs .
we suggest that these differences reflect the fact that only gcs originate from strongly bound , massive scs formed in gmcs polluted by stellar winds of very massive stars in low - mass subhalos : although both the halo field stars and the gcs were initially `` clusters '' within low - mass building blocks of the galaxy at high-@xmath11 and later stripped to become the halo populations , the differences in physical properties of their parent scs / gmcs result in the differences of abundances between the two halo populations .
we have first investigated the influences of stellar winds of massive stars on early chemical evolution of gc - forming gmcs based on grape sph chemodynamical simulations . in our numerical models , gcs form in turbulent , high - density giant molecular clouds ( gmcs ) , which are embedded in a massive dark matter halo at high redshifts .
we have focused on the chemical evolution of gc - forming gmcs in the present study .
we summarise our principal results of the models as follows .
\(1 ) chemical pollution of gc - forming gmcs by stellar winds from massive stars can result in star - to - star abundance inhomogeneities among light elements ( e.g. , c , n , and o ) of stars in gcs .
the present model with a canonical imf ( @xmath38 ) also shows a c - n anticorrelation that stars with smaller [ c / fe ] have larger [ n / fe ] in a gc .
these results imply that `` self - pollution '' of gc - forming gmcs by stellar winds from massive stars can cause abundance inhomogeneities of gcs .
\(2 ) the present models with different parameters and canonical imfs ( @xmath38 ) however can not show n - rich stars with @xmath0 } \sim 0.8 $ ] observed in some galactic gcs ( e.g. ngc 6752 ) .
the simulated number fraction of n - rich ( [ n / fe]@xmath61 ) stars is small ( @xmath62 ) in most models with @xmath38 .
the clear bimodality in [ n / fe ] can not be seen in the models , mainly because the stellar winds of the first generation of stars can not trigger a strong burst of star formation for the 2nd generation in the present models with a single , structureless gmc .
models with top - heavy imfs ( @xmath79 ) show a significantly larger fraction ( @xmath112 ) of n - rich stars . although models with top - heavy imfs are more consistent with observations , gcs in these models are suggested to be disintegrated soon after gc formation .
\(3 ) although almost all stars ( @xmath1% ) show normal he abundances ( y ) of @xmath2 , some stars later formed in gmcs can have y as high as @xmath3 in some models .
the number fraction of he - rich stars with y @xmath4 is however found to be small ( @xmath5 ) for most models .
these results imply that ( i ) the galactic gcs can have y abundance inhomogeneities and ( ii ) the degree of inhomogeneities can depend on imfs and @xmath33 of gc - forming gmcs .
\(4 ) the present models with canonical imfs can hardly explain the observed large fraction of cn - strong stars in ngc 6752 and the large fraction of he - rich stars in ngc 2808 .
we therefore discussed these apparent failures in the context of massive star formation preceding low - mass one within gc - forming gmcs ( i.e. , the bimodal star formation scenario ) and the new external pollution scenario .
the formation of the 2nd generations of stars needs to be triggered by the stellar winds of the 1st ones to explain the observed very small spread in [ fe / h ] in the bimodal star formation scenario .
the above imf problem is however demonstrated to be much less severe in the bimodal star formation scenario with stellar yield tables for rotating massive stars .
\(5 ) the origin of the observed differences in abundances patterns of light elements ( e.g. , c , n , and o ) between the galactic halo field stars and the gcs can reflect the fact that the field stars originate from low - mass , unbound star clusters formed in gmcs not polluted by stellar winds of very massive stars with @xmath113 . although both the halo field stars and the gcs were initially in the low - mass subhalos and later stripped to become the halo populations , only gcs originate from strongly bound , more massive star clusters formed within gmcs polluted efficiently by stellar winds of very massive stars .
thus , the present chemodynamical simulations , which first have investigated self - consistently star formation histories and time evolution of he , c , n , and o within gc - forming gmcs , have shown that the observed abundance inhomogeneities of gcs stars can not be explained on the basis of the stellar winds by massive stars , if a realistic imf and the yields by schaller et al ( 1992 ) are used . in the present paper , we have not considered the self - pollution of gc - forming gas clouds by ejecta of agb stars , for which top - heavy imfs are required to explain the observed large fraction of n - rich stars in some gcs ( e.g. , smith & norris ; bekki et al .
if a top - heavy imf is the only way to explain the observed degrees of abundance inhomogeneities of gcs both in the stellar wind scenario and in the agb one and if gcs with top - heavy imfs inevitably disintegrate soon after their formation , the external pollution scenario discussed briefly in this paper needs to be investigated by chemodynamical simulations in a more quantitative way both for the two scenarios .
we have also suggested that different imfs in two major epochs of star formation for gc - forming gmcs need to be explored in future chemodynamical models , in particular , with chemical yield tables for rotating massive stars .
we are grateful to the anonymous referee for valuable comments , which contribute to improve the present paper . k.b .
acknowledges the financial support of the australian research council throughout the course of this work .
the numerical simulations reported here were carried out on grape systems kindly made available by the astronomical data analysis center ( adac ) at national astronomical observatory of japan ( naoj ) .
ccccccccc standard & 1.0 & 99 & 2.35 & 120 & 0.032 & 0.028 & 0.223 + lower mass & 0.1 & 99 & 2.35 & 120 & 0.026 & 0.023 & 0.189 + lowest mass & 0.05 & 99 & 2.35 & 120 & 0.032 & 0.028 & 0.222 + nucleus & 1.0 & 0 & 2.35 & 120 & 0.054 & 0.049 & 0.330 + top - heavy imf & 1.0 & 99 & 1.35 & 120 & 0.270 & 0.332 & 0.827 + bottom - heavy imf & 1.0 & 99 & 2.35 & 40 & @xmath114 & @xmath114 & @xmath114 + | we first present chemodynamical simulations to investigate how stellar winds of massive stars influence early dynamical and chemical evolution of forming globular clusters ( gcs ) . in our numerical models , gcs form in turbulent , high - density giant molecular clouds ( gmcs ) , which are embedded in a massive dark matter halo at high redshifts .
we show how high - density , compact stellar systems are formed from gmcs influenced both by physical processes associated with star formation and by tidal fields of their host halos .
we also show that chemical pollution of gc - forming gmcs by stellar winds from massive stars can result in star - to - star abundance inhomogeneities among light elements ( e.g. , c , n , and o ) of stars in gcs .
the present model with a canonical initial mass function ( imf ) also shows a c - n anticorrelation that stars with smaller [ c / fe ] have larger [ n / fe ] in a gc .
although these results imply that `` self - pollution '' of gc - forming gmcs by stellar winds from massive stars can cause abundance inhomogeneities of gcs , the present models with different parameters and canonical imfs can not show n - rich stars with @xmath0 } \sim 0.8 $ ] observed in some gcs ( e.g. , ngc 6752 ) .
we discuss this apparent failure in the context of massive star formation preceding low - mass one within gc - forming gmcs ( `` bimodal star formation scenario '' ) .
we also show that although almost all stars ( @xmath1% ) show normal he abundances ( y ) of @xmath2 , some stars later formed in gmcs can have y as high as @xmath3 in some models . the number fraction of he - rich stars with y @xmath4 is however found to be small ( @xmath5 ) for most models .
we discuss this result in the context of the possibly large y values observed in a few galactic gcs ( e.g. , @xmath6 cen and ngc 2808 ) .
we also briefly discuss significantly larger degrees of abundance inhomogeneities of the models with stellar yield tables for rotating massive stars in the bimodal star formation scenario . |
the study of highly excited compound nuclei , characteristic of weakening quantum effects such as superfluity and shell effects @xcite , is of great interests for heavy ion reactions . in particular ,
the survival probabilities of compound superheavy nuclei , as determined by the competition between thermal neutron emission and fission rates , are related to the choice of cold or hot fusion for the synthesis of new superheavy elements @xcite .
the description of thermal nuclear properties are conventionally based on the statistical model @xcite , which is very successful , however , invokes a phenomenological level density parameter based on parameterizations of the fermi gas model . while the microscopic calculations of level densities are rare @xcite .
the level densities can also be determined by the experimental neutron evaporation spectrum , however , up to very limited excitation energies @xcite .
various models of level densities indeed can cause uncertainties in descriptions of thermal properties and stabilities of hot nuclei . on the other hand ,
the compound nuclei can be described by the finite - temperature hartree - fock - bogoliubov ( ft - hfb ) approach in a microscopic view without any free parameters @xcite .
thus it is very desirable to explore the connections between two pictures : the statistical model and the ft - hfb approach . in the ft - hfb approach
, the compound nuclei can be self - consistently described by quasiparticle excitations due to a finite temperature @xcite . in our previous work , the evolution of fission barriers and neutron gases in terms of excitation energies
have been studied based on the ft - hfb approach @xcite , which can be meaningful for the experimental synthesis of superheavy nuclei @xcite . as a further step , in this work we like to study the neutron emission rates in compound nuclei . in hot nuclei , the thermal equilibrium of neutron evaporation can be obtained by the pressure produced by the external neutron gas ( or vapor ) @xcite .
then the neutron emission lifetime is inversely proportional to the density of the neutron gas @xcite .
presently the ft - hfb approach is adopted to self - consistently take into account the interplay between single - particle motions and pairing in the thermal equilibrium . for ft - hfb descriptions of hot nuclei ,
a crucial problem is to treat the continuum contributions to external neutron gases @xcite .
this can be realized by taking the advantages of the coordinate - space hfb approach , by which the nearly uniform neutron gas in large distances can be obtained @xcite .
while the conventional hfb approach based on harmonic oscillator basis expansion can not treat such kind of surface asymptotics . in this work
, we adopted the hfb - ax solver with finite temperatures to study the properties of hot nuclei in deformed coordinate spaces @xcite .
the hfb - ax solver is based on b - spline techniques for axially symmetric deformed nuclei @xcite .
the ft - hfb equation is solved with a mesh size of 0.6 fm and the order of b - splines is taken as 12 . to study systems that need large coordinate spaces ,
the hybrid parallel scheme is adopted .
the aim of this paper is to study the neutron emissivity in hot heavy and superheavy nuclei . in section
[ theory ] , the relevant ft - hfb formulas and the neutron emission models are given . in section
[ results ] , the thermal properties of hot nuclei and calculated neutron emission rates have been discussed .
the ft - hfb equation in the coordinate space can be written as @xcite : @xmath5\left [ \begin{array}{clrr}u_i({\boldsymbol{r } } ) \\
\vspace{2pt } v_i({\boldsymbol{r}})\\ \end{array } \right]=e_i\left [ \begin{array}{clrr } u_i({\boldsymbol{r } } ) \vspace{2pt}\\ v_i({\boldsymbol{r } } ) \end{array } \right],\label{hfb}\ ] ] where @xmath6 and @xmath7 are the temperature dependent single - particle hamiltonian and pairing potential , respectively . for the particle - hole interaction channel , the mostly used skyrme forces sly4 @xcite is adopted . for the particle - particle channel
, the density dependent surface pairing interaction is used @xcite .
the pairing strengths are taken as @xmath8=512 mev @xmath9 that can reasonably reproduce the neutron pairing gap of @xmath10sn with a pairing cutoff window of 60 mev @xcite .
compared to the hfb equation at zero temperature , in the ft - hfb equation the density @xmath11 and pairing density @xmath12 have to be modified as @xcite @xmath13 where @xmath14 denotes the thermal fermi distribution ( @xmath15 t is the temperature ) .
the temperature dependence of other densities can also be derived straightforwardly .
the entropy @xmath16 is given by @xcite @xmath17.\ ] ] the total free energy @xmath18 is given by @xmath19=@xmath20 , where @xmath21 is the intrinsic energy based on the temperature dependent densities . based on ft - hfb solutions ,
the hot nuclei are in thermal equilibrium and surrounded by external gases .
the external gases , also called vapor , are contributed by unbound continuum states @xcite . to extract the physical nuclear density distribution ,
the ft - hfb equation is solved with and without nuclear interactions .
the free neutron gas contribution corresponds to the ft - hfb solutions without nuclear interactions .
this method has been used in calculations of shell corrections by subtracting the unphysical continuum levels based on the green s function approach @xcite . in ft - hfb iteration calculations ,
the total density distributions @xmath22 and the gas density distributions @xmath23 are obtained respectively .
then the fermi energies @xmath24 and @xmath25 are self - consistently determined by satisfying the particle number equation , @xmath26,\vspace{4pt } \\
n = \displaystyle \int d^3r [ \rho_n({\boldsymbol{r}})-\rho_n^{gas}({\boldsymbol{r } } ) ] .
\end{array}\ ] ] in the case of large coordinate spaces and high temperatures , the particle numbers contributed by the external gases can be nonegligible and the self - consistent subtraction procedure will become important . based on the coordinate - space ft - hfb solutions , the uniform neutron gas density distributions can be obtained . with the uniform neutron gas density @xmath27 ,
the neutron emission widths @xmath28 can be given by the nucleosythesis formula @xcite : @xmath29 where @xmath30 is the neutron capture cross section ; @xmath31 is the average velocity of particles in the external gas . for simplicity ,
the neutron cross section @xmath30 is taken as @xmath32 , i.e. , the geometrical cross section , where @xmath33 is the nuclear radius that can be obtained by ft - hfb calculations .
the neutron emission lifetime @xmath34 is related to the width by @xmath35 . to calculate the statistical average velocity @xmath31 ,
the fermi occupation number @xmath36 in terms of neutron energies @xmath37 ( i.e. , kinetic energies ) in the gas is assumed , @xmath38 differ from the procedure in ref .
@xcite , we explicitly considered the density of the neutron gas , which is actually equivalent to the bonche s formula @xcite .
although the ft - hfb approach has been extensively applied to hot nuclei , neutron star crusts and atomic condensates @xcite , practical calculations of neutron emission rates have been rarely undertaken so far . compared to calculations in ref .
@xcite that was based on spherical ft - hartree - fock , we extend the applicability of ft - hfb for neutron emission rates in deformed cases as well as superheavy compound nuclei .
statistical models have been widely used to study the thermal neutron emission rates ( or widths ) .
the neutron evaporation width @xmath28 with an excited energy of @xmath39 is given by @xcite : @xmath40 where @xmath41 is the level density in terms of excitation energies , and @xmath42 is the spin factor of neutron ; the excitation energy @xmath43 is the difference in the intrinsic energies ( @xmath44 ) from ft - hfb ; @xmath45 is the neutron separation energy .
the level density @xmath46 is based on the fermi gas model , defined as @xcite @xmath47 where @xmath48 is the level density parameter taken as @xmath49 mev@xmath50 as suggested in @xcite . to compare ft - hfb and the statistical model , the variables @xmath45 , @xmath39 , @xmath33 in eq.([stat ] )
are taken the same values as in the ft - hfb approach .
we have performed calculations for nuclei @xmath51 and @xmath52 at temperatures up to @xmath15t=2 mev .
the statistical bulk properties such as , nuclear density distributions , excitation energies , entropy and neutron fermi energies at different temperatures are discussed in detail .
these statistical properties are directly related to the formula of neutron emission rates . to see the box size dependence , large coordinate - spaces of 24 fm and 30 fm are adopted .
then the neutron emission rates are calculated microscopically with comparison with the statistical model . to study the superheavy compound nuclei ,
the neutron emission rates in selected @xmath3cn ( cold fusion ) and @xmath4fl ( hot fusion ) have also been studied . at different temperatures .
the density profiles @xmath53 and @xmath54 are shown along the cylindrical @xmath55-axis and the perpendicular @xmath56-axis , respectively .
, title="fig:",scaledwidth=50.0% ] + and @xmath52 at different temperatures .
the neutron densities are plotted in logarithmic scale .
the external neutron gases are also displayed .
, title="fig : " ] + at @xmath57=2.0 mev , obtained by solving ft - hfb equations with box sizes of 24fm and 30fm .
, title="fig:",scaledwidth=50.0% ] + firstly we studied the density distributions of the deformed nucleus @xmath0u , as displayed in fig.[1 ] . in fig.[1 ]
, we can see that as temperature increases , ( 1 ) the central density fluctuations become weaker , ( 2 ) the density distributions in the @xmath55-axis direction shrink and the associated shape deformations decrease .
these two observations both agree with the fact that shell effects are diminished as temperatures increase @xcite .
note that the thomas - fermi approximation can not take into account such density oscillations @xcite . in the case of cylindrical coordinates ,
the differences between the density profiles @xmath58 and @xmath59 actually reflect the surface deformations of axially - symmetric nuclei .
however , the deformation transition to a spherical shape can happen at higher temperatures , depending on different effective nuclear interactions @xcite . to study the temperature dependent neutron densities including external gases , the neutron density distributions of @xmath0u and @xmath1u are plotted in the logarithmic scale , as shown in fig.[2 ] . as expected , the neutron gas gradually increases with increasing temperatures . at nuclear surfaces ,
the neutron gas has a uniform density distribution ( values are given in table [ table1 ] ) .
thus the contributions to particle numbers from gases can be easily estimated . with the same temperature
, @xmath1u always has larger gas densities than @xmath0u , due to its larger fermi energies , as shown in fig.[fig5 ] .
in addition , in both @xmath0u and @xmath1u , one can see that the gas densities increase slower than exponential functions of temperatures .
this behavior has also been shown in compound superheavy nuclei @xcite .
the proton gases are not shown and their contributions are much smaller due to the coulomb potential .
and excited energies @xmath43 in @xmath51 and @xmath52 as a function of temperatures .
the solid and dashed lines represent results obtained by using and not using the self - consistent gas subtraction procedure , respectively.,title="fig:",scaledwidth=50.0% ] + of @xmath51 and @xmath52 as a function of temperatures .
the solid and dashed lines represent results obtained by using and not using the self - consistent gas subtraction procedure , respectively.,title="fig:",scaledwidth=48.0% ] + in fig.[boxsize ] , the neutron density distributions of @xmath51 at @xmath15t@xmath60 are calculated with the box sizes of 24 fm and 30 fm , respectively .
it is important to get convergent neutron gas densities as well as thermal properties by adopting large box sizes @xcite . with larger box sizes , the neutron gas contributions become much larger ( @xmath61 ) .
compared to our earlier work that adopted a box size of 20 fm @xcite , the box size adopted in this work is larger and the gas contributions are non - negligible , particularly in @xmath1u at @[email protected] mev . in our calculations with the self - consistent subtraction procedure ( see eq.[subtraction ] ) , the neutron gas density of 30 fm is slightly less that of 24 fm , within a factor of @xmath63 .
this means our calculations are converged and not sensitive to box sizes , in which the subtraction procedure is essential . to further illustrate the importance of the self - consistent subtraction procedure ,
the entropy @xmath16 and excited energies @xmath39 as a function of temperatures are shown in fig.[se ] .
we can see that the gas subtraction in the neutron - rich @xmath1u has larger influences than in @xmath0u .
the subtraction corrections can reduce the entropy and in particular the excitation energies . generally , such corrections are significant for temperatures higher than 1.5 mev .
the fermi energies ( or chemical potentials ) of @xmath2u as a function of temperatures are shown in fig.[fig5 ] .
fig.[fig5 ] shows that the fermi energies are weakly dependent on temperatures , and slightly increase within the subtraction approach , implying increased neutron gas densities . by summarizing fig.[se ] and fig.[fig5 ]
, we demonstrated that the subtraction procedure is very important , however , which has not been widely adopted in ft - hfb calculations due to additional computing costs .
the main aim of this work is to study neutron emission rates based on the neutron gas solutions in the ft - hfb approach . in this method
the neutron emission rates can be calculated microscopically without any free parameters .
the formula have been given in sec.[theory ] .
conventionally the neutron emission rates ( or widths ) are calculated by statistical models that depend on level density parameters .
table [ table1 ] displays the calculated neutron emission widths as a function of temperatures in @xmath51 and @xmath52 , by both the ft - hfb approach and the statistical model . in the statistical model ,
the neutron emission rates are sensitive to the level density parameter @xmath48 .
firstly we adopted the constant @xmath64 that was suggested in @xcite .
we can see that the agreement between the ft - hfb and the statistical model is satisfactory . for low excitation energies
, the statistical model tends to underestimate the emission rates compared to ft - hfb results .
on the contrary , statistical models ( @xmath65 ) overestimate the emission rates compared to ft - hfb with high excitation energies .
the discrepancy is due to the utilization of constant level density parameter @xmath48 that can have temperature dependence to some extent .
in fact , based on the ft - hfb approach , we can extract the level density parameter @xmath48 by three definitions @xcite : @xmath66 , @xmath67 and @xmath68 .
here we adopt @xmath67 from ft - hfb solutions to calculate the neutron emission rates by the statistical model ( @xmath69 ) . in this case , the agreement between two approaches is also good , but slightly underestimate the widths systematically . for comparison
, @xmath70 gives the smallest level density parameter and overestimates neutron emission rates . while @xmath71 gives the largest level density parameter , which is consistent with earlier calculations @xcite .
this demonstrated the excitation energy dependent level density parameter ( @xmath67 ) can be used to connect the microscopic ft - hfb and the statistical model .
experimental studies of the level density of @xmath0u at high excitation energies can provide meaningful examinations , unfortunately , which is known for @xmath72 only up to 6 mev @xcite .
ccccc kt & & + ( mev ) & @xmath27 & @xmath73 & @xmath65 & @xmath69 + + 1.0&@xmath74 & @xmath75&@xmath76&@xmath77 + 1.5&@xmath78 & @xmath79&@xmath80&@xmath81 + 2.0&@xmath82 & @xmath83&@xmath84&@xmath85 + + 1.0&@xmath86 & @xmath87&@xmath88&@xmath89 + 1.5&@xmath90 &
@xmath91&@xmath92&@xmath93 + 2.0&@xmath94 & @xmath95&@xmath96&@xmath97 + ccccc kt & & + ( mev ) & @xmath27 & @xmath73 & @xmath65 & @xmath69 + + 1.0&@xmath98 & @xmath99&@xmath100&@xmath101 + 1.5&@xmath102 & @xmath103&@xmath104&@xmath105 + 2.0&@xmath106 & @xmath107&@xmath108&@xmath109 + + 1.0&@xmath110 & @xmath75&@xmath111&@xmath112 + 1.5&@xmath113 & @xmath114&@xmath115&@xmath116 + 2.0&@xmath117 & @xmath118&@xmath119&@xmath120 + table [ table2 ] displays the calculated neutron emission rates of superheavy compound nuclei @xmath3cn and @xmath4fl , which are typical cold fusion and hot fusion compound nuclei , respectively .
again the neutron widths given by ft - hfb and by the statistical model with the excitation energy dependent level density parameter ( @xmath67 ) agree very well . however , the agreement between ft - hfb and the statistical model with the constant level density parameter is less satisfactory , compared to table [ table1 ] .
the statistical model with the constant @xmath48 gives very small widths ( @xmath65 ) at low temperatures compared to ft - hfb results .
this indicates that the constant level density parameter may not be suitable for the superheavy mass region .
the neutron emission rates of @xmath3cn ( cold fusion ) are smaller than that of @xmath4fl ( hot fusion ) with the same temperature . or say the cooling of @xmath4fl is more favorable by neutron evaporations than @xmath3cn .
this is mainly due to its smaller neutron separation energy .
the microscopic ft - hfb calculations of neutron emission rates can be useful for the synthesis of superheavy nuclei .
the level density connection between ft - hfb and statistical models as pointed out in this work can also be useful for fission studies .
in summary , the finite - temperature hfb approach has been applied to calculations of neutron emission rates of compound nuclei in the microscopic view .
the ft - hfb equation is solved in the deformed coordinate - space so that the continuum contributions to external neutron gas can be precisely obtained .
we have demonstrated that the self - consistent gas subtraction is important to properly describe statistical properties of hot nuclei .
the ft - hfb calculated neutron emission rates of compound deformed @xmath2u and superheavy nuclei agree well with the statistical model .
furthermore , we demonstrated that by adopting inputs from ft - hfb ( an excitation energy depent level density parameter , excitation energies and neutron separation energies ) , the statistical model can be connected to the microscopic ft - hfb approach .
our approach can be useful for the synthesis of superheavy nuclei and further ft - hfb study of thermal fission rates are in progress .
useful discussions with w. nazarewicz and f.r .
xu are gratefully acknowledged .
this work was supported by the research fund for the doctoral program of higher education of china ( grant no .
20130001110001 ) , and the national natural science foundation of china under grants no.11375016 , 11235001 .
we also acknowledge that computations in this work were performed in the tianhe-1a supercomputer located in the chinese national supercomputer center in tianjin . | background : : the neutron emission rates in thermal excited nuclei are conventionally described by statistical models with a phenomenological level density parameter that depends on excitation energies , deformations and mass regions . in the microscopic view of hot nuclei , the neutron emission rates can be determined by the external neutron gas densities without any free parameters .
therefore the microscopic description of thermal neutron emissions is desirable that can impact several understandings such as survival probabilities of superheavy compound nuclei and neutron emissivity in reactors .
purpose : : to describe the neutron emission rates in deformed compound nuclei , the external thermal neutron gases are self - consistently obtained based on the finite - temperature hartree - fock - bogoliubov ( ft - hfb ) approach .
methods : : the skyrme ft - hfb equation is solved by hfb - ax in deformed coordinate spaces . based on the ft - hfb approach , the thermal properties and external neutron gases are properly described with the self - consistent gas substraction procedure .
then neutron emission rates can be obtained with the densities of external neutron gases .
the results are compared with the statistical model to explore the connections between the ft - hfb approach and the statistical model .
results : : the thermal statistical properties of @xmath0u and @xmath1u are studied in detail in terms of excitation energies .
the thermal neutron emission rates in @xmath2u and superheavy compound nuclei @xmath3cn and @xmath4fl are calculated , which agree well with the statistical model by adopting variables from ft - hfb .
conclusions : : the coordinate - space ft - hfb approach can provide reliable microscopic descriptions of neutron emission rates in hot nuclei , as well as microscopic constraints on the excitation energy dependence of level density parameters for statistical models . gbksong |
in studies of disordered systems , one often encounters the problem of finding the density of single - particle states in energy regions where the states can only exist as a consequence of disorder .
an example is the density of negative energy states of a particle moving in a random potential ( the lowest energy of a free particle is zero).@xcite another example is provided by a superconductor with magnetic impurities .
the impurities break cooper pairs and fill the gap with electron states of arbitrarily small energy.@xcite the density of such states ( `` lifshitz tails '' ) is typically small , as it requires a large disorder fluctuation to induce them . in that case
the density can be estimated by the probability of the least suppressed disorder fluctuation that creates a state of a given energy ( the so - called optimal fluctuation method).@xcite in ref .
we applied the optimal fluctuation method to study the density of states and optical absorption in disordered peierls insulators . in the absence of disorder ,
the electron excitation spectrum in these quasi - one - dimensional materials has a gap which results from the electron - lattice instability and is accompanied by a periodic lattice distortion with the wave vector @xmath0.@xcite we considered half - filled chains , in which case the peierls instability results in an alternation of long and short bonds ( dimerization ) .
disorder in the electron hopping amplitudes induces electron states inside the peierls gap , transforming it into a pseudogap . in ref .
we showed that the optimal fluctuation for a disordered dimerized chain has the form of a soliton - antisoliton pair .
it induces two electron states which lie close to the center of the pseudogap and which are symmetric and antisymmetric superpositions of states localized near the soliton and antisoliton . knowing the typical form of the disorder - induced electron states we were able to find the absorption coefficient at small energies .
these results were also used to explain the coexistence of the dimerization and antiferromagnetism , observed recently in the spin - peierls material cugeo@xmath1.@xcite the probability of the optimal configuration gives , however , only an estimate of the density of states .
a more careful calculation requires finding the contribution of disorder realizations close to the optimal one . in this paper
, we perform such a calculation using the close relation between the optimal fluctuation method and the semiclassical approximation in quantum mechanics and field theory . in the path integral version of this approximation ,
the most important paths lie close to the `` saddle - points '' of the action , called instantons.@xcite in particular , the instantons for the double - well problem are the classical trajectories describing the imaginary time motion between the two wells , which corresponds to tunneling and results in level splitting .
we show how the disorder average of the density of states can be written in the form of a functional integral that represents the sum over all paths of a particle in a corresponding double - well potential .
then the optimal disorder configuration becomes precisely the instanton trajectory and the calculation of the density of states can be done using standard instanton methods .
the outline of this paper is as follows . in sec .
[ fgm ] we introduce the fluctuating gap model describing disordered peierls systems and briefly discuss the previously obtained analytical results for the disorder - averaged density of states in that model .
next , in sec .
[ funint ] we establish the relation between the disorder average in the fluctuating gap model and the quantum - mechanical average for a certain double - well problem .
we find the instanton trajectory corresponding to the optimal fluctuation and perform the `` saddle - point '' integration for the average density of states .
we discuss our results and conclude in sec . [ discussion ] .
two technical points are addressed in appendices . in appendix
a we discuss how the symmetries of the fgm hamilitonian are crucial for the validity of the optimal fluctuation method . in appendix
b we discuss the instanton calculation for a different choice of boundary conditions .
the fluctuating gap model ( fgm ) is a continuum model describing electrons in disordered quasi - one - dimensional semiconductors.@xcite in this paper we use it to describe half - filled peierls chains with disorder in the hopping amplitudes of the electrons between the chain s lattice sites .
these systems are characterized by the peierls order parameter , @xmath2 , which is the ( continuum version of the ) alternating part of the hopping amplitudes along the chain , _
i.e. _ , the difference between the hopping amplitudes on nearest odd and even bonds . in the absence of disorder
this order parameter ( also called `` dimerization '' ) is constant along the chain and equals the gap in the single - electron spectrum.@xcite disorder in the electron hopping amplitudes results in random variations of the order parameter along the chain ( which explains the name of the model ) : @xmath3 here , @xmath4 is the average value of the order parmeter and @xmath5 is the fluctuating part with a gaussian correlator , @xmath6 the single - electron states close to the fermi energy @xmath7 , are described by the wave function @xmath8 where the two amplitudes @xmath9 and @xmath10 correspond to electrons moving , respectively , to the right and to the left with the fermi velocity @xmath11 . since the density of single - electron states does not depend on spin projection , we omit the spin index of the electron wave function .
the wave functions satisfy the one - dimensional dirac equation , @xmath12 where @xmath13 and @xmath14 the pauli matrices . for the applicability of the continuum description of electrons , the random variations in the hopping amplitudes have to be relatively small .
this model was first introduced by keldysh.@xcite it was also considered in the context of the thermodynamical properties of quasi - one - dimensional charge - transfer salts,@xcite and has been applied to study the effect of disorder on the peierls transition temperature , as well as the effect of quantum lattice fluctuations on the optical spectrum of peierls materials .
recently it was used to describe the phase diagram of disordered spin - peierls systems.@xcite in the remainder of this section we discuss the dependence of the spectrum of the single - electron eigenstates on the disorder strength . by @xmath15
we denote the disorder averaged density of states per unit length : @xmath16 \rangle\;,\ ] ] where @xmath17 is the chain length . due to the charge conjugation symmetry of the dirac hamiltonian , the density of states is a symmetric function of energy : @xmath18 we also introduce @xmath19 , equal to the average number of electronic states ( per unit length ) with energy between @xmath20 and some @xmath21 : @xmath22 an analytical expression for @xmath19 was found by ovchinnikov and erikhman@xcite by means of the `` phase formalism'',@xcite @xmath23 here , @xmath24 and @xmath25 are dimensionless variables , and @xmath26 and @xmath27 are , respectively , bessel and neumann functions . in fig.[d.o.s . ]
we plot the average density of states @xmath15 for several values of the dimensionless disorder strength @xmath28 . in the absence of disorder ( @xmath29 )
, the electron spectrum has a gap between the energies @xmath30 and @xmath31 .
disorder gives rise to the appearance of electron states inside the gap .
the energy dependence of the average density of states close to the fermi energy ( @xmath32 ) is given approximately by@xcite @xmath33 ( here and below we assume @xmath34 to be positive , which is sufficient in view of eq.([symrho ] ) ) .
for @xmath35 the density of states has a pseudogap ( the peierls gap filled with disorder - induced states ) . for @xmath36
the pseudogap disappears and the density of states becomes divergent at @xmath37 .
this is a dyson - type singularity,@xcite which occurs in the band center for random hamiltonians with charge conjugation symmetry . in ref .
we showed that the energy - dependence of the average density of states at small energy and weak disorder is mainly determined by the weight of the optimal disorder fluctuation @xmath38 , which is defined as the most probable among the fluctuations that induce a state at given energy @xmath34 .
the form of the optimal fluctuation is @xmath39 \ ; , \label{soan}\ ] ] where @xmath40 and @xmath41 describe , respectively , the position and the spatial extent of the disorder fluctuation , and @xmath42 is determined by @xmath43 thus , @xmath44 , plotted in fig .
[ s - a ] , has precisely the form of a soliton - antisoliton pair discussed in refs . in the context of polaron excitations in the conjugated polymer _ trans_-polyacetylene .
the spectrum of electron states for this @xmath2 ( also plotted in fig .
[ s - a ] ) consists of a valence band ( with highest energy @xmath45 ) , a conduction band ( with lowest energy @xmath46 ) , and two localized intragap states @xmath47 with energies @xmath48 , where @xmath49 the soliton - antisoliton separation @xmath41 is fixed by the condition @xmath50 .
the two intragap states are the symmetric and antisymmetric superpositions of the midgap states localized near the soliton and antisoliton.@xcite the energy splitting @xmath51 decreases exponentially with the soliton - antisoliton separation , so that for @xmath52 , @xmath53 where @xmath54 is the correlation length . for white - noise disorder with the correlator eq.([gauss ] ) , the weight @xmath55 $ ] of the disorder configuration @xmath5 is given by @xmath56 = \exp \left ( -\frac{1}{2a } \int\!\!dx\eta^{2}(x ) \right ) \;.\ ] ] using eq.([dist ] ) we find that the weight of the optimal configuration , @xmath57 \propto \varepsilon^{\frac{2}{g}}\ ; , \label{weight}\ ] ] gives , indeed , a good estimate for the shape of the density of states inside the pseudogap for @xmath58 [ cf .
eq.([dos ] ) ] .
we would like to conclude this section by noting that , while in the random potential problem the optimal fluctuation induces one state with large negative energy,@xcite in the fgm the soliton - antisoliton fluctuation eq.([soan ] ) induces two states with energies @xmath59 .
the small difference between the energies of the symmetric and antisymmetric states is potentially dangerous for the application of the optimal fluctuation method to the fgm .
the problem arises in the calculation of the contribution of the disorder realizations close to the optimal fluctuation : @xmath60 the perturbation @xmath61 can , in principle , strongly mix the symmetric and antisymmetric intragap states , because of the small energy denominator @xmath62 appearing in the perturbation series .
such mixing would affect the value of the energy splitting between the two states .
this problem arises because of the charge conjugation symmetry of the dirac hamiltonian eq.([dirac ] ) , which implies the symmetry of the spectrum of its eigenvalues around @xmath37 .
we address this question in appendix a , where we show that , just because of the symmetry of the hamiltonian eq.([dirac ] ) , the mixing of the symmetric and antisymmetric states is small and thus , despite the small energy splitting , the optimal fluctuation method is applicable .
the result of appendix a also justifies our calculation of the optical absorption coefficient presented in ref . .
the weight of the optimal fluctuation , given by eq.([weight ] ) , is the main factor that describes the suppression of the density of states at small energy . to obtain a full expression for the density
, however , one has to take into account the contribution of the disorder realizations close to the optimal configuration . in this section
we perform this calculation using the correspondence between the averaging over disorder realizations @xmath5 and the quantum - mechanical averaging over the ground state for a certain double - well potential
. then the main suppression factor eq.([weight ] ) , as well as the correction to it , can be easily found using well - known methods developed for the double - well problem .
the density of electron states averaged over the white - noise disorder can be written in the form of the functional integral : @xmath63 \;,\ ] ] with @xmath64 and @xmath65 as defined in eq.([dirac ] ) .
the relation between the functional integral and a particular quantum mechanical problem will become apparent after we make a substitution of variables that allows us to express the disorder fluctuation @xmath5 in terms of the eigenfunction @xmath66 of eq.([dirac ] ) with energy @xmath34 .
due to time - reversal symmetry of the dirac hamiltonian its eigenfunctions @xmath66 can be chosen to satisfy @xmath67 and , therefore , can be written in the form : @xmath68 the two real variables @xmath69 and @xmath70 introduced by @xmath71 then satisfy , @xmath72 and @xmath73 we now use eq.([sub ] ) to express the disorder configuration @xmath5 in terms of @xmath74 .
then @xmath75 , defined by eq.([s ] ) , can be written as @xmath76 = s_e[\phi(x ) ] + s_b\;,\ ] ] where @xmath77 = \frac{1}{2 } \int_{-l/2}^{l/2}\!\!dx \left ( v_f^2 \left ( \frac{d\phi}{dx } \right)^2 + \left(\delta_0 - \varepsilon \cosh(2 \phi ) \right)^2 \right)\;,\ ] ] and @xmath78 @xmath79 $ ] has the form of the euclidean action of a particle with mass @xmath80 and coordinate @xmath70 moving in the imaginary time @xmath81 in the inverted double - well potential , @xmath82 plotted in fig .
[ assfig ] .
the angle @xmath83 in eq.([pot ] ) is defined by @xmath84 finally , in the functional integral eq.([fun ] ) , the disorder strength @xmath85 plays the role of @xmath86 .
thus , we have identified the quantum double - well problem that corresponds to our disorder average . a similar correspondence exists also for the random potential problem.@xcite
the instanton @xmath87 is the solution of the classical equation of motion in the inverted double - well potential with the boundary conditions @xmath88 this corresponds to infinitely long motion from the left maximum to the right one .
the energy on this classical trajectory , @xmath89 equals zero , which gives , @xmath90 as the solution with the minus sign corresponds to the trivial disorder configuration @xmath91 [ cf .
eq.([sub ] ) ] , we choose the plus sign . equation ( [ difur ] ) is easily solved , yielding @xmath92 where @xmath42 and @xmath41 are defined by @xmath93 and @xmath94 using eq.([sub ] ) we find that the instanton eq.([solution ] ) is exactly the optimal disorder configuration given by eq.([soan ] ) and that eqs .
( [ varphi ] ) , ( [ k ] ) , and ( [ kr ] ) defining the parameters of the instanton solution coincide with eqs .
( [ vfk1 ] ) and ( [ eps ] ) .
finally , for the instanton trajectory , the value of @xmath75 defined by eq.([action ] ) is : @xmath95}{a } = \frac{2}{g } \left ( 2\varphi - \tanh(2\varphi ) \right ) \;.
\label{iaction}\ ] ] we note that the boundary conditions eq.([instbc ] ) for the instanton trajectory are not , in general , the boundary conditions that one usually imposes on the eigenfunction @xmath66 .
in particular , @xmath83 defined by eq.([varphi ] ) , depends on the energy @xmath34 , while , for instance , the boundary conditions for a finite peierls chain with an even number of atoms in the continuum limit take the form @xcite @xmath96 which is equivalent to @xmath97 we discuss the instanton for these boundary conditions in appendix b. it is clear , however , that in the limit @xmath98 the density of states should not depend on the boundary conditions .
therefore , we choose here the ones that lead to the easiest calculation of the density of states , namely , @xmath99 where @xmath100 is the instanton solution given by eq.([solution ] ) with @xmath101 . with the last condition
@xmath100 is antisymmetric , so that @xmath102 in order to perform the functional integration in eq.([fun ] ) , we first have to regularize it .
we do this by dividing the range of variation of @xmath81 , @xmath103 $ ] , into @xmath104 small intervals of size @xmath105 , assuming the value of @xmath106 to be constant inside each interval .
then , @xmath107 and the integration measure , @xmath108^{\frac{1}{2}}}\;,\ ] ] is defined in such a way that the average of the unity operator is one . in the regularized version of the substitution eq.([sub ] ) , the new variables @xmath109 ( @xmath110 ) are the values of @xmath70 on the right end of each interval and @xmath111 is fixed by the boundary condition at the chain s left end .
for the moment we treat eq.([sub ] ) as a functional substitution and forget about the boundary condition at the right end .
up to second order in @xmath105 , the regularized form of the substitution reads : @xmath112 it is necessary to retain the higher order term in powers of @xmath105 for the correct calculation of the jacobian of the substitution .
the integration measure can now be written as @xmath113^{\frac{1}{2 } } } j\;,\ ] ] with @xmath114 it is important that eq.([regsub ] ) provides a one - to - one correspondence beween the disorder configuration @xmath115 and the values of the solution of the differential equation ( [ sub ] ) at the points @xmath116 .
in other words , eq.([regsub ] ) relates the disorder configuration to the electron wave function in the presence of this disorder , which gives us a convenient representation for the trace of the delta - dunction in eq.([fun ] ) .
the energy @xmath34 is the eigenvalue of the hamiltonian @xmath65 if @xmath70 at the right end satisfies the boundary condition : @xmath117 therefore , the average density of states eq.([fun ] ) can be written in the form : @xmath118 the delta - function removes the integration over @xmath119 and , using eq.([sub ] ) , one can easily calculate the derivative of @xmath119 with respect to @xmath34 , keeping both the disorder @xmath5 and the value of @xmath70 at the left end of the chain fixed : @xmath120 here , @xmath121 is given by @xmath122 so that @xmath123 , defined by eq.([j ] ) , can be written as @xmath124 we thus arrive at the following final form of the functional integral @xmath125\;,\ ] ] where the integration measure is defined by @xmath126^{\frac{1}{2 } } } \prod_{n=1}^{n - 1 } \frac{v_f d\phi_n}{\left[2 \pi a a\right]^{\frac{1}{2 } } } \;,\ ] ] and @xmath127 = j \left ( \frac{\partial \phi_n}{\partial \varepsilon } \right)_{\phi_0,\eta(x ) } \exp \left ( -\frac{s_b}{a } \right)\;.\ ] ] in this section we perform the functional integration in eq.([fun1 ] ) over configurations @xmath74 in the vicinity of the saddle - point configuration @xmath128 defined as @xmath129 the boundary conditions eq.([bc ] ) were chosen to ensure that @xmath130 , where @xmath131 is given by eq.([solution ] ) with @xmath101 .
the result of the integration is : @xmath132 where @xmath133 is given by eq.([iaction ] ) and @xmath134 is the determinant of the operator @xmath135 obtained by the second variation of the euclidean action eq.([eaction ] ) . in eq.([sadpoint ] ) @xmath123 , @xmath134 , and @xmath136 have to be calculated at @xmath137 . using eq.([difur ] )
( with the plus sign ) one easily finds that @xmath138 so that eq.([y(x ) ] ) can be written in the form @xmath139 since @xmath100 is antisymmetric , @xmath140 , so that @xmath141 and @xmath142 as for large @xmath17 , @xmath143 can be replaced by @xmath144 and @xmath145 we obtain @xmath146 we calculate the determinant @xmath134 using the standard method for calculation of determinants of schrdinger operators in one dimension , which can be found in ref . .
the result is @xmath147 the determinant is proportional to the chain size , because the soliton - antisoliton pair can appear at any place in the chain ( formally the factor @xmath17 comes from the integration over the zero mode @xcite ) .
taking all factors together , we finally obtain for the averaged density of states @xmath148 the resulting expression for the average density of electron states in the limit @xmath149 is @xmath150 for @xmath58 , this agrees with eq.([dos ] ) , confirming the validity of the optimal fluctuation method at small energies and weak disorder .
in this paper , we have used functional integration to establish a relation between the fluctuating gap model , which describes one - dimensional electron motion in the presence of both a peierls distortion and quenched disorder , and the quantum motion of a particle in a double - well potential .
averaging over disorder realizations in the model with quenched disorder corresponds to the sum over all paths in the quantum - mechanical problem .
we have shown that the instanton trajectory describing the imaginary - time motion between the two wells corresponds to the optimal disorder fluctuation .
the probability of this fluctuation determines the asymptotic behavior of the averaged electron density of states .
we showed that the most probable form of the wave function of the electron states lying deep within the pseudogap contains two peaks .
the optimal disorder fluctuation that induces such a state , has the form of a soliton - antisoliton pair and the peaks of the wave function are localized near the two kinks of this fluctuation ( see fig . [ s - a ] ) . as we demonstrated in this paper and in ref .
, the instanton approach allows for a relatively easy calculation of the density of states and absorption coefficient .
our result eq.([ansdos ] ) is valid if the density of disorder - induced states is small , which is the case when @xmath151 and @xmath58 .
above , we emphasized the strong relation between the calculation of the disorder average and the double - well problem .
however , there is also a clear difference : in the double - well problem , an important role is played by multi - instanton configurations , in which instantons are followed by anti - instantons and _ vice versa_.@xcite although the action for the configuration containing @xmath152 instantons and anti - instantons is @xmath152 times the action of a single instanton @xmath153 , the multi - instanton contribution obtains a factor @xmath154 from the sum over all possible instanton positions ( @xmath155 is the imaginary time corresponding to @xmath17 ) .
this statistical factor grows with @xmath155 , while the suppression factor @xmath156 does not decrease any further , making the multi - instanton configurations important at large enough @xmath155 .
we note that it is the sum over all possible numbers of instantons and anti - instantons that gives the energy splitting between the two lowest states in the double - well potential.@xcite by contrast , the asymptotic expression eq.([ansdos ] ) for the density of states in the fgm is given by only one instanton , which means that the largest contribution to the averaged density of states comes from a single disorder fluctuation .
the origin of the difference lies in the fact that , while in the quantum double - well problem the anti - instanton describes a real physical process ( the tunneling between the wells in the direction opposite to the one described by the instanton ) , in the fgm it corresponds to zero disorder . to see this
, we note that the instanton @xmath157 , obtained from the instanton by `` time - reversal '' , @xmath158 , is the solution of eq.([difur ] ) with the minus sign . comparing this equation with eq.([sub ] ) , we find that @xmath159 corresponds to @xmath160 . the euclidean action @xmath161 of the anti - instanton is exactly cancelled by the `` boundary '' term @xmath162 [ cf .
eq.([action ] ) ] .
it is then clear , that anti - instantons ( and , hence , the multi - instanton configurations ) can not play any role for the calculation of the density of states . formally , this happens because for @xmath163 we have : @xmath164 which tends to a constant value in the large @xmath17 limit , while the corresponding expression for the instanton solution eq.([pd ] ) diverges as @xmath165 . since the value of the determinant @xmath166 for the anti - instanton is the same as for the instanton , the anti - instanton contribution is proportional to @xmath167 and , therefore , vanishes in the @xmath168 limit
the authors are grateful to dr . i. v. kolokolov for useful discussions .
this work is supported by the `` stichting scheikundig onderzoek in nederland ( son ) '' and the `` stichting voor fundamenteel onderzoek der materie ( fom ) '' .
in this appendix we show that owing to the symmetry properties of the fgm hamiltonian a small perturbation of the optimal fluctuation @xmath169 does not strongly mix the symmetric and antisymmetric states , despite the small energy splitting between them . as was explained in sec .
[ fgm ] such a mixing , if it would occur , would invalidate our optimal fluctuation calculation .
first , we obtain the effective hamiltonian , acting on the subspace of the two intragap states , which includes the virtual excitations to all other electron states .
to this end we write the single - electron hamiltonian for the disorder realization @xmath5 , given by eq.([perturbation ] ) , in the form : @xmath170 where @xmath171 and @xmath172 is the perturbation : @xmath173 we divide the hilbert space of the hamiltonian @xmath174 into two subspaces : a and b. the subspace a consists of the symmetric and antisymmetric intragap states , which here are denoted as @xmath175 , while all the other states belong to the subspace b , so that @xmath176 where @xmath177 are the projection operators for the corresponding subspaces .
the effective energy - dependent hamiltonian , acting on the subspace a , has the form : @xmath178 where the first term is the diagonal @xmath179-by-@xmath179 matrix : @xmath180 the second term is the first - order perturbation , @xmath181 and the last term , @xmath182 describes the higher - order processes involving virtual excitations to the subspace b. we now want to show that the non - diagonal matrix elements of the effective hamiltonian @xmath183 are smaller than its diagonal matrix elements .
first , we show that the first - order nondiagonal matrix elements equal @xmath20 : on the one hand , using the relation @xmath184 [ cf . eq.([timerev ] ) ] we obtain that the first - order perturbation hamiltonian is real and symmetric : @xmath185 on the other hand , we can use the relation between the positive - energy and negative - energy eigenstates of the dirac hamiltonian , @xmath186 , to show that @xmath187 comparing eq.([hand1 ] ) with eq.([hand2 ] ) we obtain : @xmath188 next we consider the non - diagonal matrix elements of @xmath189 .
the perturbative expansion of @xmath190 in powers of @xmath191 describes virtual excitations into the subspace b , so that all intermediate states in this expansion are separated from the intragap states by a large energy ( @xmath192 ) . since @xmath193
, the expansion terms involved do not contain small energy denominators . moreover , the non - diagonal matrix elements of the operator ( [ virtex ] ) equal @xmath20 at @xmath37 .
the proof of this statement is analogous to the proof of eq.([ndzero ] ) and we shall not repeat it here .
, satisfying eq.([timerev ] ) , @xmath194 and @xmath172 are real symmetric matrices and that they both anticommute with @xmath195 .
] hence , the non - diagonal matrix elements of the operator @xmath190 are @xmath196 .
moreover , they contain an additional small factor @xmath197 .
therefore , the small disorder perturbation @xmath198 can not result in a strong mixing of the two intragap states . next
we show that the corrections to the energies of these two states due to the diagonal matrix elements of the effective hamiltonian are also small , which will justify the validity of our calculation of the optical absorption coefficient.@xcite using the expressions for the wave functions of the symmetric and antisymmetric states,@xcite the straightforward calculation of the first - order terms gives @xmath199 where @xmath200 since the condition @xmath201 implies @xmath202 [ see eq.([dist ] ) ] , the last equation can be written in the form : @xmath203 _ i.e. _ , the first - order correction to the energy of the intragap state with the energy @xmath48 is proportional to this small energy times a factor @xmath204 .
this result can be easily traced back to the fact that the energy of the midgap state of a single kink can not be changed by a small perturbation of @xmath2 ( see , _ e.g. _ , ref . ) and , therefore , it holds in all orders of the expansion of the effective hamiltonian in powers of @xmath205 .
thus , we found that , despite the small energy splitting between the symmetric and antisymmetric intragap states , a small variation of the disorder fluctuation does not strongly affect either the wave functions , or the energies of these two states .
this result is a direct consequence of symmetries of the hamiltonian eq.([dirac ] ) .
in this appendix we discuss the instanton trajectory for the boundary conditions eq.([inf ] ) , which one obtains in the continuum limit for an open chain with an even number of atoms . to avoid spurious divergences related to an infinite value of @xmath70 at the boundaries , we will use the following procedure .
first , we impose , instead of eq.([inf ] ) , the boundary conditions , @xmath206 where @xmath207 is large but finite .
next we take the limit @xmath168 and only then @xmath208 is tended to infinity . since for @xmath209 part of the instanton trajectory
@xmath210 lies outside the well of the potential eq.([pot ] ) located between @xmath211 and @xmath83 , the energy @xmath212 defined by eq.([energy ] ) is positive . for @xmath213
the equation for the instanton trajectory is , @xmath214 now , the particle s velocity becomes very large as soon as it leaves the well .
the large euclidean action eq.([eaction ] ) associated with this region is , however , completely compensated by the surface term eq.([baction ] ) in the infinite @xmath17 limit . to see this , one notes that even for infinite @xmath208 the time of motion @xmath17 along the instanton trajectory is finite , as the velocity grows very fast outside the well .
we are , however , interested in long time ( large @xmath17 ) trajectories . the only way to make @xmath17 large is to tend the positive energy @xmath212 to zero , in which case the particle would spend most of the time near the maxima @xmath211 and @xmath83 of the potential @xmath215 , where it has a small velocity .
the relation between the energy and the time of motion for large @xmath17 is , @xmath216 for @xmath207 the dependence of the energy on @xmath208 can be neglected .
when the energy @xmath212 tends to zero , the trajectory given by eq.([difur1 ] ) inside the well , _
i.e. _ , for @xmath217 , tends to the solution of eq.([difur ] ) with the plus sign . in other words , in the limit @xmath168 , this part of the trajectory becomes the instanton solution eq.([solution ] ) that we considered above . on the other hand ,
outside the well ( @xmath218 ) the solution of eq.([difur1 ] ) tends to the solution of eq.([difur ] ) with the minus sign .
the latter corresponds to zero disorder , @xmath91 , which explains the cancellation between the euclidean action coming from outside the well and the surface term .
we proved , therefore , that in the limit @xmath219 the suppression factor @xmath75 for the instanton with boundary conditions eq.([ninf ] ) is identical to eq.([iaction ] ) . the boundary conditions eq.([ninf ] )
imply that the solution @xmath100 is antisymmetric .
therefore , the jacobian @xmath123 given by eq.([jac ] ) equals unity , as before . at the same time , however , the values of the partial derivative @xmath220 and the determinant @xmath134 differ considerably from eqs .
( [ pd ] ) and ( [ det ] ) , respectively .
for instance , @xmath221 nevertheless , the product of the determinant and the partial derivative remains unchanged up to small corrections that vanish for @xmath168 .
therefore , despite the fact that the instanton trajectories for the boundary conditions eq.([inf ] ) and eq.([bc ] ) are rather different , the average density of states obtained by the saddle - point integration near these trajectories is the same for sufficiently long chains , as , of course , should be the case . | we use the optimal fluctuation method to find the density of electron states inside the pseudogap in disordered peierls chains .
the electrons are described by the one - dimensional dirac hamiltonian with randomly varying mass ( the fluctuating gap model ) .
we establish a relation between the disorder average in this model and the quantum - mechanical average for a certain double - well problem .
we show that the optimal disorder fluctuation , which has the form of a soliton - antisoliton pair , corresponds to the instanton trajectory in the double - well problem .
we use the instanton method developed for the double - well problem to find the contribution to the density of states from disorder realizations close to the optimal fluctuation .
= 23.5 cm = 15.9 cm = -15pt = 0.0 cm = 0.0 cm @figurefigure |
extensive interest has recently been devoted to the understanding of molecular motors , which play pivotal roles in cellular processes by performing mechanical work using the energy - driven conformational changes .
kinesin , myosin , f@xmath2-atpase , groel , rna polymerase , and ribosome belong to a group of biological machines that undergoes a series of conformational changes during the mechanochemical cycle where the molecular conformation is directly coupled to the chemical state of the ligand . although substantial progress has been achieved in understanding the underlying physical principles that govern molecular motors during the last decade , major issues still remain to be resolved .
specifically , some of the outstanding questions are as follows : ( a ) how is the chemical energy converted into mechanical work ? ( b ) how is the directionality of the molecular movement determined ?
( c ) how is the molecular movement coordinated or regulated ?
several biochemical experiments have quantified the kinetic steps @xcite , single molecule experiments using optical tweezers have measured the mechanical response of individual molecular motors @xcite , and an increasing number of crystal structures have provided glimpses into the mechanisms of molecular motors @xcite .
these experimental evidences , however , are not sufficient to fully address all the questions above .
for example , little is known not only about the structural details of each chemical state but also about the kinetic pathways connecting them .
hence , if feasible , a computational strategy using the coordinates from x - ray and/or nmr structures can shed light on the allosteric dynamics of molecular motors .
although some initial numerical studies @xcite have proceeded towards addressing issue ( a ) for a few cases where both open and closed structures are _ explicitly known _ , no previous attempt has been made to answer issue ( c ) . in this paper
we investigate this question in the context of the conventional kinesin where the mechanochemical coordination of the motor movement is best manifested among the motor proteins .
one of the experimentally best studied molecular motors is the conventional kinesin ( kinesin-1 ) @xcite , a relatively small sized motor protein that transports cellular materials by walking in an unidirectional hand - over - hand manner along the microtubule ( mt ) filaments . compared to other motor proteins involved in material transport such as myosin and dynein
, the conventional kinesin has a remarkable processivity , which can travel about a hundred ( @xmath38.2 @xmath4 ) steps without being dissociated from the mt .
the mechanochemical cycle conjectured from experiments suggests that there must be a dynamic coordination between the two motor domains in order to achieve such high processivity .
the quest to identify the origin of this dynamic coordination has drawn extensive attention among the kinesin community .
since hancock and howard @xcite first hypothesized that the `` internal strain '' was needed for processivity , the strain - dependent mechanochemistry became a popular subject in kinesin studies @xcite .
with the aid of optical tweezers , guydosh and block recently revisited this issue by monitoring the real - time kinesin dynamics in the presence of atp and @xmath5 , a tight binding atp analog @xcite .
they discovered that , when @xmath5 was bound to the kinesin , the pause - time of the step increased substantially and that the normal step was restored only after the obligatory backstep .
this suggests that @xmath5 is released only when the head bound with @xmath5 becomes the leading head ( l ) .
supported by this observation , they advocated a kinetic model in which the rearward strain via the neck - linker facilitates the release of the ligand from the l @xcite .
stated differently , the binding of the ligand to the l is inhibited because the rearward strain constitutes an unfavorable environment for the atp binding sites of the l. in the present study , we focus on the elucidation of the structural origin of the coordinated motion in kinesin by adopting a simple computational strategy
. better straightforward evidence of the regulation on the nucleotide binding site can be obtained when a structure in which both kinesin heads are simultaneously bound to the mt binding site is determined .
such a structure will allow us to identify the structural differences between the leading ( l ) and the trailing ( t ) head .
to date , this structure , however , has not yet been reported .
the only available structures include an isolated kinesin-1 without the mt @xcite , an isolated single - headed kinesin - like ( kif1a ) with various ligand states @xcite , and a single kif1a bound to the tubulin - dimer binding site @xcite .
therefore , we utilized existing protein data bank ( pdb ) structures and manually built a model system of the two - headed kinesin molecule with both heads bound to the tubulin binding sites ( see fig.2 and legend ) . this model was used to generate an ensemble of structures via simulations . a direct comparison between the l and t equilibrium structures shows that the tension built on the neck - linker induces the disruption of the nucleotide binding site of the l , which directly supports inferences from experimental observations @xcite .
* mechanochemical cycle of kinesin : * we begin by reviewing the mechanochemical cycle of kinesin molecule on the mt to clarify the importance of dynamic coordination between the two motor domains for kinesin processivity .
recent experiments using laser optical tweezers ( lot ) , cryo - electron microscopy , electron paramagnetic resonance , and fret , as well as the crystal structures at various states @xcite provide glimpses into the structural and dynamical details of how the kinesin molecule walks on the microbubule filaments . depending on the nucleotide state at the binding site , both the motor domain structure and the binding interface between kinesin and mt are affected .
in particular , a minor change of the motor domain coupled to the nucleotide is amplified to a substantial conformational change of the neck linker between the ordered and the disordered state .
experimental studies strongly suggest the mechanochemical cycle shown in fig.1 @xcite .
the mechanical stepping cycle of kinesin initiates with the binding of atp to the empty kinesin head strongly bound to the mt [ @xmath6 .
docking of the neck linker to the neck - linker binding motif on the leading head ( x in fig.1 ) propels the trailing head ( y in fig.1 ) in the ( + ) -direction of mt , which leads to an 8 @xmath4 mechanical step [ @xmath7 .
the interaction with the mt facilitates the dissociation of adp from the catalytic core of l [ @xmath8 .
atp is hydrolyzed and produces @xmath9 state for the t [ @xmath10 . when @xmath11 is released and the trailing head is unbound from the mt , the half - cycle is completed [ @xmath12 .
the mechanical step is achieved in a hand - over - hand fashion by alternating the binding of the two motor domains ( x and y in fig.1 ) to the mt @xcite .
high processivity of the kinesin requires this kinetic cycle to be stable ( remain within the yellow box in fig.1 ) .
a premature binding of the atp to the leading head in the state of @xmath13 should be prevented , i.e. , the condition @xmath14/(k_r^{(iii)}+k_{diss}^{(iii)})$ ] and @xmath15/(k_r^{(iv)}+k_{diss}^{(iv)})\rightarrow 0 $ ] should be satisfied in fig.1 ( see supporting information for the master equation describing the kinetic cycle ) .
atp binding to the ( iii ) or ( iv ) states can destroy the mechanochemical cycle of the kinesin .
the binding of atp on the leading head should be suppressed before the @xmath16-@xmath11 is released from the t. otherwise , both heads become adp - bound states , which have a weak binding affinity to the mt , and that leads to dissociation from the mt .
since the kinesin has a high processivity compared to other molecular motors , effective communication is required between the two heads regarding the chemical state of each of the partner motor domains . + * two - headed kinesin bound to the microtubule : * in the absence of interactions with the mt , the individual kinesin monomers fold into identical conformations . to achieve its biological function , however , folding into the native structure alone
is not sufficient .
coupled with the nucleotide and the mt , the two kinesin monomers in the dimeric complex need to alternate the acquisition of the native structure in a time - coordinated fashion for the uni - directional movement .
the currently available three - dimensional structure ( pdb i d : 3kin , structure 2 in fig.2-a ) , in which each monomer is in its native state , does not provide such a dynamic picture by failing in fulfilling the geometrical requirement of simultaneous bindings of both motor domain to the adjacent tubulin binding sites that have an 8-nm gap .
the inspection of 3-d structure suggests that a substantial increase of the distance between the two motor domains can be gained by breaking a few contacts associated with the neck - linker ( @xmath17 , @xmath18 ) and the neck - linker binding site on the motor domain ( @xmath19 ) . to this end
, we manipulated the 3-d structure of 3kin around the neck - linker of the l and created a temporary structure whose two heads bind to the mt binding sites simultaneously .
both l and t have energetic biases towards the identical native fold but the interactions with the tubulin binding sites adapt the dimeric kinesin structure into a different minimum structure , which is not known _ a priori_. we performed simulations ( see methods ) to relax this initial structure and to establish the thermal equilibrium ensemble of the kinesin molecule on mt ( see fig.3-a ) .
transient dimeric kinesin conformations corresponding to the steps ( iii ) and ( iv ) during the cycle ( fig.1 ) allow us to investigate the structural deviation between l and t of kinesin molecule .
this simple computational exercise can confirm or dismiss the experimental conjecture regarding whether the mechanochemical strain significantly induces regulation on the nucleotide binding site and also if it occurs in the l. + * catalytic core of the leading head is less native - like on the mt : * since the nucleotide binding and release dynamics is sensitively controlled by the kinesin structure , we assume that the nucleotide molecule has an optimal binding affinity to the kinesin motor domain in the native structure . for function there is a need to understand how the native structure of the kinesin motor domain is perturbed under the different topological constraints imposed on the dimeric kinesin configuration by interacting with the mt .
the equilibrium ensemble of the structures shows that the neck - linker is in the docked state for the t but undocked for l. in comparison to the native structure , the overall shape of the nucleotide binding pocket in the t is more preserved . as long as the mt constrains the two heads 8-nm apart , this configuration is dominant in the thermal ensemble ( see fig.3-a ) .
_ global shape comparison : _ there are in principle multiple ways to quantitatively compare the two motor domain structures . to assess the structural differences ,
the radius of gyration ( @xmath20 ) of the two motor domain structures from the equilibrium ensemble are computed ( see fig.3-b ) . because the neck - linker and the neck - helix adopt different configurations relative to the motor domain in each monomer
, we perform a @xmath21 analysis for the motor domains only ( residue 2 - 324 ) .
the @xmath21 distributions show that the l is slightly bigger than the t both in the size ( @xmath22 ) and in the dispersion ( @xmath23 ) .
meanwhile the @xmath21 for the native state ( 3kin ) is @xmath24 .
clearly , the sizes of both of the heads in the thermal ensemble are expanded at @xmath25 as compared to the native structure .
the size alone does not tell much about the difference between the structures .
the rmsd relative to the native structure and between the two motor domains ( residues 2 - 324 ) computed over the equilibrium ensemble gives @xmath26 , @xmath27 , @xmath28 , where @xmath29 is the rmsd between conformations x and y. if the @xmath30 helix is excluded from the rmsd calculation of motor domain ( residues 2 - 315 ) , then @xmath31 , @xmath32 , and @xmath33 .
the rmsd analysis shows that the @xmath30 helix significantly contributes more for the deviation of the leading head from its native state than the t. additional detailed comparisons with respect to the native state can be made using the structural overlap function of @xmath34 pair , @xmath35 , which is defined as @xmath36 where @xmath37 if @xmath38 , otherwise @xmath39 .
@xmath40 is the distance of @xmath34 pair in native state , where @xmath41 . by setting @xmath42 values identical in both heads ( i.e.
, both heads have the same native state ) , we compute the @xmath35 values for the trailing and the leading heads , respectively . the relative difference of the @xmath35 value between t and l , @xmath43 is defined by @xmath44 which quantitatively measures the structural difference of the two heads . based on the @xmath43 value ( fig.3-c ) , the distances between the mt binding motif of the t ( l11 , l12 , @xmath45 , @xmath46 ) and other secondary structure units ( @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 ) are 50 % more native - like than in the l. _ conserved native contacts in trailing head reveals the strain propagation pathway in leading head : _ a direct measure of similarity to the native structure is the fraction of native contacts preserved in the thermal ensemble @xcite . since we assume that atp affinity is optimized in the native state , we can readily assess the quality of the structure using this measure .
we quantify the nativeness of a pair using @xmath53 , where @xmath54 if @xmath55 , @xmath56 residues are in contact at the native state ( @xmath57 ) , and @xmath58 otherwise .
@xmath59 ( with @xmath60 or @xmath61 ) is obtained by averaging over the thermal ensemble .
when @xmath59 is averaged over all the native pairs , the average fraction of native contacts , @xmath62 is calculated as @xmath63 where @xmath64 is the total number of native pairs . for the t and l conformations , @xmath65 and @xmath66 , respectively .
the relative difference of native contacts between the two kinesin heads at the pair level , @xmath67 , is quantified similarly to eq.[eqn : chi ] as @xmath68 in fig.4 , @xmath67 is color - coded based on its value . as expected from the equilibrium ensemble , conspicuous differences are found around the structural motifs having direct contacts with neck - linker , giving @xmath69
. quantitative inspection of the other contacts is illustrated in the structure .
we color the kinesin head structure based on the @xmath67 value .
the residue pairs are colored in magenta if @xmath70 , red if @xmath71 , where the positive @xmath67 signifies that the native contacts in trailing head are more intact .
the residue pairs are colored in light - blue if @xmath72 , blue if @xmath73 .
more intact contacts , when the trailing and the leading head are compared , are visualized by yellow line in fig.5-b .
our analysis not only shows that there is higher probability of the formation of native contacts present in the t in comparison to the l , but also suggests how the tension is propagated towards the nucleotide binding site to disrupt the nativeness of the nucleotide binding pocket in the leading head .
as expected , a dense network of intact contacts are found between the neck - linker ( @xmath18 ) and the neck - linker binding motif ( @xmath19 ) .
this network continues along the @xmath52 helix , perturbing @xmath74 , @xmath75 , @xmath45 , and finally reaches the nucleotide binding site ( see si fig.6 for the nomencaltures of the secondary structures ) .
it is surprising that the disruptions of native contacts are found particularly in the nucleotide binding site , which is believed to be the trigger point for the allosteric transition .
all the important nucleotide binding motifs ( p - loop , switch-1 , switch-2 , and n4 ) are recognized by our simulational analysis using a nonlinear - hamiltonian ( see supporting information for the comparison with linear - harmonic potential represented as gaussian network model ) . + * estimate of the tension in the neck - linker : * the deformation of the leading motor domain is caused by the internal tension in the neck linker .
the tension on the neck linker is estimated using the force ( @xmath76 ) versus extension ( @xmath77 ) relationship of a worm - like chain model @xcite , @xmath78 , \label{eqn : wlc}\ ] ] where @xmath79 is the persistence length of the polymer and @xmath61 is the contour length . @xmath80 for the 15-amino acid neck - linker ( residue from 324 to 338 ) @xmath81 , and in the equilibrium ensemble of structures , @xmath82 @xmath4 . assuming that @xmath83 @xmath4 @xcite for this segment , we estimate a tension @xmath84 @xmath85 . by integrating eq.[eqn : wlc ] for @xmath86 @xmath4 . the tensional energy stored in the neck - linker
is obtained , which is 17 @xmath87 .
about 20% of the atp hydrolysis energy ( @xmath88 ) is stored in the neck - linker and directly perturbs the nucleotide binding site of the l , whereas mechanical action to the t is dissipated through the dense network of contacts formed between the neck - linker ( @xmath89 ) and the neck - linker binding site ( @xmath90 ) . for a given extension @xmath91 ,
when the length of the neck - linker is varied by @xmath92 , the variation in the length of the neck - linker can affect the effective tension as @xmath93(\delta l / l)$ ] .
for a given @xmath94 , the resulting tension change may be significant depending on the value of the extension @xmath77 .
experimentally , the kinesin dynamics has recently been studied by varying the linker length by introducing a spacer composed of amino acids @xcite . since @xmath95 @xmath4 for the insertion of a single amino acid
, the lengthening of the linker leads to a reduction of the tension by @xmath96 . in light of the force values
controlling the kinesin dynamics in lot experiments , which is @xmath97 @xmath85 and 7 @xmath85 is the stall force , a value of @xmath98 can be significant . according to hackney _ et .
_ s experimental analysis the processivity is reduced by @xmath99-fold when a single amino acid is inserted or deleted , and 6 or 12 additional amino acids resulted in to a 3@xmath1004 fold reduction in the kinetic processivity @xcite .
because of the size and time scale spanned by a typical molecular motor as well as the lack of crystal structures , theoretical study based on the structure is not a common approach such as master equation descriptions @xcite or brownian ratchet models @xcite .
knowledge of structural details , however , is the key ingredient to understand the mechanochemistry of molecular motors . in the present study
, we propose computational strategies to resolve this problem in kinesin dynamics , particularly the nucleotide binding regulation mechanism between the two motor domains . by building a kinesin model on the tubulin filament
, we explicitly identified the effect of internal tension ( @xmath84 @xmath85 ) on the front kinesin head domain and showed that the tension propagation to the leading head provokes the switch - related motifs ( p - loop , switch-1 , switch-2 ) in the nucleotide binding pocket .
assuming `` the nativeness as a criterion for the optimal nucleotide binding condition '' , we concluded that the nucleotide binding pocket in the leading head is not favorable ( or partially unfolded or cracked @xcite ) for the nucleotide binding while the trailing head is bound to the mt .
this conclusion explains the recent real - time single molecule traces of kinesin generated by guydosh and block @xcite .
the reduction of ligand affinity of the leading head due to the rearward tension benefits the high processivity in two ways .
first , the premature atp binding is inhibited before the chemical reaction ( atp hydrolysis , @xmath11 release ) in the trailing head is completed , during which the trailing head is tightly bound to the mt and the tension is built on the neck linker .
second , the release of adp is facilitated in the leading head , which accelerates step @xmath101 in fig.1 .
the latter point is consistent with uemura and ishiwata s experimental observation of a seven - fold increase of the adp dissociation constant from the monomeric kinesin head in the presence of rearward loading @xcite .
it is noteworthy that global measures characterizing the two head domains do not distinguish the qualitative differences between the two heads .
major changes in the strained region such as the kinesin neck - linker may be associated to minor global differences in the motor domain .
exactly how these minor global differences that are localized in a small volume is amplified in the process of allosteric transition is the another key issue in understanding molecular motors .
the internal tension regulates the interaction between the kinesin and the nucleotide .
conversely , the interaction between kinesin , nucleotide , and mt also switches on and off the internal tension . for conventional kinesin , mechanical and chemical mechanism
are closely correlated , producing a remarkable processivity of the kinesin movement on the mt .
the simulations were performed using two classes of energy function .
one is the standard structure based ( sb ) potential @xcite and the other is the self - organized polymer ( sop ) potential @xcite ( see supporting information and si figs .
7 - 8 ) . for these two different topology - based potential functions
, we obtained qualitatively identical results .
results from the sb potential are presented in figs.3 - 5 while results from the sop potential are in si fig.9 in supporting materials .
this suggests that the results are robust as long as the information of native topology with a nonlinear form of energy potential is used as an input . +
* energy function@xmath100structure based potential : * using the structure in fig.2-c as a starting structure , we simulated and sampled an ensemble of two - headed kinesin configurations on the mt .
we performed the langevin simulations of the sb model @xcite whose equation of the motion of each interaction center is integrated by a verlet algorithm .
the energy potential is given as @xmath102})\nonumber\\ & + \sum_{i=1}^{n_k-3}\sum_{j = i+3}^{n_k}\left[\epsilon_h\left(\left(\frac{r^o_{ij}}{r_{ij}}\right)^{12}-2\left(\frac{r^o_{ij}}{r_{ij}}\right)^6\right)\delta_{ij}+\epsilon_l\left(\frac{\sigma}{r_{ij}}\right)^{12}(1-\delta_{ij})\right]\nonumber\\ & + \sum_{i=1}^{n_k}\sum_{k=1}^{n_{tub}}\left[\epsilon_h\left(\left(\frac{r^o_{ik}}{r_{ik}}\right)^{12}-2\left(\frac{r^o_{ik}}{r_{ik}}\right)^6\right)\delta^*_{ik}+\epsilon_l\left(\frac{\sigma}{r_{ik}}\right)^{12}(1-\delta^*_{ik})\right],\end{aligned}\ ] ] where the energy hamiltonian is divided into intramolecular interactions for the kinesin molecule and intermolecular interactions at the kinesin - mt interface .
the superscripts @xmath103 and @xmath104 denote the kinesin and the kinesin - tubulin interaction , respectively .
because our focus is on the kinesin dynamics , we fixed the coordinates of the mt in space .
since the length scale of the kinesin geometry is small ( @xmath105 ) compared to that of mt ( diameter @xmath106 , persistence length @xmath107 ) , the explicit computation of the dynamics of the entire mt structure , in which the 13 protofilaments constitute the cylindrical geometry , should not qualitatively change our conclusions .
the first and the second term define the backbone interactions . the bond distance @xmath108 between the neighboring residues @xmath55 and @xmath109 are harmonically constrained with respect to the bond distance in native state @xmath110 with a strength @xmath111 @xmath112 . in the second term , the angle @xmath113 is formed between residues @xmath55 , @xmath109 , and @xmath114 with @xmath115 @xmath116 .
@xmath117 is the angle of the native state .
the third term is the dihedral angle potential with @xmath118 and @xmath119 that describes the ease of rotation around the angle formed between successive residues from @xmath55 to @xmath120 along the backbone .
the lennard - jones potential is used to account for the interactions that stabilize the native topology .
a native contact is defined from the pair of interaction centers whose distance is less than @xmath121 in native state for @xmath122 . if @xmath55 and @xmath56 residues are in contact in the native state , @xmath54 , otherwise @xmath58 . aided by @xmath123 we assign stabilizing potential for native pairs and repulsive potential for non - native pairs .
we assign @xmath124 @xmath125 for the intra and inter neck - helix ( residue@xmath126 ) interactions to secure the coiled - coil association between the neck - helices . for other kinesin residue - residue interactions
, we set @xmath127 @xmath125 regardless of the sequence identity .
the parameters determining the native topology @xmath128 and @xmath123 are determined from the crystal structure of human kinesin ( pdb i d : 3kin ) and incorporated to the trailing kinesin ( t ) and the coiled - coil whose structure is shown in fig.2-c . to constitute an identical fold condition , we transferred topological information in trailing head ( t ) to the leading head ( l ) by substituting @xmath128 , @xmath117 , @xmath129 and @xmath123 from t to l , i.e. , @xmath130 , @xmath131 , @xmath132 and @xmath133 for all @xmath55 and @xmath56 . the kinesin - tubulin interaction parameters ( @xmath134 , @xmath135 , @xmath136 , and @xmath137 )
are similarly defined as kinesin intramolecular interaction parameters except for the slightly larger native contact distance ( @xmath138 ) .
the parameters @xmath136 and @xmath137 defining the interface topology between the trailing kinesin head ( t ) and the tubulin are also transfered to the interface topology between the leading head l and the next tubulin binding site . + * simulations :* the initial structure , whose two heads are constrained to be oriented on the tubulin binding sites , is relaxed under the sb or sop - hamiltonian and subsequently the equilibrium ensemble of the structures is collected from the low friction langevin dynamics simulations at @xmath139 . the position of the interaction center is integrated using @xmath140 where @xmath141 is friction coefficient , @xmath142 is the conformation force , and @xmath143 is the random force satisfying @xmath144 where the integration time ( @xmath145 ) is discretized . in low friction langevin dynamics , natural time
is given by @xmath146 .
we chose @xmath147 and @xmath148 .
low friction is deliberately chosen for the purpose of effectively sampling the conformational space @xcite . under such conditions
the resulting dynamics as a function of time step should not be taken parallel to the real time dynamics . to produce an overdamped dynamics it is essential to integrate the motion by neglecting the inertial term as well as choosing a high friction coefficient that amounts to the water viscosity ( @xmath149 ) .
we are grateful to paul whitford and sam cho for critically reading the manuscript .
this work was funded by the nsf grant 0543906 and by the nsf - sponsored center for theoretical biological physics ( grants phy-0216576 and 0225630 ) .
* figure * 1 : mechanochemical cycle of kinesin .
the subscripts @xmath150 and @xmath151 refer to each of the kinesin head , @xmath152 denotes the empty head , and ( : mt ) is appended if the head is strongly bound to the mt .
* figure * 2 : procedure to construct the two - headed kinesin / mt - protofilament model .
* a. * three structures from protein data bank ( pdb ) are used .
1 . single headed kinesin ( kif1a ) bound on tubulin ( pdb i d : 1ia0 ) .
two - headed kinesin ( pdb i d : 3kin ) .
two consecutive tubulin complexed to the stathmin - like domain ( pdb i d : 1ffx ) .
we overlaped the chain ( a ) , ( b ) ( blue in figure ) of 3kin onto the chain k of 1ia0 and the chain a of 1ffx ( @xmath153-domain ) onto the chain ( b ) of 1ia0 , which leads to the structure in * b*. the structural homology ( @xmath0 backbone rmsd=1.6 ) between kif1a and a head of the two - headed kinesin are sufficient that one of the kinesin heads fits to the tubulin binding site . while the sequence difference between kif1a and the conventional kinesin ( sequence identity @xmath154% ) may affect the strength of interactions between kinesin and tubulin , leading to a different binding affinity of conventional kinesin from kif1a
, we assume that the binding orientation of the two - headed kinesin is similar to that of kif1a on the tubulin .
after the structure overlap , the chain ( c ) , ( d ) of 3kin is internally rotated around a few positions in the neck linker ( 324 - 338 ) until the chains c , d are placed in the vicinity of binding site of tubulin that is designed to be identical to the interface between the kinesin rear head and the tubulin .
we performed the simulation to relax the kinesin structure on the mt and obtained the structure shown in * c*. * figure * 3 : the ensemble of structures and structural comparisons between two heads using @xmath21 and @xmath43 . *
a. * the thermal ensemble of structures is illustrated using the multiple structures obtained during the simulations .
different colors are used to distinguish the motor domain ( residues 2@xmath100323 ) from the neck - linker and the tail part ( residues 324@xmath100370 ) .
substantial variations of the neck - linker / tail position in ensemble show its flexibility .
the nucleotide binding sites in the l and t are indicated by the arrows . on the upper - right corner of the panel
are two crystal structures of kinesin ( 3kin ) .
one is the view from the top ( left ) and the other is the view towards the nucleotide binding pocket ( right ) . when compared with the top view of 3kin crystal structure , it is visually clear that the nucleotide binding pocket is more intact in the t. * b. * analysis of kinesin motor domains ( 2 - 324 ) using the radius of gyration ( @xmath21 ) .
histograms of @xmath21 collected over the ensemble are fit to a gaussian distribution . for the trailing head ( red ) , @xmath155 , @xmath156 . for the leading head ( blue ) , @xmath157 , @xmath158 .
* c. * analysis using the structure overlap function .
@xmath43 value is color - coded on the right ( see the text for the definition of @xmath43 ) .
* figure * 4 : comparison of two heads using the fraction of native contact .
* a*. the average contact map for kinesin ( left ) .
the relative difference of average contact map between trailing head and leading head with respect to trailing head is shown on the right .
the red arrows mark the set of contact pairs whose @xmath67 value is greater than 0.5 ( 50% ) .
* b*. the result of @xmath67 in * a * is redrawn using 3-d plot for clarity .
* figure * 5 : the structure of kinesin on mt colored based on the protocol discussed in the main text . * a. * the residues colored in red and magenta on the trailing head ( structure on the left ) are those maintaining more contact than the residues in the leading head ( structure on the right ) .
some of the residues in colored region of the trailing head are involved with the nucleotide binding site . *
b. * the enlarged view in cartoon representation from neck - linker docking site ( left ) and from ligand binding site ( right ) .
sphere representations in orange are adp molecule . the network of contact pairs are depicted in yellow lines . , width=576 ] , width=480 ] , width=768 ] , width=864 ] , width=576 ] * supporting information * + * structural details of kinesin . *
detailed knowledge of the structure is fundamental to understand the working mechanism of a biological nanomachine . a monomer of the conventional kinesin is structurally categorized into three parts : the head ( or motor domain ) ( residues 2@xmath100323 ) , the neck linker ( residues 324@xmath100338 : @xmath17 , @xmath18 ) , and the neck - helix ( residue 339@xmath100 : @xmath159 ) .
the head region is composed of eight @xmath160-strands flanked with three @xmath153-helices [ @xmath50 , @xmath51 , @xmath161 on one side ( fig .
6a ) , and @xmath162 , @xmath46 , @xmath52 on the other side ( fig .
6b ) ] on each side of the @xmath160-sheet .
one side contains the binding motif ( @xmath162 , @xmath163 , @xmath52 , l8 , l11 , l12 ) for the microtubule ( fig .
6a ) and the other side provides a nucleotide binding pocket ( fig .
the nucleotide binding site in the kinesin head region is structurally homologous across the motor protein and the g - protein superfamilies .
the structural motifs around the nucleotide binding site , such as the p - loop ( n1 ) ( 86@xmath10093 ) , switch-1 ( n2 ) ( 199@xmath100204 ) , switch-2 ( n3 ) ( 232@xmath100237 ) , n4 ( 14 - 17 ) are accordingly designated @xcite .
the crystal structures with the different nucleotide states suggest that the presence or the absence of @xmath16-@xmath11 is sensed by these motifs and that the structural changes of the motifs are related to the allosteric transitions .
the ordered state of the neck - linker , which is extended from the n terminus of kinesin head and is composed of two beta strands ( @xmath17 and @xmath18 ) , forms contacts with the n - terminal region of the @xmath19-strand .
further extension from the neck - linker leads to the neck - helix ( @xmath159-helix ) , through which two monomers form a dimeric complex . +
* computations of residue displacement cross - correlation using elastic gaussian network model and simulation under sb potential . * for given coordinates of a complex three - dimensional structure
, the dynamical property of an object can be extracted at the zeroth order by investigating its topology .
the gaussian network model ( gnm ) is the simplest possible method to study large biomolecules using the corresponding minimal topology @xcite .
the gnm views the biomolecular construct as a collection of beads connected with harmonic springs with strength @xmath16 .
the connectivity between the beads is purely determined by the cut - off distance parameter @xmath164 .
for a pair of beads @xmath55 and @xmath56 , whose distance @xmath165 satisfies @xmath166 , the harmonic potential constrains the position of beads via @xmath167 where @xmath168 is the heaviside function , @xmath169 with @xmath170 , and @xmath171 is the @xmath172 element of kirchhoff matrix @xmath173 .
use of the harmonic potential amounts to the expansion of the potential @xmath174 at the potential minimum @xmath175 as @xmath176 . since the partition function of gnm is given by @xmath177e^{-\beta h}=\left[\det\left(\frac{\gamma\gamma}{2\pi k_bt}\right)\right]^{-3/2}$ ]
, the correlation between the spatial fluctuation of two residues is expressed using the inverse of kirchhoff matrix , @xmath178 for @xmath179 , the mean square displacement of the @xmath180 residue , @xmath181 , corresponds to the b - factor ( debye - waller temperature factor ) as @xmath182 .
a comparison between the b - factor and mean square displacement ( msd ) from the gnm determines the effective strength of the harmonic potential that stabilizes the structure .
note that the quality of the msd in gnm is solely controlled by the @xmath164 value , thus we scaled the msd with @xmath183 for gnm analysis .
we applied the gnm analysis with @xmath184 on the two - headed kinesin whose both heads fit to the adjacent tubulin binding site , and then computed the cross - correlation matrix as shown in fig.7a .
the cross - correlation value @xmath185 , @xmath186 scaled by @xmath183 , shows that except for the neck - helix region the amplitude of correlation in leading head is always larger than that of the trailing head .
this is expected since the neck - linker of the leading kinesin is detached from the motor domain .
the residues in the network with less coordination number are subject to a larger fluctuation .
the relative difference of the cross - correlation between the leading and the trailing kinesin using @xmath187 is illustrated in fig . 7b .
7c shows the auto - correlations ( or mean square displacement ) , which are the diagonal elements of the @xmath185 matrix .
gnm analysis is useful in analyzing the fluctuation dynamics of the stable structure at the residue level in the basin of attraction where the basin is modeled as a quadratic potential .
however , the expansion of the potential minima up to the quadratic term is justified only if the fluctuation @xmath188 is small .
the amplitude of fluctuation in biological systems at physiological temperatures ( @xmath189 ) is most likely to exceed the limit beyond which nonlinear response is no longer negligible . in order to take this effect into account
, the hamiltonian should be expanded beyond the linear response regime .
this procedure indeed reverses the simple idea that tirion @xcite and bahar et al .
@xcite have proposed in the context of gnm analysis .
however , _
minimal _ inclusion of the nonlinear term can be useful by increasing the susceptibility of the structure . once the nonlinear term is included ,
a simple analytical expression such as eq.[eqn : z_n ] is not available .
thus , we resort to the simulations . the analytically obtained quantities , @xmath186 , @xmath190 in fig .
7 can also be calculated over the thermal ensemble of structures obtained from simulations using a nonlinear - hamiltonian ( see fig .
the first conclusion drawn from the simulational analysis is similar to the gnm in that the leading head experiences larger fluctuations .
secondly , the position and relative amplitude of the msd peaks , reproduced using the simulation results , shows a good agreement with gnm results .
however , the direct comparison of @xmath185 ( or @xmath190 ) between fig . 7 and fig . 8
shows that the simulation results from the nonlinear - hamiltonian display a more sensitive pattern of cross - correlations .
the pronounced amplitude of @xmath185 ( or @xmath190 ) suggests a strong spatial correlation between residues @xmath55 and @xmath56 . + * alternative energy function : sop potential . *
an alternative potential function for the sb potential used in the main text is the self - organized polymer ( sop ) potential that was recently adopted for simulations of the mechanical unfolding of large molecules of rna and proteins @xcite as well as the allosteric dynamics of groel @xcite .
the energy hamiltonian is defined as @xmath191\delta_{ij}\nonumber\\ & + \sum_{i=1}^{n_k-2}\epsilon_l\left(\frac{\sigma}{r_{i , i+2}}\right)^6+\sum_{i=1}^{n_k-3}\sum_{j = i+3}^{n_k}\epsilon_l\left(\frac{\sigma}{r_{ij}}\right)^6(1-\delta_{ij})\nonumber\\ & + \sum_{i=1}^{n_k}\sum_{k=1}^{n_{tub}}\left[\epsilon_h\left(\left(\frac{r^o_{ik}}{r_{ik}}\right)^{12}-2\left(\frac{r^o_{ik}}{r_{ik}}\right)^6\right)\delta^*_{ik}+\epsilon_l\left(\frac{\sigma}{r_{ik}}\right)^6(1-\delta^*_{ik})\right ] .
\label{eq : sop } \tag{9}\end{aligned}\ ] ] the first term is for the chain connectivity of the kinesin molecule . the finite extensible nonlinear elastic ( fene ) potential @xcite is used with @xmath192@xmath193 , @xmath194 , and @xmath108 is the distance between neighboring interaction centers @xmath55 and @xmath109 .
the lennard - jones potential interactions stabilize the native topology . a native contact is defined as the pair of interaction centers whose distance is less than @xmath195 in native state for @xmath122 .
if @xmath55 and @xmath56 sites are in contact in the native state , @xmath54 , otherwise @xmath58 .
we used @xmath124 @xmath125 in the native pairs , and @xmath196 @xmath125 for non - native pairs . to ensure the non - crossing of the chain
, we used a @xmath197 power potential in the repulsion terms and set @xmath198 , which is typical @xmath199 distance .
the parameters determining the native topology , @xmath200 and @xmath123 , are adopted from the trailing kinesin ( x ) whose structure is shown in fig .
2_c_. we transferred the topological information in the trailing head ( t ) to the leading head ( l ) by substituting @xmath200 and @xmath123 from the t to l. kinesin - tubulin interaction energies are similarly defined as kinesin intramolecular interaction energies with slightly different native contact distances .
we set the cut - off distance for the native interactions between the kinesin and the tubulin as @xmath201 .
the parameters , @xmath136 and @xmath202 , defining the interface topology between the kinesin head t and the tubulin is transfered to the kinesin head l and the next tubulin binding site . using the sop potential
, we obtained qualitatively identical results as those obtained from the sb potential .
the nucleotide binding pocket of the front head is disrupted in the dimeric kinesin configuration whose both heads are bound to the tubulin binding sites .
the figures corresponding to fig .
3_a _ and _ c _ and fig .
5 are regenerated using sop model in fig .
+ * master equations for the mechanochemical cycle of kinesin described in fig . 1 . * in the limit when the dissociation of dimeric kinesin from the microtubule is negligible , the kinetic equation describing the dynamic cycle shown in fig . 1 is written as @xmath203p_{(i)}+k_rp_{(ii)}+k_{dmt}p_{(iv)}+k_ap_{(v)}\nonumber\\ \frac{dp_{(ii)}}{dt}&=-(k_r+k_d)p_{(ii)}+k_{bi}[atp]p_{(i)}\nonumber\\ \frac{dp_{(ii')}}{dt}&=-k_{dadp}p_{(ii')}+k_dp_{(ii)}\nonumber\\ \frac{dp_{(iii)}}{dt}&=-(k_h+k_{bi}^{(iii)}[atp])p_{(iii)}+k_{dadp}p_{(ii')}+k_r^{(iii)}p_{(iii')}\nonumber\\ \frac{dp_{(iv)}}{dt}&=-(k_{dmt}+k_{bi}^{(iv)}[atp])p_{(iv)}+k_hp_{(iii)}+k_r^{(iv)}p_{(iv')}\nonumber\\ \frac{dp_{(iii')}}{dt}&=-(k_r^{(iii)}+k^{(iii)}_{diss})p_{(iii')}+k_{bi}^{(iii)}[atp]p_{(iii)}\nonumber\\ \frac{dp_{(iv')}}{dt}&=-(k_r^{(iv)}+k^{(iv)}_{diss})p_{(iv')}+k_{bi}^{(iv)}[atp]p_{(iv)}\nonumber\\ \frac{dp_{(v)}}{dt}&=-k_ap_{(v)}+k_{diss}^{(iii)}p_{(iii)}+k_{diss}^{(iv)}p_{(iv ) } , \tag{10}\end{aligned}\ ] ] where @xmath204 is the probability of finding the molecule in a mechanochemical state @xmath205 with @xmath206 .
the steady state solutions by setting @xmath207 leads to @xmath208}\left(1+\frac{k_r}{k_d}\right)\frac{\mathcal{x}}{\mathcal{z}},\quad p_{(ii)}=\frac{1}{k_d}\frac{\mathcal{x}}{\mathcal{z}},\quad p_{(ii')}=\frac{1}{k_{dadp}}\frac{\mathcal{x}}{\mathcal{z}}\nonumber\\ p_{(iii)}&=\frac{\mathcal{y}}{\mathcal{z}},\quad\quad p_{(iv)}=\frac{1}{\mathcal{z}},\quad\quad p_{(iii')}=k_{m}^{(iii)}\frac{\mathcal{y}}{\mathcal{z}},\quad\quad p_{(iv')}=k_{m}^{(iv)}\frac{1}{\mathcal{z}}\nonumber\\ p_{(v)}&=\frac{1}{k_a}\left(k_{diss}^{(iii)}k_{m}^{(iii)}\mathcal{y}+k_{diss}^{(iv)}k_{m}^{(iv)}\right)\frac{1}{\mathcal{z}}\nonumber\\ \mathcal{x}&\equiv k_{dmt}\left(1+\frac{k^{(iii)}_{diss}}{k_h}k_{m}^{(iii)}\right)\left(1+\frac{k^{(iv)}_{diss}}{k_{dmt}}k_{m}^{(iv)}\right)\nonumber\\ \mathcal{y}&\equiv \frac{k_{dmt}}{k_h}\left(1+\frac{k_{diss}^{(iv)}}{k_{dmt}}k_{m}^{(iv)}\right)\nonumber\\ \mathcal{z}&\equiv\left[1+\left(1+\frac{k_{diss}^{(iii)}}{k_a}\right)k_{m}^{(iii)}\right]\mathcal{y}+\left[1+\left(1+\frac{k_{diss}^{(iv)}}{k_a}\right)k_{m}^{(iv)}\right]\nonumber\\ & + \left[\frac{1}{k_{bi}[atp]}\left(1+\frac{k_r}{k_d}\right)+k_d^{-1}+k_{dadp}^{-1}\right]\mathcal{x}\nonumber\\ k_{m}^{(iii)}&\equiv\frac{k_{bi}^{(iii)}[atp]}{k_r^{(iii)}+k_{diss}^{(iii)}},\quad k_{m}^{(iv)}\equiv\frac{k_{bi}^{(iv)}[atp]}{k_r^{(iv)}+k_{diss}^{(iv)}}. \label{eqn : ss } \tag{11}\end{aligned}\ ] ] when the average velocity at steady state is computed using @xmath209p_{(i)}-k_r p_{(ii)})=\mathcal{x}/\mathcal{z}$ ] , one can write the velocity in the form of michaelis - menten equation . @xmath210))}[atp]}{\frac{\frac{k^*}{1+\mathcal{q}([atp])}(1+\frac{k_r}{k_d})}{k_{bi}}+[atp ] } = d\frac{u_1^o[atp]}{\frac{u_1 ^ 0+w_1 ^ 0}{k_0 ^ 0}+[atp ] } = \frac{v_{max}[atp]}{k_m+[atp ] } \label{eqn : velocity } \tag{12}\ ] ] where @xmath211 , @xmath212)&\equiv\frac{k^*}{k_h}\left[\frac{1+\left(1+\frac{k_{diss}^{(iii)}}{k_a}\right)k_m^{(iii)}}{1+\frac{k_{diss}^{(iii)}}{k_h}k_m^{(iii)}}-1\right]\nonumber\\ & + \frac{k^*}{k_{dmt}}\left[\frac{1+\left(1+\frac{k_{diss}^{(iv)}}{k_a}\right)k_m^{(iv)}}{\left(1+\frac{k_{diss}^{(iii)}}{k_h}k_m^{(iii)}\right)\left(1+\frac{k_{diss}^{(iv)}}{k_h}k_m^{(iv)}\right)}-1\right],\nonumber \tag{13}\end{aligned}\ ] ] and @xmath213 @xmath4 ( the gap between neighboring tubulin binding sites ) . if the dissociation from the microtubule is suppressed by small @xmath214 and @xmath215 then @xmath216 .
depending on the rate constant , @xmath217 can be either positive or negative .
although the large dissociation constant reduces the processivity , the presence of dissociation can increase the effective velocity of kinesin if @xmath218 .
the second and the third expressions following the equality sign are given to compare our result with the ( n=2)-model of fisher et .
@xcite and michaelis - menten kinetics , respectively .
if @xmath219 the ( n=2)-model analysis on the experimental data by block and coworkers @xcite predicts @xmath220 , @xmath221 , and @xmath222 ( @xmath223 ) . this sets the lower bound for the parameters as @xmath224 , @xmath225 , @xmath226 , @xmath227 @xmath228 .
the average run length of kinesin , @xmath61 , is calculated using @xmath229 where @xmath230 is the number of mechanical steps of kinesin . if the probability of the dissociated kinesin is small ( @xmath231 ) , ( i.e. , if @xmath232_{(v)}$ ] is negligible in fig .
1 ) , then @xmath233 where @xmath234 is the dissociation rate . + 10 ma , y. z. & taylor , e. w. ( 1997 ) _ j. biol . chem . _ * 272 * , 724730 .
moyer , m. l. , gilbert , s. p. , & johnson , k. a. ( 1998 ) _ biochemistry _ * 37 * , 800813 .
visscher , k. , schnitzer , m. j. , & block , s. m. ( 1999 ) _ nature _ * 400 * , 184187 .
shaevitz , j. w. , abbondanzieri , e. a. , landick , r. , & block , s. m. ( 2003 ) _ nature _ * 426 * , 684687 .
chemla , y. r. , aathavan , k. , michaelis , j. , grimes , s. , jardine , p. j. , anderson , d. l. , & bustamante , c. ( 2005 ) _ cell _ * 122 * , 683692 .
kozielski , f. , sack , s. , marx , a. , thormhlen , m. , schnbrunn , e. , biou , v. , thompson , a. , mandelkow , e. m. , & mandelkow , e. ( 1997 ) _ cell _ * 91 * , 985994 .
nitta , r. , kikkawa , m. , okada , y. , & hirokawa , n. ( 2004 ) _ science _ * 305 * , 678683 .
sablin , e. p. , case , r. b. , dai , s. c. , hart , c. l. , ruby , a. , vale , r. d. , & fletterick , r. j. ( 1998 ) _ nature _ * 395 * , 813816 .
rayment , i. , rypniewski , w. r. , schmidt - base , k. , smith , r. , tomchick , d. r. , benning , m. m. , winkelmann , d. a. , wesenberg , g. , & holden , h. m. ( 1993 ) _ science _ * 261 * , 5058 .
abrahams , j. p. , leslie , a. g. w. , lutter , r. , & walker , j. e. ( 1994 ) _ nature _ * 370 * , 621628 .
xu , z. , horwich , a. l. , & sigler , p. b. ( 1997 ) _ nature _ * 388 * , 741 .
cramer , p. , bushnell , d. a. , & kornberg , r. d. ( 2001 ) _ science _ * 292 * , 18631876 .
yusupov , m. m. , yusupova , g. z. , baucom , a. , lieberman , k. , earnest , t. n. , cate , j. h. d. , & noller , h. f. ( 2001 ) _ science _ * 292 * , 883896 .
koga , n. & takada , s. ( 2006 ) _ proc .
sci . _ * 103 * , 53675372 .
okazaki , k. , koga , n. , takada , s. , onuchic , j. n. , & wolynes , p. g. ( 2006 ) _ proc .
sci . _ * 103 * , 1184411849 .
hyeon , c. , lorimer , g. h. , & thirumalai , d. ( 2006 ) _ proc .
sci . _ * 103 * , 1893918944 .
yu , j. , ha , t. , & schulten , k. ( 2006 ) _ biophys .
j. _ * 91 * , 20972114 .
brady , s. t. ( 1985 ) _ nature _ * 317 * , 7375 .
vale , r. d. , reese , t. s. , & sheetz , m. p. ( 1985 ) _ cell _ * 42 * , 3950 .
hancock , w. d. & howard , j. ( 1999 ) _ proc .
sci . _ * 96 * , 1314713152 .
uemura , s. & ishiwata , s. ( 2003 ) _ nature struct .
* 10 * , 308311 .
klumpp , l. m. , hoenger , a. , & gilbert , s. p. ( 2004 ) _ proc .
_ * 101 * , 34443449 .
rosenfeld , s. s. , fordyce , p. m. , jefferson , g. m. , king , p. h. , & block , s. m. ( 2003 ) _
_ * 278 * , 1855018556 .
guydosh , n. r. & block , s. m. ( 2006 ) _ proc .
_ * 103 * , 80548059 .
kikkawa , m. , okada , y. , & hirokawa , n. ( 2000 ) _ cell _ * 100 * , 241252 . rice , s. , lin , a. w. , safer , d. , hart , c. l. , naber , n. , carragher , b. o. , cain , s. m. , pechatnikova , e. , wilson - kubalek , e. m. , whittaker , m. , pate , e. , cooke , r. , taylor , e. m. , milligan , r. a. , & vale , r. d. ( 1999 ) _ nature _ * 402 * , 778784 .
sindelar , c. v. , budny , m. j. , rice , s. , naber , n. , fletterick , r. , & cooke , r. ( 2002 ) _ nature struct .
biol . _ * 9*(11 ) , 844848 .
cross , r. a. ( 2004 ) _ trends biochem .
* 29 * , 301309 .
hackney , d. d. ( 1994 ) _ proc .
sci . _ * 91 * , 68656869 .
asbury , c. l. , fehr , a. n. , & block , s. m. ( 2003 ) _ science _ * 302 * , 21302134 .
cho , s. s. , levy , y. , & wolynes , p. g. ( 2006 ) _ proc .
sci . _ * 103 * , 586591 .
marko , j. f. & siggia , e. d. ( 1996 ) _ macromolecules _ * 27 * , 981988 . schuler , b. , lipman , e. a. , steinbach , p. j. , kumke , m. , & eaton , w. a. ( 2005 ) _ proc . natl .
sci . _ * 102 * , 27542759 .
hackney , d. d. , stock , m. f. , moore , j. , & patterson , r. a. ( 2003 ) _ biochemistry _ * 42 * , 1201112018 .
terada , t. p. , sasai , m. , & yomo , t. ( 2002 ) _ proc .
_ * 99 * , 92029206
. fisher , m. e. & kolomeisky , a. b. ( 2001 ) _ proc .
sci . _ * 98 * , 77487753 .
fisher , m. e. & kim , y. c. ( 2005 ) _ proc .
sci . _ * 102 * , 1620916214 .
reimann , p. ( 2002 ) _ phys . rep .
_ * 361 * , 57265 .
miyashita , o. , onuchic , j. n. , & wolynes , p. g. ( 2003 ) _ proc .
sci . _ * 100 * , 1257012575 .
clementi , c. , nymeyer , h. , & onuchic , j. n. ( 2000 ) _ j. mol .
biol . _ * 298 * , 937953 .
hyeon , c. & thirumalai , d. ( 2007 ) _ biophys .
j. ( in press ) _ * 92 * , 731743 .
hyeon , c. , dima , r. i. , & thirumalai , d. ( 2006 ) _ structure _ * 14 * , 16331645 .
honeycutt , j. d. & thirumalai , d. ( 1992 ) _ biopolymers _ * 32 * , 695709 .
sack , s. , muller , j. , alexander , m. , manfred , t. , mandelkow , e. m , brady , s. t. , & mandelkow , e. ( 1997 ) _ biochemistry _ * 36 * , 16155 16165 .
sack , s. , kull , f. j. , & mandelkow , e. ( 1999 ) _ eur .
j. biochem . _ * 262 * , 111 .
tirion , m. m. ( 1996 ) _ phys .
lett . _ * 77*(9 ) , 19051908 .
haliloglu , t. , bahar , i. , & erman , b. ( 1997 ) _ phys .
lett . _ * 79*(16 ) , 30903093 .
tama , f. , valle , m. , frank , j. , & brooks iii , c. l. ( 2003 ) _ proc .
_ * 100 * , 93199323 .
kremer , k. & grest , g. s. ( 1990 ) _ j. chem
. phys . _ * 92 * , 50575086 .
* fig . 6 * : kinesin structure .
( _ a _ ) top view when kinesin is bound to tubulin .
the helices on the top with respect to @xmath160-sheet are colored in pink .
( _ b _ ) bottom view .
the helices on the bottom with respect to @xmath160-sheet or on the side of the tubulin binding interface are colored in lightblue .
l11 loop , which is not observable in the crystal structure because of disorder , is tentatively drawn with a dashed line .
( _ c _ ) side view .
note that the neck - linker ( @xmath235 , @xmath18 ) is connecting the @xmath159 neck - helix with the @xmath52 helix in motor domain .
( _ d _ ) view around the nucleotide binding site .
p - loop , switch-1 , switch-2 , and n4 regions , which are relevant to nucleotide binding , are colored in red with annotation . +
* : analysis of the kinesin equilibrium dynamics using the gaussian network model ( gnm ) .
( _ a _ ) the cross - correlation map of the residue fluctuation for the two - headed kinesin structure is shown in fig.2c .
( _ b _ ) the amplitudes are color - coded based on its value .
the relative difference between the trailing and the leading head with respect to leading head ( @xmath190 ) is plotted on the right .
( _ c _ ) mean square displacement of residues in the trailing head ( red ) and the leading head ( blue ) are plotted on the same plot .
a comparison of the amplitudes shows that the leading head fluctuates more than the trailing head . the relative difference between the two plots with respect to leading head ( @xmath236 )
is plotted on the right panel . + * fig . 8 * : analysis of the kinesin equilibrium dynamics using an equilibrium ensemble generated from the simulations under sb - hamiltonian .
the legends for _ a_-_c _ are identical to si fig . 7 .
+ * fig . 9
* : results of the strain induced regulation in the kinesin dimer on the mt generated using the sop potential ( eq.[eq : sop ] ) .
comparisons between the results obtained with the sop potential and with the sb potential confirm qualitatively the identical conclusions . | in the presence of atp , kinesin proceeds along the protofilament of microtubule by alternated binding of two motor domains on the tubulin binding sites .
since the processivity of kinesin is much higher than other motor proteins , it has been speculated that there exists a mechanism for allosteric regulation between the two monomers .
recent experiments suggest that atp binding to the leading head domain in kinesin is regulated by the rearward strain built on the neck - linker .
we test this hypothesis by explicitly modeling a @xmath0-based kinesin structure whose both motor domains are bound on the tubulin binding sites .
the equilibrium structures of kinesin on the microtubule show disordered and ordered neck - linker configurations for the leading and the trailing head , respectively .
the comparison of the structures between the two heads shows that several native contacts present at the nucleotide binding site in the leading head are less intact than those in the binding site of the rear head .
the network of native contacts obtained from this comparison provides the internal tension propagation pathway , which leads to the disruption of the nucleotide binding site in the leading head .
also , using an argument based on polymer theory , we estimate the internal tension built on the neck - linker to be @xmath1 pn . both of these conclusions support the experimental hypothesis .
= 22pt |
learning the covariance structure of a stochastic process from data is a fundamental prerequisite for problems such as prediction , classification and control . for example , to do prediction for an ornstein uhlenbeck ( ou ) ( uhlenbeck and ornstein @xcite ) process @xmath0 , @xmath1 $ ] with mean 0 and covariance function @xmath2 where @xmath3 are unknown ,
ying @xcite proposed the maximum likelihood ( ml ) estimators for @xmath4 and @xmath5 based on discrete observations @xmath6 with @xmath7 , and established the root-@xmath8 consistency of the corresponding ml estimator for @xmath9 .
note that since the probability measures induced by two ou processes are absolutely continuous with respect to each other if and only if their @xmath9 values are equal ( ibragimov and rozanov @xcite ) , the parameters in ( [ expcovfun1dim ] ) are asymptotically identifiable up to @xmath9 .
however , when the ou process is subject to measurement error , the so - called `` nugget '' effect ( see , for example , cressie @xcite ) may deteriorate the performance of the ml estimators .
in particular , chen , simpson and ying @xcite showed that the ml estimator for @xmath9 becomes fourth - root-@xmath8 consistent , depicting the effect of measurement error in estimating the exponential covariance parameters in ( [ expcovfun1dim ] ) . on the other hand
, they also proved that the ml estimator of the measurement - error variance has the usual root-@xmath8 consistency .
in fact , a similar phenomenon can also be found in a driftless brownian motion ( bm ) process with measurement error .
let this error - contaminated process be denoted by @xmath10 $ ] .
having observed @xmath11 , stein @xcite showed that a modified ml ( mml ) estimator of the ratio of the variance of the increments of the bm process to that of measurement error is only fourth - root-@xmath8 consistent , whereas the corresponding mml estimator of the measurement - error variance still remains root-@xmath8 consistent .
similar asymptotic results for the ml estimators of the two variances have also been established by at - sahalia , mykland and zhang @xcite . in this article
, we shall superimpose a time trend ( regression ) term on an ou process with measurement error in order to accommodate a broader range of applications . specifically , we propose the following model for a real - valued stochastic process
@xmath12 : @xmath13 where @xmath14 is a @xmath15-dimensional time trend vector , @xmath0 is a zero - mean ou process with covariance function defined in ( [ expcovfun1dim ] ) , @xmath16 is a zero - mean gaussian measurement error with @xmath17 for some unknown @xmath18 , @xmath19 is a @xmath20-dimensional constant vector , and @xmath21 , @xmath22 and @xmath23 are independent . in a computer experiment , @xmath0 in ( [ setupfullmodel ] ) can be used to describe the systematic departure of the response @xmath24 from the linear model @xmath25 and @xmath16 denotes the measurement error . for more details ,
we refer the reader to sacks , schiller and welch @xcite and ying @xcite .
model ( [ setupfullmodel ] ) can also be applied to one - dimensional geostatistical modeling and @xmath26 therein corresponds to a commonly used exponential covariance model ; see ripley @xcite and cressie @xcite for numerous examples .
denote the true time trend by @xmath27 where @xmath28 is independent of @xmath22 and @xmath23 , and define @xmath29 .
the time trend model @xmath30 in ( [ setupfullmodel ] ) is said to be correctly specified if @xmath31 and misspecified otherwise . in this article
, we shall allow @xmath32 to be misspecified , which further increases the flexibility of model ( [ setupfullmodel ] ) .
however , a misspecified time trend will usually create extra challenges in estimating covariance parameters .
this motivates us to ask how the ml estimators of the covariance parameters in model ( [ setupfullmodel ] ) perform when the corresponding time trend model is subject to misspecification . to facilitate exposition
, we assume in the sequel that @xmath33 $ ] for some @xmath34 , and the data are observed regularly at @xmath35 , @xmath36 .
in addition , we also allow that the number of regressors ( model complexity ) @xmath37 grows to infinity in order to reduce the model bias .
when @xmath38 , the domain @xmath39 $ ] has been considered by the aforementioned authors , and the setup is called fixed domain asymptotics . on the other hand , when @xmath40 , the domain @xmath41 grows to infinity as @xmath42 with a faster growing rate for a larger @xmath43 value , and the setup is referred to as the increasing domain asymptotics , even though the minimum inter - data distance @xmath44 goes to zero .
this is different from the increasing domain setup considered by zhang and zimmerman @xcite , in which the minimum distance between sampling points is bounded away from zero . by incorporating both fixed and increasing domains ,
our _ mixed _ domain asymptotic framework enables us to explore the interplay between the model misspecification / complexity and the growing rate of @xmath41 on the asymptotic behaviors of the ml estimators , thereby leading to an intriguing answer to the above question .
re - parameterizing ( [ expcovfun1dim ] ) by @xmath45 and @xmath46 , the covariance parameter vector in model ( [ setupfullmodel ] ) can be written as @xmath47 .
let @xmath48 , the parameter space , be a compact set in @xmath49 and suppose @xmath50 .
based on model ( [ setupfullmodel ] ) and observations @xmath51 , we estimate @xmath52 using the ml estimator @xmath53 , which satisfies @xmath54 where @xmath55 \label{loglikefuntrue } \\[-8pt ] \nonumber & & { } -\tfrac{1}{2}\mathbf{z}'\bigl(\mathbf{i}-\mathbf{m}(\bolds \theta)\bigr)'\bolds\sigma^{-1 } ( { \bolds\theta } ) \bigl ( \mathbf{i}-\mathbf{m}(\bolds\theta)\bigr)\mathbf{z},\end{aligned}\ ] ] is known as the profile log - likelihood function , in which @xmath56 , @xmath57 with @xmath58 and @xmath59 with @xmath60 being full rank almost surely ( a.s . ) . it is not difficult to show that the ml estimator of @xmath61 is given by @xmath62 , where @xmath63 however , since model ( [ setupfullmodel ] ) can be misspecified , investigating the asymptotic properties of @xmath64 is beyond the scope of this paper
let @xmath65 and @xmath66 .
by @xmath67 , ( [ setupfullmodelf ] ) and ( [ loglikefuntrue ] ) , we have @xmath68 where with @xmath69 , @xmath70 is the log - density function for @xmath71 . as will be seen in section [ sec2 ] , the contribution of the time trend to @xmath72
is mainly made by @xmath73 the first term above , vanishing when ( [ correct ] ) holds true , is due to model misspecification , and the second term , having an order of magnitude @xmath74 uniformly over @xmath48 ( see lemma [ lemmaprojectionsup ] ) , is related to model complexity .
we therefore introduce @xmath75 as a uniform bound for ( [ bound ] ) over @xmath48 .
let @xmath76 .
the growing rates of @xmath41 needed for @xmath77 , to achieve consistency are given in the next theorem in terms of the order of magnitude of @xmath78 .
it provides a preliminary answer to the question of whether the covariance structures of @xmath79 and @xmath80 can be learnt from data under possible model misspecification .
[ thmconsistency ] suppose @xmath81 for some @xmath82 .
then , for @xmath34 , @xmath83 theorem [ thmconsistency ] shows that as long as ( [ assumptionofmisselected ] ) holds true , @xmath84 is a consistent estimator of @xmath85 , regardless of the value of @xmath43 .
in contrast , in order for @xmath86 and @xmath87 to achieve consistency , one would require @xmath88 and @xmath89 , respectively .
in fact , these two constraints can not be weakened because we provide counterexamples in section [ sec3 ] illustrating that @xmath87 is no longer consistent when @xmath90 , and both @xmath86 and @xmath87 fail to achieve consistency if @xmath91 .
it is worth mentioning that @xmath92 is highly convoluted due to the involvement of regression terms , making it difficult to establish consistency of @xmath93 .
our strategy is to decompose the nonstochastic part of @xmath72 into several layers whose first three leading orders are @xmath94 , @xmath95 and @xmath96 , respectively , and express the remainder stochastic part as the sum of @xmath97 and two other terms that can be uniformly expressed as @xmath98 and @xmath99 ; see ( [ lemmaconvergea.s.0 ] ) .
one distinctive characteristic of these nonstochastic layers is that the coefficient associated with the @xmath100th ( @xmath101 ) leading layer only depends on @xmath102 .
when ( [ assumptionofmisselected ] ) is assumed , this hierarchical layer structure together with some uniform bounds established for the second moments of @xmath103 enables us to derive the consistency of @xmath93 in the order of @xmath84 , @xmath104 and @xmath105 by focusing on one layer and one parameter at a time .
let @xmath106 denote the trace of a matrix @xmath107 . as shown in the proof of theorem [ thmconsistency ] , the uniform bounds for @xmath103
are first expressed in terms the supremums of @xmath108 , or other similar trace terms such as those given in ( [ pfthm01eq01 ] ) .
these expressions are obtained using the idea that the sup - norms of a sufficiently smooth function can be bounded above by suitable integral norms , as suggested in lai @xcite , chan and ing @xcite and chan , huang and ing @xcite .
we then carefully calculate the orders of magnitude of the aforementioned traces , yielding uniform bounds in terms of @xmath109 or @xmath110 .
note that dahlhaus @xcite has applied the chaining lemma ( see pollard @xcite ) to obtain uniform probability bounds for some quadratic forms of a discrete time long - memory process .
however , since no rates have been reported in his bounds , his approach may not be directly applicable here .
whereas theorem [ thmconsistency ] has demonstrated the performance of @xmath53 from the perspective of consistency , the questions of what are the convergence rates of and whether there are central limit theorems ( clts ) for @xmath111 , still remain unanswered .
the next section is devoted to these questions .
in particular , it is shown in theorem [ thmctlunderincorrectmodel ] that for @xmath112 , @xmath101 , @xmath113 if @xmath114 , and @xmath115 has a limiting normal distribution if @xmath116 . since the time trend is involved , our proof of theorem [ thmctlunderincorrectmodel ] is somewhat nonstandard .
we first obtain the initial convergence rates of @xmath93 using the standard taylor expansion and an argument similar but subtler than the one used in the proof of theorem [ thmconsistency ] .
using these initial rates , we can improve the convergence results through the same argument .
we then repeat this iterative procedure until the final convergence results are established .
the rest of this article is organized as follows . in section [ sec2 ] ,
we begin by establishing the clt for @xmath111 in situations where @xmath117 is fixed and the regression model is correctly specified ( namely , ( [ correct ] ) is true ) ; see theorem [ thmctlundertruemodel ] .
we subsequently drop these two restrictions and report in theorem [ thmctlunderincorrectmodel ] the most general convergence results of this paper . in section [ sec3 ] ,
we provide two counterexamples showing that the results obtained in theorem [ thmconsistency ] are difficult to improve .
the proofs of all theorems and corollaries in the first three sections are given in section [ sectionproofoftheoremsandcorollaries ] .
the proofs of the auxiliary lemmas used in section [ sectionproofoftheoremsandcorollaries ] are provided in the supplementary material ( chang , huang and ing @xcite ) in light of space constraint . before leaving this section
, we remark that although our results are derived under the gaussianity of @xmath118 and @xmath119 , similar results can be obtained when either @xmath118 or @xmath120 is not ( but pretended to be ) gaussian , provided some fourth moment information is available . on the other hand , while we allow the time trend to be misspecified , we preclude a misspecified covariance model .
the interested reader is referred to xiu @xcite for some asymptotic results on the ml estimators when the covariance model considered in stein @xcite or at - sahalia , mykland and zhang @xcite is misspecified .
in this section , we begin with establishing the asymptotic normality of @xmath121 , in situations where the regression model is correctly specified and @xmath117 is fixed .
[ thmctlundertruemodel ] assume that ( [ correct ] ) holds and @xmath117 is a fixed nonnegative integer .
( note that these assumptions yield @xmath122 in ( [ assumptionofmisselected ] ) . ) then for @xmath34 , @xmath123 and for @xmath124 , @xmath125 one of the easiest ways to understand theorem [ thmctlundertruemodel ] is to link the result to the fisher information matrix .
straightforward calculations show that under the assumption of theorem [ thmctlundertruemodel ] , the diagonal elements of the fisher information matrix evaluated at @xmath126 are given by @xmath127 where the trace terms are solely contributed by the log - density ( log - likelihood ) function for @xmath71 ( defined in ( [ logpdfe+e ] ) ) , and the @xmath128 terms , which vanish if the time trend is known to be zero , are related to the model complexity . moreover , by ( [ sigma1sigmainverse2 ] ) , ( [ diffsigma1sigmainverse ] ) and ( [ sigma1sigmainverse4 ] ) , @xmath129 it is interesting pointing out that the denominator on the right - hand side of the first equation of ( [ fisherview ] ) coincides exactly with the limiting variance in ( [ thmtheta1consistency ] ) .
this is reminiscent of a conventional asymptotic theory for the ml estimate which says that the limiting variance of the ml estimate is the reciprocal of the corresponding fisher information number . on the other hand , while the reciprocals of the right - hand sides of the second and third identities of ( [ fisherview ] ) are the same as the limiting variances in ( [ thmtheta2consistency ] ) and ( [ thmtheta3consistency ] ) , the divergence rates of the corresponding trace terms @xmath130 and @xmath110 are much slower than @xmath8 .
in fact , they are equal to the divergence rates of the second and third leading layers of the nonstochastic part of @xmath131 ; see ( [ lemmaconvergea.s.0 ] ) .
these findings reveal that the amounts of information related to @xmath132 s have different orders of magnitude , thereby leading to different normalizing constants in the clts for @xmath133 s .
the next theorem improves theorem [ thmctlundertruemodel ] by deriving rates of convergence of @xmath134 , without requiring @xmath122 in ( [ assumptionofmisselected ] ) .
it further shows that clts for @xmath134 , are still possible if the model misspecification / complexity associated with the time trend has an order of magnitude smaller than @xmath135 and @xmath136 , respectively . [ thmctlunderincorrectmodel ] suppose that ( [ assumptionofmisselected ] ) is true .
then for @xmath34 , @xmath137 and for @xmath124 , @xmath138 in addition , for @xmath34 , @xmath139 and for @xmath124 , @xmath140 recall that @xmath141 , @xmath142 and @xmath143 . it is shown in ( [ fisherview0 ] ) and ( [ fisherview ] ) that , ignoring the constant , the amount of information regarding @xmath132 contained in @xmath71 is @xmath144 , @xmath145 . on the other hand ,
as will become clear later , @xmath146 can be used to measure the amount of information contaminated by model misspecification / complexity ( again ignoring the constant ) .
therefore , the first part of theorem [ thmctlunderincorrectmodel ] delivers nothing more than the simple idea that @xmath147 provided that @xmath148 \label{maincondition } \\[-8pt ] \nonumber & & \quad<\mbox{amount of information regarding } \theta_{0,i } \mbox { contained in } \bolds\eta+ \bolds\epsilon.\end{aligned}\ ] ] note that the second term on the right - hand side of ( [ mainidea ] ) is the best rate one can expect when the time trend is known to be zero .
the second part of theorem [ thmctlunderincorrectmodel ] further indicates that the clts for @xmath133 s in theorem [ thmctlundertruemodel ] carry over to situations where ( [ maincondition ] ) holds with the right - hand side replaced by its square root . to the best of our knowledge ,
this is one of the most general clts established for @xmath133 s . in the following , we present two specific examples illustrating how the asymptotic behavior of @xmath133 s is affected by the interaction between @xmath149 and @xmath43 . in the first example
, the model misspecification yields @xmath150 , and hence @xmath90 . according to theorem [ thmctlunderincorrectmodel ] , the clts for @xmath84 and @xmath104 hold for a certain range of @xmath43 .
[ coropolymle0 ] consider the intercept - only model of ( [ setupfullmodel ] ) with @xmath151 .
suppose that @xmath152 , where @xmath153 and @xmath154 are nonzero constants .
then for @xmath155 , @xmath156 \label{cor1theta1consistency } \\[-8pt ] \nonumber n^{1-\delta}(\hat\theta_1- \theta_{0,1 } ) & = & o_p(1 ) ; \qquad\delta\in[1/2,1 ) , \\ n^{(1+\delta)/4 } ( \hat{\theta}_2-\theta_{0,2 } ) & \displaystyle\mathop { \rightarrow}^{d } & n \bigl(0 , 2^{5/2}\theta_{0,1}^{1/2 } \theta_{0,2}^{3/2 } \bigr ) ; \qquad \delta\in[0,1/3 ) , \nonumber \\[-8pt ] \label{cor1theta2consistency } \\[-8pt ] \nonumber n^{(1-\delta)/2 } ( \hat\theta_2 -\theta_{0,2 } ) & = & o_p(1 ) ; \qquad\delta\in[1/3,1).\end{aligned}\ ] ] we remark that the scaling factor @xmath157 is introduced for the linear term , @xmath158 , so that @xmath159 does not depend on @xmath8 , where @xmath160 .
the model misspecification in the next example results in @xmath161 , yielding @xmath91 .
therefore , @xmath84 is guaranteed to be consistent in view of theorem [ thmctlunderincorrectmodel ] .
[ coroexp0 ] consider the same setup as in corollary [ coropolymle0 ] except that @xmath162 , where @xmath163 is generated from a zero - mean gaussian spatial process with covariance function @xmath164 , \ ] ] for some constants @xmath165 . then for @xmath155 ,
@xmath166 it is worth noting that @xmath87 is inconsistent under the setup of corollary [ coropolymle0 ] .
moreover , both @xmath86 and @xmath87 are inconsistent under the setup of corollary [ coroexp0 ] .
these inconsistency results will be reported in detail in the next section . before closing this section
we remark that our theoretical results on @xmath93 can be used to make statistical inference about the regression function .
for example , when ( [ correct ] ) holds and @xmath167 is a fixed integer , the convergence rate of @xmath93 obtained in theorem [ thmctlundertruemodel ] plays an indispensable role in analyzing the convergence rate of the ml estimator , @xmath62 , of @xmath61 .
recently , by making use of theorems [ thmctlundertruemodel ] and [ thmctlunderincorrectmodel ] , chang , huang and ing @xcite established the first model selection consistency result under the mixed domain asymptotic framework .
moreover , some technical results established in the proofs of theorems [ thmctlundertruemodel ] and [ thmctlunderincorrectmodel ] have been used by chang , huang and ing @xcite to develop a model selection consistency result under a misspecified covariance model .
using the examples constructed in corollaries [ coropolymle0 ] and [ coroexp0 ] , we show in this section that the constraints @xmath168 and @xmath169 imposed in theorem [ thmconsistency ] for the consistency of @xmath87 and @xmath86 , respectively , can not be relaxed .
[ coropolymle ] under the setup of corollary [ coropolymle0 ] , @xmath170 [ coroexp ] under the setup of corollary [ coroexp0 ] , @xmath171 all the above results can be illustrated by figure [ orderarea ] , in which some change point behavior of @xmath133 s ( in terms of modes of convergence ) is exhibited when @xmath172 runs through the region @xmath173 .
@cc@ to @xmath174 with respect to @xmath175 , where @xmath176 , @xmath43 is the growing rate of the domain and @xmath149 satisfies @xmath177 .
note that @xmath133 also possesses asymptotic normality when @xmath175 falls in the dark gray regions , but may fail to achieve consistency when @xmath175 falls in the white regions or on the dash lines .
in addition , the points on the lines between the light and dark gray area are referred to as the change points merely in the modes of convergence but not in the convergence rate scenario.,title="fig : " ] & to @xmath174 with respect to @xmath175 , where @xmath176 , @xmath43 is the growing rate of the domain and @xmath149 satisfies @xmath177 . note that @xmath133 also possesses asymptotic normality when @xmath175 falls in the dark gray regions , but
may fail to achieve consistency when @xmath175 falls in the white regions or on the dash lines .
in addition , the points on the lines between the light and dark gray area are referred to as the change points merely in the modes of convergence but not in the convergence rate scenario.,title="fig : " ] + & + + [ orderarea ]
in this section , we first prove the consistency of @xmath53 in section [ sec41 ] . the proofs of clts for @xmath53 with and without the restrictions of correct specification and fixed dimension on the time trend model are given in sections [ sec42 ] and [ sec43 ] , respectively .
the proofs of corollaries [ coropolymle0 ] and [ coropolymle ] and those of corollaries [ coroexp0 ] and [ coroexp ] are provided in sections [ sec44 ] and [ sec45 ] , respectively . to prove theorem [ thmconsistency ]
, we need a series of auxiliary lemmas , lemmas [ lemmaardecomposition][lemmauniform ] .
lemma [ lemmaardecomposition ] gives a modified cholesky decomposition for @xmath178 , which can be used to prove lemma [ propositionmaxandminboundsofeigenvalues ] , asserting that the eigenvalues of @xmath179 are uniformly bounded above and below .
lemmas [ lem02ttraces ] and [ lem03dcompsegetag ] provide the orders of magnitude of the cholesky factors of @xmath178 and the products of @xmath180 and these factors . based on lemmas [ propositionmaxandminboundsofeigenvalues][lem03dcompsegetag ] , lemma [ lemmaexpequations ] establishes asymptotic expressions for the key components of the nonstochastic part of @xmath72 , and lemma [ lemmaexpequations1 ] provides the orders of magnitude of @xmath181 .
lemmas [ propositionmaxandminboundsofeigenvalues ] and [ lemmaexpequations1 ] can be used in conjunction with lemma [ lemmauniform ] , which provides uniform bounds for quadratic forms in i.i.d .
random variables , to analyze the asymptotic behavior of @xmath97 ; see ( [ lemmah0result ] ) .
lemmas [ lemmaprojectionsup ] and [ lemmamisselectedmodel ] explore the effects of the time trend model on @xmath182 .
[ lemmaardecomposition ] let @xmath183 be given by ( [ sigma ] ) with @xmath184 , @xmath185 and @xmath186 .
then @xmath187 where @xmath188 @xmath189 , and @xmath190 [ propositionmaxandminboundsofeigenvalues ] let @xmath191 and @xmath192 denote the maximum and minimum eigenvalues of the matrix @xmath107
. for @xmath183 given by ( [ sigma ] ) , suppose that @xmath193 is compact .
then , @xmath194 \label{propositionboundeigen } \\[-8pt ] \nonumber & \leq & \limsup_{n\rightarrow\infty}\sup_{\bolds\theta\in\theta } \lambda_{\max } \bigl(\bolds\sigma^{-1/2}(\bolds\theta ) \bolds \sigma(\bolds\theta_0)\bolds\sigma^{-1/2}(\bolds\theta ) \bigr ) < \infty.\end{aligned}\ ] ] [ lem02ttraces ] under the setup of lemma [ lemmaardecomposition ] , for any @xmath195 , where @xmath48 is compact , and @xmath34 , the following equation holds uniformly over @xmath48 : @xmath196 [ lem03dcompsegetag ] under the setup of lemma [ lem02ttraces ] , for any @xmath197 , @xmath198 where @xmath199 , @xmath200 and @xmath201 .
in addition , for any @xmath34 , @xmath202 furthermore , for any @xmath124 , @xmath203 [ lemmaexpequations ] under the setup of lemma [ lem02ttraces ] , the following equations hold uniformly over @xmath48 : @xmath204 \label{logdetsigma } \\[-9pt ] \nonumber & & { } -\frac{1-\delta}{2}\log n + o\bigl(n^{\delta}\bigr)+ o(1 ) , \\
\operatorname{tr } \bigl(\bolds\sigma(\bolds\theta_0)\bolds \sigma^{-1}(\bolds\theta ) \bigr ) & = & \frac{\theta_{0,1}}{\theta_1}n - \frac{\theta_{0,1}}{2\theta_1 } \biggl(\frac{2\theta_2}{\theta_1 } \biggr ) ^{1/2}n^{(1+\delta)/2}\nonumber \\ \label{sigma1sigmainverse3 } & & { } + \frac{\theta_{0,2}}{(2\theta_1\theta_2)^{1/2}}n^{(1+\delta)/2 } + \frac { \theta_{0,2}(\theta_3 ^ 2-\theta_{0,3}^2)}{2\theta_2\theta_{0,3}}n^\delta \\ & & { } + o\bigl(n^\delta\bigr)+o(1).\nonumber\end{aligned}\ ] ] [ lemmaexpequations1 ] under the setup of lemma [ lem02ttraces ] , the following equations hold uniformly over @xmath48 : @xmath205 \operatorname{tr } \bigl(\bolds\sigma_\eta(\bolds\theta_0 ) \bolds\sigma^{-1}(\bolds\theta ) \bigr ) & = & \frac{\theta_{0,2}}{(2\theta _ 1\theta_2)^{1/2}}n^{(1+\delta)/2 } + \frac{\theta_{0,2}(\theta_3 ^ 2-\theta_{0,3}^2)}{2\theta_2\theta _ { 0,3}}n^\delta \nonumber \\[-9pt ] \label{sigma1sigmainverse } \\[-9pt ] \nonumber & & { } + o\bigl(n^\delta\bigr)+ o(1 ) ,
\\ \label{diffsigma1sigmainverse } \operatorname{tr } \biggl ( \biggl(\bolds\sigma^{-1}(\bolds\theta ) \frac{\partial}{\partial\theta_3 } \bolds\sigma(\bolds\theta ) \biggr)^2 \biggr ) & = & \frac{1}{\theta_3}n^\delta+ o\bigl(n^\delta\bigr).\end{aligned}\ ] ] as will be shown later , ( [ propositionboundeigen ] ) , ( [ sigma1sigmainverse2 ] ) and ( [ sigma1sigmainverse ] ) can be used to derive bounds for @xmath206 and @xmath207 . these bounds , together with ( [ diffsigma1sigmainverse ] ) , play important roles in establishing the consistency of @xmath208 . [ lemmaprojectionsup ]
let @xmath209 be full rank a.s . then under the setup of lemma [ lem02ttraces ] , @xmath210 where @xmath211 is defined in ( [ projectionmatrix ] ) .
[ lemmamisselectedmodel ] under the setup up of lemma [ lem02ttraces ] , let @xmath209 be full rank a.s .
suppose that for some @xmath212 , @xmath213 then @xmath214 before introducing lemma [ lemmauniform ] , we need some notation . for @xmath215 , define @xmath216 .
let @xmath217 be a function of @xmath218 .
for @xmath219 , define @xmath220 .
denote by @xmath221 the @xmath222-dimensional closed ball centered at @xmath223 with radius @xmath224 . for @xmath225 , define the @xmath226-dimensional sphere : @xmath227 [ lemmauniform ] assume that @xmath228 are i.i.d .
random variables with @xmath229 and @xmath230 .
let @xmath231_{1\leq i ,
j\leq n}$ ] be an @xmath232 matrix whose @xmath233th component is @xmath234 , a function of @xmath235 with a continuous partial derivative @xmath236 on @xmath221 , for @xmath237 .
define @xmath238 , where @xmath239 .
then for @xmath240 , there exists a constant @xmath241 such that @xmath242 \label{lemmauniformbound } \\[-8pt ] \nonumber & & \quad\leq c \sum_{m=1}^{r}\sum _ { \mathbf{j } \in \mathbf{j}(m , r)}\operatorname{vol}^{2}\bigl(b_{\tau } ( \bolds\lambda , { \mathbf j})\bigr ) \sup_{\bolds\xi\in b_{\tau}(\bolds \lambda ) } \operatorname{var}\bigl ( \mathbf{d}_{\mathbf{j}}q_1(\bolds\xi)\bigr),\end{aligned}\ ] ] where @xmath243 denotes the volume of @xmath48 .
first , we prove ( [ lemmatheta1consistency ] ) . by ( [ assumptionofmisselected ] ) , ( [ logdetsigma ] ) , ( [ sigma1sigmainverse3 ] ) , ( [ finitevariable ] ) and ( [ lemmamis2 ] )
, it follows that @xmath244 \label{lemmaconvergea.s.0 } \\[-8pt ] \nonumber & & { } - \biggl\{\frac{\theta_2}{\theta_1 } + \theta_3-\frac{\theta _ { 0,2}(\theta_3 ^ 2 - \theta_{0,3}^2)}{2\theta_2\theta_{0,3 } } \biggr\ } n^\delta \\ & & { } + h(\bolds\theta ) + o_p\bigl(n^\xi\bigr ) + o_p\bigl(n^\delta\bigr),\nonumber\end{aligned}\ ] ] uniformly in @xmath48 , where @xmath69 .
hence , ( [ lemmatheta1consistency ] ) is ensured by for any @xmath245 , @xmath246 as @xmath42 , where @xmath247 . since by ( [ lemmaconvergea.s.0 ] ) , @xmath248 and
since @xmath249 , ( [ lemmatheta1consistency0 ] ) follows immediately from @xmath250 since @xmath97 is continuous on @xmath48 and @xmath48 is compact , in the rest of the proof , we assume without loss of generality that @xmath251 , a closed ball centered at @xmath252 with radius @xmath253 for some @xmath224 . by ( [ lemmauniformbound ] ) with @xmath254 and @xmath255 , we obtain @xmath256 and @xmath257 \label{ing0504a } \\[-8pt ] \nonumber & & \qquad{}+ \operatorname{var } \biggl(\frac{\partial^2}{\partial\theta_1\,\partial\theta _ 2 } h(\bolds\theta ) \biggr ) + \operatorname{var } \biggl(\frac{\partial^2}{\partial\theta_1\,\partial\theta_3 } h(\bolds \theta ) \biggr ) + \operatorname{var } \biggl ( \frac{\partial^2}{\partial\theta_2\,\partial\theta_3 } h(\bolds\theta )
\biggr ) \\ & & \qquad{}+\operatorname{var } \biggl(\frac{\partial^3}{\partial\theta_1\,\partial\theta _ 2\,\partial\theta_3 }
h(\bolds\theta ) \biggr ) \biggr\},\nonumber \ ] ] for some constant @xmath241 . by ( [ propositionboundeigen ] ) , ( [ sigma1sigmainverse ] ) , @xmath258 for the nonnegative definite matrices @xmath259 and @xmath260 , and using @xmath261 twice
, we obtain @xmath262 \label{sigma1sigmainverse4 } \\[-8pt ] \nonumber & & { } -\operatorname{tr}\bigl(\bolds\sigma^{-2 } ( \bolds\theta_0 ) \bolds\sigma_\eta(\bolds\theta_0)\bigr ) \biggr\ } \\ & = & \frac{1}{\theta_{0,1}^2}n + o\bigl(n^{(1+\delta)/2}\bigr).\nonumber\end{aligned}\ ] ] equations ( [ propositionboundeigen ] ) , ( [ ing0504b ] ) and ( [ sigma1sigmainverse4 ] ) lead to @xmath263 \label{ing0505a } \\[-8pt ] \nonumber & = & \sup_{\bolds\theta\in\theta } 2\operatorname{tr } \bigl ( \bigl(\bolds \sigma^{-2}(\bolds\theta ) \bolds\sigma(\bolds\theta_0 ) \bigr)^2 \bigr)= o(n).\end{aligned}\ ] ] similarly , ( [ propositionboundeigen ] ) , ( [ sigma1sigmainverse2 ] ) and ( [ ing0504b ] ) imply @xmath264 moreover , by ( [ propositionboundeigen ] ) , ( [ diffsigma1sigmainverse ] ) and ( [ ing0504b ] ) , one gets @xmath265 in a similar way , it can be shown that @xmath266 and @xmath267 \label{ing0505e } \\[-8pt ] \nonumber & & \quad{}+ \sup_{\bolds\theta\in\theta } \operatorname{var } \biggl(\frac{\partial ^3}{\partial\theta_1\,\partial\theta_2\,\partial\theta_3 } h(\bolds \theta ) \biggr)=o\bigl(n^{\delta}\bigr).\end{aligned}\ ] ] consequently , ( [ lemmah0result ] ) follows from ( [ ing0504a])([ing0505e ] ) , and hence ( [ lemmatheta1consistency ] ) holds true . next , we prove ( [ lemmatheta2consistency ] ) , which in turn is implied by the property that for any @xmath268 , there exists an @xmath269 such that @xmath270 as @xmath42 , where @xmath271
. let @xmath272 .
since @xmath169 , by ( [ lemmaconvergea.s.0 ] ) , we have @xmath273 therefore ( [ theta2consistent ] ) is given by @xmath274 by ( [ lemmauniformbound ] ) with @xmath254 and @xmath275 , we obtain @xmath276 and @xmath277 \label{ing0505f } \\[-8pt ] \nonumber & & \qquad{}+ \operatorname{var } \biggl(\frac{\partial^2}{\partial\theta_1\,\partial\theta _ 2 } h(\bolds\theta ) \biggr ) + \operatorname{var } \biggl(\frac{\partial^2}{\partial\theta_1\,\partial\theta_3 } h(\bolds \theta ) \biggr ) + \operatorname{var } \biggl ( \frac{\partial^2}{\partial\theta_2\,\partial\theta_3 } h(\bolds\theta ) \biggr ) \\ & & \qquad{}+\operatorname{var } \biggl(\frac{\partial^3}{\partial\theta_1\,\partial\theta _ 2\,\partial\theta_3 } h(\bolds\theta ) \biggr ) \biggr\},\nonumber \ ] ] for some constant @xmath241 . in addition
, it follows from ( [ propositionboundeigen ] ) , ( [ sigma1sigmainverse2 ] ) and ( [ ing0504b ] ) that @xmath278 \label{pfthm01eq01 } \\[-8pt ] \nonumber & & \qquad { } \times\bigl(\bolds\sigma(\bolds\theta_b ) \bigl(\bolds\sigma ( \bolds\theta_b)-\bolds\sigma(\bolds\theta)\bigr ) + \bigl(\bolds\sigma ( \bolds\theta_b)-\bolds\sigma(\bolds\theta)\bigr)\bolds\sigma(\bolds \theta ) \bigr ) \bigr ) \bigr ) \\ & & \quad= o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-1}(\bolds\theta ) \bigl(\bolds\sigma^{-1}(\bolds\theta)- \bolds\sigma^{-1}(\bolds\theta_b)\bigr)\bolds\sigma(\bolds \theta_b ) \bigl(\bolds\sigma(\bolds\theta_b)-\bolds \sigma(\bolds\theta)\bigr ) \bigr ) \bigr)\nonumber \\ & & \qquad { } + o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-1}(\bolds\theta ) \bigl(\bolds\sigma^{-1}(\bolds\theta)- \bolds\sigma^{-1}(\bolds\theta_b)\bigr ) \bigl(\bolds\sigma ( \bolds\theta_b)-\bolds\sigma(\bolds\theta)\bigr)\bolds\sigma(\bolds \theta ) \bigr ) \bigr)\nonumber \\ & & \qquad { } + o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bigl(\bolds \sigma^{-1}(\bolds\theta)-\bolds\sigma^{-1}(\bolds \theta_b)\bigr)\bolds\sigma^{-1}(\bolds\theta_b ) \bolds\sigma(\bolds\theta_b ) \bigl(\bolds\sigma(\bolds \theta_b)-\bolds\sigma(\bolds\theta)\bigr ) \bigr ) \bigr)\nonumber \\ & & \qquad{}+o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bigl(\bolds \sigma^{-1}(\bolds\theta)-\bolds\sigma^{-1}(\bolds \theta_b)\bigr)\bolds\sigma^{-1}(\bolds\theta_b ) \bigl(\bolds\sigma(\bolds\theta_b)-\bolds\sigma(\bolds\theta)\bigr ) \bolds\sigma(\bolds\theta ) \bigr ) \bigr)\nonumber \\ & & \quad= o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-2}(\bolds\theta ) \bigl(\bolds\sigma(\bolds\theta_b)-
\bigr ) \bigr)\nonumber \\ & & \qquad { } + o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-1}(\bolds\theta ) \bigl(\bolds\sigma(\bolds\theta_b)- \bolds\sigma(\bolds\theta)\bigr ) \bolds\sigma^{-1}(\bolds \theta_b ) \bigl(\bolds\sigma(\bolds\theta_b)-\bolds
\sigma(\bolds\theta)\bigr ) \bigr ) \bigr)\nonumber \\ & & \qquad{}+o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-1}(\bolds\theta ) \bigl(\bolds\sigma(\bolds\theta_b)- \bolds\sigma(\bolds\theta)\bigr ) \bolds\sigma^{-1}(\bolds
\theta_b ) \bigl(\bolds\sigma(\bolds\theta_b)-\bolds
\sigma(\bolds\theta)\bigr ) \bigr ) \bigr)\nonumber \\ & & \qquad { } + o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-2}(\bolds\theta_b ) \bigl(\bolds\sigma(\bolds \theta_b)-\bolds\sigma(\bolds\theta)\bigr)^2 \bigr ) \bigr)\nonumber \\ & & \quad= o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bolds \sigma^{-2}(\bolds\theta ) \bigl(\bolds\sigma_\eta(\bolds \theta_b)-\bolds\sigma_\eta(\bolds\theta ) \bigr)^2 \bigr ) \bigr)\nonumber \\ & & \quad= o\bigl(n^{(1+\delta)/2}\bigr).\nonumber\end{aligned}\ ] ] combining ( [ ing0505f ] ) and ( [ pfthm01eq01 ] ) , with ( [ ing0505b])([ing0505e ] ) , yields ( [ lemmaconvergea.s.5 ] ) , and hence ( [ lemmatheta2consistency ] ) is established . finally , we prove ( [ lemmatheta3consistency ] ) .
it suffices to show that for any @xmath279 , there exist @xmath280 such that @xmath281 as @xmath42 , where @xmath282 and @xmath283 .
let @xmath284 .
since @xmath168 , by ( [ lemmaconvergea.s.0 ] ) , we have @xmath285 therefore , it suffices for ( [ theta3consistent ] ) to show that @xmath286 by ( [ lemmauniformbound ] ) with @xmath254 and @xmath287 , we obtain @xmath288 and @xmath289 \label{ing0505 g } \\[-8pt ] \nonumber & & \qquad{}+\operatorname{var } \biggl(\frac{\partial^2}{\partial\theta_1\,\partial\theta _ 2 } \bigl(h(\bolds\theta)-h(\bolds \theta_c ) \bigr ) \biggr ) + \operatorname{var } \biggl(\frac{\partial^2}{\partial \theta_1\,\partial\theta_3}h ( \bolds\theta ) \biggr ) + \operatorname{var } \biggl(\frac{\partial^2}{\partial \theta_2\,\partial\theta_3}h(\bolds\theta ) \biggr ) \\ \nonumber & & \qquad { } + \operatorname{var } \biggl(\frac{\partial^3}{\partial\theta_1\,\partial\theta _ 2\,\partial\theta_3}h(\bolds\theta ) \biggr ) \biggr\},\end{aligned}\ ] ] for some constant @xmath241 . in view of ( [ ing0505 g ] ) , ( [ ing0505c ] ) and ( [ ing0505e ] ) , ( [ lemmaconsistencyhsmallo ] ) is guaranteed by @xmath290 \label{lemmaconsistencyhderivatetheta12 } \sup_{\bolds\theta\in\theta } \operatorname{var } \biggl(\frac{\partial ^2}{\partial\theta_1\,\partial\theta_2 } \bigl(h ( \bolds\theta)-h(\bolds\theta_c )
\bigr ) \biggr ) & = & o \bigl(n^\delta\bigr).\end{aligned}\ ] ] in what follows , we only focus on the proof of ( [ lemmaconsistencyhderivatetheta1 ] ) since the proofs of ( [ lemmaconsistencyhderivatetheta2 ] ) and ( [ lemmaconsistencyhderivatetheta12 ] ) are similar .
note first that by an argument similar to that used to prove ( [ pfthm01eq01 ] ) , one obtains @xmath291 & & \quad= 2 \sup_{\bolds\theta\in\theta}\operatorname{tr } \biggl ( \biggl(\frac { \partial}{\partial\theta_1 } \bigl ( \bolds\sigma^{-1}(\bolds\theta)-\bolds\sigma^{-1}(\bolds \theta_c ) \bigr ) \bolds\sigma(\bolds\theta_0 ) \biggr)^2 \biggr ) \nonumber \\[-9pt ] \label{pfthm01eq02 } \\[-9pt ] \nonumber & & \quad= o \bigl ( \sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bigl(\bolds \sigma^{-1}(\bolds\theta ) \bigl(\bolds\sigma(\bolds\theta_c)- \bolds\sigma(\bolds\theta)\bigr ) \bigr)^2 \bigr ) \bigr ) \\[-2pt ] & & \quad= o \bigl(\sup_{\bolds\theta\in\theta}\operatorname{tr } \bigl ( \bigl ( \mathbf{t}_n^{-1}(\bolds\theta ) \mathbf{g}_n ( \bolds\theta ) \bigl(\bolds\sigma_{\eta}(\bolds\theta_c)-
\bolds\sigma_{\eta}(\bolds\theta ) \bigr ) \mathbf{g}_n(\bolds \theta ) ' \bigr)^2 \bigr ) \bigr).\nonumber\end{aligned}\ ] ] in addition , ( [ lem03eq01geta0 g ] ) and some algebraic manipulations yield @xmath292 \label{pfthm01eq03 } & & \quad= \biggl(\frac{\theta_2\rho_n}{\theta_{0,3}\rho_{0,n}}\bigl(1-\rho _ { 0,n}^2\bigr ) - \frac{\theta_2}{\theta_3}\bigl(1-\rho_n^2\bigr ) \biggr ) \mathbf{i } + \biggl(1-\frac{\rho_n}{\rho_{0,n } } \biggr ) ( 1-\rho_n \rho_{0,n})\bolds\sigma_\eta(\bolds\theta_c ) \\[-2pt ] & & \qquad { } + \frac{\theta_2}{\theta_{0,3 } } \biggl(1-\frac{\rho_n}{\rho_{0,n } } \biggr ) \bigl ( \mathbf{v}_0\mathbf{e}_1'+ \mathbf{e}_1\mathbf{v}_0'\bigr ) + \theta_2 \biggl(\frac{1}{\theta_{0,3}}-\frac{1}{\theta_3 } \biggr ) \rho_n^2\mathbf{e}_1\mathbf{e}_1 ' , \nonumber\end{aligned}\ ] ] where @xmath293 , and @xmath294 uniformly in @xmath48 . moreover , by ( [ tracet2 ] ) , ( [ lem03eq02vtv])([lem03eq06tracetetateta ] ) and @xmath295 which can be shown using an argument similar to that used to prove ( b.2 ) in the supplementary document ( chang , huang and ing @xcite ) , we have @xmath296 \sup_{\bolds\theta\in\theta } n^{-4(1-\delta ) } \operatorname{tr } \bigl(\bigl ( \mathbf{t}_n^{-1}(\bolds\theta)\bolds\sigma_\eta ( \bolds\theta_c)\bigr)^2 \bigr ) & = & o\bigl(n^\delta \bigr),\nonumber \nonumber \\[-9pt ] \label{ing0505i } \\[-9pt ] \nonumber \sup_{\bolds\theta\in\theta}n^{-2(1-\delta ) } \operatorname{tr } \bigl(\bigl ( \mathbf{t}_n^{-1}(\bolds\theta ) \bigl(\mathbf{v}_0 \mathbf{e}_1'+\mathbf{e}_1 \mathbf{v}_0'\bigr)\bigr)^2 \bigr ) & = & o(1 ) , \\[-2pt ] \sup_{\bolds\theta\in\theta } \operatorname{tr } \bigl(\bigl(\mathbf{t}_n^{-1 } ( \bolds\theta)\mathbf{e}_1\mathbf{e}_1 ' \bigr)^2 \bigr ) & = & o(1).\nonumber\end{aligned}\ ] ] combining ( [ pfthm01eq02])([ing0505i ] ) leads to ( [ lemmaconsistencyhderivatetheta1 ] ) and hence ( [ lemmaconsistencyhsmallo ] ) .
this completes the proof of ( [ lemmatheta3consistency ] ) .
to prove theorem [ thmctlundertruemodel ] , we need two additional lemmas , lemmas [ lemmataylorfirstorder][lemmataylorsecondorder ] , which provide the orders of magnitude of @xmath297 and @xmath298 ; @xmath299 , when the convergence rate of @xmath53 is given . on the contrary , using the orders of the magnitude of @xmath297 and @xmath298 , @xmath299 , one can also derive the convergence rate of @xmath53 ; see ( [ thmtheta1taylor])([thmtheta3taylor ] ) . as a result
, the convergence rate of @xmath53 can be sequentially improved via an initial convergence rate and applying this argument repeatedly .
[ lemmataylorfirstorder ] under the setup of lemma [ lemmamisselectedmodel ] , define for @xmath300 , @xmath301 where @xmath92 is given by ( [ loglikefuntrue ] ) .
let @xmath302 be an estimate of @xmath303 with @xmath304 , @xmath305 and @xmath306 for some constants @xmath307 $ ] , @xmath308 $ ] and @xmath309 $ ] ; @xmath34 . then for any @xmath34
, @xmath310 \label{thmg1order } \\[-8pt ] \nonumber & & { } + o_p\bigl(n^\xi\bigr ) + o(1 ) , \\ g_2\bigl((\hat\theta_1,\theta_{0,2},\hat
\theta_3)'\bigr ) & = & o_p\bigl(n^{(1+\delta)/4 } \bigr ) + o_p\bigl(n^{(1+\delta)/2-r_1}\bigr)+ o_p \bigl(n^{\delta - r_3}\bigr ) \nonumber \\[-8pt ] \label{thmg2order } \\[-8pt ] \nonumber & & { } +
o_p\bigl(n^\xi\bigr ) + o(1),\end{aligned}\ ] ] and for @xmath124 , @xmath311 \label{thmg3order } \\[-8pt ] \nonumber & & { } + o_p\bigl(n^\xi\bigr ) + o(1).\end{aligned}\ ] ] in addition , for any @xmath34 , if @xmath312 and @xmath313 , @xmath314 if @xmath315 , @xmath316 and @xmath317 , @xmath318 furthermore , for any @xmath124 , if @xmath319 , @xmath320 and @xmath321 , @xmath322 [ lemmataylorsecondorder ] under the setup of lemma [ lemmamisselectedmodel ] , let @xmath323 let @xmath302 be an estimate of @xmath324 .
suppose that @xmath325 .
then for @xmath34 , there exists a constant @xmath326 satisfying @xmath327 such that @xmath328 in addition , suppose that @xmath325 and @xmath329 , then for @xmath34 , there exists a constant @xmath330 satisfying @xmath331 such that @xmath332 furthermore , suppose that @xmath333 , then for @xmath124 , there exists a constant @xmath334 satisfying @xmath335 such that @xmath336 we shall prove ( [ thmtheta1consistency])([thmtheta3consistency ] ) by iteratively applying ( [ thmg1order])([thmg33order ] ) . for the first iteration
, we show that @xmath337 \label{thmtheta2iterat1 } \\[-8pt ] \nonumber \hat\theta_2 - \theta_{0,2 } & = & o_p \bigl(n^{-(1-\delta)/2}\bigr)\qquad \mbox{if } \delta\in[1/3,1 ) , \\
n^{\delta/2 } ( \hat{\theta}_3-\theta_{0,3 } ) & \displaystyle\mathop { \rightarrow}^{d } & n(0,2\theta_{0,3})\qquad \mbox{if } \delta \in(0,1/2 ) , \nonumber \\[-8pt ] \label{thmtheta3iterate1 } \\[-8pt ] \nonumber \hat{\theta}_3-\theta_{0,3 } & = & o_p \bigl(n^{-(1-\delta)/2}\bigr)\qquad \mbox{if } \delta\in[1/2,1).\end{aligned}\ ] ] proof of ( [ thmtheta1iterate1 ] ) taking the taylor expansion of @xmath338 at @xmath339 yields @xmath340 where @xmath341 satisfies @xmath327 .
therefore , for ( [ thmtheta1iterate1 ] ) to hold , it suffices to show that @xmath342 g_{11}\bigl(\hat{\bolds\theta_a^*}\bigr ) & = & \frac{n}{\theta_{0,1}^2 } + o_p(n),\end{aligned}\ ] ] where the first equation follows from ( [ lemmatheta2consistency ] ) and ( [ thmg1order ] ) with @xmath343 , and the second one is given by ( [ lemmatheta1consistency ] ) and ( [ thmg11order ] ) .
proof of ( [ thmtheta2iterat1 ] ) let @xmath344 .
taking the taylor expansion of @xmath345 at @xmath346 yields @xmath347 where @xmath348 satisfies @xmath349 .
therefore , for ( [ thmtheta2iterat1 ] ) to hold , it suffices to show that @xmath350 where the first two equations follow from ( [ thmg2order ] ) with @xmath351 , ( [ thmg2asymptoticnormality ] ) and ( [ thmtheta1iterate1 ] ) , and the last one is ensured by ( [ lemmatheta2consistency ] ) , ( [ thmg22order ] ) and ( [ thmtheta1iterate1 ] ) . proof of ( [ thmtheta3iterate1 ] ) taking the taylor expansion of @xmath352 at @xmath353 yields @xmath354 where @xmath355 satisfies @xmath335 .
therefore , for ( [ thmtheta3iterate1 ] ) to hold , it suffices to show that @xmath356 where the first two equations follow from ( [ thmg3order ] ) with @xmath357 , ( [ thmg3asymptoticnormality ] ) , ( [ thmtheta1iterate1 ] ) and ( [ thmtheta2iterat1 ] ) , and the last one is ensured by ( [ thmg33order ] ) .
thus , ( [ thmtheta3iterate1 ] ) is established .
for the second iteration , we show that @xmath358 \label{thmtheta1iterate2 } \\[-8pt ] \nonumber \hat{\theta}_1-\theta_{0,1 } & = & o_p \bigl(n^{-(1-\delta)}\bigr)\qquad \mbox{if } \delta\in[1/2,1 ) , \\
n^{(1+\delta)/4 } ( \hat{\theta}_2-\theta_{0,2 } ) & \displaystyle\mathop { \rightarrow}^{d } & n \bigl(0 , 2^{5/2}\theta_{0,1}^{1/2 } \theta_{0,2}^{3/2 } \bigr)\qquad \mbox{if } \delta\in[0,3/5 ) , \nonumber \\[-8pt ] \label{thmtheta2iterat2 } \\[-8pt ] \nonumber \hat\theta_2 - \theta_{0,2 } & = & o_p \bigl(n^{-(1-\delta)}\bigr)\qquad \mbox{if } \delta\in[3/5,1 ) , \\
n^{\delta/2 } ( \hat{\theta}_3-\theta_{0,3 } ) & \displaystyle\mathop { \rightarrow}^{d } & n(0,2\theta_{0,3})\qquad \mbox{if } \delta \in(0,2/3 ) , \nonumber \\[-8pt ] \label{thmtheta3iterate2 } \\[-8pt ] \nonumber \hat{\theta}_3-\theta_{0,3 } & = & o_p \bigl(n^{-(1-\delta)}\bigr)\qquad \mbox{if } \delta\in[2/3,1).\end{aligned}\ ] ] by ( [ thmg1order ] ) with @xmath359 , ( [ thmg1asymptoticnormality ] ) and ( [ thmtheta2iterat1 ] ) , we have @xmath360 the above two equations , ( [ thmg11order ] ) and ( [ thmtheta1taylor ] ) give ( [ thmtheta1iterate2 ] ) . by ( [ thmg2order ] ) with @xmath361 and @xmath362 , ( [ thmg2asymptoticnormality ] ) , ( [ thmtheta3iterate1 ] ) and ( [ thmtheta1iterate2 ] )
, we have @xmath363 combining these two equations together with ( [ thmg22order ] ) and ( [ thmtheta2taylor ] ) yields ( [ thmtheta2iterat2 ] ) . by ( [ thmg3order ] ) with @xmath364 , ( [ thmg3asymptoticnormality ] ) , ( [ thmtheta1iterate2 ] ) and ( [ thmtheta2iterat2 ] )
, we have @xmath365 which , together with ( [ thmg33order ] ) and ( [ thmtheta3taylor ] ) , lead immediately to ( [ thmtheta3iterate2 ] ) . following the same argument as in the second iteration , we can recursively
show that for each @xmath366 @xmath367 , \\
n^{\delta/2 } ( \hat{\theta}_3-\theta_{0,3 } ) & \displaystyle\mathop { \rightarrow}^{d } & n(0,2\theta_{0,3})\qquad \mbox{if } \delta\in \bigl(0,i/(i+1)\bigr ) , \\ \hat{\theta}_3-\theta_{0,3 } & = & o_p \bigl(n^{-i(1-\delta)/2}\bigr)\qquad \mbox{if } \delta\in\bigl[i/(i+1),1\bigr).\end{aligned}\ ] ] thus ( [ thmtheta1consistency])([thmtheta3consistency ] ) are proved .
we divide the proof into three parts corresponding to @xmath368 , @xmath369 and @xmath370 .
first , we consider @xmath371 .
we further divide the proof into six subparts with respect to @xmath149 in terms of a partition of @xmath372 , corresponding to @xmath373 , @xmath374 , @xmath375 , @xmath376 , @xmath377 and @xmath378 .
we shall prove each of the following six subparts separately : for @xmath378 , @xmath379 for @xmath377 , @xmath380 for @xmath376 , @xmath381 for @xmath382 , @xmath383 for @xmath384 , @xmath385 for @xmath373 , @xmath386 and if in addition @xmath387 , then @xmath388 proof of ( a1 ) applying ( [ thmg1order ] ) with @xmath389 and @xmath378 , we have @xmath390 according to ( [ lemmatheta1consistency ] ) and ( [ thmg11order ] ) , we have @xmath391 the desired conclusion ( a1 ) now follows from plugging ( [ thm2g1order ] ) and ( [ thm2g11order ] ) into ( [ thmtheta1taylor ] )
. proof of ( a2 ) applying ( [ thmg1order ] ) with @xmath389 and @xmath377 , we have @xmath392 combining this with ( [ thmtheta1taylor ] ) and ( [ thm2g11order ] ) gives @xmath393 applying ( [ thmg2order ] ) with @xmath351 , @xmath394 and @xmath377 , we obtain @xmath395 from ( [ lemmatheta2consistency ] ) and ( [ thmg22order ] ) , we have @xmath396 combining this with ( [ thmtheta2taylor ] ) and ( [ thm2g2order ] ) leads to ( [ thm2parti-2 - 2 ] ) . in addition , applying ( [ thmg1order ] ) with @xmath397 and @xmath377 , we have @xmath398 this together with ( [ thmtheta1taylor ] ) and ( [ thm2g11order ] ) gives ( [ thm2parti-2 - 1 ] ) .
proof of ( a3 ) following the same arguments as the one used in the proof of ( [ thm2parti-2 - 2 ] ) leads to ( [ thm2parti-3 - 2 ] ) . applying ( [ thmg1asymptoticnormality ] ) with @xmath351 , @xmath399 , @xmath400 and @xmath376 , we have @xmath401 this together with ( [ thmtheta1taylor ] ) and ( [ thm2g11order ] ) gives ( [ thm2parti-3 - 1 ] ) .
proof of ( a4 ) applying ( [ thmg2asymptoticnormality ] ) with @xmath402 and @xmath382 , we have @xmath403 this , ( [ thmtheta2taylor ] ) and ( [ thm2g22order ] ) imply ( [ thm2parti-4 - 2 ] ) .
moreover , ( [ thm2parti-4 - 1 ] ) can be shown by an argument similar to that used to prove ( [ thm2parti-3 - 1 ] ) .
proof of ( a5 ) the proofs of ( [ thm2parti-5 - 1 ] ) and ( [ thm2parti-5 - 2 ] ) are similar to those of ( [ thm2parti-4 - 1 ] ) and ( [ thm2parti-4 - 2 ] ) , respectively . applying ( [ thmg3order ] ) with @xmath357 , @xmath400 and @xmath376
, we have @xmath404 from ( [ lemmatheta1consistency])([lemmatheta3consistency ] ) and ( [ thmg33order ] ) , we obtain @xmath405 combining this with ( [ thmtheta3taylor ] ) and ( [ thm2g3order ] ) leads to ( [ thm2parti-5 - 3 ] ) .
proof of ( a6 ) equations ( [ thm2parti-6 - 1 ] ) and ( [ thm2parti-6 - 2 ] ) can be proved in a way similar to the proofs of ( [ thm2parti-4 - 1 ] ) and ( [ thm2parti-4 - 2 ] ) . applying ( [ thmg3asymptoticnormality ] ) with @xmath357 , @xmath400 and @xmath406 , we have @xmath407 this together with ( [ thmtheta3taylor ] ) and ( [ thm2g33order ] ) gives ( [ thm2parti-6 - 3 ] ) .
second , we consider @xmath369 .
following an argument similar to that used in the first part , we obtain for @xmath378 , @xmath408 for @xmath377 , @xmath409 for @xmath410 , @xmath411 for @xmath412 , @xmath413 for @xmath414 , @xmath415 for @xmath373 , @xmath416 third , for @xmath370 , one can similarly show that for @xmath378 , @xmath408 for @xmath417 , @xmath409 for @xmath418 , @xmath419 for @xmath376 , @xmath420 for @xmath414 , @xmath415 for @xmath373 , @xmath421
thus the proof of the theorem is complete . to prove corollaries [ coropolymle0 ] and [ coropolymle ] , the following lemma , which provides the order of magnitude of @xmath78 defined in ( [ eqrisk ] ) ,
is needed .
[ lemmapolyequations ] under the setup of lemma [ lem02ttraces ] , let @xmath422 and @xmath423 .
then for any @xmath34 , the following equations hold uniformly in @xmath48 : @xmath424 we first prove corollary [ coropolymle0 ] .
note that @xmath425 \label{cor1misselectorder } \\[-8pt ] \nonumber & & \quad= \beta_{0,1}^2\mathbf{x}'\bolds \sigma^{-1}(\bolds\theta)\mathbf{x } -\beta_{0,1}^2 \mathbf{x}'\bolds\sigma^{-1}(\bolds\theta)\mathbf{1}\bigl ( \mathbf{1}'\bolds\sigma^{-1}(\bolds\theta)\mathbf{1 } \bigr)^{-1 } \mathbf{1}'\bolds\sigma^{-1}(\bolds \theta)\mathbf{x } \\ & & \quad= \frac{\beta_{0,1}^2\theta_3 ^ 2}{24\theta_2}n^\delta+ o\bigl(n^\delta \bigr),\nonumber\end{aligned}\ ] ] uniformly in @xmath48 , where @xmath426 and the last equality is obtained from ( [ polyresult1])([polyresult3 ] ) .
therefore , ( [ cor1xi = theta ] ) holds . with the help of ( [ cor1xi = theta ] ) ,
( [ cor1theta1consistency ] ) and ( [ cor1theta2consistency ] ) follow directly from theorem [ thmctlunderincorrectmodel ] .
second , we prove corollary [ coropolymle ] . by ( [ loglikedecompose ] ) , ( [ lemmamis2 ] ) and ( [ cor1misselectorder ] ) , we have @xmath427 uniformly in @xmath48 , noting that @xmath428 . therefore , by ( [ logdetsigma ] ) and ( [ sigma1sigmainverse3 ] ) , @xmath429 \label{loglikedecomposepoly } \\[-8pt ] \nonumber & = & n\log(2\pi ) - \frac{1-\delta}{2}\log{n } + \biggl(\log\theta_1 + \frac{\theta_{0,1}}{\theta_1 } \biggr)n \\ & & { } + \biggl(\frac{2\theta_2}{\theta_1 } \biggr)^{1/2 } \biggl(1-\frac { \theta_{0,1}}{2\theta_1 } + \frac{\theta_{0,2}}{2\theta_2 } \biggr)n^{(1+\delta)/2}\nonumber \\ & & { } - \biggl\{\frac{\theta_2}{\theta_1}+\theta_3 \biggl(1-\frac{\theta _ { 0,2}}{\theta_2 } \biggr ) + \frac{\theta_{0,2}\theta_{0,3}+\theta_{0,2}\theta _ { 0,3}^*}{2\theta_2}\nonumber \\ & & { } -\frac{\theta_{0,2}}{2\theta_2\theta_{0,3}^ * } \bigl(\theta_3-\theta _ { 0,3}^ * \bigr)^2 \biggr\ } n^\delta+h(\bolds\theta)+o_p \bigl(n^\delta\bigr)+o_p(1),\nonumber\end{aligned}\ ] ] uniformly in @xmath48 , where @xmath430 .
it follows from ( [ loglikedecomposepoly ] ) and the same argument as in the proof of ( [ theta3consistent ] ) that for any @xmath279 , there exist @xmath431 such that @xmath432 as @xmath42 , where @xmath433 . thus ( [ cor1theta3inconsistent ] ) is established , and hence the proof is complete .
we first prove ( [ cor1xi=1+theta2 ] ) .
let @xmath434 . by an argument similar to that used to prove ( [ lemmamis2 ] ) , it can be shown that @xmath435 this , together with ( [ polyresult3 ] ) and ( [ sigma1sigmainverse ] ) , gives @xmath436 uniformly in @xmath48 , where @xmath437 .
in addition , an argument similar to that used to prove ( [ lemmaconvergea.s.5 ] ) yields @xmath438 hence ( [ cor1xi=1+theta2 ] ) follows . in view of
( [ cor1xi=1+theta2 ] ) and theorem [ thmctlunderincorrectmodel ] , we obtain ( [ cor3theta1 ] ) .
thus , the proof of corollary [ coroexp0 ] is complete . to prove ( [ cor3theta2inconsistent ] ) , note first that by the same line of reasoning as in ( [ loglikedecomposepoly ] ) , one gets @xmath439 uniformly in @xmath48 , where @xmath440 and @xmath441 . moreover , using arguments similar to those used in the proofs of ( [ theta2consistent ] ) and ( [ theta3consistent ] ) , respectively , one can show that for any @xmath268 , there exists an @xmath269 such that @xmath442 and for any @xmath279 , there exist @xmath280 such that @xmath443 where @xmath444 and @xmath445 . combining ( [ loglikedecomposeexp])([ing0507b ] ) yields ( [ cor3theta2inconsistent ] ) and ( [ cor3theta3inconsistent ] ) .
this completes the proof of corollary [ coroexp ] .
the authors would like to thank the associate editor and an anonymous reviewer for their insightful and constructive comments , which greatly improve the presentation of this paper .
the research of chih - hao chang and hsin - cheng huang was supported by ministry of science and technology of taiwan under grants most 103 - 2118-m-390 - 005-my2 and most 100 - 2628-m-001 - 004-my3 , respectively .
the research of ching - kang ing was supported by academia sinica investigator award .
chang , c .- h .
, huang , h .- c . and ing , c .- k .
( 2014 ) , `` asymptotic theory of generalized information criterion for geostatistical regression model selection , '' _ the annals of statistics _ , * 42 * , 2441 - 2468 . | we consider a stochastic process model with time trend and measurement error .
we establish consistency and derive the limiting distributions of the maximum likelihood ( ml ) estimators of the covariance function parameters under a general asymptotic framework , including both the fixed domain and the increasing domain frameworks , even when the time trend model is misspecified or its complexity increases with the sample size .
in particular , the convergence rates of the ml estimators are thoroughly characterized in terms of the growing rate of the domain and the degree of model misspecification / complexity .
./style / arxiv - general.cfg , |
in just the united states each year approximately 795.000 people are affected by stroke @xcite . in france strokes
are around 150.000 each year @xcite .
similar trends are shown all around the industrialized world .
a stroke can cause disabilities , such as paralysis , speech difficulties , and emotional problems .
these disabilities can significantly affect the quality of daily life for survivors .
several researchers on the field have shown that important variables in relearning motor skills and in changing the underlying neural architecture after a stroke are the quantity , duration , and intensity of training sessions ( for more information see e.g. , * ? ? ?
in this paper we present a mixed reality approach for upper limb rehabilitation focused to increase quantity , duration , and quality of training sessions .
in particular the approach answers the following challenges : ( i ) increase motivation of patients by making the training a personalized experience , for example adjusting exercises difficulty according to patients cognitive and physical capabilities ; ( ii ) take into account patients impairments by offering intuitive and easy to use interaction modalities that do not require a learning curve and that are adapted to elder people ; ( iii ) make it possible to therapists to track patient s activities and evaluate progress in order to adapt therapeutic strategies ; ( iv ) open opportunities for telemedicine and tele rehabilitation when patients leave the hospital ; ( v ) provide an economically acceptable system by reducing both equipment and management costs .
it is worth to note that the above mentioned issues address two targets : the therapist and the patient .
it is our opinion that it is important to evaluate acceptance of these systems from both , the patient s point of view and the therapists point of view .
otherwise , even if clinical efficiency is demonstrated , developed systems will be used only for academic studies and may not be widely accepted in real rehabilitation centers . for this reason , before testing the system directly with patients we decide to conduct a pilot study in conjunction with a french hospital .
aim of this pilot was understand the potential and benefits of mixed reality from the therapist s point of view .
the pilot involved 3 therapists who played the role of patients .
three sessions , one using conventional rehabilitation , another using an ad hoc developed game on a pc , and another using a mixed reality version of the same game were held . before entering in detail into the pilot study description , in following subsections we will describe in more depth what a stroke is and what are its consequences , and why a virtual approach could be useful for such a rehabilitation . after a stroke
there is a loss of brain functions due to disturbance in the blood supply to the brain .
the affected area of the brain is unable to function , leading to inability to move one or more limbs on one side of the body and perhaps to cognitive problems such aphasia or the so - called neglect effect . because of these impairments stroke sufferers are often unable to independently perform day - to - day activities such as bathing , dressing , and eating .
nearly three - quarters of all strokes occur in people over the age of 65 .
the specific abilities that will be lost or affected by stroke depend on the extent of the brain damage and most importantly where in the brain the stroke occurred .
for example , a stroke in the right hemisphere often causes paralysis in the left side of the body .
this is known as left hemiplegia .
survivors of right - hemisphere strokes may have problems with their spatial and perceptual abilities .
this may cause them to misjudge distances ( leading to a fall ) or be unable to guide their hands to pick up an object , button a shirt or tie their shoes .
they may even be unable to tell right - side up from upside - down when trying to read .
survivors of right - hemisphere strokes may also experience left - sided neglect . stemming from visual field impairments ,
left - sided neglect causes the survivor of a right - hemisphere stroke to forget or ignore objects or people on their left side . on the other hand ,
someone who has had a left - hemisphere stroke may develop aphasia .
aphasia is a catch - all term used to describe a wide range of speech and language problems .
these problems can be highly specific , affecting only one component of the patient s ability to communicate , such as the ability to move their speech - related muscles to talk properly
. the same patient may be completely unimpaired when it comes to writing , reading or understanding speech .
by contrast to survivors of right - hemisphere stroke , patients who have had a left - hemisphere stroke often develop a slow and cautious behavioral style .
they may need frequent instruction and feedback to complete tasks ( information from the * ? ? ?
as we can see from these two examples , the consequences of two stroke accidents could be deeply different .
for this reason each stroke rehabilitation program is personal , designed for a particular patient , and not a generic one . in stroke accidents rehabilitation
involves intensive and continuous training to regain as much function as possible , depending on several factors including the severity of brain lesions and the degree of cerebral plasticity .
following the hypothesis that the theories of motor learning can be applied on motor relearning most rehabilitation techniques are founded on principles of motor learning and skill acquisition established for the healthy nervous system .
acknowledged features are among others : ( i ) the motivation of the participant ; ( ii ) the use of variable practice ( i.e. practice a variety of related tasks ) ; ( iii ) training with high intensity / many repetitions ; ( iv ) and providing feedback @xcite .
in addition , therapy on the lower extremity ( i.e. , legs ) is the primary concern in early inpatient stroke therapy in order to enable mobility of the patient .
recovery of the upper extremity ( i.e. , arms ) has a slower progression and is usually gained through outpatient and home therapy @xcite .
patients with upper extremity paralysis typically regain motion starting from their shoulder . over time
, they may gradually regain motion in the elbow , wrist , and , finally , the hand .
the most important part of stroke rehabilitation is conducted during the first 6 months after the stroke and due to the cost , only 6/8 weeks of rehabilitation is done under the continued direct supervision of an expert ( i.e. , in the hospital ) . because of limitations on therapy patients must do much of the work necessary to recover arm function at home . to summarize , after a stroke the affected area of the brain is unable to function , leading to inability to move one or more limbs on one side of the body and also perhaps to cognitive problems such aphasia or the so - called neglect effect .
as a result of these impairments stroke sufferers are often unable to independently perform day - to - day activities such as bathing , dressing , and eating .
as a side effect they can develop depression or aggressiveness due to the trauma of reduced capabilities . depression and aggressiveness
also imply that stroke survivors may find it difficult to focus on a therapy programme .
in addition , while these programmes attempt to stimulate the patient with a variety of rehabilitation exercises , stroke victims commonly report that traditional rehabilitation tasks can be boring due to their repetitive nature .
we can then say that motivation is an important factor for rehabilitation success .
several researchers have shown , both in animal and human , that important variables in relearning motor skills and in changing the underlying neural architecture are the quantity , duration , and intensity of training sessions .
in particular , research in animal models suggests that with intensive therapy ( repeating individual motions hundreds of times per day ) , animals that experience strokes can recover a significant amount of their lost motor control @xcite .
similarly , recent guidelines for treatment of human patients recommend high - intensity , repetitive motions while keeping patients informed about their progress @xcite .
looking at the effects of different intensities of physical therapy treatment , several authors @xcite have reported significant improvement in activities of daily living as a result of higher intensities of treatment . to experience significant recovery
, stroke patients must perform a substantial number of daily exercises .
unfortunately , typical sessions with therapists include a relatively small number of motions @xcite .
one of the possible solutions is doing rehabilitation at home .
however , while therapists prescribe a home exercise regimen for most patients , a study indicates only 31% of patients actually perform these exercises as recommended @xcite .
thus , home based stroke rehabilitation has the potential to help patients in recovering from a stroke . on the other hand
a main problem arises when choosing which kind of home - based technology can both help and motivate patients to perform therapeutic exercises .
the challenge for post stroke rehabilitation is to create exercises able to decrease monotony of hundreds of repeated motions . in order to overcome the above mentioned problems different kinds of non traditional therapies
have been proposed .
for example , the possibility of using virtual rehabilitation has been the subject of experiments by several authors ( for example @xcite ) .
although most of studies on this topic are linked to the study of virtual reality environments recent studies have focused on the use of videogames and consoles for rehabilitation ( such as @xcite .
results of these studies can be summerized as follows : 1 .
_ personalization _ : virtual rehabilitation technology creates an environment in which the intensity of feedback and training can be systematically manipulated and enhanced in order to create the most appropriate , individualized motor learning paradigm @xcite .
rehabilitation using games can take advantage of the use of adaptation in order to create ad hoc personalized games
_ interactivity _ : an advantage present in all forms of virtual rehabilitation is the use of interactivity .
for example it has been suggested that integrating gaming features in virtual environments for rehabilitation could enhance user motivation .
virtual rehabilitation exercises can be made to be engaging , such that the patient feels immersed in the simulated world .
this is extremely important in terms of the patient motivation @xcite , which , in turn , is key to recovery . a person who enjoys what he is doing spends more time developing her skills in a given activity .
feedback _ : interactive feedback can contribute to motivation .
for example , by providing visual and auditory rewards , such as displaying gratifying messages in real time , patients are motivated to exercise @xcite 4 . _
tracking _ : the evolution of the patient s performance can be easily stored , accessed and developed without the patient s or therapist s input .
in addition , the internet can be used for data transfer , allowing a therapist to remotely monitor progress and to modify the patient s therapy program @xcite .
_ telerehabilitation _ : virtual rehabilitation can stimulate the patient using a variety of rehabilitation exercises at a low cost .
this means that rehabilitation costs can be contained if the technology used for virtual rehabilitation ( consoles and games ) is easily accessible .
in addition lower cost personal equipment ( for example pc - based ) will eventually allow rehabilitation stations to be placed in locations other than the rehabilitation center , such as a patient s home .
on the other hand , virtual rehabilitation raises important challenges that may limit its widespread adoption such as : * clinical acceptance , which relies on proved medical efficacy * therapist s attitude towards the technology ( e.g. , the therapist agrees that technology is able to replace therapists and the like ) * patient s attitude towards the technology ( e.g. , the patient may not consider a game to be real rehabilitation ) .
* challenges linked to the kind of technology used ( challenges differ from virtual reality to consoles ) . in order to overcome the last three challenges we propose an approach based on mixed reality ( see section [ sytems ] of this paper for the description of such a system )
. however , while evaluating usability for systems has become a standard @xcite attitude is a more complex element which requires an in depth assessment .
for this reason we propose to use an approach meant to evaluate acceptance of a tool based on shackel .
shackel @xcite defines a model where product acceptance is the highest concept .
the user has to make a trade off between utility , the match between user s needs and functionality , usability , ability to use functionality in practice and likeability , affective evaluation vs costs ( see fig.[fig : shackel ] ) it is worth noting that in a system for stroke rehabilitation we are dealing with two kind of acceptances .
the first one is patient s acceptance of the system . in order to be accepted by our users the system has to be considered usable , useful and likeable .
the second one is therapist s acceptance .
in addition to the above mentioned elements the system has also to show adequate cost for the therapist .
while in this paper we address directly the therapist acceptance and the patient s perspective only as a simulate one ( see section [ protocol ] ) it s our opinion that both perspectives are very important .
this section reviews existing approaches for post stroke rehabilitation .
the presentation of works is not exhaustive and the focus has been made on three types of rehabilitation systems : robot based system , virtual reality based systems and mixed reality based systems . besides , only early stages of rehabilitation have been considered which excluded for instance system dedicated to the rehabilitation of fingers such as @xcite . to avoid ambiguities , hereafter definition of each type of system
is given .
[ [ definition - of - virtual - reality ] ] definition of virtual reality + + + + + + + + + + + + + + + + + + + + + + + + + + + + + virtual reality system is commonly defined as an environment that can simulate real world situations and allows users to interact within this simulated environment using different types of devices such as head mounted device ( hmd ) and gloves .
[ [ definition - of - mixed - reality ] ] definition of mixed reality + + + + + + + + + + + + + + + + + + + + + + + + + + + mixed reality system merges real and virtual worlds to create a new context of interaction where both physical and digital objects co - exist and interact consistently in real time .
this contrasts with virtual reality where the interaction is directional from the physical world where the user is situated towards the virtual environment where the interaction holds . within mixed reality system ,
the interaction holds among objects of both virtual and physical environments .
a classical example of a mixed reality game is the mr pong @xcite where players play pong game with a virtual ball colliding with physical objects on a table .
[ [ definition - of - robot - assisted - rehabilitation ] ] definition of robot assisted rehabilitation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + robot based rehabilitation uses a training robot .
the robot complements or induces patient s movement and provides feedback .
robot systems are often coupled with a computer program used to create virtual context for actions and delivering a visual feedback .
an evaluation framework is used to analyze usability of each system from therapist , patient and financier point of view . from the therapist s point of view ,
we are interested by the following evaluation criteria : [ [ therapist - intervention ] ] therapist intervention + + + + + + + + + + + + + + + + + + + + + + this criterion informs whether the therapist can intervene during the rehabilitation sessions .
in fact , we have noticed from observations and preliminary interviews with therapists the importance of assistance and guidance delivered by the therapist .
the therapist has to intervene physically during the rehabilitation sessions to support the patient and to prevent him from developing incorrect compensating gestures such as chest balancing .
possible values of this criterion are : yes , if the therapist can intervene and no , otherwise .
[ [ changes - on - therapist - habits ] ] changes on therapist habits + + + + + + + + + + + + + + + + + + + + + + + + + + + this criterion informs about changes induced by introduction of the rehabilitation system to therapist working habits .
the introduction of a new rehabilitation system may induce changes in the way therapists were conducting therapeutic sessions .
the goal of this criterion is to evaluate the amount of the induced changes .
possible values are : negligible , if the therapists has not to change , her usual way of work ; moderate , when the therapist has to change some of her working habits and important , when the therapist has to change her way of conducting therapies .
[ [ system - setup ] ] system setup + + + + + + + + + + + + this criterion informs about the setup phase of the system .
often , the rehabilitation system needs some setup phase in order to get devices installed and configured correctly before starting rehabilitation sessions .
this criterion informs whether the therapist alone can perform this setup phase or a specialized assistant is required to setup the system .
possible values of this criterion are : therapist , if the therapist could perform the setup phase and assistant , if intervention of a specialized technician is required .
[ [ location ] ] location + + + + + + + + this criterion concerns the place where the rehabilitation sessions can be performed .
in fact , for some systems such as robot based systems and some virtual reality simulators a special room has to be dedicated to perform rehabilitation sessions .
other lightweight systems do not require a fixed infrastructure and can be used almost anywhere . from the patient s point of view
, we are interested in two evaluation criteria that are : [ [ eye - hand - focus ] ] eye - hand focus + + + + + + + + + + + + + + observations from therapeutic sessions and interviews with therapists have revealed an important point related to the interaction of the patient with the system .
in fact , post stroke patients during early stages may be reluctant to any system that requires an important cognitive effort . ideally , the attention of the patient has to be attracted and focused on a single point .
for instance , when given a mouse patients look at their hand when trying to move and not to the screen .
consequently , they find it difficult to follow two actions at the same time : the hand that is moving and the game displayed on the screen .
this criterion evaluates whether the eye and hand of the patient are attracted to the same place or to different places .
possible values for this criterion are : same place , when the hand and eyes of the patient are directed to the same place and different places , when the hand and eyes of the patient are attracted to different spaces .
[ [ invasiveness - of - the - system ] ] invasiveness of the system + + + + + + + + + + + + + + + + + + + + + + + + + + this criterion informs about the invasiveness of the system from the patient s perspective .
in fact , systems may require the patient to wear special devices to track their movement or to deliverer feedback .
these devices are more or less convenient to wear and to tolerate by patients especially the ones that suffer from impairment and muscular spasticity
. possible values for this criterion are : convenient , if the system is considered as convenient to use ; invasive , if the system is considered as interfering with the patient . for a rehabilitation system to be used in rehabilitation centers ,
it is necessary to demonstrate its medical efficiency and also to show that the system is economically sustainable .
the economical aspect of the rehabilitation system are evaluated through two criteria that are : [ [ unitary - cost - of - the - system ] ] unitary cost of the system + + + + + + + + + + + + + + + + + + + + + + + + + + this criterion evaluates the unitary cost of the system in kilo euros ( ke ) .
the objective is not to deliver a precise price , but evaluate an order of magnitude of each system cost .
thus , possible values of this criterion are : less than 1 ke per unit , 1 - 5 ke per unit , 5 - 10 ke per unit , and more than 10 ke per unit . [
[ extra - required - resources ] ] extra required resources + + + + + + + + + + + + + + + + + + + + + + + + while the previous criterion estimates the unitary cost of the system , this criterion evaluates all extra resources that must provided for running the system such as recruiting specialized personnel , making available a special and so on .
robot based approach for rehabilitation has been used since the late 90s to assist patients .
the manus robot developed at the mit @xcite is an exemplar system ( figure [ fig : manus ] ) . in this system
the patient is in front of the robot and her shoulder is strapped to a chair . the patient s impaired arm is strapped to a wrist carrier and attached to the manipulandum .
a video screen is above the training table to create a virtual context for movements and provide visual feedbacks to the patients .
several clinical trials have been conducted to evaluate the efficiency of robots based rehabilitation when compared to classical therapies .
for instance , in @xcite it has been demonstrated that robot - assisted therapy improved outcomes for long period of training when compared with usual care .
[ tab : theraper ] .therapist perspective [ cols="<,<,<,<,<",options="header " , ] always in their patient s role therapist were asked to answer the following question : `` in general , which therapy did you preferred to do ?
( classify exercises in order of preference ) '' .
this means that we asked them to order the therapies as they thought the patients will do .
all the three gave the following order : 1)mixed reality game , 2)pc game , 3)physical therapy .
hereafter the results of the therapists questionnaire ( without any order and with duplicates ) .
in synthesis they were asked as therapists to list the most negative and positive aspects of all the used systems . _ list of the most negative aspect(s ) _ : + for the pc game : + - the patient had to look at the screen while playing + - the patient had to grab the mouse + - the fact to use only the upper part of the triangle was upsetting ( here the therapist is referring to the game ) + - the screen was small + - it s difficult for the patient to move the mouse + - it s difficult for the patient to use the mouse + for the mr game : + - it takes more time to start up ( here the therapist is referring to the system / wii calibration ) + for classical therapy : + - tiresome and painful + - motivation is low + - the patient has to confront with his own limitation _ list the most positive aspect(s ) _ : + for the pc game : + - games are fun + - the patient can forget his environment + - different and may be funny for the mr game : + - accessible .
you can use whatever surface you want ( for example a table ) + - not so difficult to use as in the mouse case + - less material needed once started + none of the therapists gave particular comments on the classical therapy .
finally , as therapists they had to answer the following question : `` in general , which therapy did you preferred to do ?
( classify exercises in order of preference ) '' .
this time they were thus asked their own preferences on the therapies . all the three gave the following order : 1)mixed reality game , 2)pc game , 3)physical therapy .
as described in section [ methods ] the main goal of the experiment is to compare a mixed reality system ( mrs ) with two alternative single - user tools through an pilot study .
the two alternative tools were classical physical rehabilitation and an ad - hoc post stroke pc game .
we expected the mrs and the pc game to be more accepted in shackel s sense @xcite than classical rehabilitation when therapists are playing the patient role .
the mrs easiness of use is also expected to be higher than the pc one .
we are then analyzing perceived utility , usability , likeability ( affective evaluation ) ( see figure [ fig : shackel ] ) it is important to note that this is a pilot study involving a limited number of participants ( @xmath0 ) to prepare a larger scale experiment .
consequently , quantitative data are presented as a support for discussion and no statistically valid generalizations could be inferred at this stage .
as we can see from figure [ fig : graph ] elements linked to the environment oblivion ( immersion and flow ) have different ratings .
flow is rated low for classical rehabilitation ( @xmath1 ) while pc ( @xmath2 ) and mixed reality ( @xmath3 ) have higher scores .
sensory and imaginative immersion ratings are definitely lower for classical rehabilitation ( @xmath4 ) , next comes pc ( @xmath5 ) and finally mixed reality ( @xmath6 ) we can conclude that within the context of this experiment , both mrs and pc have been considered as more immersive , followed by classical therapy .
this is an expected result since it is already known that using games for rehabilitation creates a sense of meaningfulness for repeated movements and encourages flow experience by forgetting everything around and being focused only on the task to be performed @xcite .
lack of tension is an element characterizing tiredness and boringness .
results show that this component is almost similar for pc , mrs and classical therapy with respectively ( @xmath7 ) , ( @xmath8 ) and ( @xmath9 ) .
so , within the context of this experiment all systems have been considered equivalent in terms of tiredness and boringness .
when asked about this point , therapists highlight the fact that post stroke rehabilitation demands a lot of effort from the patient .
so , game based rehabilitation can facilitate , to some extent , the context of rehabilitation but patients will sooner or later show fatigue and pain and the therapeutic session has to stop .
if we look at negative affects , pc ( @xmath10 ) is the one rating lower , followed by mixed reality ( @xmath11 ) and the classical therapy ( @xmath12 ) is the one rated with higher negative affect .
when asked qualitatively about this fact , therapists mentioned the problem of exertion when using mrs .
in fact , there is a one to one mapping of patient movements between the physical and virtual world .
so the scaling and artificial compensation of movements is not possible as it could be performed within a virtual environment by changing speed and sensitivity of the pointing device for instance .
this was considered as a very interesting remark that we were not aware of before conducting the study .
consequently , within mrs the exertion of the patient is an important challenge to take into account and innovative solutions have to be provided to compensate patient s limited movement such as modifying dynamically size of objects or moving objects closer to patient when she is stuck .
what is very interesting is that while mr is seen as a possible cause of moderately negative affects , it is rated as the systems with the higher positive affects ( ( @xmath13 ) for mr ; ( @xmath14 ) for pc ; ( @xmath15 ) for classical therapy ) this reinforces the idea that mr has the potential to increase the volume of rehabilitation by increasing the exercises they do .
however , as said previously , this should be done while taking in account patient s exertion . finally , competences and challenges are rated in a very similar way .
this means that the effort the patient had to put to understand and perform exercises was the same . to summarize , when playing the role of patients , therapists
have considered : * that pc , mrs and classical therapy were equivalent on competence , tension and challenge components . *
both pc and mrs systems are better than classical therapy on immersion , flow , negative affect and positive affects . *
mrs has been considered as worse than pc on negative affect due to worries expressed on fatigue .
if we switch to the therapists questionnaire we can describe why the mrs has a higher potential to be accepted by patients .
in fact , one of the most underlined problems with pc rehabilitation was the use of the mouse as a pointing device .
this limitation becomes one of potentiality of mrs ( see the comment : not difficult to use as in the mouse case ) .
in addition , comments on pc usage are also related to limitations about screen dimension .
finally the most interesting comment is linked to the fact that the patients have to look in another place ( the pc screen ) in order to perform the task .
this could be a great problem when using this kind of device on some patients .
the mrs on the other hand can project on any surface .
finally it is interesting to note that all three subjects gave the same order of preferences : ( 1 ) mrs , ( 2 ) pc , ( 3 ) classical therapy both as patients and as therapists .
objective of this paper was to present a mixed reality system ( mrs ) for rehabilitation of the upper limb after stroke .
the system answers the following challenges : ( i ) increase motivation of patients by making the training a personalized experience ( ii ) take into account patients impairments by offering intuitive and easy to use interaction modalities ( iii ) make it possible to therapists to track patient s activities and to evaluate / track patient s progress ( iv ) open opportunities to telemedicine and tele rehabilitation ( v ) provide an economically acceptable system by reducing both equipment and management costs .
it s our opinion that it is important to evaluate the acceptance of these systems not only from the the patient s point of view but also from therapists point of view .
we decide then to test this system wih therapists in a pilot study conducted in conjunction with a french hospital .
the main assumption behind this experiment was that , even if clinical efficiency is demonstrated , if therapist does not accept the system it will be used only for academic studies and may not be widely accepted in real rehabilitation centers .
the pilot described in this paper involved 3 therapists who played the role of patients .
the main idea was that , because of the years they spent with patients , they are able to simulate patients reactions to the system .
three sessions , one using conventional rehabilitation , another using an ad hoc developed game on a pc , and another using a mixed reality version of the same game were held .
we expected the mrs and the pc game to be accepted more than physical rehabilitation by the therapists in the patient s role(h1 ) , and mrs easiness of use to be considered higher than the pc one(h2 ) .
results shows that while the effort the patient has to put in the exercises was practically the same for all the systems , mixed reality can be a mean to help patients to forget exertion and potentially augment the number of exercises they do .
we can say that mixed reality and pc games requires the same amount of effort than classical rehabilitation .
however , because of the characteristics of mixed reality and pc games we can suppose that patients will prefer to use this two systems with a preference for the mrs ( h2 ) .
in addition , analyzing the therapists questionnaire we can say that the mixed reality system has a higher potential to be accepted by patients ( h1 ) .
in fact , one of the most underlined problems with pc rehabilitation was the use of a mouse as a device .
this limitation becomes one of potentiality of mixed reality .
in addition the mrs can be project in whatever surface and avoid the problem to look in a place and perform the task in another .
the pilot study described in this paper was not intended to deliver statistical evidence , but simply to give approximate values guiding the set - up of following experiments .
for this reason we scheduled two more experiments .
the first one is an in depth version of the pilot study presented in this paper . for this experiment
we want to gather a greater number of data using the same role playing protocol .
the second one will drop the role playing and will analyze actual stroke rehabilitation of the upper limb with patients and therapists each in they own role .
the main idea behind this two experiment is to compare results from the two in order to understand the discrepancy between patients attitudes and therapists expectations towards the system .
will be added , a. , mubin , o. , shahid , s. , martens , j .- b . ,
2008 . designing and evaluating the tabletop game experience for senior citizens .
proceedings of the 5th nordic conference on human - computer interaction building bridges - nordichi 08 , 403 .
burke , j. w. , mcneill , m. d. j. , charles , d. k. , morrow , p. j. , crosbie , j. h. , mcdonough , s. m. , aug .
2009 . optimising engagement for stroke rehabilitation using serious games .
the visual computer 25 ( 12 ) , 10851099 .
cameiro , m. s. , badia , s. b. i. , oller , e. d. , verschure , p. f. , jan .
neurorehabilitation using the virtual reality based rehabilitation gaming system : methodology , design , psychometrics , usability and validation .
journal of neuroengineering and rehabilitation 7 ( 1 ) , 48 .
http://www.ncbi.nlm.nih.gov/pubmed/20860808 chen , y. , huang , h. , xu , w. , wallis , r. i. , sundaram , h. , rikakis , t. , ingalls , t. , olson , l. , he , j. , 2006 .
the design of a real - time , multimodal biofeedback system for stroke patient rehabilitation .
proceedings of the 14th annual acm international conference on multimedia - multimedia 06 , 763 .
http://portal.acm.org/citation.cfm?doid=1180639.1180804 , palma , phyllis , bender , anneke , december 2007 .
feasibility of using the sony playstation 2 gaming platform for an individual poststroke : a case report .
journal of neurologic physical therapy 31 ( 4 ) , 180189 .
goude , d. , bjrk , s. , rydmark , m. , jan . 2007 .
game design in virtual reality systems for stroke rehabilitation .
studies in health technology and informatics 125 ( 4 ) , 1468 .
http://www.ncbi.nlm.nih.gov/pubmed/17377254 jack , d. , boian , r. , merians , a. , tremaine , m. , burdea , g. , adamovich , s. , recce , m. , poizner , h. , sep .
2001 . virtual reality - enhanced stroke rehabilitation .
ieee transactions on neural systems and rehabilitation engineering : a publication of the ieee engineering in medicine and biology society 9 ( 3 ) , 308318 .
lo , a. c. , guarino , p. d. , g.richards , l. , haselkorn , j. k. , anddaniel g. federman , g. f. w. , ringer , r. j. , wagner , t. h. , krebs , h. i. , volpe , b. t. , bever , c. t. , bravata , p. , 2010 .
robot - assisted therapy for long - term upper - limb impairment after stroke .
new england journal of medicine 362 ( 19 ) , 17721783 .
micera , s. , carrozza , m. c. , guglielmelli , e. , cappiello , g. , zaccone , f. , freschi , c. , colombo , r. , mazzone , a. , delconte , c. , pisano , f. , minuco , g. , dario , p. , 2005 . a simple robotic system for neurorehabilitation .
autonomous robots 19 ( 3 ) , 271284 .
poels , k. , yvonnede kort , ijsselsteijn , w. , 2007 .
`` it is always a lot of fun ! '' : exploring dimensions of digital game experience using focus group methodology . in : future play 07 : proceedings of the 2007 conference on future play .
acm , new york , ny , usa , pp .
8389 .
reid , d. , hirji , t. , 2004 . the influence of a virtual reality leisure intervention program on the motivation of older adult stroke survivors : a pilot study . physical & occupational therapy in geriatrics 21 ( 4 ) , 119 .
shackel , b. , 1991 .
usability context , framework , definition , design and evaluation . in : shackel , b. , richardson , s. j. ( eds . ) , human factors for informatics usability .
cambridge university press , pp . 2131 .
van der lee , j. h. , wagenaar , r. c. , lankhorst , g. j. , vogelaar , t. w. , deville , w. l. , bouter , l. m. , november 1999 .
forced use of the upper extremity in chronic stroke patients : results from a single - blind randomized clinical trial .
stroke 30 ( 11 ) , 23692375 .
yavuzer , g. , senel , a. , atay , m. , hj , 2008 .
`` playstation eyetoy games '' improve upper extremity - related motor functioning in subacute stroke : a randomized controlled clinical trial .
european journal of hys rehabil med 44 ( 3 ) , 237244 . | the objective of this paper is to present a mixed reality system ( mrs ) for rehabilitation of the upper limb after stroke .
the system answers the following challenges : ( i ) increase motivation of patients by making the training a personalized experience ; ( ii ) take into account patients impairments by offering intuitive and easy to use interaction modalities ; ( iii ) make it possible to therapists to track patient s activities and to evaluate / track their progress ; ( iv ) open opportunities for telemedicine and tele rehabilitation ; ( v ) and provide an economically acceptable system by reducing both equipment and management costs . in order to test this system
a pilot study has been conducted in conjunction with a french hospital in order to understand the potential and benefits of mixed reality .
the pilot involved 3 therapists who played the role of patients .
three sessions , one using conventional rehabilitation , another using an ad hoc developed game on a pc , and another using a mixed reality version of the same game were held .
results have shown the mrs and the pc game to be accepted more than physical rehabilitation .
mixed reality , post stroke rehabilitation , serious games |
we will consider in this paper _ las vegas algorithms _ , introduced a few decades ago by @xcite , i.e. randomized algorithms whose runtime might vary from one execution to another , even with the same input . an important class of las vegas algorithms is the family of _ stochastic local search _ methods @xcite . they have been used in combinatorial optimization for finding optimal or near - optimal solutions for several decades @xcite , stemming from the pioneering work of lin on the traveling salesman problem @xcite .
theses methods are now widely used in combinatorial optimization to solve real - life problems when the search space is too large to be explored by complete search algorithm , such as mixed integer programming or constraint solving , c.f . @xcite . in the last years , several proposal for implementing local search algorithms on parallel computer have been proposed , the most popular being to run several competing instances of the algorithms on different cores , with different initial conditions or parameters , and let the fastest process win over others .
we thus have an algorithm with the minimal execution time among the launched processes .
this lead to so - called independent multi - walk algorithms in the local search community @xcite and portfolio algorithms in the sat community ( satisfiability of boolean formula ) @xcite .
this parallelization scheme can of course be generalized to any las vegas algorithm .
the goal of this paper is to study the parallel performances of las vegas algorithms under this independent multi - walk scheme , and to predict the performances of the parallel execution from the runtime distribution of the sequential runs of a given algorithm .
we will confront these predictions with actual speedups obtained for a parallel implementation of a local search algorithm and show that the prediction can be quite accurate , matching the actual speedup very well up to 100 parallel cores and then with a deviation limited to about 20% up to 256 cores .
the paper is organized as follows .
section 2 is devoted to present the definition of las vegas algorithms , their parallel multi - walk execution scheme , and the main idea for predicting the parallel speedups .
section [ probabilistic - model ] will detail the probabilistic model of las vegas algorithms and their parallel execution scheme .
section [ local - search ] will present the example of local search algorithms for combinatorial optimization , while section [ benchmarks ] will detail the benchmark problems and the sequential performances . then
, section [ prediction ] will apply the general probabilistic model to the benchmark results and thus predict their parallel speedup , which will be compared to actual speedups of a parallel implementation in section [ analysis ] . a short conclusion and future work end will the paper .
we borrow the following definition from @xcite , chapter 4 .
an algorithm a for a problem class @xmath0 is a ( generalized ) las vegas algorithm if and only if it has the following properties : 1 .
if for a given problem instance @xmath1 , algorithm a terminates returning a solution @xmath2 , @xmath2 is guaranteed to be a correct solution of @xmath3 .
2 . for any given instance @xmath4 ,
the run - time of a applied to @xmath3 is a random variable .
this is a slight generalization of the classical definition , as it includes algorithms which are not guaranteed to return a solution .
a large class of las vegas algorithms is the so - called family of _ metaheuristics _ , such as simulated annealing , genetic algorithms , tabu search , swarm optimization , ant - colony optimization , etc , which have been applied to different sets of problems ranging from resource allocation , scheduling , packing , layout design , frequency allocation , etc .
parallel implementation of local search metaheuristics @xcite has been studied since the early 1990s , when parallel machines started to become widely available @xcite . with the increasing availability of pc clusters in the early 2000s ,
this domain became active again @xcite .
apart from domain - decomposition methods and population - based method ( such as genetic algorithms ) , @xcite distinguishes between single - walk and multi - walk methods for local search .
single - walk methods consist in using parallelism inside a single search process , _ e.g.,_for parallelizing the exploration of the neighborhood ( see for instance @xcite for such a method making use of gpus for the parallel phase ) .
multi - walk methods ( parallel execution of multi - start methods ) consist in developing concurrent explorations of the search space , either independently or cooperatively with some communication between concurrent processes .
sophisticated cooperative strategies for multi - walk methods can be devised by using solution pools @xcite , but require shared - memory or emulation of central memory in distributed clusters , thus impacting on performance .
a key point is that a multi - walk scheme is easier to implement on parallel computers without shared memory and can lead , in theory at least , to linear speedups @xcite .
however this is only true under certain assumptions and we will see that we need to develop a more realistic model in order to cope with the performance actually observed in parallel executions .
let us now formally define a parallel multi - walk las vegas algorithm .
an algorithm a for a problem class @xmath0 is a ( parallel ) multi - walk las vegas algorithm if and only if it has the following properties : 1 .
it consists of @xmath5 instances of a sequential las vegas algorithm a for @xmath0 , say @xmath6 .
if , for a given problem instance @xmath7 , there exists at least one @xmath8 $ ] such that @xmath9 terminates , then let @xmath10 $ ] , be the instance of a terminating with the minimal runtime and let @xmath2 be the solution returned by @xmath11 .
then algorithm a terminates in the same time as @xmath11 and returns solution @xmath2 .
3 . if , for a given problem instance @xmath7 , all @xmath12 $ ] , do not terminate then a does not terminate .
the multi - walk parallel scheme is rather simple , yet it provides an interesting test - case to study how las vegas algorithms can scale - up in parallel .
indeed runtime will vary among the processes launched in parallel and the overall runtime will be that of the instance with minimal execution time ( i.e. `` long '' runs are killed by `` shorter '' ones ) .
the question is thus to quantify the ( relative ) notion of short and long runs and their probability distribution .
this might gives us a key to quantify the expected parallel speed - up .
obviously , this can be observed from the sequential behavior of the algorithm , and more precisely from the proportion of long and short runs in the sequential runtime distribution . in the following ,
we propose a probabilistic model to quantify the expected speed - up of multi - walk las vegas algorithms .
this makes it possible to give a general formula for the speed - up , depending on the sequential behavior of the algorithm .
our model is related to _ order statistics _ , a rather new domain of statistics @xcite , which is the statistics of sorted random draws .
indeed , explicit formulas have been given for several well - known distributions .
relying on an approximation of the sequential distribution , we compute the average speed - up for the multi - walk extension .
experiments show that the prediction is quite good and opens the way for defining more accurate models and apply them to larger classes of algorithms .
previous works @xcite studied the case of a particular distribution for the sequential algorithm , the exponential distribution .
this case is ideal and the best possible , as it yields a linear speed - up .
our model makes it possible to approximate las vegas algorithms by other types of distribution , such as a shifted exponential distribution or a lognormal distribution . in the last two cases the speed - up is
no longer linear , but admits a finite limit when the number of processors tends toward infinity .
we will see that it fits experimental data for some problems .
local search algorithms are stochastic processes .
they include several random components : choice of an initial configuration , choice of a move among several candidates , plateau mechanism , random restart , etc . in the following
, we will consider the _ computation time _ of an algorithm ( whatever it is ) as a random variable , and use elements of probability theory to study its multi - walk parallel version .
notice that the computation time is not necessarily the cpu - time ; it can also be the number of iterations performed during the execution of the algorithm .
consider a given algorithm on a given problem of a given size , say , the magic - square@xmath13 .
depending on the result of some random components inside the algorithm , it may find a solution after @xmath14 iterations , @xmath15 iterations , or @xmath16 iterations .
the number of iterations of the algorithm is thus a discrete random variable , let s call it @xmath17 , with values in @xmath18 .
@xmath17 can be studied through its cumulative distribution , which is by definition , the function @xmath19 s.t .
@xmath20}}$ ] , or by its distribution , which is by definition the derivative of @xmath19 : @xmath21 .
it is often more convenient to consider distributions with values in @xmath22 because it makes calculations easier .
for the same reason , although @xmath23 is defined in @xmath18 , we will use its natural extension to @xmath22 .
the expectation of the computation is then defined as @xmath24}}=\int_0^\infty t { \ensuremath{f_{y}}}(t ) dt$ ] assume that the base algorithm is concurrently run in parallel on @xmath5 cores . in other words , over each core the running process is a fork of the algorithm . the first process which finds a solution then kills all others .
and the algorithm terminates .
the @xmath25-th process corresponds to a draw of a random variable @xmath26 , following distribution @xmath23 .
the variables @xmath26 are thus independently and identically distributed ( i.i.d . ) .
the computation time of the whole parallel process is also a random variable , let @xmath27 , with a distribution @xmath28 that depends both on @xmath5 and on @xmath29 .
since all the @xmath26 are i.i.d . , the cumulative distribution @xmath30 can be computed as follows : @xmath31 } } \\ & = & { \ensuremath{\mathbb{p}[\exists i \in \{1 ... n\ } , x_i \leq x]}}\\ & = & 1 - { \ensuremath{\mathbb{p}[\forall i \in \{1 ... n\ } , x_i > x]}}\\ & = & 1 - \prod_{i=1}^{n } { \ensuremath{\mathbb{p}[x_i > x]}}\\ & = & 1 - \left ( 1 - { \ensuremath{\mathcal{f}_{y}}}(x)\right)^n \\\end{aligned}\ ] ] which leads to : @xmath32 thus , knowing the distribution for the base algorithm @xmath17 , one can calculate the distribution for @xmath33 . in the general case
, the formula shows that the parallel algorithm favors short runs , by killing the slower processes .
thus , we can expect that the distribution of @xmath33 moves toward the origin , and is more peaked . as an example , figure [ min - distribution ] shows this phenomenon when the base algorithm admits a gaussian distribution .
distribution of @xmath27 , in the case where @xmath17 admits a gaussian distribution ( cut on @xmath34 and renormalized ) .
the blue curve is @xmath17 .
the distributions of @xmath33 are in pink for @xmath35 , in yellow for @xmath36 and in green for @xmath37 .
, scaledwidth=40.0% ] the model described above gives the probability distribution of a parallelized version of any random algorithm .
we can now calculate the expectation for the parallel process with the following relation : @xmath38 } } & = & \int_0^\infty t { \ensuremath{f_{{\ensuremath{z^{(n)}}}}}}(t ) dt\\ & = & n \int_0^\infty t { \ensuremath{f_{y}}}(t ) ( 1-{\ensuremath{\mathcal{f}_{y}}}(t))^{n-1}dt \\\end{aligned}\ ] ] unfortunately , this does not lead to a general formula for @xmath39 $ ] . in the following , we will study it for different specific distributions . to measure the gain obtained by parallelizing the algorithm on @xmath5 core , we will study the speed - up @xmath40 defined as : @xmath41}}/{\ensuremath{\mathbb{e}[{\ensuremath{z^{(n)}}}]}}\ ] ] again , no general formula can be computed and the expression of the speed - up will depend on the distribution of @xmath17 .
however , it is worth noting that our computation of the speed - up is related to order statistics , see @xcite for a detailed presentation .
for instance , the first order statistics of a distribution is its minimal value , and the @xmath42 order statistic is its @xmath42-smallest value . for predicting the speedup , we are indeed interested in computing the expectation of the distribution of the minimum draw . as the above formula suggests , this may lead to heavy calculations , but recent studies such as @xcite give explicit formulas for this quantity for several classical probability distributions .
assume that @xmath17 has a shifted exponential distribution , as it has been suggested by @xcite .
@xmath43}}=&x_0 + 1/\lambda\end{aligned}\ ] ] then the above formula can be symbolically computed by hand : @xmath44 the intuitive observation of section [ min - distribution ] is easily seen on the expression of the parallel distribution , which has an initial value multiplied by @xmath5 but an exponential factor decreasing @xmath5-times faster , as shown on the curves of figure [ lois - exponentielle ] . for an exponential distribution , here in blue with @xmath45 and @xmath46 , simulations of the distribution of @xmath27for @xmath47 ( pink ) , @xmath48 ( yellow ) and @xmath49 ( green).,scaledwidth=45.0% ] and in this case
, one can symbolically compute both the expectation and speed - up for @xmath27 : @xmath38}}&= & n
\lambda \int_{x_0}^\infty t e^{- n \lambda ( t - x_0 ) } dt \\ & = & x_0+\frac{1}{n\lambda } \\ { \ensuremath{\mathcal{g}_{n}}}&= & \frac{x_0+\frac{1}{\lambda } } { x_0+\frac{1}{n\lambda } } \\\end{aligned}\ ] ] figure [ speedup - exponentielle ] shows the evolution of the speed - up when the number of cores increases .
with such a rather simple formula for the speed - up , it is worth studying what happens when the number of cores @xmath5 tends to infinity .
depending on the chosen algorithm , @xmath50 may be null or not .
if @xmath51 , then the expectation tends to @xmath14 and the speed - up is equal to @xmath5 .
this case has already been studied by @xcite . for @xmath52 ,
the speed - up admits a finite limit which is @xmath53 .
yet , this limit may be reached slowly , but depends on the value of @xmath50 and @xmath54 : obviously , the closest @xmath50 is to zero and the higher it will be . another interesting value is the coefficient of the tangent at the origin , which approximates the speed - up for a small number of cores . in case of an exponential ,
it is @xmath55 .
the higher @xmath50 and @xmath54 , the bigger is the speed - up at the beginning . in the following
, we will see that , depending on the combinations of @xmath50 and @xmath54 , different behaviors can be observed
. predicted speed - up in case of an exponential distribution , with @xmath45 and @xmath46 , w.r.t .
the number of cores.,scaledwidth=40.0% ] other distributions can be considered , depending on the behavior of the base algorithm .
we will study the case of a lognormal distribution , which is the log of a gaussian distribution , because it will be shown in section [ ms200 ] that it fits one experiment .
it has two parameters , the mean @xmath56 and the standard deviation @xmath57 . in the same way as the shifted exponential
, we shift the distribution so that it starts at a given parameter @xmath50 .
formally , a ( shifted ) lognormal distribution is defined as : @xmath58 where erfc is the complementary error function defined by @xmath59 .
figure [ loi - lognormale ] depicts lognormal distributions of @xmath27 , for several @xmath5 .
the computations for the distribution of @xmath27 , its expectation and the theoretical speed - up are quite complicated formulas . but @xcite gives an explicit formula for all the moments of lognormal order statistics with only a numerical integration step , from which we can derive a computation of the speed - up ( since the expectation of @xmath27is the first order moment for the first order statistics ) .
this allows us to draw the general shape of the speed - up , an example being given on figure [ speedup - lognormale ] .
due to the numerical integration step , which requires numerical values for the number of cores @xmath5 , we restrict the computation to integer values of @xmath5 .
this is a reasonable limitation as the number of cores is indeed an integer . for a lognormal distribution , here in blue with @xmath51 , @xmath60 and @xmath61 ,
simulations of the distribution of @xmath27for @xmath47 ( pink ) , @xmath48 ( yellow ) and @xmath49 ( green).,scaledwidth=40.0% ] predicted speed - up in case of a lognormal distribution , with @xmath51 , @xmath60 and @xmath61 , depending on the number of cores ( on the abscissa).,scaledwidth=40.0% ]
since about a decade , the interest for the family of local search methods and metaheuristics for solving large combinatorial problems has been growing and has attracted much attention from both the operations research and the artificial intelligence communities for solving real - life problems @xcite .
local search starts from a random configuration and tries to improve this configuration , little by little , through small changes in the values of the problem variables .
hence the term `` local search '' as , at each time step , only new configurations that are `` neighbors '' of the current configuration are explored
. the definition of what constitutes a neighborhood will of course be problem - dependent , but basically it consists in changing the value of a few variables only ( usually one or two ) .
the advantage of local search methods is that they will usually quickly converge towards a solution ( if the optimality criterion and the notion of neighborhood are defined correctly ... ) and not exhaustively explore the entire search space . applying local search to constraint satisfaction problems ( csp )
has been attracting some interest since about a decade @xcite , as it can tackle csps instances far beyond the reach of classical propagation - based constraint solvers @xcite .
a generic , domain - independent constraint - based local search method , named adaptive search , has been proposed by @xcite .
this meta - heuristic takes advantage of the structure of the problem in terms of constraints and variables and can guide the search more precisely than a single global cost function to optimize , such as for instance the number of violated constraints
. the algorithm also uses a short - term adaptive memory in the spirit of tabu search in order to prevent stagnation in local minima and loops .
an implementation of adaptive search ( as ) has been developed in c language as a framework library and is available as a freeware at the url : + we used this reference implementation for our experiments .
the adaptive search method can be applied to a large class of constraints ( _ e.g. _ linear and non - linear arithmetic constraints , symbolic constraints , etc . ) and naturally copes with over - constrained problems @xcite .
the input of the method is a constraint satisfaction problem ( csp for short ) , which is defined as a triple ( x;d;c ) , where x is a set of variables , d is a set of domains , i.e. , finite sets of possible values ( one domain for each variable ) , and c a set of constraints restricting the values that the variables can simultaneously take . for each constraint , an _ error function _ needs to be defined ; it gives , for each tuple of variable values , an indication of how much the constraint is violated .
this idea has also been proposed independently by @xcite , where it is called `` penalty functions '' , and then reused by the comet system @xcite , where it is called `` violations '' . for example , the error function associated with an arithmetic constraint @xmath62 , for a given constant @xmath63 , can be @xmath64 .
adaptive search relies on iterative repair , based on variable and constraint error information , seeking to reduce the error on the worst variable so far .
the basic idea is to compute the error function for each constraint , then combine for each variable the errors of all constraints in which it appears , thereby projecting constraint errors onto the relevant variables .
this combination of errors is problem - dependent , see @xcite for details and examples , but it is usually a simple sum or a sum of absolute values , although it might also be a weighted sum if constraints are given different priorities . finally , the variable with the highest error is designated as the `` culprit '' and its value is modified . in this second step ,
the well known min - conflict heuristic @xcite is used to select the value in the variable domain which is the most promising , that is , the value for which the total error in the next configuration is minimal . in order to prevent being trapped in local minima
, the adaptive search method also includes a short - term memory mechanism to store configurations to avoid ( variables can be marked tabu and `` frozen '' for a number of iterations ) .
it also integrates reset transitions to escape stagnation around local minima .
a reset consists in assigning fresh random values to some variables ( also randomly chosen ) .
a reset is guided by the number of variables being marked tabu .
it is also possible to restart from scratch when the number of iterations becomes too large ( this can be viewed as a reset of all variables but it is guided by the number of iterations ) .
the core ideas of adaptive search can be summarized as follow : * to consider for each constraint a heuristic function that is able to compute an approximated degree of satisfaction of the goals ( the current _ error _ on the constraint ) ; * to aggregate constraints on each variable and project the error on variables thus trying to repair the _
worst _ variable with the most promising value ; * to keep a short - term memory of bad configurations to avoid looping ( _ i.e. _ some sort of _ tabu list _ ) together with a reset mechanism .
we have chosen to test this method on two problems from the csplib benchmark library @xcite , and on a hard combinatorial problem abstracted from radar and sonar applications .
after briefly introducing the classical benchmarks , we detail the latter problem , called costas array . then we show the performance and the speed - ups obtained with both sequential and a multi - walk adaptive search algorithm on these problems .
we use two classical benchmarks from csplib consisting of : * the all - intervalseries problem ( prob007 in csplib ) , * the magic - squareproblem ( prob019 in csplib ) . although these benchmarks are academic , they are abstractions of real - world problems and could involve very large combinatorial search spaces , _ e.g.,_the
200@xmath65200 magic - squareproblem requires 40,000 variables whose domains range over 40,000 values . indeed
the search space in the adaptive search model ( using permutations ) is @xmath66 , _
i.e.,_more than @xmath67 configurations .
classical propagation - based constraint solvers can not solve this problem for instances higher than 20x20 . also note that we are tackling constraint _ satisfaction _
problems as optimization problems , that is , we want to minimize the global error ( representing the violation of constraints ) to value zero , therefore finding a solution means that we actually reach the bound ( zero ) of the objective function to minimize
. this problem is described as ` prob007 ` in the csplib .
this benchmark is in fact a well - known exercise in music @xcite where the goal is to compose a sequence of @xmath68 notes such that all are different and tonal intervals between consecutive notes are also distinct .
this problem is equivalent to finding a permutation of the @xmath68 first integers such that the absolute difference between two consecutive pairs of numbers are all different .
this amounts to finding a permutation @xmath69 of @xmath70 such that the list @xmath71 is a permutation of @xmath72 .
a possible solution for @xmath73 is @xmath74 since all consecutive distances are different : 3 @xmath75 6 @xmath76 0 @xmath77 7 @xmath78 2 @xmath79 4 @xmath80 5 @xmath81 1 the magic - squareproblem is catalogued as ` prob019 ` in csplib and consists in placing the numbers @xmath82 on an @xmath83 square such that each row , column and main diagonal equal the same sum ( the constant @xmath84 ) .
for instance , this figure shows a well - known solution for @xmath85 ( depicted by albrecht drer in his engraving _ melancholia i _ , 1514 ) .
a costas array is an @xmath86 grid containing @xmath68 marks such that there is exactly one mark per row and per column and the @xmath87 vectors joining the marks are all different .
we give here an example of costas array of size 5 .
it is convenient to see the costas arrayproblem ( cap ) as a permutation problem by considering an array of @xmath68 variables @xmath88 which forms a permutation of @xmath89 .
the above costas array can thus be represented by the array @xmath90 $ ] .
historically these arrays have been developed in the 1960 s to compute a set of sonar and radar frequencies avoiding noise @xcite . a very complete survey on costas arrays
can be found in @xcite .
the problem of finding a costas array of size @xmath68 is very complex since the required time grows exponentially with @xmath68 . in the 1980 s ,
several algorithms have been proposed to build a costas array given @xmath68 ( methods to produce costas arrays of order 24 to 29 can be found in @xcite ) , such as the welch construction @xcite and the golomb construction @xcite , but these methods can not built costas arrays of size @xmath91 and some higher non - prime sizes .
nowadays , after many decades of research , it remains unknown if there exist any costas arrays of size @xmath91 or @xmath92 .
another difficult problem is to enumerate all costas arrays for a given size . using the golomb and welch constructions , drakakis _ et .
al _ present in @xcite all costas arrays for @xmath93 .
they show that among the @xmath94 permutations , there are only 164 costas arrays , and 23 unique costas arrays up to rotation and reflection .
we run our benchmarks in a sequential manner in order to have about 650 runtimes for each .
sequential experiments , as well as parallel experiments , have been done on the _ griffon _ cluster of the grid5000 platform .
the following tables [ tab : seqtime ] and [ tab : seqiter ] shows the minimum , mean , median and maximum respectively among the runtimes and the number of iterations of our benchmarks .
.sequential execution times ( in seconds ) [ cols="^ , > , > , > , > " , ] it is worth noticing that our model approximates the behaviors of experimental results very closely , as shown by the predicted speed - ups matching closely the real ones .
moreover we can see that on the three benchmark programs , we needed to use three different types of distribution ( exponential , shifted exponential and lognormal ) , in order to approximate the experimental data most closely .
this shows that our model is quite general and can accommodate different types of parallel behaviors .
a quite interesting behavior is exhibited by the costas 21 problem .
our model predicts a linear speedup , up to 10,000 cores and beyond , and the experimental data gathered for this paper confirms this linear speed - up up to 256 cores .
would it scale up with a larger number of cores ?
indeed such an experiment has been done up to 8,192 cores on the jugene ibm bluegene / p at the jlich supercomputing center in germany ( with a total 294,912 cores ) , and reported in @xcite , of which figure [ fig : speedup - jugene ] is adapted .
we can see that the speed - up is indeed linear up to 8,192 cores , thus showing the adequation of the prediction model with the real data .
[ fig : speedup - jugene ] finally , let us note that our method exhibits an interesting phenomenon . for the three problems considered , the probability of returning a solution in _ no _ iterations is non - null :
since they start by a uniform random draw on the search space , there is a very small , but not null , probability that this random initialization directly returns the solution .
hence , in theory , @xmath51 and the speed - up should be linear , with an infinite limit when the number of cores tends to infinity .
intuitively , if the number of cores tends to infinity , at some point it will be large compared to the size of the search space and one of the cores is likely to immediately find the solution .
yet , in practice , observations shows that the experimental curves may be better approximated by a shifted exponential with @xmath52 , as it is the case for ai 700 . with an exponential distribution ,
this leads to non - linear speed - up with a finite limit .
indeed , the experimental speed - up for ai 700 is far from linear . on the contrary ,
costas 21 has a linear speed - up due to its @xmath95 , which makes the statistical test succeed for @xmath96 .
firstly , this suggests that the comparison between @xmath50 and @xmath97 on a number of observations is a key element for the parallel behavior .
it also means that the number of observations needed to properly approximate the sequential distribution probably depends on the problem .
we have proposed a theoretical model for predicting and analyzing the speed - ups of las vegas algorithms . it is worth noticing that our model mimics the behaviors of the experimental results very closely , as shown by the predicted speedups matching closely the real ones .
our practical experiments consisted in testing the accuracy of the model with respect to three instances of a local search algorithm for combinatorial optimization problems .
we showed that the parallel speed - ups predicted by our statistical model are accurate , matching the actual speed - ups very well up to 64 parallel cores and then with a deviation of about 10% , 15% or 30% ( depending on the benchmark problem ) up to 256 cores .
however , one limitation of our approach is that , in practice , we need to be able to compute the expectation of the minimum distribution . nevertheless , apart from the exponential distribution for which this computation is easy , recent results in the field of order statistics gives explicit formulas for a number of useful distributions : gaussian , lognormal , gamma , beta .
this provides a wide range of tools to analyze different behaviors . in this paper
we validated our approach on classical combinatorial optimization and csp benchmarks , but further research will consider a larger class of problems and algorithms , such as sat solvers and other randomized algorithms ( e.g. quick sort ) .
another interesting extension of this work would be to devise a method for predicting the speed - up from scratch , that is , without any knowledge on the algorithm distribution .
preliminary observation suggests that , given a problem and an algorithm , the general shape of the distribution is the same when the size of the instances varies .
for example , the different instances of all - intervalthat we tested all admit a shifted exponential distribution .
if this property is valid on a wide range of problems / algorithms , then we can develop a method for predicting the speed - up for large instances by learning the distribution shape on small instances ( which are easier to solve ) , and then estimating the parallel speed - up for larger instances with our model .
j. beard , j. russo , k. erickson , m. monteleone , and m. wright .
costas array generation and search methodology . _ aerospace and electronic systems , ieee transactions on _ , 430 ( 2):0 522 538 , april 2007 .
issn 0018 - 9251 .
doi : 10.1109/taes.2007.4285351 .
y. caniou , d. diaz , f. richoux , p. codognet , and s. abreu .
performance analysis of parallel constraint - based local search . in _
ppopp 2012 , 17th acm sigplan symposium on principles and practice of parallel programming _ , new orleans , la , usa , 2012 .
acm press .
poster paper .
d. diaz , f. richoux , y. caniou , p. codognet , and s. abreu .
parallel local search for the costas array problem . in _ ieee workshop on new trends in parallel computing and optimization ( pc012 ) , in conjunction with ipdps 2012 _ , shanghai , china , may 2012 . ieee press .
s. minton , m. d. johnston , a. b. philips , and p. laird .
minimizing conflicts : a heuristic repair method for constraint satisfaction and scheduling problems . _ artificial intelligence _
, 580 ( 1 - 3):0 161205 , 1992 . p. m. pardalos , l. s. pitsoulis , t. d. mavridou , and m. g. c. resende .
parallel search for combinatorial optimization : genetic algorithms , simulated annealing , tabu search and grasp . in _ proceedings of irregular _ , pages 317331 , 1995 .
t. van luong , n. melab , and e .-
. local search algorithms on graphics processing units . in _
evolutionary computation in combinatorial optimization _ , pages 264275 .
lncs 6022 , springer verlag , 2010 . | we propose a probabilistic model for the parallel execution of _ las vegas algorithms _ , _
i.e.,_randomized algorithms whose runtime might vary from one execution to another , even with the same input .
this model aims at predicting the parallel performances ( _ i.e.,_speedups ) by analysis the runtime distribution of the sequential runs of the algorithm .
then , we study in practice the case of a particular las vegas algorithm for combinatorial optimization , on three classical problems , and compare with an actual parallel implementation up to 256 cores .
we show that the prediction can be quite accurate , matching the actual speedups very well up to 100 parallel cores and then with a deviation of about 20% up to 256 cores .
theory , algorithms , performance .
las vegas algorithms , prediction , parallel speed - ups , local search , statistical modeling , runtime distributions . |
in complex control systems containing sampled - data elements , it is possible that these elements operate asynchronously . in some cases asynchronous character of operation of sampled - data elements does not influence stability of system . in other cases
any small desynchronization of the updating moments of sampled - data elements leads to dramatic changes of dynamics of a control system , and the system loses stability @xcite .
last years there is begun ( see , e.g. , @xcite ) intensive studying of the effects connected with asynchronous operation of control systems ; both necessary , and sufficient stability conditions for various classes of asynchronous systems were obtained . at the same time
no one succeed in finding general , effectively verified criteria of stability of asynchronous systems , similar to known for synchronous systems @xcite . the problem on stability of linear asynchronous systems has appeared more difficult than the problem on stability of synchronous systems . in the paper ,
attempt of formal explanation of complexity of the stability analysis problem for linear asynchronous systems is undertaken .
it is shown that there are no criteria of absolute stability of linear asynchronous systems consisting of a finite number of arithmetic operations .
consider a discrete - time linear control system whose dynamics is described by the vector difference equation @xmath0 where @xmath1 is the state vector of the system and @xmath2 is a square matrix of dimension @xmath3 with the elements @xmath4 .
the system will be called synchronous if @xmath5 .
if @xmath6 , and the set @xmath7 consists of finitely many elements @xmath8 , then the system will be called asynchronous or desynchronized .
let @xmath9 be a finite totality of square matrices of dimension @xmath3 . the system will be called absolutely stable with respect to the class of matrices @xmath10 ( cf .
@xcite ) if there exists @xmath11 such that for any sequence of matrices @xmath12 the following estimates hold : @xmath13 let us call the system absolutely exponentially stable with respect to the class of matrices @xmath10 if there exist @xmath11 and @xmath14 such that for any sequence of matrices @xmath12 the following estimates hold : @xmath15 if the class of matrices @xmath10 consists of the square matrices @xmath16 , @xmath17 , , @xmath18 of dimension @xmath3 then , for its description , it suffices to specify @xmath19 numbers : @xmath20 , @xmath21 ,
, @xmath22 , @xmath23 , @xmath24 , , @xmath25 , , @xmath26 , @xmath27 , , @xmath28 .
therefore , each class @xmath10 consisting of @xmath29 square matrices of dimension @xmath3 can be treated as a point in some space @xmath30 .
denote by @xmath31 the set of those classes @xmath10 in the space @xmath32 with respect to which the system is absolutely stable . by @xmath33
we denote the set of those classes @xmath34 with respect to which the system is absolutely exponentially stable .
now , the problem of studying the absolute stability of the system can be reformulated as the problem of description of the sets @xmath31 and @xmath33 ; the simpler in some sense the structure of the sets @xmath31 or @xmath33 the easier to obtain a criterion of absolute stability or absolute exponential stability .
the sets @xmath35 and @xmath36 allow a simple description .
indeed , each class @xmath37 consists of a single matrix .
therefore , we need to obtain conditions of stability or asymptotic stability of some difference equation @xmath38 .
the routh
hurwitz stability criterion @xcite allows to represent these conditions as a finite system of polynomial inequalities including the elements @xmath39 of the matrix @xmath40 .
verification of the obtained inequalities can be performed by a finite number of arithmetic operations over the elements of the matrix @xmath40 . in other words ,
the question whether an arbitrary class @xmath41 belongs to the set @xmath35 or @xmath36 may be resolved by a finite number of arithmetic operations .
is it possible , for @xmath42 , by a finite number of arithmetic operations to resolve the question whether an arbitrary class @xmath34 belongs to the set @xmath43 or @xmath33 ?
the answer to this question will be given in the next section .
let @xmath44 be an element of the coordinate space @xmath45 .
a finite sum @xmath46 with numerical coefficients @xmath47 is called a polynomial in variable @xmath48 .
a set @xmath49 is said to have the _
sa_-property @xcite if there exists a finite number of polynomials @xmath50 , , @xmath51 , @xmath52 ,
, @xmath53 such that @xmath54 coincides with the set of elements @xmath48 satisfying the condition @xmath55 a set @xmath54 is called semialgebraic @xcite if it is a unity of a finite number of the sets possessing the _ sa_-property . [ t:1 ]
let @xmath56 .
if a subset @xmath54 of the space @xmath32 satisfies conditions @xmath57 then it is not semialgebraic .
the proof of the theorem is given in the appendix .
semialgebraicity of a set is equivalent to the existence of a criterion ( consisting in verification of a finite number of the conditions of the form ) which allows by a finite number of arithmetic operations of addition , subtraction , multiplication and comparison of numbers to establish belonging of an element to a given set . as seen from theorem [ t:1
] , neither the set @xmath31 nor the set @xmath33 are semialgebraic .
so , the meaning of theorem [ t:1 ] is that in general , by a finite number of arithmetic operations , it is impossible to ascertain whether a desynchronized system is absolutely stable ( absolutely exponentially stable ) or not . the problem on the existence of algebraic criteria of stability is acute also for classes of desynchronized systems different from those considered above . for example , in the theory of continuous - time desynchronized systems there arises the problem of stability of the so - called regular systems @xcite ( i.e. , the systems with the infinite number of updating moments for each component ) . the discrete - time system , and the related to it sequence of matrices @xmath58 ,
will be called regular if each matrix @xmath59 from the class @xmath60 appears in the sequence @xmath61 infinitely many times .
denote by @xmath62 the greatest integer @xmath63 having the property : the set of matrices @xmath64 can be decomposed in @xmath63 subsets @xmath65 , @xmath66 , , @xmath67 such that each of them contains all the matrices @xmath68 .
clearly , the system is regular if and only if @xmath69 . the system will be called absolutely exponentially regularly stable with respect to the class of matrices @xmath60 if there exist @xmath11 , @xmath14 such that for any regular sequence of matrices @xmath12 the following estimates hold : @xmath70 denote by @xmath71 the set all the classes @xmath34 with respect to which the system is absolutely exponentially regularly stable .
[ t:2 ] let @xmath56 .
then the set @xmath71 is not semialgebraic . to prove theorem [ t:2 ]
it suffices to note that the set @xmath71 contains @xmath33 , and is contained in @xmath31 .
then by theorem [ t:1 ] it is not semialgebraic . in other words , for @xmath56 there are no semialgebraic criteria of absolutely exponentially regular stability of discrete - time desynchronized systems .
as shown above , the problem of absolute stability of the system can be reduced to the analysis of behaviour of infinite products of the matrices @xmath12 .
theorem [ t:3 ] below reduces the same problem to the descriptive - geometric question on existence in the space @xmath72 such a norm in which each matrix @xmath8 is contractive .
[ t:3 ] the system is absolutely stable in a class of matrices @xmath73
, @xmath74 if and only if there is a norm @xmath75 in @xmath72 for which the following inequalities hold : @xmath76 the system is absolutely exponentially stable in a class of matrices @xmath10 if and only if there is a norm @xmath75 in @xmath72 and a number @xmath77 for which the following inequalities hold : @xmath78 the proof of the theorem is given in the appendix . several important properties of the sets @xmath31 and @xmath33
follow from theorem [ t:3 ] .
for example , the set @xmath33 is open in @xmath32 ; the set @xmath33 belongs to the interior of the set @xmath31 .
due to openness of the set @xmath33 , if the system is absolutely exponentially stable with respect to some class @xmath79 then it is also absolutely exponentially stable with respect to any class @xmath80 , , @xmath74 of matrices @xmath59 sufficiently close to the corresponding matrices @xmath81 @xmath82 .
theorems [ t:1 ] and [ t:3 ] imply that the problem of construction , for a given set of square matrices , of a norm satisfying conditions or is algebraically unresolvable .
theorem [ t:1 ] states that in general there are no effective criteria of absolute stability of desynchronized systems .
nevertheless , such criteria may exist for some particular desynchronized systems .
let us present examples .
[ ex:1 ] denote by @xmath83 the subset of the space @xmath84 consisting of the classes @xmath85 of matrices @xmath59 of the form @xmath86 the problem on absolute stability of the system with respect to the classes @xmath87 arises in @xcite in the process of study of continuous - time systems with a special types of desynchronization of updating moments . [ t:4 ]
the system is absolutely stable with respect to the class of matrices @xmath88 of the form if and only if one of the following system of relations holds : @xmath89 , @xmath90 , @xmath91 ; @xmath89 , @xmath92 , @xmath93 , @xmath94 is arbitrary ; @xmath95 , @xmath96 is arbitrary , @xmath97 , @xmath91 ; @xmath98 , @xmath99 ; @xmath100 , @xmath101 , @xmath102 .
the proof of theorem [ t:4 ] is cumbersome and so is skipped .
let us point out that the criterion of absolute stability of the system with respect to the classes of matrices from @xmath103 , given in theorem [ t:4 ] , is semialgebraic .
[ ex:2 ] denote by @xmath104 the subset of the space @xmath84 consisting of the classes @xmath85 of matrices @xmath59 of the form for which @xmath105 .
the set @xmath106 is semialgebraic .
the criterion of absolute stability of the system with respect to the classes of matrices @xmath10 from @xmath104 consists in verification that the maximal eigenvalue of the matrix @xmath107 does not exceed @xmath108 .
this assertion is proved similarly to theorem 2 from the first part of @xcite .
_ proof of theorem [ t:1 ] _ suffices to present for the case @xmath109 .
the idea of proof is simple .
we construct two families of matrices depending on the real parameter @xmath110 $ ] : @xmath111 the set @xmath112 of all the classes @xmath10 of the form @xmath113 forms in the space @xmath114 an algebraic set .
suppose that the set @xmath54 is semialgebraic .
then the set @xmath115 is also semialgebraic .
therefore , by the theorem of whitney ( see , e.g. , @xcite ) about a finite number of connected components of a real algebraic set , a neighbourhood of the class @xmath116 in @xmath115 should be either empty or consisting of a finite number of the connected components
. we will show that it has infinitely many components of connectedness , see fig .
[ semialg ] .
so , the set @xmath54 is not semialgebraic .
has infinitely many components of connectedness ] let us pass to the theorem proof . denote by @xmath117 the euclidean norm in @xmath118 : if @xmath119 then @xmath120 . consider two families of matrices : @xmath121 [ l : a1 ] let @xmath122
. then @xmath123 .
[ l : a2 ] let @xmath124
. then @xmath125 , where @xmath126 , @xmath127 .
both lemmas follow from the equality @xmath128}{\cos\varphi}p(\varphi)$ ] @xmath129 whose validity is justified by direct calculations , see fig .
[ semialg2 ] . under the action of the map @xmath130 [ l : a3 ]
let @xmath124 , and @xmath131 or @xmath132 for @xmath133 .
then @xmath134 where @xmath135 , integers @xmath136 and @xmath137 are non - negative , @xmath138 .
prove the lemma by induction .
for @xmath139 the assertion of the lemma is evident ; suppose that it is valid for @xmath140 .
then for @xmath141 the matrix @xmath142 can be represented as @xmath143 , where @xmath144 , @xmath145 or @xmath146 , @xmath147 , @xmath148 . if @xmath149 then @xmath150 , and for the matrix @xmath40 the representation holds in which @xmath151 , @xmath152 , @xmath153 , @xmath154 . if @xmath155 and @xmath145 then @xmath156 , and for the matrix @xmath40 the representation holds in which @xmath157 , @xmath158 , @xmath159 , @xmath160 . if @xmath155 and @xmath146 then @xmath161 . here
the factor @xmath162 according to lemma [ l : a2 ] can be replaced by @xmath163 . then @xmath164 . therefore , for the matrix @xmath40 the representation holds in which @xmath165 , @xmath158 , @xmath159 , @xmath154 .
in addition , @xmath166 since @xmath144 , @xmath167 . the inductive step is completed .
lemma [ l : a3 ] is proved .
@xmath168 .
the proof of the corollary immediates from the representation and unitarity of the rotation matrix @xmath169 .
[ l : a4 ] let @xmath170
. then @xmath171 .
let @xmath172 be a sequence of matrices from @xmath173 .
then , for each @xmath174 , one of two equalities @xmath175 or @xmath176 holds . by @xmath177 , @xmath178 , where @xmath179 .
therefore , the product of matrices @xmath180 , @xmath181 , , @xmath182 can be represented in the form : @xmath183 , where @xmath184 or @xmath185 .
then , by corollary from lemma [ l : a3 ] , @xmath186 which implies absolute stability of the class of matrices @xmath173 .
lemma [ l : a4 ] is proved .
[ l : a5 ] let @xmath187 .
then @xmath188 for all sufficiently large @xmath189 .
proof . clearly the lemma will be proved
if , for each sufficiently large @xmath189 , there can be found a sequence of matrices @xmath190 such that @xmath191 .
define the sequence of matrices @xmath182 as follows : @xmath192=g(s_{n})$ ] , @xmath193=\ldots = a[(n+2)i+n]=h(s_{n})$ ] , @xmath194=g(s_{n})$ ] .
let us set @xmath195
. then @xmath196^{i}.\ ] ] since @xmath197 and @xmath198 , where @xmath199 , then @xmath200^{i}.\ ] ] consequently , by lemma [ l : a1 ] @xmath201 .
recall that @xmath202 is a projector and so @xmath203 .
therefore , @xmath204 direct calculations show that @xmath205 . henceforth , for sufficiently large values of @xmath189 the inequality @xmath206 holds . from this and from the relation
lemma [ l : a5 ] is proved .
let us complete the proof of the theorem . since by lemmas [ l : a4 ] and [ l : a5 ] @xmath171 , @xmath207 and @xmath57 then @xmath208 , @xmath209 .
but because of the points @xmath210 and @xmath211 interleave between each other then the set @xmath115 contains infinitely many different components of connectedness ( different classes @xmath173 belong to different components of connectedness ) . therefore , the set @xmath115 by the theorem of whitney @xcite is not semialgebraic .
but since @xmath112 is an algebraic set then the set @xmath54 is not semialgebraic .
theorem [ t:1 ] is proved .
_ proof of theorem [ t:3 ] . _ in one side the assertion of theorem [ t:3 ] is obvious : absolute stability and absolute exponential stability of the system with respect to the class @xmath9 immediately follow from inequalities and .
let us show that absolute exponential stability of the system with respect to the class of matrices @xmath60 implies .
let , for some @xmath77 , the relation be valid .
set @xmath212 , @xmath213 @xmath214 , where @xmath117 is the euclidean norm on @xmath72 , and the maximum is taken over all possible collections of the matrices @xmath215 .
define the norm @xmath75 as follows : @xmath216 .
the function @xmath217 is semiadditive and due to it satisfies the relations @xmath218 .
henceforth , @xmath219 if and only if @xmath220 . other properties of a norm are obvious for @xmath75 .
let us justify inequalities .
clearly , for any @xmath221 and @xmath222 the estimate @xmath223 is valid . therefore , @xmath224 from here @xmath225 for @xmath221 .
inequalities are proved .
construction of the norm @xmath75 and the proof of inequalities in the case of absolute stability of the system are carried out similarly .
theorem [ t:3 ] is proved .
10 url # 1`#1`urlprefixhref # 1#2#2 # 1#1 kleptsyn a. f. , kozyakin v. s. , krasnoselskii m. a. , and kuznetsov n. a. , _ effect of small synchronization errors on stability of complex systems .
i - iii _ , automat .
remote control , 1983 , vol .
44 , no .
7 , pp . 861867 , 1984 , vol . 45 , no . 3 , pp .
309314 , 1984 , vol .
45 , no . 8 , pp .
10141018 .
kleptsyn a. f. , _ investigation of stability of two - component desynchronized systems _ , in _
ixth all - union workshop on control problems .
theses _ , pp . 2728 , moscow : nauka , 1983 , in russian .
kleptsyn a. f. , kozyakin v. s. , krasnosel@xmath226ski m. a. , and kuznetsov n. a. , _ stability of desynchronized systems _ ,
nauk sssr , 1984 , vol .
274 , no . 5 , pp . 10531056 , in russian , translation in soviet phys .
29 ( 1984 ) , 9294 .
kleptsyn a. f. , krasnosel@xmath226ski m. a. , kuznetsov n. a. , and kozjakin v. s. , http://www.sciencedirect.com/science/article/pii/037847548490106x[_desynchronization of linear systems _ ] , math .
simulation , 1984 , vol .
26 , no . 5 , pp .
[ ] . http://www.sciencedirect.com/science/article/pii/037847548490106x kleptsyn a. f. , _ stability of desynchronized complex systems of a special type _ , avtomat .
i telemekh . , 1985 , no . 4 , pp . 169171 .
tsypkin a. z. , _ theory of linear sampling systems _ , moscow : fizmatgiz , 1963 , in russian .
aizerman m. a. and gantmacher f. r. , _ absolute stability of regulator systems _ , translated by e. polak , san francisco , calif . : holden - day inc . , 1964 .
treves j. f. , _ lectures on linear partial differential equations with constant coefficients _ , notas de matemtica , no .
27 , instituto de matemtica pura e aplicada do conselho nacional de pesquisas , rio de janeiro , 1961 . milnor j. , _ singular points of complex hypersurfaces _ , annals of mathematics studies , no . 61 , princeton , n.j .
: princeton university press , 1968 .
kozyakin v. s. , _ algebraic unsolvability of problem of absolute stability of desynchronized systems _
, avtomat .
i telemekh . , 1990 , no . 6 , pp .
4147 , in russian , translation in automat .
remote control 51 ( 1990 ) , no .
6 , part 1 , 754759 .
asarin e. a. , kozyakin v. s. , krasnosel@xmath226ski m. a. , and kuznetsov n. a. , http://eqworld.ipmnet.ru/ru/library/books/asarinkozyakinkrasnoselskijkuznecov1992ru.pdf[_analiz ustoichivosti rassinkhronizovannykh diskretnykh sistem _ ] , moscow : nauka , 1992 , in russian .
kozyakin v. s. , _ indefinability in o - minimal structures of finite sets of matrices whose infinite products converge and are bounded or unbounded _ , avtomat .
i telemekh . , 2003 , no . 9 , pp . 2441 , in russian , translation in autom .
remote control 64 ( 2003 ) , no .
9 , 13861400 .
http://dx.doi.org/10.1023/a:1026091717271 [ ] .
in the foregoing text , a few misprints in the proof of theorem [ t:1 ] occurred in the original journal version of the article @xcite were corrected , and two figures were added .
the improved text was included in the monograph @xcite .
generalization of the presented results can be found in @xcite . | in the author s article `` algebraic unsolvability of problem of absolute stability of desynchronized systems '' ( automat .
remote control 51 ( 1990 ) , no . 6 , pp .
754759 ) , it was shown that in general for linear desynchronized systems there are no algebraic criteria of absolute stability . in this paper , a few misprints occurred in the original version of the article are corrected , and two figures are added . |
in a recent paper @xcite we presented a spectral method for tensorial partial differential equations on geometries with a spherical component . we showed how to implement the ` eth'-formalism based on spin - weighted spherical harmonics following the work by newman , penrose and others @xcite .
the restriction to tensorial equations implied the use of spin - weighted functions with integer spin - weights . in the present work
we extend our method to also include functions with half - integer spin - weight on spherical geometries .
in particular , we show how the evolution of dynamical half - integer spin fields can be accomplished .
our motivation for doing so stems mainly from general relativity : spinorial fields describing elementary particles such as electrons or neutrinos are fundamental sources for the einstein equations and are studied for instance as test fields on kerr - newman backgrounds , see @xcite and references therein .
however , there are also interesting developments in condensed matter physics in relation to the description of graphene @xcite for which our method might be relevant .
the numerical treatment of the dirac field is particularly interesting as its evolution can be rather peculiar and sometimes even counter - intuitive as visualised by thaller @xcite . it is well known that when working with @xmath1 ( and other compact geometries ) coordinate singularities can lead to instabilities that spoil the accuracy of numerical schemes .
these issues can be avoided by working with spectral methods where coordinate - independent manipulation of expressions in terms of a well - defined basis of functions upon which the action of differential operators reduces to algebraic manipulation is possible @xcite .
thus , issues of coordinate singularities are dealt with automatically by the method .
our spectral method makes use of the spin - weighted spherical harmonics ( swsh ) @xcite which can be shown to be equivalent to the well - known wigner @xmath2-functions @xcite and thus provide a complete , orthonormal basis for @xmath3 . in the description of half - integer spin fields , we have found it convenient to work directly with the half - integer swsh due to the particularly simple action of the associated @xmath4 and @xmath5 differential operators which may be thought of as constituting covariant derivative operators on the sphere @xcite . as the action of the aforementioned operators on the swsh reduces to simple algebraic manipulation we may translate pde systems into coupled ( infinite dimensional ) ode systems and thus construct a spectral evolution scheme for an initial value problem ( ivp ) of interest .
our work extends that of @xcite into a spectral algorithm for spin - weighted spherical harmonics ( swsh ) with half - integer spin - weight @xmath6 .
the method we present inherits several desirable properties from @xcite .
in particular it is theoretically exact if a minimum number of grid points are used at a given band - limit @xmath7 .
furthermore the same algorithmic complexity of @xmath8 is achievable for spectral transformations . in brief , @xcite seek to calculate values for the integer swsh over @xmath1 by mapping the sphere into the @xmath0-torus ( thus allowing for fast fourier transforms to be used ) and relate the swsh to the reduced wigner @xmath9-functions evaluated at @xmath10 . for a short summary of past work related to the integer case see @xcite .
the approach to the half - integer case is in principle similar , however suitable modification of the @xmath0-torus map must be made in order to account for the periodicity of spinor fields .
in addition , the recursion relation @xcite that allows for evaluation of the wigner @xmath9-functions must be modified . as a toy model
we numerically explore the dynamics associated with a dirac equation on a 2-dimensional manifold with spatial topology of @xmath1 as an ivp .
this will allow for a test of the spin - weighted spectral transformations we present in the context of evolution equations .
common techniques for treating the numerical problem of the dirac equation consist of : fd schemes formulated on a flat - lattice in configuration space @xcite , on a grid within a finite - volume in momentum - space @xcite and using methods based on the split - step operator technique @xcite .
particular to the fd approach special care must be taken so as to avoid the fermion - doubling problem @xcite .
elimination of spurious modes introduced to the solution may be accomplished by means of nonlocal approximation for the spatial derivative operator @xcite or by staggered - grid schemes @xcite . as remarked upon in @xcite
the issue of spurious modes is not particular to the dirac equation but can occur whenever a symmetric fd approximant is used for a first derivative on a uniform grid .
we seek to avoid the above issue entirely by making use of the global approximation to functions and derivative operators that spectral methods provide @xcite .
the equations of motion ( eom ) that we derive for the dirac equation in the aforementioned geometric setting result in a coupled system of variable - coefficient linear hyperbolic pdes .
when performing a swsh decomposition this results in product terms that must be further simplified . from past experience @xcite
we have found that while it is possible to make use of clebsch - gordan expansions ( see sec .
[ sec : swshsum ] ) , i.e. , working entirely in the space of coefficients , it is far more convenient to instead work with a pseudo - spectral method ( see sec .
[ sec : psmethod ] ) , and thus we follow the latter approach in this work .
this paper is structured as follows : in sec .
[ sec : swshsum ] we recall basic properties of the half - integer swsh together with the @xmath4-formalism . in sec .
[ sec : fwdxform ] and [ sec : bwdxform ] respectively we discuss the forward and backward spectral transformations of spin - weighted functions on the sphere . in sec .
[ sec : wigrec ] we present the required modification to the recursion relation for computing the reduced wigner functions at @xmath10 .
subsequently , in sec . [
sec : consischk ] we perform consistency checks on our implementation of the algorithm .
error pairs and the property of exponential convergence are analysed . in sec .
[ sec : dirac ] we proceed by numerically solving the @xmath11 dimensional dirac equation as an ivp on a curved geometry , specifying to cases with spatial topology of @xmath1 .
we find eom adapted to the spin - weighted formalism . in sec .
[ sec : psmethod ] we briefly recall the pseudo - spectral method . in sec .
[ sec : numericaleom ] we construct numerical solutions to the eom and inspect convergence properties of the numerical solutions obtained together with conserved currents .
[ sec : conc ] concludes .
in this section we briefly summarise key properties of the spin - weighted spherical harmonics ( swsh ) that we will make use of in the presentation of our half - integer spectral algorithm and in the pseudo - spectral method for solution of dynamical systems . for further details
we refer the reader to @xcite .
geometrically , the spin - weighted functions are sections of certain line bundles over the 2-sphere @xmath1 .
they have a representation in terms of ordinary functions over patches of @xmath1 which depend on the choice of coordinates and the choice of an orthonormal frame at points of the patch .
the behaviour of this representation under changes of the frame is captured by the spin - weight assigned to the function . here
, we will deal only with representations of the spin - weighted functions given in terms of the standard polar coordinates @xmath12 and the orthonormal frame defined in terms of the coordinate derivative vectors @xmath13 equivalently , these vectors can be obtained as real and imaginary parts of the complex linear combination @xmath14 which , together with its complex conjugate , satisfies the orthonormality relations @xmath15 at all points of @xmath1 covered by the polar coordinates with respect to the standard metric on @xmath1 .
every smooth function @xmath16 on @xmath1 with spin - weight @xmath6 can be expressed as : @xmath17 where the @xmath18 are the swsh , which form a complete orthonormal basis for the complex vector space of spin - weight @xmath6 functions .
recall that the action of the @xmath4 and @xmath5 differential operators on @xmath19 serves to raise and lower the spin - weight respectively ; when viewed as maps with respect to polar coordinates we have : @xmath20= \partial_\vartheta[{}_sf]-{\mathrm{i}}\csc\vartheta\,\partial_\varphi[{}_sf ] - s\cot\vartheta { } _ sf,\\\label{eq : actionethp } \eth':{}_sf&\rightarrow { } _ { s-1}\tilde{f } , & & \eth'\left[{}_sf\right]= \partial_\vartheta[{}_sf]+{\mathrm{i}}\csc\vartheta\,\partial_\varphi[{}_sf ] + s\cot\vartheta { } _ sf.\end{aligned}\ ] ] the swsh may be written explicitly as : @xmath21 where @xmath22 is the reduced wigner @xmath9-function : @xmath23 under complex conjugation the swsh satisfy : @xmath24 the swsh satisfy the orthonormality relation : @xmath25 a particularly useful property the swsh possess is that under the action of @xmath4 and @xmath5 the action of eq .
and eq . reduces to a `` ladder algebra '' : @xmath26 & = -\sqrt{(l - s)(l+s+1)}{}_{s+1}y_{lm}(\vartheta,\varphi),\\\label{eq : ladder2 } \eth'\left[{}_sy_{lm}(\vartheta,\varphi ) \right ] & = \sqrt{(l+s)(l - s+1)}{}_{s-1}y_{lm}(\vartheta,\varphi),\end{aligned}\ ] ] which we will exploit when performing decompositions of dynamical equations , allowing for a map from a pde system to an ( infinite dimensional ) ode system ( see sec .
[ sec : dirac ] ) .
in addition to this we have the commutator expression : @xmath27\,{}_sy_{lm}(\vartheta,\varphi)=-2s\,{}_sy_{lm}(\vartheta,\varphi).\ ] ] a product of two swsh with spin - weights @xmath28 and @xmath29 is a function of spin - weight @xmath30 and , therefore , it can be written as a finite linear combination of swsh @xmath31 where @xmath32 and the coefficients are related to the usual clebsch - gordan coefficients , see e.g. @xcite .
we now describe our numerical algorithm for evaluation of the forward transform @xmath33 . as a first step we introduce the notation @xmath34 which allows for the rewriting of eq . as @xcite : @xmath35 following from a factoring of rotations @xcite .
in particular , note that @xmath36 ( cf .
we defer the details of how the @xmath37 elements are calculated together with their symmetry properties to sec .
[ sec : wigrec ] .
define the functional : @xmath38:=\int^{2\pi}_0\int^\pi_0 e^{-{\mathrm{i}}m\vartheta}e^{-{\mathrm{i}}n\varphi}{}_s f(\vartheta,\varphi ) \sin\vartheta \,d\vartheta d\varphi.\ ] ] this integral can be evaluated exactly we now describe our method which is based on @xcite . combining [ eq : imnquadpr , eq : wignerddecdelta , eq : swshdefn , eq : funcs2sphtrunc ] results in : @xmath39 where @xmath40 in order to make use of existing fft ( fast fourier transform ) algorithms in the evaluation of the quadrature in eq .
, we must extend a spin - weighted function @xmath19 with data sampled on the domain @xmath41,\varphi\in[0,2\pi]\}$ ] to a @xmath42 periodic function on @xmath43 .
this is accomplished using the intrinsic symmetries of the @xmath44 together with the expansion ansatz of eq . .
for convenience we define the domains @xmath45 ( see fig .
[ fig : extdom ] ) .
we have @xmath46 upon which @xmath47 . for @xmath48 put @xmath49 .
next , define @xmath50 by @xmath51 now , define @xmath52 by @xmath53 finally , we arrive at the extended function @xmath54 defined by @xmath55 the smooth function @xmath56 is @xmath42 periodic in @xmath57 and @xmath58 , thus we may write : @xmath59 in order to evaluate we now identify @xmath60 \equiv i_{mn}\left[{}_sf(\vartheta,\varphi)\right ] & = \int_0^{2\pi } \int_0^\pi e^{-{\mathrm{i}}m\vartheta/2}e^{-{\mathrm{i}}n\varphi/2}{}_sf(\vartheta,\varphi)\sin\vartheta\,d\vartheta d\varphi,\end{aligned}\ ] ] where we can extend the indices to all integers , i.e. , @xmath61 .
this is permissible as both functions are equal in the region of integration . upon substitution with eq .
we find : @xmath62 & = \sum_{i , j=0 } { } _ s f_{ij } \;w_\vartheta(i - m)w_\varphi(j - n),\end{aligned}\ ] ] with @xmath63 equation amounts to a two - fold discrete convolution in spectral space . by the convolution theorem
, this implies that we may equivalently consider pointwise multiplication of the inverse transforms of @xmath64 and @xmath65 with @xmath66 .
let @xmath67 and @xmath68 denote the number of grid points over @xmath57 and @xmath58 on the containerization of @xmath1 that the function @xmath69 is to be sampled at . upon extension to the doubly @xmath42 periodic domain ( see fig . [
fig : extdom ] ) we work instead with the extended function @xmath56 .
furthermore , in order to simplify numerical construction of the extension , the number of samples on the extended domain is taken to be @xmath70 and @xmath71 .
correspondingly , the spatial sampling intervals are now given by @xmath72 and @xmath73 . in order to satisfy the nyquist condition ,
such that spurious aliasing does not occur , we impose @xmath74 , where @xmath7 is the harmonic that the function @xmath69 is band - limited to . under these choices
a convenient form of the quantities discussed above that is directly amenable to numerical work is given by : @xmath75 together with : @xmath76 = & \frac{1}{(n_\vartheta'-1)(n_\varphi'-1 ) } \sum_{\gamma=0}^{n_\vartheta'-2}\sum_{\varepsilon=0}^{n_\varphi'-2 } e^{-{\mathrm{i}}\mu\gamma \delta \vartheta/2 } e^{-{\mathrm{i}}\nu\varepsilon\delta\varphi/2 } \\\label{eq : imunudft } & \times \tilde{w}_\vartheta(\gamma ) \tilde{w}_\varphi(\varepsilon ) { } _ sf(\gamma,\varepsilon).\end{aligned}\ ] ] where @xmath77 , @xmath78 .
note that in this approach , we must discard all values of @xmath79 for which @xmath80 and @xmath81 are even integers .
in order to recover @xmath82 from eq .
we take @xmath83 and @xmath84 .
overall we find an algorithmic complexity of @xmath8 as the integrals @xmath79 may be evaluated exactly in @xmath85 operations by performing a two - dimensional fft and each component of @xmath86 ( as required by eq . )
can be computed using @xmath87 floating point evaluations see sec .
[ sec : wigrec ] and also @xcite .
we remark that if the analysis of strictly real data is desired then it is possible to attain a linear increase in the execution speed of transformations ( specifically the fft component by a factor of approximately @xmath0 ) .
however , as we are primarily interested in applying transformations for the solution of equations of motions that govern the dynamics of complex fields we do not explore this further .
we now describe the algorithm for evaluation of the backward ( inverse ) transform @xmath88 .
the backward spherical harmonic transform maps the expansion coefficients @xmath89 , for @xmath90 , to a function on ( a dense subset of ) @xmath1 . because we are working with band - limited functions we can , at the analytical level ,
perfectly reconstruct the original function . to this end , eq .
must be evaluated .
as the inverse transform does not contain integrals , issues of quadrature accuracy do not arise .
define : @xmath91:= { \mathrm{i}}^{s - n } \sum_{l=|s|}^l \sqrt{\frac{2l+1}{4\pi } } \delta^l_{-m , s}{}_sa_{ln}\delta^l_{-m , n}\ ] ] which allows for : @xmath92 use of eq . permits a rewriting of eq . as : @xmath91= \begin{cases } { \mathrm{i}}^{s - n } \sum_{l=|s|}^l \sqrt{\frac{2l+1}{4\pi } } \delta^l_{-m , s}{}_sa_{ln}\delta^l_{-m , n } & m\geq \frac{1}{2}\\ { \mathrm{i}}^{s - n } \sum_{l=|s|}^l \sqrt{\frac{2l+1}{4\pi } } ( -1)^{2l+s+n}\delta^l_{m , s}{}_sa_{ln}\delta^l_{m , n } & m\leq -\frac{1}{2 } \end{cases},\ ] ] which allows for a reduction in computation time . in eq .
we may use an fft directly , discarding values of @xmath69 for which @xmath93 and @xmath94 .
if the input data to the fft library is hermitean then another linear increase ( @xmath95 ) in the execution speed of a transform is possible however , this again does not change the overall algorithmic complexity @xmath8 .
here we answer the question of how to compute the @xmath86 required in the forward and backward transforms .
we will base our numerical algorithm for computation of an arbitrary @xmath86 on recursion and exploitation of symmetries .
first we note the symmetry properties ( inherited from @xmath22 ) : @xmath96 using an approach similar to @xcite one can derive an analogous trapani - navaza ( tn ) style recursion from eq . for half - integer @xmath97 values :
@xmath98 note that by making use of the symmetries provided by [ eq : delsym1,eq : delsym2,eq : delsym3 ] similar recursion relations may be constructed connecting different combinations of subscript indices .
we visualise the possible values of @xmath86 up to some maximal band - limit @xmath7 as being arranged in a square pyramidal lattice with @xmath86 values corresponding to @xmath99 and @xmath100 occupying the top - most plane ; @xmath101 and @xmath102 the next plane down and so forth . in a recursive approach
one can thus initialise with a single value of @xmath103 and with [ eq : delrectn1,eq : delrectn2,eq : delrectn5 ] ( together with symmetries ) compute those values of @xmath86 constrained to the surface of the pyramidal lattice . for each fixed @xmath104 value
those @xmath86 that occupy the interior of the pyramidal structure can be computed using eq .
and eq . ( together with symmetries ) .
we note that symmetries allow for a reduction in the total number of elements that must be calculated explicitly via recursion to @xmath105 entries for a fixed @xmath104 plane and @xmath106 total entries for a choice of maximal band - limit @xmath7 .
in our implementation we initialise with @xmath107 and iterate such that for a given @xmath104 the @xmath108 indices obey the condition @xmath109 . working with double precision arithmetic ,
we have found that the above scheme remains stable up to a band - limit of @xmath110 .
due to exponential convergence @xcite this @xmath7 will in practical situations be far above the resolution required to accurately sample fields for evolution equations and thus is not a concern for this work .
we note however , that by instead working with ratios such as @xmath111 we have verified that it is possible to construct a stable type of hybrid recursion in analogy to @xcite , for @xmath112 .
as a first check on the consistency of the half - integer swsh algorithm presented above we construct error pairs .
here one populates coefficients with random data whereupon a transformation is applied so as to construct the spatial representation of the corresponding function ; this spatial function is transformed back to coefficients and the associated error may be examined .
this procedure may be summarised as : @xmath113 the real and imaginary parts of @xmath114 we generate by sampling from the continuous uniform random distribution on the interval @xmath115 .
the numerical error associated with the specification of eq .
is shown in fig .
[ fig : fwdbwderr_sub1 ] and fig .
[ fig : fwdbwderr_sub2 ] .
we see that our implementation is indeed consistent , accurate and stable .
furthermore the scaling @xmath116 of @xmath117 ( fig .
[ fig : fwdbwderr_sub2 ] ) for the half - integer spin - weight uniform random data matches the scaling observed for integer spin - weight gaussian random data observed in @xcite .
( colour online ) averages of numerical error pairs when performing the transform procedure of eq . .
a fixed spin - weight @xmath6 is chosen and @xmath118 sets of @xmath89 are randomly generated as described in the text . in both sub - figures the band - limit of transforms is given by @xmath119 .
( a ) averaged maximum , absolute relative error @xmath120 .
the thick black line corresponds to a scaling of @xmath121 .
( b ) averaged maximum , absolute relative rms error @xmath122 .
the thick black line corresponds to a scaling of @xmath116 . in ( a ) and
( b ) : green `` @xmath123 '' corresponds to @xmath124 ; blue `` @xmath125 '' corresponds to @xmath126 ; black `` @xmath127 '' corresponds to @xmath128 .
, title="fig : " ] [ fig : fwdbwderr_sub1 ] ( colour online ) averages of numerical error pairs when performing the transform procedure of eq . .
a fixed spin - weight @xmath6 is chosen and @xmath118 sets of @xmath89 are randomly generated as described in the text .
in both sub - figures the band - limit of transforms is given by @xmath119 .
( a ) averaged maximum , absolute relative error @xmath120 .
the thick black line corresponds to a scaling of @xmath121 .
( b ) averaged maximum , absolute relative rms error @xmath122 .
the thick black line corresponds to a scaling of @xmath116 . in ( a ) and
( b ) : green `` @xmath123 '' corresponds to @xmath124 ; blue `` @xmath125 '' corresponds to @xmath126 ; black `` @xmath127 '' corresponds to @xmath128 .
, title="fig : " ] [ fig : fwdbwderr_sub2 ] the product of two spin - weighted functions @xmath129 and @xmath130 on @xmath1 is a function with spin - weight @xmath30 .
if @xmath129 and @xmath130 are comprised of a finite number of swsh then their product is also and thus by selecting a sufficiently large @xmath7 we exactly sample the resulting product function .
we can also check the property of exponential convergence ; to this end define @xmath131 , a measure of the average magnitude of coefficients at a fixed @xmath104 value .
we expect that given smooth test functions @xmath132 should behave as @xmath133 @xmath134 for large @xmath104 @xcite .
smooth half - integer spin - weighted functions may be constructed by taking a finite number of swsh and modulating by the exponential of a smooth spin - weight @xmath135 function .
introduce : @xmath136 ^ 3\\ \label{eq : testfunc6 } { } _ { -1/2 } k(\vartheta,\varphi ) & = 0.7i\,{}_{-1/2 } y_{5/2 , -1/2}(\vartheta,\varphi)+0.9\,{}_{-1/2 } y_{3/2 , 1/2}(\vartheta,\varphi)\\ \label{eq : testfunc7 } { } _ { -1/2 } \tilde{k}(\vartheta,\varphi ) & = { } _ { 0 } \tilde{g}(\vartheta,\varphi ) \ , { } _ { -1/2 } k(\vartheta,\varphi)\end{aligned}\ ] ] is depicted and corresponds to : blue `` @xmath125 '' , @xmath137 ; red `` @xmath123 '' , @xmath138 ; black `` @xmath139 '' , @xmath140 ; green `` @xmath141 '' , @xmath142 ; cyan " @xmath127 , @xmath143 .
all transforms are performed at a band - limit @xmath144 and @xmath145 respectively .
the functions @xmath137 , @xmath138 and @xmath142 are completely sampled as the average magnitude of coefficients ( @xmath132 ) falls to numerical round - off for @xmath104 well below @xmath7 . for @xmath140 and @xmath143
we observe the expected exponential decay in @xmath132 .
( see text for details ) ] in fig .
[ fig : coeffdecay ] we show @xmath132 for various combinations that test the various properties outlined .
in particular , we observe that functions comprised of a finite number of swsh are completely captured if @xmath7 is greater than the highest harmonic of the linear combination .
products of spin - weighted functions ( each individual function comprised of a finite linear combination of swsh ) also are exactly resolved to within numerical error for appropriate @xmath7 .
furthermore , we observe the crucial property of exponential convergence for smooth functions .
in order to test the swsh half - integer algorithm we now numerically solve the dirac equation on a 3-dimensional lorentz manifold @xmath146 with topology @xmath147 . since the dirac equation on a 3-dimensional manifold is less familiar than its 4-dimensional counterpart we present a detailed derivation in appendix [ sec : dirac - equation-3 ] and simply state the result here .
we consider the manifold @xmath148 with the metric @xmath149 where @xmath12 are standard polar coordinates for the 2-sphere .
the function @xmath150 is a conformal factor relating the induced metric at every instant of time @xmath151 on @xmath1 to the standard metric of the unit 2-sphere . according to appendix [ sec : dirac - equation-3 ] the dirac equation on @xmath152
is given by the following two equations for a spinor @xmath153^t$ ] @xmath154 the spinor components @xmath155 have spin - weight @xmath156 , while @xmath157 has vanishing spin - weight . while performing a spectral decomposition of variable coefficient pdes such as eq .
and eq . by assuming functions
may be expanded as in eq .
and then products decomposed as in eq .
is possible ( see @xcite for examples of such expansions in the integer swsh case ) we have found that instead it is far simpler to use a pseudo - spectral ( ps ) approach .
we outline this as follows : given the coefficients @xmath158 and @xmath159 which represent the functions @xmath129 and @xmath130 sampled at a band - limit @xmath7 then the coefficients @xmath160 corresponding to the associated point - wise product @xmath161 can be calculated by performing the transformations : @xmath162 subsequently taking the pointwise product and transforming : @xmath163 we find an approximation to an expansion utilizing eq . directly .
we emphasize that this method also easily allows one to take into account the action of the @xmath164 operators on spin - weighted functions by embedding their action as multiplication ( see eq . and eq . ) in coefficient space , together with taking account of their spin raising and lowering properties when transforms are performed .
we now recast the pde system of eq .
and eq . as an ( infinite dimensional ) ode system using the ps method .
it is convenient to define the auxiliary fields : @xmath165 where @xmath166 corresponds to a spin - weight of @xmath167 . using these fields
we find that eq . and
eq . can be written as : @xmath168 we note that in the case of @xmath169 , [ eq : diraceom1,eq : diraceom2 ] yield a constant - coefficient pde system which may be directly decomposed as : @xmath170 in this case , a solution is readily arrived at : @xmath171,\ ] ] where @xmath172 . when @xmath173 we numerically construct the spatial representation of the function using eq . with spatial sampling to coincide with the half - integer swsh transformation .
we will also find it useful to introduce the rescaled current component : @xmath174 which with the ps method can be computed using the rescaled fields @xmath175 . from eq . and
eq . the probability @xmath176 can be computed via : @xmath177 in this section we examine numerical solutions to the eom [ eq : diraceom1,eq : diraceom2 ] under three conditions : the case @xmath169 ( corresponding to a @xmath1 spatial geometry ) , where we numerically solve the system of equations given by the decomposition of eq .
; the case of a static deformation ( time - independent ) of @xmath157 ; a time - dependent @xmath157 . in the latter two cases we work with the decomposition of eq .
in each case we are free to select arbitrary initial data for @xmath155 .
as convergence tests must be performed for any numerical calculation we begin by briefly summarising the procedure we follow ( based on @xcite ) for doing so , subsequently presenting our results . aside from presenting convergence for the fields @xmath178 we also verify that the continuity equation for the current is obeyed and probability is conserved in each case examined . in using the ps approach of sec .
[ sec : psmethod ] to compute the time - evolution of @xmath155 we solve a truncated @xmath179 ode system for @xmath180 .
suppose that we use an explicit , temporal integrator with a fixed time - step @xmath181 .
the numerically calculated solution for this choice of @xmath181 we denote as @xmath182 .
assume that there exists a taylor expansion of @xmath182 about the exact solution @xmath180 with error constant terms @xmath183 : @xmath184 suppose now that the integrator is of order @xmath185 so that @xmath186 .
thus : @xmath187 by successively rescaling @xmath181 with a constant ( here @xmath0 ) and comparing @xmath188 with @xmath180 we find : @xmath189 equation may be used in the case when an analytical solution is known .
if this is not the case we may instead successively compare @xmath188 and @xmath190 which yields : @xmath191 this is known as a self - consistent convergence test ( scct ) @xcite .
note that in both cases ( eq .
and eq . )
we may also make comparisons of functions in the spatial representation by transforming from coefficient - space .
in particular , we will consider the maximum , absolute , relative error metric : @xmath192 which allows for the comparison of the two solutions @xmath193 at sampling nodes @xmath194 , @xmath195 .
we measure error in this manner such that convergence properties of solutions and any potential instability that may result from the swsh transformation algorithm may be simultaneously inspected . having described the methods we use to test convergence of solutions
we now examine some example cases .
we fix the mass - parameter to be @xmath196 at the outset and choose a temporal range of @xmath197 $ ] throughout .
these choices have been made so as to allow for observation of multiple oscillations of @xmath180 modes ( see eq . ) while simultaneously avoiding exact integer multiple of the frequencies @xmath198 .
note that all specified initial conditions @xmath199 are constructed with data as indicated below and then the corresponding coefficients @xmath200 rescaled by an overall factor so that @xmath201 is normalized to @xmath202 at @xmath203 .
for each numerical evolution the temporal integrator we choose is the standard , explicit runge - kutta @xmath204th order method ( rk4 ) ; thus @xmath205 in eq .
. for @xmath169 we select the smooth initial data : @xmath206 with @xmath143 given by eq . and @xmath140 given by eq .
the system to solve is thus specified by eq .
and eq . upon mapping the latter @xmath207 to coefficient space using the swsh transformation algorithm .
the results of convergence tests ( based on eq . with the analytical solution of eq . )
of our numerical solution are shown in fig .
[ fig : eomf1fieldcvgce ] .
we find excellent agreement with the expected @xmath204th order convergence in time for initial data and parameters specified .
in particular , we see that the numerical error of the solution is dominated by the temporal discretization .
( spatial representation ) for the case @xmath169 .
@xmath208 is displayed at @xmath209 for a number of steps @xmath210 .
@xmath211 is calculated from eq . and the spatial representation found by the swsh transformation .
blue ( dashed ) , `` @xmath139 '' : @xmath212 , @xmath213 ; green ( dashed ) , `` @xmath141 '' : @xmath212 , @xmath214 ; red ( dashed ) , `` @xmath125 '' : @xmath215 , @xmath213 ; cyan ( dashed ) , `` @xmath216 '' : @xmath215 , @xmath214 ; black ( dashed ) , `` @xmath127 '' : @xmath217 , @xmath213 ; magenta ( dashed ) , `` @xmath218 '' : @xmath217 , @xmath214 ; blue ( solid ) , `` @xmath219 '' : @xmath220 , @xmath213 ; green ( solid ) , `` @xmath123 '' : @xmath220 , @xmath214
. the thick black line corresponds to an expected convergence of @xmath204th order in time .
( see text for discussion ) ] in fig .
[ fig : contin_s2_sub1 ] and fig .
[ fig : prob_s2_sub2 ] we provide representative examples of the maximum absolute value of the @xmath221-divergence @xmath222 ( expected to be @xmath135 by eq . ) and probability conservation @xmath223 ( again expected to be @xmath135 ) at a fixed number of time - steps ( temporal discretization ) .
this allows for the examination of the effect of varying the band - limit @xmath7 on the evolution of the smooth initial data of eq . .
( colour online ) test of the continuity equation and probability conservation for the case @xmath169 .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . )
( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : contin_s2_sub1 ] ( colour online ) test of the continuity equation and probability conservation for the case @xmath169 .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . ) ( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : prob_s2_sub2 ] in fig .
[ fig : contin_s2_sub1 ] we observe that the numerical error is greatest for @xmath144 and minimizes for @xmath212 the reason for this can be explained as follows : from the evolution equation ( eq . ) we see that coupling does not occur between distinct @xmath227 modes , only between fixed @xmath227 of @xmath228 .
hence we see that if for given @xmath227 it is the case that @xmath229 then to within numerical tolerance these coefficients should remain @xmath135 over the course of the evolution .
thus for a finite @xmath7 we effectively provide a choice of initial data which can be viewed as exact at the specified @xmath7 .
the error arising from the spectral components of @xmath155 can hence be entirely attributed to that of the swsh transformation itself , which scales with increasing @xmath7 as shown in fig . [
fig : fwdbwderr ] .
this behaviour can also be observed in fig .
[ fig : eomf1fieldcvgce ] by comparing the numerical error associated with components @xmath155 at a fixed @xmath230 for differing @xmath7 . in fig .
[ fig : contin_prob_s2 ] we examine the conservation of probability with the quantity @xmath223 where @xmath201 is calculated according to eq . and
as @xmath169 we may avoid rescaling to @xmath231 as is required in more general cases .
as no swsh transformations are required to compute this quantity once @xmath180 is known we find that upon changing @xmath7 the previous associated numerical error is not accumulated and the results for @xmath223 at differing @xmath7 coincide . for a time independent @xmath157 we select the ( finite @xmath7 ) initial data @xmath232 : @xmath233 @xmath234 and @xmath157 we choose as : @xmath235 together with the above specifications we make use of auxiliary fields as defined in eq . with @xmath236 and the eom decomposition of eq . .
in this case instead of comparing with an analytical solution we perform a self - consistent convergence based on eq . .
the result of this is presented in fig .
[ fig : eomfdiffieldcvgce ] .
( spatial representation ) for the case of static @xmath157 ( eq . ) .
@xmath237 is displayed at @xmath209 for a number of steps @xmath210 .
blue ( dashed ) , `` @xmath139 '' : @xmath212 , @xmath213 ; green ( dashed ) , `` @xmath141 '' : @xmath212 , @xmath214 ; red ( dashed ) , `` @xmath125 '' : @xmath215 , @xmath213 ; cyan ( dashed ) , `` @xmath216 '' : @xmath215 , @xmath214 ; black ( dashed ) , `` @xmath127 '' : @xmath217 , @xmath213 ; magenta ( dashed ) , `` @xmath218 '' : @xmath217 , @xmath214 ; blue ( solid ) , `` @xmath219 '' : @xmath220 , @xmath213 ; green ( solid ) , `` @xmath123 '' : @xmath220 , @xmath214 . the thick black line corresponds to an expected convergence of @xmath204th order in time .
( see text for discussion ) ] we observe the expected @xmath204th order convergence in time for the numerical solution @xmath238 at the band limits tested . in fig .
[ fig : contin_diffeo_sub1 ] and fig .
[ fig : prob_diffeo_sub2 ] we show representative examples of how the numerical solution obeys the continuity equation and probability conservation respectively can grow dramatically - during the calculation we thus only retain the coefficients @xmath180 at evenly interspersed points on the temporal grid as presented in figures . ] .
note that in order to calculate @xmath201 for this case we rescale fields as in eq . and use the full prescription of the ps method ( see sec .
[ sec : psmethod ] ) .
in contrast to the @xmath169 case we now have a variable coefficient eom and coupling between distinct @xmath227 modes of @xmath228 occurs , thus sampling at a sufficiently high band - limit @xmath7 in addition to choosing a sufficiently fine temporal grid is required in order to resolve conservation properties accurately .
( colour online ) test of the continuity equation and probability conservation for the case of static @xmath157 ( eq . ) .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . ) ( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : contin_diffeo_sub1 ] ( colour online ) test of the continuity equation and probability conservation for the case of static @xmath157 ( eq . ) .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) . ( a )
maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . )
( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : prob_diffeo_sub2 ] finally we consider a time - dependent @xmath157 .
the selection we make is linear interpolation in time between initial @xmath239 and final @xmath240 static deformations of @xmath1 .
that is , @xmath241 and @xmath242 .
the ( real ) functions @xmath243 and @xmath244 we choose to have non - zero @xmath245 and @xmath246 coefficients : @xmath247 initial data @xmath232 is selected as in eq .
and eq . .
together with the above specifications we again make use of auxiliary fields as defined in eq .
( note however that @xmath248 ) and the eom decomposition of eq . .
we again perform a self - consistent convergence based on eq . .
the result of this is presented in fig .
[ fig : eomftdepfieldcvgce ] .
( spatial representation ) for the case of time - dependent @xmath157 ( eq . ) .
@xmath237 is displayed at @xmath209 for a number of steps @xmath249 .
blue ( dashed ) , `` @xmath139 '' : @xmath212 , @xmath213 ; green ( dashed ) , `` @xmath141 '' : @xmath212 , @xmath214 ; red ( dashed ) , `` @xmath125 '' : @xmath215 , @xmath213 ; cyan ( dashed ) , `` @xmath216 '' : @xmath215 , @xmath214 ; black ( dashed ) , `` @xmath127 '' : @xmath217 , @xmath213 ; magenta ( dashed ) , `` @xmath218 '' : @xmath217 , @xmath214 ; blue ( solid ) , `` @xmath219 '' : @xmath220 , @xmath213 ; green ( solid ) , `` @xmath123 '' : @xmath220 , @xmath214 .
the thick black line corresponds to an expected convergence of @xmath204th order in time .
( see text for discussion ) ] once again , excellent agreement with the expected @xmath204th order convergence is observed with respect to the temporal scheme .
we are again in a situation with a variable coefficient eom , however now it also becomes non - autonomous .
the continuity equation , together with probability conservation must be obeyed in this case also , and indeed we find that provided a sufficiently high band - limit @xmath7 is chosen we may resolve these stated properties in our numerical solution ( see fig . [
fig : contin_tdep_sub1 ] and fig .
[ fig : prob_tdep_sub2 ] ) .
( colour online ) test of the continuity equation and probability conservation for the case of time - dependent @xmath157 ( eq . ) .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . ) ( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : contin_tdep_sub1 ] ( colour online ) test of the continuity equation and probability conservation for the case of time - dependent @xmath157 ( eq . ) .
we fix the time - step @xmath181 to be @xmath224 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath209 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 ( based on eq . ) ( b ) conservation of probability @xmath201 with time ( based on eq . ) in both sub - figures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : prob_tdep_sub2 ] having analysed the numerical properties of our implementation of a pseudo - spectral method for the dirac equation we now turn to potential applications .
we consider a @xmath250 dimensional analogue of the usual friedmann - robertson - walker ( frw ) space - time in comoving coordinates . the particular physical scenario
we wish to model is an imploding universe with pressure @xmath251 and density @xmath252 equal to zero .
we take the scale factor to be @xmath253 .
thus , we see that at @xmath203 the background spatial geometry coincides with that of @xmath1 upon which initial data for the dirac equation must be provided . for many practical purposes it is sufficient to consider an initial gaussian state when discussing the dynamics of the dirac equation . for a @xmath254-sphere embedded in @xmath255 the von mises - fisher distribution serves as an analogue to the planar gaussian distribution @xcite .
the probability density function is given by : @xmath256 where @xmath257 , @xmath258 and @xmath259 is the modified bessel function of the first kind .
the parameter @xmath260 may be thought of as analogous to the reciprocal of the variance of the gaussian distribution and @xmath261 as the mean direction about which the points @xmath262 cluster .
we choose initial data with average momentum @xmath135 where @xmath263 and @xmath264 . for the mass parameter
we take @xmath196 and evolve eq . and
eq . on the temporal range
@xmath265 $ ] . as before we use explicit rk4 as the temporal integrator and upon performing scct ( fig .
[ fig : eomfrwfieldcvgce ] ) find @xmath204th order convergence in time as expected .
( spatial representation ) for the case of @xmath266 .
@xmath237 is displayed at @xmath267 for a number of steps @xmath268 .
blue ( dashed ) , `` @xmath139 '' : @xmath212 , @xmath213 ; green ( dashed ) , `` @xmath141 '' : @xmath212 , @xmath214 ; red ( dashed ) , `` @xmath125 '' : @xmath215 , @xmath213 ; cyan ( dashed ) , `` @xmath216 '' : @xmath215 , @xmath214 ; black ( dashed ) , `` @xmath127 '' : @xmath217 , @xmath213 ; magenta ( dashed ) , `` @xmath218 '' : @xmath217 , @xmath214 ; blue ( solid ) , `` @xmath219 '' : @xmath220 , @xmath213 ; green ( solid ) , `` @xmath123 '' : @xmath220 , @xmath214 .
the thick black line corresponds to an expected convergence of @xmath204th order in time .
( see text for discussion ) ] once again the continuity equation must hold and probability must be conserved .
these properties are checked for the present numerical calculation in fig .
[ fig : contin_frw_sub1 ] and fig .
[ fig : prob_frw_sub2 ] . for each chosen band - limit
we observe growth in numerical error of several orders of magnitude as @xmath269 .
this is reasonable when we take into account that the spatial geometry is shrinking to a point ( @xmath270 as @xmath271 ) .
( colour online ) test of the continuity equation and probability conservation for the case of @xmath266 .
we fix the time - step @xmath181 to be @xmath272 .
@xmath225 indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath267 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 .
( b ) conservation of probability @xmath201 with time . in both subfigures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : contin_frw_sub1 ] ( colour online ) test of the continuity equation and probability conservation for the case of @xmath266 .
we fix the time - step @xmath181 to be @xmath272 . @xmath225
indicates the number of time - steps taken from @xmath203 ( to the maximum @xmath267 ) .
( a ) maximum over all sampled nodes in the spatial representation of @xmath226 .
( b ) conservation of probability @xmath201 with time . in both subfigures : blue `` @xmath125 '' : @xmath212 ; red `` @xmath219 '' : @xmath215 ; green `` @xmath123 '' : @xmath217 ; black `` @xmath127 '' : @xmath144 .
( see text for discussion ) , title="fig : " ] [ fig : prob_frw_sub2 ] we now fix the band - limit as @xmath144 together with temporal grid @xmath273 . for the aforementioned resolutions snapshots of the dynamics
are displayed on an aitoff - hammer projection @xcite in fig .
[ fig : diracfrwsoln ] . in fig .
[ fig : diracfrwsoln_a ] we see that the probability density @xmath274 is initially gaussian in character and subsequently due to the rotational symmetry of the initial state about the mean direction @xmath261 we observe evolution towards a ring - like structure indicating dynamics governed by dispersion ( fig .
[ fig : diracfrwsoln_b ] ) . in fig .
[ fig : diracfrwsoln_c ] we see evolution towards the antipode of @xmath261 , however we find that this does not coincide with a reconstruction of the initial condition in the limit @xmath275 ( fig .
[ fig : diracfrwsoln_d ] ) . observe that the average amplitude of @xmath276 increases dramatically ; this is to be expected due to conservation of probability @xmath201 and the form of the volume element for this particular geometry used in its calculation ( eq . ) .
[ fig : diracfrwsoln_a ] ( colour online ) aitoff - hammer projection of the probability density @xmath274 associated with the @xmath250 dimensional dirac equation on frw background .
note that during the course of the time evolution the spatial radius @xmath277 of the spherical projection in each sub - figure is distinct , furthermore the colour scaling also varies .
see text for discussion .
, title="fig : " ] [ fig : diracfrwsoln_b ] ( colour online ) aitoff - hammer projection of the probability density @xmath274 associated with the @xmath250 dimensional dirac equation on frw background .
note that during the course of the time evolution the spatial radius @xmath277 of the spherical projection in each sub - figure is distinct , furthermore the colour scaling also varies .
see text for discussion .
, title="fig : " ] [ fig : diracfrwsoln_c ] ( colour online ) aitoff - hammer projection of the probability density @xmath274 associated with the @xmath250 dimensional dirac equation on frw background .
note that during the course of the time evolution the spatial radius @xmath277 of the spherical projection in each sub - figure is distinct , furthermore the colour scaling also varies .
see text for discussion .
, title="fig : " ] [ fig : diracfrwsoln_d ] ( colour online ) aitoff - hammer projection of the probability density @xmath274 associated with the @xmath250 dimensional dirac equation on frw background . note that during the course of the time evolution the spatial radius @xmath277 of the spherical projection in each sub - figure is distinct , furthermore the colour scaling also varies .
see text for discussion .
, title="fig : " ]
we have presented a new spectral algorithm for half - integer spin - weighted functions on @xmath1 based on spin - weighted spherical harmonics ( swsh ) by extending the method for the integer case presented in @xcite .
our implementation of this spectral algorithm shows excellent agreement with the theory we have presented .
indeed we find that the numerical error scaling comparable to the integer swsh algorithm @xcite and that exponential convergence properties are displayed as anticipated .
furthermore , the expected algorithmic complexity of @xmath8 is retained .
in addition , we have outlined how one may construct the @xmath11 dirac equation on a curved space - time , adapting the result to the @xmath4-formalism and for a geometry with spatial topology of @xmath1 .
viewing the dirac equation as an ivp we demonstrated how a pseudo - spectral approach can be combined with the half - integer swsh algorithm and analysed several different situations with distinct spatial geometries , examples including linear interpolation between distinct spatial configurations with time and the case of an imploding @xmath11 frw - like model . the numerical solutions that we constructed
obey anticipated convergence rates for the temporal solver used ( explicit rk4 ) and physical invariants ( current continuity and probability ) are shown to be preserved during the time - evolution to an excellent degree .
jf is grateful to g. sparling for several enlightening discussions .
this research was partly supported by the marsden fund of the royal society of new zealand .
let @xmath278 be a @xmath279-dimensional real vector space equipped with a quadratic form @xmath280 with signature @xmath281 , i.e. , @xmath280 can be represented in diagonal form as @xmath282 .
let @xmath283 be the clifford algebra associated to @xmath284 .
this is the unique algebra with unit @xmath285 and so called structure map @xmath286 such that for every @xmath287 the relation @xmath288 holds . let @xmath289 be a basis of @xmath278 and define @xmath290 and @xmath291 .
then , by polarization of the defining relation , one finds the well - known clifford - dirac anti - commutator relations @xmath292 irreducible representations @xmath293 of the clifford algebras have dimensions @xmath294 for @xmath279 even and @xmath295 for @xmath279 odd .
it is well known ( see @xcite ) that these representations can be constructed recursively from the lower dimensional ones .
the representation space @xmath293 of the clifford algebra @xmath283 is called the spin space . depending on the particular case ,
the spin space is equipped with certain invariant structures , see @xcite,@xcite,@xcite for detailed discussions .
the commutators of the generators @xmath296 } = \frac12 \left[\gamma_a,\gamma_b\right]$ ] generate a lie algebra with commutation relations @xmath297
= \eta_{cb}\gamma_{ad } - \eta_{ca}\gamma_{bd } - \eta_{db}\gamma_{ac } + \eta_{da}\gamma_{bc}.\ ] ] these are the commutation relations for the lie algebra @xmath298 of the ( pseudo- ) orthogonal group @xmath299 associated with the bilinear form @xmath280 .
this lie algebra acts on the subspace @xmath300\subset \mathfrak{c}(\mathbb{v},\eta)$ ] as follows : for any skew bi - vector @xmath301 and every vector @xmath302 we define @xmath303 and @xmath304 , then the action of @xmath298 is via commutation in @xmath283 @xmath305 \to \gamma[{\mathbb{v}}],\quad(\gamma(w ) , \gamma(v ) ) \mapsto \gamma(w)\gamma(v ) - \gamma(v)\gamma(w).\ ] ] we can see the corresponding action on @xmath306 from the explicit calculation @xmath307 thus , the action on @xmath300 $ ] corresponds to the linear mapping defined by @xmath308 on @xmath306 , which is clearly anti - symmetric with respect to @xmath280 .
the same lie algebra also acts on the spin - space @xmath293 via the representation of the clifford algebra @xmath309 in order to make use of spinors on the manifold @xmath152 we associate at every point @xmath310 the clifford algebra @xmath311 and its representation space which we denote by @xmath312 .
the rigorous construction is invariantly described in terms of ( associated ) vector bundles but we will not go into the details here . instead we refer to @xcite for a complete description .
the collection of all spin spaces @xmath312 forms the spin bundle @xmath313 over @xmath152 and sections of this bundle are referred to as spinor fields or simply spinors .
similarly , we obtain the clifford bundle as the collection of all clifford algebras @xmath311 .
the structure maps defined at every point @xmath314 yield a bundle map @xmath315 from the tangent bundle @xmath316 to the clifford bundle , assigning to every tangent vector @xmath317 an element @xmath318 in the clifford algebra at @xmath314 .
the connection @xmath319 on @xmath152 can be lifted to a connection also denoted by @xmath319 on @xmath313 ( see @xcite ) .
it can be uniquely characterized by the fact that the structure map @xmath315 and the invariant structures on the spin spaces are covariantly constant . with this connection
one can define the dirac operator on @xmath152 as follows : we pick a basis @xmath289 of the tangent space @xmath320 at each @xmath310 . now
the anti - commutator relations are @xmath321 which hold at every @xmath310 with the appropriate definition of @xmath322 and @xmath323 .
the dirac operator @xmath324 acting on spinor fields @xmath325 is defined by @xmath326 in this paper we assume the vector space @xmath306 to have dimension 3 and to be equipped with a metric @xmath280 with signature @xmath327 .
the corresponding clifford algebra @xmath328 has a 4-dimensional real representation @xcite , which may conveniently be described within @xmath329 as the real algebra generated by the ` dirac matrices ' @xmath330 the spin space is @xmath331 ( regarded as a 4-dimensional real vector space ) and carries a sesquilinear form ( hermitean inner product ) defined by @xmath332 where @xmath333 is the hermitean conjugate of @xmath325 . for given spinors @xmath334 $ ] and @xmath335 $ ]
this product is @xmath336 the generators are symmetric with respect to this product , i.e. , we have @xmath337 given two spinors @xmath338 and @xmath325 there is a naturally defined covector @xmath339 with components @xmath340 . when @xmath341 , then this covector is @xmath342.\ ] ] hence , it is real and null whenever @xmath343 .
choosing an orthonormal frame @xmath346 on @xmath152 we can express the dirac operator on @xmath152 with respect to the chosen frame explicitly . to this end
we need the ricci rotation coefficients for the frame defined by @xmath347 the condition that the structure map be covariantly constant translates into the equation @xmath348 where the @xmath349 , which depend on the point @xmath310 , are endomorphisms of the spin space @xmath312 .
more precisely , they are linear combinations of the infinitesimal generators of @xmath344
. this equation can be solved uniquely for the @xmath350 and yields @xmath351 the connection defined in this way also leaves the complex structure and the sesquilinear product invariant . with these ` spin coefficient ' matrices we can express the dirac operator acting on an arbitrary spinor @xmath325 as @xmath352 the dirac equation on @xmath152 can be obtained from a variational principle . for
any given spinor field @xmath325 with compact support on @xmath152 we can write down the action functional @xmath353:=\int_{{\mathscr{m}}}{\mathrm{i}}\langle\psi,{\mathscr{d}}\psi\rangle + \mu \,\langle\psi,\psi\rangle\ , d{{\mathscr{m}}},\ ] ] where here and later on @xmath354 denotes the invariant volume form on a manifold @xmath152 . this defines a real number since the imaginary part can be rewritten as @xmath355 variation with respect to @xmath325 yields the dirac equation @xmath356 to every solution @xmath325 of the dirac equation the covector @xmath357 defines a conserved current @xmath358 denoted by @xmath359 taking the divergence of this equation we find @xmath360 with this current we can define a conserved quantity in the usual way .
let @xmath151 be a time coordinate for @xmath152 defined on an interval @xmath361 $ ] for some arbitrary @xmath362 and let @xmath363 be a 2-dimensional manifold .
we assume that @xmath364 is embedded into @xmath152 as a 3-dimensional submanifold @xmath365 via the embedding @xmath366
. then @xmath367 embeds @xmath363 as a space - like hyper - surface @xmath368 into @xmath152 .
let @xmath369 be the boundary of @xmath363 , which @xmath370 embeds as a 1-dimensional submanifold @xmath371 into @xmath152 .
now the boundary of @xmath365 is @xmath372 where @xmath373 .
integrating the divergence equation over @xmath365 and using the generalised stokes theorem we obtain @xmath374 where we denote the hyper - surface forms on the boundary @xmath375 by @xmath376 .
introducing the future - pointing time - like normal @xmath377 to the hyper - surfaces @xmath368 and the outward normal @xmath378 to @xmath379 we can write @xmath380 defining the scalar `` charge '' ( also interpreted as probability ) @xmath201 and its flux @xmath381 by @xmath382 we can write the balance law @xmath383 which holds for arbitrary values of @xmath362 . with appropriate boundary conditions ( for instance , when @xmath363 has no boundary as we assume below ) for the dirac field one can make the flux vanish so that the balance law turns into the conservation equation for @xmath201 @xmath384 we now specialize to the case @xmath148 with the metric @xmath385 where @xmath12 are standard polar coordinates for the 2-sphere . the function @xmath150 is a conformal factor relating the induced metric at every instant of time @xmath151 on @xmath1 to the standard metric of the unit 2-sphere .
we choose the orthonormal frame as @xmath386 the non - vanishing ricci rotation coefficients are @xmath387 from these we obtain the non - vanishing ` spin coefficient ' matrices @xmath388 in order to make use of the spin - weighted formalism we need to identify the spin - weights of the quantities involved . to this end
we consider the infinitesimal rotations of tangent vectors to the sphere as induced by the @xmath389-lorentz group @xmath299 at every point on @xmath152 .
consider the infinitesimal rotation given by @xmath390 acting as follows on the two frame vectors on the sphere
@xmath391 = \gamma_2 , \quad \gamma(e_2 ) = \gamma_2 \mapsto \frac12 \left[\gamma_{12},\gamma_2 \right ] = -\gamma_1.\ ] ] in terms of the complex null - vector @xmath392 this is @xmath393 so @xmath394 generates the frame rotations @xmath395 .
but it also acts on the spin - space at each point : for every spinor @xmath396^t$ ] we have @xmath397 -\psi_1\\ \psi_2 \end{bmatrix*}.\ ] ] therefore , assigning spin - weight @xmath398 to @xmath399 results in the spin - weights @xmath400 and @xmath401 for @xmath402 and @xmath403 , respectively .
henceforth , we denote the spinor @xmath325 therefore by @xmath153^t$ ] .
since @xmath404 is a scalar function it has spin - weight 0 . the probability or charge density is @xmath406 we take @xmath365 in the form @xmath407 with hyper - surfaces of constant time @xmath408 then the ` charge ' integral as defined in yields the expression @xmath409 since @xmath408 has no boundary , implies that for every @xmath410 @xmath411 | we present a new spectral scheme for analysing functions of half - integer spin - weight on the @xmath0-sphere and demonstrate the stability and convergence properties of our implementation .
the dynamical evolution of the dirac equation on a manifold with spatial topology of @xmath1 with a pseudo - spectral method is also demonstrated . |
friction phenomena take place across a broad range of time and length scales , from microscopic atomistic processes , as in the gliding motion of a nanocluster or a nanomotor @xcite , up to extremely macroscopic instances , as in fault dynamics and earthquake events @xcite . due to the ubiquitous nature of mechanical dissipative processes and the enormous practical relevance ,
friction has been investigated over the centuries . while the empirical laws of macroscopic friction are well known @xcite , the fundamental understanding of the tribological phenomena at the microscopic scales is still lacking from many points of view .
the basic difficulty of friction is intrinsic , involving the dissipative dynamics of large systems , often across ill - characterized interfaces , and generally violent and nonlinear .
the severity of the task is also related to the experimental difficulty to probe systems with many degrees of freedom under a forced spatial confinement , that leaves very limited access to probing the buried sliding interface . thanks to remarkable developments in nanotechnology , new inroads are being pursued and new discoveries are being made . at the nanometer scale ,
state - of - the - art ultra - high - vacuum systems and local probe studies show a dynamical behavior which is often significantly different , not just quantitatively but even qualitatively , from the ones observed in macroscopic tribology . bridging
the gap among the different length scales in tribological systems still remains an open challenge .
the phenomenological descriptions that apply to macroscopic friction can not yet be derived from the fundamental atomic principles and the interplay of processes occurring at the molecular level .
nanofriction is in somewhat better shape .
together with the current experimental possibility to perform well - defined measurements on well - characterized materials at the fundamental microscopic level of investigation of the sliding contacts , advances in the computer modeling of interatomic interactions in materials science and complex systems encompass molecular - dynamics ( md ) simulations of medium to large scale for the exploration of the tribo - dynamics with atomic resolution @xcite . despite the benefits brought about by numerical simulations of realistic 3d sliding systems ,
the resulting proliferation of detailed complex data , and the requirement of always - growing computational efforts have stimulated , in parallel , the concurrent search for simpler modeling schemes , such as , e.g. , generalized prandtl - tomlinson ( pt ) , frenkel - kontorova ( fk ) , for nanofriction , and of burridge - knopoff and earthquake - like models , for mesoscale and macroscale friction , suitable to describe the essence of the physics involved in highly nonlinear and non - equilibrium tribological phenomena in a more immediate fashion .
here we discuss current progress and open problems in the simulation and modeling of tribology at the microscopic scale , and its connection to the macroscale .
neither the pt model , described in detail in several surveys @xcite with several applications to concrete tip - based physical systems , nor the phenomenological approach based on the rate - and - state models @xcite will be considered here . with a view to emphasize the role of nonlinearity the present topical review will restrict to the following theoretical approaches to sliding friction .
section [ linear : sec ] revises the simple case of near - equilibrium linear friction in classical mechanics . section [ fk : sec ] focuses on nonlinearity in crystal sliding in the framework of the fk model and its generalizations .
atomistic models and md nanofriction simulations are presented in sec .
[ md : sec ] .
mesoscopic multicontact earthquake - like models are finally examined in sec .
[ multicontact : sec ] .
statistical mechanics accounts for the intimate mechanism of friction : a system at equilibrium has its kinetic energy uniformly distributed among all its degrees of freedom .
a sliding macroscopic object clearly is not at equilibrium : one of its degrees of freedom ( the center - of - mass motion ) has far more kinetic energy than any other .
the tendency of the system toward equilibrium will lead to the transfer of energy from that degree of freedom to all other ones : as a result the macroscopic object will slow down and its energy will be transferred to the disordered motion of the other degrees of freedom , resulting in warming up .
this is all sliding friction really is : the tendency of systems toward equilibrium energy equipartitioning among many interacting degrees of freedom .
thus , in the course of friction under an applied external force , energy is reversed into the system in the form of frictional heat .
the frictional heat is generally dissipated by some form of heat bath , such as that provided by a thermostat at temperature @xmath0 . in a frictional steady state , caused for example by submitting a slider to an external force @xmath1 , the slider dissipates energy to the bath , and therefore does not accelerate indefinitely it reaches instead a steady state characterized by an average drift velocity @xmath2 . when both @xmath2 and @xmath1 are infinitesimal , the relationship between the two quantities is linear , @xmath3 in this so - called `` viscous friction '' the proportionality constant @xmath4 is the linear friction coefficient .
it is known from classical statistical mechanics , for example of brownian motion as described by the langevin equation , that for linear friction systems which obey eq .
( [ erio-1 ] ) , the einstein relationship @xmath5 is generally valid , connecting the friction coefficient @xmath4 , which measures dissipation , to the diffusion coefficient @xmath6 , which measures fluctuations .
this expresses the fluctuation - dissipation theorem of linear , viscous friction . to simulate the classical motion of a macroscopic object moving in contact with an equilibrium bath such as the molecules of a gas or a liquid , or the phonons of a solid
, the standard implementation @xcite requires adapting newton s equations of motion with the addition of a damping force @xmath7 plus a random force @xmath8 .
the damping force represents the transfer of energy from the macroscopic object of mass @xmath9 to the heat bath , i.e. dissipation : @xmath10 this formula , equivalent to eq .
( [ erio-1 ] ) , assumes that the deviation from equilibrium is small , so that _ linear response _ holds : the restoring stokes force is linear in the perturbing velocity , and acts opposite to it to restore the @xmath11 equilibrium regime .
this linear dependence is purely the lowest order term in a taylor expansion : there is no reason to expect the linear relation ( [ eq_langevin ] ) to extend to large velocity , and indeed e.g. the drag friction of speeding objects in gases is well known to follow rayleigh s quadratic dependence on speed @xmath12 .
the random - force term represents statistically the `` kicks '' that the objects experiences due to its interaction with the thermal bath . in the frame of reference of the thermal bath @xmath13 of course .
the random term is the result of many very frequent collisions events , resulting in random forces uncorrelated with themselves except over very short time spans .
more precisely , we assume there is some maximum time @xmath14 beyond which any correlation vanish : @xmath15 in addition , the assumption of thermal equilibrium ensures us that the bath is in a steady state , so that @xmath16 is independent of @xmath17 , and depends on @xmath18 only : the statistical properties of the random force are constant in time . in most practical situations
, one can safely ignore the dynamics over a time scale of the order of @xmath14 or shorter .
we are interested instead in the integral effect of @xmath19 over some time period @xmath17 that is long compared to @xmath14 .
we can break up that integral into many pieces , each covering a duration @xmath14 : @xmath20 this integral is then a sum of many independent random terms , each drawn from the same distribution whose only relevant property is that it has zero mean value . as a result of the central - limit theorem
, the total integral obeys a gaussian distribution with null mean , and whose standard deviation scales with the number of terms in the sum , i.e. @xmath21 . by taking the equation of motion in the absence of any external driving , @xmath22 and integrating it in time
we obtain @xmath23 the first term at the right - hand side of eq .
( [ eq_diffint ] ) becomes negligible for a time @xmath24 , long enough for the object to equilibrate with the thermostat and lose memory of its initial condition . in this large-@xmath17 limit , by taking the square module of eq .
( [ eq_diffint ] ) and executing the ensemble average , we have @xmath25 the term at the left side multiplied by @xmath26 yields the average kinetic energy , which by standard equipartition needs to equal @xmath27 . by rearranging the exponentials on the right - hand side and substituting @xmath28 and @xmath29
, we obtain : @xmath30 where we have used the steadiness of the stochastic process discussed after eq .
( [ bathuncorr ] ) . assuming , as is commonly the case , that the autocorrelation time @xmath14 of the random term is short compared to @xmath31 , the integrand of eq .
( [ equipart ] ) has @xmath32 in all region @xmath33 of delays where the factor @xmath34 is significantly different from @xmath35 .
this observation further simplifies eq .
( [ equipart ] ) to the _ fluctuation - dissipation _
relation @xmath36 this expression draws an explicit link between the autocorrelation amplitude of the fluctuations and the product of the dissipation coefficient and thermostat temperature . note that the integral in eq .
( [ fluctdiss ] ) depends on both the amplitude of fluctuations @xmath37 and the time over which they remain self - correlated .
the effect on the mesoscale object increases if the random force is larger and/or if the time interval over which @xmath37 pushes in the same direction before changing is longer .
the relation ( [ fluctdiss ] ) can be equally well satisfied by weaker random forces acting for longer correlation times or stronger forces with shorter @xmath14 , which thus lead to the same statistical effects . as @xmath14 is the shortest time scale around , for all practical purpose
one can satisfy eq .
( [ fluctdiss ] ) assuming a sort of @xmath38 limit : @xmath39 providing a simple recipe for computer simulations . for simulations of models such as the pt or fk ones , the phenomenological degrees of freedom
are often coupled to a langevin thermostat of this kind , implying that each degree of freedom is actually coupled to a vast number of other bath degrees of freedom . even in md simulations of atomic - scale friction ,
langevin thermostats are applied to all or a part of the atoms involved @xcite .
of course , this approach is not rigorous , since the relevant particles colliding with each given simulated atom are already all included in the conservative and deterministic forces explicitly accounted for by the `` force field '' .
the langevin approach is quite accurate to describe small perturbations away from equilibrium , but it may fail quite badly in the strongly out - of - equilibrium nonlinear phenomena which are the target of the present paper . in the rest of this review we will deal with nonlinear frictional phenomena , which deviate violently from linearity and near - equilibrium , and where therefore eqs .
( [ erio-1 ] ) and ( [ erio-2 ] ) do not generally apply .
as it has , surprisingly , only been realized in the last few decades , even arbitrarily violently non - equilibrium and nonlinear driven phenomena adhere to an extension of the fluctuation - dissipation theorem .
that is the jarzynski ( or jarzynski crooks ) relation @xcite , whose simplest form can be briefly summarized as follows .
suppose starting from a system in state a at temperature @xmath0 , and apply an external force of arbitrary form and strength causing it to evolve , for example to slide , to another state b ; assume for simplicity b to be also a state of equilibrium .
call @xmath40 the work done by the external force , and call @xmath41 the difference of equilibrium free energy between the states b and a. clearly , @xmath42 must be valid , because some work will be wasted in going from a to b , unless that was done infinitely slowly ( adiabatically ) .
suppose now repeating the forced motion a@xmath43b many times .
each time , @xmath40 will be different .
the jarzynski equality states that @xmath44 it can be shown that in near - equilibrium conditions , eq .
( [ erio-4 ] ) is completely equivalent to the fluctuation - dissipation theorem .
the beauty of it is however that eq .
( [ erio-4 ] ) is totally general .
one particular case is useful in order to underline its far - reaching power .
suppose to take b = a , that is a final state identical to the starting one .
in that case @xmath45 this equation appears at first sight impossible to satisfy , because surely all @xmath46 : all forced motion must cost work .
the answer is that the probability @xmath47 is indeed a distribution centered around a positive @xmath48 , but with a nonzero tail extending to @xmath49 .
this tail represents rare events where work is gained rather than spent we can think of them as a sort of `` free lunches '' .
jarzynski s theorem requires that `` free lunches '' must occur precisely in such a measure to satisfy eq .
( [ erio-5 ] ) .
however it is easy to convince ourselves that they will be frequent only in microscopic systems , where @xmath47 is broad . the larger the system involved , the narrower @xmath47 will be , the rarer and rarer the occurrence of `` free lunches '' . in a macroscopic friction experiment , the occurrence of a `` free lunch
'' will be virtually impossible .
in nanoscale tribology , extensive attention has focused on the time - honored pt model , which describes a point - like tip sliding over a space - periodic crystalline surface in a minimal fashion .
we shall omit this model from the present review , since it is covered in great detail elsewhere @xcite .
we concentrate instead on its natural extension , the one - dimensional fk model @xcite , which provides a prototypical description of the mutual sliding of two perfect , extended crystalline surfaces .
first studied analytically in ref .
@xcite and later introduced independently to address the dynamics of dislocations in crystals @xcite , subsequently this model became the paradigm describing the structure and dynamics of adsorbed monolayers in the context of surface physics .
the standard fk model consists of a 1d chain of @xmath50 classical particles ( `` atoms '' ) , interacting via harmonic forces and moving in a sinusoidal potential , as sketched in fig . [
fig : fkmodel ] .
the hamiltonian is @xmath51 \,.\end{aligned}\ ] ] in eq .
( [ fkhamil ] ) , the @xmath52 term represents the kinetic energy of the particles , and the next term describes the harmonic interaction , with elastic constant @xmath53 , of nearest - neighboring atoms at equilibrium distance @xmath54 .
the final cosine term describes the `` substrate corrugation '' , i.e. the periodic potential of amplitude @xmath55 and period @xmath56 , as experienced by all particles alike . to probe static friction ,
all atoms are driven by an external force @xmath1 , which is increased adiabatically until sliding starts .
the continuum limit of the fk model , appropriate for large @xmath53 , is the exactly integrable sine - gordon ( sg ) equation , and this mapping contributed to the great success of the fk model .
the solutions of the sg model include nonlinear topological solitons ( known as `` kinks '' and `` antikinks '' ) , plus dynamical solitons ( `` breathers '' ) , beside linear vibration waves ( phonons ) . in the fk model , the sliding processes are entirely governed by its topological excitations , the kinks .
let us consider the simplest `` commensurate '' case , where before sliding the chain is in a trivial ground state ( gs ) , when @xmath50 atoms fit one in each of the @xmath57 minima of the substrate potential , so that the coverage ( i.e. the relative atomic concentration ) @xmath58 equals @xmath59 . in this case ,
the addition ( or subtraction ) of a single atom results in configurations of the chain characterized by one kink ( or antikink ) excitation . still at zero applied force , in order to reach a local minimum of the total potential energy in eq .
( [ fkhamil ] ) , the kink expands in space over a finite length , so that the resulting relaxed chain configuration consists in a local compression ( or expansion , for an antikink ) . upon application of a force ,
it is far easier to move along the chain for kinks than for atoms , since the activation energy @xmath60 for a kink displacement [ known as the peierls - nabarro ( pn ) barrier ] is systematically smaller , and often much smaller @xcite , than the amplitude @xmath55 of the energy barrier that single atoms experience in the substrate corrugation . the motion of kinks ( antikinks ) , i.e. the displacement of the extra atoms ( vacancies ) represents the mechanism for mass transport along the chain .
these displacements are responsible for the mobility , diffusivity , and conductivity within this model .
generally therefore a larger concentration of kinks is associated to a larger the overall mobility @xcite . for the simple commensurate gs ( e.g. , @xmath61 ) , which contains neither kinks nor antikinks , the onset of sliding motion under a driving force occurs via the creation of a kink - antikink pair , e.g. induced by a thermal fluctuation ,
see fig .
[ kinks ] .
if the fk chain is of finite length , kinks / antikinks are usually created at one chain free end , then they advance along the chain , eventually disappearing at the opposite end @xcite .
every kink running from one end of the chain to the other produces the advancement of the entire chain by one lattice spacing @xmath56 . for a finite film confined between two surfaces , or for an island deposited on a surface
, the general expectation is that sliding initiates likewise with the formation and entrance of a kink , or antikink at the boundary @xcite . in this two - dimensional ( 2d ) case , and more generally in @xmath6-dimensional systems , the zero - dimensional kinks of the fk model are replaced by @xmath62 dimensional misfit dislocations or domain walls , whose qualitative physics and role is essentially the same .
incommensurability between the periods @xmath56 and @xmath54 plays an important role in the fk model .
assume , in the limit of an infinite chain length , the ratio @xmath63 of the substrate period @xmath56 to the average spacing @xmath54 of the chain to be irrational .
the gs of the resulting incommensurate fk model is characterized by a sort of `` staircase '' deformation , with a regular sequence of regions where the chain is compressed ( or expanded ) to match the periodic potential , separated by kinks ( or antikinks ) , where , at regular intervals , the misfit stress is released through a localized expansion ( compression ) .
the incommensurate fk model exhibits , under fairly general conditions on @xmath64 @xcite , a critical elastic constant @xmath65 , such that if @xmath66 the chain can slide freely on the substrate at no energy cost , i.e. the static friction @xmath67 drops to zero ( and the low - velocity kinetic friction becomes extremely small ) , while , remarkably , this is no longer true when @xmath68 . in the early 1980s a rigorous mathematical theory of this phenomenon called `` the transition by breaking of analyticity '' , now widely known as the _ aubry transition _
@xcite , was developed .
a simple explanation of free sliding in the unpinned @xmath69 state is the following .
for every atom climbing up toward a corrugation potential maximum , there always is another atom moving down , with an exact energy balance of these processes . quite generally , incommensurability guarantees that the total energy ( we are at @xmath70 ) is independent of the relative position of the chain and the periodic lattice .
however , in order for the chain to slide with continuity between two consecutive positions , it is necessary that particles should be able to occupy a maximum of the potential , the worst possible position . at the aubry transition , however , realized by a relative increase of the periodic potential magnitude , or equivalently by a softening of the chain stiffness , the probability for a particle to occupy that position drops from a finite value to exactly zero .
the nature of this transition , which is structural but without any other static order parameter ( besides energy , of course ) , is dynamical , similar in that to a glass transition : simply , a part of phase space becomes unavailable , in this case by sliding .
the chain is unpinned and mobile as long as , in its gs , atoms may occupy with a finite probability all positions , including those arbitrarily close to the maxima of the substrate potential , but is immobilized when that possibility ceases .
the critical chain stiffness @xmath65 marks the crossing of the aubry transition , where the chain turns from the free sliding state to the locked ( `` pinned '' ) state with a nonzero static friction @xmath67 @xcite .
the value @xmath71 is in turn a discontinuous function on the length ratio @xmath64 characterizing the model .
the minimum value @xmath72 [ in units of @xmath73 is achieved for the golden - mean ratio @xmath74 @xcite .
the stiff - spring chain with @xmath75 can explore adiabatically the full infinite and continuous set of gss configurations by means of displacements at no energy cost .
this zero - frequency freely - sliding mode is the goldstone mode consistent with an emerging continuous translational invariance of the model , connecting continuously with an acoustical phase mode ( phason ) at finite wavelength .
by contrast , in the pinned soft - chain region @xmath76 , all particles remain trapped close to the substrate - potential minima , a configuration which exhibits a finite energy barrier against motion over the corrugation .
the locking is provided here , despite translational invariance , by the inaccessibility of forbidden configurations , which act as dynamical constraints . above @xmath71
the incommensurate - chain sliding can therefore be initiated by an arbitrarily small driving force , whereas for @xmath68 the chain and the corrugation lock together through the pinning of the kinks or superkinks separating locally lattice - matched regions .
note that the locking of the free ends in a finite - size fk chain necessarily leads to pinning , even when @xmath64 is irrational and regardless of how large @xmath53 is . even for a finite chain ,
it is nevertheless still possible to define and detect a symmetry - breaking aubry - like transition @xcite .
for the characterization of the aubry transition , a `` disorder '' parameter @xmath77 was conveniently defined @xcite as the smallest distance of atoms from the nearest maximum of the corrugation potential .
this quantity vanishes in the freely - sliding state , and is nonzero in the pinned state . at the critical pinned - to - sliding point ,
the disorder parameter exhibits a power - law behavior @xmath78 here the values of the critical exponents are functions of the irrational length ratio @xmath64 .
specifically , for the golden - mean ratio @xmath79 and @xmath80 @xcite .
equation ( [ criticalexp : eq ] ) characterizes the continuous aubry transition with a scaling behavior typical of critical phenomena , here at @xmath70 but as a function of the the stiffness parameter @xmath53 .
it is common to refer to the exponents in eq .
( [ criticalexp : eq ] ) as super - critical , since they are specific to the pinned side of the transition , @xmath81 .
sub - critical exponents were introduced for the freely - sliding state @xmath75 , as well .
to describe the response of the model to an infinitesimally small dc force @xmath1 applied to all atoms , an extra damping term @xmath82 has to be included in the equation of motion to prevent unlimited acceleration , and to achieve instead a steady - state .
the resulting effective viscosity in the subcritical region is defined as @xmath83 , in terms of steady - state average velocity @xmath84 resulting in response to @xmath1 . at the aubry critical point @xmath71 ,
the effective viscosity @xmath85 diverges . for @xmath86 , the golden - mean ratio ,
the scaling behavior of @xmath85 is @xmath87 with @xmath88 . as is the case for all scaling relations , eq .
( [ gammaaubry ] ) provides the leading divergence close to the aubry point ; at a larger distance from @xmath71 , @xmath85 deviates from eq .
( [ gammaaubry ] ) .
eventually , in the @xmath89 sg limit , @xmath85 decreases toward @xmath4 . in general ,
in the unpinned phase at @xmath75 the incommensurate fk model exhibits an effective viscosity systematically larger and thus a mobility @xmath90 consistently smaller than its maximum value @xmath91 .
this observation is illustrated by the @xmath92 limiting values of the @xmath75 curves of fig .
[ fig : aubryfig ] . exclusively in the @xmath93 sg limit , the incommensurate system moves under an infinitesimal force without any extra dissipation added to the base value @xmath4 , therefore in a frictionless sliding motion , despite the finite corrugation magnitude @xmath55 . the first prediction of vanishing static friction was formulated for the incommensurate infinite - size sufficiently hard fk chain by peyrard and aubry @xcite .
this phenomenon was subsequently re - discovered for incommensurate tribo - contacts , and named _ superlubricity _ @xcite .
this name has drawn criticism , because it could misleadingly suggest the vanishing of _ kinetic _ friction too , in analogy to superfluidity or superconductivity .
actually instead , the depinning of sliding interfaces closes just one of the channels for energy dissipation , namely the one associated to the stick - slip elastic instability at low speed .
additional dissipation channels , including the emission of vibrations such as sound waves , remain active , with the result that the actual kinetic friction force remains nonzero and growing with increasing sliding speed . all the same , the superlubric state does attain a significant reduction of the kinetic friction force ( and thus an increased mobility @xmath94 ) , compared to the pinned state @xmath76 , see fig . [
fig : aubryfig ] .
the driven fk model was usefully employed to describe the onset of sliding of a crystalline contact @xcite , even though this model can not describe rigorously the experimentally - significant real - life plastic deformations of the contact .
experimentally , superlubricity has been studied for a graphite flake sticking to the tip of an atomic force microscope ( afm ) sliding over an atomically flat graphite surface @xcite .
extremely weak friction forces of less than @xmath95 pn were detected in the vast majority of the relative flake - substrate orientations , namely those orientations generating incommensurate contacting surfaces , see fig .
[ fig : flakesuperlub ] . stick - slip motion , associated with a much higher friction force ( typically @xmath96 pn ) , was instead found in the narrow ranges of orientation angles where the flake - substrate contact was commensurate .
the above discussion ignores temperature , assuming so far @xmath70 . at nonzero temperature ,
the sliding - friction response of the fk model @xcite requires of course the addition of a thermostat ( see sec . [ linear : sec ] ) .
the common choice of a langevin thermostat for example simulates all dissipation mechanisms through a viscous force @xmath82 , and includes fluctuations by the addition of gaussian random forces whose variance is proportional to temperature @xmath0 , as sketched in sec .
[ fluctdiss : sec ] . at @xmath97 ,
thermal fluctuations can always overcome all sorts of pinning and will thus initiate sliding by nucleation of mobile defects even in the fully commensurate ( pinned ) condition , see fig .
[ kinks ] .
more generally , in the fk model the dimensionless coverage @xmath63 plays a central role , because it defines the concentration of geometrical kinks `` close to @xmath61 and of ' ' superkinks " which arise when @xmath64 deviates slightly from a background commensurate pattern that is not @xmath98 , but rather a rational @xmath99 , with @xmath100 and @xmath101 mutually prime integers . if @xmath64 is only slightly different from @xmath102 , the gs of the fk model consists of extended domains with the commensurate pattern associated to @xmath102 , separated by superkinks ( super - antikinks ) , in the form locally mismatched regions of compression ( expansion ) relative to @xmath102 .
the above concepts of pinning or superlubricity apply for an infinitesimal applied force .
additional interesting physics arises at finite force . by increasing the driving force , a fk model with a pinned gs ( either commensurate , or incommensurate but past the aubry transition )
is known to show a hierarchy of first - order dynamical phase transitions , starting from the completely immobile state , passing through several intermediate stages characterized by different running states of the kinks , to eventually reach a totally running state .
consider , for example , the ratio @xmath103 : initially the mass transport along the chain is supported by superkinks constructed on top of the background @xmath104 structure . since the average superkink - superkink distance @xmath105 is large , they interact weakly , and the atomic flow is restricted by the need for these rarefied superkinks to negotiate their pn barriers ( see fig . [ fig : th-01]a ) . for larger driving @xmath1 ,
the effective pn barriers are tilted and lowered ( in analogy to barriers of the corrugation potential ) , producing an increased single - kink mobility @xmath94 . as a result
, the zero - temperature transition from the locked state ( @xmath106 ) to the running state takes place at the force @xmath107 , where the factor @xmath108 depends on the shape of the pn potential . in terms of the dimensionless superkink concentration @xmath109 ,
the mobility becomes @xmath110 . beyond @xmath111 ,
further possibilities depend on the damping coefficient @xmath4 . at very small
damping , @xmath112 , the driven model transition leads directly into the fully running state , because running superkinks self - destroy soon after they start to move , causing an avalanche , thus driving the whole chain to a total running state similar to that shown at the right side of fig .
[ kinks ] .
when the dissipation rate is larger , @xmath113 , one instead finds intermediate stages with stable running superkinks , see fig .
[ fig : th-01 ] . the mechanism for a second rapid increase of the mobility after
depinning depends again on the value of @xmath4 ( for details see refs .
@xcite ) . in between the initial superkink - sliding stage and the fully running state
, a sort of `` traffic - jam '' intermediate regime may emerge @xcite .
a qualitatively similar picture was confirmed also for different and more complex kink patterns , such as that shown for the @xmath114 example in fig .
[ fig : th-01]b @xcite . in this case , the gs consists of domains of the @xmath115 commensurate structure , separated by superkinks at an average spacing @xmath116 . even the @xmath117 pattern could itself be viewed as a dense array of trivial kinks constructed on top of the simple @xmath104 background structure .
the force dependence of the mobility @xmath118 bears a trace of this double nature , with a state of running superkinks preceding a state of running kinks . for not too small @xmath4 therefore the mobility @xmath94 increases in two distinct steps as the driving force is increased .
a first step , at @xmath119 ( here @xmath120 defines the depinning force for the fully commensurate model @xmath121 ) occurs when the superkinks begin to slide ; then a second step , at @xmath122 , occurs in correspondence to the unpinning of the trivial kinks .
several extensions of the fk model have been proposed to describe a broad range of frictionally relevant phenomena .
most of these generalizations involve modifications of either the interactions or the system dimensionality . to address more realistic systems ,
anharmonicity of the chain interatomic potential has been studied in detail .
the resulting features include mainly new types of dynamical solitons ( supersonic waves ) , a modification of the kink - kink interaction , the breaking of the kink - antikink symmetry , and even the possibility of a chain rupture associated to the excessive stretching of an antikink @xcite . the large kink - antikink asymmetry consistent with friction experiments in layers of repulsive colloids @xcite was attributed to the strong anharmonicity in the colloid - colloid interaction @xcite .
the essence of this asymmetry is the same as that between the physical parameters of a vacancy and those of an interstitial .
research has addressed also substrates with a complex corrugation pattern @xcite , including quasiperiodic @xcite and random / disordered corrugation profiles @xcite .
modifications from the plain fk model may generate qualitatively different types of excitations , e.g. phonon branches and kinks of different kinds , as well as modifications in the kink - antikink collisions . at small driving force , where the dynamics and tribology are dominated by moving kink - like structures ,
different sliding modes appear .
the frenkel - kontorova - tomlinson ( fkt ) model @xcite introduces an harmonic coupling of the sliding atomic chain to a driving support , thus making it possible to investigate stick - slip features in a 1d extended simplified contact .
the fkt framework provided the ideal platform to investigate the tribological consequences of combined interface incommensurability , finite - size effects , mechanical stiffness of the contacting materials , and normal - load variations @xcite .
important generalizations involving increased dimensionality compared to the regular fk model bear significant implications for tribological properties such as critical exponents , size - scaling of the friction force , depinning mechanisms , and others . in particular 2d extensions of the fk model @xcite
have been applied to the modeling of the ( unlubricated ) contact of two crystals .
such is the case , for example , in quartz - crystal microbalance ( qcm ) experiments , where single - layer adsorbate islands are made slide over a crystalline substrate @xcite .
another case is that of recent experiments carried out with 2d monolayers of colloids driven over a laser - generated optical lattice @xcite .
interestingly , the 2d aubry transition of incommensurate colloids was shown by mandelli _
et al . _
@xcite to be of first order , rather than of second order as in 1d . as a consequence in 2d the free sliding and
the pinned phases retain local stability for a range of parameters that extends beyond the transition point , a point where the total energy has a crossing singularity instead of a smooth stiffness dependence as in 1d .
it is likely , although not proven to our knowledge , that the 2d fk should possess a first - order aubry transition too . among generalized 2d fk models we recall the two coupled fk chains @xcite , the `` balls and springs '' layer of particles linked in 2d by harmonic springs and moving in a 2d periodic corrugation potential , the scalar anisotropic 2d fk model consisting of a coupled array of 1d fk chains , the 2d vector anisotropic model ( namely the zigzag fk model where
the transverse motion of atoms is included @xcite ) , the 2d vector isotropic fk model @xcite , and finally the 2d tribology model @xcite ( see also ref .
@xcite and references therein ) .
these approaches which generalize the fk model have been of use for the study of the transient dynamics at the onset of sliding .
capturing these transient phenomena is often highly nontrivial in fully realistic md simulations ( see e.g. ref .
an interesting example of such a transient is the depinning of an atomic monolayer driven across a 2d periodic substrate profile of hexagonal symmetry @xcite .
the formation , by nucleation , of an island of moving atoms in a `` sea '' of quasi - stationary particles mediates the transition from the locked to the running state .
the moving island expands rapidly along the direction of the driving force , and grows at a slower rate in the orthogonal direction . within the island
, the 2d crystal retains its approximate ordered hexagonal structure , thanks to its stiffness supported by the intra - atomic forces . as a result ,
at the onset of depinning , the model exhibits regions of almost perfect hexagonal - lattice order delimited by a closed boundary of dislocations .
kinks in 1d and dislocation lines in 2d exhibit peculiar tribological properties . in noncontact experiments ,
an oscillating afm tip was seen to dissipate significantly more when hovering above a dislocation line of a incommensurate adsorbates than above in - registry regions @xcite .
the larger softness and mobility of the dislocation regions @xcite accounts for this effect .
an explicit demonstration of this mechanism was carried out by the study of a mismatched fk chain , whose dynamics was forced and simultaneously probed by an oscillating localized model tip @xcite .
this approach illustrates the ability of the fk model to capture the local modifications of the dissipation properties .
in contrast , if retardation effects related to the finite speed of sound across a material need to be taken into account , more sophisticated models are called for .
the investigation of systems _ confined _ between two shearing sliders , such as single particles @xcite or harmonic chains @xcite embedded between competing periodic potentials , have led to the discovery of several nonlinear tribological phenomena involving either stick - slip dynamics or the formation of peculiar `` synchronized '' sliding regimes @xcite .
the fk model can be generalized with the addition of a second , different sinusoidal corrugation potential , as sketched in fig .
[ velcm : fig]c .
when the second potential is spatially advanced relatively to the first as a function of time , the model realizes the simplest idealization of a slider - solid lubricant - slider confined geometry . in this extended model
the lattice mismatch was shown to generate peculiar and robust `` quantized '' sliding regimes @xcite , where the chain deformations are synchronized to the relative motion of the two corrugations , in such a way that the chain s ( i.e. the solid lubricant s ) average drift velocity acquires nontrivial fixed ratios to the externally - imposed sliding velocity .
specifically , the ratio of the lubricant speed to that of the slider @xmath123 remains locked to specific `` plateau '' values across broad ranges of most model parameters , including the potential magnitude of the two sliders , the chain stiffness ( see fig .
[ velcm : fig]a ) , the dissipation rate @xmath4 , and even the external velocity @xmath124 itself .
the speed ratio is ultimately determined by geometry alone : @xmath125 , where @xmath126 is the incommensurability ratio between the chain spacing @xmath127 and that , @xmath128 , of the closest slider @xcite .
the plateau mechanism is the fact that the kinks formed by the mismatch of the chain with one slider ( the slider whose spatial periodicity is closest to that of the chain ) , are rigidly dragged at velocity @xmath124 by the other slider .
the kink density being geometrically determined and lower than the chain density implies that the overall velocity ratio shares exactly the same properties .
the exactness of the velocity plateaus implies a sort of `` dynamical incompressibility '' , identically zero compliance to perturbations trying to modify @xmath129 from its quantized - plateau value .
this robustness of the plateaus can be demonstrated e.g. by adding a constant force @xmath130 , pushing all particles in the chain : as long as @xmath130 is small enough , it does perturb the dynamics of the velocity - plateau attractor , but not the value of @xmath129 . eventually , above a critical force @xmath131 , the driven model abandons the plateau dynamics .
this transition , explored by increasing the external driving force @xmath132 , exhibits a broad hysteresis , and shares many features of the _ static_-friction depinning transition , except that here it takes place `` on the fly '' @xcite . disregarding details , this transition is then formally equivalent to the standard aubry depinning transition @xcite , with the moving kinks of the lubricant - substrate interface taking here the role of particles .
the robustness of the quantized plateau stands even after replacement of the sinusoidal corrugation potential of eq .
( [ fkhamil ] ) with a deformed profile : the remoissenet - peyrard non - sinusoidal potential even extends the velocity plateau in the space of model parameters @xcite .
the quantized sliding regime of the crystalline solid lubricant was also investigated in a significantly less idealized 2d model , including the perpendicular degree of freedom @xcite .
md simulations carried out for a monolayer or multilayer lubricant film where atoms interacting via lennard - jones potentials can also move perpendicularly to the sliding direction , as sketched in fig .
[ model_quantized : fig ] showed quantized plateaus in this case too .
these plateaus were shown to be resilient against a variations in the loading forces across a broad range , against thermal fluctuations , and also against the presence of quenched disorder in the substrates .
this quantized sliding state was also characterized by significantly lower values of kinetic friction @xmath133 ( the average force needed to maintain the advancement of the top slider ) @xcite , than the regular non - quantized regime , see fig .
[ hyst - diss - kink : fig ] .
quantized sliding has been again demonstrated more recently in a 3d model where the lubricant is represented by a layer of lennard - jones atoms @xcite .
the quantized - sliding state and its boundaries were fully characterized in the special case of perfectly - aligned crystalline layers .
we note however there are reasons to expect that incommensurately mismatched epitaxial layers could relax to a mutually rotated alignment @xcite .
the quantized - sliding state in such rotated arrangements is the subject of active investigation . more generally , no experimental observations of the quantized sliding predicted for solid lubricants has appeared so far .
layered systems such as graphene and bn appear to offer a good opportunity for the future study of these curious phenomena .
the simple models considered above yielded precious qualitative and often semi - quantitative understanding of several features of friction . to address subtler physical behavior in specific systems ,
it is nevertheless necessary and desirable to include atomistic structural and mechanical details of the interface .
md simulations can help make an inroad such detail , also offering a level of detail that can in some instances replace experiment .
thanks to advances in computing algorithms and hardware , recent years have witnessed a remarkable increase in our ability to simulate tribologic processes in realistic nano - frictional systems , and obtain detailed microscopic information .
a md simulation is _ de facto _ a controlled computational experiment , where the overall atomic dynamics is provided by the numerical solutions of suitably generalized newton s equation of motion relying on interatomic forces derived by specific realistic interparticle - interaction potentials .
tribological simulations require a careful selection of the geometric arrangement of the sliding interface , e.g. as in fig . [
simulation : fig ] , and of the applied boundary conditions .
influential review articles @xcite cover the atomistic md simulation of friction , with a focus on technical aspects such as the construction of a realistic interface , suitable techniques for the application of load , shear , and temperature control .
the simplest approach to temperature control , namely adding a langevin thermostat to newton s equations @xcite , has been adopted broadly , but more refined approaches have been proposed and adopted for friction simulations , as discussed in sect .
[ thermostat : sect ] below .
physically relevant quantities , including the average friction force , the slider and the lubricant mean velocities , several correlation functions , and the heat flow can be evaluated numerically by carrying out suitable averages over the model dynamics of a sliding interface , as long as it is followed for a sufficiently long time .
the modeling of friction must first of all address correctly ordinary equilibrium and near - equilibrium phenomena , where the fluctuation - dissipation theorem ( sec .
[ linear : sec ] ) governs the smooth conversion of mechanical energy into heat , but most importantly it must also deal with inherently nonlinear dissipative phenomena such as instabilities , stick - slip , and all kinds of hysteretic response to external driving forces , characteristic of non - equilibrium dynamics .
the choice of realistic interatomic forces is often a major problem . indicating with @xmath134 the total interaction energy as a function of all atomic coordinates @xmath135 , the force on atom @xmath136 is @xmath137 , fully determined in terms of @xmath138 .
unfortunately , the adiabatic energy @xmath138 results from the solution of the quantum ground state of the electrons a practically complicated problem whose quantitative outcome may moreover be of uncertain quality .
the reason why _ ab - initio _
md , e.g. of the car - parrinello type @xcite , is not generally used in sliding friction is that it can neither handle large systems , exceeding few hundreds of atoms , nor run for tribologically significant duration , usually in excess of @xmath139 ns . on the other hand ,
the physical situations where a first - principles description of interatomic forces is mandatory are not too common in the frictional phenomena studied so far . as a consequence ,
most md models for friction rely on more or less refined interatomic `` force fields '' , ranging from sophisticated energy surfaces modeled on calculations at _ ab - initio _ density - functional or tight - binding level @xcite , to empirical distance- and angle - dependent many - body classical potentials , to basic pairwise potentials ( e.g. morse or lennard - jones ) , to the simplest elastic - springs models , which represent generalizations of the fk model . concretely , the scientific literature documents many realistic force fields , ready to address several classes of materials and their combinations @xcite .
while these force fields allow qualitative atomistic simulations of tribological systems , their limitations often prevent quantitative accuracy . in particular , in the course of such a violent frictional process as wear , atoms are likely to modify their chemical coordination and even their charge state : phenomena and radical chemical changes usually impossible to describe with empirical force fields .
mechanochemistry and tribochemistry are time - honored areas @xcite offering obvious examples where empirical force - fields would fail , and simulations must by necessity be conducted by electronic - structure based first - principles methods .
also , even if for a specific system , element or compound , a satisfactory force field has been arrived at , the mere replacement of one atomic species with another one generally requires a complete and usually difficult re - parameterization of the whole force field . as a result ,
quantitatively accurate nanofrictional investigations remain a substantial challenge because of opposite limitations in the use of first principles versus empirical force fields . a promising compromise could possibly be provided by the so - called reactive potentials @xcite , capable of describing some chemical reactions , including interface wear with satisfactory computational efficiency in large - scale atomic simulations , compared to semi - empirical and first - principles approaches .
retardation effects due to the finiteness of the speed of sound are usually irrelevant in slow - speed experiments ( @xmath140 mm / s ) . for larger speed
, retardation effects related to the finite speed of sound can be taken into account explicitly in md modeling , provided that rigid layers are either omitted or introduced with special care .
other effects of nonlocality in time , such as retardation due to the finite speed of light @xcite are usually omitted in the md force fields altogether , as they lead to negligible corrections in all conditions where sliding involves a proper material contact .
such retardation effects do play a role in noncontact geometries , such as in experiments probing lateral casimir forces @xcite , whose strength can become relevant at large sliding speed @xcite . as we already mentioned above in sec .
[ linear : sec ] , any kind of sliding friction involves mechanical work , some of which is then transformed into heat ( the rest going into structural transformations , wear , etc . ) .
the heat is then transported away by phonons ( and electrons in the case of metallic sliders ) and eventually dissipated to the environment .
likewise , all excitations generated at the sliding interface in simulations should be allowed to propagate away from it , and to disperse in the bulk of both sliders . instead , due to the small simulation size , this energy may unphysically pile up in the rather small portion of solid representative of the `` bulk '' of the substrates , where these excitations are scattered and back - reflected by the simulation - cell boundary , instead of being properly dissipated away . in order to prevent continuous heating and attain a steady state of the tribological system , the joule heat must then be removed at a steady rate . in the fk and pt models , a viscous damping term @xmath82 , eq .
( [ eq_langevin ] ) , is generally introduced for this purpose . in these minimal models
however , the value of @xmath4 is well known to affect the dynamical and frictional properties , but there is unfortunately no clear prescription for the choice of @xmath4 . in md atomistic simulations ,
the heat removal is often achieved by means of equilibrium thermostats , e.g. nos - hoover or langevin , see sec .
[ fluctdiss : sec ] . in this way
however an unphysical energy sink is spread throughout the simulation cell . as a result , the atoms at the interface fail to follow their actual conservative trajectories , but evolve through an unphysically damped dynamics , with unknown and generally undesired effects on the overall tribological properties @xcite . in order to address and mitigate this problem , modifications of the equations of motion for the atoms inside the microscopically small simulation cell were proposed with the target of reproducing the frictional dynamics of a realistic macroscopic system , after the integration of extra `` environment '' variables .
one possible approach is the application of langevin equations with a damping coefficient that changes as a function of the position and velocity of each atom in the lubricant , in accordance with the dissipation known for the atoms adsorbed on a surface @xcite .
this method involves modifying the standard langevin equations @xcite .
another approach to improve the simulation of dissipation within blocks in reciprocal motion requires modifying the damping term ( [ eq_langevin ] ) to a form @xmath141 where @xmath142 is the average center - mass velocity of the atoms forming the sliding block to which particle @xmath143 belongs locally @xcite . another more rigorous , physically appealing approach is the recently - implemented dissipation scheme , drawing on earlier , long - known formulations @xcite and subsequent derivations @xcite , describing the correct embedding of the newtonian simulation cluster inside a larger heat bath made of the same material . upon integrating out the heat bath degrees of freedom , atoms in the boundary layer that borders between the cluster and the heat bath
are subjected to additional non - conservative and non - markovian forces that mimic the surrounding bath through a so - called memory kernel .
an approximate but very practical scheme replaces this memory kernel by a simple viscous damping @xmath4 , here applied exclusively to atoms in the boundary layer .
the magnitude of the parameter @xmath4 is optimized variationally by minimizing , with surprising accuracy , the energy reflected across the boundary @xcite .
this dissipation scheme has been implemented recently in nanofriction simulations where it was shown to improve greatly over other conceptually and practically inadequate thermostats . besides the limitations of system size and simulation times that are obvious and will be discussed later
, there is another limitation concerning temperature , that is rarely mentioned .
all classical frictional simulations , atomistic or otherwise , are only valid at sufficiently high temperature .
they become in principle invalid at low temperatures where the mechanical degrees of freedom of solids progressively undergo `` quantum freezing '' , and both mechanics and thermodynamics deviate from classical .
unfortunately there is at the moment no available route to include appropriately these quantum effects in dynamical and frictional simulations .
each core of a present - day cpu executes @xmath144 floating - point operations per second ( flops ) .
md simulations usually benefit effectively of medium - scale parallelization .
approximately linear scaling can be achieved up to @xmath145 cores , thus a md simulation can execute @xmath146 flops routinely .
the evaluation of the forces is usually the most cpu - intensive part of a md simulation . for each atom , depending on the force - field complexity and range , this evaluation can require @xmath147 operations , or even more . as a result , the number of time - integration steps @xmath148 multiplied by the number of simulated particles @xmath50 , is @xmath149 per computer runtime second . given that simulations of atomic - scale friction require time - steps in the femtosecond region , a medium - size simulation involving @xmath150 simulated particles , can advance at an estimated speed of @xmath151 fs each real - life second , namely @xmath152s each simulation day .
clearly , speed scales down for more refined force fields , and for larger systems size , although this increase may be mitigated by a larger - scale parallelization .
we can compare these estimates with typical sizes , duration , and speeds in frictional experiments . in macroscopic tribology experiments ,
sliding speeds often range in the @xmath153 m / s region : each microsecond the slider would progress by @xmath154 to @xmath155 m , namely @xmath156 lattice spacings of standard crystalline surfaces . in such conditions
@xmath157s may suffice to generate a good statistics of atomic - scale events , although it may still be insufficient to address e.g. the diffusion of wear particles or additives in the interface , or phenomena associated to surface steps and/or point defects .
by contrast , in nanoscale afm experiments the tip usually advances at much lower speeds @xmath158 : over a typical run it is possible to simulate a tiny @xmath139 pm displacement , far too small to explore even a single atomic - scale event , let alone averaging over a steady state .
for this reason , in all conditions where long equilibration times and/or slow diffusive phenomena and/or long - distance correlations can be expected , models should be preferred to realistic but expensive md .
however , md simulations can provide so much physical insight that they make sense even if carried out at much higher speeds than in real - life afm or surface force apparatus ( sfa ) experiments : in practice , currently the sliding speeds of most atomistic tribology simulations are in the @xmath159 m / s region
. here however we should distinguish between static and kinetic friction , and for the latter between smooth - sliding and stick - slip regimes .
smooth kinetic friction generally increases with speed ( velocity strengthening ) , but sometimes decreases with increasing speed in certain intervals . in the former case ,
simulating smooth high - speed frictional sliding is not fundamentally different from the real sliding at low speed , with appropriate changes in frictional forces with @xmath84 .
velocity weakening conditions , alternatively , tend to lead to an intrinsic instability of smooth sliding , which is therefore not often pertinent to real situations .
as a result , for nanoscale systems , md simulations is of value in the description of smooth dry kinetic friction despite the huge velocity gap . on the other hand , static friction the smallest force needed to set a slider in motion is also dependent on the simulation time ( a longer wait may lead to depinning when a short wait might not ) , and generally dependent on system size , often increasing with sub - linear scaling with the slider s contact area . to address this kind of behavior in md simulations , it is often necessary to resort to scaling arguments in order to extrapolate the large - area static friction from small - size md simulations @xcite . returning to the simulation time problem ,
let us come to stick - slip in md simulation , and to the desirability to describe the stick - slip to smooth - sliding transition as a function of parameters such as speed . in afm and sfa experiments , stick - slip and
its associated characteristically high friction and mechanical hysteresis tend to transition into smooth sliding when the speed exceeds @xmath160 m/s ; in contrast , in md modeling the same transition is observed in the @xmath159 m / s region .
this 6-order - of - magnitude discrepancy in speed between experiments and simulations is well known and has been largely discussed @xcite in connection with the effective mass distributions and spring - force constants , that are vastly different , and highly simplified in simulations .
attempts to fill the time and speed gaps can rely on methods , such as hyperdynamics , parallel - replica dynamics , on - the - fly kinetic monte carlo , and temperature - accelerated dynamics which have been developed in the last decades @xcite .
however , caution should generally be exerted in that some of these schemes and methods are meant to accelerate the establishment of equilibrium but not always to treat properly the actual frictional - loss mechanisms .
concerning stick - slip friction , another problem is that , unlike simulations , real experiments contain mesoscale or macroscale component intrinsically involved in the mechanical instabilities of which stick - slip consists . here
the comforting observation is that stick - slip is nearly independent of speed , so that so long as a simulation is long enough to realize a sufficient number of slip events , the results may already be good enough @xcite .
one can even describe stick - slip friction adiabatically , e.g. , from a sequence of totally static calculations , where a periodic back - and - forth sliding path is trodden , the area of hysteresis cycle generated by two different to and from instabilities representing the friction @xcite .
a serious aspect of stick - slip friction which md simulation is unable to attack is ageing .
the slip is a fast event , well described by md , but sticking is a long waiting time , during which the frictional contact settles very slowly .
the longer the sticking time , the larger the static friction force necessary to cause the slip . typically experiments show a logarithmic increase of static friction with time @xcite .
rate and state friction approaches , widely used in geophysics @xcite , describe phenomenologically frictional ageing , but a quantitative microscopic description is still lacking .
mechanisms invoked to account for contact ageing include chemical strengthening at the interface in nanoscale systems @xcite , and plastic creep phenomena in macroscopic systems @xcite .
contact ageing is observed also in other disordered systems out of equilibrium , including glasses and granular matter . in seismology finally , as will be discussed later , it is generally accepted that ageing is responsible for aftershocks , as also shown by some models @xcite .
if md simulation may be satisfactory in nanoscale friction , it is clearly not capable of describing mesoscale and macroscale tribology .
the insurmountable difficulties of the fully atomistic treatment of all typical and large length scales that are responsible for the dynamical processes in large scale systems has prompted in recent years increasing efforts towards multiscale approaches .
the main idea is that , at a sufficient distance from the sliding interface , continuum mechanics should describe all processes to a fair level of approximation .
finite - elements simulations of the continuum mechanics may provide a practical model for the elastic and plastic deformations . using finite element methods , one can increase the coarse - graining level while moving away from the sliding interface , thus keeping the computational effort under control .
several groups @xcite combined the md description of the sliding interface , where local deformations at the atomic length scale and highly nonlinear phenomena occur , with a continuum - mechanics description in the `` bulk '' regions where strains are continuous and small .
the main difficulty faced by this class of approaches is the correct choice of the matching between the atomistic region with the continuum part @xcite . because at the continuum level the detail of lattice vibrations can not be represented in full , the matching conditions
should at least minimize the reflection of the acoustic phonons at the atomistic - continuum interface .
in other words , the matching should allow the transmission of sound deformations in both directions with sufficient accuracy : this is necessary for a proper disposal of the joule heat into the bulk .
simulations can provide direct insight in the dynamical processes at the atomistic level , that are at the origin of friction , allowing a connection of these microscopic facts with their macroscopic counterparts .
case studies , in which the system is well describable by both experimental and theoretical sides , are of extreme importance firstly to permit a crosscheck between the two , and then to make use of simulations in order to highlight particular aspects that can not be accessed by experiments . here
we summarize the main results of a few selected simulations sampled from the expanding literature of friction simulation , certainly not claiming an exhaustive review of the field .
the sliding of rare - gas overlayers deposited on metallic substrates at low temperature has contributed much to the understanding of how friction scales with the contact - area size , the substrate corrugation , and the sliding velocity .
rare - gas atoms condense into 2d solid islands showing a faceted - circular shape , arranged on multiple layers at low temperatures or on a single layer at diffusion - enabled temperatures @xcite .
the friction characteristics of these solid islands on the substrate , resulting from inertial sliding , has been probed experimentally by qcm apparatuses , revealing a complex interplay among friction , coverage , and temperature @xcite .
the rare - gas lattice spacing inside the island , generally incommensurate but sometimes commensurate with that of the substrate , plays a very important role in determining the pinning or free sliding that controls the frictional behavior . a generally overlooked aspect which has been highlighted only recently @xcite is the larger thermal expansion coefficient of rare - gas layers than that of a metal substrate , causing a temperature - dependent lattice mismatch at the interface with possible incommensurate - commensurate transitions . due to this mechanism ,
md simulations have predicted the possible appearance of static - friction peaks in the correspondence of a long - range commensurate phase occurring at a particular temperature @xcite . in the case of periodic monolayers ,
a change in the lattice mismatch can be also induced by an adhesion - driven densification of the adsorbate @xcite , again eventually encountering a commensurate phase @xcite .
simulations of rare - gas incommensurate adsorbates , whose linear substrate - induced misfit dislocations ( `` solitons '' ) must flow during sliding , have revealed the role of their entrance in the depinning of the island @xcite , and of their dissipation through anharmonic coupling to phonons , in kinetic friction @xcite .
finite - size effect are in this case of absolute relevance , effectively generating ( or enhancing ) static friction through a pinning barrier arising at the interface edge , which solitons must overcome to establish motion @xcite .
the edge - related origin of the pinning mechanisms implies that static friction @xmath67 can grow with the island size @xmath161 at most as @xmath162 , i.e. as its perimeter @xmath163 , if the pinning points were uniformly distributed along the island or cluster edge . as shown in fig .
[ varini.fig2 ] , a different shape of the deposited nano - object can generally lead to a different scaling exponent .
similar sublinear scaling exponents were identified in dynamic friction experiments in which gold nanoclusters of variable size / shape were dragged at low speed over a graphite substrate by an afm tip @xcite . scaling exponents of both the rare - gas island / metal surfaces ( theoretical ) and dragged gold clusters / graphite ( experimental ) are in the order of @xmath164 .
this indicates that not all points at the boundary provide pinning with equal efficiency .
a scaling close to @xmath165 might rather indicate a random efficiency of boundary points , whereby only @xmath166 provide effective pinning .
nowadays computational capabilities even permit the atomistic simulation of an entire afm tip , enabling the understanding of several mechanisms which are not describable by simplified pt - like models ( see fig .
[ slufi2009_fig1 ] ) .
for example , it is possible to highlight the formation / rupture dynamics of contacts in multi - asperity interfaces , and consequently estimate the true contact area as a function of the apparent one . besides
, it is possible to investigate the effect of the tip plasticity and elasticity , which are of fundamental importance to define the load - dependent contact area @xcite , and as channels for dissipation and wear @xcite .
this approach enables the bottom - up derivation of the linear scaling laws of macroscopic friction with size , and their transition to the sublinear ones for incommensurate nanosized contacts .
we can now understand that such transition takes place when the contact roughness becomes large compared to the range of interfacial interactions @xcite . in the study of repeated scratching of metallic surfaces by hard afm tips , widely employed in the field of micro / nano machining ,
md simulations have uncovered strongly non - linear trends of the frictional force with the feed ( i.e. the distance from the first groove ) , induced by lateral forces exerting on the tip due to the substrate plasticity @xcite .
it is also important to mention the simulations of nanotubes ( nt ) , either made of carbon or hexagonal bn , which , due to their extraordinary mechanical and electronic properties , have been investigated with enormous interest in the last decades .
almost defectless nt can be formed nowadays with lengths of the order of 1 cm @xcite , and precise measurements of their mechanical and frictional properties have started to appear in literature @xcite .
simulations of concentric nanotubes in relative motion ( telescopic sliding ) , have revealed the occurrence of well - defined velocities at which friction is enhanced , corresponding to a washboard frequency resonating with longitudinal @xcite or circular @xcite phonon modes , leading to enhanced energy dissipation .
the frictional response becomes highly non - linear while approaching the critical velocity and , contrary to macroscopic systems , washboard resonances can arise at multiple velocities , especially for incommensurate interfaces where more than one length scale may be in common to the contacting surfaces @xcite .
the exceptional electro - mechanical properties of nts have also been investigated by various tip - based techniques , revealing a strong friction anisotropy dictated by nts orientation . in this respect ,
simulation - assisted experiments of a sliding nanosized tip over cnts reveal that transversal friction is enhanced by a hindered - rolling motion of the nt , with a consequent frictional dissipation that is absent in the longitudinal sliding @xcite .
the same elastic deformation have been reportedly responsible of a reverse stick - slip effect in the case of an afm probe sliding over a super - lattice cnt forest @xcite . here
, simulations reveal that the fast sticking is induced by the penetration of the tip into the valley between the nts and its interaction with the tubes on both sides , causing an elastic shell buckling of the cnts .
in contrast , the gradual slipping occurs over much longer distance because it includes both the sliding on the top of the nt and the energy release at both sides of the graphitic wall . when two sliding surfaces are separated by a thick lubricant film , as it ordinarily happens under weak - load conditions , the tribological response of the confined system is typically determined by the fluid viscosity . in these cases of hydrodynamic lubrication ,
friction can be computed based on the navier - stokes equations , which prescribe a monotonically increasing kinetic friction as a function of the relative sliding speed @xcite . by contrast , at high load and low driving velocity
, the lubricant may not maintain a broad gap between the sliding surfaces , with the result that solid - solid contact eventually occurs .
prior to full squeezeout under pressure , as confirmed by experiments and simulations , the intervening boundary film usually changes from liquid to solid or nearly solid , exhibiting a layered structure prone to develop finite static friction and a high - dissipative stick - slip dynamics in such `` boundary - lubrication '' regime . both sfa measurements @xcite and md investigations @xcite have demonstrated sharp upward jumps of friction at squeezeout transitions where the number of lubricant layers decreases from @xmath50 to @xmath167 .
boundary - lubricated systems often display stick - slip dynamics during tribological measurements associated with a significant value of the friction dissipation . as the load increases , however , it becomes harder and harder to squeeze out an extra lubricant layer .
this hardening and increased difficulty of squeezeout reflects the tendency to crystallization of the initially liquid lubricant and the increased cost of the `` crater '' whose formation constitutes the nucleation barrier of the transition @xcite .
once it has happened , generally this @xmath168 `` re - layering '' transition gives rise to an upward friction jump . in principle however , upon re - layering of the solidified ( structured ) confined film , the two - dimensional ( 2d ) parallel crystalline - like order could occasionally change under pressure toward a more favorable mismatched ( incommensurate ) substrate - lubricant geometry . in that case , as a result , sliding friction might actually switch downward from highly - frictional stick - slip to smooth dynamical regimes characteristic of incommensurate superlubric interfaces , with a highly mobile 2d soliton pattern , of the type sketched in fig .
[ solid_lub ] .
so far this type of event has only been observed in simulations @xcite . in sfa experiments ,
boundary - lubricated systems often display stick - slip dynamics during tribological measurements , associated with a significant value of the friction dissipation .
one can not directly access the detailed film and interface rearrangements giving rise to the stick - slip mesoscopic intermittent dynamics .
the mechanisms at play for the stick - slip dynamics in the boundary - lubrication regime have been studied by md investigations @xcite .
several realistic models for lubrication layers were simulated @xcite .
the issue whether frictional shearing occurs through the middle of the solid lubricant film , possibly accompanied by melting - freezing , or if it forms a smooth shear band , or it occurs at the substrate - lubricant boundary , is one which can be addressed by computer modeling .
depending on the relative strength of the potentials governing the lubricant - lubricant and lubricant - substrate interactions , a thin confined film may exhibit a solid - like or liquid - like behavior under shear .
if the interaction with the substrates is weaker than the lubricant - lubricant one , then sliding takes place mainly at the surface - lubricant interfaces .
the lubricant film is then allowed to maintain or acquire a solid order .
if both the solid lubricant and the substrate are characterized by nearly - perfect crystalline structures and these structures are mismatched and/or misaligned @xcite , then smooth superlubric sliding with reduced kinetic friction ensues : in such conditions , solid lubrication can provide quite low friction . in practice
however neither the substrates nor the lubricant are likely to maintain undefected crystalline order . defects and/or impurities between the sliding surfaces , even if diluted to a weak concentration , may suffice to induce pinning and finite static friction @xcite , thereby eliminating superlubricity . in the opposite condition of prevailing lubricant - substrate interactions ,
the surfaces are covered and protected from wear by lubricant monolayers : sliding occurs inside the lubricant bulk . in such condition
the lubricant film can be led to melt during sliding ; alternatively , the layering imposed by the surfaces can remain solid , with slips occurring in a layer - over - layer sliding @xcite .
simulations are of particular value in the exploration of extreme frictional regimes , that are difficult to access experimentally . among such extreme regimes ,
researchers have investigated or are investigating high temperature , high speed , high pressure , and high plate charging in ionic liquid lubrication .
although for most of these conditions there still is no experimental evidence to discuss , simulation has made some interesting predictions that should become of future reference .
_ high temperature .
_ close to the substrate melting point @xmath169 , its crystal surface may or may not undergo surface melting the formation , in full thermal equilibrium , of a thin liquid or quasi - liquid film at the substrate - vacuum interface @xcite .
either event influences importantly the contact of an afm tip with the surface .
surface melting gives rise to a local jump - to - contact of the film with the afm tip , as was found both in experiments @xcite and in md simulations @xcite . in that case
, friction is expected to become hydrodynamic and uninteresting .
more interesting is the case where the substrate surface does not undergo surface melting , such as is the case for particularly well packed , stable surfaces like pb(111 ) or nacl(100 ) @xcite . for an afm tip sliding on nacl(100 ) ,
frictional md simulations suggested two quite different outcomes depending on the frictional mode . a sharp penetrating tip plowing the solid surface experiences a large friction , which drops sharply when the substrate temperature is only slightly below @xmath169 , so that the joule heat suffices to raise temperature locally and form a liquid drop accompanying and lubricating the moving tip . a blunt tip sliding wearlessly experiences instead a very small friction at low temperature , counterintuitively surging and becoming large close to @xmath169 , where the nonmelting surface lattice softens a phenomenon analogous to that exhibited by flux lattices in type ii superconductors @xcite .
_ high speed .
_ friction at high speed , of order of @xmath59 m / s or higher , is common in several technologically relevant situations but is rarely addressed in nanoscale , atomistically characterized situations , where velocity is more typically @xmath170 m/s , many orders of magnitude smaller . as anticipated in sec .
[ sizeissues : sec ] , md simulation is an ideal tool for the study of friction in fast - sliding of nanosized systems . using gold clusters on graphite as test system
, simulation has explored high - speed friction , and especially differences and similarities from low speed , examining the slowing down of a ballistically kicked cluster .
both kinetic frictions are similarly viscous proportional to velocity .
however , they show just the opposite thermal dependence .
whereas low speed ( diffusive ) friction decreases upon heating , when diffusion increases , the high speed ( ballistic ) friction rises with temperature , when thermal fluctuations of the contact increase @xcite . _ high pressure . _
the local uniaxial pressure transmitted to a local contact by the overall load on a slider may reach a hundred kbar , but is generally not very well characterized , and the effects of pressure insufficiently explored .
md simulation makes suggestions of different kinds .
first , pressure may provoke structural transformation of a crystalline substrate ( or slider ) from its initial crystal structure to another .
as a recent simulation has shown @xcite this will reflect in a frictional jump , either up or down .
second , pressure may bring a solid compound close enough to the chemical stability limit for the frictional perturbation to cause bond breaking and the beginning of chemical decomposition @xcite .
third , pressure may lead to electronic or magnetic transformations , such as insulator - metal transitions , and this may also in principle influence friction . _ high plate charging in ionic - liquid lubrication .
_ ionic liquids are salts whose ions have such a large size that @xmath169 falls below room temperature .
experimental data have shown that friction across contacts lubricated by ionic liquids depends on the state of electrical charging of the sliders @xcite .
md simulations applied to heavily simplified ionic liquid models indicate how this dependence can be ascribed to electrically induced structural modifications at the slider - lubricant interface @xcite . for extreme plate charging , these modifications may even modify the lubricant thickness , and also affect its whole molecular structure , with strong predicted consequences on friction @xcite .
on meso- and macro - scales the interface between two bodies is quite generally far from uniform .
when rough surfaces come into dry contact , the actual contacts occur at asperities of different sizes , typically characterized by a fractal distribution . even for a contact of ideally flat surfaces of polycrystal bodies ,
different regions of the interface will be characterized by different local values of the static friction due to structural or orientational domains @xcite . for a lubricated contact , different values of the local static friction may appear due to patches of solidified lubricant or due to capillary bridges .
all these cases can be rationalized with the help of an earthquakelike ( eq ) model first introduced by burridge and knopoff @xcite to describe real earthquakes .
the nature of two problems earthquakes and friction is very similar : differences are restricted to their spatial - temporal scales : kilometers and years to millenia in geology compared to nanometers and seconds to hours in tribology .
the eq model , also known as the spring - and - block model or the multi - contact model , has been successfully used in many studies of friction @xcite ; similar schemes have been used also to model the failure of fiber bundles and faults @xcite . in eq models ,
two corrugated surfaces make contact only at a discrete set of points , as shown schematically in fig .
[ earthquakemodel ] .
when the slider moves , a single point contact elongates elastically as a spring , as long as the local shear force @xmath171 ( @xmath172 is the contact stretching and @xmath173 is its elastic constant ) remains below a threshold value @xmath174 ; then the contact breaks and slips for some distance , as indeed was observed in tip - based microscopy experiments as well as in md simulations . then , either immediately or after some delay time , the contact is reformed again with zero or lower stretching and a new threshold value .
the simplified version of the eq model assumes that all contacts have the same threshold @xmath175 ; such a model however corresponds to a singular case and may lead to unphysical results @xcite . in a real situation
, the contacts always have different thresholds with a continuous distribution @xmath176 of their static threshold elongations .
therefore , when the upper block begins to advance , the forces acting locally on each contact increase : at successive moments , the contacts begin to snap in a sequence : weaker contacts break earlier , while the strongest contact resists to the last .
eq - like models are usually studied by simulation .
nonetheless , the kinetics of the eq model can be described by a master equation ( me ) , occasionally known as boltzmann equation or kinetic equation . in concrete , indicate the distribution of contact s stretchings @xmath177 by @xmath178 , when the sliding plate reaches position @xmath179 ( see fig .
[ earthquakemodel ] ) : the evolution of @xmath178 is described by the equation @xcite @xmath180 q(x ; x ) = \delta ( x ) \ , \int_{-\infty}^{\infty } d\xi \ , p(\xi ) \ , q(\xi ; x ) \ , , \label{int - eq02}\ ] ] where @xmath181 is the fraction of contacts which break at stretching @xmath177 as a consequence of the plate advancing by @xmath182 .
the `` rate '' @xmath183 and the distribution of the breaking thresholds @xmath184 are connected by the relation @xmath185 indicating that the fraction of contacts which snap when @xmath179 increases by @xmath182 equals those that have their thresholds between @xmath177 and @xmath186 , divided by the total fraction @xmath187 of contacts still unbroken at stretching @xmath177 .
the eq model can be extended to account for thermal effects as well as the ageing of contacts @xcite ; the latter requires an additional equation to describe the increase of threshold values with the time of stationary contact .
analytic solutions of the me are available and , in the smooth - sliding regime , they provide us with the velocity and temperature dependence of the kinetic friction force @xmath133 @xcite .
contrary to the amontons
coulomb laws , which state that ( macroscopic ) friction is independent of velocity , the friction force in the eq model depends on the sliding speed . at small
driving velocity @xmath133 increases linearly with speed , @xmath188 . indeed , if the slider moves slowly enough , thermal fluctuations will soon or later break all the contacts
the slower the slider the longer time any contact will be given to undergo a fluctuation exceeding its respective threshold ; therefore the smaller the resulting kinetic friction force is .
this linear @xmath189 dependence could be represented as a ( characteristically large ) effective viscosity @xmath190 of a ultrathin lubricant film @xcite . at high driving velocities
the friction force exhibits the opposite behavior , it decreases when @xmath84 grows , @xmath191 , due to relaxation , which one can also call an ageing effect . after snapping
, a contact slips for a short time , then it stops and is reformed again , growing in size and strength .
the faster the slider moves , the shorter time the contacts are left to be reformed and to grow .
overall therefore , the kinetic friction force @xmath192 increases with @xmath84 at low @xmath84 , up to a maximum at @xmath193 and then decreases . at very high velocities
@xmath192 should eventually grow again due to an increased damping in the slider bulk . at intermediate speeds typical of experiments ,
the interplay of thermal and ageing effects generates a weak ( approximately logarithmic ) @xmath189 dependence , approximately consistent with the amontons - coulomb laws , although the proper @xmath189 dependence is rather difficult to detect in experiments @xcite . on the decreasing ( `` velocity - weakening '' ) branch of the @xmath189 dependence at @xmath194
the slider motion may become unstable and change from smooth sliding to stick - slip motion . if the slider velocity increases due to a fluctuation , the friction force decreases , and the slider accelerates .
this effect is studied usually with the help of phenomenological approach ( e.g. , see @xcite and references therein ) .
a detailed study of the eq model @xcite shows that stick - slip for the multi - contact interface may appear , if and only if two necessary conditions are satisfied .
first , the interface must exhibit an elastic instability .
when the slider moves , the contacts break but then are formed again , but only if the reformed contacts build up a force sufficient to balance the driving force , the motion will be stable .
otherwise the slider will develop an elastic instability , and will keep accelerating until the overall pulling spring force ( of elasticity @xmath53 , see fig . [ earthquakemodel ] ) decreases enough to regain stability .
second , the contacts must undergo ageing .
once these conditions are satisfied , stick - slip will exists for an interval of driving velocities only , @xmath195 , while for lower ( @xmath196 ) and higher ( @xmath197 ) speeds the motion is smooth @xcite . the me approach discussed above however assumed a rigid slider which is not a proper model of a realistic extended system . for a nonrigid slider ,
its elasticity produces contact - contact interaction : as soon as a contact fails , the forces on nearby contacts must increase by an amount @xmath198 .
this @xmath199 was shown @xcite to depend on the distance @xmath200 from the failed contact as @xmath201 at short distance @xmath202 , and as @xmath203 at long distance . the crossover length @xmath204 , known as the elastic correlation length @xcite , depends on properties of both the slider ( its young modulus @xmath205 ) and the interface ( the mean separation @xmath206 between nearby contacts and their average rigidity @xmath173 ) .
accordingly , a simpler model can be formulated that considers the slider as rigid across regions of lateral size @xmath207 , with the micro - contacts inside each @xmath208-sized area treated as a single effective macro - contact . for this `` @xmath208-contact ''
the parameters can be evaluated by solving a specific me .
numerics also showed , in addition , that a large fraction of the extra inter - contact force concentrates behind and in front of the snapped contact , implying that effectively the interface can be treated as a one - dimensional chain of @xmath208-contacts , at least approximately .
now , if a @xmath208-contact undergoes elastic instability , namely if at a certain threshold stress the @xmath208-contact snaps and then advances , then the surrounding @xmath208-contacts acquire an extra chance to also fail : this mechanism results in a sequence of snaps propagating forward and backward along the interface as in a domino effect .
the resulting dynamics of the chain of @xmath208-contacts could then be described by the frenkel - kontorova model of sec .
[ fk : sec ] above , but replacing the sinusoidal substrate potential with a sawtooth profile , i.e. a periodically - repeated array of inclined lines @xcite .
this approach allows one to calculate the maximum and minimum shear stress for the propagation of this self - healing crack ( the minimum shear stress coincides with the griffith threshold @xcite ) , and also the dependence of the crack velocity on the applied stress .
when a @xmath208- sized contact fails at some point along the chain subject to a uniform shear stress , two self - healing cracks leave the initial snapping contact , propagating in opposite directions as divergent solitons similar to the kink - antikink pair of fig .
[ kinks ] , until these cracks either reach the boundary or meet another crack generated elsewhere . on the other hand , nonuniform shear stress is relevant for experiments such as those carried out by fineberg s group @xcite with the slider pushed at its trailing edge : at this location the shear stress is maximum and across the block it decreases with the distance from the trailing edge . in this system ,
the leftmost @xmath208-contact is the most likely to fail first , as the pushing force is increased .
this failure will result in an increased stress concentrating on the successive @xmath208-contact , which will fail as well .
this process will repeat itself until a self - consistent stress remains below the breaking threshold everywhere in the slider . as a result , the self - healing crack initiated at the trailing edge
will propagate along the interface for a certain distance @xmath209 ( that can be calculated @xcite ) , releasing the stress at its tail side , while accumulating extra stress at its forward side .
when the pushing force further increases , a second crack starts at the trailing edge , and can trigger a failure sequence in the pre - formed stressed state , thus propagating to some extra distance .
these multiple cracks repeat themselves until they reach the slider leading edge , resulting in a major collective slip .
thus , at the sliding onset , several cracks advance along the interface , with the whole slider undergoing multiple small forward slips ( the so - called precursors ) . in experiment these precursors could be detected and could help predicting the eventual large `` earthquake '' @xcite .
as mentioned at the beginning of this section , the eq model was formulated initially to explain earthquakes .
actual earthquakes follow two approximate empirical laws : the one named after gutenberg - richter @xcite states that the number of earthquakes with the magnitude @xmath210 scales with the corresponding magnitudes according to a power law ; the omori law @xcite states that aftershocks occur with a frequency decreasing roughly with the inverse time after the main shock .
eq - like models discussed above can provide a rationale for both these laws of seismology .
specifically , the gutenberg - richter law can be understood as a direct consequence of contact ageing @xcite ; the omori law can be interpreted in terms of cracks propagating to a finite distance after a major earthquake , the stress is not released in full : a certain amount of stress remains stored at a distance @xmath211 from the main shock @xcite , where an aftershock is likely to occur some time later .
the fascinating and multidisciplinary topic of microscopic friction , where physics , engineering , chemistry and materials science meet to study the process of converting mechanical energy irreversibly into heat , still lacks fundamental understanding , and increasingly calls for well - designed experiments and simulations carried out at well - characterized interfaces .
although afm , sfa , and qcm setups are providing insight in the high nonlinear out - of - equilibrium interface processes at the small length scales , these advanced experimental techniques still provide averaged tribological data .
overall physical quantities , such as the average static and kinetic friction , the mean velocity and slip times , do not allow to tackle easily the problem of relating the mesoscopic frictional response of a driven system to the detailed microscopic dynamics and structural rearrangements occurring at the confined interface under shear . in this respect , by explicitly following and analyzing the dynamics of all degrees of freedom at play in controlled numerical `` experiments '' with interface geometry , sliding conditions , and interparticle interactions can be tuned ,
mathematical modeling and computer simulations have proven remarkably useful in the investigation of tribologic processes at the atomic scale and are likely to extend their role in future frictional studies . even though a large number of open questions remains to be addressed
, these modeling frameworks have provided effective insight in the nonlinear microscopic mechanisms underlying the complex nature of static and kinetic friction , the role of metastability , of crystalline incommensurability , and of the interface geometry .
each theoretical approach , from simplified descriptions , to extended realistic md and hybrid multiscale simulations , has limitations and strengths , with specific abilities to address specific aspects of the physical problem under consideration .
thus , a robust prior understanding of the theoretical background is a basic first step in deciding which modeling features deserve a specific attention and which ones are rather irrelevant details , and then in selecting the best methodological approach for a given problem . concluding , it is worth recalling that novel experimental approaches address the intrinsic tribological difficulties of dealing with a buried interface with a very limited control of the physical parameters of the frictional system : artificial systems consisting in optically trapped charged particles , either cold ions in empty space or colloidal particles in a fluid solvent , forced to slide over a laser - generated periodic potential profile .
indeed , especially in 2d colloid sliding it is possible to follow each particle in real time , like in md simulations . by knowing and , on top of that , tuning the properties of a sliding interface
, our physical understanding can expand significantly and open up possibilities to control friction in nano- and micro - sized systems and devices , with serious possibilities of bridging between nanoscale and mesoscale sizes and phenomena .
useful discussion and collaboration with s. zapperi m. urbakh , j. scheibert , m. peyrard , b. persson , g.e .
santoro , r. capozza , a.r .
bishop , and a. benassi , is gratefully acknowledged .
this work is partly funded by the erc advanced grant no .
320796-modphysfrict , the swiss national science foundation sinergia crsii2_136287 , and by cost action mp1303 .
o.b . acknowledges partial support from the egide / dnipro grant no .
28225uh and from the nasu `` resurs '' program . | the nonlinear dynamics associated with sliding friction forms a broad interdisciplinary research field that involves complex dynamical processes and patterns covering a broad range of time and length scales .
progress in experimental techniques and computational resources has stimulated the development of more refined and accurate mathematical and numerical models , capable of capturing many of the essentially nonlinear phenomena involved in friction . |
early ideas on properties of gluonic mesons or glueballs as bound states of gauge bosons within qcd remained inconclusive at first because of an apparent lack of phenomenological evidence for such resonances beyond the prominent mesons @xmath2 , baryons @xmath3 and antibaryons @xmath4 .
an outline of expected scenarios for ( mainly ) binary glueballs , i.e. consisting predominantly of two gauge bosons , was presented in ref .
specifically a low mass and high mass scenario were distinguished in ref .
@xcite , referring to the mass of the lowest lying glueball with quantum numbers @xmath5 .
the latter is expected to be close to 1 gev in the low mass scenario , whereas it is to be sought above or around 1.5 gev in the high mass case . for an appraisal of these ideas
we refer to ref .
lattice gauge theory , in particular without quark flavors indicates preference for the high mass scenario @xcite , @xcite , @xcite , @xcite . in the presence of quarks lattice simulations
are at the moment inconclusive @xcite , @xcite .
in a recent paper @xcite the spectroscopic assignment of the three lowest lying binary gluonic mesons denoted @xmath6 shown in table [ tabsum ] was proposed and discussed .
@xmath7 two main modes of production of @xmath8 were elaborated on in ref .
@xcite : \a ) two to two pseudoscalar meson scattering in the i = 0 channel : @xmath9 the wide structure interrupted by two narrow width resonances , which manifest themselves as almost complete negative interference is referred to as red dragon in ref .
it is interpreted as @xmath10 interfering with the two scalar , isoscalar @xmath2 mesons , ( @xmath11 , @xmath12 ) forming the three flavor nonet .
this contrasts with the assignment proposed in ref .
@xcite where @xmath13 is identified with @xmath10 together with a nontrivial admixture of scalar @xmath2 states .
the production characteristics is represented by the modulus square of the s wave scattering amplitude in the three channels in table [ elast ] , which we reproduce in figure [ gbfig3 ] from ref .
@xcite .
\b ) production in the reaction @xmath14 at central rapidity .
a most remarkable line shape over the whole region of @xmath10 displaying a width of approximately 1 gev is observed in the double pomeron dominated reaction @xmath15 where the momentum transfers from @xmath16 to @xmath17 and from @xmath18 to @xmath19 are constrained to be quite small compared to the center of mass energy .
the line shape as observed by the axial field spectrometer collaboration @xcite is shown in figure [ figcent ] .
it represents the production of the @xmath20 glueball state ( called red dragon ) in ref .
@xcite .
-2 mm -2 mm -2 mm -2 mm it is characterized by a different distribution over the two pion invariant mass than the two to two scattering cross sections in figure [ gbfig3 ] . the difference in shape is a manifestation of the chameleon nature of the leading scalar glueball as the unique genuinely wide resonance . the destructive interference pattern at the @xmath21 scalar resonances @xmath22 and @xmath13 is clearly visible in both production line shapes .
the production of @xmath10 in the decays \c ) @xmath23 \d ) @xmath24 \e ) @xmath25 where it was systematically searched for , remains unresolved for the time beeing .
the axial field spectrometer collaboration @xcite performs a spin analysis of the @xmath26 system up to an invariant mass of 2.3 gev , involving s and d waves .
this analysis @xcite reveals a spin 2 resonance with the mass and width parameters @xmath27 .
these parameters agree very well with those of @xmath28 in table [ tabsum ] .
there is a controversy over the spin of the @xmath29 ( j=0 , 2 ) @xcite .
we refer to the review of particle physics of the particle data group @xcite and to ref .
@xcite for more details .
further experiments and/or analysis are necessary to resolve this issue .
we consider hadron hadron and nucleus nucleus collisions at energies high enough to be dominated by pomeron exchange . in this energy domain , multiplicities of produced particles , including heavy flavors such as strangeness and charm , could exhibit limiting behaviour in their ratios . in this regime
we expect the glueball production to be dominant without imposing rapidity gaps .
we propose to study consequences of the hypothesis that correspondingly dominant particle production proceeds from a gluon rich environment .
this then translates into a major component of @xmath10 production . in this process the lower mass region @xmath30
is assumed to be dominant , similar to the shape of the red dragon observed in central production in the reaction @xmath14 shown in figure [ figcent ] .
it is a general property of a wide resonance that its line shape is not universal but rather chameleon like .
this is supported by the contrast between the line shapes in figures 1 and 2 .
the line shape in figure 2 shows an exponential dependence on the invariant mass , suggestive of a thermal distribution .
we therefore deduce that the production of low masses of the @xmath20 glueball state is preferred in a high energy and high density hadronic environment .
thus the two pion decay mode of @xmath31 becomes dominant , giving rise to characteristic two pion s - wave correlations relative to directly produced pions , which in an approximately thermal and noninteracting model remain uncorrelated .
the glueball associated s - wave component should dominate over direct production of vector mesons @xmath32 , with the main two pion decay proceeding through the @xmath33 .
the selection of central events , i.e. events with a small impact parameter enhances the gluon initiated production of @xmath31 .
we study the following production : @xmath34{lll } \begin{array}{l } h \ + \
h ' \vspace*{0.3 cm } \\ a \
+ \ a ' \end{array } \ & \rightarrow & \ n \ gb \ ( \ 0^{++ } \ ) \ + \ n \ \pi \ + \ x \end{array } \end{array}\ ] ] and the two decay channels : @xmath35{lll } gb \
( \ 0^{++ } \ ) \ \rightarrow \ \pi \ \pi , k \ \overline{k } \end{array } \end{array}\ ] ] @xmath36{lll } gb \
( \ 0^{++ } \ ) \ \rightarrow \ \pi^{+ } \pi^{- } \ \ell^{+ } \ell^{- } \end{array } \end{array}\ ] ] in eq .
( [ eq:1a ] ) n denotes the number of scalar glueballs and n the number of pions emitted independently of glueballs . the semileptonic glueball decay in eq . ( [ eq:3a ] )
proceeds as follows @xmath37 the decay channel in equation ( [ eq:3a ] ) is not characteristic for a glueball .
the gluonic zweig rule suppresses the mixing of glueball and @xmath2 states , but it does not apply to a low mass @xmath20 state because of the strong effective coupling .
this is reflected by the large width ( @xmath38 1 gev ) of the @xmath20 glueball @xcite .
we estimate that the branching fraction for the decay in eq .
( [ eq:3a ] ) is of the order of @xmath39 .
this branching fraction is of the same order as @xmath40 .
the latter is extracted from central @xmath41 production of hadrons @xcite .
the measurement of the @xmath20 glueball state in the high track density environment in the final state of heavy ion collisions could be pursued by searching for the typical inteference pattern displayed in figures ( [ gbfig3 ] ) and ( [ figcent ] ) in the invariant mass distributions of @xmath42 , @xmath43 , @xmath44 and @xmath45 . the decay channel to kaons is less affected by background , however it can reconstruct only the tail of the @xmath20 invariant mass distribution .
especially the @xmath42 mode singles out natural spin parity ( e.g. @xmath20 , @xmath46 , @xmath47 , ... ) .
this fact compensates for the loss of statistics as compared to the @xmath43 channel .
in particular the @xmath48 channel is absent .
+ the characteristic interference pattern could be enhanced applying cuts on kinematic variables like the transverse momentum ( @xmath49 ) and the opening angle of the decay products .
further we put forward the question , whether background subtraction differential in relative momentum of the decay products ( e.g. @xmath42 ) , renders possible an angular momentum analysis overcoming the serious combinatorial background .
this analysis has been succesfully performed in pp collisions @xcite albeit imposing severe rapidity gaps .
+ the measurement of the @xmath46 glueball state through e.g. the decay channel @xmath50 may be possible .
the associated production of the gb @xmath46 is suppressed compared to the @xmath20 state due to its mass .
nevertheless it could be identified especially in the @xmath51 channels because of two things : first it would dominate the production of @xmath46 @xmath2 states of comparable mass and second it would decay through a normal decay width given in table 1 into @xmath26 and @xmath51 . due to the well defined shape of the @xmath52 glueball and its small width it is possible to suppress the background in a more efficient way than for the @xmath20 glueball .
this is achieved through cuts applied to the decay kinematics of the meson pairs ( e.g. @xmath53 ) .
+ the production of the @xmath20 state could have important consequences for the invariant mass distribution of @xmath54 pairs , through the decay channel @xmath55 .
the invariant mass distribution of @xmath41 pairs produced in s+au and pb+au reactions at 200 and 158 gev per nucleon respectively shows a significant excess over expectations based on known sources of @xmath41 pair production @xcite as shown in figure [ na45 ] .
+ we emphasize that the contribution denoted known sources in figure [ na45 ] represents the best knowledge as processed in monte carlo studies of @xmath56 and @xmath57 collisions at comparatively lower energy density than in central pb pb collisions at 158 gev / c per nucleon .
figure [ gb_rho_nocuts ] shows the invariant mass distribution of @xmath41 pairs resulting from the decay @xmath58 and from the assumed copious production and decay of the @xmath20 glueball state : @xmath59 .
+ in this calculation it was assumed that the product of production cross section times branching fraction into @xmath60 of the @xmath61 meson and the @xmath20 glueball state are in the ratio of 1 : 20 . the branching ratio of the @xmath62 is @xmath63 and comparable with the one of @xmath64 .
furthermore it was assumed that the @xmath20 glueball state is distributed in mass as shown in figure [ figcent ] .
the rapidity distribution is taken as gaussian with a width of 0.6 units .
for the transverse mass distribution we choose an exponential shape with inverse slope of 150 mev .
the @xmath65 meson was assumed to have the same rapidity distribution as the @xmath20 and its inverse slope in @xmath66 is taken as 200 mev .
additionally in order to simulate the experimental response we introduced an error of @xmath67 = 5@xmath68 for the momenta of the decay products @xmath41 . in figure [ gb_rho_ptcut ] the influence of the rejection of all dilepton transverse momenta below 0.2 gev is shown .
this cut follows the actual procedure applied to the experimental data of figure [ na45 ] .
the @xmath41 pairs from the decay of the @xmath20 glueball state significantly populate the region of invariant mass where the excess is seen by the na45 experiment .
the decay channel @xmath69 occurs with a branching ratio typically 100 times larger than for the decay in eq .
( [ eq:3a ] ) .
the resulting @xmath70 s would have predominantly very low transverse momenta displayed in figure [ gb_to_gamma ] .
an abnormal production of @xmath70 s from the above decay is negligible compared to bremsstrahlung and @xmath71 decay .
we discussed consequences of dominant @xmath72 glueball production in high energy collisions of hadrons and nuclei .
it is conceivable that the glueball production becomes a dominant part in central nuclear collisions .
the existence of the qgp phase under thermodynamic conditions above a transition temperature of 150 - 200 mev associates directly a gluon rich environment favoring the glueball production mechanism in the nuclear reaction case over hadronic collisions at the same energy . the characteristic interference pattern of the @xmath20 glueball with the @xmath22 and @xmath13 could serve as the signature of this state decaying into a pair of mesons @xmath42 , @xmath43 , @xmath44 and @xmath45 without imposing large rapidity gaps .
however the two meson channels suffer from combinatorial problems , which may be reduced for the @xmath51 channels .
a promising decay channel appears : @xmath73 .
this is not a characteristic glueball decay channel but enhanced through @xmath2 mixing by the large width of glueball @xmath20 .
because the branching fraction of this channel is of the order @xmath39 the opposite sign dilepton signal becomes significant compared to the @xmath74 contribution .
the expected signal is dominant in the dilepton mass region between 200 and 800 mev .
we conjecture that this enhancement is observed in s+au and pb+au collisions at 200 and 158 gev per nucleon respectively at low impact parameter and high energy density .
a dominant @xmath20 glueball component brings into focus an enhanced production of glueball @xmath46(1700 ) . due to the well defined shape of the @xmath46(1700 ) glueball and its small width it appears possible to suppress the background in a more efficient way than for the @xmath20 glueball .
p. de forcrand et al . , _ phys .
lett . _ * b152 * ( 1985 ) 107 ; + c. michael and m. teper , _ nucl .
_ * b314 * ( 1989 ) 347 ; + g. s. bali , a. hulsebos , a. c. irving , c. michael , k. schilling and p. stephenson , _ phys .
_ * b309 * ( 1993 ) 378 ; + h. chen , j. sexton , a. vaccarino and d. weingarten , _ nucl .
b ( proc . suppl . )
_ * 34 * ( 1994 ) 357 . | we elaborate on the hypothesis that in high energy hadron hadron and nucleus nucleus collisions the lowest mass glueballs are copiously produced from the gluon rich environment especially at high energy density .
we discuss the particular glueball decay modes : @xmath0 and @xmath1 . + * * s. kabana + laboratory for high energy physics + and + * p. minkowski * + institute for theoretical physics + university of bern + sidlerstrasse 5 + ch - 3012 bern , switzerland + 15 .
june 1999 |
liquid crystal anchoring effects have been intensively studied both experimentally and theoretically during the past decades @xcite .
such an interest is well understood since most liquid crystal devices are cells comprising orienting surfaces with a liquid crystal in between .
typically , aligning surfaces provide a uniform orientation of the liquid crystal director in the cell interior . on the phenomenological level ,
liquid crystal anchoring can be described by two basic parameters : the easy axis direction , @xmath0 , and the anchoring coefficient @xmath1 .
these two parameters are critical design parameters for every liquid crystal device @xcite . a variety of _ experimental _ methods have been proposed to measure anchoring parameters , in particular the anchoring coefficient @xmath1 @xcite .
most of them measure surface director deviations in an external field and involve rather complicated optical setups . in spite of the practical importance , the mechanism of the director alignment is still not well understood .
experimental techniques always involve optical measurements .
they test the entire liquid crystal cell and therefore can not provide a satisfactory description of the thin interface region .
theoretical investigations of anchoring are also quite controversial .
for example , the usual continuous phenomenological theory has divergent surface terms in the elastic free energy expansion @xcite .
one of the approaches for the systematic investigation of anchoring phenomena is computer simulation of liquid crystals in confined geometries .
indeed , computer simulation is a well established method to treat bulk elastic coefficients @xcite , the surface anchoring strength @xcite , and structures of disclination cores @xcite .
this means that computer simulation allows investigation of details of the liquid crystalline structure which can not be resolved experimentally .
several papers have already been published trying to search for reasonable surface potentials to use in simulations .
however , most of them do not characterize aligning surfaces using well established parameters .
questions about the formation of a solid interface layer , values of the easy axis angle and anchoring coefficient are still open .
the reason for this is probably a lack of reliable methods to measure these parameters by computer simulation . in this paper
, we propose a technique to measure the surface anchoring strength @xmath1 by computer simulation .
the method itself is based on the study of the _ director fluctuations _ in the liquid crystal cell .
a similar approach has already been used for the _ experimental _ characterization of the interface @xcite and is based on measuring the light scattering caused by the director fluctuations . in computer simulation ,
the director fluctuations can be studied directly , using ensemble averages of functions of the second - rank order tensor components .
a fit to the fluctuation amplitudes with equations predicted by elastic theory then allows determination of the surface anchoring strength .
large length- and time - scale fluctuations of the director @xmath2 can be described in the continuum model of liquid crystals , based on the phenomenological elastic constants . in this approach ,
the hydrodynamic equations for the director and the boundary conditions can be obtained by minimization of the cell free energy @xcite : @xmath3 where @xmath4 is a hydrodynamic term with an effective viscosity coefficient @xmath5 , @xmath6 is the frank elastic free energy , and @xmath7 is the surface anchoring energy . in
what follows we use the one - elastic - constant approximation , i.e. @xmath8 .
then the frank free energy can be brought into the form : @xmath9 d{\bf r } \:.\ ] ] the liquid crystal cell is bounded by surfaces @xmath10 which provide some kind of anchoring condition @xcite . below we consider _ homeotropic _ anchoring , that is , normal to the surface .
_ planar _ anchoring can be treated in the same way .
we assume that the interaction of the director with the cell surfaces has the rapini - papoular form @xcite : @xmath11 where unit vectors @xmath12 define directions of the easy axes at @xmath10 : @xmath13 , @xmath14 ; @xmath1 is the anchoring energy , and @xmath15 .
equations for the director and boundary conditions can be obtained by minimization of the cell free energy eqn ( [ eqn : f_total ] ) . in the case of homeotropic anchoring on both surfaces the stationary director distribution in the cell is homeotropic , i.e. @xmath16 .
therefore , we have to investigate small perturbations of the director around the distribution : @xmath17 minimizing the total free energy ( [ eqn : f_total ] ) and linearizing the equations for the director and boundary conditions with respect to @xmath18 , we obtain equations @xmath19 and boundary conditions @xmath20 for the fluctuations . taking into account eqn .
( [ eqn : director ] ) , the expression for the free energy fluctuations can be rewritten in the form of the average of the self - conjugate ( hermitian ) operator @xmath21 : @xmath22 the eigenfunctions of the operator @xmath21 , which satisfy boundary conditions ( [ eqn : boundary ] ) , form a complete set of orthogonal functions characterized by wave vector @xmath23 .
therefore , @xmath24 can be expanded in a series of these orthogonal functions : @xmath25 \ : , \label{eqn : fluct}\end{aligned}\ ] ] where @xmath26 , and @xmath27 here we have introduced the dimensionless wave vector @xmath28 and anchoring parameter @xmath29 where @xmath30 is the extrapolation length @xcite .
the wave vectors @xmath31 form a discrete spectrum because of confinement in the @xmath32 direction which depends on the anchoring energy @xmath1 ( see appendix for details ) .
the explicit form of the @xmath31 spectrum is given by the secular equation : @xmath33 each individual mode can now be seen from eqn ( [ eqn : hydro ] ) to relax exponentially with a relaxation time given by @xmath34 substituting expansion ( [ eqn : fluct ] ) into the free energy ( [ eqn : denergy ] ) , and performing the integration over the cell volume we obtain : @xmath35 integrating , we took into account the orthogonality of the eigenfunctions in the expansion ( [ eqn : fluct ] ) with different eigenvectors @xmath23 , which allowed us to reduce the summations over @xmath23 , @xmath36 to a single sum over @xmath23 .
application of the equipartition theorem of classical statistical mechanics , just as for elastic fluctuations in bulk liquid crystals @xcite , gives the fluctuation amplitudes : @xmath37 where @xmath38 denotes an ensemble average .
in molecular simulations , rather than measuring director fluctuations , it is more convenient to measure fluctuations of the second - rank order tensor components ( following forster @xcite ) .
we define the real - space order tensor density @xmath39 where @xmath40 , in terms of the orientation vectors @xmath41 of each molecule @xmath42 ( we consider only uniaxial molecules ) .
if we assume that there is no variation in the _ degree _ of ordering , we may write @xmath43 where @xmath44 is the order parameter , i.e. the largest eigenvalue of @xmath45 . if the director @xmath46 is taken to lie along the @xmath32 axis throughout the sample , the off - diagonal components @xmath47 , @xmath48 , are proportional to the fluctuations of the corresponding director components @xmath49 this is the situation for homeotropic anchoring at both surfaces
. for planar anchoring , with the director along @xmath50 and the surface normal along @xmath32 , the components @xmath51 , @xmath52 are important ( and non - equivalent ) .
measurements are performed directly in reciprocal space .
the fourier transform of the real - space order tensor is @xmath53 then the fluctuations @xmath54 can be easily measured from simulations : @xmath55 \ : .
\label{eqn : simul}\ ] ] we explicitly relate simulation - measured fluctuation modes with theoretically predicted amplitudes of the director fluctuations for @xmath56 @xmath57\ ] ] where @xmath58 .
note that the @xmath31 take discrete ( but not equally spaced ) values as discussed earlier , while the @xmath59 values are unrestricted . since @xmath24 is real , using eqn ( [ eqn : fluct ] ) we have @xmath60^{\ast } = -\frac{\xi^{2}+\chi^{2}}{\left ( \xi + { \rm i}\chi \right ) ^{2 } } \delta { \bf n}^{(+)}(-{\bf q}_{\bot},q_{z } ) \:.\ ] ] taking this equation into account , and the fact that fluctuations with different wavevectors are independent , i.e. @xmath61 if @xmath62 , the corresponding ensemble average of the squared order parameter can be rewritten as @xmath63 we measure @xmath44 and @xmath64 from simulations , eqn ( [ eqn : simul ] ) , and then compare with the theoretical prediction , eqn ( [ eqn : theory ] ) , which is parametrized by @xmath65 , @xmath30 and @xmath66 .
both the permitted @xmath67 spectrum , and the variation of @xmath68 with @xmath69 are sensitive to the anchoring strength parameter @xmath70 .
fluctuation amplitudes given by eq .
( [ eqn : theory ] ) have features that simplify the fitting procedure .
first , terms with small @xmath31 values dominate because of the @xmath71 in the denominator of eq ( [ eqn : theory ] ) .
therefore , it is always possible to truncate this sum and use only the first values of the @xmath31 spectrum .
then , for large @xmath59 or for @xmath72 , eq . ( [ eqn : theory ] ) can be simplified so that the dependence on @xmath59 is explicit : @xmath73 ^{2}\sum_{q_{z}}~\frac{\chi ^{2}}{q_{z}^{2}\left ( 2\xi + \chi ^{2}+\xi ^{2}\right ) } .
\label{eqn : theory : approx}\ ] ] therefore , for large @xmath59 , the fluctuation amplitude @xmath74 has a characteristic oscillation with the period given by @xmath75 .
this means that we can adjust the cell thickness @xmath65 independently of parameters @xmath76 and @xmath66 by examining characteristic wavelength of the fluctuation amplitude @xmath77 .
to test the technique proposed , we simulated a liquid crystal confined between parallel walls ( slab geometry ) , with finite homeotropic anchoring at the walls .
the director fluctuations occur around the preferred , _ uniform _ , alignment perpendicular to the walls .
we performed monte carlo ( mc ) simulation of the liquid crystal system .
we used a molecular model which has been studied earlier in this geometry @xcite .
the molecules in this study were modelled as hard ellipsoids of revolution of elongation @xmath78 , where @xmath79 is the length of the semi - major axis and @xmath80 the length of the two equal semi - minor axes . the phase diagram and properties of this family of models
are well studied @xcite .
it is useful to express the density as a fraction of the close - packed density @xmath81 of perfectly aligned hard ellipsoids , assuming an affinely - transformed face - centred cubic or hexagonal close packed lattice . in this case
, the isotropic - nematic phase transition occurs at quite a low density , @xmath82 , and the simulations are performed at a state point corresponding to @xmath83 , for which the nematic order parameter is @xmath84 . for this model ,
temperature is not a significant thermodynamic quantity , so it is possible to choose @xmath85 throughout .
the slab geometry is defined by two hard parallel confining walls , which can not be penetrated by the _ centres _ of the ellipsoidal molecules .
packing considerations generate homeotropic ordering at the surface .
surface anchoring in this system has been studied recently @xcite for a system with wall separation @xmath86 , by applying an orienting perturbation at one of the walls and observing the response at the other .
this yielded an estimate of the extrapolation length @xmath87 . in the current work , simulations were carried out for systems of @xmath88 particles with wall separations @xmath89 , which ( from the above estimate of @xmath30 ) would correspond to surface anchoring parameters in the range @xmath90 . to be sure that we have the same state point for each simulation , we adjusted the density to have the same @xmath91 component of the pressure tensor for all systems .
then a sequence of runs was carried out using the constant-@xmath92 ensemble , allowing typically @xmath93 mc sweeps for equilibration and @xmath94 sweeps for accumulation of averages ( one sweep is one attempted move per particle ) .
the simulation results were analysed to give a density profile , and an order tensor profile , which are shown in figs [ fig:1 ] , [ fig:2 ] . from these profiles
we can see that the walls are sufficiently well separated , and the variation of the order parameter across the slab is small , even in spite of the large change in local density near the walls .
density profiles as functions of @xmath32-coordinate .
distances are normalized by the molecular semi - axis length @xmath79 and the density is expressed relative to the closed - packed density @xmath95 for perfectly aligned ellipsoids .
the profiles are symmetrical , only one side of the slab is shown .
the results for different wall separations are almost indistinguishable on this scale .
, width=302 ] profiles of nematic order parameter @xmath44 for different wall separations .
distances are normalized by the molecular semi - axis length @xmath79 .
the profiles are symmetrical , only one side of the slab is shown .
note the highly expanded vertical scale .
, width=302 ] the order tensor fluctuations in reciprocal space were calculated using expression ( [ eqn : simul ] ) . to fit the simulation results with the elastic theory we have to remember that the size of the simulation box @xmath96 is not necessarily equal to the liquid crystal cell thickness @xmath65 appearing in the elastic theory .
the former is a physical quantity , which , in statistical mechanics , is determined by the positions at which the liquid number density becomes identically zero .
the latter appears in the elastic theory : it is determined by the positions at which the orientational elastic boundary conditions are applied .
physically , the difference may be ascribed to partial penetration of the walls by the liquid crystal molecules , formation of an ordered ( or solid ) layer near the walls , or other molecular - scale features .
we assume that we may write @xmath97 , where the value @xmath98 ( which may be positive or negative ) is characteristic of the wall , independent of @xmath99 , and may be determined in our fitting process .
the best estimate of the wall - induced separation distance @xmath98 was obtained examining the ratios @xmath100 for different @xmath101 since they reveal more structure for large @xmath59 .
plotting the results in this way removes the overall scaling of the amplitudes of fluctuations , which are sensitive to changes in @xmath30 , while the shapes of the curves , and the characteristic oscillation wavelengths , are sensitive to the choice of @xmath98 .
these ratios with @xmath102 , @xmath103 and corresponding fitting curves are plotted in fig .
[ fig:3 ] .
the best fits were obtained with @xmath104 . in this figure
we also plot theoretical curves with @xmath105 : the clear discrepancy with the simulation results indicates that the simulation box size @xmath96 is indeed significantly different from the actual cell thickness .
ratio of the director fluctuations as function of wavevector ( normalized by the molecular semi - axis length @xmath79 ) for different wall separations .
symbols : monte carlo results ; solid lines : elastic theory ; dashed lines : elastic theory , without correction for the difference between the elastic - theory cell thickness @xmath65 and the simulation box size @xmath99 .
, width=302 ] following the determination of @xmath98 , we adjusted the extrapolation length @xmath30 to obtain the best fit to the fluctuation data : the fluctuation amplitudes with small @xmath59 are most sensitive to this quantity .
together with the corresponding fitting curves for the different slab thicknesses , our results are plotted in fig .
[ fig:4 ] .
the best fits were obtained with a bulk elastic constant @xmath106 and an extrapolation length @xmath107 .
the theoretical fitting curves agree well with the simulation results , for small values of @xmath59 , as one would expect for a theory valid for long wavelength fluctuations . at higher
@xmath59 , the structure ( emphasized in the inset of fig . [ fig:4 ] by a multiplying factor @xmath108 ) is not perfectly reproduced , but the agreement is satisfactory .
this is not surprising , since we expect the elastic theory to become less accurate at higher @xmath109 .
finally , we note that the extrapolation distance relative to the simulation wall position is @xmath110 , which compares moderately well with the value @xmath111 obtained in the previous study of this system @xcite .
it should be noted that the director configuration of that work does not allow one to determine , separately , @xmath98 and @xmath30 , so the quoted value of @xmath30 really represents @xmath112 .
we have to point out some possible limitations of the method .
the first one is computational time .
the effect of the surfaces is to dampen the amplitude of long - wavelength modes ; these have the longest relaxation times , according to eqn ( [ eqn : time ] ) , and so it is essential to carry out very long runs to adequately sample them .
we have paid some attention to estimating the error bars on the measured values of @xmath113 , as indicated in fig [ fig:4 ] : the larger values at low @xmath109 follow directly from this effect .
we shall return to examine the time - dependence of fluctuations in a future publication .
the second limitation is the actual sensitivity of the measured averages to the variation in the anchoring strength and cell thickness .
one might expect that in practice it is not possible to measure large values of the anchoring parameter @xmath114 , so we need reasonably thin cells .
as the cell thickness @xmath99 becomes large , the fluctuation spectrum becomes insensitive to @xmath115 .
however , it is important that the walls do not become too close : the _ bulk _ region should be sufficiently large compared to the _ interfacial _ region . only in this case
can we assume that the scalar order parameter @xmath44 in the liquid crystal bulk is constant for large scale fluctuation modes .
director fluctuations ( arbitrary units ) as function of wavevector ( normalized by the molecular semi - axis length @xmath79 ) for different wall separations .
symbols : monte carlo results ; error estimates are indicated at some representative points , at higher wave - vectors the errors are smaller than the plotting symbols .
solid lines : elastic theory , fitted to parameters discussed in the text .
inset : fluctuations multiplied by @xmath116 to emphasize structure at higher wavenumbers .
successive curves are offset by 0.5 for clarity .
, width=302 ] to summarize , analysis of the director fluctuations in nematic liquid crystal slabs allowed us to measure the surface anchoring strength parameter .
the method has been tested for a system of hard ellipsoids of revolution of elongation @xmath117 confined between hard walls with homeotropic anchoring .
careful analysis of fluctuations in slabs of different thickness has allowed us to resolve , for the first time , the position of the elastic boundary condition relative to the simulation wall , as well as measuring the extrapolation length .
the elastic theory gives a good description at low wavenumbers , but is less accurate at higher wavenumbers .
this research was supported by epsrc . d.a .
acknowledges support through grant ors/99007015 of the overseas research students award ; g.g . acknowledges the support of a british council grant ; m.p.a .
is grateful to the alexander von humboldt foundation .
some of this work was carried out while visiting the max planck institute for polymer research , mainz , and the institute of physics , university of mainz ; the authors are grateful to kurt kremer and kurt binder for their hospitality , and conversations with f. schmid and h. lange are gratefully acknowledged .
to obtain the spectrum of the @xmath31 modes of the fluctuations we need to select from the eigenfunctions of the operator @xmath21 those which satisfy the boundary conditions ( [ eqn : boundary ] ) .
substituting eqn ( [ eqn : fluct ] ) into the boundary conditions ( [ eqn : boundary ] ) , we obtain a set of linear equations @xmath118 this set of linear homogeneous equations for @xmath119 has a nontrivial solution if its determinant equals zero .
this condition leads to the secular equation for the @xmath31 vector ( [ eqn : secular ] ) and relation ( [ eqn : nplusminus ] ) . in the case of strong anchoring , @xmath120 ,
the @xmath31 spectrum is equidistant : @xmath121 , where @xmath122 is a positive integer . for finite anchoring coefficient @xmath123
, we have a shift in this spectrum .
the magnitude of the shift depends on the anchoring parameter @xmath123 .
indeed , for sufficiently strong anchoring parameters , @xmath124 , asymptotically : @xmath125 which is equivalent to replacing @xmath65 by @xmath126 , @xmath76 being the extrapolation length .
the spectrum of the @xmath129,@xmath130 wavevectors depends on the system geometry . again
, if we have periodic boundary conditions in @xmath50 and @xmath131 directions , the @xmath129 and @xmath130 have a discrete spectrum , on a fine grid @xmath132 , @xmath133 , otherwise @xmath134 are unrestricted .
it is also easy to show that the eigenfunctions which correspond to different eigenvalues @xmath31 and @xmath135 are orthogonal .
indeed , using eqn.([eqn : nplusminus ] ) we can rewrite @xmath136 \delta { \bf n}^{(+)}({\bf q}_{\bot},q_{z } ) \:.\end{aligned}\ ] ] it is easy to check using the secular equation ( [ eqn : secular ] ) that functions @xmath137 are orthogonal , i.e.@xmath138 where @xmath139 is the kronecker delta .
therefore , the eigenfunctions @xmath140 are orthogonal and can be normalized using eqn ( [ eqn : norm ] ) . | we propose a simple and reliable method to measure the liquid crystal surface anchoring strength by molecular simulation .
the method is based on the measurement of the long - range fluctuation modes of the director in confined geometry . as an example
, molecular simulations of a liquid crystal in slab geometry between parallel walls with homeotropic anchoring have been carried out using the monte carlo technique . by studying different slab thicknesses ,
we are able to calculate separately the position of the elastic boundary condition , and the extrapolation length . |
the analysis of financial data by methods developed for physical systems has a long tradition@xcite and has recently attracted the interest of physicists . among the reasons for this interest
is the scientific challenge of understanding the dynamics of a strongly fluctuating complex system with a large number of interacting elements .
in addition , it is possible that the experience gained by studying complex physical systems might yield new results in economics .
financial markets are complex dynamical systems with many interacting elements that can be grouped into two categories : ( i ) the _ traders _ such as individual investors , mutual funds , brokerage firms , and banks and ( ii ) the _ assets _ such as bonds , stocks , futures , and options . interactions between these elements lead to transactions mediated by the stock exchange .
the details of each transaction are recorded for later analysis .
the dynamics of a financial market are difficult to understand not only because of the complexity of its internal elements but also because of the many intractable external factors acting on it , which may even differ from market to market . remarkably , the statistical properties of certain observables appear to be similar for quite different markets@xcite , consistent with the possibility that there may exist `` universal '' results .
the most challenging difficulty in the study of a financial market is that the nature of the interactions between the different elements comprising the system is unknown , as is the way in which external factors affect it .
therefore , as a starting point , one may resort to empirical studies to help uncover the regularities or `` empirical laws '' that may govern financial markets .
the interactions between the different elements comprising financial markets generate many observables such as the transaction price , the share volume traded , the trading frequency , and the values of market indices [ fig .
[ sp500_brownian ] ] .
a number of studies investigated the time series of returns on varying time scales @xmath0 in order to probe the nature of the stochastic process underlying it . for a time series @xmath9 of prices or market index values ,
the return @xmath10 over a time scale @xmath0 is defined as the forward change in the logarithm of @xmath9@xcite , @xmath11 for small changes in @xmath9 , the return @xmath12 is approximately the forward relative change , @xmath13 in 1900 , bachelier proposed the first model for the stochastic process of returns an uncorrelated random walk with independent , identically gaussian distributed ( _ i.i.d _ ) random variables@xcite .
this model is natural if one considers the return over a time scale @xmath0 to be the result of many independent `` shocks '' , which then lead by the central limit theorem to a gaussian distribution of returns @xcite .
however , empirical studies @xcite show that the distribution of returns@xcite has pronounced tails in striking contrast to that of a gaussian . to illustrate this fact ,
we show in fig .
[ gt_brownian ] the 10@xmath3min returns of the s&p 500 market index@xcite for 1986 - 1987 and contrast it with a sequence of _ i.i.d . _ gaussian random variables . both are normalized to have unit variance .
clearly , large events are very frequent in the data , a fact largely underestimated by a gaussian process . despite this empirical fact
, the gaussian assumption for the distribution of returns is widely used in theoretical finance because of the simplifications it provides in analytical calculation ; indeed , it is one of the assumptions used in the classic black - scholes option pricing formula@xcite . in his pioneering analysis of cotton prices
, mandelbrot observed that in addition to being non - gaussian , the process of returns shows another interesting property : `` time scaling '' that is , the distributions of returns for various choices of @xmath0 , ranging from 1 day up to 1 month have similar functional forms@xcite .
motivated by ( i ) pronounced tails , and ( ii ) a stable functional form for different time scales , mandelbrot@xcite proposed that the distribution of returns is consistent with a lvy stable distribution@xcite that is , the returns can be modeled as a lvy stable process .
lvy stable distributions arise from the generalization of the central limit theorem to random variables which do not have a finite second moment [ see appendix a ] .
conclusive results on the distribution of returns are difficult to obtain , and require a large amount of data to study the rare events that give rise to the tails .
more recently , the availability of high frequency data on financial market indices , and the advent of improved computing capabilities , has facilitated the probing of the asymptotic behavior of the distribution .
for these reasons , recent empirical studies@xcite of the s&p 500 index@xcite analyze typically @xmath14@xmath15 data points , in contrast to approximately 2000 data points analyzed in the classic work of mandelbrot@xcite .
reference @xcite reports that the central part of the distribution of s&p 500 returns appears to be well fit by a lvy distribution , but the asymptotic behavior of the distribution of returns shows faster decay than predicted by a lvy distribution . hence , ref .
@xcite proposed a truncated lvy distribution a lvy distribution in the central part followed by an approximately exponential truncation as a model for the distribution of returns .
the exponential truncation ensures the existence of a finite second moment , and hence the truncated lvy distribution is not a stable distribution @xcite . the truncated lvy process with _ i.i.d .
_ random variables has slow convergence to gaussian behavior due to the lvy distribution in the center , which could explain the observed time scaling for a considerable range of time scales@xcite .
in addition to the probability distribution , a complementary aspect for the characterization of any stochastic process is the quantification of correlations .
studies of the autocorrelation function of returns show exponential decay with characteristic decay times @xmath16 of only 4@xmath3min@xcite . as is clear from fig .
[ acorsp](a ) , for time scales beyond 20 @xmath3min the correlation function is at the level of noise , in agreement with the _ efficient market hypothesis _ which states that is not possible to predict future stock prices from their previous values@xcite . if price - correlations were not short - range , one could devise a way to make money from the market indefinitely .
it is important to note that lack of linear correlation does not imply an _ i.i.d .
_ process for the returns , since there may exist higher - order correlations [ fig [ acorsp](b ) ] .
indeed , the amplitude of the returns , referred to in economics as the _ volatility_@xcite , shows long - range time correlations that persist up to several months , and are characterized by an asymptotic power - law decay .
a recent preliminary study reported that the distributions of 5@xmath3min returns for 1000 individual stocks and the s&p 500 index decay as a power - law with an exponent well outside the stable lvy regime@xcite .
consistent results were found by studies both on stock markets @xcite and on foreign exchange markets@xcite .
these results raise two important questions : first , the distribution of returns has a finite second moment , thus , we would expect it to converge to a gaussian because of the central limit theorem . on the other hand ,
preliminary studies suggest the distributions of returns retain their power - law functional form for long time scales .
so , we can ask which of these two scenarios is correct ?
we find that the distributions of returns retain their functional form for time scales up to approximately 4 days , after which we find results consistent with a slow convergence to gaussian behavior .
second , power - law distributions are not stable distributions , but the distribution of returns retains its functional form for a range of time scales .
it is then natural to ask how can this _ scaling behavior _ possibly arise ?
one possible explanation is the recently - proposed exponentially - truncated lvy distribution@xcite .
however , the truncated lvy process is constructed out of _ i.i.d . _ random variables and hence is not consistent with the empirically - observed long persistence in the autocorrelation function of the volatility of returns . moreover
, our data support the possibility that the asymptotic nature of the distribution is a power - law with an exponent outside the lvy regime .
also , we will argue that the scaling behavior observed in the distribution of returns may be connected to the slow decay of the volatility correlations .
the organization of the paper is as follows .
section iii describes the data analyzed .
sections iv and v study the distribution of returns of the s&p 500 index on time scales @xmath17 day and @xmath18 day , respectively .
section vi discusses how time correlations in volatility are related to the time scaling of the distributions , and sect .
vii presents concluding remarks .
first , we analyze the s&p 500 index , which comprises 500 companies chosen for market size , liquidity , and industry group representation in the us . the s&p 500 is a market - value weighted index ( stock price times number of shares outstanding ) , with each stock s weight proportional to its market value .
the s&p 500 index is one of the most widely used benchmarks of u.s . equity performance . in our study , we first analyze database ( i ) which contains `` high - frequency '' data that covers the 13 years period 19841996 , with a recording frequency of less than 1 min .
the total number of records in this database exceeds @xmath19 . to investigate longer time scales
, we study two other databases . database ( ii )
contains daily records of the s&p 500 index for the 35-year period 19621996 , and database ( iii ) contains monthly records for the 71-year period 19261996 . in order to test if our results are limited to the s&p 500 index , we perform a parallel analysis on two other market indices .
database ( iv ) contains 3560 daily records of the nikkei index of the tokyo stock exchange for the 14-year period 19841997 , and database ( v ) contains 4649 daily records of the hang - seng index of the hong kong stock exchange for the 18-year period 19801997 .
first , we analyze the values of the s&p500 index from the high - frequency data for the 13-year period 19841996 , which extends the database studied in ref . @xcite by an additional 7 years . the data are typically recorded at 15 second intervals .
we first sample the data at 1 min intervals and generate a time series @xmath9 with approximately 1.2 million data points . from the time series @xmath9 , we compute the return @xmath21 which is the relative change in the index , defined in eq .
( [ return ] ) . in order to compare
the behavior of the distribution for different time scales @xmath0 , we define a normalized return @xmath22 @xmath23 here , the time averaged volatility @xmath24 is defined through @xmath25 and @xmath26 denotes an average over the entire length of the time series .
figure [ sp500_hist](a ) shows the cumulative distribution of returns for @xmath20 min . for both positive and negative tails ,
we find a power - law asymptotic behavior @xmath27 similar to what was found for individual stocks@xcite . for the region @xmath28
, regression fits yield @xmath29 well outside the lvy stable range , @xmath30 .
consistent values for @xmath7 are also obtained from the density function . for a more accurate estimation of the asymptotic behavior , we use the modified hill estimator [ fig . [ sp500_hill](a ,
we obtain estimates for the asymptotic slope in the region @xmath28 : @xmath31 for the region @xmath32 , regression fits yield smaller estimates of @xmath7 , consistent with the possibility of a lvy distribution in the central region .
the values of @xmath7 obtained in this range are quite sensitive to the bounds of the region used for fitting .
our estimates range from @xmath33 up to @xmath34 for different fitting regions in the interval @xmath35 .
for example , in the region @xmath36 , we obtain @xmath37 which are consistent with the result @xmath38 found for small values of @xmath39 in ref .
note that in ref .
@xcite the estimates of @xmath7 were calculated using the scaling form of the return probability to the origin @xmath40 .
it is possible that for the financial data analyzed here , @xmath40 is not the optimal statistic , because of the discreteness of the individual - company distributions that comprise it@xcite .
it is also possible that our values of @xmath7 for small values of @xmath39 could be due to the discreteness in the returns of the individual companies comprising the s&p 500 .
next , we study the distribution of normalized returns for longer time scales . figure [ sp500_scl](a ) shows the cumulative distribution of normalized s&p 500 returns for time scales up to 512 min ( approximately 1.5 days ) .
the distribution appears to retain its power - law functional form for these time scales .
we verify this scaling behavior by analyzing the moments of the distribution of normalized returns @xmath39 , @xmath41 where @xmath42 denotes an average over all the normalized returns for all the bins . since @xmath5 , we expect @xmath43 to diverge for @xmath44 , and hence we compute @xmath43 for @xmath45 .
figure [ sp500_scl](b ) shows the moments of the normalized returns @xmath39 for different time scales from 5 min up to 1 day .
the moments do not vary significantly for the above time scales , confirming the apparent scaling behavior of the distribution observed in fig .
[ sp500_scl](a ) .
for time scales beyond 1 day , we use database ( ii ) which contains daily - sampled records of the s&p 500 index for the 35-year period 19621996 .
figure [ sp500_1d](a ) shows the agreement between distributions of normalized s&p 500 daily - returns from database ( i ) , which contains 1@xmath3min sampled data , and database ( ii ) , which contains daily - sampled data .
regression fits for the region @xmath47 give estimates of @xmath5 .
figure [ sp500_1d](b ) shows the scaling behavior of the distribution for @xmath48 1 , 2 , and 4 days . for these choices of @xmath0 ,
the scaling behavior is also visible for the moments [ fig .
[ sp500_1d](c ) ] .
figure [ sp_scl_brk](a ) shows the distribution of the s&p 500 returns for @xmath49 and 16 days .
the data are now consistent with a slow convergence to gaussian behavior .
this is also visible for the moments [ fig .
[ sp_scl_brk](b ) ] .
the s&p 500 is but one of the many stock market indices .
hence , we investigate if the above results regarding the power - law asymptotic behavior of the distribution of returns hold for other market indices as well .
figure [ universality ] compares the distributions of daily returns for the nikkei index of the tokyo stock exchange and the hang - seng index of the hong kong stock exchange with that of the s&p 500 .
the distributions have similar functional forms , suggesting the possibility of `` universal '' behavior of these distributions .
in addition , the estimates of @xmath7 from regression fits , @xmath50 are in good agreement for the three cases .
the behavior of the time - averaged volatility @xmath51 as a function of the time scale @xmath0 is shown in fig .
[ acorsp](c ) .
we find a power - law dependence , @xmath52 we estimate @xmath53 for time scales @xmath54 min . this value is larger than @xmath55 due to the exponentially - damped time correlations , which are significant up to approximately 20 min . beyond 20 min , @xmath56 , indicating the absence of correlations in the returns , in agreement with fig .
[ acorsp](a ) .
the time - averaged volatility is also consistent with essentially uncorrelated behavior for the daily and monthly returns .
we have presented evidence that the distributions of returns retain the same functional form for a range of time scales[see fig .
[ alpha.fig ] and table [ alpha.exp ] ] . here ,
we investigate possible causes of this scaling behavior .
previous explanations of scaling relied on lvy stable@xcite and exponentially - truncated lvy processes@xcite . however , the empirical data that we analyze are not consistent with either of these two processes . here
, we compare the rate of convergence of the probability of the returns to that of a computer - generated time series which has the same distribution but is statistically independent by construction .
this way , we will be able to study the convergence to gaussian behavior of independent random variables distributed as a power - law , with an exponent @xmath5 .
first , we generate a time series @xmath57 distributed as @xmath58 .
we next calculate the new random variables @xmath59 , and compute the cumulative distributions of @xmath60 for increasing values of @xmath61 .
these distributions show faster convergence with increasing @xmath61 than the distributions of returns [ fig .
[ sim_pow4](a ) ] .
this convergence is also visible in the moments .
figures [ sim_pow4](a , b ) show that for @xmath62 , both the moments and the cumulative distribution show gaussian behavior .
in contrast , for the distribution of returns , we observe significantly slower convergence to gaussian behavior : in the case of the s&p 500 index , one observes a possible onset of convergence for @xmath63 4 days ( 1560 mins ) , starting from 1@xmath3min returns .
these results confirm the existence of time dependencies in the returns .
next , we show that the scaling behavior observed for the s&p 500 index no longer holds when we destroy the dependencies between the returns at different times .
we start with the 1@xmath3min returns and then destroy all the time dependencies that might be present by shuffling the time series of @xmath64 , thereby creating a new time series @xmath65 which contains _ statistically - independent _ returns . by adding up @xmath61 consecutive returns of the shuffled series @xmath65 , we construct the @xmath61@xmath3min returns @xmath66 .
figure [ sp_shuffle](a ) shows the cumulative distribution of @xmath67 for increasing values of @xmath61 .
we find a progressive convergence to gaussian behavior with increasing @xmath61 .
this convergence to gaussian behavior is also clear in the moments of @xmath67 , which rapidly approach the gaussian values with increasing @xmath61 [ fig .
[ sp_shuffle](b ) ] .
this rapid convergence confirms that the time dependencies cause the observed scaling behavior .
we have presented a detailed analysis of the distribution of returns for market indices , for time intervals @xmath0 ranging over roughly 4 orders of magnitude , from 1@xmath3min up to 1 month ( @xmath68 16,000 min ) .
we find that the distribution of returns is consistent with a power - law asymptotic behavior , characterized by an exponent @xmath5 , well outside the stable lvy regime @xmath69 . for time scales @xmath70 , where @xmath8 days , our results are consistent with slow convergence to gaussian behavior .
we have also demonstrated that the scaling behavior does not hold if we destroy all the time dependencies by shuffling .
the breakdown of the scaling behavior of the distribution of returns upon shuffling the time series suggests that the long - range volatility correlations , which persist up to several months , may be one possible reason for the observed scaling behavior .
recent studies @xcite show that the distribution of volatility is consistent with an asymptotic power - law behavior with exponent 3 , just as observed for the distribution of returns .
this finding suggests that the process of returns may be written as @xmath71 where @xmath72 denotes the return at time @xmath73 , @xmath74 denotes the volatility , and @xmath75 is an _ i.i.d . _
random variable _ independent _ of @xmath74 .
since the asymptotic behavior of the distributions of @xmath74 and @xmath72 is consistent with power - law behavior , @xmath75 should have an asymptotic behavior with faster decay than either @xmath72 or @xmath74 .
in fact , eq . ( [ garch ] ) is central to all the arch models @xcite , with @xmath75 assumed to be gaussian distributed .
different arch processes assume different recursion relations for @xmath74 . in the standard arch model , @xmath76 , leading to a power law distribution of returns with exponent depending on the parameters @xmath7 and @xmath77 .
however , the standard arch process predicts a volatility correlation that decays exponentially , since @xmath74 depends only on the previous event , and can not account for the observed long - range persistence in @xmath74 . to try to remedy this
, one can require @xmath74 to depend not only on the previous value of @xmath72 but on a finite number of past events .
this generalization is called the garch model .
dependence of @xmath74 on the finite past leads not to a power - law decay ( as is observed empirically ) , but to volatility correlations that decay exponentially with larger decay times as the number of events `` remembered '' is increased . in order to explain the long range persistence of the autocorrelation function of the volatility
, one must assume that @xmath74 depends on all the past rather than a finite number of past events @xcite .
such a description would be consistent with the empirical finding of long - range correlations in the volatility , and the observation that the distributions of @xmath72 and @xmath74 have similar asymptotic forms . if the process of returns were governed by the volatility , as in eq .
( [ garch ] ) , then the volatility would seem to be the more fundamental process .
in fact , it is possible that the volatility is a measure of the amount of information arriving into the market , and that the statistical properties of the returns may be `` driven '' by this information .
we thank j .- p .
bouchaud , m. barthlemy , s. v. buldyrev , p. cizeau , x. gabaix , i. grosse , s. havlin , k. illinski , p. ch .
ivanov , c. king , c .- k .
peng , b. rosenow , d. sornette , d. stauffer , s. solomon , j. voit and especially r. n. mantegna for stimulating discussions and helpful suggestions .
the authors also thank bob tompolski for his help throughout this work .
mm thanks dfg and lana thanks fct / portugal for financial support . the center for polymer studies
is supported by the nsf .
lvy stable distributions arise from the generalization of the central limit theorem to a wider class of distributions . consider the partial sum @xmath78 of independent identically distributed ( _ i.i.d .
_ ) random variables @xmath79 . if the @xmath79 s have finite second moment , the central limit theorem holds and @xmath80 is distributed as a gaussian in the limit @xmath81 .
if the random variables @xmath79 are characterized by a distribution having asymptotic power - law behavior @xmath82 where @xmath83 , then @xmath80 will converge to a lvy stable stochastic process of index @xmath7 in the limit @xmath84 .
except for special cases , such as the cauchy distribution , lvy stable distributions can not be expressed in closed form .
they are often expressed in terms of their fourier transforms or characteristic functions , which we denote @xmath85 , where @xmath86 denotes the fourier transformed variable .
the general form of a characteristic function of a lvy stable distribution is @xmath87 & $ [ \alpha\neq 1]$ \cr i\mu
q-\gamma|q|\left[1+i\beta{q\over |q|}{2\over\pi}\ln|q|\right ] & $ [ \alpha=1]$},\ ] ] where @xmath88 , @xmath89 is a positive number , @xmath90 is the mean , and @xmath77 is an asymmetry parameter . for symmetric lvy distributions ( @xmath91 ) ,
one has the functional form @xmath92 for @xmath93 , one obtains the cauchy distribution and for the limiting case @xmath94 , one obtains the gaussian distribution . by construction ,
lvy distributions are stable , that is , the sum of two independent random variables @xmath95 and @xmath96 , characterized by the same lvy distribution of index @xmath7 , is itself characterized by a lvy distribution of the same index .
the functional form of the distribution is maintained , if we sum up independent , identically distributed lvy stable random variables . for lvy distributions ,
the asymptotic behavior of @xmath97 for @xmath98 is a power - law , @xmath99 hence , the second moment diverges .
specifically , @xmath100 diverges for @xmath101 when @xmath102 .
in particular , all lvy stable processes with @xmath102 have _ infinite _ variance . thus , non - gaussian stable stochastic processes do not have a characteristic scale .
although well - defined mathematically , these distributions are difficult to use and raise fundamental problems when applied to real systems where the second moment is often related to the properties of the system . in finance , an infinite variance would make risk estimation and derivative pricing impossible .
a common problem when studying a distribution that decays as a power law is how to obtain an accurate estimate of the exponent characterizing the asymptotic behavior . here
, we review the methods of hill@xcite .
the basic idea is to calculate the _ inverse _ of the local logarithmic slope @xmath103 of the cumulative distribution @xmath104 , @xmath105 we then estimate the inverse asymptotic slope @xmath106 by extrapolating @xmath103 as @xmath107 . we start with the normalized returns @xmath39 and proceed in the following steps : * step ii : * the cumulative distribution
is then expressed in terms of the sorted returns as @xmath112 figure [ hill_explain ] is a schematic of the cumulative distribution thus obtained .
the _ inverse _ local slopes @xmath113 can be written as @xmath114 using eq .
( [ defp ] ) , the above expression can be well approximated for large @xmath115 as @xmath116 yielding estimates of the local inverse slopes . *
step iii : * we obtain the inverse local slopes through eq .
( [ def_gamm1 ] ) .
we can then compute an average of the inverse slopes over @xmath117 points , @xmath118 where the choice of the averaging window length @xmath117 varies depending on the number of events @xmath111 available .
* step iv : * we plot the locally averaged inverse slopes @xmath119 obtained in step iii as a function of the inverse normalized returns @xmath120 [ see , e.g. , fig . [ sp500_hill ] ]
. we can then define two methods of estimating @xmath7 . in the first method
, we extrapolate @xmath103 as a function of @xmath120 to @xmath121 , similarly to the method of successive slopes@xcite ; this procedure yields the inverse asymptotic slope @xmath106 . in the second method ,
we average over all events for @xmath120 smaller than a given threshold @xcite , with the average yielding the inverse slope @xmath106 . to test the hill estimator
, we analyze two surrogate data sets with known asymptotic behavior : ( a ) an independent random variable with @xmath122 , and ( b ) an independent random variable with @xmath123 . as shown in figs .
[ hill_explain](b , c ) , the method yields the correct results @xmath124 and @xmath125 , respectively .
d. chowdhury and d. stauffer , eur .
j. b * 8 * , 477 ( 1999 ) ; i. chang and d. stauffer , physica a * 264 * , 1 ( 1999 ) ; d. stauffer and t. j. p. penna , physica a * 256 * , 284 ( 1998 ) ; d. stauffer , p. m. c. de oliveria and a. t. bernardes , int . j. theor .
finance * 2 * , 83 ( 1999 ) .
some previous studies were conducted with the return defined as the difference in the price and not the difference in the logarithm of the price .
however , for events that are not too large , the two quantities have the same statistics .
the s&p 500 index is the sum of the market capitalizations ( share price multiplied with the number of outstanding shares ) of 500 companies representing the major industries .
note that these are not necessarily the largest , and that the composition of the index changes slowly in time .
t. bollerslev , r. y. chou , and k. f. kroner , j. econometrics * 52 * , 5 ( 1992 ) ; g. w. schwert , j. finance * 44 * , 1115 ( 1989 ) ; a. r. gallant , p. e. rossi and g. tauchen , review of financial studies * 5 * , 199 ( 1992 ) ; b. le baron , j. business * 65 * , 199 ( 1992 ) ; k. chan , k. c. chan and g. a. karolyi , review of financial studies * 4 * , 657 ( 1991 ) . the discreteness in the returns is due to the fact that stocks are priced in units of fractions of usd , usually 1/8 , 1/16 , or 1/32 .
the discreteness is not so evident in the s&p 500 index as it is the sum of 500 stocks , each with different weights . .
the values of the exponent @xmath7 , for different time scales @xmath0 , for the s&p 500 index : ( a ) power - law regression fit to the cumulative distribution , and ( b ) hill estimator .
the daggered values are computed using database ( ii ) , which contains daily - sampled records , while the values without the dagger are computed using database ( i ) , which contains records with a 1@xmath3min sampling .
note that we use the conversion 1 day=390 min .
[ cols="^ , < , < , < , < " , ] | we study the distribution of fluctuations over a time scale @xmath0 ( i.e. , the returns ) of the s&p 500 index by analyzing three distinct databases . database
( i ) contains approximately 1 million records sampled at 1 min intervals for the 13-year period 19841996 , database ( ii ) contains 8686 daily records for the 35-year period 19621996 , and database ( iii ) contains 852 monthly records for the 71-year period 19261996 .
we compute the probability distributions of returns over a time scale @xmath0 , where @xmath1 varies approximately over a factor of @xmath2from 1@xmath3min up to more than 1@xmath3month .
we find that the distributions for @xmath4 4 days ( 1560 mins ) are consistent with a power - law asymptotic behavior , characterized by an exponent @xmath5 , well outside the stable lvy regime @xmath6 . to test the robustness of the s&p result
, we perform a parallel analysis on two other financial market indices .
database ( iv ) contains 3560 daily records of the nikkei index for the 14-year period 1984 - 97 , and database ( v ) contains 4649 daily records of the hang - seng index for the 18-year period 1980 - 97 .
we find estimates of @xmath7 consistent with those describing the distribution of s&p 500 daily - returns .
one possible reason for the scaling of these distributions is the long persistence of the autocorrelation function of the volatility . for time scales
longer than @xmath8 days , our results are consistent with slow convergence to gaussian behavior . 2 |
the determination of the shape of a baryon is an interesting problem that addresses complex technical issues , both from the experimental and theoretical point of view @xcite . recently ,
for example , there has been particular interest ( and controversy ) about possible deviations from a spherically symmetric shape of the nucleon . in this work
we will address another interesting question , namely whether the @xmath0 baryon has a spherical shape or not , and maybe more importantly , what we mean by that .
information on a baryon s shape is encoded in its electromagnetic form factors , which can be measured , at least in principle .
the electromagnetic form factors describe how a particle interacts with a photon , and the number of independent form factors depends on the particle s spin .
a particle with spin-1/2 , such as the nucleon , is characterized by only two form factors , namely the electric charge @xmath3 and magnetic dipole @xmath4 , whereas a particle with spin-3/2 , such as the @xmath0 , has altogether four form factors , namely , in addition to @xmath3 and @xmath4 , also electric quadrupole @xmath5 and magnetic octupole @xmath6 form factors .
higher - order form factors and moments can provide a measure of the deformation of an extended particle .
for instance , nonvanishing values of the electric quadrupole moment and of the magnetic octupole moment , which are proportional to @xmath7 and @xmath8 , respectively , indicate a deviation of the charge and magnetic dipole distributions from the spherically symmetric form @xcite . at this point
, it is important to distinguish between `` spectroscopic '' and `` intrinsic '' moments , the former being the observable values of the corresponding electromagnetic form factor at zero momentum transfer , whereas the latter refer to quantities calculated from charge or magnetic density distributions , which are not directly observable @xcite .
for instance , nucleons or spin-1/2 nuclei do not possess an electric quadrupole or magnetic octupole moment , and therefore they can not be deformed if finite higher spectroscopic moments are used as criteria for deformation
. on the other hand , their intrinsic density distribution may not be spherically symmetric , giving rise to nonvanishing intrinsic higher moments @xcite .
the situation is somewhat simpler in the case of the @xmath0 , the lightest baryon with spin-3/2 and first candidate for a nonvanishing electric quadrupole moment , because its intrinsic and spectroscopic quadrupole moments differ only by a constant factor @xcite .
thus , there is a strong motivation to estimate the @xmath0 electric quadrupole moment .
previous studies on the deformation of the @xmath0 can be found in refs .
@xcite .
it is only in the nonrelativistic limit that the spatial distributions of the charge and magnetic densities are related to the electromagnetic form factors through a fourier transform , such that information about the shape of the density distributions can be accessed through form factor data : in the breit frame , the charge distribution is the fourier transform of @xmath3 , and the magnetic dipole density is the fourier transform of @xmath4 @xcite .
however , this interpretation has its limitations . in a relativistic description ,
the spatial distribution of charge or magnetic densities depends on the reference frame .
when an absorbed photon imparts only a small momentum to the baryon , this frame dependence can be ignored , but in the general relativistic case there is no direct relation between form factors and coordinate - space densities .
it was to address this difficulty that alternative concepts to measure deformation were proposed , such as transverse densities and moments @xcite and spin - dependent density distributions @xcite .
transverse density distributions were introduced recently within the context of a light - front description of the electromagnetic current . in this formulation ,
the charge and magnetic density distribution of a particle is described as seen from a light - front moving towards the particle .
the longitudinal direction is effectively eliminated by projecting the density onto the transverse plane , which is not subject to lorentz contraction . for spin polarizations in the transverse plane , radially asymmetric transverse density distributions can be obtained , which may then be analyzed in terms of multipoles .
however , it is not clear how exactly they are connected to the intrinsic deformation of the particle s density distribution in its rest frame .
for instance , just as in a number of previous model calculations , a recent lattice qcd study found a negative value of the electric quadrupole moment of the @xmath2 baryon in the state with spin projection 3/2 , implying an oblate intrinsic deformation relative to the spin direction , whereas the corresponding transverse charge density field pattern shows a prolate shape @xcite .
it has been suggested that information about deformation should be deduced by comparing the transverse moments to the so - called `` natural '' moments of structureless , elementary particles . in this formulation ,
a particle is _ defined _ as elementary if its light - cone helicity is nontrivially conserved at tree level .
the argument given to support this assumption is that in a composite particle the elementary constituents can jump between orbital states and thereby change the particle s total helicity , whereas this is not possible if the particle has no constituents @xcite .
for such a structureless particle with spin 3/2 and charge @xmath9 ( specified in units of the proton charge ) , the natural moments labeled here with the superscript @xmath10are @xcite g^(nat)_e2(0)= - 3 e _
, g^(nat)_m3(0)= - e _ , [ eqgnatural ] besides g^(nat)_e0(0)= e _
, g^(nat)_m1(0)= 3 e_. [ eqgnatural2 ] in the example of the @xmath2 with spin projection 3/2 , a negative sign of @xmath11 is consistent with a prolate transverse charge density distribution if @xmath12 .
in general , the transverse electric quadrupole moment is only a function of the anomalous electromagnetic moments @xcite , and a nonzero value indicates a deviation from a pointlike structure when viewed from the light - front .
the question remains what the transverse moments can tell us about the intrinsic density deformation in the rest frame . in the following
we present the intrinsic three - dimensional rest - frame densities obtained in covariant model calculations of the @xmath0 baryon , and we will return to this question when we discuss our results . in this work we used relativistic quark - diquark model wave functions of the @xmath0 baryon that were constructed within the covariant spectator theory @xcite to calculate charge and magnetic density distributions , both in momentum and coordinate space .
thus we have direct information about their shape .
in particular , we compared two models , one which includes only @xmath13-waves and is therefore spherically symmetric , and another that includes @xmath14-wave components that induce a small deformation .
we determined then how this shape information manifests itself in the higher moments , as well as in the corresponding transverse moments .
we want to emphasize that our moments are not calculated at tree level and contain anomalous contributions . therefore they do not reduce to the natural values of eqs .
( [ eqgnatural ] ) and eqs .
( [ eqgnatural2 ] ) . with a spherically symmetric spatial wave function , where the quark and diquark are in a relative @xmath13-wave @xcite , we obtained g_e2(0)= 0 , g_m3(0)=0 .
[ eqsp01 ] this result does not depend on any specific model parameters , but holds in general as long as only @xmath13-waves are present .
once higher orbital angular momentum components are included in the quark - diquark wave function , nonvanishing values of @xmath15 and @xmath16 are generated @xcite .
this indicates that it is possible to relate deviations from spherical symmetry in the charge ( dipole moment ) distribution to the value of the electric quadrupole ( magnetic octupole ) moment , similar to what was found in the nonrelativistic limit .
the calculated values of the transverse quadrupole and octupole moments on the other hand do not seem to give any clear indication on the deformation of the density distributions .
this paper is organized as follows : the covariant spectator quark model is introduced in sec .
[ secspecquarkmodel ] . in sec .
[ secffvsha ] , we relate the electromagnetic form factors with the experimentally accessible polarized helicity amplitudes . in sec .
[ sectrans ] we discuss the deformation of the @xmath0 as determined from transverse densities , and in sec . [ secclassic ] we present the results obtained from the usual three - dimensional densities . in sec .
[ secconclusions ] we draw our conclusions .
we apply a quark model obtained in the covariant spectator formalism @xcite , and parameterized to describe the @xmath0 baryon , as discussed in detail in refs .
the @xmath0 wave function is a mixture of an @xmath13 state ( @xmath17 ) and two @xmath14 states ( @xmath18 , coupled to core spin @xmath19 and @xmath20 ) for the quark - diquark system @xcite , of the general form _ =
[ eqpsidel ] in this equation , @xmath21 is the admixture coefficient of the @xmath22 state ( quark core with spin-3/2 ) and @xmath23 the admixture coefficient of the @xmath24 state ( quark core with spin-1/2 ) . the momentum and spin indices are suppressed for simplicity .
the @xmath13- and @xmath14-state wave function components are written in terms of spin , orbital angular momentum , and isospin operators , multiplied by scalar functions , @xmath25 , @xmath26 , and @xmath27 . in our covariant spectator model ,
the diquark four - momentum is on - mass - shell , and therefore these scalar functions depend only on the square of the quark four - momentum , @xmath28 , where @xmath29 and @xmath30 denote the @xmath0 and the diquark four - momentum , respectively . assuming each of the @xmath0 wave function components in ( [ eqpsidel ] ) to be normalized to 1 , the overall normalization constant becomes @xmath31 .
this specific form of the wave function was introduced in refs .
@xcite and two different parameterizations were studied in refs . @xcite . here
we will use the model of ref .
@xcite , because it gives a more consistent description of the valence quark contribution of the @xmath0 to the @xmath1 reaction , both in the physical region and in the regimes accessible in lattice qcd calculations with heavy pions . in this model ,
the two @xmath14-state probabilities are both about 0.89% ( @xmath32 and @xmath33 ) . for more details on the model
we refer to refs .
@xcite .
the internal structure of the constituent quarks is described in terms of quark electromagnetic form factors , parametrized through a vector meson dominance mechanism and included in an effective quark current @xmath34 @xcite . employing the wave function ( [ eqpsidel ] ) and the quark current @xmath34 , in the covariant spectator formalism we write the electromagnetic current @xcite as j^&= & 3 _ _ s _ k _ ( p_+,k;s^ ) j_q^_(p_-,k;s ) [ eqcurrent1 ] + & = & -|u_(p_+,s^ ) \ { ^ . + & & . + } u_(p_-,s ) , where @xmath35 ( @xmath36 ) represents the final ( initial ) four - momentum , @xmath37 is the transferred momentum , @xmath38 , @xmath39 is the mass of the @xmath0 , and @xmath40 the diquark polarizations . for the covariant integration over
the on - mass - shell diquark momentum @xmath30 we use the abbreviation _
k , with @xmath41 , where @xmath42 is a model parameter that corresponds to a mean value of the spectator diquark mass @xcite .
the asymptotic states @xmath43 are the rarita - schwinger vector states @xcite . throughout this paper
we follow the convention used in our previous work that the diquark polarization indices , @xmath44 , on the wave functions are suppressed .
the multipole @xmath0 form factors can be written as linear combinations of @xmath45 , @xmath46 @xcite . for @xmath47 , to first order in the admixture coefficients @xmath21 and @xmath23 , one finds @xcite g_e0(0 ) & = & n^2 e _
+ g_m1(0 ) & = & n^2 ( e_+ _ ) + g_e2(0 ) & = & 3 ( an^2 ) e__d3^ + g_m3(0 ) & = & ( e_+ _ ) n^2 , [ eqdff ] where e_= ( 1 + |t_3 ) , & & _ = ( _ + + _ - _ + = 2_u-_d , & & _ -=23_u+13_d , [ eqdefk ] with @xmath48 , and @xmath49 is the nucleon mass .
the factors @xmath50 and @xmath51 are defined in terms of the overlap integrals between the initial @xmath13-state and the final @xmath14-state , as _ d3^&= & _ 0 _ k b(k , q , p_+)_d3 ( p_+,k ) _
s(p_-,k ) + i_d1^&= & _ 0 _ k b(k , q , p_+ ) _ d1 ( p_+,k ) _
s(p_-,k ) , with @xmath52 .
the function @xmath53 , whose detailed form is given in ref .
@xcite , reduces to @xmath54 in the limit @xmath55 , where @xmath56 is the familiar spherical harmonic .
the model described above was applied in ref .
@xcite to calculate the @xmath0 electromagnetic form factors , and its results were compared successfully to the recent lattice qcd simulations of refs .
although the model is still incomplete because important degrees of freedom , such as meson ( pion in particular ) cloud effects , are not included , it agrees well with the lattice qcd data for @xmath3 and @xmath4 @xcite and is also consistent with the unquenched @xmath5 data @xcite
. this success can be due to an effective suppression of pion cloud effects in the @xmath57 reaction ) @xcite , the difference between the results with the pion cloud and the results when the pion cloud effect is removed differ by only about 5% @xcite .
] , in contrast to the @xmath1 transition where the opening of the @xmath58 channel is crucial @xcite .
it is also possible that effective pion cloud effects are already included adequately through the vector meson dominance mechanism which models the effective quark current @xmath34 .
the electromagnetic form factors of a baryon are invariant functions of @xmath59 .
they are independent of the reference frame and of the initial or final polarization of the baryon .
note that these functions are not directly measured in an experiment .
what can be measured are cross sections in a particular frame , from which helicity transition amplitudes between two different or equal polarization states of the baryon can be deduced . in a process like @xmath57
, there are only 3 independent components of the current , as a consequence of current conservation .
these components can be chosen to be @xmath60 and @xmath61 , or , alternatively , @xmath62 and @xmath63 , with @xmath64 .
note that @xmath65 is associated with the photon polarizations @xmath66 that involve a change of the baryon polarization ( @xmath67 ) , and @xmath68 with @xmath69 where the baryon polarization is conserved .
the transition amplitude for spin projections @xmath70 and @xmath71 is j^(s^,s ) = - |u_(p_+,s^ ) u_(p_-,s ) , where the operator @xmath72 is implicitly defined through eq .
( [ eqcurrent1 ] ) , and @xmath73 are the photon polarization vectors , with @xmath74 and @xmath75 .
we will work in the breit frame , where the photon four - momentum is @xmath76 , with @xmath77 , the photon three - momentum @xmath78 points along the positive @xmath79-direction , and the initial and final total momenta are @xmath80 . the spin nonflip components of the current ( @xmath81 ) are @xcite for @xmath82 .
here we use @xmath83 . ]
j^0(s , s ) = g_e0(q^2 ) - f_s(s ) g_e2(q^2 ) , [ eqj0bf ] for @xmath84 , with @xmath85 and @xmath86 .
one can combine the two independent amplitudes as a symmetric combination of matrix elements j^0_s = = g_e0(q^2 ) , and an asymmetric combination j^0_a = = - g_e2(q^2 ) . in the limit @xmath55 , the first equation yields @xmath87 , whereas @xmath11 can not be obtained directly from the amplitudes at @xmath47 because @xmath88 goes to zero .
similarly , the magnetic form factors are obtained from the spin - flip current matrix elements @xmath89 .
again , there are only two independent amplitudes related with @xmath4 and @xmath6 for @xmath90 ( see refs .
@xcite for details ) . in a nonrelativistic formalism , the baryon s shape , and in particular any possible deformation deviation from a spherically symmetric form
would depend on the spin projection along the @xmath79-axis .
one can then define an electric charge distribution , @xmath91 , associated with each spin projection ( @xmath92 or @xmath93 ) and define an intrinsic electric quadrupole momentum @xcite as q_(s)= d^3 * r * _ e(*r*,s ) * r*^2 y_20 ( ) .
when @xmath94 , its sign indicates whether the system is oblate ( @xmath95 ) or prolate ( @xmath96 ) if @xmath97 ( the opposite shape if @xmath98 ) .
note in particular that the shapes for @xmath99 and for @xmath100 can be different ( see for instance ref .
@xcite ) . whenever the quadrupole magnetic moment is referred to without explicitly mentioning the polarization state , the maximum projection
is assumed @xcite .
the interpretation of electromagnetic form factors as fourier transforms of charge and magnetic distribution densities is valid only in the nonrelativistic limit .
if the absorbed photon imparts a significant momentum transfer to the struck system , the boost of the final state wave function relative to the one of the initial state can no longer be neglected , which spoils this simple interpretation . in order to still be able to extract information about distribution densities from the measured form factors , the concept of a transverse density distribution
was introduced @xcite . by going to an infinite momentum frame
, the dependence on the longitudinal component of the momentum is eliminated , and the deformation is defined in terms of densities in the space of the two transverse impact parameters , @xmath101 and @xmath102 . taking the transverse spin projection ( @xmath103 ) oriented in the @xmath104 direction , the transverse electric quadrupole moment @xmath105 is @xcite _ ^(s_=+ ) & = & \ { 2 + .
[ eqqt ] for @xmath106 one obtains @xmath107 . in the previous equation ,
@xmath108 $ ] is interpreted as the electric quadrupole moment induced by the magnetic moment in the light - front frame , and @xmath109 $ ] as the polarization effect due to the internal structure , also seen in that frame @xcite . using the natural values for @xmath11 and @xmath110 , eqs .
( [ eqgnatural])([eqgnatural2 ] ) , from eq .
( [ eqqt ] ) we conclude that @xmath111 for a structureless particle with spin 3/2 seen from the light - front . .transverse electric quadrupole moment @xmath112 in units of @xmath113 , and transverse magnetic octupole moment @xmath114 in units of @xmath115 , for the @xmath2
. [ cols="<,^,^,^,^",options="header " , ] similarly , the transverse magnetic octupole moment becomes @xcite _ ^(s_=
+ ) & = & \ { - g_m1(0 ) -g_e2(0 ) + . + & & .
g_m3(0 ) + e _ } ( ) . [ eqot ] for @xmath116
, one has @xmath117 .
a spin-3/2 particle without structure in the light - front has a vanishing transverse magnetic octupole moment . the authors of refs .
@xcite suggest that the electric quadrupole and magnetic octupole moments of the @xmath0 should be compared with their _ natural _ values , given in eqs .
( [ eqgnatural])([eqgnatural2 ] ) , and that deformation should be defined in terms of a positive or negative deviation from those reference values . equations ( [ eqqt])([eqot ] ) depend indeed on these differences , but what they describe is not the full deformation in three - dimensional coordinate space . instead ,
because the transverse density is defined in the two - dimensional @xmath118 plane , they measure an asymmetry of the density between the spin direction ( along the @xmath104-axis ) and the perpendicular direction ( along the @xmath119-axis ) in the @xmath120-plane . to see how much information about the deformation of the @xmath0
is contained in these higher transverse moments , we have calculated them for two covariant spectator quark - diquark models with significantly different shapes .
the first , model ii of ref .
@xcite , includes only @xmath13-states in the quark - diquark wave function , which is therefore spherically symmetric .
we call it here model `` spectator - s '' . the second model , presented in ref .
@xcite , is deformed , because apart from @xmath13-states it includes also @xmath14-states .
we refer to it here as model `` spectator - sd '' .
the spherical model spectator - s yields @xcite for the @xmath2 g_e0(0)=1 , g_m1(0)=3.29 , and the quadrupole and octupole moments vanish . for model spectator - sd one obtains @xcite & & g_e0(0 ) 1 , g_m1(0)= 3.27 + & & g_e2(0 ) = -1.70 , g_m3(0)= -1.72 .
[ eqvalues ] from these values we can calculate the transverse electric quadrupole and the magnetic octupole moments . the results are presented in table [ tableqdel ] , together with lattice qcd data obtained by the mit - nicosia group @xcite with three different methods , for pion masses in the range @xmath121 mev .
the positive sign of @xmath112 for all lattice calculations suggests a transverse distribution elongated in the spin direction .
the same transverse deformation is produced by model spectator - sd .
however , the transverse quadrupole moment for the pure @xmath13-wave model spectator - s is not zero as one might expect .
instead , it also predicts a deformation in the spin direction although not so strong as in the previous case .
thus , whereas zero or nonzero values of electric quadrupole and magnetic octupole moments distinguish clearly between spherical and deformed @xmath0 states , the corresponding transverse moments do not provide the same information .
this is a quantitative illustration that the transverse moments are not an unambiguous measure of deformation . as for the transverse octupole moment
, our result , @xmath122 , suggests a deformation perpendicular to the spin axis .
this is true for both spectator quark models , with and without @xmath14-states .
the numerical values are very close , which means that @xmath123 does not discriminate much between models with spherical or deformed wave functions . at the moment ,
no lattice calculations of @xmath123 are available .
but we can use the form factor data of the adelaide group @xcite at @xmath124 gev@xmath125 in eq .
( [ eqot ] ) , replacing @xmath9 by @xmath126 , for a rough estimate .
we obtain @xmath127 , which is consistent with our result although with large statistical uncertainty and also suggests a deformation perpendicular to the spin direction . for a more rigorous comparison with our predictions we have to wait for future lattice calculations .
in this section , we calculate the @xmath0 charge densities from wave functions obtained in a covariant spectator quark model , and we illustrate to what extent the distortion caused by the @xmath14-wave contributions manifests itself both in the momentum - space and coordinate - space densities . there is no need to make use of electromagnetic form factors to characterize deformation in this case , because the densities are calculated directly from the wave functions .
thereby we sidestep the usual problems in relating densities and form factors , namely that the latter involve wave functions in different reference frames .
the former is the fourier transform of the latter only in the nonrelativistic limit , whereas we are interested in the general , relativistic case .
because the covariant spectator theory is more naturally formulated in momentum space , in the following we start our discussion of @xmath0 charge densities in the momentum - space representation where the wave functions were developed @xcite .
then we perform a fourier transform of the wave function and discuss the densities in coordinate space . the components of the @xmath0 wave function of eq .
( [ eqpsidel ] ) can be written @xcite as & & _ s(p , k;s)= -_s(p , k ) ( _ p^)^ u_(p , s ) , [ eqpsis ] + & & _ d3(p , k;s)= _ d3(p , k ) _ d3(p , k;s ) , + & & _ d1(p , k;s)= _ d1(p , k ) _
d1(p , k;s ) , [ eqpsid1 ] where the isospin state , which is a common factor in all wave functions , is omitted , and we use the notation _
d(2s)(p , k;s)= - 3 ( _ p^)^ ( p_s ) _
d^ u_(p , s ) , [ eqpsidx ] where @xmath128 is the core spin and @xmath29 the four - momentum of the @xmath0 . in eqs .
( [ eqpsis ] ) and ( [ eqpsidx ] ) , @xmath129 is the diquark polarization vector in the fixed - axis representation @xcite . in the last equation ,
@xmath130 is a projector onto the state @xmath131 or @xmath132 , and @xmath133 is the spectator @xmath14-state operator .
more details can be found in ref .
@xcite . the scalar functions @xmath25 , @xmath26 , and @xmath27 regulate the momentum distribution of the quark - diquark system .
because both the @xmath0 baryon and its diquark constituent are on - mass - shell , these functions depend only on .
it is convenient to express them in terms of the dimensionless variable = , where @xmath42 is the diquark mass .
following ref .
@xcite , we use the parametrizations & & _ s(p , k)= + & & _ d3(p , k)= [ eqpsird3 ] + & & _ d1(p , k)= \ { - } .
[ eqpsird1 ] the momentum range parameters @xmath134 ( @xmath46 ) , given in units of @xmath42 , determine the long- and short - range dependence of the wave functions in coordinate space .
the normalization constants @xmath135 , @xmath136 , and @xmath137 are determined by the conditions @xmath138 , @xmath139 , and @xmath140 , where @xmath141 is the total four - momentum of the @xmath0 baryon in its rest frame .
the coefficient @xmath142 is determined through the orthogonality between the @xmath0 and nucleon states @xcite . in this work ,
we use the model of ref .
@xcite , with the parameters @xmath143 , @xmath144 , @xmath145 , and @xmath146 .
one obtains @xmath147 , and the @xmath14-state admixture coefficients are @xmath32 and @xmath33 .
similarly to the case of the nucleon in ref .
@xcite , the momentum - space charge density of the @xmath0 in its rest frame is defined as _ ( * k*,s)= _ _ s _ ^(|p , k;s ) j_q _ ( |p , k;s ) , [ eqrho ] where @xmath148 is the spin projection of the @xmath0 state , and @xmath149 the charge operator @xcite .
remember that implicitly the wave function @xmath150 depends also on the diquark polarization @xmath44 . substituting ( [ eqpsidel ] ) into ( [ eqrho ] ) one gets & & _ ( * k*,s)= n^2 _ ,
s(*k*,s ) + & & + a^2 n^2 _ , d3(*k*,s)+ b^2 n^2 _ ,
d1(*k*,s ) + & & + 2 an^2 _ , sd3 ( * k*,s ) , [ eqrho1 ] where _ , x(*k*,s)= _ _ s _ x^(|p , k ; s ) j_q _ x ( |p , k ; s ) , [ eqrhox ] for @xmath151 , and _ , s d3(*k*,s)= _ _ s _
d3^(|p , k ; s ) j_q _ s ( |p , k ; s ) .
[ eqrhosd3 ] the @xmath13 and @xmath22 states can either be in the initial or final state , hence the factor of 2 in the last term of eq .
( [ eqrho1 ] ) . after performing the spin and isospin algebra ,
it is helpful to isolate the spherically symmetric contribution @xmath152 from the angle - dependent terms in the total density , and one obtains & & _ ( * k*,s ) = _ ^s(*k * ) + & & + 2 e_a n^2 f_s(s ) * k*^2 y_20(z ) + & & - e_b^2 n^2 f_s(s )
* k*^4|_d1(|p ,
k)|^2 y_20(z ) , [ eqrhototal ] where @xmath153 , the function @xmath154 is proportional to the spherical harmonic @xmath155 , and [ eq : rhoks ] & & _
^s(*k*)= e_n^2 + & & . in the rest frame , @xmath156 depends only on @xmath157 .
the second term in ( [ eqrhototal ] ) is the overlap between the @xmath13 and the @xmath22 state . it does not vanish because the states have the same core spin-3/2
the @xmath24 state on the other hand is always orthogonal to @xmath13 and @xmath22 because its definition includes a spin-1/2 projector .
the last term comes from the overlap between the initial and final @xmath24 states .
the two last terms of eq .
( [ eqrhototal ] ) vanish when the angular integration is performed .
the radially symmetric part of the density can also be written as _
^s(*k*)= , [ eqrhos ] and the angle - dependent asymmetric component as _ ^a(*k*)&= & + & = & 2 a e_n^2 * k*^2 y_20(z ) + & & - b^2 e_n^2 * k*^4|_d1(|p , k)|^2 y_20(z ) .
[ eqrhoa ] using ( [ eq : rhoks ] ) and ( [ eqrhoa ] ) we can rewrite @xmath158 as _ ( * k*,s)= _ ^s(*k * ) + f_s(s ) _ ^a(*k * ) .
[ eqnewrhot ] from this equation and from @xmath159 it is clear that the deformation density for @xmath100 has always the opposite sign of the one for @xmath99 . for small @xmath14-state admixture coefficients @xmath21 and @xmath23 , the @xmath13 to @xmath22 transition term dominates the asymmetry .
this is consistent with a calculation of the form factors in first order in @xmath21 and @xmath23 where the @xmath22 state is responsible for a nonzero electric quadrupole form factor @xcite . the factors @xmath160 and @xmath161 in eq .
( [ eqrhototal ] ) are always greater than or equal to zero , and therefore the @xmath22- and @xmath24-state contributions to the deformation enter with opposite signs .
the overall factor multiplying @xmath162 can have either sign , depending on the specific parametrization of the wave functions and on the spin projection @xmath70 . to illustrate the deformation of the @xmath0 graphically , in fig .
[ figkall ] we show a polar representation of @xmath158 , for @xmath163 in the @xmath164-@xmath165 plane ( the densities are invariant under rotations about the @xmath165-axis ) .
the positive @xmath165 direction , corresponding to polar angle @xmath166 , points upwards . for fixed values of @xmath167 ,
the length of a straight line from the origin to a given point on a displayed curve is the respective density , and its angle with the upward direction is the polar angle @xmath168 . in this representation , a distribution @xmath169 with no @xmath168-dependence yields a perfect circle .
the three panels of fig .
[ figkall ] show @xmath158 for @xmath170 , @xmath171 , and @xmath172 gev , respectively .
the perfect circle ( solid line ) shows the symmetric distribution @xmath173 . in each case
, the dashed line represents @xmath174 , and the dotted line @xmath100 . as determined by eq .
( [ eqnewrhot ] ) the deviations from a spherically symmetric density for @xmath100 and @xmath99 are equal but with opposite signs . in the case of the @xmath175 density ,
the deformation is along @xmath165 for the smaller momenta k=0.2 and k=1.0 gev , but it is along @xmath164 for k=2.0 gev .
this change in the shape of the deformation with increasing momentum can be understood from the behavior of the @xmath176 and @xmath177 terms in the density of eq .
( [ eqrhototal ] ) .
figure [ figrho ] shows the k - dependence of the magnitudes of @xmath178 and @xmath179 , the factors that multiply @xmath180 with opposite signs , in comparison with the symmetric term @xmath173 . for small k ,
the @xmath176 term dominates strongly over the @xmath177 term , which is expected already because it is of first order in the small @xmath14-wave admixture parameter @xmath21 , whereas @xmath177 is quadratic in @xmath23 .
as k increases , the @xmath176 contribution falls faster than @xmath177 , due to the faster falloff of the @xmath13-state wave function .
around @xmath181 gev both contributions become equal in magnitude and cancel against each other . for larger values of
k @xmath177 dominates over @xmath176 , and changes the deformation from being along @xmath164 to being along @xmath165 .
however , the coefficients of @xmath182 at such high momenta are already very small ( note the logarithmic scale in fig .
[ figrho ] ) .
figure [ figrho ] shows also that , at high momenta , the spherically symmetric part of the density , @xmath173 , is itself dominated by its @xmath177 component .
comparing eqs .
( [ eqrhototal ] ) and ( [ eq : rhoks ] ) we see that the magnitude of the @xmath177 coefficient of @xmath162 in ( [ eqrhototal ] ) , represented by the dotted line in fig .
[ figrho ] , is the same as the @xmath177 contribution to the spherically symmetric part of the density in ( [ eq : rhoks ] ) .
with increasing momentum k , the dotted line seems to converge to the solid line of the total symmetric contribution .
however , it reaches only about 90% of @xmath173 , the small @xmath183 contribution being responsible for the remainder .
this ratio of the @xmath177 contribution to the total spherically symmetric part can be obtained using the asymptotic ratio @xmath184 .
figure [ figk3d ] shows contour plots of momentum space charge densities in the @xmath164@xmath165 plane .
the deformation of the total density @xmath185 , displayed in the upper panel , is barely visible in this plot , because the spherically symmetric contribution dominates strongly .
the lower panel shows only the much smaller asymmetric part @xmath186 , which enhances the density along the @xmath165-direction .
the origin of this enhancement is analyzed in fig .
[ figk3d2 ] , where the upper panel shows the @xmath177 , and the lower panel the @xmath176 contribution .
the @xmath177 channel density is deformed along @xmath164 but at the low momenta shown in the figure it is overwhelmed by the more than one order of magnitude larger @xmath176 density deformation along @xmath165 . in summary , at low and intermediate momenta the momentum space density distribution shows a small deformation in the @xmath165-direction for the @xmath0 state with spin projection @xmath174 , and a small deformation in the @xmath165-direction for the @xmath100 state . for high momenta the shape of the deformation is reversed , but this is hardly relevant because the density is already negligibly small in this region .
we proceed now to the calculation of charge densities of the @xmath0 in coordinate space .
all densities will be determined in the rest frame of the @xmath0 , where the total four - momentum is @xmath187 .
using the modified momentum - space wave function _ ( * k * ; s ) , [ eqtilpsi ] where the relativistic phase - space factor is absorbed into the definition of the wave function , we can write the fourier transform in the same form as for nonrelativistic wave functions : _ ( * r * ; s ) = e^i * k * _ ( * k * ; s ) .
[ eqft ] with this convention for the factors of @xmath188 , the inverse transform is _
( * k * ; s ) = e^ -i * k * _ ( * r * ; s ) . in complete analogy to the momentum - space expression ( [ eqrho ] ) , the charge density in coordinate space is given by _
( * r*;s)= _ _ s _ ^(*r*,s ) j_q _ ( * r*,s ) . as before , one can decompose @xmath189 into angular momentum components @xmath190 . in the fixed - axis polarization state basis
we are using , there is no @xmath30-dependence in the diquark polarization vector @xmath191 .
this makes the fourier transform of the @xmath13-state ( [ eqpsis ] ) in the rest frame particularly simple , _ s(*r*;s)= - r_s ( ) _ ^u^(|p , s ) , [ eqtilpsisr ] where we have introduced the shorthand @xmath192 to represent @xmath191 , and r_s ( ) = e^ i * k * , [ eqrs0 ] with @xmath193 .
the factor @xmath192 depends on the diquark polarization @xmath44 , which is not shown explicitly .
the isospin states are not affected by the transformation and are also suppressed for simplicity .
there is no angle - dependence in the rest frame wave function @xmath194 , therefore @xmath195 is also an @xmath13-wave , depending only on @xmath196 .
boosting the wave function to another frame would induce an angle - dependence into @xmath194 and consequently also into @xmath195 . however , this kind of relativistic deformation due to lorentz contraction of the system along the direction of motion is a separate issue .
we are interested in intrinsic deformations of the @xmath0 , which are already present in the rest frame . using the familiar expansion of a plane wave into partial waves , e^i * k * = 4_l=0^+ _ m =- l^l i^l y_l m ( ) y_l m^ ( ) j_l ( ) , [ eq : planewave ] where @xmath197 are the spherical bessel functions , eq .
( [ eqrs0 ] ) becomes r_s ( ) = 4 j_0 ( ) .
[ eq : rs ] the general structure of the complete @xmath14-state wave functions ( without isospin ) can be written as _
d(*k*;s ) = _ d(k , s ) , where @xmath14 stands for @xmath24 or @xmath22 .
@xmath198 is the spin wave function of the @xmath14-states . in the rest frame
, it can be represented as @xcite _ d1 ( k , s ) & = & + ^2 _ ^ + & & _
m_l s_1 2 m_l ; s_1| s y_2 m _
l ( ) u^(|p , s_1 ) + _ d3 ( k , s ) & = & -^2 _ ^ + & & _
m_l s_1 y_2 m _
l ( ) u^(|p , s_1 ) + where @xmath199 is the spin-1/2 state @xcite u^(p , s)= _ 5 ( ^- ) u(p , s ) .
the angle dependence is contained exclusively in the factors @xmath200 , and the spin states are completely independent of @xmath163 .
the @xmath14-state wave functions in coordinate space are then given by the fourier transform , which yields _
d1(*r*;s ) & = & + + & & _
m_l s_1 2 m_l ; s_1| s y_d1^m_l ( * r * ) + _ d3(*r*;s ) & = & - + & & _
m_l s_1 2 m_l ; s_1| s y_d3^m_l ( * r * ) , + & & [ eq : psid13 ] where _ d^m_l ( * r * ) = e^ i * k * * k*^2 y_2 m_l ( ) .
again , in the rest frame @xmath201 is also angle independent , and the fourier transform simplifies to _ d^m_l ( * r * ) = -r_d ( ) y_l m_l ( ) , [ eq : yd ] where r_d ( ) = 4 ^2 j_2 ( ) , [ eqr ] and the minus sign factored out in ( [ eq : yd ] ) comes from the @xmath202 in ( [ eq : planewave ] ) .
we calculated the functions @xmath195 , @xmath203 , and @xmath204 numerically , and the results are presented in fig .
[ figfourier1 ] .
the @xmath13-state dominates at small distances , but the two @xmath14-state wave functions become comparable in size around @xmath205 fm and dominate for larger values of @xmath206 .
the @xmath14-waves start out with opposite signs , but the @xmath24 wave changes sign at @xmath207 fm .
as r goes to zero , the relativistic @xmath14-state wave functions are weakly singular , namely @xmath208 .
this behavior , which is reminiscent of the singular radial dependence of the dirac wave functions of the hydrogen atom @xcite , does not cause any problems because the densities remain integrable .
the origin of these singularities is the slower falloff with increasing relative momentum of the relativistic momentum - space wave functions compared to nonrelativistic wave functions . if we calculate the fourier transform of the @xmath14-state wave functions in the nonrelativistic limit , at small r we obtain the regular behavior @xmath209 .
the total coordinate - space charge density is & & _ ( * r*,s ) = n^2 _ , s(*r*,s ) + & & + a^2 n^2 _ , d3(*r*,s)+ b^2 n^2 _ , d1(*r*,s ) + & & + 2 an^2 _ , sd3 ( * k*,s ) .
[ eqrho2 ] the various components are defined in analogy with ( [ eqrhox ] ) and ( [ eqrhosd3 ] ) .
the density associated with the @xmath13-state is _ , s(*r*,s)= e _ r_s^2 . for the other cases , one uses & & _ _ s _ p^(_s ) _ p^ ( _ s)= - g^ + , + & & |u_(p , s ) u^(p , s ) = |u_(p , s)u^(p , s)= -1 , + & & |u_(p , s ) u^(p , s)= |u^(p , s ) u_(p , s ) = 0 , assuming the same polarization in the initial and final states .
the results for @xmath22 , @xmath24 , and the transition @xmath13 to @xmath22 are & & _ , d3 ( * r * , + ) = e _
r_d3 ^ 2 , + & & _ , d1 ( * r * , + ) = e_r_d1 ^ 2 , + & & _ , s d3 ( * r * , + ) = - e_r_d3 r_s y_20(z ) , for @xmath99 , and & & _ , d3 ( * r * , + ) = e _
r_d3 ^ 2 , + & & _ , d1 ( * r * , + ) = e_r_d1 ^ 2 , + & & _ , s d3 ( * r * , + ) = e_r_d3 r_s y_20 ( z ) , for @xmath100 .
the function @xmath210 , with @xmath153 , was defined previously , but the angle @xmath168 is now to be understood as the angle between @xmath211 and the @xmath79-axis . more details are given in appendix [ aprho ] .
the total density becomes _
( * r*,s ) = _ ^s(*r * ) + f_s(s ) _ ^a(*r * ) , where _ ^s(*r * ) & = & n^2 e _ is again the angle- and spin - projection - independent contribution , and _ ^a(*r*)&= & - 2 e_a n^2 r_s r_d3 y_2 0 ( z ) + & & - e_b^2 n^2 r_d1 ^ 2 y_2 0 ( z ) , [ eqrho3 ] the angle - dependent asymmetric component .
equation ( [ eqrho3 ] ) is the coordinate - space analogue of eq .
( [ eqrhoa ] ) .
the different sign of the term containing the @xmath13 and @xmath22 wave functions is due to eq .
( [ eq : yd ] ) where @xmath212 is defined such that it does not contain the factor @xmath202 .
again there are two independent terms proportional to @xmath213 that cause deformation , one associated with the @xmath24 state , and another with a @xmath13 to @xmath22 transition .
the main difference to the analogous momentum - space expression is that both terms come now with a minus sign . because @xmath204 enters squared , the total effect of the two @xmath14-state contributions can be inferred from the sign of the wave functions @xmath195 and @xmath203 and the sign of @xmath21 .
when @xmath21 is positive , as in model spectator - sd , and @xmath195 , @xmath203 have the same sign , the two terms in eq .
( [ eqrho3 ] ) have also the same sign and reinforce each other s contribution to the deformation .
this is the case in the region up to @xmath214 fm shown in fig .
[ figfourier1 ] , where both @xmath195 and @xmath203 are positive .
we checked that both @xmath195 and @xmath203 stay positive in that region when the wave function parameters are varied within a broad range , such that the direction of the deformation remains a robust result . for @xmath99
this means that @xmath215 is multiplied by an overall negative factor , implying an oblate shape , as shown in fig .
[ figrall ] for r = 0.2 , 1.0 , and 2.0 fm . for @xmath216 the coefficient of @xmath215 has the opposite sign , and the deformation is prolate . at @xmath217
fm the deformation is too small to be visible in the graph , but it becomes more pronounced as r increases .
figure [ figrhol ] shows the magnitudes of the individual coefficients of @xmath215 , together with the radially symmetric @xmath218 .
the @xmath176 contribution dominates over @xmath177 at small r , because it is of first order in the small @xmath14-state admixture coefficient @xmath21 , whereas @xmath177 is of second order in @xmath23 .
however , with increasing r the @xmath13-wave falls more rapidly than the @xmath14-waves , and the @xmath24-term dominates for @xmath219 fm .
finally , fig .
[ figspace3d ] presents contour plots of the charge densities of the @xmath174 state in the @xmath104-@xmath79 plane .
the upper panel shows the total density , the lower panel only the much smaller asymmetric contribution which is responsible for the oblate shape .
our results for the shape of the @xmath2 are in agreement with those of previous studies , such as the results reported in ref .
@xcite using the cloudy bag model , and the constituent quark model calculations of ref .
the oblate shape for @xmath2 is also consistent with the negative sign of @xmath11 found in lattice qcd simulations @xcite .
the formalism presented here can be applied to other systems .
for instance , from an analysis of the form factors of the @xmath220 baryon , it can be concluded that the charge density distribution of the @xmath220 also has an oblate shape @xcite .
the @xmath0 is the lowest - mass baryon that can possess a nonvanishing electric quadrupole moment . in a nonrelativistic framework
, the electric quadrupole moment can be used as an indicator for a particle s deviation from a spherically symmetric shape . however , in the general , relativistic case the connection between shape and higher spectroscopic moments is more complicated , which led to the proposal of alternative methods to measure deformation .
one of these methods suggests to extract information about deformation from transverse densities , calculated in the transverse impact - parameter space @xmath118 in the infinite - momentum frame .
it has the advantage that the transverse density moments @xmath221 and @xmath123 are zero for a structureless particle in the transverse plane , and therefore can be used to measure the extension of the particle in impact - parameter space .
however , they do not allow the classification of the shape in the particle s rest frame . in this work we used a different relativistic formalism , the covariant spectator theory , to investigate the relation between moments of charge or magnetic density distributions and the intrinsic shape of these distributions in the @xmath0 baryon s rest frame .
we used two covariant quark - diquark momentum - space wave functions for the @xmath0 , one consisting of pure @xmath13-waves only , called `` spectator - s '' , and another which includes @xmath13- and @xmath14-waves , called `` spectator - sd '' .
the electric and magnetic moments and form factors can be calculated directly from these wave functions , as well as the momentum space densities .
the coordinate space densities are then obtained from the fourier - transformed wave functions .
we arrived at the following results : for the @xmath13-wave model spectator - s one obtains @xmath222 , and the electric and magnetic density distributions are spherically symmetric .
the @xmath14-wave admixture in model spectator - sd on the other hand produces a spatial deformation of the @xmath2 density distribution , and the quadrupole and octupole moments become @xmath223 and @xmath224 , respectively .
the negative value of the quadrupole moment corresponds to an oblate density distribution in coordinate space .
we conclude therefore that the higher moments are good indicators that allow to distinguish deformed from spherically symmetric systems . using the same wave functions and their respective electric and magnetic moments
, we also calculated the corresponding transverse density quadrupole and octupole moments ( @xmath221 and @xmath123 ) . for each moment , the obtained values ( see table [ tableqdel ] ) for models spectator - s and spectator - sd are nonvanishing and of the same sign , and thus do not show a clear distinction between spherically symmetric and deformed cases .
since the transverse moments are nonzero one can conclude that the system is not seen as a point in impact parameter space when viewed from the light - front frame , but it is not clear how further information on its shape could be extracted . in this sense ,
the usual three - dimensional density distribution complements the information contained in the transverse densities , and it is also closer to our intuitive notion of deformation . for the specific case of the @xmath2 baryon with spin projection @xmath175 ,
the covariant model spectator - sd predicts an oblate shape of its density distribution in coordinate space .
this is in agreement with previously obtained results , both from other quark model calculations and lattice qcd simulations .
all model parameters were determined through fits to the available lattice qcd data for the @xmath225 transition form factors at large pion mass , where the uncertain pion cloud effects are minimal ( the experimental data are also well predicted ) @xcite .
in particular , the coefficients @xmath21 and @xmath23 are determined by the lattice transition form factors for @xmath226 , namely @xmath21 is determined by the electric quadrupole form factor @xmath227 and @xmath23 by the coulomb quadrupole form factor @xmath228 .
further improvements in the statistical quality of the lattice data might alter the magnitudes of @xmath21 and @xmath23 and therefore the extent of the deformation we predict for the @xmath0 , but the signs are not in doubt . in this sense ,
the nature of the deformation , namely that @xmath2 is oblate rather than prolate , is a robust prediction of our model . g. r. thanks carl carlson for helpful discussions . this work was supported in part by the european union under the hadronphysics3 grant no .
283286 , and by the fundao para a cincia e a tecnologia ( fct ) , under grant no .
ptdc / fis/113940/2009 .
one of us ( g. r. ) was also supported by fct under grant no .
sfrh / bpd/26886/2006 .
we list here more explicit expressions of the coordinate - space @xmath0 wave functions @xmath229 and @xmath230 . performing the sums in eq .
( [ eq : psid13 ] ) we get and _ d3 ( * r * ; + ) & = & - _ d3 ^ 0 ( * r * ) _ ^ u^ ( |p,+ ) + & & + _ d3^+1 ( * r * ) _ ^ u^ ( |p,+ ) + & & - _ d3^+2 ( * r * ) _ ^ u^ ( |p,- ) , + & & + _ d3 ( * r * ; + ) & = & - _ d3 ^ -1 ( * r * ) _ ^ u^ ( |p,+ ) + & & + _
d3 ^ 0 ( * r * ) _
^ u^ ( |p , + ) + & & - _ d3^+2 ( * r * ) _ ^ u^ ( |p,- ) .
f. gross , g. ramalho and m. t. pea , phys . rev .
d * 85 * , 093005 ( 2012 ) [ arxiv:1201.6336 [ hep - ph ] ] ; g. ramalho and k. tsushima , phys .
d * 81 * , 074020 ( 2010 ) [ arxiv:1002.3386 [ hep - ph ] ] ; g. ramalho and m. t. pea , phys .
d * 84 * , 033007 ( 2011 ) [ arxiv:1105.2223 [ hep - ph ] ] ; g. ramalho , d. jido and k. tsushima , phys .
d * 85 * , 093014 ( 2012 ) [ arxiv:1202.2299 [ hep - ph ] ] .
g. ramalho , f. gross , m. t. pea and k. tsushima , in _ proceedings of the 4th workshop on exclusive reactions at high momentum transfer _ , edited by a. radyushkin ( world scientific , singapore , 2011 ) , p. 287 ; [ arxiv:1008.0371 [ hep - ph ] ] . | using a covariant spectator quark model that describes the recent lattice qcd data for the @xmath0 electromagnetic form factors and all available experimental data on @xmath1 transitions , we analyze the charge and magnetic dipole distributions of the @xmath0 baryon and discuss its shape .
we conclude that the quadrupole moment of the @xmath0 is a good indicator of the deformation and that the @xmath2 charge distribution has an oblate shape .
we also calculate transverse moments and find that they do not lead to unambiguous conclusions about the underlying shape . |
since the first detection of an exoplanet orbiting around a pulsar @xcite and around a main sequence star @xcite using radial velocity measurements , several other techniques , such as microlensing , coronagraphy , and polarimetric observations have been developed for the search of other worlds .
transit observations have contributed to a large fraction of the about @xmath14 confirmed planets .
several systematic surveys have been implemented for the detection of exoplanet candidates by transit observations .
important examples of these surveys include space - based missions such as corot @xcite and kepler @xcite satellites .
the high photometric precision of their observations allowed to boost the number of candidates and confirmed extrasolar planets detected in the last five years .
the next generation of space - based missions , such as the orbiting gaia @xcite and the upcoming jwst @xcite , plato @xcite , and tess @xcite are expected to increase the detection rate , and broaden the limits of the currently observed distribution of exoplanetary parameters , despite that some of these missions have not the detection of exoplanets as a main goal . on the other hand ,
ground - based photometric surveys such as superwasp @xcite and hat @xcite are focused on exoplanets with large radii and orbiting close to their host star , as these objects are particularly suited for these instruments . in order to improve the photometric precision of ground - based follow - up observations of exoplanetary transits ,
a technique consisting in slightly defocusing the image has been introduced and widely used in recent works @xcite .
the achieved results often compete with those of space - based surveys , and increase the opportunity for small telescopes to produce light curves with a remarkable photometric precision .
following the results obtained with the defocused photometry technique , we started an observational campaign of known exoplanetary transits , already detected by the superwasp and hat surveys .
photometric follow - up of these objects in different bandpasses is useful to study a possible dependence of the planetary radius as a function of the wavelength .
this project makes use of all three telescopes operating at the san pedro mrtir national astronomical observatory ( see sect .
[ obs ] for details ) , which we employ for the first time in the field of transit observations . in this paper
we focus on two particular targets : and , which are briefly presented in the following subsections .
llccccl diameter & focal ratio & ccd size & resolution & field of view & read out noise & gain + @xmath4 & @xmath15 & @xmath16 px & @xmath17px & @xmath18 & @xmath19 & 1.8 + @xmath7 & @xmath20 & @xmath21 px & @xmath22px & @xmath23 & @xmath24 & 4.0 + @xmath0 & @xmath25 & @xmath26 px & @xmath27px & @xmath28 & @xmath29 & 1.8 + wasp-39bis a saturn - mass planet recently discovered by the superwasp survey , and confirmed by @xcite with a photometric and spectroscopic follow - up observations .
the photometry was carried on with the pan - starrs-@xmath30 filter on the the @xmath31 faulkes telescope north , and in the gunn-@xmath6 filter on the @xmath32 euler telescope .
the spectroscopy was carried out with the sophie @xcite and the coralie spectrographs , installed on the @xmath33 telescope at the _ haute - provence _
observatory ( ohp ) and on the @xmath32 euler telescope , respectively .
@xcite obtain a period of @xmath34 days , a @xmath35 in heliocentric julian date ( hjd ) , a transit duration of @xmath36 minutes , and the following physical parameters : a mass of @xmath37 , a radius of @xmath38 , and thus a resulting mean density of @xmath39 .
this is one of the lowest values for the currently known extrasolar planets .
the reported semi - major axis is @xmath40 au , and evidences of a highly inflated radius are also found .
this work presents the first photometric follow - up of wasp-39bsince the discovery .
wasp-43bis a very short - period hot jupiter - like planet discovered by @xcite using observations from the wasp - south camera array , complemented with photometric observations with the @xmath41 trappist telescope in the @xmath42 band and the euler telescope in the gunn @xmath43 band .
radial velocity measurements were also obtained with coralie .
@xcite find a period of @xmath44 days , an initial epoch of @xmath45 hjd , and a duration of @xmath46 min .
the authors derive a mass of @xmath47 , a radius of @xmath48 leading to a density of @xmath49 .
the reported semi - major axis is @xmath50 au . in order to improve the characterization of wasp-43band its host star , for which @xcite find evidences of excess rotation , @xcite observed over thirty eclipses between transits and occultations , using mainly the trappist telescope in the @xmath42 and @xmath30 bands , respectively .
the survey was completed with transit observations obtained with the euler telescope in the gunn @xmath51 band and with the vlt / hawk - i instrument installed on the @xmath52 ut4 telescope in paranal .
the work allowed significant improvements in the determination of the stellar density and in the refinement of the parameters of the planetary system .
in particular , @xcite find a mass of @xmath53 and a radius of @xmath54 .
occultations also detected the thermal emission of the planet .
thermal emission is also the subject of spitzer observations , carried out by @xcite , which highlighted the possibility of thermal inversion in the atmosphere . @xcite
uniformly analyzed occultations of nine exoplanets including wasp-43b , with the goal to characterize their @xmath55 to @xmath56 ratio , which is found to be @xmath57 in the case of wasp-43b , and which is attributed to the upper limit value of @xmath58 .
other observations using the grond instrument on the mpg / eso @xmath59 telescope in the @xmath60 , @xmath51 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , and @xmath65 bands @xcite allowed to refine the period to a value of @xmath66 .
this value agrees with spectroscopic observations carried on by @xcite .
these observations also confirmed planetary dayside thermal emission with poor heat redistribution , as observed by @xcite .
cccc date & filter & exp . &
telescope + & + 2014 - 03 - 17 & @xmath1 & @xmath67@xmath68 & @xmath0 + 2014 - 03 - 17 & @xmath3 & @xmath69@xmath70 & @xmath4 + 2014 - 03 - 21 & @xmath2 & @xmath71 & @xmath4 + & + 2014 - 02 - 13 & @xmath2 & @xmath72@xmath71 & @xmath4 + 2014 - 03 - 07 & @xmath6 & @xmath73 & @xmath7 + 2014 - 03 - 07 & @xmath2 & @xmath74 & @xmath4 + 2014 - 03 - 20 & @xmath3 & @xmath72@xmath71 & @xmath4 + 2014 - 03 - 21 & @xmath75 & @xmath72@xmath71 & @xmath4 + 2014 - 03 - 29 & @xmath6 & @xmath76 & @xmath7 + 2014 - 05 - 12 &
@xmath3 & @xmath72@xmath71 & @xmath4 + finally , an x - ray study by @xcite on the host star of wasp-43busing xmm - newton observations found that the planet is exposed to a high - energy radiation field ; and water abundance measurements were recently reported by @xcite . in sect .
[ obs ] we describe the used telescopes and our observations of wasp-39band wasp-43b , while in sect .
[ data ] and [ sect : p ] we focus on data reduction techniques and on a discussion about the period of the two objects .
results are presented in sect .
[ disc ] , while conclusions are described in sect .
observations were carried on using all three equatorial ritchey - chrtien telescopes installed at the the san pedro mrtir national astronomical observatory ( spm - oan ) .
these facilities are located at an altitude of @xmath77 in the middle of the peninsula of baja california in northwestern mexico ( n , w ) .
the @xmath4 and the @xmath0 are operated on - site , and the @xmath7 is a robotic telescope monitored and operated remotely .
we principally used the @xmath4 telescope , an @xmath78 which provides an @xmath18 field of view on the _ mexman _ instrument .
we decided to window the _ esopo _ ccd to avoid vignetting and to reduce the readout time .
the instrument comes with a set of standard @xmath79 johnson filters among others . in this paper
we present observations carried out using the @xmath5 filters . and the @xmath0 telescopes , in two random ccd images , and using an inverse color scale .
the target and field stars a , b , and c are also shown .
star a was chosen as a reference for photometry for both the telescopes .
[ w39-field ] ] and the @xmath7 telescopes , in two random ccd images , and using an inverse color scale .
the defocusing is particularly appreciable in the @xmath7 telescope images .
the target field stars a , b , and c are also shown .
star a was chosen as a reference for photometry for both the telescopes .
[ w43-field ] ] observations were also obtained with the robotic @xmath7 telescope and its multi - channel imager _ ratir
this instrument , designed for gamma ray burst observations and follow - up , is equipped with three dichroics to transmit the light to four detectors named c0 , c1 , c2 , and c3 , equipped with optical and infrared filters . the ccds work in @xmath80 binning mode by default .
in this paper we present observations carried out using the gunn-@xmath6 only , corresponding to the c1 detector which provides a @xmath81 field of view . taking advantage from free observing time during maintenance operations , we had the possibility to use the @xmath0 telescope , improved with active optics on the primary mirror , which is normally dedicated to spectroscopy .
we used this telescope in direct - imaging mode .
the observation presented in this paper was carried out using the @xmath1 filter on the _ marconi _ ccd providing a @xmath82 field of view .
additional details about telescopes and detectors are listed in table [ tab : tel ] , while table [ tab : log ] shows the log of the observations .
telescope in @xmath2 and @xmath3 bands and with the @xmath0 telescope in the @xmath1 band . for displaying purposes ,
the light curves were reported to the time of the last observation ( 2014 - 03 - 21 ) , and an arbitrarily shift is applied .
the fit of the light curves ( solid lines ) is also shown , as well as the residuals in the lower box .
[ w39-curves ] ] telescope in @xmath5 bands and with the @xmath7 telescope in the gunn-@xmath6 band , reported to the time of the observation of 2014 - 03 - 29 , and arbitrarily shifted .
the fit and the residuals are also shown .
the dashed line represent an interpolation of the model . [ w43-curves ] ] we performed three observations of wasp-39b , covering two subsequent transits .
its parent star is a g - type dwarf with a reported magnitude of @xmath83 @xcite .
for the transit of 2014 - 03 - 17 we had simultaneous access to the @xmath4 and to the @xmath0 telescopes .
[ w39-field ] shows the target and its nearby stars as seen in the fields of the two telescopes , some of which ( a , b and c ) were tested as comparison stars , as seen by the fields of the two telescopes .
we observed in the @xmath3 band with the smaller telescope , obtaining 103 frames with exposure times of @xmath71 and @xmath70 , and in the @xmath1 band with the larger one , in order to take advantage from its sensitivity , obtaining 57 frames with exposure times of @xmath84 and @xmath68 .
as the object was particularly faint in this filter , we decided to use a @xmath85 binning and a small defocus .
the transit of 2014 - 03 - 21 was observed with the @xmath4 telescope only .
we obtained 142 frames in the @xmath2 band with an exposure time of @xmath71 , slightly defocusing the telescope to avoid saturation .
the night conditions during all observations suffered from very thin high haze and a nearly - full moon , with a seeing of about @xmath86@xmath87 .
depending on the instrument , the band , and seeing conditions , different defocusing levels were used . moreover , rapid changes in the atmospheric conditions required a variation of the exposure time in order to avoid saturation .
due to its short period with respect to wasp-39b , we had the opportunity to carry out more observations of wasp-43b .
we obtained a total of seven transits spanning six nights .
[ w43-field ] shows two fields containing the target and nearby stars as seen with the @xmath4 and @xmath7 telescopes .
four transits were observed with the @xmath4 telescope only : on 2014 - 02 - 13 , 2014 - 03 - 20 , 2014 - 03 - 21 , 2014 - 05 - 12 in the @xmath2 , @xmath3 , @xmath75 , and again @xmath3 filters , respectively , with exposure times spanning from @xmath74 to @xmath71 .
another transit was obtained with the @xmath7 telescope only ( 2014 - 03 - 29 ) , strongly defocused , in the gunn-@xmath6 filter , and with a @xmath76 exposure time .
this transit was unfortunately affected by clouds from the first contact until almost the mid time . finally , we observed with both the @xmath4 and @xmath7 telescopes on 2014 - 03 - 07 , using the @xmath2 and gunn-@xmath6 filters respectively , and with exposure times of @xmath74 and @xmath73 .
all observations were taken in condition of a nearly - full moon .
ccd images of both transits were debiased and flat - fielded using standard iraf procedures from the ccdproc package .
master bias and master sky flat fields were obtained from images taken at the beginning and at the end of each observing night .
cosmic rays were also removed using the method implemented by the lakos package @xcite , then images were aligned using the imalign iraf procedure .
we decided to use different aperture photometry routines to obtain the light curves : a custom iraf routine and the ` defot ` routine @xcite written in the idl language , that we modified to work with the fits headers generated by the spm - oan telescopes .
we found that the results of these two pipelines are in good agreement . in the case of images with a small amount of defocus , it was possible to calculate the full width half maximum ( fwhm ) of the point spread function by fitting an elliptical gaussian .
we used this information to dynamically adapt the aperture radius to a value of 2.5 times the calculated fwhm .
we found that this technique increases the quality of the photometry with respect to the fixed - radius aperture . in the case of images with a consistent defocus
, several fixed apertures were tested in order to find the optimal value .
several field stars were tested in order to find a reference star to perform differential photometry , and a non - saturated star with a color index and magnitude values as close as possible to those of the target star was chosen .
the chosen reference stars for wasp-39band wasp-43bare labeled as `` a '' in fig .
[ w39-field ] and fig .
[ w43-field ] , respectively .
we then used the target and the reference star to produce differential light curves .
the resulting light curves may present some residual trends , which were removed with a first - order airmass correction described by @xcite .
the timestamps of the light curve were converted to the dynamical time - based system ( @xmath88 , here therefore @xmath89 ) using the transformation given by @xcite .
we used the aperture photometry technique presented above for both wasp-39b , whose three final light curves are presented in fig . [ w39-curves ] , and wasp-43b , whose seven final light curves are shown in fig . [ w43-curves ] .
lccc parameter & filter & wasp-39 & wasp-43 + @xmath6 [ ] & & @xmath90 & @xmath91 + @xmath92 & & @xmath93 & @xmath94 + & @xmath1 & @xmath95 & + & @xmath75 & & @xmath96 + @xmath97 & @xmath2 & @xmath98 & @xmath99 + & @xmath3 & @xmath98 & @xmath100 + & @xmath6 & & @xmath12 + @xmath101 [ d ] & & 4.055259 & 0.81347437 + @xmath102 & & 0 & 0 + @xmath103 [ ] & & 0 & 0 + & @xmath1 & 0.950 & + & @xmath75 & & 0.750 + @xmath104 & @xmath2 & 0.425 & 0.599 + & @xmath3 & 0.335 & 0.451 + & @xmath6 & & 0.485 + & @xmath1 & -0.086 & + & @xmath75 & & 0.040 + @xmath105 & @xmath2 & 0.246 & 0.137 + & @xmath3 & 0.250 & 0.193 + & @xmath6 & & 0.183 + the light curves were fitted using the idl software transit analysis package ( tap ) implemented by @xcite , which uses a markov chain monte carlo ( mcmc ) method to find the best fit parameters for the @xcite model .
this code allows to fit multiple light curves simultaneously , which is particularly useful for fixing parameters such as the orbital inclination @xmath6 and the scaled semi - major axis @xmath92 to the same values for a group of curves , and thus obtain a global fit for these quantities .
other parameters , such as the mid - transit time @xmath106 and the scaled planetary radius @xmath97 are allowed to be fitted individually for each curve .
this technique allows to fit the same orbital model to a group of light curves of the same object obtained in different bands , and to find potential planetary radius differences as a function of the wavelength .
all light curves of each system were fitted simultaneously .
we used a set of fixed values in the mcmc analysis for several parameters : the period @xmath101 , taken from @xcite for wasp-39band from @xcite for wasp-43b ; the eccentricity @xmath102 and the argument of periastron @xmath103 , both set @xmath107 .
tap allows the use of linear and quadratic models for the stellar limb darkening .
a quadratic model is assumed in this analysis , and the two terms @xmath104 ( linear ) and @xmath105 ( quadratic ) are fixed in the mcmc analysis to the values corresponding to the filter used for each light curve .
the values were obtained from the exofast @xcite online tool , which are interpolated from stellar atmosphere models of @xcite .
we then fitted for each system : the orbital inclination @xmath6 , the scaled semi - major axis @xmath92 , and a value of the scaled radius @xmath97 for each one of the used filters .
tap was initialized with the most recent parameters reported in @xcite for wasp-39band @xcite for wasp-43b .
the mid - transit time @xmath106 for each light curve was also fitted ( in the case of multiple light curves of the same object obtained on the same night using different telescopes , the same @xmath106 was fitted for both the light curves ) . for the mcmc analysis ,
the @xmath106 is initialized with a value estimated by tap from the input light curve .
the best - fit models and corresponding residuals are shown in figs . [ w39-curves ] and [ w43-curves ] , and the values of the parameters are shown in table [ params ] .
using the fitted values for the mid - transit time @xmath106 , we retrieved the period @xmath101 of wasp-39band wasp-43bby fitting the following linear law : @xmath108 where @xmath109 is the initial epoch and @xmath110 is the number of periods since @xmath109 . @xmath109 and @xmath101 are best - fit parameters .
calculated @xmath106 and linear fit of the wasp-39bperiod , including data from @xcite , from the etd - tresca database , and from our observations .
[ w39-lin ] ] further analysis of the ephemerids was carried out in order to investigate a long - term variation of the period with respect to the time .
a constant decrease of the period for close - orbiting planets may be indicative of processes that remove orbital energy such as tidal dissipation @xcite . a simple model assuming a constant variation of the period is given by the following quadratic equation @xmath111 proposed by @xcite , where @xmath112 .
for the present analysis , a levenberg - marquardt least - squares fitting algorithm implemented in the code of @xcite was initially used to find the best - fit parameters of eqs .
[ eq : p ] and [ eq : dp ] .
however , this algorithm can be trapped in a local minimum of @xmath113 .
for this reason , a monte carlo code was set - up to search for the @xmath113 minimum and sample the parameter space .
both approaches were tested on the data provided by @xcite , finding good agreement with their results .
we find that both approaches are consistent and give similar best - fit values .
comforted by these results , we decided to apply this method to our data , and we describe the results in the following subsections .
calculated @xmath106 , linear and quadratic fit of the wasp-43bperiod including data from @xcite , and from our observations .
see sect .
[ w43-quad - discussion ] for curve details .
[ w43-quad ] ] concerning wasp-39b , the @xmath106 reported by @xcite and the one calculated from our observations are fitted using eq .
[ eq : p ] .
this gives a * @xmath114 ( bjd ) , * @xmath115 days .
this fit has a * @xmath116 , where @xmath113 stands for the `` reduced @xmath113 '' , i.e. @xmath117 , where @xmath118 are the degrees of freedom ( number of data @xmath119 number of parameters ) .
however , due to the significantly small number of observations , we decided to increase the number of transits considered in our analysis by introducing additional data from the exoplanet transit database ( etd ) webpage @xcite , in order to improve the robustness of our results .
we selected a total of three light curves . according to this database ,
the timings of the light curves are reported in hjd .
they were transformed to bjd using the software provided by @xcite .
we used tap to obtain only @xmath106 for these light curves by running the mcmc analysis with all other parameters to be fixed to the values provided by @xcite . by adding these new data to previous ones , we obtain a total of 7 observations
the linear fit of eq .
[ eq : p ] was repeated with these additional transits , and the following results were obtained : * @xmath120 ( bjd ) , * @xmath121 days , and a * @xmath122 , indicating a period larger than that reported by @xcite . the residuals for this fit are shown in fig . [ w39-lin ] .
as the last two tresca points show large differences with respect to the fit , we tested the fit process without their contribution , obtaining the following results : * @xmath123 ( bjd ) , * @xmath124 days , and a * @xmath125 , and in this case too we report a larger period with respect to @xcite results . in both cases the results
are consistent , but we suggest to improve their robustness with additional observations . an attempt to fit eq .
[ eq : dp ] was made , but the small amount of data do not allow a reasonable result for the fit .
more data are needed as well as precise timings , in order to assess if there is a long - term constant variation of the period , or to detect the presence of other transit timing variations .
a sufficient amount of wasp-43btransit timings is available in the literature spanning several years of observations @xcite .
some previous studies of possible long - term variations have been carried out by @xcite and @xcite , suggesting that the period of wasp-43bis slowly decreasing with time .
in this work , we repeat the analysis combining our data with the timings available in the literature .
a best - fit model was obtained using the marquardt - levenberg algorithm previously described . for the linear fit ,
best - fit values are * @xmath126 ( bjd ) , and * @xmath127 days , giving a * @xmath128 .
for the quadratic fit , best fit values are * @xmath129 ( bjd ) , * @xmath130 days , and a * @xmath131 , corresponding to * @xmath132 .
this gives a * @xmath133 .
the small difference in @xmath113 between the linear and quadratic fits shows that , statistically , there is no significant improvement by using the latter , as shown in fig .
[ w43-quad ] . using the values provided by @xcite
, we obtain * @xmath134 days , and * @xmath135 , giving a * @xmath136 .
the parameters reported by @xcite are * @xmath137 days , and we have * @xmath138 , giving a * @xmath139 .
the fit was also tested with the monte carlo approach which find a a minimum @xmath113 corresponding to @xmath140 and @xmath141 , with a @xmath142 .
this analysis showed that the @xmath113 minimum is very close to a value of @xmath143 .
these fits are visualized in fig .
[ w43-quad ] .
the model of @xcite fits most of the points reasonably well , except for those corresponding to our observations .
the model of @xcite shows more deviations with respect to recent observations .
for comparison , we also plot a third model using the same period reported by @xcite but with a * @xmath144 , which is in the order of magnitude of previously reported values , giving a * @xmath145 .
this corresponds to the same orbital period of @xcite but with a smaller value of @xmath146 .
it can be seen that it fits reasonably points from earlier observations .
however , a detailed analysis of transit time variations has not been considered in this work , which could explain some of the observed variations @xcite .
the present analysis then confirms the results of @xcite , also showing that a quadratic model does not improve significantly the fit of the transit timings for wasp-43b .
we report for the first time a light curve of wasp-39b in the @xmath1 band obtained with the @xmath0 telescope at spm - oan .
observations in this band are important to study the presence of high - hazes in the atmosphere , as rayleigh scattering becomes important at smaller wavelengths , which can produce larger observed radii in this band @xcite .
the resulting light curve is shown in fig . [ w39-curves ] .
although there is more dispersion in the data compared to the @xmath3 and @xmath2 bands , a depth flux of approximately @xmath147 is observed .
the best - fit values obtained in the previous section give a relative radius of @xmath148 .
no significant variations of the planetary radius between the @xmath1 , @xmath2 , and @xmath3 bands is found .
it is also noted that the @xmath1 and @xmath3 band light curves were obtained in the same night , and no evident asymmetries are observed that could suggest the presence of tails or other features around the planet .
additional observations in the @xmath1 band can help to reduce the uncertainty for this highly inflated extrasolar planet .
the analysis of the ephemerids in the previous section gives a period for wasp-39bapproximately @xmath149 larger than that of @xcite . ) .
the value reported by @xcite in the @xmath6 filter is also shown .
[ w43-radius ] ] wasp-43bshows a slight dependence of the @xmath97 as a function of the used filter ( see fig . [ w43-radius ] ) , which should be investigated with additional observations . in particular , we find for filter @xmath6 a value higher by @xmath13 with respect to the value reported by @xcite in the same filter .
( @xmath12 against @xmath150 ) .
although , as our two @xmath6 filter curves are not complete and show signatures of red noise , we suggest to carry on more observations to confirm the significancy of this result .
nevertheless , we find a tendency for the @xmath151 similar to that of @xcite in the sense that the planetary radius is significantly higher at the @xmath6 band with respect to the value in other bands .
the semi - major axis is slightly smaller , but comparable to the value obtained by @xcite and @xcite .
the analysis of the ephemerids gives a period consistent with previous results , and we confirm that there is no improvement while using a quadratic model for fitting the ephemerids .
further analysis and observations would be required to ascertain if a constant period variation exists for wasp-43b .
given the proximity of this planet to its host star , this is particularly relevant to explore potential mechanisms that remove orbital energy such as tidal dissipation .
+ the present work is part of an ongoing survey started in 2014 . a total of 23 extrasolar planets were observed mainly with the @xmath4 and the @xmath7 telescopes giving a total of 40 good quality light curves in several filters .
upgrades in the instrumentation , refinement of the observing technique , and the possibility to benefit from dark and photometric nights will increase the quality of the results in the near future , and will complement other telescope data for multi - site studies and analysis of this class of objects .
moreover , three new @xmath152 telescopes will be completed by 2016 at the san pedro mrtir observatory , which are part of the taos ii project @xcite .
the characteristics of this survey @xcite allow to set up specific pipelines for the early detection of new extrasolar planets , for example with a binning of the light curves as already tested for taos data @xcite .
also , a project for the installation of a @xmath153 telescope at the san pedro mrtir observatory has recently reached the phase of preliminary design .
current and _ in fieri _ spm - oan telescopes then represent a unique opportunity for the extrasolar planet community to complement the northern hemisphere observatories involved in this research area .
multi - filter exoplanet transit observations of wasp-39band wasp-43b , carried on for the first time with all three san pedro mrtir telescopes , have shown the capability of these facilities of @xmath4 , @xmath7 and @xmath0 for this kind of investigation .
the fit of the two objects shows a scatter of @xmath8@xmath9 rms in condition of full or nearly - full moon .
this makes these instruments suitable for future ground - based observing campaigns , in particular for the follow - up of alerts triggered by the upcoming projects currently in development .
a tendency of a higher planet radius in the @xmath6 band is obtained , consistent with the tendency reported in recent works .
additional observations of this object are needed to confirm a slight dependence of this parameter on the observing filter .
research carried out thanks to the support of unam - dgapa - papiit project in115413 .
fgrf acknowledges the support of a conacyt scholarship for postgraduate studies in mexico .
we also acknowledge dr . wolfgang steffen burg for the discussion about future exoplanetary - related projects , john southworth for his useful comments about the form ant the content of this manuscript , and the oan technical and service staff for daily support .
we finally acknowledge the anonymous referee for the useful remarks . ,
n. , klein , c. , fox , o. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
8446 , society of photo - optical instrumentation engineers ( spie ) conference series , a. , gonzlez , j. j. , kutyrev , a. s. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 8446 , society of photo - optical instrumentation engineers ( spie ) conference series , j. z. , johnson , j. a. , tonry , j. , et al .
2011 , transit analysis package ( tap and autokep ) : idl graphical user interfaces for extrasolar planet transit photometry , astrophysics source code library , ascl:1106.014 , j. c. , wang , s .- y . ,
lehner , m. j. , jorden , p. , & fryer , m. 2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
8446 , society of photo - optical instrumentation engineers ( spie ) conference series , m. j. , wang , s .- y .
, alcock , c. a. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
8444 , society of photo - optical instrumentation engineers ( spie ) conference series , s. , kohler , d. , bouchy , f. , et al .
2008 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 7014 , society of photo - optical instrumentation engineers ( spie ) conference series , a. m. , richer , m. g. , bloom , j. s. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol
. 8444 , society of photo - optical instrumentation engineers ( spie ) conference series | three optical telescopes located at the san pedro mrtir national observatory were used for the first time to obtain multi - filter defocused photometry of the transiting extrasolar planets wasp-39band wasp-43b .
we observed wasp-39bwith the @xmath0 telescope in the @xmath1 filter for the first time , and additional observations were carried out in the @xmath2 and @xmath3 filters using the @xmath4 telescope .
wasp-43b was observed in @xmath5 with the same instrument , and in the @xmath6 filter with the robotic @xmath7 telescope .
we reduced the data using different pipelines and performed aperture photometry with the help of custom routines , in order to obtain the light curves .
the fit of the light curves ( @xmath8@xmath9 rms ) , and of the period analysis , allowed a revision of the orbital and physical parameters , revealing for wasp-39ba period ( @xmath10 days ) which is @xmath11 seconds larger than previously reported .
moreover , we find for wasp-43ba planet / star radius ( @xmath12 ) which is @xmath13 larger in the @xmath6 filter with respect to previous works , and that should be confirmed with additional observations . finally , we confirm no evidence of constant period variations in wasp-43b . |
static bodies resulting from combined gravitational attraction and electrical ( electrostatic ) repulsion have been playing a central role in many investigations , eg .
@xcite and references therein , connected with general relativistic treatment of gravitational contraction . as a part of einstein equation , one has to specify energy - momentum tensor which , for a perfect fluid , has the form @xmath0 where @xmath1 , @xmath2 , @xmath3 and @xmath4 are mass density , pressure , four - velocity and metric ( in geometrized units : @xmath5 ) .
in addition , one has to specify the equation of state relating the pressure to the density @xmath6 . according to the oppenheimer - snyder scenario @xcite , in a static spherically symmetric perfect fluid ball with the above energy momentum tensor , after stationary phase and at the end of nuclear burning ,
the central pressure is no longer able to counterbalance the gravitational attraction and the ball starts to collapse . by setting @xmath7
the energy momentum tensor assumes the form characteristic for dust ball @xmath8 .
eventually , as the ball shrinks one can expect formation of a black hole .
the majumdar papapetrou formalism @xcite can be applied to study properties of spacetimes generated by @xmath8 where @xmath1 represents static electrically counterpoised dust ( or extremal charged dust , ecd ) , i.e. pressureless matter in equilibrium under its own gravitational attraction and electrical repulsion @xcite .
special attention has been paid to ecd distributions leading to spacetimes that are regular everywhere , but with exterior arbitrarily close to that of an extremal reissner
nordstrm ( ern ) black hole @xcite .
this paper considers the coupled einstein - maxwell field equations in the majumdar
papapetrou approach . with the energy - momentum tensor given by ( [ eq : t1 ] ) with @xmath7
, our study will concentrate on diverse ( analytic ) spherically symmetric forms of ecd distributions and corresponding solutions will be found .
the theoretical framework is presented in section [ sec : rev ] . in section [ sec : bif ] , we present the static solutions to einstein - maxwell equations for diverse ecd distributions obtained numerically , and in section [ sec : delta ] for the @xmath9-shell distribution , all of which show interesting bifurcating properties . in section [ sec : uni ] , we discuss the bifurcating behaviour and show that diverse distributions can be treated on equal footing .
similar behaviour found in gauge theories and einstein yang mills systems is discussed , and we give some details on numerical methods for finding bifurcating solutions . in section [ sec : qbh ] , we show that with adjusting of the source strength and appropriate rescaling of the solutions , the spacetimes with exteriors arbitrarily close to the ern case can be obtained .
conclusions are given in section [ sec : concl ] , and extension of the present work is proposed .
one usually starts by defining a static spherically symmetric line element of the form @xmath10 where @xmath11 .
for an asymptotically flat spacetime , one requires @xmath12 and for a nonsingular spacetime @xmath13 .
papapetrou has shown @xcite that the line element can be written as @xmath14 and that it is possible to connect functions @xmath15 and @xmath16 in such a way as to satisfy ( as one possibility ) @xmath17 or @xmath18 . together with majumdar
s assumption @xcite about connection between @xmath19 ( or @xmath20 ) metric component and a scalar component of the electromagnetic potential @xmath21 comprising the electromagnetic energy - momentum tensor @xmath22 , the line element @xmath23 in the majumdar
papapetrou form can be written in the harmonic coordinates @xmath24 as @xmath25 ( in the original notation of @xcite ) .
we will assume the spherical symmetry as a general symmetry requirement for our problem .
the essential ingredient to the line element ( [ eq : ds2:maj ] ) , as shown by majumdar , are additional assumptions which should be made about a specific form for the energy - momentum tensor @xmath26 that enters the einstein equations @xmath27 ( in geometrized units ) .
the component of @xmath28 due to electromagnetic fields is given by @xmath29 where @xmath30 is the antisymmetric electromagnetic field tensor , comma denoting the ordinary derivative to differ from semicolon denoting the covariant derivative .
it satisfies the empty space equation @xmath31 . a more general situation differing from the one described above corresponds to @xmath32 where the superscript @xmath33 denotes matter .
the absence of @xmath34 leads to reissner
nordstrm solutions with @xmath35 for @xmath36 .
it is shown in ref .
@xcite that for charged sphere of radius @xmath37 , for @xmath38 large enough , @xmath39 could be set equal to zero , whereas for the interior solution , i.e. @xmath40 , one is not permitted to put @xmath41 .
it follows that electric charge contributes to the gravitational mass of the system .
in addition , mass as given in @xmath42 above satisfies a positivity condition @xmath43 . citing ref .
@xcite ` a charged sphere must have a positive mass ' . in the case of the complete @xmath28 given in ( [ eq : t2 ] )
one assumes also a perfect fluid of classical hydrodynamics for which , in general , energy - momentum tensor is given by ( [ eq : t1 ] ) . with @xmath7 , the energy momentum tensor ( [ eq : t1 ] ) reduces to @xmath44 , where @xmath1 is now an invariant dust density and @xmath45 is the four - velocity which will be assumed to be in non - moving ( co - moving ) coordinate system so that @xmath46 the electromagnetic part of @xmath26 is given , as before , with ( [ eq : tem ] ) .
the four - current density of the the charge / matter / dust distribution is given by @xmath47 at this point , the crucial assumption is made by setting the electric charge density @xmath48 equal to the matter density @xmath1 , or somewhat more general @xmath49 .
now the metric ( [ eq : ds2:maj ] ) can be written in the form @xmath50 the einstein field equations incorporating the above introduced ingredients are given by @xmath51 , \end{aligned}\ ] ] and by using the majumdar ( or majumdar papapetrou ) assertion , the generalized ( nonlinear ) poisson equation ( with @xmath52 ) @xmath53 assumes a simple form @xmath54 here @xmath55 and @xmath1 are functions of the co - moving coordinates @xmath56 determined from the zeroth component of the electromagnetic potential @xmath21 , i.e. @xmath57 confining our attention to the spherically symmetric case , the majumdar papapetrou metric can be written as @xmath58 the metric function @xmath55 is a function of @xmath59 only , so ( [ eq : lapu ] ) reduces to @xmath60 where the prime denotes the differentiation with respect to @xmath59 . the line element ( [ eq : ds2:u : r ] ) can be expressed in the standard form ( [ eq : ds2:ab ] ) through the coordinate transformation @xmath61 with the following relations among the profile functions : @xmath62 and @xmath63 in regions of space where @xmath64 , the nonlinear equation ( [ eq : lapu : r ] ) reduces to a homogeneous equation with the general solution @xmath65 where @xmath66 and @xmath39 are integration constants . using ( [ eq : r = ur])([eq : a : u ] ) , to express the line element in the standard form ( [ eq : ds2:ab ] ) , one obtains @xmath67 . if the region of space we are considering extends to @xmath68 , according to the requirements ( [ eq : ab : inf ] ) , we set @xmath69 , and only @xmath39 remains as a free parameter .
the line element is then @xmath70 which we recognize as the extremal reissner
nordstrm ( ern ) spacetime .
the general reissner nordstrm ( rn ) solution specifies the asymptotically flat spacetime metric around an electrically charged source . expressing the rn line element in the standard form ( [ eq : ds2:ab ] )
, one has @xmath71 , where @xmath39 and @xmath72 are the adm mass and charge of the source . for @xmath73 ,
the rn spacetime is regular everywhere except at @xmath74 , while for @xmath75 , in addition to the singularity at @xmath74 , it exhibits two horizons located at @xmath76 . in the ern case , i.e. @xmath77 ( gravitational attraction balances electrical repulsion )
, only one horizon at @xmath78 is present .
it is important to note that the solution ( [ eq : uh ] ) is valid in the range @xmath79 which is according to eq .
( [ eq : r = ur ] ) mapped to @xmath80 .
that is to say that the solution ( [ eq : uh ] ) specifies only the external part of the ern spacetime ( see @xcite ) . we shall proceed to solve the field equation ( [ eq : lapu : r ] ) for several distributions of electrically counterpoised dust ( or extremal charged dust , ecd ) @xmath81 .
the ecd distributions that will be considered effectively vanish at large @xmath59 , so we expect our spacetimes to behave like ( [ eq : ds2:ern ] ) for @xmath82 .
the parameter @xmath83 , characterizing the asymptotic behaviour of @xmath84 at large @xmath59 , is the adm mass seen by the distant observer .
the behaviour of @xmath84 as @xmath85 can be characterized by the parameter @xmath86 .
the metric ( [ eq : ds2:u : r ] ) is regular at @xmath87 only if @xmath88 and , according to ( [ eq : r = ur ] ) , the range @xmath79 is mapped to @xmath89 . on the other hand , if @xmath90 the function @xmath84 is infinite at @xmath87 which indicates the singularity . as the range @xmath79 is now mapped to @xmath91 , this singularity is , in the @xmath38-coordinate , located at @xmath92 . therefore , we may understand the parameter @xmath93 as the ` mass below horizon ' . if @xmath94 we have @xmath95 , while in case of non - vanishing ecd density @xmath1 we have @xmath96 where @xmath97 can be understood as the contribution of ecd to the adm mass of the configuration .
the ` ecd mass ' @xmath97 is the space integral of the ecd density @xmath1 : @xmath98 we conclude this section by pointing out , for later convenience , that if certain @xmath84 and @xmath81 solve the equation ( [ eq : lapu : r ] ) , one is allowed to rescale the functions @xmath99 which , as a consequence , rescale the mass parameters according to @xmath100 these relations will help us compare field configurations corresponding to different mass parameters .
our approach in constructing solutions to the majumdar
papapetrou systems is to assume certain ecd distribution @xmath81 and then integrate the differential equation ( [ eq : lapu : r ] ) numerically to obtain the metric function @xmath84 , and is therefore different from the approach used in @xcite where @xmath81 is analytically reconstructed from the assumed metric and the requirement that @xmath81 may not be negative . at the expense of requiring numerical procedures
our approach allows complete freedom in choosing the shape of @xmath81 , which we found more natural .
however , in accord with the requirement that our solutions be asymptotically flat , we choose ecd distributions @xmath81 that are well localized in the @xmath59-coordinate .
more specifically , we require that the total mass / charge in a linear theory @xmath101 is finite , i.e. that @xmath81 falls off more rapidly than @xmath102 . as our first example we take @xmath81 of the form @xmath103 where @xmath104 is the length scale that in further text we set equal to @xmath105 , @xmath106 has the role of an adjustable source strength factor ( in appropriate units ) and @xmath107 is a normalization constant that renders @xmath108 for @xmath109 and @xmath110 . as ( [ eq : lapu : r ] ) is a second order differential equation we have to impose two boundary conditions onto a solution . as the first boundary condition
, we use the requirement that the spacetime is asymptotically flat , i.e. @xmath111 , and as the second we chose to fix the parameter @xmath112 .
we obtain the solutions by numerical integration @xcite of the differential equation . starting with the solution of ( [ eq : lapu : r ] ) for source strength @xmath113 ,
i.e. @xmath114 , we slowly increase the value of @xmath106 .
increasing of @xmath106 evolves the solution toward @xmath115 , but one may only increase the value of @xmath106 up to a critical value @xmath116 .
the dependence of @xmath117 on @xmath106 in solutions obtained in this way is shown as the lower part of the solution tracks in figure [ fig : eta ] .
the critical points are labeled @xmath118 , @xmath119 , @xmath120 , and @xmath121 , and the corresponding values of @xmath116 and @xmath117 are given in table [ tbl : eta ] .
the dependence of @xmath122 ( adm mass ) on source strength @xmath106 in solutions to ( [ eq : lapu : r ] ) with ecd distribution given by ( [ eq : rho1 ] ) and boundary conditions @xmath88 ( leading to regular spacetimes ) and @xmath123 ( leading to singular spacetimes ) .
the numerical values of @xmath106 and @xmath122 for the solutions labeled by letters @xmath118@xmath121 are given in table [ tbl : eta ] . ]
.[tbl : eta ] numerical values of @xmath106 and @xmath122 for the solutions labeled by letters in figure [ fig : eta ] .
[ cols="<,<,<,<,<",options="header " , ] the @xmath38-space metric components @xmath124 and @xmath125 ( dashed lines ) for the critical regular solutions obtained with the first three mp - normalized sources of figure [ fig : uni : eta ] and table [ tbl : eta : uni ] ( dashed lines ) , and for @xmath126 ern metric components ( dotted line ) .
( also to be compared with metric components in figure [ fig : del : ab ] . ) ]
using a mp - normalized ecd distribution @xmath127 , the critical regular solution to the field equation ( [ eq : lapu : r ] ) is obtained with the source strength @xmath128 .
the @xmath38-coordinate metric components of the critical regular solutions to ( [ eq : lapu : r ] ) and the ecd distribution ( [ eq : rho : ab : mp ] ) with parameters @xmath129 and @xmath130 are shown in figure [ fig : uni : abc ] . at large @xmath38 ,
metric components @xmath131 and @xmath125 coalesce into the @xmath126 ern metric , while at the intermediate values where major part of the ecd is distributed , they are manifestly regular and do not show significant dependence on the choice of the shape of the ecd distribution . by using @xmath132 , these solutions can ,
due to the bifurcating behaviour , either evolve toward flat space along @xmath133 branch , or toward higher mass configurations along the @xmath134 branch .
any of these solutions can be rescaled to describe a unit mass configuration if the rescaling according to ( [ eq : alpha1 ] ) and ( [ eq : alpha2 ] ) is carried out with the scaling parameter @xmath135 .
starting from the critical solutions shown in figure [ fig : uni : abc ] we followed the @xmath134-branch to obtain the field configurations corresponding to @xmath136 and @xmath137 which we then rescaled to restore the unit mass configurations .
the metric components of these solutions are shown in figure [ fig : uni : ab10 ] .
the @xmath38-space metric components obtained by using the mp - normalized sources as in figure [ fig : uni : abc ] to generate @xmath136 ( upper panel ) and @xmath137 ( lower panel ) regular solutions which were then rescaled to describe unit mass objects .
@xmath126 ern metric ( dotted line ) . ]
the @xmath38-space metric components obtained by using the mp - normalized sources as in figure [ fig : uni : abc ] to generate @xmath136 ( upper panel ) and @xmath137 ( lower panel ) regular solutions which were then rescaled to describe unit mass objects . @xmath126 ern metric ( dotted line ) . ]
it is plausible that , by following the @xmath134-branch toward higher masses and then rescaling the solutions to describe the @xmath126 configurations , the minimum of the function @xmath138 is deeper and closer to @xmath139 .
asymptotically , in the region @xmath140 the metric components @xmath141 and @xmath142 follow the ern metric and as the point @xmath139 is approached from @xmath140 there appears to be a double zero both in @xmath138 and @xmath142 . in the @xmath143 region we would asymptotically have @xmath144 which would not allow for timelike intervals . however , since our spacetimes are regular everywhere and such a situation is realized only asymptotically , we can consider a radially moving photon for which @xmath145 and @xmath146 , so @xmath147 .
the ratio of the metric components @xmath148 for the @xmath136/rescaled solutions , related to the time required for the photon to transverse a unit distance in the @xmath38-coordinate , is shown in figure [ fig : uni : photon ] .
the distributions of the ecd for the solutions of figure [ fig : uni : ab10 ] are shown in figure [ fig : uni : rho ] .
asymptotically , as we would go higher on the @xmath134-branch and rescale to restore unit mass configurations , the density would be completely pulled within the @xmath139 region .
the same was obtained by @xcite .
the ratio @xmath149 for the solutions shown in figure [ fig : uni : ab10 ] ( dashed lines ) and for the ern metric ( dotted lines ) . ] the density @xmath150 for the solutions in figure [ fig : uni : ab10 ] . the densities that effectively vanish at @xmath151 correspond to @xmath137/rescaled solutions , while those that effectively vanish at @xmath152 correspond to @xmath136/rescaled solutions . ]
the majumdar papapetrou formalism provides a good environment to study solutions to einstein
maxwell equations where matter is assumed to be described by electrically counterpoised dust ( ecd ) . in this paper , we have obtained and analyzed regular and quasi black hole solutions stemming from the m p formalism and obtained for diverse spherically symmetric ecd distributions by numerical integration of the nonlinear field equations . as an immediate consequence of nonlinearity , bifurcating solutions have been identified with respect to the amount of adm mass allocated in the mass source term .
also an upper bound to the source strength has been found above which no solution exists .
although ecd distributions have assumed analytically different ( spherically symmetric ) forms , we have been able to reformulate the sources to treat them on equal footing . from this treatment
we have been able to obtain regular solutions that come arbitrarily close to black hole solutions , the so called quasi black holes .
investigation of such bodies could be useful in investigations of interiors of black holes that still hide many unanswered questions .
bifurcation is not an unusual feature in gauge field theory @xcite or in gravity @xcite .
it is encouraging that here we are able to show that the bifurcation described in this work is not an artifact of a particular choice of charge / matter / energy density .
stability and bifurcation are closely related problems , so investigation of the above solutions with regard to stability is a natural extension of this work , bearing in mind the citation from ref .
@xcite ` stability analysis may be more expensive than the calculation of the solutions themselves . '
this work is supported by the croatian ministry of science and technology through the grant zp 0036038 .
authors would like to thank for hospitality the abdus salam international centre for theoretical physics , trieste , italy , where part of this work was carried out . | static spherically symmetric distributions of electrically counterpoised dust ( ecd ) are used to construct solutions to einstein - maxwell equations in majumdar papapetrou formalism .
unexpected bifurcating behaviour of solutions with regard to source strength is found for localized , as well as for the delta - function ecd distributions .
unified treatment of general ecd distributions is accomplished and it is shown that for certain source strengths one class of regular solutions approaches minkowski spacetime , while the other comes arbitrarily close to black hole solutions . |
the essence of the model @xcite is the order parameter equation for the complex amplitude @xmath13 of parametric layer oscillations @xmath14 at the frequency @xmath9 coupled with the equation for the thickness of the layer @xmath15 averaged over the period of vibrations , @xmath16 here @xmath17 is the normalized amplitude of forcing at the resonant frequency @xmath0 .
linear terms in eq .
( [ eq0a ] ) can be obtained from the complex growth rate for infinitesimal periodic layer perturbations @xmath18 $ ] .
expanding @xmath19 for small @xmath20 , and keeping only two leading terms in the expansion @xmath21 we obtain the linear terms in eq .
( [ eq0a ] ) , where @xmath22 characterizes ratio of dispersion to diffusion and parameter @xmath23 , @xmath24 characterizes the frequency of the driving . in eq .
( [ eq0b ] ) @xmath25 are transport coefficients for @xmath15 .
slowly - varying thickness of the layer @xmath15 controls the dissipation rate ( the last term in eq.([eq0a ] ) ) . the second equation ( [ eq0b ] )
describes re - distribution of the averaged thickness due to the diffusive flux @xmath26 , and an additional flux @xmath27 caused by the spatially nonuniform vibrations of the granular material .
this coupled model was used in refs.@xcite to describe the pattern selection near the threshold of the primary bifurcation .
it was shown that at small @xmath28 ( which corresponds to low frequencies and thick layers ) , the primary bifurcation is subcritical and leads to the emergence of square patterns . for higher frequencies and/or thinner layers , transition is supercritical and leads to roll patterns . at intermediate frequencies ( @xmath29 ) , the stable localized solutions of eqs.([eq0a]),([eq0b ] ) corresponding to isolated _ oscillons _ and variety of bound states were found in agreement with experiment . in this paper
we focus on high - acceleration patterns at high frequencies . in ref .
@xcite it was indicated that the density transport coefficient @xmath30 is proportional to the energy of plate vibration ( @xmath31 ) , and , therefore , it should increase with the driving frequency @xmath0 . as a result , for high frequencies the coupling between @xmath15 and @xmath13 becomes less relevant , and one can assume @xmath32const and exclude it from eq.([eq0a ] ) by rescaling .
then the model can be reduced to a single order - parameter equation @xmath33 eq.([eq1 ] ) also describes the evolution of the order parameter for the parametric instability in vertically oscillating fluid layers ( see @xcite ) .
it is convenient to shift the phase of the complex order parameter via @xmath34 with @xmath35 .
the equations for real and imaginary part @xmath36 are : @xmath37 where @xmath38 .
at @xmath39 , eqs . ( [ eq2a]),([eq2b ] ) has only one trivial uniform state @xmath40 , at @xmath41 , two new uniform states appear , @xmath42 . the onset of these states corresponds to the period doubling of the layer flights sequence , observed in experiments @xcite and predicted by the simple inelastic ball model @xcite .
signs @xmath43 reflect two relative phases of layer flights with respect to container vibrations@xcite .
the trivial flat state @xmath44 loses stability if the following condition fulfilled ( compare @xcite ) : @xmath45 because of the symmetry of eq.([eq1 ] ) , small perturbations from the trivial state lead to the formation of a periodic sequence of rolls or stripes .
let us analyze the stability of the non - trivial states @xmath46 with respect to small perturbations with wavenumber @xmath20 @xmath47 .
\label{eq4}\ ] ] at the threshold the uniform state loses its stability with respect to periodic modulations with the critical wavenumber @xmath48 at @xmath49 ( correspondingly , @xmath50 ) , where @xmath51 small perturbations with every direction of the wavevector grow at the same rate . the resultant selected pattern is determined by the nonlinear competition between the modes . in the presence of the reflection symmetry
@xmath52 , quadratic nonlinearity is absent , and cubic nonlinearity near the trivial state favors stripes corresponding to a single mode . near the fixed points @xmath53 the reflection symmetry for perturbations @xmath54
is broken , and hexagons emerge at the threshold of the instability . to clarify this point
we perform weakly - nonlinear analysis of eqs .
( [ eq2a]),([eq2b ] ) for @xmath55 , and @xmath56 . at @xmath57 , the variables @xmath58 and @xmath59 are related as in linear system : @xmath60 + c.c . ,
\label{eq11}\end{aligned}\ ] ] where @xmath61 and @xmath62/ ( b k^2_c-2\omega)$ ] .
corresponding adjoint eigenvector is @xmath63 where @xmath64/ b k^2_c$ ] . substituting eq .
( [ eq11 ] ) into eqs .
( [ eq2a]),([eq2b ] ) and performing the orthogonalization , we obtain equations for the slowly - varying complex amplitudes @xmath65 , @xmath66 ( we assume only three waves with triangular symmetry , favored by quadratic nonlinearity ) @xmath67 where the coefficients @xmath68 are @xmath69 eqs .
( [ eq14 ] ) are well - studied ( see @xcite ) .
there are three critical values of @xmath70 : @xmath71 and @xmath72 .
the hexagons are stable for @xmath73 , and the stripes are stable for @xmath74 .
thus , near @xmath75 the model exhibit stable hexagons @xcite .
since we have two symmetric fixed points , both up- and down - hexagons co - exist . for smaller @xmath76 stripes
are stable , and for larger @xmath76 , flat layers are stable , in agreement with observations@xcite , as well as direct numerical simulations of eq.([eq1])@xcite .
the above analysis requires the values of @xmath77 to be small . for parameters @xmath78 ,
this requirement is satisfied for @xmath79 , but not for @xmath80 .
the estimates can be improved by substituting @xmath81 instead of @xmath75 in ( [ coef ] ) .
the resulting range of stable hexagons is plotted in the phase diagram @xmath82 ( see below fig.[phase ] ) .
at @xmath41 , eqs . ( [ eq2a]),([eq2b ] )
have an interface solution connecting two uniform states @xmath83 .
let @xmath84 be the coordinate perpendicular to the interface and @xmath85 is parallel to the interface .
the interface is the stationary solution to eqs .
( [ eq2a]),([eq2b ] ) : @xmath86 for @xmath87 eqs .
( [ int_s ] ) have a solution of the form @xmath88 . for @xmath89
the solution is available only numerically .
we used shooting matching technique to find this solution in the range of parameters @xmath90 and @xmath17 .
a typical solution is shown in fig .
[ int_num ] . as one sees from the figure , for @xmath91 the asymptotic behavior of the interface exhibits decaying oscillations .
= 8.2 cm let us now consider the perturbed solution @xmath92 .
\label{eq15}\ ] ] where @xmath20 is the transverse modulation wavenumber and @xmath93 is the corresponding growth rate .
for @xmath94 we obtain linear equations @xmath95 where the matrix @xmath96 is of the form @xmath97 in order to determine the spectrum of eigenvalues @xmath98 we have solved eq .
( [ eq16 ] ) along with stationary eqs .
( [ int_s ] ) numerically using numerical matching - shooting technique .
we have found that for small @xmath99 the interface is stable with respect to transverse undulations , i.e. @xmath100 .
however , for the values of @xmath99 above certain critical value @xmath101 the interface exhibits transverse instability : @xmath102 in the band of wavenumbers @xmath103 , see fig . [ int_spec ] .
= 8.2 cm this instability is confirmed by direct numerical simulations of eq.([eq1 ] ) .
an example of the evolution of slightly perturbed interface is shown in fig .
[ interf_fig ] .
small perturbations grow to form a `` decorated '' interface . with time
these decorations evolve slowly and eventually form a labyrinthine pattern .
the neutral curve for this instability can be determined as follows .
numerical analysis shows that at the threshold the most unstable wavenumber is @xmath104 and we can expect that for @xmath105 @xmath106 , see fig . [ int_spec ] . expanding eqs.([eq16 ] ) in power series of @xmath107 : @xmath108 in the zeroth order in @xmath20 we obtain @xmath109 .
the corresponding solution is the translation mode @xmath110 . in the first order in @xmath107
we arrive at the linear inhomogeneous problem @xmath111 a bounded solution to eq .
( [ eq18 ] ) exists if the r.h.s .
is orthogonal to the localized mode of the adjoint operator @xmath112 .
the adjoint operator @xmath113 is of the form @xmath114 since the operator @xmath113 is not self - adjoint , the adjoint mode does not coincide with the translation mode , and , therefore , must be obtained numerically .
the orthogonality condition fixes the relation between @xmath115 and @xmath20 : @xmath116 where @xmath117 is `` surface tension '' of the interface .
the instability , corresponding to the negative surface tension of the interface , onsets at @xmath118 .
although the interface solution @xmath119 is not localized ( see fig . [ int_num ] ) , the integrals in eq .
( [ coef3 ] ) converge because the adjoint functions @xmath120 are localized .
the numerically - obtained adjoint eigen - mode is shown in fig .
the neutral curve is shown in fig .
[ phase ] .
direct numerical simulations of eq.([eq1 ] ) show that this instability leads to so called `` decorated '' interfaces ( see fig .
[ interf_fig],a ) , however at later stages the undulations grow and finally form labyrinthine patterns ( fig .
[ interf_fig ] , b - d ) .
= 8.2 cm = 6.cm the emergence of the labyrinthine patterns as a result of interface instability contradicts experiments @xcite where stable decorated interfaces are typically observed .
thus , the model eq .
( [ eq1 ] ) does not capture saturation of the interface instability and requires modification .
we also checked that the coupling to the density field introduced in ref .
@xcite also does not provide desired saturation .
let us now discuss possible mechanisms of saturation of the transversal instability of the interface , which is not captured by eq .
( [ eq1 ] ) .
the reason for proliferation of the labyrinths is that local stabilization mechanisms can not saturate the negative surface tension instability " of the interface .
indeed , the behavior of small perturbation @xmath121 of almost flat interface is described by the following linear equation : @xmath122 , where @xmath117 is the surface tension
. the linear growth must be counter - balanced by nonlinear terms .
due to translation invariance the local nonlinearity can be function of gradients of @xmath121 only . in the lowest order the first term is @xmath123 , since the quadratic term does not saturate monotonic instability . since for the modulation wavenumber @xmath124 one has @xmath125 and @xmath126 , the linear instability always overcomes local nonlinearity at small wavenumbers . as a result
small perturbations grow and saturation happens due to non - local interactions as in a labyrinthine pattern which is stabilized by non - local repulsion of the fronts ( see @xcite ) .
as we see , in our model the interface instability does not result in stable decorated interfaces .
one of the possibility for the saturation of this instability could be local convection induced by the interface . indeed , due to shear motion of grains at the interface ( flat regions moves in anti - phase ) local convective flow can be created .
the scale of this flow can be large then the interface thickness .
one of the possible manifestation of the convection beside grain transport is reduction of the effective `` viscosity '' of flat layer , or increase in the sensitivity to external vibrations@xcite . as we can see from fig .
[ phase ] increase in vibration magnitude @xmath17 suppresses the interface instability .
as the instability develops , the length of the interface grows , which increases the convection and finally suppresses the instability .
this feedback mechanism results in saturation of the interface instability and creates stable decorations .
although the details of this coupling are not known yet , it can be implemented within the order - parameter approach in a phenomenological way . in eq .
[ eq1 ] the forcing @xmath127 must be replaced by @xmath128 , where @xmath129 includes the effect of local convective flow and @xmath130 .
we take the simplest possible form of the coupling to the convective flow : @xmath131 where @xmath132 characterizes the typical scale of the convective flow and @xmath70 is the amplitude of the coupling , and @xmath133 is the area of integration domain .
since @xmath134 is nonzero only at the interface , this integral is proportional to the total length of the interface . with the growth of the interface length the value of
effective forcing " @xmath135 grows as well , and eventually leaves the instability domain .
we performed numerical simulations with eq .
( [ eq1 ] ) modified in this fashion .
the convolution integral ( [ w ] ) was calculated using the fast fourier transform .
the results are presented in fig .
[ supos ] .
remarkably , for certain range of parameters the instability indeed saturates due to nonlocal term eq .
( [ w ] ) , and we find interfaces with stable decorations , see fig .
[ supos],a .
in addition to the interface solution which exists for @xmath41 , eq .
( [ eq1 ] ) for even higher values of @xmath76 possesses a variety of other nontrivial two - dimensional solutions including localized `` super - oscillons '' . the typical solution and its corresponding localized linear mode are shown in fig .
[ sup_osc ] .
= 8.2 cm = 8.2 cm although super - oscillon appears similar to the conventional oscillon " discovered in ref .
@xcite , there is a significant difference .
asymptotic behavior of the super - oscillon is @xmath136 for @xmath137 whereas the regular oscillon has @xmath138 . as a result super - oscillons
always have the same phase as their background .
oppositely - phased super - oscillons must be separated by the interface connecting two opposite phases of the period - doubled flat layer whereas oppositely - phased oscillons can form stable bound states ( dipoles , @xcite ) .
super - oscillons exist in a relatively narrow domain of parameters of eq .
( [ eq1 ] ) .
decrease of @xmath17 leads to destabilization of the localized eigen - mode leading to subsequent nucleation of new `` super - oscillons '' on the periphery of the unstable super - oscillon , with the increase of @xmath17 super - oscillons disappear in a saddle - node bifurcation .
interestingly , for certain values of @xmath139 we also find an interesting structure consisting of a decorated interface and a chain of super - oscillons bound to the decorations , see fig .
[ supos],b .
these objects can exist completely independently of the interface , fig .
[ supos ] c , as one may expect for eq .
( [ eq1 ] ) .
thus , we conclude that extended eq .
( [ eq1 ] ) qualitatively describes observed phenomenology .
the results of linear stability analysis can be summarized in the phase diagram shown in fig .
[ phase ] . remarkably , the structure of the phase diagram fig.[phase ] is qualitatively similar to that of the experiments for high frequencies of vibration , see fig .
[ phd ] and refs .
increasing vibration amplitude @xmath17 leads to transition from a trivial state to stripes , hexagons , decorated interfaces , and finally , to stable interfaces .
transition from unstable to stable interfaces also occurs with decreasing @xmath99 ( increasing vibration frequency @xmath0 ) , in agreement with ref.@xcite . in experiments , at yet higher @xmath6 , quarter - harmonic patterns appear , however these patterns are not described by our model .
our analysis also predicts co - existence of stripes and hexagons in a very narrow parameter range ( see sec .
super - oscillons were found in a narrow domain close to line 7 of fig .
[ phase ] .
note that square patterns which are emerge at lower frequencies , are not included in the phase diagram .
they are obtained within the full model eqs.([eq0a]),([eq0b ] ) with additional density field @xmath15 , see @xcite .
near the line @xmath140 ( see fig . [ phase ] ) eq .
( [ eq1 ] ) can be simplified . in the vicinity of this line @xmath141 and @xmath142 . in the leading order we can obtain from eq .
( [ eq2b ] ) @xmath143 , and eq .
( [ eq2a ] ) yields @xcite : @xmath144 rescaling the variables @xmath145 @xmath146 @xmath147 , we arrive to the swift - hohenberg equation ( she ) @xmath148 where @xmath149 the description by she is valid if @xmath150 which implies additional condition for the validity of this approach @xmath151 at @xmath152 .
although this equation is simpler than eq .
( [ eq1 ] ) , it captures many essential features of the original system dynamics , including existence and stability of stripes and hexagons in different parameter regions ( see @xcite ) , existence of the interface solutions , interface instability , super - oscillons and emergence of labyrinthine patterns @xcite .
indeed , simple analysis shows that the growth rate of the instability of the uniform state @xmath153 as a function of the perturbation wavenumber is determined by the formula @xmath154 , so it becomes unstable at @xmath155 at critical wavenumber @xmath156 . as in the original model , near the threshold of this instability , subcritical hexagonal patterns are preferred .
interface stability can also be analyzed more simply as the linearized operator corresponding to the model ( [ sh1 ] ) is self - adjoint .
the threshold value of @xmath157 is obtained from the following solvability condition @xmath158 which yields @xmath159 .
it yields the equation for the neutral curve of the interface instability in the swift - hohenberg limit in the form : @xmath160 fig .
[ interf_fig1 ] , a - d shows the development of the interface instability within she with subsequent transition to labyrinthine patterns , similar to the dynamics of the full parametric equation ( [ eq1 ] ) .
again , to saturate the instability , an additional non - local mechanism is required .
as it was shown recently , swift - hohenberg equation also possesses localized solutions ( see @xcite ) .
these localized states are analogous to the super - oscillons of eq .
( [ eq1 ] ) .
let us focus on the effect of additional subharmonic driving on the motion of the interface . in the original model eq .
( [ eq1 ] ) the interface does not move due to symmetry between left and right halves of the interface .
however , if the plate oscillates with two frequencies , @xmath0 and @xmath9 , the symmetry between these two states connected by the interface , is broken , and interface moves .
the velocity of interface motion depends on the relative phase of the sub - harmonic forcing with respect to the forcing at @xmath161 .
the effect of a small external subharmonic driving applied on the back - ground of primary harmonic driving with the frequency @xmath0 can be described by the following equation ( cf .
( [ eq1 ] ) ) @xmath162 where @xmath163 characterizes the amplitude of the sub - harmonic pumping , and @xmath164 determines its relative phase . for small @xmath163 ,
we look for moving interface solution in the form @xmath165 and @xmath166 , or , alternatively , @xmath167 solvability condition ( compare with eq .
( [ coef3 ] ) ) fixes the interface velocity as a function of amplitude and the phase of external subharmonic driving @xmath168 where @xmath169 has the meaning of susceptibility of the interface for external forcing and @xmath170 is some phase shift .
the explicit answer is possible to obtain for @xmath87 when @xmath171 , and @xmath172 , which yields the interface velocity @xmath173 . in general , @xmath174 , @xmath120 , and hence @xmath175 ,
can be found numerically . the interface velocity as function of @xmath176
is shown in fig .
[ interf_speed ] .
thus , from this analysis we conclude that additional subharmonic driving results in controlled motion of the interface .
the velocity of the interface depends on the amplitude of the driving , and the direction is determined by the relative phase @xmath164 .
we performed experiments with a thin layer of granular material subjected to periodic driving .
our experimental setup is similar to that of @xcite .
we used @xmath177 mm diameter bronze or copper balls , and the layer thickness in our experiments was 10 particles .
we performed experiments in a rectangular cell of @xmath178 @xmath179 .
we varied the acceleration @xmath6 and the frequency @xmath0 of the primary driving signal as well as acceleration @xmath17 , frequency @xmath180 and the phase @xmath164 of the secondary ( additional ) driving signal . the interface position and vertical accelerations
were acquired using a high - speed video camera and accelerometers and further analyzed on a pentium computer . to maintain and measure the proper acceleration at @xmath0 and @xmath9 we employ the lock - in technique with our signal originating from an accelerometer attached to the bottom of the cell .
this allows for the simultaneous _ real - time _ feed back and control of the device . to reduce the interstitial gas effects@xcite we reduced the pressure to 2 mtorr .
our visual data is acquired using a high speed digital camera ( kodak sr-1000c ) , in addition to high speed recording this also allows for a synchronization between the patterns and the image acquisition . the experimental phase diagram is shown in fig .
it is similar to that of ref .
@xcite , however the transition from hexagons to interfaces is elaborated in more detail .
the dashed line indicates the stability line for interfaces with respect to periodic undulations . to determine this stability limit
, we kept the value of acceleration @xmath6 fixed and increased the driving frequency @xmath0 . for each value of the frequency we extracted the interface width by averaging up to 10 images of the interface . in order to find that amplitude of the periodic undulations @xmath181 we subtracted from the obtained value the thickness of the flat interface @xmath182 . the resultant amplitude or undulations @xmath181 as function of driving frequency @xmath0 is shown in fig . [ wid ] .
as one sees from the figure , the amplitude of undulations @xmath181 decreases gradually approaching the critical value of the frequency . however , a small hysteresis at the transition point can not be excluded because of large error bars .
= 12 cm = 7.2 cm as it follows from the phase diagram , for small amplitude of the vertical acceleration @xmath183 stripes are the only stable pattern ( note that we focus on high - frequency patterns when squares do not exist , compare @xcite ) . at slightly higher @xmath184 stripes and hexagons coexist . for higher @xmath185
the hexagons becomes stable .
due to the sub - harmonic character of motion both up and down hexagons may co - exist separated by a line phase defect , fig .
[ int_sup],a .
there exists a narrow band , from @xmath186 to @xmath187 , where localized states , or _
super - oscillons _ , appear both within the bulk of the material and also pinned to the the front between the interface and the bulk fig .
[ int_sup],b . for even higher values of the acceleration @xmath188 the super - oscillons disappear ( fig . [ int_sup],c ) leading to isolated interfaces with periodic decorations .
= 8.2 cm in the absence of additional driving the interface drifts toward the middle of the cell ( see fig .
[ int_sup ] ) .
we attribute this effect to the feedback between the oscillating granular layer and the plate vibrations due to the finite ratio of the mass of the granular material to the mass of the vibrating plate . even in the absence of subharmonic drive
, the vibrating cell can acquire subharmonic motion from the periodic impacts of the granular layer on the bottom plate at half the driving frequency .
if the interface is located in the middle of the cell , the masses of material on both sides of the interface are equal , and due to the anti - phase character of the layer motion on both sides , an additional subharmonic driving is not produced . the displacement of the interface @xmath189 from the center of the cell leads to a mass difference @xmath190 on opposite sides of the interface which in turn causes an additional subharmonic driving proportional to @xmath190 . in a rectangular cell @xmath191 .
our experiments show that the interface moves in such a way to decrease the subharmonic response , and the feedback provides an additional term @xmath192 in the r.h.s . of eq .
( [ speed ] ) , yielding @xmath193 the relaxation time constant @xmath194 depends on the mass ratio ( this also holds for the circular cell if @xmath189 is small compared to the cell radius ) .
thus , in the absence of an additional subharmonic drive ( @xmath195 ) , the interface will eventually divide the cell into two regions of equal area ( see fig .
[ int_sup ] ) . in order to verify the prediction of eq .
( [ eq2_1 ] ) we performed the following experiment .
we positioned the interface off center by applying an additional subharmonic drive .
then we turned off the subharmonic drive and immediately measured the amplitude @xmath196 of the plate acceleration at the subharmonic frequency@xcite .
the results are presented in fig .
[ fig_rel ] .
the subharmonic acceleration of the cell decreases exponentially as the interface propagates to the center of the cell .
the measured relaxation time @xmath194 of the subharmonic acceleration increases with the mass ratio of the granular layer and of the cell with all other parameters fixed .
the mass of the granular layer was varied by using two different cell sizes : circular , diameter @xmath197 cm , and rectangular , @xmath198 cm , while keeping the thickness of the layer unchanged .
for these cells we found that the relaxation time @xmath194 in the rectangular cell is about four times greater than for the circular cell ( see fig .
[ fig_rel ] ) .
this is consistent with the ratio of the total masses of granular material ( 52 grams in rectangular cell and 198 grams in circular one ) . in a separate experiment
the mass of the cell was changed by attaching an additional weight of 250 grams to the moving shaft , which weights 2300 grams .
this led to an increase of the corresponding relaxation times of 15 - 25 % .
the relaxation time @xmath194 increases rapidly with @xmath6 ( see fig .
[ fig_rel ] inset ) .
= 3.5 in when an additional subharmonic driving is applied , the interface is displaced from the middle of the cell . for small amplitude of the subharmonic driving @xmath17 ,
the stationary interface position is @xmath199 , since the restoring force balances the external driving force . the equilibrium position @xmath189 as function of @xmath164 is shown in fig .
[ fig_accl]a .
the solid line depicts the sinusoidal fit predicted by the theory . because of the finite mass ratio effect
, the amplitude of the measured plate acceleration @xmath196 at frequency @xmath180 also shows periodic behavior with @xmath164 , ( see fig .
[ fig_accl]b ) , enabling us to infer the interface displacement from the acceleration measurements . for even larger amplitude of subharmonic
driving ( more than 4 - 5 % of the primary driving ) extended patterns ( hexagons ) re - emerge throughout the cell .
the velocity @xmath200 which the interface would have in an infinite system , can be found from the measurements of the relaxation time @xmath194 and maximum displacement @xmath201 at a given amplitude of subharmonic acceleration @xmath17 , @xmath202 , see eq.([eq2_1 ] ) .
we verified that in the rectangular cell the displacement @xmath189 depends linearly on @xmath17 almost up to values at which the interface disappears at the short side wall of the cell ( see fig .
[ fig_dis ] , inset ) .
figure [ fig_dis ] shows the susceptibility @xmath203 as a function of the amplitude of the primary acceleration @xmath6 .
the susceptibility decreases with @xmath6 .
the cusp - like features in the @xmath6-dependence of @xmath203 ( and @xmath194 , see inset to fig .
[ fig_rel ] ) , are presumably related to the commensurability between the lateral size of the cell and the wavelength of the interface decorations .
= 4 in = 4 in we developed an alternative experimental technique which allowed us to measure simultaneously the relaxation time @xmath194 and the `` asymptotic '' velocity @xmath204 .
this was achieved by a small detuning @xmath205 of the additional frequency @xmath180 from the exact subharmonic frequency @xmath9 , i.e. @xmath206 .
it is equivalent to the linear increase of phase shift @xmath164 with the rate @xmath207 .
this linear growth of the phase results in a periodic motion of the interface with frequency @xmath208 and amplitude @xmath209 ( see eq ( [ eq2_1 ] ) ) .
the measurements of the response functions " @xmath210 are presented in fig .
[ fig_de ] . from the dependence of @xmath201 on @xmath205
we can extract parameters @xmath211 and @xmath194 by a fit to the theoretical function .
the measurements are in a very good agreement with previous independent measurements of relaxation time @xmath194 and susceptibility @xmath203 . for comparison with the previous results ,
we indicate the values for @xmath194 and @xmath203 , obtained from the response function measurements of figs .
[ fig_rel ] and [ fig_dis ] ( stars ) .
the measurements agree within @xmath212 % .
in this paper we studied the dynamics of thin vibrated granular layers in the framework of phenomenological model based on the ginzburg - landau type equation for the order parameter characterizing the amplitude of sub - harmonic sand vibrations . although rigorous derivation of this model is not possible up to date , our approach gives a significant insight in the problem in hand and results in testable predictions .
our model is complementary to more elaborate fluid dynamics - like continuum descriptions or direct molecular dynamics simulations and allows us to carry out analytical calculations of various regimes and predict their stability .
a challenging problem is to establish a quantitative relation between our order parameter model and other models and experiments .
we have shown in this paper that the generic parametric ginzburg - landau model eq.([eq1 ] ) captures not only patterns of vibrated sand near the primary bifurcation , but also large acceleration patterns such as hexagons , interfaces , and super - oscillons .
the structure of the phase diagram fig.[phase ] for high frequencies and amplitudes of vibrations is qualitatively similar to that of previous experiments refs.@xcite , as well as the experiments reported here .
increasing vibration amplitude leads to transition from a trivial state to stripes , hexagons , decorated interfaces , and finally , to stable interfaces .
interface decorations were found to be a result of the transversal instability of a flat interface which is analogous to the negative surface tension .
we found that eq .
( [ eq1 ] ) is not sufficient to describe the saturation of interface instability .
we propose a possible non - local mechanism of saturation of this instability which takes into account the dependence of the overall length of the interface and magnitude of parametric forcing , which may occur due to a large scale convective flow induced by the sand motion near the interface .
we described analytically the motion of the interface under the symmetry - breaking influence of the small subharmonic driving . in our experimental study we found that on a qualitative level the theoretical phase diagram is similar to the experimental one
the experiments also confirmed existence of super - oscillons and their bound states with the interface as it was predicted by the theory .
further experimental study is necessary to elucidate the specific mechanism for saturation of the interface instability .
stimulated by the theoretical prediction , we also performed experimental studies of interface motion under additional subharmonic driving .
the experiment confirmed that the direction and magnitude of the interface displacement depend sensitively on the amplitude and the relative phase of the subharmonic driving .
moreover , we found that period - doubling motion of the flat layers produces subharmonic driving because of the finite ratio of the mass of the granular layer and the cell .
this in turn leads to the restoring force driving the interface towards the middle of the cell .
we thank l. kadanoff , r. behringer , h. jaeger , h. swinney and p. umbanhowar for useful discussions .
this research was supported by the us department of energy , grant # w-31 - 109-eng-38 , and by nsf , stcs # dmr91 - 20000 .
c.j . acknowledges support from joint anl - uofc collaborative grant .
thanks the engineering research program of the office of basic energy sciences at the u.s .
department of energy , grants # de - fg03 - 96er14592 , de - fg03 - 95er14516 for support . c. bizon , m.d .
shattuck , j.b .
swift , w.d .
mccormick , h.l .
swinney , , 57 ( 1998 ) j.r
. de bruyn , c. bizon , m.d .
shattuck , d. goldman , and h.l .
swinney , , 1421 ( 1998 ) s. luding , e. clement , j. rajchenbach , and j. duran , europhys . lett . * 36 * , 247 ( 1996 ) .
d. rothman , ( 1998 ) ; e. cerda , f. melo , and s. rica , , 4570 ( 1997 ) ; t. shinbrot , nature * 389 * , 574 ( 1997 ) ; j. eggers and h. riecke , , 4476 ( 1999 ) ; s. c. venkataramani and e. ott , , 3495 ( 1998 ) . c. crawford and h. riecke , physica d * 129 * , 83 ( 1999 ) l.s .
tsimring and i.s .
aranson , , 213 ( 1997 ) ; i.s .
aranson and l.s .
tsimring , physica a * 249 * , 103 ( 1998 ) .
i.s . aranson , l.s .
tsimring , and v.m .
vinokur , 1327 ( 1999 ) .
das and d. blair , phys .
a * 242 * , 326 ( 1998 ) .
note that eq.([eq1 ] ) also describes the evolution of the order parameter for the parametric instability in vertically oscillating fluids ( see @xcite ) .
however , unlike granular materials , fluid layer can not leave the surface and therefore the non - trivial uniform states are prohibited .
w. zhang and j. vinals , , 690 ( 1995 ) .
kiyashko , l.n .
korzinov , m.i .
rabinovich , and l.s .
tsimring , , 5037 ( 1996 ) .
a. mehta and j.m .
luck , , 393 ( 1990 ) .
m. cross and p.c .
hohenberg , 851 ( 1993 ) .
this analysis is similar to that of ref .
@xcite where stability of hexagons was demonstrated for the she . in a certain limit
( [ eq1 ] ) can be reduced to she ( see below ) .
numerical experiments were performed using quasi - spectral split - step method in a @xmath213 domain with @xmath214 mesh points and periodic boundary conditions .
petrich and r.e .
goldstein , , 1120 ( 1994 ) ; d.m .
petrich , d.j.muraki , and r.e .
goldstein , , 3933 ( 1996 ) .
the importance of large - scale flow was indirectly demonstrated also in ref .
p.b.umbanhowar , ph .
thesis , u.texas , austin , 1996 .
she for oscillatory systems with strong resonance coupling was derived by p. coullet and k. emilson , physica d * 61 * , 119 ( 1992 ) .
g. dewel , s. mtens , mf .
hilali , p. borckmans and c.b .
price , , 4647 ( 1995 ) . in ref .
@xcite modified swift - hohenberg equations was used to describe squares and oscillons .
i. s. aranson , b. a. malomed , l. m. pismen , and l. s. tsimring , submitted to ( 1999 ) .
k. ouchi and h. fujisaka , , 3895 ( 1996 ) ; k. staliunas and v.j .
snchez - morcillo , phys .
a * 241 * , 28 ( 1998 ) .
pak e. vandoorn and r.p .
behringer , 4643 ( 1995 ) .
the measured acceleration @xmath196 may differ from the applied subharmonic ( sinusoidal ) driving @xmath163 since the granular material moves inside the cell .
we measure @xmath163 independently by removing the granular material from the cell . | theoretical and experimental study of high - acceleration patterns in vibrated granular layers is presented . the order parameter model based on parametric ginzburg - landau equation
is used to describe strongly nonlinear excitations including hexagons , interface between flat anti - phase domains and new localized objects , _ super - oscillons_. the experiments confirmed the existence of super - oscillons and bound states of super - oscillons and interfaces . on the basis of order parameter model
we predict analytically and confirm experimentally that the additional subharmonic driving results in controlled motion of the interfaces .
the collective dynamics of granular materials is a subject of current interest @xcite .
the intrinsic dissipative nature of the interactions between the constituent macroscopic particles give rise to several basic properties specific to granular substances and setting granular matter apart from the conventional gaseous , liquid , or solid states .
driven granular systems manifest collective fluid - like behavior : convection , surface waves , and pattern formation ( see e.g. @xcite ) .
one of the most fascinating examples of this collective dynamics is the appearance of long - range coherent patterns and localized excitations in vertically - vibrated thin granular layers @xcite .
the particular pattern is determined by the interplay between driving frequency @xmath0 and acceleration of the container @xmath1 ( @xmath2 is the amplitude of oscillations , @xmath3 is the gravity acceleration ) @xcite .
patterns appear at @xmath4 almost independently of the driving frequency @xmath0 . at small frequencies
@xmath5 @xcite the transition is subcritical ( hysteretic ) , leading to formation of squares . in the hysteretic region ,
localized excitations such as individual _
oscillons _ and various bound states of oscillons appear as @xmath6 is decreased . for higher frequencies
@xmath7 the onset pattern is stripes , and at the frequency slightly higher than @xmath8 the transition becomes supercritical . both squares and stripes , as well as oscillons , oscillate at half of the driving frequency @xmath9 . at higher acceleration ( @xmath10 ) , stripes and
squares become unstable , and hexagons appear instead .
further increase of acceleration at @xmath11 converts hexagons into a domain - like structure of flat layers oscillating with frequency @xmath9 with opposite phases .
depending on parameters , interfaces separating flat domains , are either smooth or
decorated by periodic undulations . for @xmath12 various quarter - harmonic patters emerge .
the pattern formation in thin layers of granular material was studied theoretically by several groups .
direct molecular dynamics simulations @xcite ( see also @xcite ) reproduced a majority of patterns observed in experiments and many features of the bifurcation diagram , although until now have not yielded oscillons and interfaces .
hydrodynamic and phenomenological models @xcite reproduced certain experimental features , however neither of them offered a systematic description of the whole rich variety of the observed phenomena . in refs .
@xcite we introduced the order parameter characterizing the complex amplitude of sub - harmonic oscillations .
the equations of motion following from the symmetry arguments and mass conservation reproduced essential phenomenology of patterns near the threshold of primary bifurcation : stripes , squares , and , localized objects , oscillons . in this paper we describe high acceleration patterns on the basis of the same order parameter model and
compare it with experimental observations .
our preliminary results were published earlier in refs .
@xcite . here
we show that at large amplitude of driving both hexagons and interfaces emerge .
we find morphological instability leading to the formation of `` decorated '' interfaces .
we study the motion of the interface under the influence of a small subharmonic component in the driving acceleration .
we also find a new localized structure , `` super - oscillon '' , which exists for high acceleration values .
we discuss possible mechanisms of saturation of the interface instability .
our experimental results demonstrate the existence of super - oscillons and bound states of super - oscillons and interfaces .
they also confirm theoretical predictions of the externally controlled interface motion .
the structure of the article is following . in sec .
i we introduce our phenomenological order - parameter model . in sec .
ii we develop the weakly - nonlinear theory for the flat period - doubled state of the vibrated layer . in sec .
iii we analyze the interface solution and study its stability with respect to transverse perturbations . in sec .
iv we discuss new localized solutions . in sec .
v we present the combined theoretical phase diagram of the model . in sec .
vi we demonstrate that in a certain limit our model can be reduced to the real swift - hohenberg equation .
this equation also possesses a similar variety of patterns including stripes , hexagons , stable and unstable interface solutions , as well as localized solutions . in sec .
vii we demonstrate that additional subharmonic driving results in drift of the interface with the velocity determined by the amplitude of the driving and the direction determined by the relative phase . in sec .
viii we present our experimental results .
it includes the study phase diagram and the effect of additional subharmonic driving .
section ix summarizes our conclusions . |
studies of nonlinear periodically driven systems are important to understanding the fundamental physics of many useful and interesting phenomena in applications of lasers @xcite , driven ratchets @xcite , hydrophilic particles on the surface of water waves @xcite , josephson junctions @xcite , etc . if bodies moving in periodically - driven systems are allowed to interact , they can display a wealth of interesting physical phenomena associated with complex systems @xcite .
studies of the interaction among oscillators , bodies , nodes , etc .
, in periodic systems , and of how bodies in such systems collectively react to environmental forcing , can be valuable in a variety of fields from neuroscience@xcite to driven josephson junction arrays @xcite .
interest in spatio - temporally periodic potentials ( stp ) began in 1951 , when kapitza published two papers @xcite on a planar pendulum with an oscillating suspension point , which is often referred to as the parametrically driven pendulum , or simply kapitza s pendulum .
the most interesting feature of this simple system is that under certain conditions , the pendulum stands stably in the inverted position .
the change in stability of the inverted position through the oscillation of the suspension point is an example of what is known as dynamic stabilization , in which an inherently unstable system can be stabilized by periodic forcing .
the motion of a single particle immersed in a stp potential may be solved analytically using a simple approximation .
for instance , for small @xmath0 we may approximate @xmath1 with a harmonic oscillator potential resulting in equations of motion that can then be solved using floquet theory @xcite .
linearization of the equation of motion in the limit of zero dissipation for one - dimensional motion allows the equation of motion to be expressed as the hill equation , @xmath2 . if @xmath3 , this equation reduces to the mathieu equation @xcite .
both the hill and mathieu equations have been studied extensively due to the interesting properties that they display and due to the many applications that can be associated with these equations , including the quantum pendulum @xcite , ion traps @xcite , and oscillations of a floating mass in a liquid @xcite . when the oscillations of @xmath4 are fast compared to the natural frequency of oscillations when @xmath4 is set to its maximum value ,
the method of averaging can be used @xcite .
the interesting physical behavior of kapita s stable inverted pendulum has enticed several researchers to study the nonlinear case both experimentally @xcite and numerically @xcite . for systems composed of particles in a stp potential ,
only a small number of publications have examined the nonlinear multiple particle dynamics accounting for the multi - particle interactions .
some examples of papers treating this subject include a study of the motion of hydrophobic / hydrophilic particles on the surface of faraday waves @xcite , multiple charged particles in an stp potential generated by an electric curtain @xcite , and multiple particles in a periodically forced straining flow @xcite .
these previous studies have considered very large numbers of particles and they have focused on the overall particle motion . in the current paper , we instead examine the dynamics of a relatively small number of particles in a stp potential using a dynamical systems point of view .
specifically , we seek to relate the nonlinear systems dynamics with multiple particles to the bifurcations and stability of single particles .
the current computational study examines a one - dimensional ( 1d ) system with multiple particles interacting through a repulsive electrostatic @xmath5 potential in an external stp potential field .
the stp potential is @xmath6 which produces equations of motion analogous to the parametrically driven pendulum in the horizontal plane .
the coefficient @xmath7 is the potential amplitude , and the distance coordinate @xmath0 and time coordinate @xmath8 are non - dimensionalized using the wavenumber @xmath9 and the stp driving frequency @xmath10 , respectively .
the driving force ( @xmath11 ) has the form of a standing wave with oscillation amplitude @xmath7 .
the dimensionless wavelength @xmath12 and the oscillation period @xmath13 are both equal to @xmath14 .
the system is assumed to be periodic over @xmath15 , where @xmath16 is an integer , so we can define the concentration @xmath17 as @xmath18 where @xmath19 is the number of particles in the simulation . for simplicity ,
all particles are assumed to carry the same charge and mass , where the non - dimensionalization is performed such that the dimensionless mass is equal to unity .
damping is proportional to particle velocity with a dimensionless damping coefficient @xmath20 . in the current paper , we focus on the effect of the parameter @xmath7 , and therefore maintain constant values of the other dimensionless parameters - the damping parameter and the dimensionless particle charge .
these latter two parameters are set equal to @xmath21 and @xmath22 throughout the paper .
these values for @xmath20 and @xmath23 are chosen because they are realistic for systems similar to those discussed in @xcite , after being dimensionalized .
the force on a particle located at @xmath24 imposed by a particle located at @xmath25 , denoted by @xmath26 , is calculated with periodic boundary conditions . to address the forces imposed by long range interactions
, we consider an infinite sequence of image systems using ewald summation method @xcite , giving @xmath27 where @xmath28 .
the sum in ( [ eqn : particle_force_sum ] ) is convergent and may be written as a polygamma function @xmath29 with series expansion @xmath30 using ( [ eqn : polygamma ] ) , we can express @xmath26 as @xmath31 the equation of motion for the @xmath32 particle is given by @xmath33 where the second term on the rhs is due to the imposed stp potential field and the third term on the rhs is the particle interactions .
an example of the system containing seven particles may be found in fig .
[ fig : sup_still ] ( multimedia view ) . and
@xmath34 , depicted as particles on the surface of a standing wave ( multimedia view ) .
[ fig : sup_still],width=8 ] for @xmath19 particles in an autonomous system , the degrees of freedom ( or dimension ) of the phase space is @xmath35 , where @xmath36 is the dimension of the physical system . in stp problems , there is an explicit time dependence in the potential and therefore the system is non - autonomous . non - autonomous systems may be transformed into autonomous form by introducing an extra degree of freedom , which in a first - order system is given by @xmath37 .
though this seems to be a trivial representation of time , this autonomous formulation is necessary when distinguishing types of bifurcations .
this augmented system thus has @xmath38 degrees of freedom , which constitutes the `` full phase space '' .
the standard choice for making poincar sections in driven systems is a time map taken at the system driving period .
time maps are stroboscopic views of a trajectory expressed as @xmath39 when @xmath16 is a positive integer .
a poincar section includes any point where a continuous trajectory transversely intersects a subspace of the full phase space @xcite .
time maps , as defined above , will produce poincar sections with dimension one less than the dimension of the full phase space . in a time map ,
the path of a particle is always transverse to the @xmath40 plane , and therefore a point of intersection of the trajectory with this plane is a convenient sub - space that satisfies the criterion necessary to be a poincar section .
it is well known that as a system approaches a bifurcation point , it may take longer for transients of the relevant quantity to die out or for the system to recover from an external perturbation @xcite .
this behavior is known as the critical slowing down phenomena .
most real systems are subject to some natural perturbations , and these perturbations can become particularly apparent near the bifurcation points . measuring the increase of the variance in a physical quantity can therefore be used as a method to predict the presence of a bifurcation point @xcite .
the model system considered in the current paper has no external perturbations , aside from computer round - off error . in the limit @xmath41 , the damped system
would be expected to settle into an attractor , but the finite time of real simulations ensures the presence of small fluctuations in `` residual '' transients . in other words ,
multiple particle systems have a `` large '' number of degrees of freedom , therefore some small trace of the initial transient behavior ( residual transients ) will most likely be detectable .
the amount of residual transients may be found in the kinetic energy fluctuations .
it is known that the kinetic energy fluctuations may contain some information about the `` effective number of degrees of freedom '' @xcite .
the more degrees of freedom , the more residual transients will be present .
this correspondence between the effective degrees of freedom and the kinetic energy fluctuations is what makes the kinetic energy fluctuations an interesting quantity to examine .
the square of the deviation of the particle kinetic energy is given by @xmath42 , @xmath43 and @xmath44 , @xmath45 denote the @xmath32 and @xmath46 particle velocities , respectively .
the average is calculated as @xmath47 .
the normalized squared deviation of the kinetic energy is given by @xmath48
the dynamics of a single particle immersed in the one - dimensional stp potential @xmath49 , given by ( 1 ) , are similar to the dynamics of the parametric pendulum . in this paper
, we only discuss dynamics for the first bifurcation sequence leading to the chaotic regime , even though there are many consecutive regimes of stable limit cycles bifurcating into chaotic trajectories . in fig .
[ fig : bif_sin_1d ] , the first bifurcation sequence is shown for an ensemble of initial conditions .
table [ table : bif ] lists the type of bifurcations , the critical values of @xmath7 at which each bifurcation occurs , and the period of the limit cycle following each bifurcation .
the table is truncated after the @xmath50 bifurcation due to numerical resolution limitations for distinguishing bifurcation onset in a small volume of the phase space . for @xmath51 , a particle will move toward and equilibrate at the antinodes of the potential @xmath52 ( i.e. , the maxima of @xmath53 ) . as @xmath7 is increased , the fixed points in the poincar section bifurcate in a supercritical flip bifurcation leading to a period-2 limit cycle for @xmath54 .
this transition is not a hopf bifurcation because the explicit time dependence in the equations of motion must be considered as a degree of freedom to the phase space .
consequently , what might appear as a fixed point in the @xmath40 phase space in a bifurcation diagram is actually a period-1 trajectory in the full phase space .
we prove this using floquet theory , by numerically calculating the stability multipliers . both of the two non - trivial stability multipliers have no imaginary component close to the bifurcation point . at the bifurcation , one stability multiplier becomes smaller than -1 while the other remains close to zero , indicating a period - doubling supercritical flip bifurcation .
this stability multiplier passing through -1 is shown in fig .
[ fig : n1_fluctuations]a , where it is denoted with a roman numeral i. the values of @xmath7 for which the first six bifurcations occur , shown in table [ table : bif ] , indicate a period - doubling cascade route to chaos .
the computed values yield a feigenbaum constant of 4.00 with an upper error bound of 6.00 and a lower bound of 2.89 . the accepted value of 4.669 for period - doubling bifurcations
@xcite is within the error bounds .
the feigenbaum constant @xmath55 is evaluated with @xmath56 where @xmath57 is the @xmath58 critical value of @xmath7 for which a period - doubling bifurcation occurs .
the two lines coming out of the chaotic region in fig . [ fig : bif_sin_1d ] are each attractors representing stable propagating trajectories , one with a positive velocity and one with a negative velocity .
these propagating trajectories travel across @xmath12 once per period of the driving potential field .
.bifurcations [ cols="^,^,^,^",options="header " , ] [ table : bif ] bins ) of the final poincar section of 1830 trajectories with different initial conditions for 300 different values of @xmath7 .
the gray scale ( color online ) represents the base 10 logarithm of the number of particles in a bin .
the roman numerals are listed here for comparison with @xmath59 shown in fig .
[ fig : n1_fluctuations ] .
[ fig : bif_sin_1d],width=8 ] before going to the multi - particle case , it is informative to compare the bifurcation diagram ( fig .
[ fig : bif_sin_1d ] ) to the calculation of @xmath59 for a single particle , which is shown in fig .
[ fig : n1_fluctuations ] .
we also show a floquet stability analysis of the fixed point at @xmath60 through @xmath61 for comparison .
floquet stability analysis is a powerful tool in analyzing bifurcations , but it is not easily applied to multiple particle systems .
it has been applied to coupled kapitza pendulums by @xcite . for a description of single particle stability analyses ,
we refer the reader to @xcite and @xcite , which are both studies of similar systems and use the floquet technique to study bifurcations . in fig .
[ fig : n1_fluctuations]a , the real and imaginary components of the stability multiplier that causes the bifurcation ( one of the two complex floquet stability multipliers @xmath62 and @xmath63 ) are plotted as @xmath7 is increased through @xmath61 . in fig .
[ fig : n1_fluctuations]b , @xmath59 is plotted as @xmath7 is increased through the full range shown in the bifurcation diagram in fig .
[ fig : bif_sin_1d ] . in fig .
[ fig : bif_sin_1d ] and fig .
[ fig : n1_fluctuations ] , the key regions associated with different system behaviors have been identified using roman numerals . for small values of @xmath7 ,
[ fig : n1_fluctuations]b shows a wide range of scattered points . however , the particle exhibits very little motion within this range of small @xmath7 values .
as @xmath7 is increased , there exists a peak in the fluctuations near @xmath61 , which is a consequence of the critical slowing down phenomenon ( region i ) .
as @xmath7 is increased past @xmath61 , the fluctuation amplitude is relatively constant until @xmath7 approaches @xmath64 , where a kink is observed ( region ii ) .
when @xmath7 is in the chaotic and near - chaotic regimes ( regions iii , iv , v ) , the fluctuations increase in amplitude and are irregular , as shown in the inset in the figure .
the two regions where the fluctuation amplitude decreases markedly in this inset correspond to the two periodic windows seen in fig .
[ fig : bif_sin_1d ] . at the end of the chaotic regime
, there is a discontinuous jump in the fluctuations to a comparatively small and relatively constant value ( region vi ) .
this last section shows the transition to propagating trajectories , and we will see that this feature is present in all cases where this transition occurs . under closer inspection , region vi overlaps with region
v because , just as in the bifurcation diagram , the propagating trajectories exist simultaneously with the chaotic regime for a small range of @xmath7 .
fixed point as a function of the potential oscillation amplitude as it is increased through the first bifurcation point .
( b ) kinetic energy fluctuations .
roman numerals for comparison with the bifurcation diagram in fig .
[ fig : bif_sin_1d ] [ fig : n1_fluctuations],width=321 ] the bifurcation diagrams for multiple interacting particles , with @xmath65 , are shown in fig . [
fig : multi_bifs ] .
the increased degrees of freedom that occur for @xmath66 make it difficult to investigate an ensemble of initial conditions that exhaustively fill the phase space .
we use random positions distributed with even probability across @xmath0 ( with @xmath67 ) as initial conditions for each run to explore a set of possible initial conditions .
the bifurcation plots are made by taking the last poincar section after 150 driving cycles of a simulation , projecting it onto the position axis , and then plotting the positions against the value of @xmath7 used in that simulation .
for very small @xmath7 , the final poincar sections are scattered because , for these values , transients die out very slowly . for larger @xmath7 ,
there are clearly defined points in the poincar sections that denote limit cycles in the full phase space . for the rest of the paper , the stable limit cycles in the poincar sections
are referred to more generally as attractors . at first glance ,
[ fig : multi_bifs ] appears to indicate a larger number of particles for odd values of @xmath19 than it does for even values of @xmath19 .
as @xmath7 is further increased , a bifurcation occurs for all the cases shown , although it is difficult to see in fig .
[ fig : multi_bifs](e ) . for other values of @xmath7 , the diagrams in fig .
[ fig : multi_bifs ] appear as scattered points , implying either chaotic motion , high - period trajectories , or motion on a torus ( after a neimark bifurcation ) .
past this scattered regime , the system collapses into a new stable regime that is qualitatively similar to the propagating trajectories that occur after the chaotic regime in the single particle case . the discrepancy between the number of possible attractors for odd and even values of @xmath19
can be explained by considering the relationship between the number of particles and the number of antinodes in one period of the potential . for the trajectory @xmath68 , the number of attractors that are observed for values of @xmath7 before the first bifurcation is equal to the number of particles when @xmath19 is even and twice the number of particles when @xmath19 is odd . in other words , there is one possible final state configuration of the whole system when @xmath19 is even and two possible final state configurations of the system when @xmath19 is odd . to see why this occurs
, we examine in detail the cases of three and four particles per period .
[ fig : even_odd](a)(b ) and ( c)(d ) show cartoons of the final particle configurations for cases @xmath69 and @xmath70 , respectively . in these cartoons ,
the particle position in the periodic domain is drawn as an angle , so that motion of the particle in the @xmath0 domain corresponds in the figure to rotation about a circle .
the intersection points of the circle with a horizontal bisection line occurs at the antinodes of the potential @xmath49 . in this figure , at least one particle
can always be found at the antinode of @xmath49 . from the single particle case ,
we know the antinodes can act as attractors , where individual particles remain motionless in the @xmath40 phase plane . in the case of four particles , shown in fig .
[ fig : even_odd](a ) , two particles may occupy the antinodes of @xmath49 and the other two particles oscillate about the nodes of @xmath49 .
one might imagine that a rotation of this configuration by @xmath71 ( shown in fig .
[ fig : even_odd](b ) ) might be a fixed point in the poincar section of the full phase space , and indeed it is , but it is not a stable configuration and , unless perfectly configured , it will collapse into the configuration shown in fig .
[ fig : even_odd](a ) . for three particles ,
one particle sits at either antinode and the two remaining particles compete over the other antinode , as shown in fig .
[ fig : even_odd](c ) and ( d ) .
which antinode a particle is attracted to depends on the initial conditions . in both the four and three particle cases , the particles at the antinodes are stationary .
stable configuration which is an asymptotically stable fixed point in the poincar sections of the full phase space .
part ( b ) is an @xmath69 unstable configuration which is the unstable fixed point in the poincar sections of the full phase space .
parts ( c ) and ( d ) show two different possible stable configurations for @xmath70 , both of which are stable fixed points in the poincar sections of the full phase space .
[ fig : even_odd],width=8 ] drawing lines between the adjacent average particle positions in the circular topology creates a regular convex polygon inscribing the circle . from our description of the particle behavior
above , at least one vertex of the polygon must be at an antinode . for a regular polygon inscribing the circle with an even number of vertices
, each antinode may touch a vertex and it is symmetric under all rotations obeying this rule . for a polygon with an odd number of vertices , however ,
only one vertex can occupy an antinode so that rotations by @xmath72 flip the symmetry about a line vertically bisecting the circle .
there are always two unique possible stable configurations when @xmath19 is odd , but only one when @xmath19 is even .
this picture changes for even values of @xmath19 when @xmath73 . for @xmath73 ,
the stable configuration no longer occurs for a pair of particles located at each antinode , but rather for pairs oscillating on either side of the antinodes much like what is shown in fig .
[ fig : even_odd](b ) but with an extra particle on the top and bottom of the circle . in fig .
[ fig : multi_bif_vals ] , the value of @xmath7 at which the first bifurcation occurs is plotted as a function of @xmath19 , with separate lines for @xmath19 even ( red squares ) and @xmath19 odd ( blue triangles ) . a striking characteristic of fig .
[ fig : multi_bif_vals ] is that between @xmath74 and @xmath75 , there is a cross - over point at which the even @xmath19 line jumps upward and crosses through the odd @xmath19 line .
this sudden jump in the even @xmath19 line between @xmath74 and @xmath75 occurs due to a change in type of first bifurcation .
the first bifurcation when @xmath74 , as well as the first bifurcations for all lower even values of @xmath19 , are neimark bifurcations ( a.k.a bifurcation to a torus ) in which @xmath19 stability multipliers cross the unit circle with nonzero imaginary components .
half of those stability multipliers ( @xmath76 ) that cross the unit circle have positive imaginary components and the other half have the complementary negative imaginary components . for @xmath75 and all higher even values of @xmath19 , the first bifurcation becomes a cyclic fold bifurcation , although the bifurcations for odd values of @xmath19 remain supercritical flip bifurcations .
we can qualitatively understand the transition which occurs when @xmath19 is changed from six to eight in fig . [ fig : multi_bif_vals ] by observing how the bifurcation diagrams , and therefore the stable attractors , depend on the particle number . in fig .
[ fig : multi_bifs](e ) , after the first bifurcation , the particle motions are quenched by the presence of other attractors .
for the sake of discussion we distinguish the two competing sets of attractors based on whether they are found at the far left or far right edges of the diagrams respectively . comparing the cases of @xmath77 , @xmath69 and @xmath74 in fig .
[ fig : multi_bifs ] we see similar rightmost attractors that appear abruptly at different values of @xmath7 in each case . as @xmath19 ( even )
increases , the rightmost attractors increasingly impinge on the leftmost attractors .
this impingement is responsible for the crossing of the lines in fig .
[ fig : multi_bif_vals ] .
when @xmath19 increases from six to eight , the rightmost attractors extinguishes the left most attractors and becomes the first available attractors for @xmath78 ( even ) .
these new first attractors , previously the rightmost , represents a fundamentally different type of limit cycle in the full phase space than what was previously first available .
therefore when this attractor first bifurcates , it falls outside of the original progression , causing the jump in @xmath61 as @xmath19 is changed from six to eight shown in fig .
[ fig : multi_bif_vals ] . for all of the cases shown in fig .
[ fig : multi_bifs ] , the system eventually collapses back into clearly defined attractors which have the form of `` propagating '' trajectories , as was also observed for the single particle case .
these attractors display non - zero net particle motion of one particle when @xmath19 is odd , but no net motion when @xmath19 is even .
the particles travel in either the @xmath79 direction before a collision - like event .
after the `` collision '' , they travel in the opposite direction having exchanged some kinetic energy with the particle with which the collision occurred .
when @xmath19 is odd the transport of one particle occurs either in the @xmath79 direction depending on the initial conditions .
there is no possible counter - propagating particle pair for one of the particles with @xmath19 odd . in fig .
[ fig : multi_flucts ] , the squared fractional deviation of the kinetic energy @xmath59 is plotted for all of the bifurcation diagrams shown in fig .
[ fig : multi_bifs ] . these plots
all exhibit a discontinuity at the point corresponding to transition to a state with the propagating trajectories .
the discontinuity is not as clear in fig .
[ fig : multi_flucts](e ) because ( as can be observed in the corresponding bifurcation diagram fig .
[ fig : multi_bifs](e ) , the propagating trajectory begins before the first bifurcation . in fig .
[ fig : multi_flucts](e ) , the curve starting below and crossing at @xmath80 indicates the values of @xmath59 for the propagating trajectory . when the periodicity of the system is increased ( @xmath81 ) ,
two distinctly different possibilities exist .
one possibility is an integer concentration , over a larger periodic system ( e.g. @xmath82 and @xmath69 : @xmath83 ) .
the second possibility is a fractional concentration ( e.g. @xmath82 and @xmath70 : @xmath84 ) . when considering long - range particle interactions , it is reasonable to assume that increasing the system size might change the dynamics even if the concentration is held fixed . the first possibility above results in system dynamics very similar to those discussed in the current paper , whereas the second possibility would result in a completely different symmetry of the system having very different dynamics .
the effect of period number @xmath16 on the system dynamics was examined in the current study by running simulations of equal concentration over larger system sizes ( i.e. , large values of @xmath16 ) .
the system dynamics in these larger systems is observed to be qualitatively the same as for the smaller system sizes reported in the paper , although the exact values of @xmath7 for which bifurcations occur is observed to change slightly .
we find that for concentrations larger than two , the effect of scaling the system size while maintaining the concentration is negligible even when comparing critical @xmath7 values .
we investigate the dynamics of multiple particles with long - range interactions in a stp system by examining poincar sections and fluctuations of the kinetic energy ( @xmath59 ) for different numbers of particles .
our results are fundamentally interesting because of their importance in understanding complexity in time - dependent systems .
the possible dynamics that exist in a wide range of different system configurations make the problem challenging , but even in the small area of the parameter space discussed in this paper we have found a variety of interesting dynamics . for instance , it is shown that the particle number @xmath19 influences the stability and the number of possible final states in a system having integer concentrations .
the possible limit cycles of the system are shown to be sensitive to whether @xmath19 is even or odd , and the influence of the particle number on the type of bifurcation is discussed .
the squared fractional deviation kinetic energy is examined as a function of the potential amplitude ( @xmath85 ) , and it is found to exhibit interesting features at and near transition points . in particular , discontinuities in @xmath86 and @xmath59 mark transitions between oscillatory and propagating modes , respectively .
the measure @xmath59 may be useful for future experimental investigations of these systems .
our work has demonstrated interesting and complex behaviors of multiple particles with the coulomb interaction in a stp potential . in particular , the dynamics of the system is sensitive to particle concentration and the dynamics can be described by the squared fractional deviation of the particle kinetic energies .
the latter is particularly valuable for studying bifurcations in real systems .
for example , in the aforementioned studies of the motions of hydrophobic / hydrophilic particles on the surface of faraday waves , the particles will interact due to capillary forces caused by their distortion of the local surface of the water , rather than through the coulomb interaction , which leads to particle clustering @xcite
. it may be convenient to study this type of behavior using the squared fractional deviation of the systems kinetic energy because this measure will decrease in the event of clustering as it measures the effective number of degrees of freedom @xcite .
studies of multiple charged particles in a standing - wave electric curtain and in acoustic waves are also important areas of research for applications of dust - particle mitigation , e.g. , from a solar panel @xcite . charged particles interacting in standing - wave electric curtains and standing - wave acoustic fields exhibit complicated dynamics that may be illuminated by studying the squared fractional deviation of the particle kinetic energy . for example , in @xcite it was observed that for charged particles in a standing - wave acoustic field , relative motion of smaller particles is faster than that of larger particles , so that the large particles act as collectors within some agglomeration volume .
any small particles present in the agglomeration volume are likely to aggregate to a larger particle , and this aggregation is desirable for applications such as cleaning particles from surfaces
. a sweep of the acoustic driving parameters to find the configuration for which maximal aggregations occurs could clearly be found and described in terms of the minimal squared fractional deviation of the particle kinetic energies , again due its measure of th effective number of degrees of freedom . in general , we hope our work may stimulate further research of stp systems with interacting particles and shed some light on their complicated and exciting dynamics .
this work was supported by nasa space grant consortium under grant numbers nnx10ak67h , nnx08az07a , and nnx13ad40a .
we also want to acknowledge the vermont advanced computing core which is supported by nasa ( nnx 06ac88 g ) , at the university of vermont for providing the high performance computing resources used for the work in this paper .
41ifxundefined [ 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 * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/revmodphys.78.1213 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.78.2632 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop _ _ , ed .
( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1017/s0022112008005326 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) link:\doibase 10.1038/nature08227 [ * * , ( ) ] @noop * * , ( ) @noop _ _ , ed .
( , , ) link:\doibase 10.1103/physreve.58.7231 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4826267 [ * * , ( ) ] @noop * * , ( ) | the spatio - temporally periodic ( stp ) potential is interesting in physics due to the intimate coupling between its time and spatial components . in this paper
we begin with a brief discussion of the dynamical behaviors of a single particle in a stp potential and then examine the dynamics of multiple particles interacting in a stp potential via the electric coulomb potential . for the multiple particles case ,
we focus on the occurrence of bifurcations when the amplitude of the stp potential varies .
it is found that the particle concentration of the system plays an important role ; the type of bifurcations that occur and the number of attractors present in the poincar sections depend on whether the number of particles in the simulation is even or odd .
in addition to the nonlinear dynamical approach we also discuss dependence of the squared fractional deviation of particles kinetic energy of the multiple particle system on the amplitude of the stp potential which can be used to elucidate certain transitions of states ; this approach is simple and useful particularly for experimental studies of complicated interacting systems . |
it is an elementary fact that the euler characteristic is multiplicative in fiber bundles . according to a classical result of chern , hirzebruch and serre @xcite the same holds for the signature ,
provided that the fundamental group of the base acts trivially on the cohomology of the fiber .
atiyah @xcite and , independently , kodaira @xcite showed that this assumption on the monodromy is necessary , by exhibiting surface bundles over surfaces with non - zero signature . in the case of bundles
whose fiber is a sphere or torus , it is easy to see that the signature must vanish
. therefore , only the signature of surface bundles of higher genus is interesting . for a closed oriented surface @xmath0 of genus @xmath1
, teichmller theory implies that the identity component of the group of orientation - preserving diffeomorphisms is contractible .
it follows that every oriented bundle with fiber @xmath0 over a base @xmath2 is determined by ( the conjugacy class of ) its monodromy representation @xmath3 where @xmath4 is the mapping class group of @xmath0 , consisting of isotopy classes of orientation - preserving diffeomorphisms .
if the base @xmath2 is also @xmath5-dimensional , then the signature of the total space @xmath6 is four times the first chern number of the flat symplectic bundle obtained by composing @xmath7 with the action of @xmath4 on the homology of @xmath0 , see @xcite . in particular
, the signature vanishes if the genus of @xmath2 is @xmath8 or @xmath9 .
the signature also vanishes for all bundles with fiber genus @xmath5 , because of igusa s theorem @xmath10 .
thus , we may assume that the fiber genus @xmath11 is @xmath12 . combining the work of meyer @xcite and of harer @xcite , one sees that the signature of the total space @xmath6 is given by the homology class of @xmath13 $ ] in the homology of @xmath4 .
more precisely , the second integral homology of the mapping class group is infinite cyclic , generated by the meyer signature cocycle corresponding to the signature of the total space .
this means that determining the maximal signature of a surface bundle with given fiber and base genus is equivalent to calculating the gromov - thurston norm in the second homology of the mapping class group .
this is essentially problem 2.18 in kirby s list @xcite . to address this problem ,
consider the function @xmath14 using seiberg - witten gauge theory , the first nontrivial lower bound for this function was proved in @xcite : @xmath15 the only systematic upper bound for this function was proved in @xcite , where it was shown that for every fiber genus @xmath16 there is a surface bundle over a surface of genus @xmath17 with signature @xmath18 . pulling back to coverings of the base
, one has @xmath19 a non - explicit improvement of in some cases was given in @xcite . in this paper
we obtain new upper bounds for the function @xmath20 by constructing examples of surface bundles in which the base genus is comparatively small .
we found these examples by first constructing lefschetz fibrations with singular fibers corresponding to expressions of products of dehn twists as products of commutators , and then taking differences of lefschetz fibrations with the same singular fibers to obtain smooth surface bundles .
we have chosen to present the examples in the way we originally found them , although it would have been possible , after the fact , to eliminate the lefschetz fibrations from the presentation and write down the monodromy representations of the surface bundles directly .
we believe that the subtraction of lefschetz fibrations presented in section [ second ] , also used in @xcite , is of interest in its own right , in addition to being a useful stepping stone in the construction of surface bundles .
our first main theorem is the following improvement of : [ main1 ] for every @xmath16 there is a surface bundle of genus @xmath11 over the surface of genus @xmath21 with signature @xmath18 .
in particular , @xmath22 .
notice that all these examples over @xmath23 have the same signature . by considering sections of our fibrations we can construct surface bundles with fiber genus @xmath11 over @xmath23 for which the signature grows linearly with @xmath11 .
more precisely , we have : [ main2 ] for every @xmath24 there are surface bundles of fiber genus @xmath11 over the surface of genus @xmath21 with signature at least @xmath25 .
this result allows us to prove upper bounds for @xmath20 which have the same shape as the lower bound , in that the fiber genus appears in the denominator .
we only formulate these upper bounds in the asymptotic case , when @xmath26 becomes large .
it is easy to see that the limit @xmath27 exists and is finite for all @xmath11 .
the inequality implies @xmath28 . using our new examples
, we will prove : [ main3 ] if @xmath16 is odd , then @xmath29 . if @xmath30 is even , then @xmath31 .
this paper is organized as follows . in section [ second ]
we review the basic facts about lefschetz fibrations and describe the subtraction operation " for them in detail .
section [ third ] is devoted to the proof of various identities in the mapping class group expressing certain products of dehn twists as products of commutators . in section [ fourth ]
we calculate the signatures of the corresponding lefschetz fibrations using the meyer signature cocycle @xcite . in the last section
we give the proofs of the theorems stated above .
we begin by recalling the definition and basic properties of lefschetz fibrations . more details can be found in @xcite .
let @xmath6 be a compact oriented @xmath18-manifold , and @xmath2 a compact oriented surface . a smooth map @xmath32 is called a lefschetz fibration if it is surjective and if for each critical point @xmath33 there are local complex coordinates @xmath34 on @xmath6 around @xmath35 and @xmath36 on @xmath2 around @xmath37 compatible with the orientations and such that @xmath38 .
it follows that a lefschetz fibration has at most finitely many critical points @xmath39 .
it is easy to see that by a slight perturbation one can achieve that @xmath40 is injective on its critical set @xmath41 .
we will always assume that this additional property holds .
the _ genus _ of @xmath40 is defined to be the genus of a regular fiber . if @xmath2 is connected , the genus is well - defined .
even when @xmath2 is not connected , we will assume that all regular fibers have the same genus .
fibers of @xmath40 passing through elements of @xmath42 are _ singular _ fibers .
notice that if @xmath43 denotes an open tubular neighborhood of the set of critical values @xmath44 , then the restriction of @xmath40 to @xmath45 is a smooth surface bundle over the surface - with - boundary @xmath46 .
a singular fiber @xmath47 , where @xmath48 , can be described by its _ monodromy _ , which is an element in the mapping class group @xmath4 . to determine this element , however , we need to fix a base point @xmath49 , an identification of @xmath50 with the closed oriented surface @xmath0 of genus @xmath11 , and a loop @xmath51 in @xmath2 based at @xmath52 which has linking number @xmath53 with @xmath54 .
the restriction of @xmath40 to the preimage of this loop is an @xmath0-bundle over @xmath55 which can be described by a single element @xmath56 .
in fact , by performing this procedure for all loops in @xmath57 we get a map @xmath58 .
it can be shown that @xmath59 is a right - handed dehn twist along a simple closed curve @xmath60 called the _ vanishing cycle _ corresponding to the singular fiber @xmath47 .
notice that , even after fixing @xmath61 and the identification @xmath62 , both @xmath59 and @xmath63 depend on the chosen loop @xmath51 .
it is convenient to fix the following conventions .
suppose that all @xmath54 lie on the boundary of a disk @xmath64 centered at @xmath61 .
let @xmath65 denote the radial curve in @xmath66 connecting @xmath52 with @xmath54 and form @xmath51 as the boundary of an appropriate neighborhood of @xmath65 , cf .
figure [ loops ] . by fixing a generating system @xmath67 of @xmath68
, the map @xmath69 can be encoded by a sequence @xmath70 , where @xmath71 and @xmath72 tell us the monodromy of the fibration along @xmath73 and @xmath74 .
it is easy to see that these elements satisfy the relation @xmath75\cdot\pi _ { i=1}^s t_i = 1 $ ] in the mapping class group .
conversely , for @xmath1 a word of the form @xmath75\cdot\pi _ { i=1}^s t_{i}$ ] representing 1 in @xmath4 ( with @xmath59 being right - handed dehn twists ) gives rise to a lefschetz fibration of genus @xmath11 over a surface @xmath2 of genus @xmath76 . as we noted already ,
the vanishing cycles and the corresponding dehn twists depend on the chosen loops @xmath51 .
it is easy to see that a cyclic permutation of the indices can be compensated by changing the identification @xmath77 , so the resulting lefschetz fibration remains the same .
one can also change the word by _
elementary transformations _ without changing the lefschetz fibration , i. e. the path @xmath51 can be changed as indicated by figure [ trafo ] . by applying an elementary transformation as shown by the figure
, we replace @xmath59 and @xmath78 by @xmath78 and @xmath79 .
( notice that this change has no effect on the product of these elements . )
the new vanishing cycles are easy to determine since for any mapping class @xmath76 the conjugate @xmath80 of the dehn twist @xmath81 is simply the dehn twist @xmath82 .
it is not hard to prove that if two words give rise to equivalent fibrations then the words can be transformed into each other by applying combinations of the two operations just described .
a singular fiber @xmath47 is _ nonseparating _ if the corresponding vanishing cycle @xmath63 is nonseparating , equivalently its homology class is nonzero in @xmath83 . if @xmath63 is a separating curve , equivalently its homology class is zero , then @xmath47 is called _
separating_. a vanishing cycle @xmath63 and the corresponding singular fiber are of _ type _
@xmath8 if @xmath63 is nonseparating ; they are of _ type _
@xmath84\ } $ ] if the vanishing cycle separates the surface of genus @xmath11 into two components with genera @xmath85 and @xmath86 .
although the vanishing cycle depends on the chosen path @xmath51 , its type is independent of this choice . from the classification of surfaces
, one can prove that for two simple closed curves of the same type there exists a diffeomorphism of the ambient surface mapping one into the other .
this implies that two singular fibers of the same type have fiber- and orientation - preservingly diffeomorphic tubular neighborhoods .
the combinatorial data of a lefschetz fibration can be encoded as follows : [ d : comb ] the vector @xmath87})\in { { \mathbb { z}}}^{[\frac{h}{2}]+1}$ ] associated to the lefschetz fibration @xmath32 is constructed by taking @xmath88 to be the number of singular fibers of type @xmath85 ( @xmath89 $ ] ) .
following @xcite we say that two fibrations @xmath90 ( @xmath91 ) are combinatorially equivalent if @xmath92 .
the construction we use to produce new examples of surface bundles is a procedure for taking the difference of two combinatorially equivalent lefschetz fibrations .
if @xmath93 and @xmath94 are combinatorially equivalent as in definition [ d : comb ] , with critical values @xmath95 and @xmath96 respectively , then a surface bundle @xmath97 can be constructed in the following way : order the @xmath98 s and @xmath99 s so that singular fibers with coinciding lower index have the same type .
fix an orientation- and fiber - preserving diffeomorphism @xmath100 between the boundaries of tubular neighborhoods of fibers with lower index @xmath101 ( @xmath102 ) .
the union of these maps will be denoted by @xmath103 .
now glue @xmath104 to @xmath105 using @xmath103 .
notice that by reversing the orientation on @xmath94 , the map @xmath103 becomes orientation - reversing , hence the resulting manifold @xmath106 inherits a natural orientation . since @xmath107 is fiber - preserving , @xmath108 admits a smooth fibration with fibers of genus @xmath11 over a compact surface @xmath2 which we will denote by @xmath109 . [
l : sub ] if @xmath110 and @xmath111 are combinatorially equivalent lefschetz fibrations with @xmath112 singular fibers , then @xmath106 is a smooth surface bundle with signature @xmath113 over the surface @xmath114 with euler characteristic @xmath115 . by construction
, @xmath108 is an oriented smooth surface bundle over a surface @xmath2 .
the claim about the euler characteristic of the base is obvious .
the claim about the signature is an instance of novikov additivity .
note that we did not assume @xmath93 and @xmath94 to be connected .
this means that basepoints have to be chosen in each component of @xmath116 , and the vectors of combinatorial data have to be summed over all components to determine combinatorial equivalence .
if @xmath93 or @xmath94 happens to be connected , then so is @xmath117 .
the main property we used in the above construction is that the manifolds @xmath118 and @xmath119 have diffeomorphic boundaries and , after reversing the orientation of one of them , this diffeomorphism can be chosen to be fiber - preserving and orientation - reversing .
a variation of this construction goes as follows : suppose that partitions of the critical values @xmath120 and @xmath121 are given together with a system of disjoint disks @xmath122 ( @xmath123 and @xmath124 ) such that each disk contains exactly one equivalence class of the partitions
. suppose furthermore that we can pair up these disks in a way that the surface bundles @xmath125 are isomorphic to @xmath126 for all @xmath127 .
then @xmath94 can be subtracted from @xmath93 along the disks @xmath128 , i. e. the manifold @xmath129 admits the structure of a surface bundle .
the signature @xmath130 is again given by @xmath131 , while the euler characteristic of the base is equal to @xmath132 .
the definition of @xmath117 is a special case of this latter construction , corresponding to the situation when each equivalence class of the partition consists of a unique critical value . by considering a partition with larger equivalence classes we get smaller
@xmath133 which results in a smaller genus for the base .
notice that in the special case of @xmath117 the assumption @xmath134 can be easily checked by determining the type of the singular fibers over the disks . in general , however , the types of the singular fibers over the disks do not specify the diffeomorphism type of the above fibration , since fibers of the same type can be glued together in many different ways resulting various fibrations over @xmath135 .
there is a generalisation of lefschetz fibrations , called achiral lefschetz fibrations , where one allows singular fibers whose monodromies are left - handed dehn twists , cf .
@xcite . keeping track of the chirality of the singular fibers , it is clear that the subtraction operation described above generalises to the category of achiral lefschetz fibrations .
we conclude this section by discussing the relation between the word specifying a lefschetz fibration and sections of the fibration .
suppose that @xmath136 is a given lefschetz fibration .
a map @xmath137 is called a _ section _ if @xmath138 . the _ self - intersection _ ( or _ square _ ) of the section @xmath139 is simply the self - intersection number of the homology class @xmath140\in h_2(x ; { { \mathbb { z}}})$ ] . in the following @xmath141
denotes the mapping class group of the closed oriented surface of genus @xmath11 with one marked point and @xmath142 denotes the mapping class group with respect to one boundary component ( fixed pointwise ) .
notice that by collapsing the boundary circle to a point we get a natural surjection @xmath143 with kernel the subgroup generated by the dehn twist @xmath144 along a curve isotopic to the boundary circle ( cf .
@xcite , for example ) .
moreover , by forgetting the marked point we have an obvious map @xmath145 .
the following two well - known facts show how the existence of a section ( and its square ) is reflected in the monodromy representation of a lefschetz fibration .
suppose that the monodromy representation of @xmath32 is given by the relator @xmath146\cdot\pi _ { i=1}^st_i$ ] representing 1 in @xmath4 .
the fibration admits a section if and only if @xmath59 and @xmath147 admit lifts @xmath148 such that @xmath149 \cdot \pi _ { i=1}^s { \tilde { t}}_i$ ] represents @xmath9 in @xmath141 .
a section of @xmath136 is given once such a lift is fixed .
suppose now that a fibration @xmath32 with a section is given , so a lift @xmath149\cdot \pi _ { i=1}^s { \tilde { t}}_i$ ] of @xmath150\cdot \pi _ { i=1}^s t_i $ ] is fixed .
take a lift @xmath151 of @xmath152 ( and @xmath153 of @xmath154 ) in @xmath142 and consider @xmath155\cdot\pi _
{ i=1}^s t_i ' \in \gamma_h ^1 $ ] . by the discussion above
, this product is in @xmath156 , hence it is equal to @xmath157 for some @xmath158 .
@xcite ) the self - intersection number of the section given by the above lift is equal to @xmath159 .
[ secc ] next we would like to show that after subtracting lefschetz fibrations with sections , under favourable circumstances the resulting fibration admits a section whose self - intersection number is equal to the difference of the self - intersection numbers of the sections of the individual fibrations .
for this , suppose that two fibrations @xmath160 ( @xmath91 ) are given by their monodromy representations @xmath161\cdot\pi t_i $ ] and @xmath162\cdot\pi s_j $ ] respectively .
suppose furthermore that the disks @xmath163 along which the subtraction operation will be performed contain the singular fibers corresponding to the dehn twists @xmath164 ( and @xmath165 resp . ) .
[ osszerag ] if the lifts @xmath166 giving rise to the sections coincide with @xmath167 ( @xmath168 ) in @xmath141 , then the difference of the two fibrations admits a section .
the self - intersection of this section is given by the difference of the self - intersection of the individual pieces .
the assumption shows that there is a diffeomorphism @xmath169 mapping the sections into each other .
now the statement is obvious notice that the self - intersection is the difference of the two self - intersections since in the subtracting operation we change the orientation of @xmath94 .
the assumption on coinciding lifts can not be relaxed , as the following example shows .
take two copies of the trivial bundle @xmath170 , fix two sections in each and blow up one section in each copy . in this way we get two lefschetz fibrations ( each with a single singular fiber ) for which the subtraction operation ( along the singular fibers ) applies and gives @xmath171 back .
the section blown up , however , can be glued only to the section in the other copy also blown up , because otherwise we would find a homology class in @xmath172 with odd square , which is clearly impossible .
surface bundles with sections of self - intersection zero can be summed along their sections by performing a fiberwise connected sum .
this is an instance of gompf s symplectic sum operation , but for our purposes the symplectic aspect is irrelevant . [ secrag ]
if @xmath173 with @xmath91 are two surface bundles with fiber genera @xmath174 over the same base surface and both fibrations admit sections with self - intersection zero , then there is a surface bundle over @xmath175 with fiber genus @xmath176 and signature @xmath177 .
the signature is additive when summing along embedded surfaces of self - intersection zero .
let @xmath178 be an oriented surface of genus @xmath11 with @xmath112 marked points and @xmath179 boundary components .
the _ mapping class group _ @xmath180 of @xmath0 consists of the isotopy classes of orientation - preserving diffeomorphisms of @xmath0 which are the identity on each boundary component and preserve the set of marked points .
the isotopies are not allowed to permute marked points or to rotate boundary components .
the groups @xmath181 , @xmath182 and @xmath183 will be denoted by @xmath184 , @xmath185 and @xmath4 , respectively .
we say that two simple closed curves @xmath186 and @xmath187 on @xmath0 are _ topologically equivalent _ if there exists a diffeomorphism of @xmath0 mapping @xmath186 to @xmath187 . for a group @xmath188 and @xmath189 ,
the commutator @xmath190 $ ] denotes the element @xmath191 and @xmath192 denotes the conjugate @xmath193 .
it follows easily from the definition of a dehn twist that if @xmath186 is a simple closed curve on @xmath0 and @xmath40 is an orientation - preserving diffeomorphism of @xmath0 , then @xmath194 in @xmath180 .
if @xmath186 and @xmath187 are two topologically equivalent simple closed curves on @xmath0 , then @xmath195 is a commutator .
more precisely , if @xmath196 then @xmath197 $ ] .
let @xmath186 and @xmath187 be two simple closed curves on @xmath0 .
if @xmath186 is disjoint from @xmath187 , then the supports of the dehn twists @xmath198 and @xmath199 can be chosen to be disjoint .
hence , @xmath198 commutes with @xmath199 . if @xmath186 intersects @xmath187 transversely at one point , then it is easy to see that @xmath200 .
it follows that @xmath198 and @xmath199 satisfy the _
braid relation _
@xmath201 .
the following two relations in the mapping class group are also well - known .
the first one is the _ lantern relation _ ( cf .
let @xmath202 be a sphere with four boundary components @xmath203 , @xmath204 , @xmath205 and @xmath206 .
suppose that @xmath202 is embedded in @xmath0 .
then there are three simple closed curves @xmath207 , @xmath208 , @xmath209 on @xmath202 , as illustrated in figure [ lantern ] ( i ) , which satisfy the lantern relation @xmath210 the second relation is the _ two - holed torus relation _ or _
chain relation_. let @xmath211 , @xmath212 , @xmath213 be three nonseparating simple closed curves on @xmath0 such that @xmath212 intersects @xmath211 and @xmath213 transversely only once , @xmath211 is disjoint from @xmath213 and @xmath214 does not disconnect @xmath0 .
a regular neighbourhood of @xmath215 is a torus with two nonseparating boundary components , say @xmath216 and @xmath217 ( cf . figure [ lantern ] ( ii ) ) .
clearly , @xmath216 and @xmath217 are disjoint from @xmath211 , @xmath212 , @xmath213 , and from each other . by using the braid relation and the fact that @xmath218 commutes with @xmath219 , the relation given by proposition 3 in @xcite
is easily shown to be equivalent to the two - holed torus relation @xmath220 now we describe various commutator relations in mapping class groups
. these relations will be used in the next section to construct the lefschetz fibrations used in the course of the proofs of the theorems stated in section [ first ] .
[ proposition ] let @xmath186 , @xmath187 , @xmath221 and @xmath222 be four simple closed curves on @xmath0 such that @xmath186 is disjoint from @xmath187 , @xmath221 is disjoint from @xmath222 , and the complements of @xmath223 and @xmath224 in @xmath0 are connected .
then @xmath225 is a commutator . by the classification of surfaces
, there exists a diffeomorphism @xmath76 of @xmath0 such that @xmath226 and @xmath227
. then @xmath228 \ .\ ] ] [ thm1 ] let @xmath16 and let @xmath186 be a simple closed curve on @xmath0 . in the mapping class group @xmath180 of @xmath0
* @xmath229 can be written as a product of two commutators , * if @xmath186 is nonseparating , then @xmath230 can be written as a product of three commutators .
suppose that the surface of genus @xmath231 with two holes in figure [ feners ] is embedded in @xmath0 .
consider the curves on @xmath0 given in the figure .
the sphere @xmath202 with four holes of the lantern relation , see figure [ lantern ] , can be embedded in @xmath0 so that the curves @xmath203 , @xmath204 , @xmath205 , @xmath206 , @xmath207 , @xmath208 , @xmath209 become respectively @xmath212 , @xmath211 , @xmath216 , @xmath217 , @xmath232 , @xmath213 , @xmath233 .
this gives us the relation @xmath234 similarly , two other embeddings of @xmath202 give the relations @xmath235 and @xmath236 if we multiply both sides of by @xmath237 , use and cancel @xmath238 , we obtain @xmath239 or , equivalently , @xmath240 similarly , the equalities and yield the equality @xmath241 applying lemma [ proposition ] to and proves that @xmath242 and @xmath243 are products of two commutators .
any nonseparating simple closed curve is topologically equivalent to @xmath216 .
if @xmath186 is a separating simple closed curve on @xmath0 , then the surface of genus @xmath5 on the right hand side of @xmath232 can be embedded in @xmath0 so that @xmath232 is topologically equivalent to @xmath186 .
now , the proof of ( a ) follows from the fact that a conjugate of a commutator is again a commutator .
similarly , two more embeddings of the lantern give the relations @xmath244 multiplying by @xmath245 from the left and using gives @xmath246 by combining and , we get @xmath247 cancelling @xmath219 and @xmath248 yields @xmath249 any simple closed curve on the left hand side is disjoint from each closed curve on the right .
notice also that the complements of @xmath250 , @xmath251 , @xmath252 , @xmath253 , @xmath254 and of @xmath255 are all connected .
lemma [ proposition ] now implies that @xmath256 is a product of three commutators , implying ( b ) .
[ negy ] let @xmath1 and let @xmath186 and @xmath187 be two simple closed curves intersecting each other transversely at one point on @xmath0 .
then @xmath257 is a product of three commutators .
suppose that the two - holed torus of figure [ lantern ] ( ii ) is embedded in @xmath0 in such a way that @xmath216 and @xmath217 are nonseparating on @xmath0 .
the curve @xmath212 intersects @xmath258 transversely at one point .
since @xmath186 intersects @xmath187 transversely at one point also and since any two such pairs are topologically equivalent , we can assume that @xmath259 and @xmath260 .
by the two - holed torus relation , we have @xmath261 .
let us denote @xmath262 by @xmath59 .
then , we obtain @xmath263 if @xmath264 and @xmath265 , we have the equality @xmath266 now , @xmath267 is a commutator and @xmath268 is a product of two commutators .
this observation completes the proof of proposition [ negy ] .
the relations expressing certain products of dehn twists as products of commutators proved in section [ third ] allow us to construct corresponding lefschetz fibrations .
these fibrations , and their signatures , depend on the choices we make for the diffeomorpisms occurring in the commutator relations . in this section
the genus of the fiber @xmath0 is @xmath16 .
the base @xmath2 of genus @xmath76 will be denoted by @xmath269 if it is closed , and by @xmath270 if it has @xmath179 boundary components . for a smooth surface bundle @xmath271 , the signature
is completely determined by the corresponding monodromy representation .
we shall pass back and forth between surface bundles over bases with boundary and lefschetz fibrations over closed bases using the following well - known fact , see @xcite : [ p : ss ] the signature of a fibered neighbourhood of a nonseparating , respectively separating , singular fiber in a lefschetz fibration is equal to @xmath8 , respectively to @xmath272 .
now fix a symplectic basis for @xmath273 , so that the monodromy representation @xmath274 of @xmath6 composed with the action of the mapping class group on homology @xmath275 yields a symplectic representation @xmath276 of the fundamental group of the base .
the following result of meyer @xcite allows us to calculate the signature : [ t : meyer ] let @xmath277 be an oriented surface bundle with monodromy representation @xmath278 . fix a standard presentation of @xmath279 as follows : @xmath280\prod_{j=1}^rc_j=1\rangle \ , \ ] ] and let @xmath281 by the cocycle defined in @xcite .
then the signature of @xmath6 is given by the formula @xmath282 where @xmath283 , @xmath284 , @xmath285 and @xmath286 $ ] .
here is a first application of this formula : [ p : sdehn ] there is a lefschetz fibration @xmath287 with a unique singular fiber and with signature @xmath272 , whether the vanishing cycle is separating or not .
it is well - known that a dehn twist can be written as a product of two commutators , see @xcite .
we need to make explicit choices for these commutators . to this end
we consider curves @xmath186 , @xmath211 , @xmath212 , @xmath213 , @xmath233 , @xmath288 and @xmath289 on a genus @xmath231 subsurface of @xmath0 as in figure [ ko ] .
further , we add curves according to figure [ koadd ] . if we choose the genus @xmath231 subsurface suitably , the vanishing cycle @xmath290 is topologically equivalent to the curve @xmath186 .
define diffeomorphisms @xmath291 and @xmath292 of @xmath0 as follows .
if @xmath186 is nonseparating , set @xmath293 and @xmath294 if @xmath186 is separating , set @xmath295 and @xmath296 one can check that @xmath297 , @xmath298 and @xmath299 .
the lantern relation as in theorem 2 of @xcite implies that @xmath300 the monodromy representation of the complement of the singular fiber is given by mapping the standard generators of @xmath301 to @xmath219 , @xmath292 , @xmath291 , @xmath302 and @xmath198 respectively , as @xmath303[\phi_1,t_{a_1}^{-1}t_{b_1}]t_{a}=1 \ .\ ] ] evaluating the signature cocycle , theorem [ t : meyer ] shows that the complement of the singular fiber has signature @xmath272 if @xmath290 is nonseparating , and has signature @xmath8 if @xmath290 is separating .
now proposition [ p : ss ] and novikov additivity complete the proof .
_ mutatis mutandis _ , this calculation generalizes to prove the next three propositions : [ p : sdehn^2 ] there is a lefschetz fibration @xmath287 with two singular fibers whose monodromies are dehn twists with the same nonseparating vanishing cycle and signature equal to @xmath304 .
[ p : sdehn^4 ] there is a lefschetz fibration @xmath305 with four singular fibers whose monodromies are dehn twists with the same nonseparating vanishing cycle and signature equal to @xmath306 .
[ p : sta^4tb^4 ] let @xmath186 and @xmath187 be two nonseparating simple closed curves on @xmath0 which intersect transversely and precisely at one point .
there is a lefschetz fibration @xmath307 with signature @xmath306 which has eight singular fibers , four of which have monodromy a dehn twist along @xmath186 and four of which have monodromy a dehn twist along @xmath187 . in all these proofs the signature of the lefschetz fibration is the same as that of the complement of the singular fibers , because all the vanishing cycles are nonseparating , cf . proposition [ p : ss ] .
we take curves @xmath232 , @xmath308 , @xmath309 , @xmath310 on a genus @xmath231 subsurface of @xmath0 as in figure [ feners ] .
we also add curves @xmath311 , @xmath312 , @xmath313 as in figure [ koadd ] , and @xmath222 and @xmath314 as in figure [ commadd ] . for each of the propositions , the vanishing cycles @xmath186 and @xmath187 are topologically equivalent to certain curves @xmath315 and @xmath316 .
we fix the latter explicitly , and construct some diffeomorphisms as required by the proofs in section [ third ] , so that we can write the monodromy representation of the complement of the singular fibers as a relator in the mapping class group
. then the calculation is done by implementing the formula in theorem [ t : meyer ] with the following data . for proposition [ p : sdehn^2 ]
the base is @xmath317 and @xmath315 is taken to be @xmath216 .
the relator giving the monodromy representation is @xmath318[t_{b_1}^{-1}t_{a_2},\phi_2]t_{a_4}^2=1 \ , \ ] ] with @xmath319 and @xmath320 for proposition [ p : sdehn^4 ] the base is @xmath321 and @xmath315 is taken to be @xmath216 again .
the relator giving the monodromy representation is @xmath322[t_{b_7}^{-1}t_{a_8},\phi_2 ] [ t_{b_2}^{-1}t_{a_1},\phi_3]t_{a_4}^4=1\ ] ] with @xmath323 @xmath324 and @xmath325 for proposition [ p : sta^4tb^4 ] consider the curves @xmath326 on a genus @xmath5 subsurface of the fiber as in figure [ lantern ] ( ii ) .
the base surface is @xmath327 and the curve @xmath315 is taken to be @xmath212 and @xmath316 is taken to be @xmath258 .
we first compute the signature corresponding to the relator @xmath328 t_{a_2}^4(t_{a_1}t_{a_2}t_{a_1}^{-1})^4t_{a_1}^2=1 \ , \ ] ] with @xmath329 which is the monodromy of a fibration @xmath330 with @xmath331 singular fibers .
the signature of @xmath332 is equal to @xmath333 .
subtracting off the fibration of proposition [ p : sdehn^2 ] from @xmath332 gives the claim .
as the expression of a given element in @xmath4 as a product of commutators is not unique , it is conceivable that the signature of the corresponding lefschetz fibrations might be different for different choices of commutators .
then a surface bundle of nonzero signature could be constructed by subtracting the lefschetz fibrations corresponding to different choices from each other .
we now prove the theorems about the minimal genus function @xmath20 stated in the introduction .
we apply the subtraction operation to the lefschetz fibrations @xmath334 and @xmath335 as in proposition [ p : sta^4tb^4 ] and [ p : sdehn^4 ] , respectively . in @xmath93
we group the singular fibers into two groups each containing four singular fibers with coinciding vanishing cycles ; in @xmath94 the singular fibers form one group . now subtracting two copies of @xmath94 according to the above pattern we get a surface bundle @xmath336 of fiber genus @xmath11 with @xmath337 ( cf .
propositions [ p : sdehn^4 ] and [ p : sta^4tb^4 ] ) .
thus @xmath338 , and the claim now follows by pulling @xmath339 back to unramified coverings of @xmath340 of degree @xmath341 .
notice that the relators defining the fibrations we used in the proof of theorem [ main1 ] represent 1 in the mapping class group @xmath142 of a surface with one boundary component . according to proposition [ secc ] , this fact shows that the fibrations given by the relators @xmath342t_a^4 t_b^4 $ ] and @xmath343t_a^4 $ ] admit sections with vanishing self - intersection .
since the lifts of the various dehn twists are chosen to be dehn twists in @xmath142 , proposition [ osszerag ] implies that @xmath336 also admits a section with zero self - intersection for all @xmath11 .
now write @xmath11 as @xmath344 where @xmath345 , and apply lemma [ secrag ] to @xmath346 copies of @xmath347 together with the product @xmath348 .
the resulting surface bundle @xmath349 of fiber genus @xmath11 has @xmath350 .
now every surface of odd genus is a covering of a genus @xmath231 surface .
it was shown in lemma 4.1 of @xcite , that after replacing a given surface bundle by a pullback to some covering of the base , the resulting surface bundle admits fiberwise coverings of any given degree . from the multiplicativity of the signature in coverings and the multiplicativity of the euler characteristic of the fiber in fiberwise coverings , for odd @xmath11 we obtain @xmath353 for
even @xmath11 consider the fibration @xmath354 of fiber genus @xmath355 with signature @xmath356 we got by taking fiberwise coverings .
it is easy to see that since @xmath357 admits a section of zero self - intersection , so does @xmath358 .
summing @xmath359 and the product fibration @xmath360 along their sections ( as in lemma [ secrag ] ) , we get a fibration over @xmath361 with fiber genus @xmath11 and signature @xmath356 .
these examples yield the bound @xmath31 once @xmath11 is even .
consequently the proof of theorem [ main3 ] is complete . for certain values of @xmath11 , the examples of kodaira @xcite give a better upper bound , namely @xmath362 .
our construction has the advantage of covering all possible values of @xmath11 ( and @xmath26 ) . | we construct examples of lefschetz fibrations with prescribed singular fibers . by taking differences of pairs of such fibrations with the same singular fibers ,
we obtain new examples of surface bundles over surfaces with non - zero signature . from these
we derive new upper bounds for the minimal genus of a surface representing a given element in the second homology of a mapping class group .
# 1#2 |
the minimal supersymmetric standard model ( mssm ) predicts at least one light neutral scalar higgs particle with mass below @xmath2 gev . in order to experimentally detect possible signals of higgs bosons and to trace back as far as possible the physical origin of a produced scalar particle ,
detailed studies for the decay and production processes of higgs boson are required . as has been discovered several years ago [ 1 - 3 ] , radiative corrections in the mssm higgs sector are large and have to be taken into account for phenomenological studies .
several methods have been developed to calculate the radiative corrections to the mssm higgs boson masses , production and decay rates .
three main approaches have been used for the calculation of 1-loop corrections : * the effective potential approach ( epa ) @xcite : this method computes the dominant correction to the higgs boson mass spectrum and coupling constants in a simple and fast way and is thus most suitable for numerical calculations , e.g. in monte carlo studies . in its practical realizations ,
it is based on the following approximations : * * only the corrections to the 2-point green s functions are included , and only the contributions from top / stop and bottom / sbottom are taken into account * * the green s functions are evaluated at momentum @xmath3 , thus neglecting the momentum dependence .
+ the epa leads to expressions for the cross sections and decay branching ratios in an effective born approximation , where the tree level quantities are replaced by the corrected higgs boson masses and mixing angles . *
the renormalization group approach ( rge ) @xcite .
this method also leads to an effective born approximation in the formulae for cross sections .
the 1-loop corrected higgs boson masses and couplings are obtained by using rg equations .
large logarithmic terms can be resummed , but realistic mass spectra are not covered by this method since it relies on an effective susy scale .
the momentum dependence of the self - energies and 3- and 4-point functions are neglected also in this approach . *
the diagrammatic calculation ( feynman diagram calculation , fdc ) @xcite : the masses are calculated from the pole positions of the higgs propagators , and the cross sections are obtained from the full set of 1-loop diagrams contributing to the amplitudes . the version of ref .
@xcite is based on the on - shell renormalization scheme .
it takes into account : * * the most general form of the mssm lagrangian with soft breaking terms , * * the virtual contributions from all the particles of the mssm spectrum , * * all 2- , 3- and 4-point green s functions for a given process with higgs particles , * * the momentum dependence of the green s functions , * * the leading reducible diagrams of higher orders corrections . +
this method is technically most complicated , but also most accurate at the 1-loop level and can be used as the reference frame for simpler methods .
moreover , it allows to study the full parameter dependence of cross sections and decay rates .
the experimental searches for higgs bosons at lep1 @xcite and studies for the future searches at higher energies @xcite conventionally make use of the most compact effective potential approximation .
the phenomenological study in ref .
@xcite for lep2 and higher energies , was done in the fdc , but still not complete at 1-loop order .
a check of the quality of the simpler approximations and estimates of their reliability in various energy and parameter ranges have also not yet been performed so far . in this paper
we present the first complete 1-loop diagrammatic results for the cross sections for the neutral higgs production processes @xmath0 and @xmath1 .
in addition , we compare the results from the complete diagrammatic calculation with the corresponding ones of the simpler and compact effective potential approximation and discuss the typical size of the differences .
recently some papers on the leading 2-loop corrections to the cp - even mssm higgs boson masses have been published @xcite .
the main conclusion is that 2-loop corrections are also significant and tend to compensate partially the effects of 1-loop corrections .
the calculations are based on the epa and rg methods . since one of the main emphasis of our study is to figure out the difference between complete and approximate results in a given order , we have not implemented the 2-loop terms .
they would improve the 1-loop fdc results in the same way as the approximations and thus do not influence the remaining differences which can only be obtained by an explicit diagrammatic calculation . in section [ sec :
outline ] we give a short description of the computational schemes for the masses and cross sections .
this is followed in section [ sec : results ] by a discussion of the numerical results and a comparison of the various approximations .
the tree level potential for the neutral mssm higgs bosons can be written as : @xmath4 diagonalization of the mass matrices for the cp - even and the cp - odd scalars , following from the potential ( [ eq : pot0 ] ) , leads to three physical particles : two cp - even higgs bosons @xmath5,and one cp - odd higgs boson , and defines their tree - level masses , and , with @xmath6 , and the mixing angles @xmath7 , @xmath8 .
the 1-loop radiative corrections significantly modify the tree level relations between the masses and mixing angles .
the way of calculating the radiative corrections in the epa and fdc methods is briefly described as follows : in the epa , the tree level potential @xmath9 is improved by adding the 1-loop terms @xcite : @xmath10 where @xmath11 @xmath12 is the tree level potential evaluated with couplings renormalized at the scale @xmath13 , and @xmath14 denotes the supertrace over the third generation of quark and squark fields contributing to the generalized squared mass matrix @xmath15 .
the full 1-loop potential @xmath16 is rediagonalized yielding the 1-loop corrected physical masses , and the effective mixing angle . thereby and
@xmath17 are usually considered as the free input parameters .
explicit formulae for the 1-loop corrected cp - even higgs mass matrices can be found in first paper of ref .
the physical higgs boson masses and 1-loop mixing angles are used in the born formulae for the production cross sections and decay rates . in many experimental neutral mssm higgs searches a simplified version of the epa ( )
was used in the analysis of the experimental data . in this approximation
only the leading correction coming from the top and stop quark loops to the higgs mass matrix is included . in addition , the effects of the sfermion mixing are neglected . in this case
the corrected higgs mass matrix has the simple form : @xmath18 where @xmath19 summarizes the leading 1-loop corrections : @xmath20 this yields : @xmath21 @xmath22 the 1-loop corrected higgs boson masses , and the effective mixing angle thus depend on only one extra susy parameter , namely the product @xmath23 of the scalar top mass eigenvalues , in addition to the tree level parameters and @xmath17 . in the fdc approach the 1-loop physical higgs boson masses are obtained as the pole positions of the dressed scalar propagators . in order to obtain accurate values of the masses , also
the leading reducible terms of higher order have to be taken into account . @xmath24 and @xmath25
are given by the solution of the equation of the form : @xmath26=0 \label{eq : fdcmass}\end{aligned}\ ] ] for the calculations of the cross sections we need the full set of scalar 2-point functions , and also the 3- and 4-point green s functions are taken into account . in figure
[ fig : feyn ] the diagrams contributing to the @xmath27 process are collected .
specially important is the diagram f ) in figure [ fig : feyn ] , illustrating the -mixing on the external scalar line , as this mixing corresponds in a good approximation to the effect of introducing the effective mixing angle into the born formulae .
the formulae for the cross sections obtained in the fdc now differ from the born expressions , because not only the effective values of the masses are corrected but also new form factors and momentum dependent effects are considered .
the analytic formulae for the cross sections for the processes @xmath28 , including all but the box and 1-loop @xmath29 vertex contributions , can be found in the first paper of ref .
effects of box contributions are discussed in the third paper of ref . @xcite . in the present paper
we include also the 1-loop contributions to the @xmath29 vertex in the cross section calculation .
note that the diagrams h ) , i ) , j ) of figure [ fig : feyn ] are not part of the phenomenological study in @xcite .
in this section we present the results for @xmath30 and @xmath31 production from the complete diagrammatic 1-loop calculation ( fdc ) and discuss the quality of simpler approximations : * the effective potential approach ( epa ) , as described above , * the simplified epa with sfermion mixing and bottom / sbottom loop neglected ( ) , based on eqs .
( [ eq : emassmat]-[eq : ealpha ] ) . in order to give an orientation about the accuracy of the approximations
we show comparisons of the various methods in the different regions of the parameter space . from the theoretical point of view , the most convenient parameters for the higgs sector are the mass of the cp - odd higgs boson and the ratio @xmath32 . from the experimental point of view
it is more natural to use the mass of the lighter cp - even higgs instead of the formal quantity @xmath17 .
this , however , is technically more complicated for the calculational scheme . in the first part of this discussion we follow the conventional way of presentation and choose @xmath17 as a free input parameter . in the second part we replace @xmath17 in favor of the physical neutral higgs mass
this has the advantage to directly get the theoretical results for the production cross sections for the case that a higgs boson is found and its mass has been determined . as a first step
, we put together the predictions of the various methods for the physical mass for given and @xmath17 .
figure [ fig : mass1 ] is based on the parameters listed in table [ tab : par ] .
.parameters used for the numerical analysis .
[ tab : par ] [ cols="^,^,^,^,^,^,^,^",options="header " , ] we have analyzed also the dependence of the differences between the epa and the fdc predictions on the susy parameters : sfermion and gaugino masses , @xmath33 parameter and sfermion mixing parameters . in most cases
the variation of those parameters does not have a large effect on the size of the differences between the epa and fdc .
the only exception is the gaugino mass , which is completely absent in the epa . for a relatively heavy ( @xmath34 gev )
the increase of @xmath35 causes a slow change of , , ( see figure [ fig : gcr ] ) .
this effect can be quite dramatic for small @xmath36 gev and @xmath35 around 1 tev , but such low values are however already excluded by the lep1 measurements .
we now turn to the more physical parametrization of the cross sections in terms of the two higgs boson masses and .
this parameterization is more clumsy in the calculations , but it has the advantage of physically well defined input quantities avoiding possible confusions from different renormalization schemes .
technically , @xmath17 is calculated for given , and the other input parameters .
varying ( and other input quantities fixed ) we thus obtain @xmath17 and @xmath37 , @xmath38 as functions of .
the predictions for the cross sections are , however , in general not unique because there can be two solutions for @xmath17 , thus yielding two branches for fixed values of .
the solutions for @xmath17 are shown in figure [ fig : tbmass ] for the input parameters of table [ tab : par ] , again for the fdc and the simpler approximations .
as can be seen from the figure , the differences between the values obtained in the epa and fdc can reach 20% .
also significant differences can occur for the cross sections , as displayed in figure [ fig : tbcr ] where the predictions of epa and fdc for the @xmath37 and @xmath38 are plotted as functions of .
the typical size of differences between the methods is 10 - 20% for = 205 gev .
we presented the results from a complete 1-loop diagrammatic calculation in the mssm for the cross sections for neutral higgs production via @xmath0 and @xmath1 , including box diagrams and @xmath39 vertex contributions which were not part of previous studies for higgs searches in the literature .
comparisons between the fdc predictions with the simpler epa approximation have shown that the epa has an accuracy of typically 10 - 20% in the parameter regions where the cross sections are large .
the differences become bigger with increasing energy
. they may be specially important for future high energy @xmath40 colliders .
the use of the physical input variables @xmath41 @xmath42 avoids ambiguities from the definition of @xmath17 in higher order , but the observed differences remain of the same size . for a better accuracy
, the full fdc would be required .
it will also be necessary to incorporate the leading 2nd order terms since they are at least as big as the non - leading 1-loop contributions .
+ the library of fortran codes for the calculation of the 1-loop radiative corrections in the on - shell renormalization scheme to the mssm neutral higgs production and decay rates @xcite can be found at the url address + _ http://itpaxp1.physik.uni-karlsruhe.de/@xmath43rosiek/neutral_higgs.html _
j. ellis , g. ridolfi and f. zwirner , _ phys .
lett . _ * 262b * ( 1991 ) 477 ; + r. barbieri and m. frigeni , _ phys .
lett . _ * 258b * ( 1991 ) 395 ; + a. brignole , j. ellis , g. ridolfi and f. zwirner _ phys . lett . _ * 271b * ( 1991 ) 123 .
m. carena , k. sasaki and c.e.m .
wagner , _ nucl .
phys . _ * 381b * ( 1992 ) 66 ; + h.e .
haber , r. hempfling , _ phys .
* d48 * ( 1993 ) 4280 : + p.h .
chankowski , s. pokorski and j. rosiek , _ phys .
lett . _ * 281b * ( 1992 ) 100 .
aleph collaboration , d. buskulic , _ phys .
lett . _ * 313b * ( 1993 ) , 312 ; + delphi collaboration , p. abreu , _ nucl .
_ * 373b * ( 1992 ) , 3 ; + l3 collaboration , o. adriani , _ z. phys .
_ * 57c * ( 1993 ) , 355 ; + opal collaboration , r. akers , preprint cern , ppe/94 - 104 ( 1994 ) ; + j. f. grivaz , plenary talk at international europhysics conference on high energies , brussels , july 27 - august 2 , 1995 ; preprint lal-95 - 83 ( to appear in the proceedings ) . , m. carena and p. zerwas et al . , to appear in the proceedings of lep2 workshop , cern yellow report , eds .
g. altarelli et al . ;
+ _ higgs particles _ , in : @xmath40 collisions at 500 gev , desy 92 - 123a , b ; desy 93 - 123c , ed .
p. zerwas .
r. hempfling , a. hoang , _ phys .
_ * b331 * ( 1994 ) 99 ; + m. carena , j.r .
espinosa , m. quiros , c.e.m .
wagner , _ phys .
* b355 * ( 1995 ) 209 ; + m. carena , m. quiros , c.e.m .
wagner , preprint cern - th-95 - 157 , hep - ph/9508343 . | we present the first complete 1-loop diagrammatic calculation of the cross sections for the neutral higgs production processes @xmath0 and @xmath1 in the minimal supersymmetric standard model .
we compare the results from the diagrammatic calculation with the corresponding ones of the simpler and compact effective potential approximation and discuss the typical size of the differences .
-2 cm # 1 |
the interaction between the _ elements _ of social , technological , biological , chemical and physical systems usually defines complex networks .
the study of topological properties in such networks has recently generated a lot of interest among the scientific community @xcite .
part of this interest comes from the attempt to understand the behavior of technology based communication networks such as the internet @xcite , the world wide web @xcite , e - mail networks @xcite or phone call networks @xcite .
the study of communication processes is also of interest in other fields , notably in the design of organizations @xcite .
it is estimated that more than one - half of the u.s .
work force is dedicated to information processing , rather than to make or sell things in the narrow sense @xcite .
tools taken from statistical mechanics are used to understand not only the topological properties of these communication networks , but also their dynamical properties .
particularly interesting is the phenomenon of congestion . in has been observed , both in real networks @xcite and in model communication networks , @xcite that the system only behaves efficiently when the amount of information handled is small enough .
the network collapses above a certain threshold and some information is accumulated and remains undelivered over large time periods or it is simply lost .
this transition from a _ free _ to a _ congested _ regime is indeed a phase transition and could be related to the appearance of the @xmath4 noise observed in internet flow data @xcite .
understanding congestion is also important from an engineering point of view .
for instance , in october of 1986 , during the first internet collapse reported in the literature , the speed of the connection between the lawrence berkeley laboratory and the uc at berkeley separated by 200 meters dropped by a factor of 100 @xcite . although the problem of congestion , and in particular its prevention and control @xcite , has been studied because of its implications in digital communications , a deep understanding of the physics of congestion for general communication processes and beyond particular protocols is still lacking .
a general collection of models that captures the essential features of communication processes has been recently proposed @xcite . in these models ,
agents nodes are organized in a hierarchical network a cayley tree and interchange information _
packets_. each agent has a certain capability that decreases as the number of packets to deliver increases .
when the capability is inversely proportional to the number of accumulated packets , a continuous phase transition was found between the free and the congested phases .
this transition was characterized by means of an order parameter .
the aim of this paper is to study the collection of models mentioned above , since only hierarchical lattices and a very particular congestion behavior
capability inversely proportional to the number of accumulated packets were considered .
an extension to consider the existence of cost to maintain communication channels has already been done @xcite .
first , the network model is extended beyond purely hierarchical cayley trees .
in particular , 1-dimensional and 2-dimensional lattices are considered .
it should be noted that simple models such as cayley trees and regular lattices can capture the main characteristics of the dynamics of information flow in complex networks .
hierarchical trees have been considered to model the tcp / ip protocol in the internet @xcite .
on the other hand , computer based communication networks have been described by placing routers and servers in square lattices @xcite .
moreover it is shown that , independently of the topology , the collection of models can be split into three groups according to how the network collapses . in the first group , agents deliver more packets as they are more congested although their capability , as defined in @xcite , decreases , and the network never collapses . in the second group ,
agents deliver always the same number of packets independently of their load number of packets to deliver ; this behavior leads to a continuous phase transition as reported for hierarchical lattices in @xcite .
finally , when agents deliver fewer packets as their loads increase , the transition to the congested phase is discontinuous and the network collapses in an inhomogeneous way giving rise to congestion nuclei . to characterize these behaviors , the order parameter defined in @xcite
is used , but also the power spectrum of the fluctuations and a susceptibility - like function .
most of the current effort is focused on the quantitative study of the continuous transition and the critical behavior associated to such a transition .
the paper is organized as follows .
section 2 presents the collection of models and describes some details concerning the extension beyond hierarchical lattices , including 1d and 2d lattices . in section 3 , these models are studied numerically and analytically .
the scaling of the critical point with the size of the system , the behavior of the order parameter , and the divergence of the characteristic time at the critical point are studied in detail .
finally , section 4 includes the discussion of the results and the conclusions .
the model considers three basic components : ( i ) the physical support for the communication process agents and communication channels , ( ii ) the discrete information packets that are interchanged , and ( iii ) the limited capability of the agents to handle such packets .
the communication network is mapped onto a graph where nodes mimic the communicating agents ( for instance , employees in a company , routers and servers in a computer network , etc . ) and the links between them represent communication lines . in particular , the three different topologies depicted in fig . [ nets ] are considered : 1d lattices of length @xmath5 , 2d square lattices of side length @xmath5 and size @xmath6 , and hierarchical cayley trees with branching @xmath7 and a number of generations or levels @xmath8 , hereafter denoted @xmath9 . ;
( b ) 2d lattice with @xmath10 and @xmath11 ; and ( c ) @xmath12 cayley tree with branching @xmath13 and @xmath14 levels or generations . ]
the dynamics of the model is as follows . at each time step @xmath15
, an information packet is created at every node with probability @xmath16 .
therefore @xmath16 is the control parameter : small values of @xmath16 correspond to low density of packets and high values of @xmath16 correspond to high density of packets .
when a new packet is created , a destination node , different from the origin node , is chosen randomly in the network .
thus , during the following time steps @xmath17 , the packet travels toward its destination . once the packet reaches the destination node , it is delivered and disappears from the network .
the time that a packet remains in the network is related not only to the distance between the source and the target nodes , but also to the amount of packets in its path .
in particular , at each time step , all the packets move from their current position , @xmath18 , to the next node in their path , @xmath19 , with a probability @xmath20 .
this probability @xmath20 is called the _ quality of the channel _ between @xmath18 and @xmath19 , and was defined in @xcite as @xmath21 where @xmath22 represents the _ capability _ of agent @xmath18 at each time step .
the quality of a channel is , thus , the geometric average of the capabilities of the two nodes involved , so that when one of the agents has capability 0 , the channel is disabled .
high qualities ( @xmath23 ) imply that packets move easily while low qualities ( @xmath24 ) imply that it takes a long time for a packet to jump from one node to the next .
the general equation proposed for @xmath22 is : @xmath25 where @xmath26 is the total number of packets currently at node @xmath18 .
the function @xmath27 determines how the capability evolves when the number of packets at a given node changes , @xmath28 with @xmath29 .
equation ( [ f_1 ] ) defines a complete collection of models with agents that behave differently depending on the exponent @xmath0 .
the average number of packets delivered during one time step by a node @xmath18 to another node @xmath19 is proportional to @xmath30 . assuming that @xmath31 the former expression reads @xmath32 .
the proportionality is exact both in the hierarchical lattice and in large enough 1d and 2d lattices , where adjacent nodes are statistically equivalent .
therefore , for @xmath2 the number of delivered packets increases with the number of accumulated packets . for @xmath3
the number of delivered packets decreases as the number of accumulated packets increases .
finally , for the particular case @xmath1 , the number of delivered packets is independent of the number of accumulated packets .
note that this particular case is consistent with simple models of queues @xcite .
the last point that needs to be explained to completely define the current model is the routing algorithm or , in other words , the set of rules that the nodes follow to select where to send a certain packet . in all the topologies considered ( 1d , 2d and cayley ) the packets follow paths of minimum length from their origin to their destination open boundary conditions are set in both 1d and 2d networks . in 1d lattices and cayley trees
it is trivial to follow a minimum path because there is only one minimum path between two arbitrary nodes . in 2d lattices , however , there are many minimum paths .
if a node can choose between two neighbors when sending a given packet , and both neighbors belong to a minimum path between the origin and the destination of the packet , one of them is chosen randomly with equal probability .
this algorithm is indeed the simplest one and the interpretation of the results is clearer than for more complex routing algorithms .
however , congested nodes will still have lower probability of receiving packets because of the definition of the quality of a channel .
therefore packets will avoid those nodes to some extent , as would happen in more sophisticated routing algorithms .
for certain parameters of the model ( in particular @xmath1 @xcite ) and for similar models of traffic @xcite , a transition from a _ free _ regime , where all the packets reach their destination , to a _ congested _ , regime where some packets are accumulated in the network , has been found .
this transition is properly described by means of the following order parameter @xmath33 introduced previously @xcite : @xmath34 in this equation @xmath35 and @xmath36 indicates average over time windows of width @xmath37 .
these averages can be over one or many realizations , yielding the same result .
essentially , the order parameter represents the ratio between undelivered and generated packets calculated at long enough times such that @xmath38 .
thus , @xmath39 is only a function of the probability of packet generation per node and time step , @xmath16 .
the power spectrum @xmath40 of the total number of packets in the network @xmath41 is also used here to further understand the phase transition . by means of the power spectrum the behavior of the time correlations of the system can be studied .
let us study separately the critical case @xmath1 and the non critical cases @xmath2 and @xmath3 .
a continuous transition between the free regime and the congested regime occurs in hierarchical networks for @xmath1 , as reported in @xcite . for small values of @xmath16 ,
all the packets reach their destination and the total number of packets @xmath42 fluctuates around a finite value . in this case
the order parameter is @xmath43 .
however , as @xmath16 increases , a critical point @xmath44 is reached , where the fluctuations in @xmath42 become very large and the characteristic time of the system diverges ( critical slowing down ) . beyond this point ,
some packets remain undelivered and @xmath45 grows linearly with @xmath15 .
the same qualitative behavior is observed here for 1d and 2d lattices , although there are quantitative differences . from an engineering point of view
it is interesting to study first the behavior of @xmath44 as the number of nodes in the network @xmath46 changes , because it will provide valuable information about the scalability of the network .
note that @xmath47 is the maximum number of packets that the network can handle per time step and , thus , it is a measure of the capacity of the network . for the hierarchical cayley tree , a mean field calculation of @xmath44 was obtained in @xcite : @xmath48 where the approximation holds for @xmath49 .
it is also possible to derive a mean field expression of @xmath44 for the 1d lattice .
since the most congested node is , from symmetry arguments , the central one the node at @xmath50 , the network will collapse when the amount of packets received by this central node is higher than the maximum number of packets that it is able to deliver .
since in a large enough network it is safe to assume that the central node will be congested similarly to its neighbors , @xmath51 , the maximum number of delivered packets should be 1 .
on the other hand , the number of packets arriving to the central node at each time step is the number of packets that are generated at each time step at the left half of the network and have their destination at the right half and conversely , this is @xmath52 .
then the critical condition is given by @xmath53 these mean field expressions ( [ pcb ] ) and ( [ pc1 ] ) can be compared with simulations .
nevertheless , the fluctuations of @xmath42 become very large near @xmath44 and it is difficult to calculate the value of the order parameter . instead
, a susceptibility - like function @xmath54 can be defined by analogy with equilibrium thermal critical phenomena , and used to estimate more accurately the value of the critical probability of packet generation , @xmath44 .
the susceptibility @xmath55 is related to the fluctuations of the order parameter by @xmath56 where @xmath57 is the width of a time window , and @xmath58 is the standard deviation of the order parameter estimated from the analysis of many different time windows of width @xmath57 .
thus a calculation implies a long realization of @xmath42 , its division into windows of width @xmath57 , calculation of the average value of the order parameter in each window and finally the determination of the standard deviation of these values .
the susceptibility shows ( fig .
[ suscep ] ) a singularity at @xmath44 as @xmath57 grows , as expected . and ( b ) a ( 5,4 ) cayley tree , for different time windows @xmath57 .
the vertical dotted line corresponds to the mean field calculation of the critical point ( equations ( [ pc1 ] ) and ( [ pcb ] ) respectively.),title="fig : " ] + and ( b ) a ( 5,4 ) cayley tree , for different time windows @xmath57 .
the vertical dotted line corresponds to the mean field calculation of the critical point ( equations ( [ pc1 ] ) and ( [ pcb ] ) respectively.),title="fig : " ] + fig .
[ fig_pc ] shows that there is good agreement between expressions ( [ pcb ] ) and ( [ pc1 ] ) and the values of @xmath44 obtained numerically by means of susceptibility measures for different network sizes .
obtained for the ( a ) 1d and ( b ) the cayley trees .
the error bars of the numerical points are smaller than the size of the symbols.,title="fig:",height=172 ] + obtained for the ( a ) 1d and ( b ) the cayley trees .
the error bars of the numerical points are smaller than the size of the symbols.,title="fig:",height=182 ] + for the 2d lattice it is more difficult to obtain even a mean field expression for @xmath44 .
however , since for 1d lattices and cayley trees the scaling relation @xmath59 holds where @xmath46 is the size of the network , one may expect the same behavior for the 2d lattice . using the susceptibility to numerically determine @xmath44 from simulations ,
one finds that this turns out to be incorrect .
although it is difficult to obtain a precise value of the exponent , fig [ 2d ] shows that it is close to 0.6 instead of 1.0 : @xmath60 lattice .
( b ) scaling of the critical probability of packet generation as a function of the size of the system for 2d square lattices.,title="fig : " ] + lattice .
( b ) scaling of the critical probability of packet generation as a function of the size of the system for 2d square lattices.,title="fig : " ] + this result suggests that the existence of multiple paths to get from the origin to the destination has important consequences , not only in shifting the value of @xmath44 , but actually changing its critical scaling behavior .
the behavior of the order parameter is studied next .
it is possible to derive an analytical expression for the simplest 1d case where there are only two nodes that exchange packets . since from symmetry considerations @xmath61 , the average number of packets eliminated in one time step is @xmath62 , while the number of generated packets is @xmath63 .
thus @xmath64 and with the present formulation of the model it is not possible to reach the super - critical congested regime .
however , @xmath16 can be extended to be the average number of generated packets per node at each step ( instead of a probability ) and in this case it can actually be as large as needed . as a result , the order parameter for the super - critical phase is @xmath65 . as observed in fig .
[ par_ordre ] , the general form @xmath66 fits very accurately the behavior of the order parameter not only for this simple 1d lattice with @xmath67 or the cayley tree @xcite , but also for any 1d lattice .
two - dimensional lattices again deviate from this behavior , although the deviation is small . .
the solid line corresponds to the analytical calculation for two nodes exchanging information packets .
symbols correspond to simulations performed in 1d , 2d and cayley lattices.,title="fig : " ] + in particular , near @xmath44 , equation ( [ ord_eq ] ) implies @xmath68 and thus the critical exponent for the order parameter is equal to 1 at least for the 1d lattice and the cayley tree .
the analysis of the power spectrum shows that in the sub - critical regime , i.e. in the free phase , the spectrum is well fitted by a lorentzian characterized by a frequency , @xmath69 , corresponding to exponentially decaying correlations with a characteristic time @xmath70 , as depicted in fig .
[ spectra ] .
as @xmath16 approaches @xmath44 , it is observed that @xmath71 and the power spectrum becomes @xmath72 for the whole range of frequencies .
alternatively the characteristic time diverges as @xmath73 ( critical slowing down ) .
this qualitative behavior is common to all the topologies of the underlying network , as shown in fig .
[ spectra ] . for different topologies and different values of the control parameter @xmath74 for the 1d case ( @xmath75 ) , 2d case ( @xmath76 ) , and cayley @xmath77 .
power spectra have been obtained averaging over 100 realizations of @xmath42 .
dotted lines represent a power law with exponent -2 .
right : characteristic frequency as a function of the control parameter @xmath78 for the different topologies .
as observed , the characteristic frequency tends to 0 as @xmath79 following a power law .
the straight lines correspond to fittings of equation ( [ forfc]).,title="fig : " ] + it is interesting to study how the characteristic frequency drops to 0 for each network topology . near the critical point
, one expects the scaling behavior @xmath80 the value of the critical exponent @xmath81 can be estimated by fitting eq .
( [ forfc ] ) to values of @xmath82 close enough to the critical point , as shown in fig .
[ spectra ] .
note that we fit @xmath44 and @xmath81 simultaneously .
this procedure yields very accurate values of @xmath44 but the values of @xmath81 are subject to large errors .
[ spectra ] yields @xmath83 for a 1d network with @xmath75 , @xmath84 for a 2d network with @xmath85 , and @xmath86 for a ( 7,5 ) cayley tree .
the determination of @xmath81 is interesting not only from an academic point of view , but also from an engineering perspective .
indeed , this exponent is related to divergences of other relevant quantities near the critical point . any characteristic time @xmath70the average time to deliver a packet , for instance
will diverge as @xmath87 and similarly the total number of packets @xmath88 since @xmath89 from little s law @xcite .
this law states that , in steady state , the number of delivered packets and the number of generated packets are equal .
the estimation of @xmath81 is particularly interesting in electronic communication protocols . indeed , eq .
( [ fort ] ) is used to determine the waiting time before a packet is considered lost in the network and therefore sent again @xcite . in practice , the exponent @xmath90 predicted by classical queue theory @xcite is assumed , while our current results suggest that more complex settings can lead to exponents even larger than 2 .
we have shown that the number of packets delivered by node @xmath18 is @xmath32 and thus , when @xmath2 , it increases with the number of packets that this node accumulates .
it is difficult to imagine a real scenario with this characteristic .
however , this case has been included to understand the critical behavior when @xmath91 , i.e. , to show the relationship between criticality and the amount of packets that can be delivered when load increases . as a consequence of the increase of the deliver capability with the load ,
the transition to collapse will never occur because , at some point in time , the number of accumulated packets will be large enough and the number of delivered and created packets will balance each other .
thus , the order parameter will be zero for any value of the control parameter @xmath16 , and the correlations will decay exponentially . as shown in fig .
[ fc_alt1 ] , the characteristic frequency tends asymptotically to @xmath92 as @xmath16 increases .
this asymptotic value depends on the size of the system . for a 1d lattice with a high density of packets ( @xmath93 ) , the number of packets that are delivered by a node is @xmath32 while the number of packets that are being delivered to this node is proportional to @xmath5 ( for instance , for the central node
, this number is simply @xmath52 ) .
therefore , @xmath94 .
the total number of packets is @xmath95 and according to little s law @xmath96 on the other hand , for @xmath97 the scaling of the characteristic frequency is given by @xmath98 since the packets success to jump from one node to the next at each time step , and therefore the characteristic time is directly the average path length between nodes .
therefore , although there is no phase transition in this case @xmath2 , there is a cross - over from a low density behavior to a high density behavior , as shown in fig .
[ fc_alt1 ] . this crossover
is also observed in 2d lattices and cayley trees . as a function of the probability of packet generation @xmath16 , for @xmath99 and different sizes of a 1d lattice . as observed , @xmath69 never becomes 0 as happens in the critical @xmath1 case .
inset : characteristic frequency at @xmath97 , @xmath100 ( squares ) , and characteristic frequency at large @xmath16 , @xmath101 ( circles ) .
the lines represent the fittings provided by equation ( [ fc01d ] ) @xmath102 , and equation ( [ fc*1d ] ) @xmath103 , respectively.,title="fig : " ] + the phase transition observed for @xmath1 is recovered when @xmath3 . above a certain threshold ,
some packets are accumulated in the network and the order parameter differs from 0 .
however , the number of packets delivered by a node @xmath18 in this case @xmath3 decreases with the number of packets accumulated at that node .
therefore , when some packets are accumulated , @xmath26 grows and finally no packets are delivered at all .
thus suddenly above the transition , which is discontinuous , the order parameter becomes 1 .
the change in the order of the phase transition affects the spreading of the collapse over the network . in the critical case @xmath1
, the collapse starts at the most _ central _ node and then spreads from this point to the rest of the network . in this case
@xmath3 , the reinforcement effect the fact that the more collapsed a node is , the more collapsed will get in the future leads to the formation of many congestion nuclei generated by fluctuations , that spread over the whole network .
figure [ patterns ] illustrates the formation of these congestion nuclei for 2d lattices with @xmath104 and @xmath105 , and @xmath106 and @xmath107 , in the non critical case @xmath3 .
dark regions represent regions with small congestion levels while bright regions correspond to highly congested regions .
left : @xmath104 and @xmath105 .
right : @xmath106 and @xmath107.,title="fig : " ] , in the non critical case @xmath3 .
dark regions represent regions with small congestion levels while bright regions correspond to highly congested regions .
left : @xmath104 and @xmath105 .
right : @xmath106 and @xmath107.,title="fig : " ] +
a collection of models recently proposed for hierarchical networks @xcite has been examined in detail for several network topologies including 1d and 2d lattices , to characterize the phase transition to collapse .
new congestion scenarios have been analyzed .
first , it has been shown that the congestion behavior is governed by the ability of agents to deliver information packets when their load increases .
when agents deliver packets at a constant ratio independently of their load number of packets to deliver , a continuous transition as the one reported in @xcite for cayley trees is also observed for 1d and 2d lattices , the order parameter being the fraction of accumulated packets .
when the number of delivered packets increases with the load , no phase transition is observed .
conversely , when the number of delivered packets decreases with the load , the order parameter jumps from zero to one and the transition becomes discontinuous .
these different behaviors are tuned by a single parameter @xmath0 , which is the exponent that determines how the capability of nodes evolves with the number of accumulated packets .
when @xmath2 there is no transition to the congested regime , for @xmath1 the transition is continuous and for @xmath3 the transition is discontinuous .
note that the continuous transition ( reported in @xcite and in some models of queues @xcite ) is only a particular case between a no - congestion behavior and a discontinuous transition behavior .
thus , the critical behavior is intimately related to the independence between load and deliver capability of nodes .
these different behaviors have been analyzed separately .
the critical case @xmath1 presents the most interesting behavior . to properly understand the physics of the collapse process
it is necessary to know how the system approaches the critical point .
the order parameter becomes useless in this region due to the emergence of large fluctuations , and then we have defined a susceptibility - like function that is related , by analogy to equilibrium critical phenomena , to fluctuations of the order parameter .
this quantity shows a peak at the critical point and allows to study the dependence on the size and topology of the system . for 1d lattices and cayley trees ,
it has been shown ( by means of simulation and also with mean field calculations ) that @xmath44 scales as the inverse of the size of the system , @xmath59 . however , for 2d lattices a different scaling @xmath108 is obtained .
it is suggested that the existence of multiple paths from the origin to the destination of the packets is the responsible for this change in the scaling behavior . on the other hand ,
the study of the power spectrum of the number of the packets in the network as a function of time has provided valuable information about the variation of temporal correlations in the system . in particular , below @xmath44 correlations decay exponentially , with a characteristic time @xmath70 that diverges at @xmath44 .
the exponent of this divergence is significantly larger than what is expected from queue theory .
this can be relevant to accurately forecast the temporal behavior in complex communication settings . for @xmath2 , no phase transition is observed .
instead , a cross - over from a low - density to a high - density regime occurs . in the low density regime , the characteristic frequency
is determined by the average distance between nodes , while in the high density regime the characteristic frequency is determined by the capability of agents to deliver information packets . in 1d lattices ,
the cross - over changes the scaling from @xmath109 at low densities to @xmath110 at high densities . finally ,
when @xmath3 the transition to the congested regime is discontinuous , since the number of delivered packets decreases with the load and for long times no packets are delivered at all .
thus the order parameter jumps from 0 to 1 .
moreover , due to the reinforcement effect the fact that the more collapsed a node is , the more collapsed it will get in the future leads to the formation of many congestion nuclei generated by fluctuations , that spread over the whole network .
the existence and distribution of such congestion nuclei can also be of interest from an engineering point of view .
the authors are gratefully acknowledged to l.a.n .
amaral , a. cabrales , l. danon , x. guardiola , c.j .
prez , m. sales , and f. vega - redondo .
this work has been supported by dges of the spanish government , grants no .
ppq2000 - 1339 and no .
bfm2000 - 0626 , and eu tmr grants no .
ist-2001 - 33493 and no . erbfmrxct980183
r.g . also acknowledges financial support from the generalitat de catalunya . | we study the dynamical properties of a collection of models for communication processes , characterized by a single parameter @xmath0 representing the relation between information load of the nodes and its ability to deliver this information .
the critical transition to congestion reported so far occurs only for @xmath1 .
this case is well analyzed for different network topologies .
we focus of the properties of the order parameter , the susceptibility and the time correlations when approaching the critical point . for @xmath2
no transition to congestion is observed but it remains a cross - over from a low - density to a high - density state . for @xmath3
the transition to congestion is discontinuous and congestion nuclei arise . |
uniform shear flow is a prototype non - equilibrium state admitting detailed study at both the macroscopic and microscopic levels via theory and computer simulation .
this is an idealized version of shear flow between parallel plates in which the velocity profile is exactly linear in a coordinate orthogonal to the flow direction ( as in couette flow ) and with a spatially uniform temperature and pressure ( in contrast to couette flow ) .
there is a single scalar control parameter , the shear rate @xmath3 , which measures how far the system is driven from equilibrium .
this flow is generated by periodic boundary conditions in the local lagrangian frame ( lees - edwards boundary conditions ) that can be implemented at the levels of hydrodynamics , kinetic theory , and newtonian mechanics @xcite .
although these boundary conditions are non - local and therefore not reproducible in real experiments , they are ideally suited for computer simulation of this special nonequilibrium state and for more penetrating theoretical analysis . in this way
, a quantitative study of rheological properties usually associated with complex molecular systems has been performed for simple atomic fluids @xcite .
the most complete studies have been via molecular dynamics simulation of newtonian dynamics at high densities and , more recently , by monte carlo simulation of the boltzmann equation at low densities @xcite .
molecular dynamics simulations have revealed a transition from fluid symmetry to an ordered state at sufficiently high shear rates @xcite , which has been attributed to a short wavelength hydrodynamic instability @xcite .
the objective here is to show that uniform shear flow also is unstable at sufficiently long wavelengths , for any finite value of the shear rate @xcite .
this instability has not been seen in earlier computer simulations due to the small system sizes considered , with consequent restrictions to shorter wavelengths . the instability is identified theoretically from a hydrodynamic analysis both near and far from equilibrium .
this analysis is confirmed quantitatively at short times by monte carlo simulations of an associated low density kinetic equation .
the asymptotic evolution of this instability also is explored via monte carlo simulation showing transition to a non - steady , spatially inhomogeneous state superimposed upon the uniform shear flow .
finally , the instability is demonstrated by new molecular dynamics simulations on larger systems to allow consideration of long wavelengths . this instability does not invalidate or compromise in any way previous studies of uniform shear flow , since we also establish that the flow is stable at sufficiently short wavelengths .
the boundary conditions generate viscous heating so that uniform shear flow is not stationary .
this viscous heating can be controlled by the introduction of a non - conservative external force that acts as a uniform thermostat .
the simplest choice is a stokes law drag force on each particle proportional to its velocity relative to the local macroscopic flow .
the proportionality `` constant '' then is adjusted to compensate for the heating .
there are several possibilities in the detailed implementation of the thermostat , leading to the same properties for the stationary state but different hydrodynamics for small perturbations from that state .
the theory and monte carlo simulations are carried out for both global and local thermostats .
the role of the thermostat is studied in the appendix where it is shown that the qualitative features of the instability are not sensitive to the choice of thermostat . in the next section ,
the usual navier - stokes hydrodynamic equations are considered .
these equations are restricted to small gradients of the hydrodynamic fields relative to equilibrium , and consequently the shear rate must be small in this analysis .
otherwise , the density and interatomic force law can be considered arbitrary within the fluid phase . the stationary solution for uniform shear flow
is identified , and the linear hydrodynamic equations for small perturbations of this solution are studied .
the five hydrodynamic modes are determined in detail for the special case of spatial perturbations orthogonal to the flow .
a critical wavevector , @xmath4 , is determined such that for wavevectors @xmath5 the perturbations grow as a function of time .
the critical wavevector vanishes as the shear rate @xmath3 goes to zero , but for any finite value of the shear rate there are sufficiently small wavevectors ( long wavelengths ) such that the perturbation is unstable .
these hydrodynamic predictions are tested by comparison to monte carlo simulations at the more fundamental kinetic theory level .
a model kinetic equation has been analyzed for states near uniform shear flow , without restriction on the shear rate @xcite .
the hydrodynamic equations for small deviations from uniform shear flow determine the critical wavevector , @xmath6 , for values of the shear rate @xmath3 beyond the limitations of the navier - stokes equations where efficient monte carlo simulations are possible .
the theoretical prediction of the growth of initial perturbations is compared with a direct monte carlo simulation of a solution to the kinetic equation .
the results confirm both the hydrodynamic analysis and the prediction of an instability for times up to about @xmath7 , where @xmath1 is the mean free time , after which the initial growth has exceeded the limitations of the linear stability analysis .
the simulation results are continued up to @xmath2 to explore the ultimate stabilization by non - linear effects .
the asymptotic state for the hydrodynamic fields appears to be a system size dependent standing wave with a period of about @xmath7 .
further details and discussion are given in section [ sec : monte ] .
in section [ sec : mole ] some preliminary attempts to see the instability at high densities using molecular dynamics for the hard sphere fluid are discussed . the system dimension in the direction of the spatial perturbation
is increased by an order of magnitude relative to previous simulations .
a long wavelength perturbation is found to grow on a very long simulation time , with no indication of approach to the steady uniform shear flow .
a quantitative comparison with theory at the required larger densities and shear rates is now possible using a recently developed kinetic model for the hard sphere enskog equation for application @xcite , although the details have not been worked out at this point .
the results of the theory and simulations are summarized and discussed in section [ sec : dis ] .
on sufficiently large space and time scales the dynamics of a fluid is well - described by hydrodynamic equations obtained from the exact conservation laws for the average mass , energy , and momentum densities , together with approximate constitutive equations for the associated fluxes . for the analysis here we use the density , @xmath8 , temperature , @xmath9 , and flow velocity , @xmath10 , as dependent variables rather than the density , energy density , and momentum .
the general form of these conservation laws is @xcite @xmath11 @xmath12 @xmath13 where @xmath14 is the material derivative .
the parameters occurring in these equations are the mass density , @xmath15 , the specific heat at constant volume , @xmath16 , the pressure , @xmath17 , the ratio of specific heats at constant pressure and volume , @xmath18 , and the coefficient of expansion , @xmath19 .
these parameters are the same functions of the local density and temperature as in equilibrium .
finally , the irreversible heat and momentum fluxes are denoted by @xmath20 and @xmath21 , respectively , and @xmath22 is the rate at which work is done by the external force representing the thermostat .
its detailed form will not be required in this section .
the above equations are incomplete until constitutive equations for the fluxes are specified in terms of the hydrodynamic fields .
however , the special solution of uniform shear flow exists independent of this choice .
it is defined by a spatially constant temperature , density , heat flux and momentum flux , and a flow velocity whose only non - vanishing component is @xmath23 .
the shear rate , @xmath3 , provides the single control parameter measuring the deviation from equilibrium .
the boundary conditions are simple periodic conditions in the local lagrangian coordinate frame , @xmath24 .
substitution of these assumptions for @xmath25 , @xmath26 , and @xmath27 into the above conservation laws shows that ( [ 1 ] ) and ( [ 3 ] ) are satisfied , while ( [ 2 ] ) reduces to @xmath28 this expresses the temperature evolution as a competition between the viscous heating effect , @xmath29 @xmath30 , and the cooling by the thermostat , @xmath31 .
a steady state is obtained by choosing the thermostat to cancel the viscous heating , @xmath32 the various thermostats described in the appendix all satisfy ( [ 5 ] ) but differ for states away from the steady state .
next consider the equations for small deviations of the hydrodynamic variables from the uniform shear flow state , retaining only terms linear in these deviations . to proceed
it is necessary to specify the constitutive equations for the heat and momentum fluxes . in this section , attention
is limited to small spatial gradients , including the shear rate , so that fourier s law and newton s viscosity law apply , @xmath33 @xmath34 here @xmath35 is the thermal conductivity , @xmath36 is the shear viscosity , and @xmath37 is the bulk viscosity .
it follows immediately that @xmath38 @xmath39 @xmath40 an abbreviated notation has been introduced where the subscript @xmath41 on a quantity indicates it is evaluated at @xmath42 . also , @xmath43 , @xmath44 , @xmath45 , @xmath46 , etc . with these results ,
the closed set of linear hydrodynamic equations for perturbations of uniform shear flow at small shear rates are given by @xmath47 @xmath48 @xmath49 in this section we choose a local thermostat for which @xmath50 in the temperature equation compensates for the excess viscous heating due to perturbations of the temperature and density , @xmath51 this does not imply a constant local temperature , however , except for spatially homogeneous deviations from uniform shear flow , although it does lead to a constant average temperature or kinetic energy for the whole system .
these differential equations have constant coefficients which suggests an equivalent algebraic form using fourier and laplace transformation .
consider first the fourier transform .
the boundary conditions are periodic in the local lagrangian frame given by @xmath52 , where @xmath53 , so it is appropriate to define the transform with respect to the variable @xmath54 , @xmath55 where @xmath56 denotes the set of perturbations , considered as a function @xmath54 in the first equality .
the periodic boundary conditions require @xmath57 , where @xmath58 are integers and @xmath59 are the linear dimensions of the system . however , since the time derivative in the hydrodynamic equations is taken at constant @xmath60 , the representation following the second equality is useful with @xmath61 . the fourier transformed hydrodynamic equations are @xmath62 the three matrices @xmath63 , @xmath64 , and @xmath65 are @xmath66 @xmath67 @xmath68 where @xmath69 is the unit vector along @xmath70 , @xmath71 , @xmath72 , and @xmath73 . to simplify the analysis attention
is restricted here and below to spatial perturbations only along the velocity gradient direction , i.e. @xmath74 . in this case
the linear hydrodynamic equations have time - independent coefficients ( i.e. , @xmath75 , @xmath76 and the matrices @xmath64 and @xmath65 simplify to @xmath77 @xmath78 equation ( [ feq ] ) can be solved by laplace transformation , @xmath79 with the result @xmath80_{\alpha \beta } ^{-1}\delta \widetilde{y}_\beta ( { \bf k},t=0 ) . \label{soly1}\ ] ] the eigenvalues @xmath81 of the matrix @xmath82 define the five simple hydrodynamic poles at @xmath83 .
the resulting five exponentials in time represent the hydrodynamic modes for relaxation of the perturbations around uniform shear flow . at @xmath84 ( perturbations of equilibrium )
they are the two sound modes , a heat mode , and a two - fold degenerate shear mode . for finite shear rate ,
the modes are more complicated and have qualitative differences . to illustrate , consider first the case of @xmath85 at fixed , finite @xmath3 , @xmath86 the coefficients in these expressions are real , @xmath87
@xmath88 @xmath89 there are two diffusive modes , @xmath90 , but the other modes are non - analytic about @xmath91 and represent more complex spatial dependence .
this behavior can be traced to the fact that the matrix @xmath92 is not normal and can not be diagonalized .
thus , at fixed @xmath93 there is a crossover in the transformation of @xmath94 from normal diagonal form to jordan form at small @xmath95 .
this is reflected in the eigenvalues if they are expanded in @xmath95 at fixed @xmath3 , as above .
for similar reasons , care must be used in representing @xmath96 as an expansion in the hydrodynamic modes with constant coefficients , since at small wavevectors there is a crossover to a mode expansion whose coefficients have algebraic time dependence .
the two propagating modes in ( [ mode2 ] ) are unstable , since @xmath97 .
the above navier - stokes analysis applies for small but finite shear rate , and small @xmath95 ( long wavelength ) .
the expansion in @xmath95 verifies that the asymptotic long wavelength modes are always unstable . at larger values of @xmath95
the modes are again stable , as follows from an exact evaluation of the eigenvalues .
there is a critical wavevector , @xmath4 , such that for @xmath98 the modes are stable whereas they are unstable otherwise .
these qualitative results apply without restriction to the atomic force law , density , or temperature .
figure [ fig_b1 ] shows @xmath4 as a function of @xmath3 for the special case of hard spheres at three densities , @xmath99 , @xmath100 , and @xmath101 ( @xmath95 and @xmath3 measured in units of the inverse mean free path and mean free time for the hard sphere boltzmann equation ) .
the thermodynamic properties are calculated using the percus - yevick approximation , while the transport coefficients are calculated from the enskog kinetic theory .
the instability is due to three matrix elements , @xmath102 , @xmath103 , and @xmath104 , and is therefore present at order @xmath95 .
the density perturbation is constant to this order and we choose @xmath105 to simplify the discussion .
the relevant variables are then the temperature perturbation @xmath106 , the longitudinal velocity perturbation @xmath107 , and the transverse velocity perturbation @xmath108 which to this order obey the equations @xmath109 @xmath110 the first equation has a coupling to the transverse velocity field due to the reference shear flow ; the second equation provides a feedback to the temperature equation through the same shear flow .
these couplings alone would lead simply to a renormalization of the sound velocity .
however , the shear flow also couples the transverse field to the longitudinal field for an additional feedback mechanism to the temperature equation through the pressure gradient .
this second mechanism is responsible for the instability .
for very long wavelengths the above equations can be simplified to give @xmath111 which exhibits the unstable modes of ( [ mode2 ] ) .
the hydrodynamic description can be derived from a more fundamental level of kinetic theory . in principle , this also allows derivation of hydrodynamic equations without the restriction of the navier - stokes approximation to small shear rates . a model kinetic theory for the practical calculation of such generalized hydrodynamic equations
is given in ref .
the resulting equations are limited to long wavelengths , as in the navier - stokes case , but the reference state of uniform shear flow can have a large shear rate .
an analysis of the hydrodynamic modes shows that there is a critical wavevector similar to that of figure [ fig_b1 ] .
it is possible to test this hydrodynamic description by a direct simulation of the more fundamental solution to the kinetic equation . for practical reasons the simulation is more efficient at larger wavevectors and shear rates than can be justified by navier - stokes hydrodynamics , and
this is the primary reason for considering the more complex generalized hydrodynamics .
the kinetic equation is a single relaxation time bgk equation @xcite given by @xmath112 here @xmath113 is the external force representing the thermostat , @xmath114 is the local equilibrium distribution , and @xmath115 is an average collision rate . the exact stationary solution for uniform shear flow in the presence of a thermostat has been studied in detail @xcite .
a variant of the chapman - enskog method can be used to study normal solutions for states near uniform shear flow , and to obtain the corresponding hydrodynamic equations for small perturbations relative to this state .
the complete details of this solution and the hydrodynamic modes as a function of the shear rate can be found in @xcite .
the method of simulation is a variant of the bird direct simulation method @xcite .
this is a monte carlo technique composed of two parts at each discrete time step : free streaming and collision .
the volume of the system is divided into cells of dimension smaller than the mean free path , and @xmath116 particles are introduced at @xmath117 with positions and velocities sampled statistically from a specified initial distribution function .
the distribution of particles is calculated at @xmath118 , with @xmath119 much smaller than the mean free time , as follows .
first the particles are displaced by @xmath120 .
next the velocity of each particle @xmath121 is replaced with probability @xmath122 by a random velocity sampled from the local equilibrium distribution , @xmath123 .
here @xmath124 , @xmath125 , and @xmath126 are the density , temperature , and flow velocity in the cell containing particle @xmath127 . in this collision stage ,
strict conservation of momentum and energy may be violated due to statistical fluctuations . to compensate for this artificial effect ,
the velocities of the particles in each cell are conveniently displaced and rescaled .
the whole process is then repeated for each subsequent time step . in this way ,
the one particle distribution function @xmath128 ( coarse grained over the cells ) is determined , from which the hydrodynamic fields can be computed directly as averages . in our simulations we have considered a system of size @xmath129 , with @xmath130 , along the @xmath131-direction at a shear rate @xmath132 .
here @xmath133 is the mean free time with @xmath134 ( maxwell molecules ) and @xmath135 is the thermal velocity . in the remainder of this section , we take @xmath136 , @xmath137 , @xmath138 , and @xmath139 .
lees - edwards boundary conditions are used to drive the shear flow @xcite .
these are simple periodic boundary conditions on both the position and velocity variables in the local lagrangian frame at @xmath140 .
both local and global thermostats have been studied .
the local thermostat is the same as that of section [ sec : navier ] except that the pressure tensor is no longer limited to its navier - stokes form .
in addition , two global thermostats are considered for more efficient implementation of the simulation .
these are described in the appendix .
the analysis there and the results of the simulation show that the instability is not sensitive to the choice of thermostat .
consequently , only the results using the global thermostats are presented here . starting from a distribution corresponding to uniform shear flow @xcite ,
the initial condition has been prepared by displacing and rescaling velocities so that @xmath141 with @xmath142 and @xmath143
. the technical parameters of the simulations are @xmath144 particles , a time step @xmath145 , and a cell width @xmath146 .
the data have been averaged over @xmath147 different realizations in the simulations of figs .
[ fig_d1 ] , [ fig_d3 ] , [ fig_d6 ] , and [ fig_d5 ] , and over 2 realizations in those of figs . [ fig_d2 ] and [ fig_d4 ] .
first we consider the short time dependence at a fixed position @xmath148 .
figure [ fig_d1 ] shows @xmath149 and @xmath150 .
the dashed lines are the results from the hydrodynamic analysis of the bgk model near uniform shear flow @xcite .
the good agreement up to @xmath151 shows that the instability is not just a consequence of the assumptions behind the hydrodynamic description .
the subsequent differences between simulation and theory are due to the fact that the latter is limited to small deviations from uniform shear flow . for longer times
we find both large amplitudes for @xmath152 and large deviations of the distribution function from that of the unperturbed state . to investigate the asymptotic state the system , we have performed the simulations for much longer times .
figure [ fig_d2 ] shows @xmath153 and @xmath154 for time up to @xmath155 .
both the velocity and the temperature oscillate in time and are modulated by a slowly varying amplitude relative to their asymptotic average values .
the maximum value @xmath156 corresponds to @xmath157 , which is close to the value @xmath4 at which the uniform shear flow with @xmath158 would become marginally stable @xcite .
thus the initial dynamics tends toward stabilization but does not ever cross over into the stable domain .
consequently , the asymptotic dynamics is not simply that of stationary uniform shear flow at a different temperature , but rather a quasi - stationary state with different spatial structure .
the thermostat used in the above simulations allows a global change in the average temperature , which is responsible for the amplitude modulations at long times . in order to have a more controlled asymptotic state ,
we have considered a variation of this thermostat that maintains the average temperature constant ( see the appendix ) .
the quantities @xmath159 and @xmath160 are plotted in fig .
[ fig_d3 ] . after a transient period of length @xmath161
, a stable oscillatory behavior of the velocity appears with a period @xmath162 .
the shape of @xmath163 over one cycle is shown in fig .
[ fig_d4 ] . here
@xmath164 , where @xmath165 , and the curves are averages over 20 successive cycles .
the value of @xmath166 has been chosen as to assure that the transient time is over and also with the criterion that @xmath167 at @xmath168 .
inspection of the results shows several regularities .
first , the following symmetry relation appears : @xmath169 where the minus sign applies to the velocity and the plus sign applies to the density and the temperature .
next , at times @xmath170 , where @xmath171 have extrema , @xmath172 , @xmath173 , and @xmath174 seem to have nodes .
note also that the vector @xmath175 rotates anti - clockwise and that most of the time @xmath176 is correlated to @xmath177 and @xmath178 is correlated to @xmath179 .
the above results indicate the wave character of the asymptotic state . to confirm this point ,
we have analyzed the profiles of the hydrodynamic quantities at relevant times .
the results are consistent with two independent invariance relations : @xmath180 @xmath181 their combination yields @xmath182 , which implies eq .
( [ d2 ] ) .
figure [ fig_d6 ] shows the spatial variation of @xmath183 and @xmath184 at @xmath185 .
not shown are times @xmath186 because they can be reproduced by use of the relation ( [ d3 ] ) . as observed in fig .
[ fig_d4 ] for the special point @xmath148 it is seen that a high ( low ) temperature is generally correlated to a high ( low ) density .
the spatial distribution of the temperature is highly non - uniform even though the thermostat maintains a constant average temperature .
figure [ fig_d5 ] shows a vector representation of the components of @xmath187 throughout the system at the values of @xmath188 . as anticipated from fig .
[ fig_d4 ] , @xmath189 rotates anti - clockwise throughout the system .
the layers @xmath190 are always nodes of @xmath191 and extrema of @xmath106 and @xmath192 .
the pattern indicated by figs .
[ fig_d6 ] and [ fig_d5 ] can be described as a periodic standing wave represented by the superposition of two symmetrical waves travelling in opposite directions . in summary , for initial values of @xmath95 and @xmath3 in the predicted unstable domain small perturbations of the hydrodynamic fields
grow according to the linear hydrodynamic equations for @xmath193 .
subsequently , non - linear effects invalidate this theoretical analysis .
the simulations show a transient period up to about @xmath161 , after which a quasi - stationary state is observed for @xmath194 . in this asymptotic state
the vector quantities oscillate at a period approximately twice that of the scalar fields .
the oscillations are spatially non - uniform for all fields considered .
the most extensive prior studies of uniform shear flow have been for dense systems via molecular dynamics ( md ) simulations .
no signature of the instability discussed here has been noted in these previous results .
there are two significant differences between the theory and simulation discussed above and the detailed implementation of earlier md simulations .
the first , and perhaps most important , has been the consideration of small system sizes relative to the wavelengths necessary to see the instability .
typically , system size is determined by the simulation time such that a sound wave will not traverse the system and generate correlations . at high densities
this has led to consideration of system sizes small compared to the critical wavevector for instability . a second difference from the discussion above
is the method for imposing a thermostat . in md simulations
it is efficient to impose the temperature control by a global re - scaling of the velocities only after as many as @xmath195 collisions .
however , at the shear rates considered this implies significant heating between applications of the thermstat and the temperature is more like a sawtooth in time rather than constant .
thus it is difficult to give a direct theoretical correspondence to the md simulation results . to demonstrate the existence of the instability using md simulations we have considered a rectangular unit cell with one side
expanded @xmath196 times larger than the other two ( @xmath197 hard spheres of diameter @xmath198 at a density of @xmath199 in a cell of size @xmath200 .
the lees - edwards boundary conditions are applied on the @xmath201 surfaces so that the velocity gradient is along the larger dimension of the system , allowing study of much longer wavelengths in the direction of the gradient than have previously been considered .
a second difference from previous simulations is a more frequent application of the velocity scaling to control the themperature and better represent continuous cooling . in our simulations
the velocities are rescaled whenever the temperature differs by 5% from its set value .
the simulation cell is divided into a fixed number of subcells , the local velocity field in each subcell is calculated and the excess kinetic energy computed relative to the local velocity field .
the velocities of the particles in each subcell are then rescaled so that the total excess kinetic energy of each subcell is equal to the set value .
the amount of rescaling is determined locally , and thus corresponds to a local thermostat , @xmath202 $ ] , as discussed in the appendix .
our implementation follows hess @xcite in that we assume uniformity of all quantities in the directions perpendicular to the velocity gradient so that the subcells are thin slabs and the only spatial variations are in the direction of the velocity gradient , as in the monte carlo simulations . due to the large number of particles in the simulations , the length of the simulations is relatively modest @xmath203 collisions .
the shear rate was fixed at @xmath204 and an initial perturbation with @xmath205 was monitored .
our theoretical estimates indicate this should correspond to conditions of instability .
figure [ fig_jfl_1]shows the @xmath206-component of the velocity field growing steadily throughout the simulation , clearly indicating the instability .
conversely , perturbations with a wavelength one quarter of this value appear to be stable as expected .
similar results are observed for the density field as well .
our primary conclusion from these preliminary results is that the instability can be observed and studied by md simulations if larger system sizes are considered and more care is taken with application of the thermostat .
uniform shear flow has been a prototype state for the study of fluids far from equilibrium , using both theoretical and computer simulation methods . until recently
, it has been assumed that this state is stable except at high densities and short wavelengths .
the new results reported here and in @xcite show that this simple macroscopic state is unstable at sufficiently large wavelengths .
previous studies via simulation have not seen this instability due to finite system sizes .
however , theoretical analysis at the hydrodynamic level clearly shows the mechanism and parameter space for this instability . in the present work this theoretical analysis is tested both qualitatively and quantitatively . at the qualitative level ,
both monte carlo simulations of a kinetic theory description and molecular dynamics simulation of the newtonian dynamics show clearly that this flow is unstable at long wavelengths .
the monte carlo simulation also confirms quantitatively the predictions of the theory on the time scale for which the linear analysis is valid . at longer times
the monte carlo simulation shows clearly that the asymptotic state is spatially non - uniform with a periodic variation in time .
a precise theoretical description of this final state has not been developed at this time .
the analysis here has been limited to spatial perturbations along the gradient of the velocity in the stationary state .
more general perturbations will lead to more complex flows due to the coupling to the convective flow .
a theoretical analysis of this more general case is in progress but is significantly more complex . due to the fact that the @xmath91 matrix of the hydrodynamic equations is non - diagonalizable ( non - normal )
, the corresponding eigenvalues do not fix a unique eigenspace .
the resulting dynamics is not simply a superposition of decaying or growing modes , but rather includes as well algebraic growth .
the resulting analysis of conditions for instability is more complex and will be reported elsewhere .
the research of m.l . and j.d .
was supported in part by nsf grant phy 9312723 .
the research of j.m.m . and
a.s . was supported in part by the dgicyt ( spain ) and by the junta de extremadura - fondo social europeo through grants no .
pb94 - 1021 and no .
eia94 - 39 , respectively .
partial support for this research also was provided by the division of sponsored research at the university of florida .
the analysis of section [ sec : navier ] made use of a specific choice for the thermostat .
other choices are possible and more convenient for computer simulation . in this work we use two different types of thermostats ,
both obtained from an external force at the microscopic level of the form @xmath207 .
\label{force}\ ] ] the corresponding source term @xmath208 in the temperature equation ( [ 2 ] ) is @xmath209 where @xmath210 is the kinetic energy density .
the thermostat parameter , @xmath211 , is always chosen to ensure constant temperature in the reference state , @xmath212 , as follows from ( [ 5 ] ) . in this work
we use three different thermostats , one local and two global .
the local thermostat is defined by @xmath213 , to linear order , so that @xmath214 the coefficients @xmath215 and @xmath216 are chosen such that all viscous heating proportional to @xmath217 and @xmath218 is compensated by the source term .
while there is still a local temperature variation , this thermostat holds the average temperature constant .
it is this local thermostat that is used in section [ sec : navier ] , with @xmath219 approximated by its navier - stokes form .
the other two thermostats considered are global , i.e. the parameter @xmath220 is spatially constant .
the simplest case is @xmath221 with a corresponding source term @xmath222 in this case there are greater effects due to viscous heating by the perturbations than with the choice @xmath223 , and even the average temperature of the system changes except at the stationary state .
a second global thermostat eliminates this global change using the form , @xmath224 , where @xmath225 and @xmath226 are the volume averaged density and temperature .
the source term in this case is @xmath227 .
\label{source3}\ ] ] in fourier representation the sources @xmath228 and @xmath229 are the same except for the @xmath91 dynamics .
both theory and the monte carlo simulations of section [ sec : monte ] indicate that the details of the perturbed dynamics depend on the thermostat used , but that the mechanisim for the instability is insensitive to the choice of thermostat . to illustrate this we repeat the analysis of section [ sec : navier ] using the global thermostat ( [ source2 ] ) ( again in the navier - stokes limit ) .
there are two new terms in the temperature equation .
one is due to temperature perturbations and leads to viscous heating even for uniform perturbations .
the second is due to density perturbations and provides a direct @xmath95 - independent coupling of the temperature and density equations ( at zero shear rate the density and temperature are only coupled indirectly via the gradients of the longitudinal velocity component and the pressure ) .
the matrices @xmath64 and @xmath65 are unchanged from section [ sec : navier ] , but these new couplings replace @xmath230 and @xmath231 in ( [ mata ] ) with non - zero values proportional to @xmath232 , @xmath233 where @xmath234 , @xmath235 .
the matrix @xmath92 is now diagonalizable and an expansion of the hydrodynamic modes in powers of @xmath95 rather than @xmath236 is obtained , @xmath237 with the coefficients , @xmath238 @xmath239 } { m\left [ n_sp_{s , n}\left ( a^2\eta _ { s , t}+w_t\right ) -p_{s , t}\left ( a^2n_s\eta _ { s , n}+n_sw_n+2\eta _ sa^2\right ) \right ] } , \ ] ] @xmath240 } { \rho _ s\left ( a^2\eta _ { s , t}+w_t\right ) } , \ ] ] @xmath241 , \ ] ] @xmath242 there are two diffusive modes , two propagating sound modes , and a time - modulated diffusive mode .
the sound modes are unstable at long wavelengths for shear rates satisfying @xmath243 .
this possibility has been explored in reference @xcite for the special case of hard spheres . in that case
both @xmath244 and @xmath245 are positive and the modes are unstable for sufficiently small shear rates .
the corresponding critical wavevector , @xmath4 , for this thermostat is determined from the exact eigenvalues and shown in figure [ fig_c1 ] for the same densities as in figure [ fig_b1 ] .
the domain of instability at long wavelengths is similar to that of figure [ fig_b1 ] using the thermostat @xmath223 .
the thermostat of this appendix allows greater viscous heating than that of section [ sec : navier ] for states perturbed from uniform shear flow .
the hydrodynamic modes are quite different , reflecting a sensitivity to the choice of thermostat .
these qualitative differences can be traced to the mathematical differences between diagonalizable and non - diagonalizable matrices , @xmath246 .
nevertheless , the mechanism for the instability described at the end of section [ sec : navier ] remains effective in both cases .
these conclusions are not limited to the navier - stokes approximation , but are confirmed as well for the kinetic model results described in section [ sec : monte ] for larger shear rates .
j. w. dufty , j. j. brey , and a. santos in _ molecular dynamics simulation of statistical mechanical systems _ , edited by g. ciccotti and w. g. hoover ( north holland publ .
co. , amsterdam , 1986 ) , pp .
. for extensive references see d. j. evans and w. hoover , ann .
fluid mech . * 18 * , 243 ( 1986 ) ; w. loose and s. hess in _ microscopic simulation of complex flows _ , m. mareschal ed .
( plenum press , new york , 1990 ) , pp .
267 - 278 .
j. f. lutsko and j. w. dufty , phys .
lett . * 57 * , 2775 ( 1986 ) ; j. f. lutsko , j. w. dufty , and s. p. das , phys . rev . a * 39
* , 1311 ( 1989 ) ; t. kirkpatrick and j. c. nieuwoudt , phys . rev .
lett . * 56*,885 ( 1986 ) . | the stability of idealized shear flow at long wavelengths is studied in detail . a hydrodynamic analysis at the level of the navier - stokes equation for small shear rates is given to identify the origin and universality of an instability at any finite shear rate for sufficiently long wavelength perturbations .
the analysis is extended to larger shear rates using a low density model kinetic equation .
direct monte carlo simulation of this equation is compared with a hydrodynamic description including non newtonian rheological effects .
the hydrodynamic description of the instability is in good agreement with the direct monte carlo simulation for @xmath0 , where @xmath1 is the mean free time .
longer time simulations up to @xmath2 are used to identify the asymptotic state as a spatially non - uniform quasi - stationary state . finally , preliminary results from molecular dynamics simulation showing the instability are presented and discussed . |
the optimal base problem is all about finding an efficient representation for a given collection of positive integers .
one measure for the efficiency of such a representation is the sum of the digits of the numbers .
consider for example the decimal numbers @xmath1 .
the sum of their digits is 25 .
taking binary representation we have @xmath2 @xmath3 and the sum of digits is 13 , which is smaller .
taking ternary representation gives @xmath4 with an even smaller sum of digits , 12 .
considering the _ mixed radix _
base @xmath5 , the numbers are represented as @xmath6 @xmath7 and the sum of the digits is 9 .
the optimal base problem is to find a ( possibly mixed radix ) base for a given sequence of numbers to minimize the size of the representation of the numbers . when measuring size as `` sum of digits '' , the base @xmath8 is indeed optimal for the numbers of @xmath9 . in this paper
we present the optimal base problem and illustrate why it is relevant to the encoding of pseudo - boolean constraints to sat .
we also present an algorithm and show that our implementation is superior to current implementations .
pseudo - boolean constraints take the form @xmath10 , where @xmath11 are integer coefficients , @xmath12 are boolean literals ( i.e. , boolean variables or their negation ) , and @xmath13 is an integer .
we assume that constraints are in pseudo - boolean normal form @xcite , that is , the coefficients @xmath14 and @xmath13 are always positive and boolean variables occur at most once in @xmath15 .
pseudo - boolean constraints are well studied and arise in many different contexts , for example in verification @xcite and in operations research @xcite .
typically we are interested in the satisfiability of a conjunction of pseudo - boolean constraints .
since 2005 there is a series of pseudo - boolean evaluations @xcite which aim to assess the state of the art in the field of pseudo - boolean solvers .
we adopt these competition problems as a benchmark for the techniques proposed in this paper .
pseudo - boolean constraint satisfaction problems are often reduced to sat .
many works describe techniques to encode these constraints to propositional formulas @xcite .
the pseudo - boolean solver minisat@xmath0 ( @xcite , cf.http://minisat.se ) chooses between three techniques to generate sat encodings for pseudo - boolean constraints .
these convert the constraint to : ( a ) a bdd structure , ( b ) a network of sorters , and ( c ) a network of ( binary ) adders .
the network of adders is the most concise encoding , but it has the weakest propagation properties and often leads to higher sat solving times than the bdd based encoding , which , on the other hand , generates the largest encoding .
the encoding based on sorting networks is often the one applied and is the one we consider in this paper .
r42 mm @xmath16 to demonstrate how sorters can be used to translate pseudo - boolean constraints , consider the constraint @xmath17 where the sum of the coefficients is 8 . on the right , we illustrate an @xmath18 sorter where @xmath19 are each fed into a single input , @xmath20 into two of the inputs , and @xmath21 into three of the inputs .
the outputs are @xmath22 .
first , we represent the sorting network as a boolean formula , @xmath23 , which in general , for @xmath24 inputs , will be of size @xmath25 @xcite .
then , to assert @xmath26 we take the conjunction of @xmath23 with the formula @xmath27 .
but what happens if the coefficients in a constraint are larger than in this example ?
how should we encode @xmath28 ?
how should we handle very large coefficients ( larger than 1,000,000 ) ? to this end , the authors in @xcite generalize the above idea and propose to decompose the constraint into a number of interconnected sorting networks .
each sorter represents a digit in a mixed radix base .
this construction is governed by the choice of a suitable mixed radix base and the objective is to find a base which minimizes the size of the sorting networks . here
the optimal base problem comes in , as the size of the networks is directly related to the size of the representation of the coefficients .
we consider the sum of the digits ( of coefficients ) and other measures for the size of the representations and study their influence on the quality of the encoding . in minisat@xmath0 the search for an optimal base
is performed using a brute force algorithm and the resulting base is constructed from prime numbers up to 17 . the starting point for this paper
is the following remark from @xcite ( footnote 8) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this is an ad - hoc solution that should be improved in the future .
finding the optimal base is a challenging optimization problem in its own right .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in this paper we take the challenge and present an algorithm which scales to find an optimal base consisting of elements with values up to 1,000,000 .
we illustrate that in many cases finding a better base leads also to better sat solving time .
section [ section : obp ] provides preliminary definitions and formalizes the optimal base problem .
section [ sec : encoding ] describes how minisat@xmath0decomposes a pseudo - boolean constraint with respect to a given mixed radix base to generate a corresponding propositional encoding , so that the constraint has a solution precisely when the encoding has a model .
section [ section:4 ] is about ( three ) alternative measures with respect to which an optimal base can be found .
sections [ sec : ob1][sec : ob3 ] introduce our algorithm based on classic ai search methods ( such as cost underapproximation ) in three steps : heuristic pruning , best - first branch and bound , and base abstraction .
sections [ sec : exp ] and [ relwork ] present an experimental evaluation and some related work . section [ sec : conc ] concludes .
proofs are given in the appendix .
in the classic base @xmath29 radix system , positive integers are represented as finite sequences of digits @xmath30 where for each digit @xmath31 , and for the most significant digit , @xmath32 .
the integer value associated with @xmath33 is @xmath34 .
a mixed radix system is a generalization where a base is an infinite radix sequence @xmath35 of integers where for each radix , @xmath36 and for each digit , @xmath37 .
the integer value associated with @xmath33 is @xmath38 where @xmath39 and for @xmath40 , @xmath41 .
the sequence @xmath42 specifies the weighted contribution of each digit position and is called the _
weight sequence of @xmath8_. a finite mixed radix base is a finite sequence @xmath43 with the same restrictions as for the infinite case except that numbers always have @xmath44 digits ( possibly padded with zeroes ) and there is no bound on the value of the most significant digit , @xmath45 .
in this paper we focus on the representation of finite multisets of natural numbers in finite mixed radix bases .
let @xmath46 denote the set of finite mixed radix bases and @xmath47 the set of finite non - empty multisets of natural numbers .
we often view multisets as ordered ( and hence refer to their first element , second element , etc . ) . for a finite sequence or multiset @xmath9 of natural numbers ,
we denote its length by @xmath48 , its maximal element by @xmath49 , its @xmath50 element by @xmath51 , and the multiplication of its elements by @xmath52 ( if @xmath9 is the empty sequence then @xmath53 ) .
if a base consists of prime numbers only , then we say that it is a prime base .
the set of prime bases is denoted @xmath54 .
let @xmath55 with @xmath56 .
we denote by @xmath57 the representation of a natural number @xmath58 in base @xmath8 .
the most significant digit of @xmath59 , denoted @xmath60 , is @xmath45 .
if @xmath61 then we say that @xmath8 is redundant for @xmath58 .
let @xmath62 with @xmath63 .
we denote the @xmath64 matrix of digits of elements from @xmath9 in base @xmath8 as @xmath65 .
namely , the @xmath50 row in @xmath65 is the vector @xmath66 .
the most significant digit column of @xmath65 is the @xmath44 column of the matrix and denoted @xmath67 .
if @xmath68 , then we say that @xmath8 is redundant for @xmath9 .
this is equivalently characterized by @xmath69 .
[ def : nrb ] let @xmath62 .
we denote the set of non - redundant bases for @xmath9 , @xmath70 .
the set of non - redundant prime bases for @xmath9 is denoted @xmath71 .
the set of non - redundant ( prime ) bases for @xmath9 , containing elements no larger than @xmath72 , is denoted @xmath73 ( @xmath74 ) .
the set of bases in @xmath75/@xmath73/@xmath74 , is often viewed as a tree with root @xmath76 ( the empty base ) and an edge from @xmath8 to @xmath77 if and only if @xmath77 is obtained from @xmath8 by extending it with a single integer value .
[ def : sumdigits ] let @xmath62 and @xmath55 .
the sum of the digits of the numbers from @xmath9 in base @xmath8 is denoted @xmath78 .
the usual binary `` base 2 '' and ternary `` base 3 '' are represented as the infinite sequences @xmath79 and @xmath80 . the finite sequence @xmath81 and the empty sequence @xmath82 are also bases .
the empty base is often called the `` unary base '' ( every number in this base has a single digit ) . let @xmath83 . then , @xmath84 , @xmath85 , @xmath86 , and @xmath87 . let @xmath62 .
a cost function for @xmath9 is a function @xmath88 which associates bases with real numbers .
an example is @xmath89 . in this paper
we are concerned with the following _ optimal base problem_. let @xmath90 and @xmath91 a cost function .
we say that a base @xmath8 is an _
optimal base for @xmath9 _ with respect to @xmath91 , if for all bases @xmath77 , @xmath92 .
the corresponding _ optimal base problem _ is to find an optimal base @xmath8 for @xmath9 .
the following two lemmata confirm that for the @xmath93 cost function , we may restrict attention to non - redundant bases involving prime numbers only .
[ l1 ] let @xmath90 and consider the @xmath93 cost function . then , @xmath9 has an optimal base in @xmath75 .
[ lem : primes ] let @xmath90 and consider the @xmath93 cost function . then , @xmath9 has an optimal base in @xmath71 .
how hard is it to solve an instance of the optimal base problem ( namely , for @xmath62 ) ?
the following lemma provides a polynomial ( in @xmath49 ) upper bound on the size of the search space .
this in turn suggests a pseudo - polynomial time brute force algorithm ( to traverse the search space ) .
[ zeta ] let @xmath62 with @xmath94 .
then , @xmath95 where @xmath96 and where @xmath97 is the riemann zeta function .
chor _ et al .
_ prove in @xcite that the number of ordered factorizations of a natural number @xmath24 is less than @xmath98 .
the number of bases for all of the numbers in @xmath9 is hence bounded by @xmath99 , which is bounded by @xmath100 .
this section presents the construction underlying the sorter based encoding of pseudo - boolean constraints applied in minisat@xmath0@xcite .
it is governed by the choice of a mixed radix base @xmath8 , the optimal selection of which is the topic of this paper .
the construction sets up a series of sorting networks to encode the digits , in base @xmath8 , of the sum of the terms on the left side of a constraint @xmath101 .
the encoding then compares these digits with those from @xmath102 from the right side .
we present the construction , step by step , through an example where @xmath103 and @xmath104 .
+ r30 mm @xmath105 the coefficients of @xmath26 form a multiset @xmath106 and their representation in base @xmath8 , a @xmath107 matrix , @xmath65 , depicted on the right .
the rows of the matrix correspond to the representation of the coefficients in base @xmath8 .
representing the coefficients as four digit numbers in base @xmath103 and considering the values @xmath108 of the digit positions , we obtain a decomposition for the left side of @xmath26 : @xmath109 to encode the sums at each digit position ( @xmath110 ) , we set up a series of four sorting networks as depicted below .
given values for the variables , the sorted outputs from these networks represented unary numbers @xmath112 such that the left side of @xmath26 takes the value @xmath113 .
for the outputs @xmath112 to represent the digits of a number in base @xmath103 , we need to encode also the `` carry '' operation from each digit position to the next .
the first 3 outputs must represent valid digits for @xmath8 , i.e. , unary numbers less than @xmath114 respectively . in our example
the single potential violation to this restriction is @xmath115 , which is represented in 6 bits . to this end
we add two components to the encoding : ( 1 ) each third output of the second network ( @xmath116 and @xmath117 in the diagram ) is fed into the third network as an additional ( carry ) input ; and ( 2 ) clauses are added to encode that the output of the second network is to be considered modulo 3 .
we call these additional clauses a _ normalizer_. the normalizer defines two outputs @xmath119 and introduces clauses specifying that the ( unary ) value of @xmath120 equals the ( unary ) value of @xmath121 .
the outputs from these four units now specify a number in base @xmath8 , each digit represented in unary notation . this number is now compared ( via an encoding of the lexicographic order ) to @xmath122 ( the value from the right - hand side of @xmath26 ) .
we now return to the objective of this paper : for a given pseudo - boolean constraint , how can we choose a mixed radix base with respect to which the encoding of the constraint via sorting networks will be optimal ?
we consider here three alternative cost functions with respect to which an optimal base can be found .
these cost functions capture with increasing degree of precision the actual size of the encodings .
the first cost function , @xmath93 as introduced in definition [ def : sumdigits ] , provides a coarse measure on the size of the encoding .
it approximates ( from below ) the total number of input bits in the network of sorting networks underlying the encoding .
an advantage in using this cost function is that there always exists an optimal base which is prime .
the disadvantage is that it ignores the carry bits in the construction , and as such is not always a precise measure for optimality . in @xcite ,
the authors propose to apply a cost function which considers also the carry bits .
this is the second cost function we consider and we call it @xmath123 .
[ cost2 ] let @xmath62 , @xmath55 with @xmath56 and @xmath124 the corresponding @xmath64 matrix of digits .
denote the sequences @xmath125 ( sums ) and @xmath126 ( carries ) defined by : @xmath127 for @xmath128 , @xmath129 , and @xmath130 for + @xmath128 .
the `` sum of digits with carry '' cost function is defined by the equation on the right .
@xmath131 the following example illustrates the @xmath123 cost function and that it provides a better measure of base optimality for the ( size of the ) encoding of pseudo - boolean constraints .
[ runningd ] consider the encoding of a pseudo - boolean constraint with coefficients @xmath132 with respect to bases : @xmath133 , @xmath134 , and @xmath135 .
figure [ fig:3cost ] depicts the sizes of the sorting networks for each of these bases .
the upper tables illustrate the representation of the coefficients in the corresponding bases . in the lower tables ,
the rows labeled `` sum '' indicate the number of bits per network and ( to their right ) their total sum which is the @xmath93 cost .
with respect to the @xmath93 cost function , all three bases are optimal for @xmath9 , with a total of 9 inputs .
the algorithm might as well return @xmath136 .
the rows labeled `` carry '' indicate the number of carry bits in each of the constructions and ( to their right ) their totals . with respect to the @xmath123 cost function , bases @xmath137 and @xmath138 are optimal for @xmath9 , with a total of @xmath139 bits while @xmath136 involves @xmath140 bits .
the algorithm might as well return @xmath137 .
the following example shows that when searching for an optimal base with respect to the @xmath123 cost function , one must consider also non - prime bases .
consider again the pseudo boolean constraint @xmath142 from section [ sec : encoding ] .
the encoding with respect to @xmath133 results in 4 sorting networks with 10 inputs from the coefficients and 2 carries .
so a total of 12 bits .
the encoding with respect to @xmath143 is smaller .
it has the same 10 inputs from the coefficients but no carry bits .
base @xmath138 is optimal and non - prime .
we consider a third cost function which we call the @xmath144 cost function .
sorting networks are constructed from `` comparators '' @xcite and in the encoding each comparator is modeled using six cnf clauses
. this function counts the number of comparators in the construction .
let @xmath145 denote the number of comparators in an @xmath146 sorting network . for small values of @xmath147
, @xmath145 takes the values @xmath148 and @xmath149 respectively which correspond to the sizes of the optimal networks of these sizes @xcite . for larger values
, the construction uses batcher s odd - even sorting networks @xcite for which @xmath150 .
[ num_comparators ] consider the same setting as in definition [ cost2 ] .
then , @xmath151 consider again the setting of example [ runningd ] .
in figure [ fig:3cost ] the rows labeled `` comp '' indicate the number of comparators in each of the sorting networks and their totals .
the construction with the minimal number of comparators is that obtained with respect to the base @xmath134 with 10 comparators .
it is interesting to remark the following relationship between the three cost functions : the @xmath93 function is the most `` abstract '' .
it is only based on the representation of numbers in a mixed radix base .
the @xmath123 function considers also properties of addition in mixed - radix bases ( resulting in the carry bits ) .
finally , the @xmath144 function considers also implementation details of the odd - even sorting networks applied in the underlying minisat@xmath0 construction . in section [ sec : exp ]
we evaluate how the alternative choices for a cost function influence the size and quality of the encodings obtained with respect to corresponding optimal bases .
this section introduces a simple , heuristic - based , depth - first , tree search algorithm to solve the optimal base problem .
the search space is the domain of non - redundant bases as presented in definition [ def : nrb ] .
the starting point is the brute force algorithm applied in minisat@xmath0 . for a sequence of integers @xmath9 ,
minisat@xmath0 applies a depth - first traversal of @xmath152 to find the base with the optimal value for the cost function @xmath153 .
our first contribution is to introduce a heuristic function and to identify branches in the search space which can be pruned early on in the search .
each tree node @xmath8 encountered during the traversal is inspected to check if given the best node encountered so far , @xmath154 , it is possible to determine that all descendants of @xmath8 are guaranteed to be less optimal than @xmath154 . in this case
, the subtree rooted at @xmath8 may be pruned .
the resulting algorithm improves on the one of minisat@xmath0 and provides the basis for the further improvements introduced in sections [ sec : ob2 ] and [ sec : ob3 ] .
we need first a definition .
let @xmath62 , @xmath155 , and @xmath91 a cost function .
we say that : ( 1 ) @xmath77 extends @xmath8 , denoted @xmath156 , if @xmath8 is a prefix of @xmath77 , ( 2 ) @xmath157 is a partial cost function for @xmath91 if @xmath158 , and ( 3 ) @xmath159 is an admissible heuristic function for @xmath91 and @xmath157 if @xmath160 .
the intuition is that @xmath161 signifies a part of the cost of @xmath8 which will be a part of the cost of any extension of @xmath8 , and that @xmath162 is an under - approximation on the additional cost of extending @xmath8 ( in any way ) given the partial cost of @xmath8 .
we denote @xmath163 .
if @xmath157 is a partial cost function and @xmath159 is an admissible heuristic function , then @xmath164 is an under - approximation of @xmath165 .
the next lemma provides the basis for heuristic pruning using the three cost functions introduced above .
[ lem : h ] the following are admissible heuristics for the cases when : 1 .
@xmath89 : @xmath166 .
@xmath167 : @xmath166 .
@xmath168 : @xmath169 . in the first two settings we take @xmath170 .
+ in the case of @xmath144 we take the trivial heuristic estimate @xmath171 /*input*/ & multiset s + /*init*/ & base bestb = @xmath172 + /*dfs*/ & depth - first traverse @xmath173 + & at each node @xmath8 , for the next value p @xmath174 b.extenders(s ) do + & base newb = b.extend(p ) + & if ( @xmath175 ) * prune * + & else if ( @xmath176 ) bestb = newb + /*output*/ & return bestb ; + the algorithm , which we call ` dfshp ` for depth - first search with heuristic pruning , is now stated as figure [ fig : alg1 ] where the input to the algorithm is a ` multiset ` of integers ` s ` and the output is an optimal base .
the algorithm applies a depth - first traversal of @xmath75 in search of an optimal base .
we assume given : a cost function @xmath91 , a partial cost function @xmath157 and an admissible heuristic @xmath159 .
we denote @xmath163 .
the abstract data type ` base ` has two operations : ` extend(int ) ` and ` extenders(multiset ) ` . for a base ` b ` and an integer ` p ` , ` b.extend(p ) ` is the base obtained by extending ` b ` by ` p ` . for a multiset ` s ` , ` b.extenders(s ) ` is the set of integer values ` p ` by which ` b ` can be extended to a non - redundant base for ` s ` , i.e. , such that @xmath177 .
the definition of this operation may have additional arguments to indicate if we seek a prime base or one containing elements no larger than @xmath72 .
initialization ( /*init*/ in the figure ) assigns to the variable ` bestb ` a finite binary base of size @xmath178 .
this variable will always denote the best base encountered so far ( or the initial finite binary base ) . throughout the traversal , when visiting a node ` newb ` we first check if the subtree rooted at ` newb ` should be pruned .
if this is not the case , then we check if a better `` best base so far '' has been found . once the entire ( with pruning )
search space has been considered , the optimal base is in ` bestb ` . to establish a bound on the complexity of the algorithm , denote the number of different integers in @xmath9 by @xmath179 and @xmath94 .
the algorithm has space complexity @xmath180 , for the depth first search on a tree with height bound by @xmath181 ( an element of @xmath75 will have at most @xmath182 elements ) .
for each base considered during the traversal , we have to calculate @xmath183 which incurs a cost of @xmath184 . to see why , consider that when extending a base @xmath8 by a new element giving base @xmath77 , the first columns of @xmath185 are the same as those in @xmath65 ( and thus also the costs incurred by them ) .
only the cost incurred by the most significant digit column of @xmath65 needs to be recomputed for @xmath185 due to base extension of @xmath8 to @xmath77 .
performing the computation for this column , we compute a new digit for the @xmath179 different values in @xmath9 .
finally , by lemma [ zeta ] , there are @xmath186 bases and therefore , the total runtime is @xmath187 . given that @xmath188 , we can conclude that runtime is bounded by @xmath189 .
in this section we further improve the search algorithm for an optimal base .
the search algorithm is , as before , a traversal of the search space using the same partial cost and heuristic functions as before to prune the tree .
the difference is that instead of a depth first search , we maintain a priority queue of nodes for expansion and apply a best - first , branch and bound search strategy .
figure [ fig : alg2 ] illustrates our enhanced search algorithm .
we call it ` b&b ` . the abstract data
type ` priority_queue ` maintains bases prioritized by the value of @xmath190 .
operations ` popmin ( ) ` , ` push(base ) ` and ` peek ( ) ` ( peeks at the minimal entry ) are the usual .
the reason to box the text `` ` priority_queue ` '' in the figure will become apparent in the next section .
base findbase(multiset s ) + /*1*/ base bestb = @xmath172 ; q = @xmath191 ; + /*2*/ while ( @xmath192 @xmath193 ) + /*3*/ base b = q.popmin ( ) ; + /*4*/ foreach ( p @xmath174 b.extenders(s ) ) + /*5*/ base newb = b.extend(p ) ; + /*6*/ if ( @xmath194 ) + /*7*/ \ { q.push(newb ) ; if ( @xmath195 ) bestb = newb ; } + /*8*/ return bestb ; + on line /*1*/ in the figure , we initialize the variable ` bestb ` to a finite binary base of size @xmath178 ( same as in figure [ fig : alg1 ] ) and initialize the queue to contain the root of the search space ( the empty base ) . as long as there are still nodes to be expanded in the queue that are potentially interesting ( line /*2*/ ) , we select ( at line /*3*/ ) the best candidate base ` b ` from the frontier of the tree in construction for further expansion .
now the search tree is expanded for each of the relevant integers ( calculated at line /*4*/ ) . for each child ` newb ` of ` b ` ( line /*5*/ ) , we check if pruning at ` newb ` should occur ( line /*6*/ ) and if not we check if a better bound has been found ( line /*7*/ ) finally , when the loop terminates , we have found the optimal base and return it ( line /*8*/ ) .
this section introduces an abstraction on the search space , classifying bases according to their product . instead of maintaining ( during the search ) a priority queue of all bases ( nodes ) that still need to be explored , we maintain a special priority queue in which there will only ever be at most one base with the same product .
so , the queue will never contain two different bases @xmath137 and @xmath138 such that @xmath196 . in case a second base , with the same product as one already in ,
is inserted to the queue , then only the base with the minimal value of @xmath190 is maintained on the queue .
we call this type of priority queue a _ hashed priority queue _ because it can conveniently be implemented as a hash table .
the intuition comes from a study of the @xmath93 cost function for which we can prove the following * property 1 * on bases : consider two bases @xmath137 and @xmath138 such that @xmath196 and such that @xmath197 . then for any extension of @xmath137 and of @xmath138 by the same sequence @xmath198 , @xmath199 .
in particular , if one of @xmath137 or @xmath138 can be extended to an optimal base , then @xmath137 can .
a direct implication is that when maintaining the frontier of the search space as a priority queue , we only need one representative of the class of bases which have the same product ( the one with the minimal value of @xmath190 ) .
a second * property 2 * is more subtle and true for any cost function that has the first property : assume that in the algorithm described as figure [ fig : alg2 ] we at some stage remove a base @xmath137 from the priority queue .
this implies that if in the future we encounter any base @xmath138 such that @xmath196 , then we can be sure that @xmath200 and immediately prune the search tree from @xmath138 .
our third and final algorithm , which we call ` hashb&b ` ( best - first , branch and bound , with hash priority queue ) is identical to the algorithm presented in figure [ fig : alg2 ] , except that the the boxed priority queue introduced at line /*1*/ is replaced by a . the abstract data type ` hash_priority_queue ` maintains bases prioritized by the value of @xmath190 . operations ` popmin ( ) ` and ` peek ( ) ` are as usual . operation ` push(b_1 ) ` works as follows : ( a ) if there is no base @xmath138 in the queue such that @xmath201 , then add @xmath137 . otherwise , ( b ) if @xmath202 then do not add @xmath137 .
otherwise , ( c ) remove @xmath138 from the queue and add @xmath137 . [ algisgood ] + ( 1 ) the @xmath93 cost function satisfies * property 1 * ; and ( 2 ) the ` hashb&b ` algorithm finds an optimal base for any cost function which satisfies * property 1*. we conjecture that the other cost functions do not satisfy * property 1 * , and hence can not guarantee that the ` hashb&b ` algorithm always finds an optimal base .
however , in our extensive experimentation , all bases found ( when searching for an optimal prime base ) are indeed optimal .
a direct implication of the above improvements is that we can now provide a tighter bound on the complexity of the search algorithm .
let us denote the number of different integers in @xmath9 by @xmath179 and @xmath94 .
first note that in the worst case the hashed priority queue will contain @xmath203 elements ( one for each possible value of a base product , which is never more than @xmath203 ) .
assuming that we use a fibonacci heap , we have a @xmath180 cost ( amortized ) per ` popmin ( ) ` operation and in total a @xmath204 cost for popping elements off the queue during the search for an optimal base .
now focus on the cost of operations performed when extracting a base @xmath8 from the queue . denoting @xmath205
, @xmath8 has at most @xmath206 children ( integers which extend it ) . for each child
we have to calculate @xmath183 which incurs a cost of @xmath184 and possibly to insert it to the queue . pushing
an element onto a hashed priority queue ( in all three cases ) is a constant time operation ( amortized ) , and hence the total cost for dealing with a child is @xmath184 . finally , consider the total number of children created during the search which corresponds to the following sum : @xmath207 so , in total we get @xmath208 .
when we restrict the extenders to be prime numbers then we can further improve this bound to @xmath209 by reasoning about the density of the primes . a proof can be found in the appendix
experiments are performed using an extension to minisat@xmath0@xcite where the only change to the tool is to plug in our optimal base algorithm .
the reader is invited to experiment with the implementation via its web interface .
all experiments are performed on a quad - opteron 848 at 2.2 ghz , 16 gb ram , running linux .
our benchmark suite originates from 1945 pseudo - boolean evaluation @xcite instances from the years 20062009 containing a total of 74,442,661 individual pseudo - boolean constraints . after normalizing and removing constraints with @xmath210 coefficients
we are left with 115,891 different optimal base problems where the maximal coefficient is @xmath211 .
we then focus on 734 pb instances where at least one optimal base problem from the instance yields a base with an element that is non - prime or greater than 17 .
when solving pb instances , in all experiments , a 30 minute timeout is imposed as in the pseudo - boolean competitions . when solving an optimal base problem , a 10 minute timeout is applied .
[ [ experiment-1-impact - of - optimal - bases ] ] experiment 1 ( impact of optimal bases ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the first experiment illustrates the advantage in searching for an optimal base for pseudo - boolean solving .
we compare sizes and solving times when encoding w.r.t . the binary base vs. w.r.t . an optimal base ( using the ` hashb&b ` algorithm with the @xmath144 cost function ) .
encoding w.r.t . the binary base , we solve 435 pb instances ( within the time limit ) with an average time of 146 seconds and average cnf size of 1.2 million clauses . using an optimal base ,
we solve 445 instances with an average time of 108 seconds , and average cnf size of 0.8 million clauses .
[ [ experiment-2-base - search - time ] ] experiment 2 ( base search time ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + here we focus on the search time for an optimal base in six configurations using the @xmath123 cost function . configurations ` m17 ` , ` dfshp17 ` , and ` b&b17 ` , are respectively , the minisat@xmath0 implementation , our ` dfshp ` and our ` b&b ` algorithms , all three searching for an optimal base from @xmath212 , i.e. , with prime elements up to 17 .
configurations ` hashb&b1,000,000 ` , ` hashb&b10,000 ` , and ` hashb&b17 ` are our ` hashb&b ` algorithm searching for a base from @xmath213 with bounds of @xmath214 1,000,000 , @xmath215 10,000 , and @xmath215 17 , respectively .
results are summarized in fig .
[ fig : results1 ] which is obtained as follows .
we cluster the optimal base problems according to the values @xmath216 where @xmath217 is the maximal coefficient in a problem .
then , for each cluster we take the average runtime for the problems in the cluster .
the value @xmath218 is chosen to minimize the standard deviation from the averages ( over all clusters ) .
these are the points on the graphs .
configuration ` m17 ` times out on 28 problems . for ` dfshp17 ` , the maximal search time is 200 seconds .
configuration ` b&b17 ` times out for 1 problem .
the ` hashb&b ` configurations have maximal runtimes of 350 seconds , 14 seconds and 0.16 seconds , respectively for the bounds 1,000,000 , 10,000 and 17 .
-axis with 50k ms on the left and 8k ms on the right .
configuration ` dfshp17 ` ( yellow ) is lowest on left and highest on right , setting the reference point to compare the two graphs .
, title="fig : " ] -axis with 50k ms on the left and 8k ms on the right .
configuration ` dfshp17 ` ( yellow ) is lowest on left and highest on right , setting the reference point to compare the two graphs .
, title="fig : " ] fig .
[ fig : results1 ] shows that : ( left ) even with primes up to 1,000,000 , ` hashb&b ` is faster than the algorithm from minisat@xmath0 with the limit of 17 ; and ( right ) even with primes up to 10,000 , the search time using ` hashb&b ` is essentially negligible .
[ [ experiment-3-impact - on - pb - solving ] ] experiment 3 ( impact on pb solving ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + fig .
[ fig : results2 ] illustrates the influence of improved base search on sat solving for pb constraints .
both graphs depict the number of instances solved ( the @xmath219-axis ) within a time limit ( the @xmath220-axis ) . on the left ,
total solving time ( with base search ) , and on the right , sat solving time only .
[ fig : results2 ] both graphs consider the @xmath221 instances of interest and compare sat solving times with bases found using five configurations .
the first is minisat@xmath0with configuration ` m17 ` , the second is with respect to the binary base , the third to fifth are ` hashb&b ` searching for bases from @xmath222 with cost functions : @xmath93 , @xmath123 , and @xmath144 , respectively .
the average total / solve run - times ( in sec ) are 150/140 , 146/146 , 122/121 , 116/115 and 108/107 ( left to right ) .
the total number of instances solved are 431 , 435 , 442 , 442 and 445 ( left to right ) .
the average cnf sizes ( in millions of clauses ) for the entire test set / the set where all algorithms solved / the set where no algorithm solved are 7.7/1.0/18 , 9.5/1.2/23 , 8.4/1.1/20 , 7.2/0.8/17 and 7.2/0.8/17 ( left to right ) .
the graphs of fig .
[ fig : results2 ] and average solving times clearly show : * ( 1 ) * sat solving time dominates base finding time , * ( 2 ) * minisat@xmath0 is outperformed by the trivial binary base , * ( 3 ) * total solving times with our algorithms are faster than with the binary base , and * ( 4 ) * the most specific cost function ( comparator cost ) outperforms the other cost functions both in solving time and size . finally , note that sum of digits with its nice mathematical properties , simplicity , and application independence solves as many instances as cost carry . [ [ experiment-4-impact - of - high - prime - factors ] ] experiment 4 ( impact of high prime factors ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + r73 mm this experiment is about the effects of restricting the maximal prime value in a base ( i.e. the value @xmath223 of minisat@xmath0 ) .
an analysis of the our benchmark suite indicates that coefficients with small prime factors are overrepresented . to introduce instances where coefficients have larger prime factors we select 43 instances from the suite and multiply their coefficients to introduce the prime factor 31 raised to the power @xmath224 .
we also introduce a slack variable to avoid gcd - based simplification .
this gives us a collection of 258 new instances .
we used the ` b&b ` algorithm with the @xmath123 cost function applying the limit @xmath223 ( as in minisat@xmath0 ) and @xmath225 .
results indicate that for @xmath225 , both cnf size and sat - solving time are independent of the factor @xmath226 introduced for @xmath227 .
however , for @xmath223 , both measures increase as the power @xmath228 increases .
results on cnf sizes are reported in fig.[fig : resultsgencactus ] which plots for 4 different settings the number of instances encoded ( @xmath219-axis ) within a cnf with that many clauses ( @xmath220-axis ) .
recent work @xcite encodes pseudo - boolean constraints via `` totalizers '' similar to sorting networks , determined by the representation of the coefficients in an underlying base . here
the authors choose the standard base 2 representation of numbers .
it is straightforward to generalize their approach for an arbitrary mixed base , and our algorithm is directly applicable . in
@xcite the author considers the @xmath93 cost function and analyzes the size of representing the natural numbers up to @xmath24 with ( a particular class of ) mixed radix bases .
our lemma [ lem : primes ] may lead to a contribution in that context .
it has been recognized now for some years that decomposing the coefficients in a pseudo - boolean constraint with respect to a mixed radix base can lead to smaller sat encodings .
however , it remained an open problem to determine if it is feasible to find such an optimal base for constraints with large coefficients . in lack of a better solution
, the implementation in the minisat@xmath0 tool applies a brute force search considering prime base elements less than 17 . to close this open problem ,
we first formalize the optimal base problem and then significantly improve the search algorithm currently applied in minisat@xmath0 .
our algorithm scales and easily finds optimal bases with elements up to 1,000,000 .
we also illustrate that , for the measure of optimality applied in minisat@xmath0 , one must consider also non - prime base elements .
however , choosing the more simple @xmath93 measure , it is sufficient to restrict the search to prime bases . with the implementation of our search algorithm it is possible , for the first time , to study the influence of basing sat encodings on optimal bases .
we show that for a wide range of benchmarks , minisat@xmath0 does actually find an optimal base consisting of elements less than 17 .
we also show that many pseudo - boolean instances have optimal bases with larger elements and that this does influence the subsequent cnf sizes and sat solving times , especially when coefficients contain larger prime factors .
we thank daniel berend and carmel domshlak for useful discussions .
10 o. bailleux , y. boufkhad , and o. roussel .
a translation of pseudo boolean constraints to sat .
, 2(1 - 4):191200 , 2006 .
o. bailleux , y. boufkhad , and o. roussel .
new encodings of pseudo - boolean constraints into cnf . in _ proc . theory and applications of satisfiability testing ( sat 09 ) _ , pages 181194 , 2009 .
p. barth . .
kluwer academic publishers , norwell , ma , usa , 1996 .
k. e. batcher .
sorting networks and their applications . in _
afips spring joint computing conference _ , volume 32 of _ afips conference proceedings _ , pages 307314 .
thomson book company , washington d.c .
, 1968 .
r. e. bixby , e. a. boyd , and r. r. indovina . : a test set of mixed integer programming problems .
, 25:16 , 1992 .
r. e. bryant , s. k. lahiri , and s. a. seshia .
deciding clu logic formulas via boolean and pseudo - boolean encodings .
in _ proc .
workshop on constraints in formal verification ( cfv 02 ) _ , 2002 .
b. chor , p. lemke , and z. mador . on the number of ordered factorizations of natural numbers . , 214(1 - 3):123133 , 2000 .
n. en and n. srensson .
translating pseudo - boolean constraints into sat .
, 2(1 - 4):126 , 2006 .
d. knuth . .
addison - wesley , 1973 .
v. m. manquinho and o. roussel .
the first evaluation of pseudo - boolean solvers ( pb05 ) .
, 2(1 - 4):103143 , 2006 .
n. sidorov .
sum - of - digits function for certain nonstationary bases .
, 96(5):36093615 , 1999 .
let @xmath62 and let @xmath55 be an optimal base for @xmath9 with @xmath69 .
let @xmath77 be the base obtained by removing the last element from @xmath8 . we show that @xmath229 and that @xmath230 .
the claim then follows .
assume falsely that for @xmath231,@xmath232 .
then we get the contradiction @xmath233 . from the definition of @xmath93 and the above contradiction we deduce that @xmath234 .
[ pnib ] let @xmath235 be a base with @xmath236 and let @xmath237 .
then , the unique representation of @xmath58 in @xmath8 is obtained as @xmath238 such that : ( 1 ) @xmath239 , ( 2 ) @xmath240 for @xmath241 , and ( 3 ) @xmath242 .
[ pprimesonly ] let @xmath246 and let @xmath247 and @xmath248 be bases which are identical except that at some position @xmath249 , two consecutive base elements in @xmath137 are replaced in @xmath138 by their multiplication .
formally : @xmath250 for @xmath251 , @xmath252 , and @xmath253 for @xmath254 .
then , the sum of the digits in @xmath255 is greater or equal to the sum of the digits in @xmath256 .
let @xmath138 be a base with @xmath262 elements of the form @xmath263 where the element @xmath264 at position @xmath265 is non - prime and @xmath266 .
so , taking @xmath267 , we are in the setting of proposition [ pprimesonly ] .
the result follows : @xmath268 consider the notation of definition [ cost2 ] .
let @xmath62 , @xmath55 with @xmath56 and @xmath124 . denote the sequences @xmath269 ( sums ) and @xmath270 ( carries ) defined by : @xmath271 for @xmath128 , @xmath272 , and @xmath273 for @xmath128 .
we denote also @xmath274 .
[ psubbase2 ] let @xmath277 , and @xmath8 a base .
then for every @xmath278 such that @xmath279 there exists @xmath280 let @xmath281 be any monotonically increasing function such that for @xmath282 , @xmath283 .
let @xmath62 .
then @xmath284 given by : let @xmath288 and @xmath9 be as defined above and @xmath8 , @xmath77 bases such that @xmath156 , @xmath56 , and @xmath289 . for every base @xmath290 we can see that @xmath291 ( the size of a finite set is a natural number ) .
therefore it is sufficient to prove that @xmath292 . from proposition [ psubbase1 ]
we get that @xmath293 . from the definition of @xmath294 and of @xmath288 we see that @xmath295 for @xmath296 , denote @xmath297 . + let @xmath298 .
by proposition [ psubbase2 ] there exists @xmath299 such that @xmath300 and therefore @xmath301 .
this implies that @xmath302 in total we have @xmath303 the proof that @xmath304 is similar noting that @xmath305 .
( of lemma [ lem : h ] ) + if @xmath306 , then @xmath307 . by proposition [ lhc ] both definitions give a heuristic cost function .
the proof for the case of @xmath144 and @xmath93 is of the same structure as proposition [ lhc ] .
the case of @xmath308 is the most complicated one .
( * property 1 * ) [ dbme ] a heuristic cost function @xmath309 is called _ base mul equivalent _ if for every @xmath62 and for every bases @xmath137 , @xmath138 such that @xmath196 and @xmath197 the following holds : [ pbestrep ] let @xmath62 and let @xmath309 be a base mul equivalent heuristic cost function . for every base @xmath137 extracted from the queue and for every base @xmath138 such that @xmath310 then @xmath311 .
assume that the claim is true for every @xmath243 and assume that during the run of the algorithm we extract a base @xmath137 from the queue with @xmath314 .
assume falsely that there exists a base @xmath138 such that @xmath315 and @xmath316.this means that @xmath138 was not evaluated yet ( otherwise @xmath317 ) .
therefore the father of @xmath138 ( in the tree of bases ) was never extracted from the queue .
let @xmath318 be the closest ancestor of @xmath138 that was extracted . denote by @xmath319 the child of @xmath318 which is also the ancestor of @xmath138 ( potentially @xmath138 itself ) .
so , @xmath319 was evaluated .
observe that @xmath318 and @xmath319 are unique because the search space ( of bases ) is a tree and @xmath320 .
so , @xmath321 and that is a contradiction to the existence of @xmath138 . for @xmath330 ,
first notice that @xmath331 .
this follows directly from the definition of admissible heuristics ( for the @xmath93 case ) .
hence , @xmath332 . from propositions [ pnib ] and [ pbaseextenstion ] , we have that @xmath333 for @xmath334 , we define @xmath335 .
so by the complete induction assumption for @xmath336 we get that @xmath337 . by the fact that @xmath338 we can deduce that @xmath339 . by the definition of admissible heuristics for @xmath93 : @xmath340 therefore , @xmath341 . combining it with the fact that @xmath342 we have that @xmath343 .
finally from the inductive assumption we get that @xmath344 .
\2 ) let @xmath345 a base mul equivalent heuristic cost function and @xmath62 .
we denote by @xmath154 the best base found by the algorithm at each point of the run .
let @xmath8 be the first base extracted from the queue such that @xmath346 .
this is the condition that terminates the main loop of the algorithm , so we need to prove that @xmath154 is the optimal base for @xmath9 .
assume falsely that there exists an optimal base @xmath347 such that @xmath348 .
let @xmath349 be the nearest ancestor of @xmath77 such that the its base equivalence class representative @xmath120 was extracted from queue ( @xmath350 otherwise @xmath351 ) . by proposition [ pbestrep ]
we know that @xmath352 and by * property 1 * that for any base c @xmath353 .
in particular for the case where @xmath354 . by choice of @xmath349
we get that the equivalence class representative of @xmath355 was not extracted ( and it is the same class of @xmath356 ) .
therefore , @xmath357 , which is a contradiction .
we use the prime number theorem which states that the density of the primes near @xmath219 is @xmath358 .
the number of prime bases evaluated in the worst case scenario is : @xmath359 but @xmath360 and so the total number of bases evaluated during a run of the algorithm is bounded by @xmath361 and the overall complexity is @xmath362 .
let @xmath363 with @xmath364 .
it is standard to define @xmath365 and @xmath366 as natural numbers such that @xmath367 and @xmath368 . in our proofs
we note that @xmath365 is the maximal number such that there exists @xmath369 with @xmath370 . by definition if @xmath374 , @xmath375 and @xmath376 .
then , @xmath377 , @xmath378 and @xmath379 .
now , @xmath380 because otherwise it would be a contradiction to the maximality of @xmath381 . if @xmath382 then @xmath383 .
assume that @xmath384 .
then @xmath385 and from that we deduce @xmath386 ( otherwise it would be a contradiction to the maximality of @xmath13 ) . on the other hand , @xmath387 @xmath388 @xmath389 . from the definition of modulus we get that @xmath390 and so @xmath391 , and therefore @xmath392 . in total
we get that @xmath393 and because @xmath13 and @xmath394 are natural numbers we get the equality .
let @xmath396 , @xmath397 , @xmath398 , and @xmath399 . by definition
we can see that @xmath400 , @xmath378 , and @xmath379 . therefore @xmath401
@xmath388 @xmath402 @xmath388 @xmath403 @xmath388 @xmath404 @xmath388 @xmath405 @xmath388 @xmath406 @xmath388 @xmath407 @xmath388 @xmath408 and this is true by proposition [ pdiv ] .
assume the assumption is true for every base @xmath8 such that @xmath415 .
let @xmath56 such that @xmath416 .
define the base @xmath417 with @xmath418 .
we can see that for @xmath419 , @xmath420 , and that for @xmath421 , @xmath422 . from this
we can see that for @xmath423 , @xmath424 and also that @xmath425 . back to the main claim by the induction
we know that @xmath426
. therefore @xmath427 . + by proposition [ pdivmod ] we can see that @xmath428 . and
therefore we get that @xmath429 .
( of proposition [ psubbase1 ] ) + first we notice that because @xmath156 then for @xmath430 we get that @xmath431 and therefore @xmath432 . by proposition [ pnib ]
we see that @xmath433 . by definition of @xmath65 and by proposition [ pnib ] and the definition of @xmath434 we prove this proposition by induction on @xmath435 .
( of proposition [ psubbase2 ] ) + let @xmath58 and @xmath8 be as defined above .
let @xmath440 such that @xmath441 .
if there exists an @xmath442 such that @xmath443 then we get that @xmath444 and in any case ( @xmath445 or @xmath446 ) @xmath447 .
otherwise let @xmath448 be the maximal index such that @xmath449 .
if @xmath446 then @xmath450 .
consider the case when @xmath445 then @xmath451 such that @xmath452 and so by dividing both sides by @xmath453 we get that @xmath454 , which means that @xmath455 . 1 . for @xmath251 by definition @xmath456 which means that @xmath457 and therefore @xmath458 . for @xmath260
again we get that @xmath459 and @xmath460 which again means that @xmath259 ) .
we can see that @xmath461 . by proposition [ pnib ]
we know that + @xmath462 + @xmath463 + @xmath464 + therefore @xmath465 + because @xmath466 we deduce that @xmath467 + and in total we get that + @xmath468 ( of proposition [ pbaseextenstion ] ) + let @xmath322 and @xmath8,@xmath77 bases such that @xmath156 .
let @xmath469 ( @xmath470 , otherwise there is no such index ) . for @xmath471 by proposition [ pnib ] we get that @xmath472 .
if @xmath227 then by proposition [ pnib ] and definition of @xmath473 we can deduce @xmath474 . | this paper formalizes the _ optimal base problem _ , presents an algorithm to solve it , and describes its application to the encoding of pseudo - boolean constraints to sat .
we demonstrate the impact of integrating our algorithm within the pseudo - boolean constraint solver minisat@xmath0 .
experimentation indicates that our algorithm scales to bases involving numbers up to 1,000,000 , improving on the restriction in minisat@xmath0 to prime numbers up to 17 .
we show that , while for many examples primes up to 17 do suffice , encoding with respect to optimal bases reduces the cnf sizes and improves the subsequent sat solving time for many examples . |
the way in which superconductivity ( sc ) occurs in high-@xmath0 cuprates challenges old concepts in condensed matter physics .
the puzzling doping dome - shape for the superconducting critical temperature ( @xmath0 ) is tied to the anomalous properties observed in the normal state . in underdoped ( ud ) the decay of @xmath0 with decreasing doping is correlated to the increasing pseudogap ( pg ) feature observed above @xmath0 .
the so - called pg phase shows properties which are not expected to occur in fermi liquids.@xcite the anomalous properties observed in ud weaken with increasing doping towards overdoped ( od ) ; however whether the conventional fermi liquid ( fl ) applies in this doping region is controversial .
recent angle - dependent magnetoresistance ( admr ) experiments in od tl2201 brought insights to this discussion.@xcite these experiments , differently to the resistivity , have the advantage to separate the scattering rate ( @xmath2 ) in two distinct components . while one component is isotropic ( @xmath3 ) on the fermi surface ( fs ) and rather constant with doping , the other one is strongly anisotropic ( @xmath4 ) and resembles the anisotropy of the pg showing maximum values near the antinode .
moreover , @xmath5 decreases with increasing doping and vanishes in highly od samples . these and others experiments have been recently interpreted ( see ref . [ ] and references therein ) in terms of that the pg phase is distint to sc , ended at the quatum critical point ( qcp ) , and its fluctuations as the responsible for pairing and transport properties .
the present paper shows that two scattering rates , with similar characteristics to those observed in admr experiments , are obtained from the @xmath1 model .
the paper is organized as follows . in sec .
ii , a summary of the method is given .
the relationship between dynamical self - enegry contributions and the two inelastic scattering rates is shown .
iii contains the results . in subsection
a results for high doping values are presented and compared with available experiments in this doping region . in subsection b predictions for low doping
are studied and discussed . in subsection
c the implications of the two obtained scattering rates on the resistivity are studied and confronted with measurements .
discussion and conclusion are given in sec .
in leading order of large @xmath6 expansion ( @xmath6 is the number of spin components ) the @xmath1 model predicts a phase diagram with close similarities to the phase diagram of hole doped cuprates.@xcite the pseudogap , which is associated to the flux - phase ( fp),@xcite competes and coexists with sc . the large-@xmath6 mean - field treatment of the @xmath1 model yields a quasiparticle ( qp ) dispersion @xmath7 where the parameters @xmath8 , @xmath9 and @xmath10 are the hopping between nearest - neighbor , next nearest - neighbor and the nearest - neighbor heisenberg coupling respectively .
the contribution @xmath11 to the mean - field band and the chemical potential @xmath12 must be obtained self - consistently@xcite from @xmath13 and @xmath14 where @xmath15 is the fermi function , @xmath16 the doping away from half - filling and , @xmath17 the number of sites .
hereafter , @xmath18 , @xmath19 , suitable for cuprates , are used and the lattice constant @xmath20 of the square lattice is considered as length unit . in this approach ,
the mean - field homogeneous state becomes unstable when the static ( @xmath21 ) flux susceptibility @xmath22^{-1}\end{aligned}\ ] ] diverges.@xcite in eq.(4 ) @xmath23 is an electronic polarizability calculated with a form factor @xmath24 . in fig.5(b )
, disregarding sc , the solid line shows , in the doping - temperature ( @xmath25 ) plane , the temperature @xmath26 which indicates the onset of fp instability . at @xmath26
a flux - mode ( @xmath27 ) reaches @xmath21 freezing the fp.@xcite at @xmath28 a phase transition occurs at the qcp placed at the critical doping @xmath29 ( fig.5(b ) ) . since the instability takes place at @xmath30 the form factor @xmath31 transforms into @xmath32 which indicates the @xmath33-wave character of the fp .
although the relevance of the fp for the physical case @xmath34 , for instance in the form of a phase without long - range but strong @xmath33-wave short - range order , is under dispute,@xcite the fp scenario possesses the main properties to be identified with the phenomenological @xmath33 cdw proposal.@xcite for discussing the quasiparticle inelastic scattering rate it is necessary to calculate self - energy corrections collecting @xmath35 fluctuations beyond mean field . as showed in ref .
[ ] the self - energy @xmath36 contains contributions from six different channels and their mixing : the usual charge channel named @xmath37 , a non - double - occupancy channel named @xmath38 and , four charge channels driven by @xmath10 .
however , as discussed in ref .
[ ] and summarized below , the relevant contributions to @xmath36 can be written as @xmath39 where @xmath40 \right . \nonumber\\ & & \hspace{-2.5 cm } + \ ; 2\;\omega \ ; im[d_{\lambda r}({\mathbf{q}},\omega-\epsilon_{{k - q } } ) ] + \left .
im[_{\lambda \lambda}({\mathbf{q}},\omega-\epsilon_{{k - q } } ) ] \right\ } \nonumber\\ & & \hspace{-1cm}\times \left[n_{f}(-\epsilon_{{k - q } } ) + n_{b}(\omega-\epsilon_{{k - q}})\right],\end{aligned}\ ] ] and @xmath41\end{aligned}\ ] ] in the above expressions , @xmath42 and @xmath43 is the bose function .
the physical meaning of eq.(5 ) is as follows .
@xmath44 corresponds to the usual charge ( @xmath37 ) sector , non - double - occupancy ( @xmath38 ) sector and the mixing of both . for @xmath19
there is no important influence of @xmath10-contributions in @xmath44 showing that the usual charge sector is weakly coupled to the @xmath10-channels .
on the other hand , @xmath10-channels play an important role at low energy and low doping in the proximity to the flux phase instability .
@xmath45 ( eq.(7 ) ) is obtained after projecting the self - energy on the eigenvector corresponding to the flux instability .
eq.(7 ) shows the coupling between carriers and fp fluctuations . for the explicit expressions of @xmath46 , @xmath47 and @xmath48 see ref .
[ ] . motivated from above discussion and the results presented below the two scattering rates @xmath49 and @xmath5 ( discussed in the introduction ) are associated to @xmath44 and @xmath50 , respectively , as follows @xmath51 @xmath52 where @xmath53 is a momentum on the fs .
fig.1 shows , for two dopings and different temperatures , @xmath54 and @xmath55 on the fs labeled by the angle @xmath56 ranging from the antinode ( @xmath57 ) to the node ( @xmath58 ) ( see inset in panel ( b2 ) ) .
note that @xmath59 and @xmath60 lay in od as the samples studied in admr experiments .
similarly to the behavior observed in admr@xcite @xmath3 is very isotropic on the fs and @xmath4 is strongly anisotropic showing the maximum near the antinode .
note that @xmath4 follows , approximately , the proposed@xcite shape @xmath61 ( dashed line in panels ( b1 ) and ( b2 ) ) .
et al . _ ( ref . [ ] ) have shown similar results for moderate onsite coulomb repulsion on the hubbard model in one loop renormalization group approximation .
on the fermi surface for @xmath59 and @xmath60 respectively , and for @xmath62 and @xmath63 .
( b1 ) and ( b2 ) the same than panels ( a1 ) and ( a2 ) for the anisotropic scattering rate @xmath4 .
dashed line in ( b1 ) and ( b2 ) is a guide for the eyes @xmath64 showing a similar trend for @xmath5 .
inset in ( b2 ) defines the angle @xmath56 ranging for the antinode to the node . in the calculation of @xmath65 , @xmath66 was used in the analytical continuation . ,
width=302 ] in fig.2(a ) , @xmath67 and @xmath68 ( where @xmath69 means the average on the fs ) are plotted as a function of doping for a fixed temperature @xmath70 .
this temperature is close to the reported @xmath71 in fig.2 of ref .
[ ] if the accepted value @xmath72 is considered . as in the experiments
, @xmath68 is only weakly doping dependent and , @xmath67 is strongly decreasing with increasing doping .
inset of fig.2(a ) shows , in the @xmath25 plane , that @xmath4 dominates over @xmath3 at low doping and low temperatures ( blue region ) while , @xmath5 dominates on the red region . in fig.2(b ) , the ratio @xmath73 at @xmath70 is plotted as a function of doping ( solid line ) and compared with the experimental data ( solid circles ) taken from ref .
although the experimental data seem to decay faster than in the theory , both results show a similar trend with increasing doping .
note that while theory predicts that @xmath4 becomes smaller than @xmath3 for @xmath74 ( see also fig.2 ( a ) ) , the same occurs in the experiment for @xmath75 . beyond a quantitative comparison it is important to discuss about the physical interpretation . in refs .
[ ] it was claimed that @xmath4 vanishes just at the doping value where sc emerges from od following dashed line in fig.2(b ) , concluding that the source for sc and for the anisotropic scattering rate is the same and associated to pg fluctuations . in present case
@xmath4 is caused by the scattering between carriers and short - range fp fluctuations .
thus , since @xmath50 proves the proximity to the pg via the coupling between carriers and the soft flux - mode of momentum @xmath30 , @xmath5 is strongly anisotropic on the fs and decreases with increasing doping beyond the qcp .
however , sc comes from the instantaneous ( no bosonic glue ) short - range antiferromagnetic exchange @xmath10@xcite leading to a different origin for sc and the scattering rate .
finally , it is worth to mention that considering the error bars ( fig.2(b ) ) @xmath5 could also decay following , for instance , dotted line which is nearly paralel to the theoretical predicted solid line . and @xmath76 as a function of doping at @xmath70 .
the inset shows , in the @xmath25 plane , that @xmath5 is larger than @xmath49 at low doping and low temperatures ( blue region ) , in the red region @xmath49 is larger .
( b ) ( @xmath77 ) ratio at @xmath70 as a function of doping ( solid line ) .
solid circles are the experimental data at similar temperature shown for comparison ( see text for discussion ) . dashed and dotted lines are possible trends according to the error bars .
, width=302 ] in fig .
3 the temperature behavior of both scattering rates is shown . at low temperature @xmath78 with @xmath79 ( fig .
3(a ) ) for all dopings .
a similar quadratic temperature behavior for @xmath49 was observed for the isotropic scattering rate in od tl2201.@xcite in contrast , @xmath67 ( fig .
3(b ) ) shows , at high dopings ( see results for @xmath80 and @xmath81 ) , a different behavior : at low temperature , @xmath82 with @xmath83 which is close to the @xmath84-linear law discussed in refs .
[ ] for similar dopings .
therefore , it is concluded that , besides the anisotroy on the fs , at high doping the temperature dependence for both , @xmath5 and @xmath49 , agrees also , qualitatively , with the experiment . as a function of temperature for several dopings .
( b ) the same than ( a ) for @xmath76 .
the studied dopings are indicated in each panel .
arrows in ( b ) indicate the temperature where an upturn is observed for low dopings ( see text ) . , width=302 ] to the knowledge of the authors there are no admr experiments in the ud region , however for low doping ( @xmath85 ) the present approach shows predictions which in principle can be tested by admr experiments if they are possible . with decreasing
temperature @xmath67 shows an upturn below a given temperature ( @xmath86 ) indicated by arrows ( see results for @xmath87 , @xmath88 and , @xmath89 in fig.3(b ) ) . @xmath86
increases with decreasing doping following the same trend ( see dotted - dashed line in fig.5(b ) ) than @xmath26 .
interestingly , although @xmath86 marks a smooth crossover and not a true phase transition , dotted - dashed line terminates at the qcp . while for large dopings @xmath4 decreases with decreasing temperature ( panels ( b1 ) and ( b2 ) in fig.1 ) , for @xmath90 ( fig .
4(b1 ) ) and @xmath91 ( fig .
4(b2 ) ) , @xmath4 shows a reentrant behavior near the antinode .
it is important to mention that the length from the antinode of the re - entrant ( marked in fig .
4(b1 ) and ( b2 ) ) expands with decreasing temperature and doping resembling the behavior of the fermi - arcs observed in angle - resolved photoemission spectroscopy@xcite ( arpes ) , i.e. , when larger is the re - entrant length shorter the arcs .
see last section for further discussion about a possible link between admr and arpes experiments . finally , fig.4
( a1 ) and ( a2 ) together with results for @xmath87 in fig .
3(a ) show that the evolution of @xmath49 from ud to od is rather smooth . and @xmath91 . with decreasing temperature
a re - entrant behavior occurs near the antinode for @xmath5 ( panels ( b1 ) and ( b2 ) ) .
, width=302 ] in spite of intense studies the origin of the resistivity in cuprates is far from closed .
whether the resistivity is composed by two different contributions with different temperature and doping dependence,@xcite or a single contribution is currently under dispute.@xcite therefore , based on present results , it is worth to discuss a possible connection between the two scattering rates observed in admr and resistivity measurements . in first approximation @xmath92 where @xmath93 and @xmath94 is the plasma frequency . using the obtained results for @xmath95 as a function of temperature ( fig.5(a ) ) a curvature mapping ( @xmath96 ) in the @xmath25 plane
is presented in fig.5(b ) where positive ( negative ) curvature is indicated in red ( blue ) .
it is possible to divide the behavior of @xmath95 in two regimes .
( a ) @xmath97 : since at low temperatures @xmath5 dominates ( inset in fig .
2(a ) ) , an upturn also occurs for @xmath95 ( fig .
5(a ) ) . note that for @xmath95 the upturn is shifted to lower temperatures respect to the upturn of the isolated @xmath5 being more pronounced below dashed line in fig .
a similar upturn for the resistivity was found experimentally.@xcite while ando _
, for bslco , lsco , and ybco found a weak and negative curvature above the upturn temperature , daou _ et al .
_ , reported a @xmath84-linear law for nd - lsco .
in present case @xmath98 with @xmath99 indicates a weak positive curvature ( light red region ) closer to the results of ref .
at this point it is important to make the following remark .
as discussed above the upturn is caused by the interaction between quasiparticles and short - range fluctuations of the fp .
if long - range order sets in at @xmath26 , in the context of present approximation @xmath95 is expected to diverge at @xmath26 .
this divergence is due to the fact that fp fluctuations were not self - consistently included in @xmath50.@xcite the inclusion of these fluctuations would weaken the upturn leading , for @xmath100 , to a @xmath33 cdw metal .
( b ) @xmath101 : close to @xmath102 and at low temperature @xmath95 shows a downturn ( blue region in fig .
5(b ) ) which fades out , with increasing temperature and doping , faster than in experiment.@xcite for larger dopings @xmath95 shows a positive @xmath84-curvature .
for low temperature @xmath95 could be approximated by a single power law @xmath98 . with increasing doping @xmath103 increases from @xmath104 near @xmath105 to @xmath106 at high doping ( solid line in fig .
similar behavior is widely discussed in the literature @xcite and interpreted as the emergence of the fl in od .
the experimental values for @xmath103 ( solid circles ) taken from ref .
[ ] were included for comparison , showing a rather good agreement between theory and experiment ( see also ref . [ ] ) . .
( b ) curvature mapping ( @xmath96 ) in the @xmath25 plane .
red ( blue ) region indicates positive ( negative ) curvature .
, width=302 ] the doping dependence of the resistivity is also important . in fig.6(b )
the @xmath107 for bi2201@xcite ( circles ) and for tl2201@xcite ( diamonds ) is plotted as a function of doping ( left axis ) together with @xmath108 ( right axis ) at @xmath109 .
this figure shows that both , @xmath110 and @xmath95 , follow similar doping dependence suggesting that the scattering rate is the main cause for the doping behavior of the resistivity . comparing the scales of right and left axis it is possible to see that they are shifted by a constant @xmath111 , since @xmath112 , then @xmath113 which is somewhat lower but on the order of the experiment@xcite and also consistent with present dispersion @xmath114 ( eq.(1 ) ) .
finally , the resistivity for the high doping case @xmath115 is estimated . using @xmath116
, @xmath110 has the following similarities with measurements in high od lsco:@xcite ( a ) for @xmath117 , @xmath118 with @xmath106 , while for higher temperatures @xmath119 .
( b ) the quadratic coefficient @xmath120 is @xmath121 which is on the order of magnitude reported in ref .
[ ] . this high value of @xmath120 , which is about two orders of magnitude larger than the expected value for conventional metals , supports the interpretation that , even at very high dopings , cuprates must be considered in the strong - coupling regime.@xcite in summary , considering that the experimental behavior of the resistivity is rather controversial@xcite and no existing theory describing systematically all observed features , present results can be considered satisfactory .
assuming the approximated form @xmath122 .
solid circles are taken from the experiment for comparison .
( b ) logarithm of the experimental resistivity ( @xmath107 ) ( left axis ) and @xmath108 ( right axis ) as a function of doping at @xmath109 .
, width=302 ]
in the framework of the @xmath1 model in large @xmath6 approximation , dynamical self - energy corrections were computed beyond mean field . the existence of two distinct components for the quasiparticle inelastic scattering rate was shown .
while one component is very anisotropic on the fs and disappears at high dopings , the other component is isotropic on the fs and rather constant with doping . in od ,
the two components behave qualitatively similar to those observed in admr experiments . in addition , predictions for ud region which , in principle , can be tested in admr experiments were discussed
. the doping and temperature behavior of the estimated resistivity has similarities with the corresponding transport measurements concluding that the picture presented here may contribute on discussions about the origin of the resistivity and its possible link to the findings observed in admr experiments .
it is known that the large-@xmath6 expansion weakens antiferromagnetic fluctuations respect to charge fluctuations .
however , it is important to mention that admr experiments were performed in the od region where the large-@xmath6 approach is expected to be more reliable . the observed agreement between experiments and theory
is interpreted here as indication that , at least , part of the physics is captured by present theory . for instance , for high doping present calculation predicts @xmath123 as observed in the experiments .
importantly , besides the temperature behavior , an unusual high value for the coefficient @xmath120 is obtained in agreement with measurements .
this is interpreted as an indication that , even in od , strong coupling effects occur in cuprates . before closing
, a possible link between admr and arpes experiments is discussed .
it was recently shown that ( a ) @xmath50 , which proves the proximity to the pg , dominates at low doping and low energy leading to fermi - arcs effects.@xcite ( b ) @xmath124 , which contains non - double - occupancy effects of the @xmath1 model , dominates at high energy leading to high energy features@xcite which resemble the waterfall effects observed in arpes.@xcite thus , present paper suggests also a possible common origin for the features seen in admr and arpes experiments . in @xmath35 , @xmath50 ( eq.(7 ) ) proves the instability only via the flux - mode which reaches @xmath21 at @xmath26 . in this order
, @xmath50 does not contain , self - consistently , self - energy effects which are beyond the scope of @xmath125-expansion . as discussed in text these effects
are expected to play a role only near @xmath26 . | the recent findings about two distinct quasiparticle inelastic scattering rates in angle - dependent magnetoresistance ( admr ) experiments in overdoped high-@xmath0 cuprates superconductors have motivated many discussions related to the link between superconductivity , pseudogap , and transport properties in these materials . after computing dynamical self - energy corrections in the framework of the @xmath1 model the inelastic scattering rate was introduced as usual .
two distinct scattering rates were obtained showing the main features observed in admr experiments .
predictions for underdoped cuprates are discussed .
the implicances of these two scattering rates on the resistivity were also studied as a function of doping and temperature and confronted with experimental measurements . |
a magnetic flux enclosed in electron trajectories induces electron wave function interference ; this is the well - known magnetic aharonov - bohm ( ab ) effect @xcite .
the ab effect describes a quantum phenomenon in which a charged particle is affected by the vector potential of an electromagnetic field .
the ab effect has inspired breakthroughs in modern physics .
photons are known to be neutral particles that do not directly interact with magnetic fields .
however , the magnetic ab effect for photons has been proposed and realized by magnetic - optical effects @xcite , dynamical modulation @xcite , and photon - phonon interaction photphot .
this is attributed to an effective magnetic field originating from the fictitious gauge field felt by photons , where photons behave as electrons in a magnetic field .
the concept of effective magnetic field for photons provides new opportunities in optics ; it stimulates interest in the exploration of fundamental physics and in the creation of applications moeffect , dynamicmodulation , photphot , hafezi2011,hafezi2013,chong .
in particular , parity - time ( @xmath0 ) symmetric optical systems proposed with effective magnetic flux allow nonreciprocal light transport longhi2015ol , lxq . the @xmath0-symmetry has attracted tremendous interest over the last decade .
a @xmath0-symmetric system may possess an entirely real spectrum even though it is non - hermitian bender98,znojil99,dorey01,ali02 .
the @xmath0-symmetric system is invariant under the combined parity ( @xmath2 ) and time - reversal ( @xmath3 ) operators , and its potential fulfills @xmath4 . in 2007 ,
a coupled optical waveguide system with an engineered refractive index and gain / loss profile was proposed to realize a @xmath0-symmetric structure @xcite .
the proposal was based on a classical analogy , namely , that maxwell s equations , describing light propagation under paraxial approximation , are formally equal to a schrdinger equation @xcite .
thereafter , a number of intriguing phenomena can be predicted and experimentally verified , either in a passive or in an active @xmath0-symmetric optical structure , including parity - time ( @xmath0 ) symmetry breaking aguo , power oscillation @xcite , coherent perfect absorbers cpa , hchen , spectral singularities @xcite , unidirectional invisibility @xcite , and nonreciprocal wave propagation @xcite .
recently , an active @xmath0-symmetric optical system has been realized using two coupled microcavities @xcite , and optical gain played a key role .
optical isolators @xcite and @xmath1-symmetric lasers @xcite were demonstrated ; the gain induced a large optical nonlinearity .
both the coupled optical waveguides and microcavities were described by a tight - binding model .
the tight - binding model demonstrated analytical and numerical tractability for study of the @xmath0 symmetry longhi1,longhi2,znojil1,znojil2,jin , bendix .
the @xmath0-symmetric phase diagram , as well as the wave packet dynamics in @xmath0-symmetric systems with open boundary conditions , have been investigated @xcite .
although the properties of a lattice with open boundary conditions and the properties of a lattice with periodical boundary conditions are similar as large systems approach a size limit , the differences are notable when the system size is small joglekar2012 .
currently , most experimentally accessible @xmath0-symmetric systems are small in size @xcite ; hence , studying @xmath0-symmetric systems under periodical boundary conditions is worthwhile , and some pioneering works have already focused on this @xcite . in this paper ,
we study the influence of magnetic flux on the @xmath0 symmetry in coupled resonators in ring configuration ( fig .
[ fig1]a ) .
the system is modeled by a magnetic tight - binding lattice , with a peierls phase factor in the hopping between adjacent sites .
the one dimensional ring system has a single pair of balanced gain and loss located at opposite positions ; when threaded by magnetic flux , the ring system remains @xmath5-symmetric .
we reveal that the @xmath0 symmetry is sensitive to the enclosed magnetic flux , especially when the gain / loss or system size is large .
we find a threshold for the gain / loss , above which , any nontrivial magnetic flux breaks the @xmath0 symmetry .
we also enumerate the eigenvalues that have broken @xmath0 symmetry .
the threshold gain / loss , as well as the @xmath0-symmetry breaking levels , are tunable by the magnetic flux .
we present a @xmath0-symmetric phase diagram in the gain / loss and magnetic flux parameter spaces .
the maximally tolerable magnetic flux for the exact @xmath0-symmetric phase is found for a @xmath6-site ring system .
our findings offer insights regarding magnetic flux in @xmath0-symmetric systems and might provide useful applications of @xmath7 symmetry in quantum metrology .
this paper is organized as follows .
we formulate a @xmath0-symmetric hamiltonian threaded by a magnetic flux in sec .
we show the energy spectrum and phase diagram of a four - site ring system in sec .
we elucidate in detail that the @xmath0 symmetry is significantly affected by the enclosed magnetic flux through studying a @xmath8-site ring system in sec .
our conclusions are drawn in sec .
we consider a discrete @xmath0-symmetric ring system threaded by a magnetic flux .
the system is schematically illustrated in fig .
[ fig1]a using coupled resonators , which constitute a @xmath8-site lattice system described by a tight - binding model
. the enclosed magnetic flux in the ring system is denoted as @xmath9 .
the magnetic flux acts globally : circling photons or charged particles accumulate a phase factor @xmath10 in each round because of the ab effect .
the plus / minus sign represents a clockwise / counterclockwise direction of motion .
the magnetic flux affects the system periodically ; the period is @xmath11 , which corresponds to one quantum of the effective magnetic flux .
a balanced pair of gain and loss is symmetrically located at opposite positions on sites @xmath12 and @xmath13 , respectively .
the hamiltonian for this @xmath8-site ring system is given by : @xmath14where @xmath15 ( @xmath16 ) is the creation ( annihilation ) operator for site @xmath17 , the periodic boundary condition requires @xmath18 .
the hopping strength between adjacent sites is set to unity without loss of generality . the balanced gain and loss are conjugate imaginary potentials on sites @xmath12 and @xmath13 , the rate is @xmath19 ( @xmath20 ) .
the enclosed magnetic field effectively induces an additional phase factor @xmath21 in the hoppings ( fig .
[ fig1]b ) , where @xmath22 is an averaged additional phase between adjacent sites . the parity operator @xmath2 is defined as @xmath23 , the time - reversal operator @xmath24 is defined as @xmath25 .
we note that in the presence of a nontrivial phase factor ( @xmath26 ) , the system hamiltonian @xmath27 remains invariant under the combined @xmath1 operator , i.e. , @xmath28 .
the ring system is @xmath0-symmetric with respect to its central axis ( dashed blue line shown in fig .
[ fig1]a ) .
when @xmath27 is in its exact @xmath0-symmetric phase , its spectrum is entirely real and all eigenstates are @xmath0-symmetric .
when @xmath27 is in its broken @xmath0-symmetric phase , complex conjugate pairs emerge and corresponding eigenstates are no longer @xmath1-symmetric . in the following ,
we discuss how the exact @xmath1-symmetric phase is affected by the parameters of the ring system , in particular , the enclosed magnetic flux and gain / loss .
we first consider the simplest case : a four - site @xmath7-symmetric ring system with @xmath29 .
the hamiltonian for a four - site ring system is written in the form : @xmath30}}=-\sum_{i=1}^{4}(e^{i\phi } a_{i}^{\dagger } a_{i+1}+\mathrm{h.c.})+i\gamma ( a_{1}^{\dagger } a_{1}-a_{3}^{\dagger } a_{3}),\]]where the periodical boundary condition requires @xmath31 .
the phase factor @xmath21 in front of the hoppings between adjacent sites indicates that the enclosed magnetic flux in the ring system is equal to @xmath32 .
we can diagonalize the @xmath33 matrix to acquire the spectrum of the ring system .
the eigenvalue @xmath34 satisfies @xmath35 . solving the equation ,
we obtain four eigenvalues @xmath36 for nontrivial magnetic flux
@xmath37 ( @xmath38 ) changes , the system spectrum changes as the magnetic flux @xmath9 . at @xmath39 , the energy levels shift and form
two pairs of doubly degenerate states with energy @xmath40 when @xmath41 ( @xmath38 ) , where the @xmath0 symmetry of the eigenstates is extremely sensitive to the balanced gain and loss , i.e. , any nonzero gain / loss ( @xmath42 ) breaks the @xmath0 symmetry . and @xmath43 .
eigenvalues in region i : real ; region ii : complex conjugate pairs ; and region iii : pure imaginary conjugate pairs .
region i is the exact @xmath0-symmetric phase .
region ii and iii compose the broken @xmath0-symmetric phase .
the boundaries ( blue curves ) indicate the exceptional points where two eigenstates coalescence occurs .
the red dots represent the coalescence of three eigenstates , and the green circle represents a hermitian system without coalescence of eigenstates.,width=7 ] in the general case , a magnetic flux @xmath9 breaks the @xmath0 symmetry of the eigenstates in the situation that @xmath44 .
the @xmath0-symmetric phase diagram of a four - site ring system is shown in fig .
[ region ] .
region i represents the exact @xmath1-symmetric phase , regions ii and iii compose the broken @xmath1-symmetric phase .
the magnetic flux @xmath9 plays different roles as the system gain / loss @xmath45 varies in regions of different parameters . for gain / loss @xmath39
, the dispersion relation is @xmath46 with eigenvector @xmath47 , @xmath48 , @xmath49 , @xmath50 . in this case , the magnetic flux @xmath9 shifts the energy spectrum with eigenstates unchanged .
region i : as @xmath19 increases , the ring system is in its exact @xmath5-symmetric phase when @xmath51 in @xmath52 .
region ii : the @xmath0 symmetry is broken when @xmath53 , eigenvalues that are complex conjugate pairs appear and the magnetic flux @xmath9 changes both the energy ( real part of eigenvalues ) and the amplification / decay ( imaginary part of eigenvalues ) of the eigenstates . in this case , the region with eigenvalues that are complex conjugate pairs expands in @xmath54 but shrinks in @xmath55 as @xmath19 increases .
region iii : the magnetic flux @xmath9 changes only the amplification / decay of the eigenstates when @xmath56 in @xmath57 .
this is because the eigenvalues are purely imaginary conjugate pairs in this situation . in the trivial magnetic flux
@xmath58 ( @xmath38 ) case , one complex conjugate pair becomes a two - fold degenerate state with an eigenvalue of zero in @xmath59 ; three energy levels coalesce at @xmath60 ( red dots ) with an eigenvalue of zero , and the @xmath0-symmetric ring system becomes a nondiagonalizable hamiltonian including a @xmath61 jordan block .
figure [ region ] shows the situation in which nontrivial magnetic flux satisfies @xmath62 ( excluding the green circle and red dots , e.g. , @xmath63 , @xmath64 ) ; this is shown in fig .
[ region ] by the blue curves that serve as the boundaries of different phases . in this case , two energy levels coalesce , and the @xmath1-symmetric ring system is a nondiagonalizable hamiltonian composed of two @xmath65 jordan blocks .
the magnetic flux is gauge - invariant and acts globally in the ring system .
taking the local transformation @xmath66 , the magnetic flux is unchanged and the hamiltonian @xmath27 changes into @xmath67 with a nonreciprocal coupling between sites @xmath12 and @xmath8 .
the hamiltonian @xmath68 is given by @xmath69which can be realized in coupled optical resonators by introducing synthetic magnetic flux , and a balanced gain and loss in the resonators .
the synthetic magnetic flux is introduced through an optical path imbalance method in the coupling process @xcite .
consider a system with @xmath8 coupled resonators in a ring configuration .
the coupling between resonator @xmath12 and @xmath70 is an effective coupling induced by an other auxiliary resonator ( fig .
[ fig1]b ) .
the nonreciprocal phase factor @xmath10 in the hopping between resonator @xmath12 and @xmath8 is caused by the optical path length difference @xmath71 , which is the difference in between the forward - going and backward - going optical paths of the auxiliary resonator in the coupling process .
the effective magnetic flux introduced is equal to @xmath72 where @xmath73 is the optical wavelength hafezi2014 .
therefore , @xmath9 is proportional to the path length difference @xmath71 , and is tunable through changing the position of auxiliary resonator .
the eigenvalues of the @xmath0-symmetric ring system @xmath67 is calculated as follows .
we denote the wave function for eigenvalue @xmath74 as @xmath75 .
the wave function @xmath75 is assumed as a superposition of forward- and backward - going waves,@xmath76 the schrdinger equation for site @xmath17 in the ring system ( excluding site @xmath12 , @xmath13 , and @xmath6 ) is given as : @xmath77substituting the wave functions eq .
( [ wa ] ) into eq .
( [ scheq ] ) , we obtain the eigenvalue @xmath78 .
the schrdinger equations for sites @xmath12 , @xmath13 , and @xmath8 are : @xmath79from the continuity of wave function on site @xmath13 , the wave function @xmath80 should satisfy @xmath81after simplification of the schrdinger equations and the continuity equation shown in eqs .
( [ schrodingereq]-[continuity ] ) , we derive a critical equation for eigenvector @xmath82 as an implicit function of the gain / loss @xmath19 and the enclosed magnetic flux @xmath9 .
the critical equation for eigenvector @xmath82 has the form : @xmath83 in the situation of a trivial magnetic flux @xmath58 ( @xmath38 ) , the gain / loss affects only one pair of energy levels ; it leaves the others unchanged .
the spectrum of a @xmath8-site ring system includes @xmath84 pairs of two - fold degenerate energy levels , i.e. , @xmath85 with @xmath86 $ ] ( indicated by blue lines in fig . [ fig_illus ] ) ; and one pair of gain / loss dependent energy levels , i.e. , @xmath87 ( indicated by dashed red lines in fig .
[ fig_illus ] ) . when the gain / loss @xmath59 , the ring system is in its broken @xmath0-symmetric phase , and the system spectrum has one conjugate pair .
in order to analyze the influence of magnetic flux on the system spectrum , we denote the left side of the critical equation eq .
( [ critical_eq ] ) as @xmath88 , which is a function of parameters @xmath19 , @xmath9 for eigenvector @xmath89 ( note that @xmath90 is not the eigenvector when @xmath42 ) , @xmath91 the real eigenvector @xmath82 with @xmath92 corresponds to the real eigenvalue @xmath74 of the @xmath0-symmetric @xmath70-site ring system .
for a situation with gain / loss @xmath59 , we have @xmath93 for any nontrivial magnetic flux @xmath37 ( @xmath38 ) ; this indicates that the spectrum of the ring system is entirely constituted by conjugate pairs without any real eigenenergy . for magnetic flux
@xmath41 ( @xmath94 ) , we have @xmath95 , the real eigenvector @xmath82 is absent , i.e. , the system spectrum is entirely constituted by conjugate pairs at @xmath42 .
the @xmath0-symmetric @xmath8-site ring system has at most one conjugate pair in its spectrum when magnetic flux is @xmath58 ( @xmath38 ) , but has at most @xmath96 conjugate pairs when magnetic flux is @xmath97 ( @xmath38 ) .
this implies the system spectrum is sensitive to magnetic flux , and the number of conjugate pairs appreciably varies with the magnetic flux . in the following ,
we systemically investigate how @xmath98 symmetry of the eigenstates is affected by the magnetic flux and gain / loss .
we first consider a trivial case of effective magnetic flux with @xmath99 ( @xmath38 ) : the function @xmath100 reduces to @xmath101\sin ^{2}\left ( nk\right ) $ ] .
@xmath102 is the critical equation for the eigenvector @xmath82 .
we find the eigenvalues are @xmath85 with @xmath103 $ ] and @xmath87 .
the gain / loss @xmath19 only changes two energy levels , @xmath104 , the eigenvector for @xmath105 is @xmath106hence , when @xmath107 , the system spectrum may be entirely real even if @xmath37 ( @xmath38 ) .
this indicates a threshold gain / loss value , @xmath108 .
when the gain / loss is below the threshold ( @xmath109 ) , the ring system can be in an exact @xmath0-symmetric phase .
as magnetic flux increases from @xmath110 to @xmath49 , the energy levels with broken @xmath0 symmetry simultaneously emerge from @xmath12 to @xmath96 pairs .
if @xmath111 , the @xmath0 symmetry is extremely sensitive to the magnetic flux ; any nontrivial magnetic flux @xmath37 ( @xmath94 ) breaks the @xmath0 symmetry and causes conjugate pairs to emerge simultaneously . above the threshold
gain / loss ( @xmath111 ) , the ring system remains in an exact @xmath0-symmetric phase only when the magnetic flux is @xmath58 ( @xmath38 ) .
moreover , the minimal number of complex eigenvalue pairs in the presence of nontrivial magnetic flux is @xmath112,\]]where @xmath113 $ ] stands for the integer part of @xmath114 . as shown in fig .
[ fig_illus]a , when @xmath115 , the ring system keeps in the exact @xmath0-symmetric phase in the presence of nontrivial magnetic flux . in fig .
fig_illusb , @xmath116 , @xmath117 ; the minimal number of energy levels with broken @xmath0 symmetry is @xmath118 for nontrivial magnetic flux ; the nontrivial magnetic flux breaks the @xmath0 symmetry of the energy levels with @xmath119 .
we plot function @xmath120 for a ring system with @xmath121 in fig .
we can see how the eigenvector is changed into a complex number by the variations of gain / loss @xmath19 and magnetic flux @xmath122 . in fig .
[ f ] , @xmath120 for ring system at @xmath39 , @xmath123 , @xmath124 , @xmath125 , @xmath126 , @xmath127 are plotted .
the black circles stand for the eigenvector @xmath128 for @xmath129 , @xmath124 , @xmath125 , @xmath126 .
note that all @xmath8 eigenvalues become conjugate pairs for nontrivial magnetic flux when @xmath130 , and for nonzero gain / loss when @xmath41 ( @xmath94 ) .
figure [ bands ] shows the energy bands found by numerical diagonalization of @xmath67 in eq .
( [ hsc ] ) with @xmath131 ; the corresponding eigenvector @xmath82 is in accord with the eigenvector from the critical equation eq .
( [ critical_eq ] ) . for trivial magnetic flux
, the @xmath0 symmetric ring system has only one complex pair when the gain / loss @xmath132 . for nontrivial flux
, the energy level degeneracy ( fig . [ fig_illus ] ) disappears .
this is because the nonreciprocal hopping between the coupled resonators breaks the time - reversal symmetry in the tunneling .
the number of conjugate pairs changes from @xmath12 ( @xmath133 ) to @xmath96 as magnetic flux @xmath9 increases from @xmath110 to @xmath49 for gain / loss below ( above ) the threshold .
figure [ bands ] shows the magnetic flux acting globally , and the spectrum structure substantially varies with the magnetic flux at different values of gain / loss .
now , we focus on the influence of magnetic flux on the @xmath0-symmetric phase diagram . in order to get the @xmath0-symmetric phase diagram
, we define @xmath134 as the maximal magnetic flux that keeps a ring system in its exact @xmath0-symmetric phase .
the maximal magnetic flux @xmath135 depends on the non - hermitian gain / loss @xmath45 and indicates the boundary of the @xmath0-symmetric phase diagram in the @xmath43 , @xmath19 parameter spaces . for a gain / loss above the threshold , @xmath136 , any nontrivial magnetic flux @xmath37 ( @xmath38 ) breaks the @xmath7 symmetry .
in this situation , @xmath134 is zero .
the @xmath0 symmetry is extremely sensitive to magnetic flux for large values of @xmath19 or @xmath96 ( the gain / loss threshold @xmath137 depends on @xmath96 ) . for a gain / loss below the threshold , @xmath138 , e.g. @xmath139 ( fig .
[ f ] , indicated by dotted green line ) , the @xmath0 symmetry is robust to the magnetic flux @xmath9 when @xmath140 is in the region between the two horizontal black lines shown in fig .
in this situation , the magnetic flux does not break @xmath0 symmetry , and the system spectrum is entirely real .
if @xmath19 or @xmath9 increases from @xmath110 , the system goes through a @xmath0-symmetric breaking phase transmission when @xmath141 for @xmath142 $ ] .
then the entirely real spectrum disappears and conjugate pairs emerge . for each @xmath19 ,
the ring system is in its exact @xmath1-symmetric phase when the magnetic flux is below the maximal value @xmath134 .
to gain insight regarding @xmath0 symmetry under magnetic flux , we set the function @xmath120 to reach its maximum at @xmath143 in the region @xmath142 $ ] , where @xmath144 can be calculated from @xmath145 .
we plot @xmath143 and @xmath146 in fig . [ phase_diagram ] ; the blue lines are obtained by numerically solving the equation @xmath147 .
the red circles are obtained by numerically diagonalizing the ring system hamiltonian ; the results from these two methods are in accord with each other . in fig .
[ phase_diagram]a , we plot @xmath148 as a function of @xmath19 for the ring system with @xmath131 .
as @xmath45 increases from @xmath149 to @xmath150 , @xmath143 increases from @xmath151 to @xmath152 , the maximal value of @xmath153 decreases from @xmath12 to @xmath110 , and the maximum magnetic flux @xmath134 decreases from @xmath49 to @xmath110 ( fig . [ phase_diagram]b ) .
figure [ phase_diagram]b shows the @xmath5-symmetry phase diagram of the ring system .
we can see the sharp changes at @xmath154 in fig .
[ phase_diagram]a , and @xmath155 at @xmath156 in fig .
[ phase_diagram]b .
the maximal magnetic flux @xmath157 keeps the ring system in the exact @xmath0-symmetric phase ; it satisfies @xmath158 .
we numerically calculate the maximum magnetic flux @xmath134 , which increases as the gain / loss @xmath19 decreases .
the maximal magnetic flux @xmath159 as a function of @xmath19 is shown in fig . [ phi_c ] ; the colored lines and markers are numerical results from the theoretical analysis and exact diagonalization of @xmath67 , respectively . for weak @xmath19 ( @xmath160 ) , we approximately have @xmath161 .
correspondingly , the maximal magnetic flux is approximately linearly dependent on the gain / loss @xmath19 as @xmath162 ( indicated by dashed black lines in fig .
[ phi_c ] ) .
the maximum magnetic flux @xmath134 decreases as ring system size @xmath96 increases . the @xmath0-symmetric phase transition is closely related to the magnetic flux in the coupled resonators enclosed magnetic flux . through tuning the coupling position of the auxiliary resonator between resonator @xmath12 and @xmath8 ( fig . [ fig1]b )
, the path lengths of forward- and backward - going directions change , which linearly affects the magnetic flux .
the @xmath98 symmetry breaking point varies with the magnetic flux ; the gain / loss threshold for @xmath0 symmetry breaking can be greatly reduced by increasing the enclosed magnetic flux from zero to half a quantum .
the @xmath1 symmetry breaking energy levels change from one pair to @xmath96 pairs when the enclosed magnetic flux is so tuned . the @xmath0 symmetry breaking can be observed at low balanced gain and loss values in coupled resonators with enclosed magnetic flux .
we investigate @xmath0-symmetric non - hermitian coupled resonators in ring configuration threaded by an effective magnetic flux .
the ring system is described by a @xmath8-site tight - binding model ; the ring system has a balanced pair of gain and loss located at two opposite sites .
we demonstrate that the system s @xmath0 symmetry is extremely sensitive to the enclosed magnetic flux when the gain / loss is above a threshold @xmath163 .
we find the minimal number of conjugate pairs emerging in the system spectrum when the @xmath0 symmetry is breaking in the presence of nontrivial magnetic flux .
the system eigenvalues all become conjugate pairs at a magnetic flux @xmath41 ( @xmath164 ) for any nonzero gain / loss @xmath19 ; or at gain / loss twice larger than the hopping strength ( @xmath59 ) for any nontrivial magnetic flux @xmath9 .
we show the @xmath0-symmetric phase diagram in the parameter spaces of @xmath19 and @xmath9 .
the results indicate the maximal magnetic flux approximately linearly depends on @xmath19 as @xmath165 in the weak @xmath19 region ( @xmath166 ) .
the maximal magnetic flux decreases as the gain / loss or system size increases .
our findings indicate the @xmath0 symmetry is very sensitive to the nonlocal vector potential in this @xmath1-symmetric non - hermitian system
. these results could be useful in quantum metrology in the future .
we acknowledge the support of national natural science foundation of china ( cnsf grant no .
11374163 ) , national basic research program of china ( 973 program grant no .
2012cb921900 ) , and the baiqing plan foundation of nankai university ( grant no .
zb15006104 ) . | we study a parity - time ( @xmath0 ) symmetric ring lattice , with one pair of balanced gain and loss located at opposite positions .
the system remains @xmath0-symmetric when threaded by a magnetic flux ; however , the @xmath0 symmetry is sensitive to the magnetic flux , in the presence of a large balanced gain and loss , or in a large system .
we find a threshold gain / loss , above which , any nontrivial magnetic flux breaks the @xmath1 symmetry .
we obtain the maximally tolerable magnetic flux for the exact @xmath0-symmetric phase , which is approximately linearly dependent on a weak gain / loss . |
the transient regime of an ising system starting from equilibrium at a uniform temperature and then suddenly coupled to a thermostat at a different temperature , is an issue involving both practical problems and theoretical aspects of non equilibrium .
for instance , as it is well known , cooling a system at a temperature smaller @xcite or equal @xcite to the critical one , gives rise to interesting dynamical out - equilibrium phenomena , where the transient regimes become indefinitely long .
this approach , even for an artificial monte carlo dynamics , provides a deep insight in important physical problems , such as aging phenomena and coarsening of magnetic domains .
we remark that within the usual canonical approach used in these studies ( e.g. metropolis ) , the interaction with the reservoir yields a uniform thermalization , a fact avoiding heat transport .
therefore , in a canonical frame , the investigation of heat propagation during a out - equilibrium warming process is not even possible .
this excludes interesting issues related to transport : which are the relevant parameters of the warming process , the role of boundary conditions , the space - time scales , the evolution of correlations aligned or transversal to the heat flow , etc . @xcite . in this canonical context heat transport
can be studied only at interface separating two systems at different fixed temperatures @xcite .
thus , in order to investigate heat transport phenomena in a transient regime , the appropriate dynamical approach should be the microcanonical one .
this approach allows to couple the system to a thermostat and to investigate how the conserved energy flows within the system itself .
however , so far , this approach has been largely neglected , possibly because of the need of a unique dynamical frame in a wide range of temperatures and/or of underlying geometries , not matched by standard microcanonical rules @xcite .
now , a new efficient microcanonical dynamics has been recently introduced in @xcite , allowing for enhancements in the study of heat flow for a spin lattice coupled to thermostats at arbitrary fixed temperatures .
indeed , among the advantages of this dynamics ( including e.g. the simple definition of local temperatures , the direct computability of conductivity , the possibility of applying to a great variety of discrete topologies and coupling patterns ) , there is also a property that the previous rules lacked , i.e. an efficient behavior in the whole range of temperatures from @xmath0 to @xmath1 .
we remark , that previous results for such a dynamics on homogeneous or disordered systems , are limited to the _ stationary _ regime , which is established after a long - time interaction between the systems and the thermostats @xcite . on the other hand , as already mentioned , here we focus on genuine non - stationary phenomena emerging when the system is suddenly put in contact with a thermostat at different temperature . in particular
, the system under study is an initially ordered rectangular lattice , where at time @xmath0 one side is coupled to a reservoir at fixed temperature @xmath2 , and consequently a heat diffusion takes place in the direction orthogonal to the hot side .
this rectangle , in order to have a permanent state of transience , should be thought of as infinite in the diffusing direction . in practice
, we shall take this size @xmath3 greater than the grid thermalized by the expanding heat front during a long observation time .
periodic boundary conditions are assumed in the transverse direction , so that the geometry reduces to a semi - infinite cylinder .
some basic concepts of out - equilibrium physics , which have been developed for instance in the study of coarsening process , will be very useful also in our approach .
in particular , dynamical scaling is a general tool that on one hand gives a simple description of the system evolution , and on the other it provides a theoretical basis for the introduction of universal concepts , such as correlation length and dynamical exponents . in this perspective , since for our dynamics a diffusive behavior has been proven @xcite , a generic observable @xmath4 is expected to depend on the time @xmath5 and on the distance @xmath6 from the reservoir according to the dynamical scaling : @xmath7 notice that the rescaling procedure allows to introduce a stationary approach even in the transient problem ; indeed , data obtained at different times can be compared to each other by a simple redefinition of distances : @xmath8 where @xmath9 is a characteristic diffusive length .
the most interesting feature we will evidence is that , in the presence of non - stationary heat propagation , the approach to equilibrium may be very slow , even when heating a system at high temperature . in this context
an interesting observable is the correlation @xmath10 , measured at time @xmath5 between the value of spins at sites @xmath11 and @xmath12 of a line orthogonal to the heat flow at distance @xmath6 from the hot side . for @xmath13
this correlation corresponds to the magnetic energy .
since we are interested in the relaxation towards the equilibrium , we shall consider the quantity @xmath14 , where @xmath15 is the correlation function of a system at equilibrium at the temperature of the heat reservoir .
thus , @xmath16 tends to 0 in the limit @xmath17 ; moreover , according to the scaling ansatz @xmath18 is a function @xmath19 . by numerical simulations ,
first we check the scaling hypothesis , then we study the behavior of the scaling functions @xmath20 .
in particular , after extensive monte carlo simulations , we obtain @xmath21 , where @xmath22 . in this context
two simple behaviors emerge , namely @xmath23 and @xmath24 , for large and small temperatures respectively
. we will show that a dynamics characterized by @xmath25 provides a reasonable description of the system evolution at infinite temperature . on the other hand ,
@xmath26 is expected to be a good behavior for finite temperature systems . in this perspective
the complex dependence on @xmath2 displayed by the exponents @xmath27 emerges as due to a dynamical crossover between the dynamics pertaining to finite and infinite temperature respectively .
we notice that the diffusive behavior characterizing the dynamical scaling , as well as the exponential decay of the correlations as a function of @xmath28 , are expected features of the model ; indeed , the correlation decay is typically exponential in the ising model as soon as the magnetic domains are regular , i.e. outside the critical regime . on the other hand ,
the way the correlations tend to their equilibrium values , i.e. by a power law in @xmath29 , is not so obvious .
it seems that the system persists in a transient regime even for times much larger than the characteristic timescale of diffusion , which is proportional to @xmath30 . in some way , even at fixed distance from the thermostat there is not a characteristic time for which the equilibrium could be considered as reached .
we will finally show that such a slow relaxation is a typical feature of open infinite systems where non - steady heat transport is present for all times .
indeed , in a finite system , a rapid decay to the equilibrium value takes place as soon as the diffusion characteristic length @xmath31 is of the order of magnitude of the system size @xmath3 .
for this reason the dynamics considered here can be adopted for the study of the stationary properties of finite systems , as it was done in @xcite .
the paper is organized as follows : in sec .
[ microdyn ] we briefly review the microcanonical dynamics exploited to simulate the evolution of the system , then , in sec . [ sec : numerics ] we present the numerical results focusing on diffusion features , on correlations and on finite - size effects ; finally , sec . [ sec : concl ] is left for conclusions .
in order to be self - contained , we briefly recall the dynamics introduced in @xcite , whose main novelty consists in assigning to the _ links _ , beyond the usual magnetic energy @xmath32 , a sort of `` kinetic '' term @xmath33 which is a non - negative definite quantity able to compensate the variations of magnetic energy due to the spin flips involving the adjacent nodes @xmath34 and @xmath35 .
this makes an essential difference with respect to the creutz microcanonical procedure @xcite , where a _ bounded _ amount of kinetic energy is assigned to each _ node _ : in that case , indeed , both the energy boundedness and the connectivity of the nodes entail many limitations , ranging from geometrical or topological constraints to non - ergodicity at low energy density .
such limitations , as shown in @xcite , are efficiently solved by the new dynamics .
an elementary move for it , starting from a random distribution of link energies , consists in the following : 1 .
select randomly one of the four spin configurations for a randomly extracted link @xmath36 , and evaluate the magnetic energy variation @xmath37 due to this choice .
if the coupling @xmath38 , as in the present case , @xmath37 is an integer whose value depends on the connectivity of the nodes @xmath34 and @xmath35 ; 2 . if @xmath39 , accept the choice and increase the link kinetic energy @xmath33 of @xmath40 ; if @xmath41 , accept the choice and decrease @xmath33 of @xmath40 only if the link energy remains non negative , otherwise reject the choice .
the unit time step consists of @xmath42 moves , where @xmath42 is the total number of links . since
a link is defined only by the adjacent nodes , this rule applies on arbitrary non - oriented connected graphs , providing a great generality to the dynamics .
we remark that at low energies the dynamics is rather time - consuming as for most of the extracted links the move is typically rejected ; yet , differently from the creutz procedure , here the dynamics works at any energy . in @xcite , where a cylindrical lattice was considered ,
many points have been numerically or theoretically supported , for both homogeneous and disordered cases .
for instance , starting from the fact that magnetic and kinetic energies are non - correlated observables , that the boltzmann distribution is recovered at equilibrium and that the system results to be ergodic at all finite temperatures , it is possible to define the temperature @xmath43 by the averaged kinetic energy @xmath44 : this is a link observable , and therefore a local quantity working as a temperature also in stationary states far from equilibrium , i.e. states forced by thermostats at different temperatures .
in order to study the transient regime , we deal here with the same simple cylindrical geometry considered in @xcite .
it consists of a rectangular lattice of @xmath45 sites , periodical in the @xmath46 direction and open in the @xmath47 direction : the first and last columns , discrete circles of @xmath48 sites , are therefore open borders , in contact with heat reservoirs constituted by supplementary columns ( two for each side are enough ) .
these reservoirs are kept at constant temperatures @xmath49 and @xmath50 by using the metropolis dynamics at equilibrium with appropriate temperatures . in most cases , for our present exigences , only the thermostat at the left border is fixed , while the right border is open and virtually placed at infinity ( i.e. very large @xmath51 ) . in the following we report and discuss the results obtained by means of monte carlo ( mc ) simulations performed on cylinders initially set at zero temperature , that is displaying an ordered magnetic configuration and zero kinetic energy on links .
a given site @xmath34 will be denoted by the couple @xmath52 , where the integers @xmath6 and @xmath11 represent the distance from the reservoir and the position along the column respectively . moreover , in order to focus on the main features of the transient , presently we limit to the homogeneous case , i.e. the coupling constants are set equal to @xmath53 for any couple of neighboring sites .
time is measured in mc steps , each corresponding to a total extraction of @xmath45 links , with consequent operations @xmath54 and @xmath55 , as described in the previous section . in this way
we get temporal series for the main observables , such as the average magnetization @xmath56 and the average energy @xmath57 , for the @xmath6-th column at time @xmath5 , being @xmath58 where @xmath59 is the value of the spin on site @xmath60 at time @xmath5 and @xmath61 is the average over different thermal realizations ( typically in our simulation we average over @xmath62 thermal histories ) .
thus , as the simulation is onset , heat starts to propagate within the cylinder : the absolute average magnetization decreases and the average energies of columns close to the thermostat increase .
the first part of simulations is devoted to the analysis of heat propagation , and the longitudinal size is taken rather large and equal to @xmath63 , so that heat can flow for long times without feeling the influence of the second border . in the following part we focus on finite - size effects and we consider relatively small longitudinal sizes , that is @xmath64 . in any case , @xmath48 is kept fixed and equal to @xmath65 ; this large value is useful to avoid finite size effects and improves the statistics of the measures . in the presence of a static heat flux ,
the system behaves diffusively , as already noticed in @xcite .
the present simulations confirm this picture even in a transient regime .
in particular , relevant physical quantities such as the total energy along the @xmath46 direction @xmath66 grows as @xmath67 , that is the characteristic law of diffusive processes ; this asymptotic regime is reached only at very large times due to the extensive routines of the dynamics ( see inset of fig . [
fig : diffusione ] ) . moreover , fig .
[ fig : diffusione ] evidences that focusing on the direction of heat propagation , the system can be studied within the framework of dynamical scaling which is typical of out - equilibrium systems .
more precisely , the quantities ( e.g. the local energy ) depending on both time @xmath5 and distance @xmath6 can be plotted in terms of a single rescaled variable @xmath68 , where @xmath69 is the characteristic length of the process , so that data obtained at different spaces and times can be plotted together on a single curve .
the fact that in our system @xmath70 grows as @xmath67 confirms the diffusive behavior of the model .
of course , an equivalent scaling ( which has been used in our plot ) consists in introducing a rescaled time @xmath71 .
moreover , we remark that the scaling approach is very general and does not depend on the observable we have considered .
indeed we have checked that different observables , such as the magnetization and the correlation functions , display an analogous diffusive dynamical scaling .
[ fig : diffusione ] finally , it is worth deepening the relation between the dynamical scaling introduced above and the generalized fourier equation used in @xcite for the description of the model @xmath72 where @xmath73 is a diffusivity factor which , in quasi - equilibrium states , is just the ratio between the conductivity and the specific heat , so that the standard fourier equation is recovered .
we recall that the previous equation is able to describe the heat propagation in very general conditions , independently of the existence of a local temperature @xcite .
now , it is straightforward to verify that posing @xmath74 leads to @xmath75 i.e. the equation can be rewritten in terms of @xmath76 only . moreover , assuming that @xmath73 is a sufficiently smooth function writable as @xmath77 , from eq .
[ eq : diff2 ] we can derive that the energy can be written as @xmath78 , namely @xmath79 ( at least at long times ) . as we will see in the following subsection , this estimate is consistent with our numerical results .
the transient regime presents interesting properties not only along the direction of heat flow but also along the orthogonal lines . in this framework ,
the simplest observable to consider is the two - point correlation function between spins of the same column at distance @xmath28 , i.e. : @xmath80 clearly , @xmath81 is simply the negative of the energy @xmath57 of the @xmath6-th column , i.e. @xmath82 . since we are interested in the long - time asymptotic behavior , we consider the quantity @xmath83 where @xmath84 is the equilibrium two - point correlation function between spins at distance @xmath28 in a 2-dimensional system evaluated at the same temperature of the thermostat .
therefore , if the system approaches to the equilibrium , we have @xmath85 , as it has been indeed verified in all our numerical simulations . in the ising model outside the critical regime , when simply connected domains are prevalent , the correlation functions decay exponentially with the distance @xmath28 , i.e. as @xmath86 , where @xmath87 is defined as the correlation length . at equilibrium , @xmath87 depends on the temperature , while , in the coarsening processes , it grows as @xmath67 , due to the diffusion of the magnetic domains @xcite .
[ fig : tri ] fig .
[ fig : tri ] shows the behavior of @xmath16 when the thermostat is at very large temperature , i.e. infinite for all practical purposes . in this case
the equilibrium correlation function vanishes .
the dynamical scaling picture is verified , since data evaluated at different columns ( e.g. @xmath88 and @xmath89 ) collapse by using the rescaled variable @xmath90 , i.e. @xmath91 . moreover , for large enough @xmath29 , simulations evidence that @xmath92 therefore , the correlation function presents an exponential decay with the distance @xmath28 and this is peculiar of the ising model outside the critical regime . on the other hand ,
an unexpected and interesting feature is the slow decay of @xmath93 as a function of @xmath29 , consistently with the predictions drawn in the previous section for @xmath13 .
while in the standard mc approach a slow equilibration is typical of _ freezing _ only ( below the critical temperature ) , here we can appreciate a similar behavior also during the warming phase .
in fact , a rapid equilibration would involve an exponential decay in @xmath29 .
in particular , the heating process can not be considered concluded even for those sites whose distance from the thermostat is much smaller than @xmath70 , i.e. the characteristic length of heat diffusion . from a different point of view
, the very slow approach to equilibrium can be described in terms of correlation length @xmath87 which , in our model , turns out to tend to zero logarithmically in @xmath29 ; indeed the correlation function can be written as : @xmath94 = \exp(-n/\lambda ) .
\label{logdecay}\ ] ] we will show that the unexpected slow relaxation at high temperature emerges in open systems , while in the next section we verify that in small systems , where the boundary effects can not be discarded and heat can not flow freely , the thermalization process is exponentially fast at large enough times . the scaling function in eq .
[ logdecay ] presents a very simple structure .
let us define @xmath95 and expand @xmath96 for small @xmath76 ( i.e. large times ) ; we get @xmath97 then , formula ( [ scal_inf ] ) can be recast as @xmath98 if @xmath99 , giving a description , at infinite temperature , of the asymptotic behavior of the model in terms of the analytic properties of @xmath96 .
now , when the border is thermalized at a finite temperature , the main features characterizing the heat transport are preserved : as evidenced in fig .
@xmath100 , scaling still holds with a definite diffusive length , the correlation functions decay exponentially and the approach to equilibrium is slow since the scaling functions vanish as a power law .
however , here the analytic form of @xmath93 is not so simple ; in fact we have @xmath21 , and fitting procedures suggest that @xmath101 as shown in the inset of fig .
@xmath100 .
[ fig : collassat5 ] the exponents characterizing the slow decays appear to be temperature dependent with a non - trivial behavior .
however , fig .
@xmath100 evidences that for high ( practically infinite ) temperature , @xmath25 , while , for low enough temperatures , @xmath24 independently of @xmath28 .
this suggests that two simple ( asymptotic ) behaviors are present in the model : for infinite temperature the scaling function should be given by @xmath102,\ ] ] hence , at infinite temperature , the dynamics is characterized by indefinitely shortening correlation length @xmath103 .
this is clearly consistent with the fact that at infinite temperature there is not an equilibrium characteristic length , since the correlation function between any couple of sites is vanishing . on the other hand , at small temperature
, @xmath93 decays as @xmath104 .
however , since the exponential decay of @xmath105 as a function of the distance @xmath28 is an expected feature , we have @xmath106 where a characteristic length @xmath107 naturally emerges .
we remark that at finite temperature the equilibrium correlation functions decay with a characteristic length @xmath108 .
since a similar behavior is expected to be reached asymptotically also out of equilibrium , we expect that , eventually , @xmath109 .
therefore the above regimes ( [ tinf],[tfin ] ) provide a reasonable description for infinite and finite temperature respectively . in this perspective , the complex behavior characterizing intermediate temperatures would result from an interplay between these regimes .
in particular , when the thermostat is at a finite temperature @xmath2 , @xmath19 evolves according to eq .
( [ tinf ] ) until the correlation length @xmath110 becomes comparable to @xmath109 ; afterwards an evolution described by ( [ tfin ] ) takes place . at very large temperatures
@xmath109 is almost zero and , in the considered time regimes , @xmath111 , therefore only the infinite temperature evolution ( [ tinf ] ) is observed . at low temperatures
the correlation @xmath107 is so large that @xmath110 becomes quickly comparable to @xmath107 and @xmath112 evolves according to ( [ tfin ] ) .
finally , the temperature dependent exponents @xmath27 observed at intermediate temperatures should be the result of the crossover between the two regimes which is ongoing in the considered time window .
let us now discuss the finite - size effects in the processes of heat transfer from the thermostat to the system .
finite - size effects can clarify some features outlined in the previous analysis , in particular the mutual influence between the transport regime and the way the asymptotic values are approached .
@xmath113 shows the behavior of @xmath114 as a function of time for the column @xmath88 and a size @xmath64 . differently from the behavior depicted in fig .
[ fig : diffusione ] , here the dynamics speeds up , approaching quickly the equilibrium value , for large enough times , i.e. when @xmath70 is comparable to the lattice size .
therefore , the presence of very long transients described by a power law seems to be typical of infinite systems , while finite systems thermalize in a rapid ( possibly exponential ) way .
this differences ensure that the dynamics may be used for the study of the stationary properties of finite systems as it has been done in @xcite
. moreover , fig .
@xmath113 evidences that the characteristic time for equilibration increases with the temperature becoming very large at @xmath115 .
[ fig : finite ]
the dynamical insight on the warming process in the ising model we have studied has two main aspects : first , a direct confirmation of such expected features as the diffusive behavior in the heat propagation , or the exponential decay of correlations in the transversal direction ; second , a clear indication about the role of heat flux in keeping the time behavior of correlations well distinct from their time - asymptotic `` equilibrium '' value .
indeed , in the indefinite propagation taking place when the longitudinal size is much longer than the observation window , the approach to the limit obeys a power law in time .
remarkably , a slow relaxation seems to be present for all temperatures .
in fact , one could in principle conjecture the existence of a special temperature ( possibly related to the critical one ) separating a slow and a fast dynamical regime ; yet the persistent slowness of the dynamics proves the non - existence of such a dynamical threshold . in particular
, the explanation of the non - trivial behavior of the temperature dependent exponents , in terms of finite and infinite temperature regimes , confirms the fact that other intermediate thresholds do not exist .
this work has been partially supported by the miur project prin 2008 `` nonlinearity and disorder in classical and quantum processes '' , and by the firb project rbfr08ekev .
ea is grateful to the italian foundation `` angelo della riccia '' for financial support .
humayun k , bray aj , 1991 _
j. phys .
a _ * 24 * 1915 | we study the warming process of a semi - infinite cylindrical ising lattice initially ordered and coupled at the boundary to a heat reservoir .
the adoption of a proper microcanonical dynamics allows a detailed study of the time evolution of the system .
as expected , thermal propagation displays a diffusive character and the spatial correlations decay exponentially in the direction orthogonal to the heat flow .
however , we show that the approach to equilibrium presents an unexpected slow behavior . in particular , when the thermostat is at infinite temperature , correlations decay to their asymptotic values by a power law .
this can be rephrased in terms of a correlation length vanishing logarithmically with time . at finite temperature , the approach to equilibrium is also a power law , but the exponents depend on the temperature in a non - trivial way .
this complex behavior could be explained in terms of two dynamical regimes characterizing finite and infinite temperatures , respectively .
when finite sizes are considered , we evidence the emergence of a much more rapid equilibration , and this confirms that the microcanonical dynamics can be successfully applied on finite structures .
indeed , the slowness exhibited by correlations in approaching the asymptotic values are expected to be related to the presence of an unsteady heat flow in an infinite system . |
high - order harmonic generation ( hhg ) is an extremely nonlinear optical process that provides coherent radiation in the extreme ultraviolet spectral region on an ultrafast time scale .
such radiation is of great interest for applications such as coherent diffractive imaging of nanostructures @xcite , pump - probe spectroscopy for studying dynamical electronic processes inside molecules @xcite and attosecond science @xcite .
the range of generated wavelengths covered by the harmonics is usually limited , on the long wavelength side by reabsorption in the generating medium , and on the short wavelength side by lack of phase matching due to ionization .
for example , the maximum harmonic order ( the shortest wavelength ) which can be phase matched in argon is the 35@xmath0 order ( @xmath124 nm ) when using femtosecond pulses from an 800 nm drive laser @xcite .
the laser intensities required for generating radiation of even shorter wavelengths are so high that a significant fraction of the gaseous medium becomes ionized . as a result of the generated transient plasma
, there is an additional refractive index contribution which renders phase matching techniques ineffective due to wave - vector mismatch . a technique that is generally seen as suitable for overcoming this ionization limitation and that allows one to efficiently generate higher order harmonics is so - called _ quasi - phase matching _ ( qpm ) @xcite .
with the introduction of a spatially periodic structure of suitable periodicity , an extra wave vector can be provided that compensates the wave - vector mismatch .
various schemes to demonstrate or make use of qpm in hhg have been proposed and developed .
the idea is that efficient hhg is only enabled in the in - phase region and is suppressed in the out - of - phase region . as an example
, a train of tunable counter - propagating pulses in a hollow - core waveguide has been used to introduce a spatially periodic ionization pattern for the generation of photon energies of several hundred electron volts @xcite .
other schemes include waveguides with a spatially periodic diameter @xcite , multimode beating in a capillary @xcite and polarization - beating in a birefringent waveguide @xcite .
recently , a small set of periodically spaced dual - gas jets , consisting of the generating gas ( argon ) and a passive gas ( hydrogen ) , has shown to shift harmonic generation up to the 41@xmath2 order ( @xmath119 nm ) @xcite .
all these approaches are based on the assumption that , depending on the specific set of parameters used , there is a certain spatial period , @xmath3 , that provides phase matching .
this assumption is much in analogy with second - order nonlinear conversion in periodically poled crystals where a particular conversion process with given input and output wavelengths requires a well - defined poling period .
however , selecting an appropriate periodic structure for achieving qpm in hhg is actually rather problematic , since the wave - vector mismatch is strongly time dependent throughout the entire drive laser pulse .
a fixed periodic structure may only be effective in achieving qpm during a certain short time interval within the drive laser pulse , such that the time - integrated ( total ) output pulse energy may still remain low . to demonstrate qpm for hhg in experiments using the schemes mentioned above , the selection of a fixed - periodic structure for maximizing the total high - order harmonic ( hh ) output is usually based on an unproven assumption , namely , that the quasi - phase matching period has to be chosen equal to twice the coherence length calculated for the time - instance where the intensity of the drive laser pulse is maximum ( at peak intensity ) .
this is based on the argument that the nonlinearly induced dipole moment of the individual atoms is the highest at that time @xcite . however , this assumption neglects any radiation that may be generated before and after the peak of the drive laser pulse .
a direct experimental clarification is not known to us and difficult to carry out , due to the ultrashort time scales . instead , one is usually detecting or optimizing any phase matching in hhg via the output pulse energy .
thus , for a proper discussion of optimizing qpm in hhg , one requires a time - dependent model of the output followed by a calculation of the output pulse energy . here , we present the first time - dependent model for quasi - phase matching in high - order harmonic generation .
we call this an extended model because it combines previously known physical effects into a more encompassing theory that delivers somewhat deviating predictions for optimizing the harmonic output . using a one - dimensional model , we include the full wave - vector mismatch with its ultrafast time dependence to calculate the hh field vs. time .
the pulse energy is then obtained by time integration . for definiteness , and as a typical example , we consider hhg in argon gas where a periodic density pattern is provided , e.g. , with a grid in a supersonic jet .
we show that the maximum harmonic pulse energy is obtained when choosing the qpm period for phase matching in the leading edge of the drive laser pulse , rather than at the peak .
as our basis for investigating quasi - phase matching , we recall the well - known model of constant et al .
@xcite and kazamias et al .
@xcite , which describes standard phase matching as achievable via pressure adjustment at weakly ionizing intensities .
the instantaneous field for a specific harmonic order , @xmath4 , generated at the exit of the medium , @xmath5 , can be expressed as @xmath6\mathrm{exp}\left[i\phi_q(z , t)\right]dz.\ ] ] here , @xmath7 is the length of the medium , @xmath8 is the propagation coordinate , @xmath9 is the ionization fraction , @xmath10 is the initial atomic number density , @xmath11 $ ] is the absorption length , where @xmath12 is the ionization cross section @xcite , and @xmath13 is the accumulated phase difference between the drive laser field and the generated harmonic field at the exit of the medium .
the drive laser field is implicitly included in the induced atomic dipole moment amplitude , @xmath14 . based on the full quantum theory developed by lewenstein et al .
@xcite , @xmath15 exhibits a strong dependence on time through its variation with the drive laser intensity , @xmath16 , and can be described approximately by the following power law @xcite .
@xmath17 here , @xmath18 is the threshold laser intensity to be reached before there is a noticeable generation of the @xmath19 harmonic .
the factor @xmath20 in eq . originates from ionization - induced ground state depletion , which indicates that the generated harmonic field scales with the number of neutral atoms .
the ionization fraction , @xmath9 , can be calculated using the ammosov - delone - krainov ( adk ) model @xcite . , ( red line ) and the harmonic dipole amplitude , @xmath21 , for the 39@xmath0 harmonic ( @xmath22 nm , blue line ) calculated for argon and a drive laser pulse ( gaussian temporal shape ) with a central wavelength at 800 nm , a pulse duration of @xmath23 fs ( fwhm ) and a peak intensity of @xmath24 .
the drive laser pulse envelop ( black dashed line ) is plotted for reference.,scaledwidth=80.0% ] to give an example for the typical temporal variation of the ionization fraction and atomic harmonic dipole amplitude , fig .
[ fig : ionization_fraction ] shows a calculation of the ionization fraction and the 39@xmath0 harmonic ( @xmath22 nm ) dipole amplitude , @xmath25 , of argon for a drive laser pulse with a peak intensity of @xmath24 .
the laser pulse has a gaussian shape with a pulse duration of @xmath23 fs ( fwhm ) and a central laser wavelength of 800 nm .
as can be seen from the red curve , the ionization fraction reaches a value of about 18% at the peak of the laser intensity and raises further to about 34% towards the end of the laser pulse .
it can further be noticed that the ionization fraction grows with each optical half - cycle resembling a stepwise increase ( here 30 steps in total ) .
the dipole amplitude exhibits a strong dependence on the drive laser intensity through the threshold and power law in eq . .
the term @xmath13 in eq . ,
is directly related to the wave - vector mismatch , @xmath26 , between the drive laser and the generated harmonics .
the accumulated phase difference after a propagation length of @xmath7 can be described as @xmath27 here , as will be detailed below , the mismatch , @xmath26 , originates from the dispersion due to neutral gas atoms ( @xmath28 ) , electrons in the plasma that are generated via ionization ( @xmath29 ) , the gouy phase shift , which depends on the focusing geometry of the drive laser ( @xmath30 ) , and the intrinsic harmonic dipole phase ( @xmath31 ) , and can be written as follows @xcite : @xmath32 the contribution of the neutral atomic dispersion to the total wave - vector mismatch is positive and can be expressed by @xmath33\frac{2\pi{p(z)}}{\lambda}\delta{n}.\ ] ] here , @xmath34 is the pressure of the gas in mbar , @xmath35 is the wavelength of the drive laser and @xmath36 is the difference between the indices of refraction of the gas per mbar at the drive laser wavelength , @xmath35 , and the high harmonic wavelength , @xmath37 .
the dependence on @xmath38 shows that the atomic dispersion decreases during the drive laser pulse .
the contribution due to the free electrons in the plasma , on the contrary , is negative and becomes @xmath39 where @xmath40 is the initial number density of atoms per mbar , and @xmath41 is the classical radius of the electron .
this contribution via @xmath38 grows in magnitude vs. time .
the wave - vector mismatch that arises from the gouy phase shift , resulting from the focusing of the drive laser beam , only varies in space .
it also has a negative sign and can be expressed by @xmath42 where @xmath43 is the position of the focus and @xmath44 is the rayleigh range .
finally , the contribution of the intrinsic harmonic dipole phase originates from the intensity dependence of the trajectories that the transiently ionized electrons follow before recombination and harmonic generation , which can be described as @xcite @xmath45 ^ 2}\alpha_qi_0,\ ] ] where @xmath46 is a coefficient related to the trajectories which the electrons take .
this expression can be chosen to be either positive or negative depending on the position of the focus with regard to the generating medium . placing the medium in front of ( behind ) the focus , where @xmath47 ( @xmath48 ) , results in a negative ( positive ) term . when the focus is placed in the center of the medium ( @xmath49 ) , such as to maximize the drive laser intensity ,
the value of @xmath50 becomes equal to zero .
the terms described so far can now be used to calculate the spatial and temporal evolution of the harmonic field , as expressed in eq . .
the instantaneous harmonic intensity of the @xmath19 harmonic at the exit of the medium is obtained from @xmath51 , where @xmath52 and @xmath53 are the speed of light and the vacuum permittivity respectively .
the on - axis harmonic pulse energy of the @xmath19 harmonic , @xmath54 , can be obtained by temporal integration over the entire pulse and multiplying with a cross - sectional area , @xmath55 , @xmath56 the described model has been widely and successfully applied to analyze measurements of hhg under various experimental conditions such as guided - wave geometries @xcite , or loose focusing geometries @xcite . in the next section , we will use the model in its simplest form , namely , for the case of a homogeneous medium ( argon ) with a loosely focused drive laser beam . in detail
, we will present the consequence of the time dependence in the wave - vector mismatch .
in the case of a loose focusing geometry for hhg , the rayleigh range is chosen to be long , typically in the order of tens of millimeters such that it becomes much longer than the length of the medium ( @xmath57 ) . although
loose focusing reduces the drive laser intensity , the advantage is that a large interaction volume can be provided for maximizing the total harmonic pulse energy .
usually a relatively low gas pressure is required to maintain phase matching @xcite . in this geometry
, the drive laser intensity stays approximately constant along the entire generating medium .
therefore , the harmonic dipole amplitude , @xmath14 , the interaction area , @xmath58 , and also the ionization fraction , @xmath38 , become space independent . due to the long rayleigh range
, the geometry - induced wave - vector contribution can be simplified as @xmath59 and becomes independent of @xmath8 .
in addition , the harmonic dipole phase contribution , @xmath50 , becomes much smaller than the other terms @xcite and can be neglected . as a result
, the total wave - vector mismatch can be regarded as a space - independent parameter , @xmath60 , such that the phase term , @xmath13 , in eq . can be simply represented as the product of @xmath60 and @xmath8 : @xmath61 here , the time - dependent coherence length , @xmath62 , is introduced for characterizing the wave - vector mismatch in comparison to other length parameters , specifically the length of the medium .
the physical meaning of @xmath63 is that the @xmath19 harmonic radiation generated at the end of this length is shifted out of phase with respect to the harmonic radiation generated at the beginning of the length .
this leads to back conversion in further propagation , often also termed as destructive inteference . by implementing a time - dependent coherence length , eq .
can be integrated analytically with the instantaneous hh intensity being @xcite @xmath64 ^ 2d_q(t)^2\frac{4l_{abs}^2}{1 + 4\pi^2\left[l_{abs}^2/l_{coh}(t)^2\right]}\\ \times\left[1+\mathrm{exp}\left(-\frac{l_{med}}{l_{abs}}\right)-2\mathrm{cos}\left(\pi\frac{l_{med}}{l_{coh}(t)}\right)\mathrm{exp}\left(-\frac{l_{med}}{2l_{abs}}\right)\right ] .
\label{eq : anaylitical solution of high harmonic field}\end{gathered}\ ] ] in this expression , reabsorption of harmonic radiation is taken into account via an absorption length , @xmath65 . from eq .
it can be seen that , ideally , the maximum harmonic output can be obtained when perfect phase matching is achieved , namely , @xmath66 ( equivalent to an infinitely long coherence length ) during the entire drive laser pulse . in particular , when absorption is neglected ( infinitely long absorption length ) the harmonic intensity would grow proportionally to the square of the medium length . for the experimentally more relevant case
that phase matching is not ideally fulfilled ( @xmath67 ) and can be approximated by a temporally constant and finite coherence length , i.e. , when the ionization fraction does not change much vs. time , and when additionally absorption can not be neglected , a coarse criterion , @xmath68 , has been proposed @xcite .
if this criterion is fulfilled , the harmonics are efficiently generated and the output is mainly limited by reabsorption .
quantitatively , for instance , when @xmath69 and @xmath70 , the harmonic intensity reaches more than half of the maximum intensity of the ideal case ( @xmath71 , @xmath72 ) . however , when generating high - order harmonics with an increased intensity and appreciable gas density , the named criterion becomes inapplicable because the coherence length becomes markedly . in that case , the coherence length varies strongly on an ultrafast time scale due to a dramatic change of the ionization fraction of the medium during the drive pulse .
therefore , phase matching is only achieved in a very short time interval during the drive pulse . as a result , for obtaining a maximum harmonic output pulse energy
, it remains questionable to what fraction and at what instance during the drive laser pulse the criterion should be fulfilled .
a further complication is that the harmonic dipole amplitude , @xmath73 , is also a time - dependent parameter that will influence the hh output .
for instance , even if the above length criterion is fulfilled for a certain time interval , the output pulse energy may still be far below optimum if @xmath74 is small during that interval . in view of these effects and to translate them properly into a situation aiming for quasi - phase matching in sec .
[ sec : hhg in a periodic density modulated medium ] , it is important to analyze the time dependence of the coherence length as well as of the harmonic dipole amplitude over the entire drive laser pulse . here , for definiteness , we take two harmonic orders for hhg in argon as an example , namely , a moderate harmonic order , the @xmath75 harmonic ( @xmath76 nm , hh21 ) , and a relatively high harmonic order , the @xmath77 harmonic ( @xmath22 nm , hh39 ) .
harmonic ( a ) and the @xmath77 harmonic ( b ) as a function of time .
assumed is harmonic generation in 30 mbar of argon in a 5 mm long medium ( @xmath78 mm , dashed blue lines ) with a gaussian - shaped laser pulse of @xmath79 peak intensity ( blue dashed curves ) .
the absorption lengths ( dashed pink lines ) and the minimum required laser intensity ( solid orange lines ) are calculated for both harmonics and plotted for comparison .
the green areas show within which time interval there is a sufficiently high drive laser intensity for hhg ( @xmath80 ) .
the red areas indicate the time interval during which the phase - matching condition is approximately fulfilled , i.e. , the harmonic output from the entire medium adds constructively ( @xmath81 ) , if the respective harmonic is generated.,title="fig:",scaledwidth=80.0% ] harmonic ( a ) and the @xmath77 harmonic ( b ) as a function of time .
assumed is harmonic generation in 30 mbar of argon in a 5 mm long medium ( @xmath78 mm , dashed blue lines ) with a gaussian - shaped laser pulse of @xmath79 peak intensity ( blue dashed curves ) .
the absorption lengths ( dashed pink lines ) and the minimum required laser intensity ( solid orange lines ) are calculated for both harmonics and plotted for comparison .
the green areas show within which time interval there is a sufficiently high drive laser intensity for hhg ( @xmath80 ) .
the red areas indicate the time interval during which the phase - matching condition is approximately fulfilled , i.e. , the harmonic output from the entire medium adds constructively ( @xmath81 ) , if the respective harmonic is generated.,title="fig:",scaledwidth=80.0% ] fig .
[ fig : coherence length for 21st hh and 39th hh ] shows the time - dependent coherence lengths ( black curves ) for the @xmath75 harmonic ( a ) and the @xmath77 harmonic ( b ) calculated for a hhg experiment in a typical loose focusing configuration , assuming a drive laser peak intensity of @xmath82 , a rayleigh range of @xmath44 = 18 mm as achieved by a f/150 focusing lens , a medium length of @xmath78 mm and an argon gas pressure of 30 mbar . the laser pulse ( also shown in fig .
[ fig : coherence length for 21st hh and 39th hh ] ) is again taken to have a gaussian shape ( blue dashed curve ) with a pulse duration of @xmath83 ( fwhm ) .
the lengths of the medium ( dashed blue lines ) and the calculated absorption lengths ( dashed pink line ) for both harmonics are also plotted for comparison .
furthermore , the green areas in fig . [ fig : coherence length for 21st hh and 39th hh ] ( a ) and ( b ) indicate the time interval of sufficiently high intensity in which , potentially , each respective harmonic can be generated ( @xmath84 ) .
it can be clearly seen that for both harmonics , the coherence length shows an ultrafast time dependence .
specifically , starting from a certain mismatch , there is first a rapid increase of @xmath63 in some part of the leading edge of the drive laser pulse .
initially , the dispersion of free electrons due to the negligible ionization fraction is relatively small , and the other terms in eq .
however , with the increase of the drive laser intensity , the ionization fraction rapidly increases further .
consequently , the dispersion in the plasma due to the accumulation of free electrons overwhelms the other terms in eq .
, resulting in a quick drop of the coherence length .
the short duration of the peak in the coherence length leaves only an extremely short interval ( a few femtoseconds long ) where the dispersion is close to zero ( phase matching ) .
for the @xmath75 harmonic ( fig .
[ fig : coherence length for 21st hh and 39th hh ] ( a ) ) , the time interval of phase matched generation is indicated by the red area .
this interval is found in the leading edge of the drive laser pulse ( where the criterion @xmath85 , is fulfilled ) .
it can be seen that this interval overlaps with the time interval where the drive laser intensity is sufficiently high for generating a noticeable @xmath75-order dipole moment , which is indicated as the green area .
the overlap of the red with the green area indicates that the @xmath75 harmonic can be efficiently generated at least within the short overlap time interval . for the @xmath77 harmonic ( fig .
[ fig : coherence length for 21st hh and 39th hh ] ( b ) ) , the coherence length reaches a very large value and also becomes longer than the length of the medium in a short time interval at the leading edge of the pulse ( red area ) .
however , in contrast to the @xmath75 harmonic , this interval is located outside the time interval in which the @xmath77 harmonic can be generated ( green area , where @xmath86 ) .
thus , the red ( phase matching ) and green ( generation ) intervals do not overlap , which indicates that the ionization fraction induced by the minimum required drive laser intensity for generating the @xmath77 harmonic is too high for phase matching .
the wave - vector mismatch , @xmath87 , due to the large amount of free electrons dominates over the other wave - vector contributions . in this example , for argon , this occurs for harmonic orders higher than @xmath88 .
consequently , the @xmath77 harmonic will not be efficiently generated at any moment throughout the entire drive laser pulse .
( black ) and the 39@xmath0 harmonic ( red ) as a function of the medium length . for @xmath89 ,
the 39@xmath0 harmonic output is about four to five orders of magnitude weaker.,scaledwidth=70.0% ] also , to quantitatively illustrate the difference between the two considered harmonics , we have calculated the on - axis harmonic output pulse energy , @xmath54 , as a function of interaction length using eqs . and .
this is plotted in fig .
[ fig : hh intensity for 21st hh and 39th hh ] .
note that the cross - sectional area , @xmath55 , is treated as a constant and can be calculated from the waist of the drive laser beam at the focus .
it can be seen that the output energy of the @xmath75 harmonic ( black curve ) grows strongly within the first few mm of the medium and saturates after roughly 3 to 4 mm .
this observation shows that the harmonic is efficiently generated and only limited by reabsorption .
on the other hand , the output pulse energy of the @xmath77 harmonic ( red curve , multiplied by 5000 in order to be visible in the graph ) does not show any obvious growth with the length of the medium and is ( beyond @xmath90 ) approximately four to five orders of magnitude weaker .
a few oscillations can be seen along the length of the medium , which is due to varying levels of destructive interference . from a series of additional calculations ,
we conclude that for moderate harmonic orders up to @xmath88 in this example , the length criterion for efficient output ( @xmath68 ) can be fulfilled over at least some certain short time interval falling within the interval that the drive laser intensity is above the generation threshold .
this allows for harmonic generation that is limited only by absorption .
however , for harmonics of @xmath91 ( such as in the example above , @xmath92 ) , the criterion can never be fulfilled at any instance during the drive laser pulse , which leads to a generation efficiency that is decreased by many orders of magnitude .
a solution to overcome such low efficiencies for higher orders is to implement qpm techniques into the hhg configuration . in section [ sec :
hhg in a periodic density modulated medium ] , we present a specific maximally simplified qpm scheme based on providing a periodic density modulation .
the basic idea of quasi - phase matching ( qpm ) is to introduce some periodic modulation of the local strength or phase of the generation process throughout the medium . to counteract the destructive interference between
locally generated harmonic fields , i.e. , generated in subsequent coherence lengths , a straightforward approach would be to suppress the harmonic generation within every other coherence length . by repeating the suppression periodically in the spatial domain ,
such modulation is equivalent to providing an additional wave - vector contribution in the fourier domain , @xmath93 , that compensates the remaining wave - vector mismatch ( @xmath94 in eq . ) . as was shown above for harmonics in argon with an order beyond @xmath88 , the mismatch that needs to be compensated originates predominately from the plasma dispersion .
the total wave - vector mismatch @xmath95 , for hhg including qpm can be expressed as @xmath96 . , and a modulation depth , @xmath97 .
the lengths of the high - density and low - density regions are @xmath98 and @xmath99 , respectively.,scaledwidth=80.0% ] the latter expression looks simple .
however , two points have to be noted . while the mismatch @xmath26 changes on an ultrafast time scale , the additional wave - vector , @xmath93 , essentially has a temporally constant value in the case of using a periodic density modulated medium because it is impossible to change any gas density distribution on an ultrashort time scale as well .
consequently , quasi - phase matching will only be fulfilled transiently again , for a certain short time interval within the drive laser pulse , as discussed above .
moreover , because the density of the medium is spatially modulated , the mismatch attains an additional , periodic space dependence
. we will consider here a periodic modulation of the gas density along the propagation axis , as is depicted in fig .
[ fig : density modulated medium ] , where we assume a rectangular variation for simplicity .
similar density distribution might be achieved , e.g. , by placing a periodic grid in a supersonic gas jet . to express the density modulation quantitatively
, the length of the medium can be written as : @xmath100 , where @xmath101 is the number of modulation periods , @xmath3 is the modulation period , and @xmath98 and @xmath99 are the lengths of the high - density and low - density regions , respectively . the corresponding modulated gas density profile along the propagation axis of the drive laser , @xmath8 , is shown in fig .
[ fig : density modulated medium ] ( b ) using the simplifying assumption of homogeneous densities within @xmath98 and @xmath99 .
we specify the modulation depth of the gas density with a factor @xmath97 . when @xmath97=1 , the gas density is fully modulated , while if @xmath97=0 , there is no modulation imposed on the gas . to obtain the instantaneous on - axis harmonic field , @xmath102 , in a density modulated medium with a loose focusing geometry
, one has to take into account that several additional parameters become space dependent in the integral in eq .
, specifically , the gas density , @xmath10 , the phase term , @xmath13 and the absorption length , @xmath103 . to evaluate eq . , we rewrite the integral as a sum over the field contributions from the high - density and low - density regions as @xmath104,\ ] ] where @xmath105 and @xmath106 are the @xmath19 harmonic field contributions generated in the high - density and low - density regions of the @xmath107 modulation period , respectively . the phase term , @xmath108 , can be calculated from the wave - vector mismatch in the regions of different densities as follows @xmath109\\ + \delta{k}_{high}(t)\cdot{z}\;\;\;\;\;\;\;\;\rm{in\;any\;high{-}density\;region } , \\
( n-1)\left[\delta{k}_{high}(t)l_{high}+\delta{k}_{low}(t)l_{low}\right]\\ + \delta{k}_{high}(t)l_{high}+\delta{k}_{low}(t)\cdot{z}\;\;\;\;\;\;\;\ ; \rm{in\;any\;low{-}density\;region}. \end{aligned } \right.\ ] ] here , @xmath110 and @xmath111 is the wave - vector mismatch in the high - density and low - density regions , respectively .
similarly , the coherence length in the high - density and low - density regions can be written as : @xmath112 and @xmath113 . next , in order to investigate the influence of qpm on efficient harmonic generation rather than on limitations due to reabsorption , we restrict ourselves to the situation that the length of the medium is shorter than the absorption length . as was described in sec .
[ sec : hhg in a homogeneous density medium ] , this can be achieved with a sufficiently low gas pressure in a loose focusing geometry .
indeed , the absorption lengths for the harmonics of higher - valued order ( @xmath91 ) are typically much longer than the length of the medium ( 5 mm ) over a broad tuning range of the gas pressure ( density ) .
for example , the absorption length for the @xmath77 harmonic at a pressure of 30 mbar is about 12 mm long even at non - modulated gas density . using the equation for the sum of a geometric series , eq .
, the expression for the instantaneous on - axis harmonic field can be simplified , and reads @xmath114 where @xmath115 is the sum of generated harmonic field contributions from only the high - density regions : @xmath116\frac{1-\mathrm{exp}\left[in\pi\left(\frac{l_{high}}{l_{coh}^{high}(t)}+\frac{l_{low}}{l_{coh}^{low}(t)}\right)\right]}{1-\mathrm{exp}\left[i\pi\left(\frac{l_{high}}{l_{coh}^{high}(t)}+\frac{l_{low}}{l_{coh}^{low}(t)}\right)\right ] } , \label{eq : e high}\ ] ] and where @xmath117 is the generated harmonic field contribution from only the low - density regions : @xmath118\mathrm{exp}\left[i\frac{\pi{l_{high}}}{l_{coh}^{high}(t)}\right]\\ \times\frac{1-\mathrm{exp}\left[in\pi\left(\frac{l_{high}}{l_{coh}^{high}(t)}+\frac{l_{low}}{l_{coh}^{low}(t)}\right)\right]}{1-\mathrm{exp}\left[i\pi\left(\frac{l_{high}}{l_{coh}^{high}(t)}+\frac{l_{low}}{l_{coh}^{low}(t)}\right)\right]}. \label{eq : e low}\end{gathered}\ ] ] with these expressions , the instantaneous on - axis intensity , @xmath119 , for the @xmath19 harmonic from a density modulated medium becomes @xmath120 the corresponding on - axis hh pulse energy of the @xmath19 harmonic , @xmath54 , can be obtained again by integration over the entire pulse ( eq . ) . from eqs . and
, it can be seen again that the coherence length ( in both the high - density and low - density regions , @xmath121 and @xmath122 ) varies strongly on an ultrafast time scale via the ionization fraction .
therefore , similar to standard phase matching discussed above for lower - order harmonics , qpm can only be transiently achieved in a short time interval within the drive laser pulse . however , by choosing the value of the modulation period , this time interval can be shifted to a proper instance , such that the hh pulse energy is maximized .
in this section , we describe quasi - phase matched generation of the @xmath77 harmonic ( @xmath22 mm ) .
we consider this specific harmonic again , because it represents the typical case where standard phase matching can not be achieved , as was described in sec .
[ sec : hhg in a homogeneous density medium ] .
we assume the same loose focusing parameters given in sec . [
sec : hhg in a homogeneous density medium ] ( a drive laser peak intensity @xmath123 @xmath124 , a rayleigh range @xmath44 = 18 mm achieved by a f/150 focusing lens , @xmath78 mm ) and the same argon gas pressure ( 30 mbar ) . based on the qpm model described in sec .
[ sec : hhg in a periodic density modulated medium ] , we calculate the instantaneous on - axis hh intensity , @xmath125 , and the harmonic pulse energy , @xmath126 , by applying different modulation periods , @xmath3 , and modulation depths , @xmath97 .
we note that , to re - phase the harmonic wave with the fundamental wave after each coherence length in the high - density regions , the low - density regions have to provide a phase shift of @xmath127 between the generated harmonic wave and the fundamental wave .
this means that also the low - density regions have to be as long as a coherence length , which is dependent of the modulation depth , @xmath97 . as a result ,
the desired modulation period should be equal to the sum of the coherence lengths calculated in both high - density and low - density regions , @xmath128 . in order to choose an appropriate modulation period , @xmath3 , the time - dependent coherence length for both high - density and low - density regions needs to be calculated .
harmonic ( @xmath22 mm ) as a function of time for various modulation depths , @xmath97 ( 1 ( purple ) , 0.8 ( red ) , 0.6 ( grey ) , 0.4 ( cyan ) , 0.2 ( orange ) and 0 ( black ) ) .
the ionization fraction ( dark green curve ) and the temporal profile of the drive laser intensity ( blue dashed curve ) are plotted for reference .
the red dashed line is the threshold intensity for the 39@xmath0 harmonic generation and the green area indicates the time interval ( from @xmath129 to @xmath130 ) during which the generation of the @xmath77 harmonic is possible .
the generating medium is 5 mm long and the gas pressure is set at 30 mbar in the high - density region .
the drive laser pulse has a peak intensity of @xmath79.,scaledwidth=80.0% ] fig .
[ fig : different coherence lengths for 39hh ] shows the calculated coherence lengths for the @xmath77 harmonic in the low - density regions for a set of values ( @xmath131 ) and high - density region ( @xmath132 ) as a function of time .
the modulation depth , @xmath97 , is varied from 0 to 1 in steps of 0.2 .
the drive laser pulse is shown as the blue dashed curve .
the green area indicates again the time interval where the drive laser intensity is above the intensity threshold ( red dashed line ) for generating the @xmath77 harmonic . in fig .
[ fig : different coherence lengths for 39hh ] , for easier discussion , three characteristic times are introduced and shown along the top axis , where @xmath133 represents the on - set of the generation of the @xmath77 harmonic , @xmath134 is the moment at which the drive laser pulse reaches its peak intensity , and @xmath135 indicates the time when the drive laser intensity drops below the intensity threshold for generating the @xmath77 harmonic . for full modulation ( @xmath136 , purple line ) ,
the coherence length for the low - density region becomes independent of time due to the absence of any gas that can be ionized . in this limiting case , the coherence length is only given by the wave - vector mismatch , @xmath137 , due to the gouy phase shift , which is set by the focusing geometry .
the coherence lengths for the modulation depths with @xmath138 are strongly time - dependent and drop rapidly during the pulse due to the increasing ionization fraction ( dark green curve ) , specifically within the time interval where the @xmath77 harmonic is generated ( green area ) . the information in the figure can be used as follows : if the harmonic output is to be maximized for a particular case , one can read the required qpm period from the figure .
for instance , if the output is to be maximized in a short time interval at the peak of the drive laser pulse , the crossing of the coherence lengths with the dashed vertical line at that time ( @xmath139 ) gives the length of both the high - density and low - density regions that yields qpm at that time .
for instance , for @xmath140 , one obtains @xmath141 mm ( red curve ) and @xmath142 mm ( black curve ) , resulting in a modulation period of @xmath143 mm . now that we have illustrated how to choose the modulation period that maximizes @xmath144 in eq . at the peak of the drive laser pulse , @xmath139 , we present in fig .
[ fig : hh39 output at different time ] a calculation of how the hh intensity ( @xmath145 ) generated at the drive laser peak grows vs. the propagation coordinate through the medium , @xmath8 . for comparison
, we also plot the spatial growth of @xmath145 generated at other characteristic instances of the drive laser pulse in the leading edge ( @xmath146 - 5.7 fs ) and the trailing edge ( @xmath1475.7 fs ) . with the modulation period , @xmath3 , set to provide qpm at @xmath139
, we also display the influence of the modulation contrast by choosing three different modulation depths . for reference
, we also plot the growth of @xmath145 generated at the peak of the drive laser pulse in a homogeneous medium , i.e. , without density modulation .
, throughout the density modulated medium for ( a ) @xmath97=1 , ( b ) @xmath97=0.8 and ( c ) @xmath97=0.6 when the modulation period , @xmath3 , is chosen to induce qpm at the peak of the drive laser pulse ( @xmath139=0 fs ) .
the black curves show the growth of @xmath145 generated at the drive laser peak , while the other curves show the growth of @xmath145 at an earlier and later time in the pulse ( @xmath148=-5.7 fs ( red ) and @xmath149=5.7 fs ( blue ) ) .
the propagation coordinate is expressed relative to the modulation period .
the pink dashed line represents @xmath150 generated at @xmath139 in a homogeneous medium , without density modulation ( @xmath97=0).,title="fig:",scaledwidth=62.0% ] , throughout the density modulated medium for ( a ) @xmath97=1 , ( b ) @xmath97=0.8 and ( c ) @xmath97=0.6 when the modulation period , @xmath3 , is chosen to induce qpm at the peak of the drive laser pulse ( @xmath139=0 fs ) .
the black curves show the growth of @xmath145 generated at the drive laser peak , while the other curves show the growth of @xmath145 at an earlier and later time in the pulse ( @xmath148=-5.7 fs ( red ) and @xmath149=5.7 fs ( blue ) ) .
the propagation coordinate is expressed relative to the modulation period .
the pink dashed line represents @xmath150 generated at @xmath139 in a homogeneous medium , without density modulation ( @xmath97=0).,title="fig:",scaledwidth=62.0% ] , throughout the density modulated medium for ( a ) @xmath97=1 , ( b ) @xmath97=0.8 and ( c ) @xmath97=0.6 when the modulation period , @xmath3 , is chosen to induce qpm at the peak of the drive laser pulse ( @xmath139=0 fs ) .
the black curves show the growth of @xmath145 generated at the drive laser peak , while the other curves show the growth of @xmath145 at an earlier and later time in the pulse ( @xmath148=-5.7 fs ( red ) and @xmath149=5.7 fs ( blue ) ) .
the propagation coordinate is expressed relative to the modulation period .
the pink dashed line represents @xmath150 generated at @xmath139 in a homogeneous medium , without density modulation ( @xmath97=0).,title="fig:",scaledwidth=62.0% ] from fig .
[ fig : hh39 output at different time ] it can be clearly seen that , in the absence of density modulation , the harmonic intensity generated at @xmath1510 fs does not grow vs. the interaction length but only oscillates periodically between 0 and some smaller value .
this oscillation , indicating periodic generation and back conversion , is caused by a rather short coherence length ( in this case about 270 @xmath152 ) .
however , when the medium is periodically modulated , the harmonic intensity of the @xmath77 harmonic generated at @xmath139=0 fs ( black curves ) shows a clear growth of the intensity with the medium length with all shown examples of the modulation depth .
specifically , the intensity generated at @xmath139 increases in synchronism with the increasing number of modulation periods .
such increase of the intensity implies that qpm is successfully achieved due to a correctly chosen modulation period . however , the harmonic intensity generated at the other times of the drive pulse ( red and blue curves ) oscillates between 0 and very low values along the medium length because for these generation times the modulation period is incorrectly chosen . among the black curves in fig .
[ fig : hh39 output at different time ] ( @xmath153 ) , the highest harmonic intensity is found in fig .
[ fig : hh39 output at different time ] ( a ) , where a full modulation is applied ( @xmath97=1 ) .
this can be explained as follows : due to the absence of gas in the low - density regions , a full suppression of the out - of - phase generation is achieved . for other modulation depths ,
the @xmath77 harmonic generated from the remaining gas in the low - density region will still partially interfere destructively with the harmonics generated in the high - density region .
moreover , it can be noted that the maximum @xmath145 , generated earlier in the pulse ( at @xmath133 , red curves ) is always higher than that generated with the same intensity later in the pulse ( at @xmath135 , blue curves ) , although the harmonic dipole amplitude , @xmath154 , is the same at both times .
this is due to a larger number of neutral atoms contributing to the hhg in the leading edge of the pulse than that in the trailing edge , because the ionization fraction reduces the neutral atomic density .
the results above show that choosing a particular modulation period only maximizes the hh output for a certain temporal fraction of the drive laser pulse , but , that also at other time intervals there is output , particularly in the leading edge . in an experiment , however , one is usually observing the hh pulse energy , while the instantaneous hh intensity is rather difficult to monitor @xcite . to optimize the pulse energy , @xmath155
, we integrate the instantaneous harmonic intensity , @xmath156 , over the entire drive laser pulse and vary the modulation period .
since , from the previous calculations , the strongest growth in the instantaneous harmonic intensity is found in a fully density modulated medium ( @xmath97=1 ) , we restrict ourselves to this example . in previous experiments ,
generally a modulation period has been selected that is equal to the sum of the coherence lengths calculated for high - density and low - density regions at the time that the drive laser pulse reaches the peak intensity .
the motivation for such choice was that at that time the highest harmonic dipole amplitude , @xmath154 , can be achieved @xcite .
however , this choice appears questionable because then the growth of the harmonic intensity is placed in the time of the highest ionization rate ( quickest change of plasma dispersion ) vs. time which yields the shortest possible time interval of quasi - phase matching . on the other hand ,
if qpm is set to occur over a longer time interval ( when the ionization rate is not maximum ) , one has to take into account a smaller instantaneous intensity . to determine the optimal modulation period for the highest on - axis harmonic pulse energy
, we again apply the same loose focusing configuration for the experiment described in sec .
[ sec : hhg in a homogeneous density medium ] .
we calculate the harmonic pulse energy for the @xmath77 harmonic by selecting different modulation periods that provide qpm at various different time intervals of sufficiently high intensity ( green area in fig . [
fig : different coherence lengths for 39hh ] ) . for more clarity
, we have replotted the green area on a finer time scale in fig .
[ fig : total hh39 output ] ( a ) . .
the purple horizontal line is the coherence length in the low - density regions , while the black curve is the coherence length in the high - density regions .
the modulation depth is set to @xmath97=1 . for the calculation of the hh output pulse energy , seven different modulation periods
are chosen corresponding to providing qpm at seven selected instances in the drive laser pulse ( from @xmath133 to @xmath135 , dotted lines ) .
( b ) on - axis hh pulse energy for the @xmath77 harmonic versus the length of the medium obtained with the seven different modulation periods .
the highest output is achieved with @xmath3 chosen to provide transient qpm in the rising edge of the pulse ( @xmath157 , thick blue line ) .
the pink dashed line of the lowest output is obtained without density modulation.,title="fig:",scaledwidth=80.0% ] .
the purple horizontal line is the coherence length in the low - density regions , while the black curve is the coherence length in the high - density regions .
the modulation depth is set to @xmath97=1 . for the calculation of the hh output pulse energy , seven different modulation periods
are chosen corresponding to providing qpm at seven selected instances in the drive laser pulse ( from @xmath133 to @xmath135 , dotted lines ) .
( b ) on - axis hh pulse energy for the @xmath77 harmonic versus the length of the medium obtained with the seven different modulation periods .
the highest output is achieved with @xmath3 chosen to provide transient qpm in the rising edge of the pulse ( @xmath157 , thick blue line ) .
the pink dashed line of the lowest output is obtained without density modulation.,title="fig:",scaledwidth=80.0% ] it can be seen that between @xmath133 and @xmath148 , the coherence length in the low - density regions ( @xmath122 ) remains independent of time due to the absence of gas that can be ionized , while the coherence length in the high - density regions ( @xmath121 ) decreases due to the increasing ionization fraction . here
, we choose various different modulation periods that provide qpm at seven different instances in the drive pulse : at five instances in the leading edge , denoted as @xmath158 , @xmath159 , @xmath160 , @xmath161 and @xmath162 , respectively , and at another two instances in the trailing edge , denoted as @xmath163 and @xmath164 .
the calculated lengths for the high - density and low - density regions as well as the corresponding modulation periods are listed in tab .
[ tb : modulation period ] .
.list of coherence lengths in the high - density and low - density regions and the modulation periods ( @xmath3 ) chosen to provide qpm at seven different instances within the drive laser pulse when the 39@xmath0 harmonic is generated . [
cols="^,^,^,^",options="header " , ] [ tb : modulation period ] in fig .
[ fig : total hh39 output ] ( b ) , we show the calculated on - axis harmonic pulse energy , @xmath126 , versus the propagation coordinate through the medium for the listed modulation periods . for comparison with the case where qpm is absent , we calculated the hh pulse energy from a homogeneous medium as well .
it can be seen that the output pulse energy is very sensitive to the chosen modulation period . for a homogeneous medium ( without density modulation , pink dashed curve )
, the output oscillates along the medium length and only reaches a small value . in the density
modulated medium , the output grows stepwise along the propagation direction with each modulation period , which indicates successful qpm in spite of its transient nature .
the most important finding from fig .
[ fig : total hh39 output ] ( b ) is that , due to this transient nature , the highest on - axis harmonic pulse energy is obtained by applying a modulation period that provides transient qpm in the _ leading edge _ rather than at the peak intensity of the drive laser pulse , here with @xmath160 . in this example with the @xmath77 harmonic in argon , the optimum qpm period is about 50% bigger than that for the peak of the laser pulse and provides 25% more output pulse energy . in summary ,
the optimum modulation period for the highest on - axis hh pulse energy , @xmath126 , can only be approximately be determined from the remaining wave - vector mismatch at the peak of the drive laser pulse .
although at this time the highest instantaneous output intensity is reached and even higher output is expected with larger qpm period .
we assign this to the fact that before the peak the time interval of successful qpm longer and overcompensates a some what lower intensity in the rising edge .
we have developed a one - dimensional , time - dependent model for quasi - phase matching ( qpm ) of high - order harmonic generation .
the model is used to calculate the optimum quasi - phase - matching period , taking into account the highly dynamic change of index during harmonic generation .
the model is presented using an example where standard phase matching can not be achieved , here the @xmath77 harmonic to be generated in argon . using a spatially periodic density modulation ( periodic structure )
, the quasi - phase matching occurs transiently at some instance during the drive laser pulse .
this is due to strong ionization driven by the need for high intensities and therefore leading to an ultrafast dependence of the coherence length .
the choice of the modulation period determines the specific moment and duration of quasi - phase matching .
we show that simply choosing the modulation period to provide phase matching at the peak intensity of the pulse , when the highest amplitude of the nonlinearly induced high - order harmonic dipole amplitude is reached , does not provide the highest harmonic output pulse energy . to find the optimum modulation period , it is essential to analyze the time dependence of the wave - vector mismatch and the harmonic dipole amplitude along the entire pulse duration . according to our model , the optimal modulation is obtained with longer periods than usually assumed , such that transient quasi - phase matching is provided in the leading edge of the pulse .
although we have presented results from the model using a specific choice of parameters for an idealized situation , we note that the model can also be applied to more complicated modulation profiles such as would be present e.g. , in a density modulated cluster jet .
the model may also be applied to more complicated gas distributions such as multi dual - gas jet arrays @xcite , by recalculating the wave - vector mismatch and the induced dipole amplitudes for the different gas species used .
further improvements of this model would include an extension to three dimensions ( including the transverse direction ) together with an improved quantum mechanical calculation of time - dependent single - atom response compared to the power - law approximation used in this paper for determining a more accurate qpm period .
this research was supported by the dutch technology foundation stw , which is part of the netherlands organization for scientific research ( nwo ) , and partly funded by the ministry of economic affairs ( project no . | we present a model for quasi - phase matching ( qpm ) in high - order harmonic generation ( hhg ) . using a one - dimensional description ,
we analyze the time - dependent , ultrafast wave - vector balance to calculate the on - axis harmonic output versus time , from which we obtain the output pulse energy .
considering , as an example , periodically patterned argon gas , as may be provided with a grid in a cluster jet , we calculate the harmonic output during different time intervals within the drive laser pulse duration .
we find that identifying a suitable single spatial period is not straightforward due to the complex and ultrafast plasma dynamics that underlies hhg at increased intensities .
the maximum on - axis harmonic pulse energy is obtained when choosing the qpm period to phase match hhg at the leading edge of the drive laser pulse . 10 s. witte , v. t. tenner , d. w. e. noom , and k. s. e. eikema , `` lensless diffractive imaging with ultra - broadband table - top sources : from infrared to extreme - ultraviolet wavelengths , '' light .
sci .
appl . * 3 * , e163 ( 2014 ) .
y. mairesse , j. higuet , n. dudovich , d. shafir , b. fabre , e. mevel , e. constant , s. patchkovskii , z. walters , m. y. ivanov , and o. smirnova , `` high harmonic spectroscopy of multichannel dynamics in strong - field ionization , '' phys .
rev . lett . * 104 * , 213601 ( 2010 ) .
s. baker , j. s. robinson , c. a. haworth , h. teng , r. a. smith , c. c. chirila , m. lein , j. w. g. tisch , and j. p. marangos , `` probing proton dynamics in molecules on an attosecond time scale , '' science * 312 * , 424 ( 2006 ) .
t. popmintchev , m. c. chen , a. bahabad , m. gerrity , p. sidorenko , o. cohen , i. p. christov , m. m. murnane , and h. c. kapteyn , `` phase matching of high harmonic generation in the soft and hard x - ray regions of the spectrum , '' proc .
natl .
acad .
sci .
u. s. a. * 106 * , 10516 ( 2009 ) .
x. zhang , a. l. lytle , t. popmintchev , x. zhou , h. c. kapteyn , m. m. murnane , and o. cohen , `` quasi - phase - matching and quantum - path control of high - harmonic generation using counterpropagating light , '' nat .
phys . * 3
* , 270 ( 2007 ) .
e. a. gibson , a. paul , n. wagner , r. tobey , d. gaudiosi , s. backus , i. p. christov , a. aquila , e. m. gullikson , d. t. attwood , m. m. murnane , and h. c. kapteyn , `` coherent soft x - ray generation in the water window with quasi - phase matching , '' science * 302 * , 95 ( 2003 ) .
m. zepf , b. dromey , m. landreman , p. foster , and s. m. hooker , `` bright quasi - phase - matched soft - x - ray harmonic radiation from argon ions , '' phys .
rev .
lett . * 99 * , 143901 ( 2007 ) .
l. z. liu , k. okeeffe , and s. m. hooker , `` quasi - phase - matching of high - order - harmonic generation using polarization beating in optical waveguides , '' phys .
rev .
a * 85 * , 053823 ( 2012 ) .
l. z. liu , k. okeeffe , and s. m. hooker , `` quasi - phase - matching of high - order - harmonic generation using multimode polarization beating , '' phys .
rev .
a * 87 * , 023810 ( 2013 ) .
a. hage , b. landgraf , m. taylor , m. wunsche , t. gangolf , h. hoppner , m. j. prandolini , r. riedel , m. schulz , f. tavella , a. willner , m. yeung , g. g. paulus , c. spielmann , b. dromey , and m. zepf , `` new design of a multi - jet target for quasi phase matching , '' rev .
sci .
instrum . * 85 * , 103105 ( 2014 ) .
a. willner , f. tavella , m. yeung , t. dzelzainis , c. kamperidis , m. bakarezos , d. adams , m. schulz , r. riedel , m. c. hoffmann , w. hu , j. rossbach , m. drescher , n. a. papadogiannis , m. tatarakis , b. dromey , and m. zepf , `` coherent control of high harmonic generation via dual - gas multijet arrays , '' phys .
rev .
lett . * 107 * , 175002 ( 2011 ) .
a. willner , f. tavella , m. yeung , t. dzelzainis , c. kamperidis , m. bakarezos , d. adams , r. riedel , m. schulz , m. c. hoffmann , w. hu , j. rossbach , m. drescher , v. s. yakovlev , n. a. papadogiannis , m. tatarakis , b. dromey , and m. zepf , `` efficient control of quantum paths via dual - gas high harmonic generation , '' new j. phys .
* 13 * , 113001 ( 2011 ) .
m. lewenstein , p. balcou , m. y. ivanov , a. lhuillier , and p. b. corkum , `` theory of high - harmonic generation by low - frequency laser fields , '' phys .
rev .
a * 49 * , 2117 ( 1994 ) . e. constant , d. garzella , p. breger , e. mvel , c. dorrer , c. le blanc , f. salin , and p. agostini , `` optimizing high harmonic generation in absorbing gases : model and experiment , '' phys
. rev .
lett . * 82 * , 1668 ( 1999 ) .
s. kazamias , d. douillet , c. valentin , f. weihe , f. auge , t. lefrou , g. grillon , s. sebban , and p. balcou , `` observation of high - contrast coherence fringes in high - order harmonic generation , '' phys .
rev .
a * 68 * , 033819 ( 2003 ) .
w. f. chan , g. cooper , x. guo , g. r. burton , and c. e. brion , `` absolute optical oscillator strengths for the electronic excitation of atoms at high resolution .
iii .
the photoabsorption of argon , krypton , and xenon , '' phys .
rev .
a * 46 * , 149 ( 1992 ) .
s. kazamias , s. daboussi , o. guilbaud , k. cassou , d. ros , b. cros , and g. maynard , `` pressure - induced phase matching in high - order harmonic generation , '' phys .
rev .
a * 83 * , 063405 ( 2011 ) .
m. v. ammosov , n. b. delone , and v. p. krainov , `` tunnel ionization of complex atoms and of atomic ions in an alternating electromagnetic field , '' soviet physics - jetp * 64 * , 1191 ( 1986 ) .
h. dachraoui , t. auguste , a. helmstedt , p. bartz , m. michelswirth , n. mueller , w. pfeiffer , p. salieres , and u. heinzmann , `` interplay between absorption , dispersion and refraction in high - order harmonic generation , '' j. phys .
b * 42 * ( 2009 ) .
t. auguste , b. carre , and p. salieres , `` quasi - phase - matching of high - order harmonics using a modulated atomic density , '' phys .
rev .
a * 76 * , 011802(r ) ( 2007 ) . c. g. durfee iii , a. r. rundquist , s. backus , c. herne , m. m. murnane , and h. c. kapteyn , `` phase matching of high - order harmonics in hollow waveguides , '' phys .
rev . lett . * 83 * , 2187 ( 1999 ) .
j. f. hergott , m. kovacev , h. merdji , c. hubert , y. mairesse , e. jean , p. breger , p. agostini , b. carre , and p. salieres , `` extreme - ultraviolet high - order harmonic pulses in the microjoule range , '' phys .
rev .
a * 66 * , 021801(r ) ( 2002 ) .
s. kazamias , d. douillet , f. weihe , c. valentin , a. rousse , s. sebban , g. grillon , f. auge , d. hulin , and p. balcou , `` global optimization of high harmonic generation , '' phys .
rev .
lett .
* 90 * , 193901 ( 2003 ) .
k. midorikawa , y. nabekawa , and a. suda , `` xuv multiphoton processes with intense high - order harmonics , '' prog .
quant .
electron * 32 * , 43 ( 2008 ) .
p. rudawski , c. m. heyl , f. brizuela , j. schwenke , a. persson , e. mansten , r. rakowski , l. rading , f. campi , b. kim , p. johnsson , and a. lhuillier , `` a high - flux high - order harmonic source , '' rev .
. instrum . * 84 * , 073103 ( 2013 ) .
y. tamaki , j. itatani , m. obara , and k. midorikawa , `` optimization of conversion efficiency and spatial quality of high - order harmonic generation , '' phys .
rev .
a * 62 * , 063802 ( 2000 ) .
p. m. paul , e. s. toma , p. breger , g. mullot , f. auge , p. balcou , h. g. muller , and p. agostini , `` observation of a train of attosecond pulses from high harmonic generation , '' science * 292 * , 1689 ( 2001 ) .
l. z. liu , k. okeeffe , and s. m. hooker , `` optical rotation quasi - phase - matching for circularly polarized high harmonic generation , '' opt .
lett . * 37 * , 2415 ( 2012 ) . |
from the @xmath3-wave symmetry of the order parameter of cuprate superconductors , one expects an enhanced sensitivity of @xmath0-axis transport in the superconducting state to disorder , due to the enhancement of the quasiparticle density of states along the gap node directions , and due to impurity assisted hopping @xcite .
both mechanisms lead to as yet unobserved @xmath4-dependences of the @xmath0-axis superfluid density @xmath5 at low @xmath6 , with coefficients that strongly depend on the scattering rate @xmath7 . on the other hand , underdoped cuprates are sufficiently disordered for ( quantum ) fluctuations of the order parameter phase to play a prominent role @xcite , leading to a @xmath6-linear behavior of @xmath8 . here
, we report on the effect of controlled point disorder on interlayer tunneling of cooper pairs in underdoped bi@xmath1sr@xmath1cacu@xmath1o@xmath2 .
single - crystalline rods of underdoped bi@xmath1sr@xmath1cacu@xmath1o@xmath2 were grown using the travelling solvent floating zone technique under 25 mbar oxygen partial pressure @xcite .
the samples used for the study , with @xmath9 k , were cleaved from the same crystalline piece .
the crystals were then irradiated with 2.5 mev electrons using the van der graaf accelerator of the laboratoire des solides irradis .
the irradiation produces homogeneously distributed frenkel pairs , which have previously been identified as strong scattering centers @xcite .
the josephson plasma resonance ( jpr ) frequency @xmath10 of the crystals was then measured using the cavity perturbation technique , exploiting @xmath11 harmonic modes to access different frequencies @xcite , and the bolometric technique using a waveguide in the @xmath12 travelling wave mode @xcite .
the latter technique allows for swept - frequency measurements , necessary to elucidate the weak @xmath13 dependence at low @xmath6 @xcite .
note that @xmath14 is proportional to the @xmath0-axis critical current @xmath15 and to the @xmath0-axis superfluid density : @xmath16 , with @xmath17 the low - frequency dielectric constant and @xmath18 is the @xmath0-axis penetration depth .
sharp jpr resonant peaks were measured for all samples under study .
both @xmath19 and @xmath20 decrease with irradiation dose .
the sensitivity of @xmath10 to even weak additional disorder contradicts the model of coherent interlayer cooper pair tunneling @xcite . within the framework of a @xmath3-wave bcs model
, the measured proportionality between @xmath14 and @xmath19 can be understood as resulting either from ( i ) the dependence of the interlayer josephson current @xmath21 on the gap magnitude @xmath22 and the quasiparticle conductivity @xmath23 @xcite , or ( ii ) from the decrease of the _ in - plane superfluid density due to strong phase fluctuations .
_ the origin of the temperature and disorder - dependence of @xmath10 can be pinpointed using fig .
[ fig : jpr - vs - t](b ) . normalising all results to @xmath20 and plotting
these versus reduced temperature , reveals a common @xmath6-dependence independent of disorder .
this contradicts the prediction of ref .
@xcite that @xmath25 should follow different powers of @xmath6 depending on the ratio of @xmath6 , @xmath22 , and @xmath7 . however , it agrees with a dominant role of quantum phase fluctuations .
then , @xmath26 where @xmath27 , @xmath28 with @xmath29 the in - plane penetration depth , @xmath30 is the cuo@xmath1 layer spacing , and @xmath31 is the cuo@xmath1 plane sheet conductivity @xcite .
the temperature dependence of the jpr frequency arises from the temperature of the in - plane phase stiffness @xmath32 .
figure 2 shows that eq .
( [ eq : paramekanti ] ) describes the results very satisfactorily . summarizing ,
the temperature- and disorder dependence of the @xmath0-axis cooper pair tunnel current in underdoped bi@xmath1sr@xmath1cacu@xmath1o@xmath2 is well described assuming a strong effect of quantum phase fluctuations . a @xmath3-wave model without fluctuations
can not account for the @xmath6-dependence of @xmath0-axis coupling . -irradiated bi@xmath1sr@xmath1cacu@xmath1o@xmath2 crystals , plotted as @xmath33 in order to bring out the low temperature @xmath6-dependence .
the line is a fit to eq .
[ eq : paramekanti ] @xcite with parameters as indicated.,title="fig:",width=302 ] [ fig : normalize ] t. xiang and j.m .
wheatley , phys .
lett . * 77 * , 4632 ( 1996 ) .
ioffe and a. millis , science * 285 * , 1241 ( 1999 ) . ming li _
b * 66 * , 024502 ( 2002 ) .
f. rullier - albenque _ et al .
_ , europhys . lett . * 50 * , 81 ( 2000 ) .
et al . _ , phys .
lett . * 83 * , 3928 ( 1999 ) .
s. colson _
. lett . * 90 * , 137002 ( 2003 ) .
et al . _ ,
* 82 * , 5345 ( 1999 ) .
y. sun and k. maki , phys .
b * 51 * , 6059 ( 1995 ) .
a. paramekanti , phys .
b * 65 , 104521 ( 2002 ) . * | the josephson plasma resonance is used to study the @xmath0-axis supercurrent in the superconducting state of underdoped bi@xmath1sr@xmath1cacu@xmath1o@xmath2 with varying degrees of controlled point - like disorder , introduced by high - energy electron irradiation . as disorder
is increased , the josephson plasma frequency decreases proportionally to the critical temperature . the temperature dependence of the plasma frequency does not depend on the irradiation dose , and is in quantitative agreement with a model for quantum fluctuations of the superconducting phase in the cuo@xmath1 layers .
disorder , interlayer coupling , josephson plasma resonance , quantum fluctuations 74.40.+k , 74.50.+r,74.62.-c , 74.62.dh |
one of the apparently simple problems for which there is no consensus yet on which are the correct continuum equations of motion is the flux of grains on a granular surface ( see for example ref .
this flux can be driven by a fluid , like air , by gravity or by an initial impact .
it modifies the granular surface itself either by deposition or by erosion .
we will consider in this paper a gravity driven flow with variable initial energy and consider deposition only .
in addition we want to be in the limit of small flux .
one realization is the slow buildup of a pile from a point source .
a discrete model was proposed in ref .
@xcite in which particles jump down stairs and deposit if their energy is below a certain material dependent threshold . from this model
a set of continuum equations was derived in ref.@xcite and applied to the problem of impact craters .
this continuum description , however , made an assumption on the time dependence of the slope that only contained the threshold in an indirect way , namely by inserting the known steady state result .
also all the jumps were of length unity even if their kinetic energy was different . in this paper
we want to deal with these defects of the equations in ref .
@xcite by inserting the two mentioned effects independently into the equations and investigating the consequences analytically and numerically .
we will introduce an extension to the model proposed in ref . @xcite .
there , particles of mass @xmath0 are added at the top of a pile , and they move , under the action of gravity @xmath1 , until their kinetic energy @xmath2 falls below a certain threshold @xmath3 ( related to the friction between grains ) where they then stop . at each collision the kinetic energy decreases by a factor proportional to the restitution coefficient of the material @xmath4 . mathematically , it reads : @xmath5 it was assumed in ref .
@xcite that the traveled distance after each particle jump is constant . here
, trying to model a more realistic situation , we assume that this distance is proportional to the kinetic energy of the particle .
so , we propose to modify equation ( [ eq : enera_cla ] ) to : @xmath6 where @xmath7 is a proportionality coefficient ( in units of @xmath8 ) .
then , writing eq .
( [ eq : enera_mod ] ) in differential form we obtain : @xmath9 where @xmath10 . to this equation one
adds in ref.@xcite an evolution equation for the slope of the form : @xmath11 where @xmath12 controls the rate of relaxation of the slope .
boundary conditions were imposed trying to reproduce real experimental situations of heap formation : @xmath13 a value which is proportional to the height from which the grains fall on the pile , and @xmath14 . the equations ( [ eq : enera_diff ] ) and ( [ eq : anga_diff ] ) , constitute a pair of linear differential equations with constant coefficients . from them ,
the angle of a pile can be easily calculated : @xmath15 and then , integrating equation ( [ eq : anga_sol ] ) the shape of the pile is : @xmath16 this solution has two remarkable properties as shown in figure 1 .
first of all the oscillatory behaviour superimposed to the usual linear decay of the pile .
these oscillations arise from the existence of a length scale @xmath12 which controles the rate of the relaxation of the slope while independently the particles jump a distance ( @xmath17 ) determined only by their kinetic energy .
this phenomenon is similar to the formation of ripples due to saltation @xcite , and it could be the first theoretical prediction of `` gravity '' induced ripples
. another important feature of eq .
( 6 ) is that it can reproduce a crater formation on the top of pile .
in fact , in the limit @xmath18 , equation ( [ eq : alta_sol ] ) transforms into : @xmath19 showing the existence of a parabolic - like crater whose depth linearly increases with the initial kinetic energy of the particles .
in this model , we add to the equation for the energies from ref .
@xcite @xmath20 a new equation for the angles based on the assumption that the equilibrium local slope of the pile is constant , i.e. @xmath21 where @xmath22 is the slope of the pile .
the evolution of the slope proceeds in the following way : if a particle arrives at @xmath23 , it has two possibilities ( as described for a similar model in ref .
@xcite ) , if @xmath24 , it moves to @xmath25 , then : @xmath26 or , it stops when @xmath27 and @xmath28 equations ( [ eq : angb_phi_d ] ) and ( [ eq : angb_phi_c ] ) may be written in the following differential form : @xmath29 we next solve the coupled differential equations ( [ eq : enerb_diff ] ) and ( [ eq : angb_phi_diff ] ) numerically .
it is more convenient to write both equations in the same form .
so dividing eq .
( [ eq : enerb_diff ] ) by @xmath30 , and noting that @xmath31 , we obtain the following equations : @xmath32 @xmath33 where @xmath34 and @xmath35 .
then , once @xmath36 is known the height of the pile can be calculated through integration of @xmath22 and obtained numerically . to numerically solve the equations ( [ eq : enerb_ns ] ) and ( [ eq : phib_ns ] ) the function @xmath37 was replaced by the continous function @xmath38 .
profiles calculated using different values of @xmath39 are shown in figure 2 . from the figure
we conclude that if @xmath40 all the numerical results are equivalent . to be on the secure side we used in the rest of the paper @xmath41 .
figure 3 represents typical sandpile profiles obtained for different values of the parameters .
the bottom curve ( where the crater is better defined ) represents a system where the grains have higher initial energy . in figure 4
is plotted the equilibrium angle of the pile @xmath42 as a function of @xmath43 for different values of @xmath44 . as previously calculated in ref .
@xcite , the angle of the pile is equal to @xmath43 showing the equivalence between our approach and that proposed in @xcite .
the depth @xmath45 and the width @xmath46 of the craters , should only depend on the three parameters involved in the model @xmath47 and @xmath3 .
in fact , in figures 5 and 6 , @xmath45 and @xmath46 obtained using different sets of parameters @xmath48 collapse on the same curve following the scaling relations : @xmath49 and @xmath50 where @xmath51 , @xmath52 , @xmath53 and @xmath54 and @xmath55 are scaling functions with the following properties @xmath56 if @xmath57 and @xmath58 if @xmath59 , idem for @xmath1 .
this indicates that around @xmath60 and @xmath61 we have critical behaviour with simple integer exponents .
physically equations ( [ eq : scaling_deltah ] ) and ( [ eq : scaling_deltax ] ) imply that in granular systems with negligible restitution coefficent ( @xmath62 ) the depth of the crater increases linearly with @xmath63 and with @xmath64 , in good agreement with eq.[eq : alta_sol_lim ] while the width of the craters increases parabolically ( @xmath51 ) .
therefore systems with small @xmath44 , i.e. ( @xmath65 ) or @xmath66 do not develop a crater .
we have presented two sets of equations of motion for the limit of dilute granular surface flow .
the first set included a jump length proportional to the energy of the grains .
this gave ripple formation on the critical slope , a phenomenon which has not yet been observed experimentally .
the second set was a cleaner way to include the deposition threshold than done in ref .
the results confirm that the previously used equation ( introduced in ref .
@xcite ) gave qualitatively the right answer for the shape of the pile which means that the picture of a characteristic length for relaxation to the equilibrium angle is correct .
we thank sfb 381 for financial support . 99 j. -p .
bouchaud , m. e. cates , j. r. prakash , and s. f. edwards , phys .
74 , 1982 ( 1995 ) physics of dry granular media - nato asi series e 350 , edited by h. j. herrmann , j. -p .
hovi , and s. luding ( kluwer academic publishers , dordrecht , 1998 ) t. boutreaux and p.g .
de gennes , j. phys .
i , france * 6 * 1285 ( 1996 ) j.j .
alonso , j .-
hovi and h. herrmann , phys .
rev e * 57 * , 672 ( 1998 ) y. grasselli , h. herrmann , g. oron and s. zapperi , preprint ( 1998 ) h. nishimori and n. ouchi , phys .
lett * 71 * , 197 ( 1993 )
graphical representation of equation 6 . we chose @xmath67,@xmath68 and @xmath69 .
* figure 5 * data collapse of the depth of the craters using eq : ( [ eq : scaling_deltah]).parameters : ( @xmath77 ) = @xmath78 ( 0.01 , 0.15 , 0.05 ) ; @xmath79 ( 0.01 , 0.25 , 0.05 ) ; @xmath80 ( 0.01 , 0.25 , 0.10 ) ; @xmath81 ( 0.01 , 0.25 , 0.15 ) ; @xmath82 ( 0.03 , 0.20 , 0.10 ) ; @xmath83 ( 0.03 , 0.25 , 0.15 ) ; @xmath79 ( 0.05 , 0.30 , 0.10 ) . * figure 6 * data collapse of the depth of the crater using eq : ( [ eq : scaling_deltax ] ) . parameters : ( @xmath77 ) = @xmath78 ( 0.01 , 0.15 , 0.05 ) ; @xmath79 ( 0.01 , 0.25 , 0.05 ) ; @xmath80 ( 0.01 , 0.25 , 0.10 ) ; @xmath81 ( 0.01 , 0.25 , 0.15 ) ; @xmath82 ( 0.03 , 0.20 , 0.10 ) ; @xmath83 ( 0.03 , 0.25 , 0.15 ) ; @xmath79 ( 0.05 , 0.30 , 0.10 ) . | we introduce two sets of continuum equations to describe granular flow on a free surface and study their properties .
the equations derived from a microscopic picture that includes jumps and a mobility threshold , account for ripple and crater formation .
_ keywords : surface flow . granular materials _ |
since the discovery of the milky way s thick disk in 1983 @xcite , there has been a growing body of evidence that the thick disk is a structurally , chemically , and kinematically distinct component of the galaxy .
structurally , the milky way s thick disk has a significantly larger scale height than the thin disk ( for reviews of the thick disk scale height see @xcite and references therein ) , and a possibly somewhat longer scale length @xcite .
chemically , thick disk stars are more metal - poor and older than stars in the thin disk @xcite .
they are also significantly enhanced in @xmath2-elements , compared to thin disk stars of comparable iron abundance @xcite .
kinematically , thick disk stars have both a larger velocity dispersion and slower net rotation than stars in the thin disk @xcite .
all of these facts lead to the conclusion that the thick disk is a relic of the young galactic disk . as such
, it provides an excellent probe of models of disk galaxy formation ( see recent reviews by @xcite ) . originally detected in s0 galaxies @xcite ,
thick disk components have since been detected in many galaxies , including s0 s @xcite , sb s @xcite , and later type galaxies @xcite . across all hubble types , photometric decompositions
consistently find that thick disk scale heights are 2 - 6 times larger than thin disk scale heights and that the thick disks scale lengths are slightly larger ( @xcite , but see also @xcite ) .
thick and thin disks are likely to have distinct formation mechanisms , given the systematic differences between their properties .
the structure , dynamics , and chemical abundance of the thin disk strongly suggest that the majority of its stars formed gradually from a rotating disk of high angular momentum gas @xcite .
in contrast , the formation of the thick disk is still poorly understood and is likely to be more complex .
the large scale height of the thick disk suggests that its stars were either ( 1 ) vertically `` heated '' from a previously thinner disk @xcite , ( 2 ) formed from gas with a large scale height @xcite , or ( 3 ) directly deposited at large scale heights during the accretion of smaller satellite galaxies @xcite .
most models of thick disk formation fall into one of these three cases . in the first
, the thick disk stars form initially in a thin disk . in the second
, they form within the thick disk itself . in the third , they form outside of the galaxy entirely .
all three of the above scenarios are compatible with the structure of the milky way s thick disk .
however , they all imply very different behavior for the early evolution of disk galaxies .
the first case would suggest that disks form largely through smooth gas accretion that is occasionally punctuated by minor merging events .
the second case would suggest either that disks form primarily through smooth monolithic collapse , with thick disk stars precipitating out of the collapsing gas cloud @xcite , or that the thick disk forms from a disk of gas that has been energetically heated by star formation @xcite .
the third case would suggest that disks can form entirely from merging sub - units , in spite of their highly ordered present day structure .
the degeneracies between these models must be broken before the properties of the thick disk can be used as a constraint on theories of galaxy formation .
past attempts to distinguish among formation models have often relied on the chemical abundances of thick disk stars .
the low metallicities of thick disk stars suggest that they formed early in the evolution of the galaxy .
their @xmath2-enhancement suggests that the time scale for their formation was sufficiently rapid to suppress iron enrichment by type ia supernovae .
unfortunately , the resulting implication that the thick disk formed early and rapidly can be accommodated by all of the possible formation scenarios .
@xcite suggest that metallicity measurements of more than @xmath3 stars would be required to definitively constrain the origin of the milky way s thick disk .
metallicities of thick disk stars therefore remain a somewhat weak discriminant among competing models .
in contrast to metallicities , stellar kinematics show more promise for constraining the origin of the thick disk . if the thick disk results from dynamical heating of a young previously thin disk , both components should have similar rotation curves , assuming similar angular momenta for the gas that falls in before and after the heating event .
if the thick disk forms from gas as it collapses into the galaxy , then angular momentum conservation requires that the rotation of the thick disk must always lag the rotation of the thin disk , by an amount that depends on the relative scale lengths of the thick and thin disks .
finally , if the thick disk is assembled from stars originally formed in merging satellites , then the relative kinematics of the thick and thin disks are likely to be highly variable .
the rotation of the milky way thick disk is marginally consistent with all of these scenarios @xcite .
measurements have consistently shown the galactic thick disk to be co - rotating with the thin disk , but with modest lag .
the exact value of the lag varies by a factor of two depending on the study , ranging from @xmath4 km s@xmath5 @xcite up to @xmath6 km s@xmath5 @xcite .
the smaller rotational lags can be easily produced in most thick disk formation models , but the larger lags observed recently are more difficult to explain without satellite accretion .
however , rather than ruling out out alternative models , the larger lags are often interpreted as contamination from a single disrupted satellite .
although data within the galaxy have not been sufficient to definitively constrain the origin of the thick disk , the milky way is only one single case .
instead , measurements of thick disk kinematics in _ several _ galaxies are necessary to exploit the full power of kinematic discriminants among formation models .
scenarios involving mergers will naturally lead to much more kinematic variation among the population of thick disks than if thick disks formed during monolithic collapse . even among merging models , we would expect the relative kinematics of the thick and thin disks to vary more if the thick disk stars are directly accreted than if the thick disk was built from vertically heated thin disk stars . by measuring the relative kinematics of thick and thin disks in a large sample of galaxies
, one can therefore discriminate among thick disk formation models far more effectively than by studying the milky way alone .
in this paper we present the first results of our on - going study of the kinematic properties of thick disks in a large sample of late - type edge - on galaxies @xcite .
we have measured of the kinematics of thick and thin disk stars in two edge - on late - type disk galaxies .
we first present the sample selection and observations in [ observationsec ] .
we then describe the data reduction and extraction of the rotation curves in [ reductionsec ] . in [ rcsec ]
we decouple the observed rotation curves to derive the rotation of the thick and thin disks . in [ losvdsec ]
we measure radially resolved velocity dispersion profiles above and below the plane .
we then discuss the implication of our initial results for models of disk galaxy formation in [ discussionsec ] and summarize our conclusions in [ conclusionsec ] .
we have carried out long - slit spectroscopic observations of two galaxies drawn from the @xcite sample of edge - on late - type galaxies .
the original sample of 49 galaxies was selected from the flat galaxy catalog @xcite and imaged in @xmath7 , @xmath8 , and @xmath9 @xcite .
@xcite used this imaging to demonstrate the ubiquity of thick disks around late - type galaxies .
we have since used two - dimensional decompositions of the galaxy images to measure structural parameters for the thick and thin disks .
this decomposition will be presented in detail in a future paper @xcite , and we now give a brief overview of our decompositions . when fitting models to our galaxies , we adopt the method of @xcite and use a levenberg - marquardt @xmath10 minimization routine .
each disk is modeled as having a surface brightness @xmath11 where @xmath12 is a modified bessel function of the first order , @xmath13 is the edge - on central surface brightness , @xmath8 is the projected radius along the major axis , and @xmath14 is the exponential scale height .
we weight individual pixels by the inverse _ model_.
this method allows one to fit regions of very low surface brightness without being overly skewed by brighter midplane features . comparing the formal reduced @xmath10 values of the single disk and two - disk models using an @xmath15-test @xcite
, we find that the two - disk models are strongly justified .
to derive estimates for our parameter uncertainties , we fit a series of models varying the functional form of the vertical profiles ( e.g. sech(@xmath16 ) , sech@xmath17 ) for both the thick and thin disk as well as comparing models which were fitted with the midplane region masked .
we estimate our uncertainties by using the full range of convergent values in all the models .
our final fitted parameters and uncertainties are listed in table [ galtable ] .
plots of our residuals collapsed along the major axis are shown in figure [ resids ] .
our residuals clearly show that subtracting off a single disk leaves excess flux at high @xmath16 while our preferred models fit well across the full range of latitudes .
this procedure is discussed more extensively in @xcite .
lcccccccccc fgc 227 & 5493 & 6261 & 107 & 21.5@xmath18 & 10.0@xmath19 & 1.0@xmath20 & 22.8@xmath19 & 9.3@xmath21 & 2.0@xmath22 & 0.50 + fgc 1415 & 1522 & 2683 & 87 & 20.9@xmath23 & 14.8@xmath24 & 1.2@xmath23 & 22.2@xmath23 & 20.0@xmath25 & 3.2@xmath23 & 0.44 + to define a sample for measuring the thick disk kinematics , we selected several initial candidates for spectroscopic follow - up .
we chose to focus our initial efforts on galaxies with rotation speeds near @xmath26 .
these galaxies are rotating fast enough that their kinematics can be measured with moderate resolution spectroscopy .
however , they are also low enough mass that they fall below the @xmath27 transition where dust lanes become common @xcite , allowing us to probe their thin disk kinematics and structure more reliably . from the galaxies with @xmath26 , we further restricted our sample to the most nearby , spatially well - resolved galaxies . the resulting sample was then submitted to the noao gemini north observing queue .
the choice of the specific galaxies targeted was then left to the gemini observers , to maximize the chances that one of our targets would be visible at any night during the observing semester . in this paper
we present results for the two galaxies that were selected by the gemini queue
fgc 227 and fgc 1415 .
the properties of both galaxies are listed in table 1 , details of our observations are listed in table [ obstable ] , and images of the galaxies are shown in figure [ slit_pos ] .
lccccccc fgc 227 & 02:00:56.7 & 19:42:26 & 9 & - & - & 20% & 3x20 + fgc 227 & 02:00:56.5 & 19:42:25 & 9 & 3.0 & 1.3 & 50% & 28x20 + fgc 1415 & 12:20:27.7 & 01:28:09 & 146 & - & - & 30% & 3x20 + fgc 1415 &
12:20:28.0 & 01:28:12 & 146 & 5.4 & 1.0 & 70% & 46x20 + we obtained long - slit spectroscopy of the ca ii triplet ( 8498 , 8542 , and 8662 ) to measure stellar kinematics .
we chose two slit locations for each galaxy , oriented with the slit parallel the major axis of the galaxy .
the first slit position was located along the midplane , where the light is dominated by thin disk stars .
the second slit position was located 34.5 thin disk scale heights above the midplane , where @xmath28% of the light came from the thick disk , based on the thick and thin disk decompositions of @xcite .
the locations of the slits are superimposed on the galaxy images shown in figure [ slit_pos ] .
typical surface brightnesses in the midplane and high - latitude slit positions were 21.5@xmath29 and 23@xmath29 , respectively , in the @xmath8-band .
the galaxies were observed with the gemini north gmos spectrograph using an 0.5@xmath30 wide slit and the r400-g5305 grating set with a central wavelength of @xmath31 .
the ccd was binned by 2 in the spatial direction during readout .
the resulting spectra have spectral resolution of 3.4 ( fwhm ) near the ca ii triplet region as measured by cuar arc spectra .
this resolution allows lines to be centroided to @xmath010@xmath32 .
accurate placement of our slit is of great importance because we are combining images taken over several nights .
our offsets from the midplane are relatively small , and an error in placement of order an arcsecond could compromise our results . in our remote observing program ,
we instructed the gemini observing specialists to align the slit along the galaxy midplane , after which , a pre - programmed offset iterator would shift the slit to the appropriate off - plane locations . to place the slit , the observing specialist first took an exposure with no dispersion element to locate the galaxy midplane .
the longslit was then moved into the light path and another exposure was taken .
if the slit did not fall on the galaxy midplane , the telescope was offset to the correct position .
inspecting the returned acquisition images , we find that the observing specialists were able to accurately place the slit along the midplane within @xmath33 arcseconds ( 1 - 2 gmos pixels ) . because our spectra are dithered along the slit with offsets of around 5 - 10 arcseconds , we have checked the overall accuracy of gemini offset commands and found that they are correct to @xmath34 pixel ( 0.15 arcseconds ) .
overall , we find that the slit was always placed within 0.21@xmath30 of our target position which ensures our target regions were always in the 0.5@xmath30 slit .
observations of fgc 227 and fgc 1415 were executed during the spring and fall observing semesters of 2003 , respectively .
the image quality was in the 70th-85th percentile , corresponding to typical seeing of @xmath35 .
because our galaxies are large ( @xmath36 ) , as are the offsets between the two slit observing positions ( @xmath37 ) , poor seeing has little effect on the resulting spectra .
the midplane slit positions were observed with three 20 minute exposures for both galaxies .
the thick disk of fgc 1415 was observed in 46@xmath3820 minute exposures over 9 separate nights , for a total of 15.3 hours .
the thick disk of fgc 227 was observed in 28@xmath3820 minute exposures over 8 separate nights , for a total of 9.3 hours .
the spectra were dithered along the slit in a random pattern between exposures , with typical shifts of 5@xmath30 .
the gmos ccd s were bias corrected using both bias images and fits to 31 columns of overscan .
flat fielding was performed with the facility calibration unit ( fcu ) and the twilight sky .
fcu flats were interleaved with the science exposures approximately every hour to ensure complete removal of any variation in the ccd fringing pattern .
twilight flats were combined and used to correct for uneven illumination of the ccd . the spectra were wavelength calibrated using night sky emission lines .
we combined the sky line data of @xcite with the iraf sky line list to construct an atlas of lines ( or line blends ) that would be bright and centroidable at our instrumental resolution .
this procedure yielded @xmath39 lines for chip 2 ( containing the ca ii triplet ) and @xmath40 lines for chip 3 .
we fit the resulting wavelength solution with a 6th degree chebychev polynomial .
rms errors in wavelength calibration were of order 0.3 , or @xmath41 at the wavelengths of interest .
analysis of the ca ii triplet absorption line features requires accurate sky subtraction due to the high density of sky lines at long wavelengths .
each column of the wavelength - rectified spectrum was fit with a low order ( @xmath42 ) polynomial in regions uncontaminated by the galaxy .
the resulting polynomial was then subtracted from the entire column .
unfortunately , in our deepest off - plane integrations , clear errors in sky - subtraction remain .
these errors are quite small ( typically only 5% of the sky ) , and are not noticible in low signal - to - noise images .
however , they are a substantial source of uncertainty in the off - plane spectra , where the surface brightness of the galaxy is less than 10% of the sky and only 1% of the brightest skylines . measurements of an image of the slit suggest that these errors result from small - scale ( 10% ) variations in slit width , that lead to systematic errors when removing extremely bright sky lines .
while nod - and - shuffle techniques can greatly reduce this problem , our targets are too large to effectively use this observing mode .
after sky subtraction , frames were corrected for atmospheric extinction .
no absolute flux calibration was applied .
the spectra were then doppler corrected to the local standard of rest , spatially aligned , and combined using a sigma - clipping algorithm to eliminate cosmic rays . before extracting one - dimensional spectra , we re - binned our galaxy spectra to a scale linear in @xmath43 .
such re - binning allows doppler velocity shifts to be performed as linear shifts in @xmath44 .
the logarithmic bin size was set to 25 km / s in order to preserve the total number of pixels .
spectra were extracted by summing 30 - 300 pixels perpendicular to the dispersion direction , corresponding to radial bins of 4 - 40@xmath30 .
we varied the extraction bin size to ensure that each spectra reaches an average signal - to - noise ratio of 10 - 15 per pixel .
in the case of fgc 227 s shallower off - plane observations , we were forced to bin to a lower s / n ( @xmath45 ) to obtain reasonable sampling across the length of the galaxy . once extracted , the galaxy spectra were normalized by dividing by a low - order polynomial .
error spectra were computed by analyzing regions of blank sky on the reduced images , and are clearly dominated by the sky - subtraction problems discussed above ( see figure [ ex_fit ] ) . in regions where the sky spectra is smooth , our noise spectra closely matches expectations from poisson statistics and read noise .
columns that showed excessive systematic residuals caused by skylines ( @xmath46 above what would be expected from poisson noise ) were flagged and masked from later analysis .
the spectra of galaxies contain detailed information on their stellar kinematics .
methods to extract this kinematic information include cross - correlating the galaxy spectra with a template star @xcite , or directly fitting a redshifted stellar template @xcite .
direct fitting has the advantage of permitting variable weighting across pixels .
we therefore adopt the direct fitting method to allow us to mask bright skylines that overlap some of the ca ii triplet features .
we fit the extracted galaxy spectra with a redshifted and broadened k - giant stellar template in the restframe region of 8480 - 8700 , similar to the methods of @xcite and @xcite .
in particular , we construct a model galaxy spectra @xmath47 as @xmath48 where @xmath49 is the normalized template stellar spectra binned logarithmically in wavelength , z is a redshift , @xmath50 is a gaussian broadening function , @xmath51 is a low - order polynomial , @xmath52 is a flat continuum , and @xmath53 denotes convolution .
the polynomial and flat continuum are used here to correct for any errors made when normalizing the galaxy and template spectra .
the order of the polynomial is kept small to ensure that absorption features are not affected .
while the broadening function can be used to measure the line - of - sight stellar velocity dispersion , our large radial bins introduce additional broadening into our spectra .
we therefore defer an analysis of the stellar velocity dispersion to
[ losvdsec ] , and choose to hold the width of the gaussian broadening function constant during the fits .
we find the best - fit redshift and polynomial coefficients simultaneously using a levenberg - marquardt routine to minimize @xmath10 : @xmath54 where @xmath55 is the extracted galaxy spectra and @xmath56 is the weighting derived from the noise spectra with masked columns set to zero weight .
uncertainties were calculated from the covariance matrix and rescaled so that the reduced chi - squared equals unity ( i.e. , we assume our fits are of good quality ) .
this step is mostly conservative in that it greatly increases error bars on several points and leaves most virtually unchanged .
our rescaling also prevents any residual unmasked systematic errors from the sky subtraction from overly influencing our error spectra and fits .
an example spectra and template fit can be seen in figure [ ex_fit ] .
the raw extracted rotation curves are presented in figures [ rcs_227 ] and [ rcs_1415 ] .
photometry suggests that our sample galaxies have both a thin and a thick disk , both of which contribute significantly to the measured dynamics at each slit position ( table [ obstable ] ) .
we now attempt to decouple the kinematics of these two components .
we first introduce a convenient analytical equation for the rotation curves : @xmath57 where @xmath58 is the recessional velocity at the galaxy s center @xmath59 , @xmath60 is the asymptotic velocity at the flat part of the rotation curve , @xmath61 is the transition radius between rising and flat sections of the rotation curve , @xmath62 governs the degree of sharpness in the transition zone , and @xmath44 is a scaled radial parameter equal to @xmath63 .
this equation is equivalent to @xcite s model 2 with the parameter @xmath64 set to zero .
this equation is purely phenomenological and is flexible enough to fit a wide variety of rotation curves .
we create model rotation curves at each slit position as follows .
we first adopt independent rotation curves for the thick and thin disks , @xmath65 and @xmath66 , using the parameterized rotation curve of equation [ vceqn ] .
we then assume our observations are some linear combination of these two rotation curves , i.e. , that each velocity measurement is given by @xmath67 where @xmath68 is the fraction of luminosity from the thick disk at the slit position .
we fix the relative contributions of the thick and thin disks at each slit position using the photometric decompositions from @xcite ; as described in [ observationsec ] .
we adopt four different possible models , shown in tables [ rc227_fits ] & [ rc1415_fits ] and figure [ resids ] .
the first is our `` preferred '' model , corresponding to our best fit photometric decomposition .
we also include two extreme models , corresponding to the brightest and the faintest thick disks that are compatible with the photometric data .
for the `` faint thick '' model , we assume that the thin disk is larger and brighter , by decreasing @xmath69 and increasing @xmath70 and @xmath71 by their uncertainties , and that the thick disk is smaller and fainter , by adjusting the same parameters in the opposite direction .
we then recalculate @xmath72 . for the `` bright thick '' model ,
we make the opposite parameter shifts .
the `` faint thick '' and `` bright thick '' models are truly extreme fits to the photometric data , and result in reduced chi - squared ( @xmath73 ) values that are statistically much worse than the `` preferred '' models ( @xmath74 ) , particularly for the `` faint thick '' model . for completeness
, we also derive a `` simple '' model assuming that the midplane and above plane spectra are entirely dominated by the thin and thick disks , respectively .
these latter fits are shown in figure [ simple_rcs ] .
we do not include radial gradients in @xmath72 , because our photometric decompositions suggest the thick and thin disks have comparable scale lengths .
likewise , we neglect seeing , which should have no effect on the midplane value of @xmath72 .
seeing affects the off - plane values only marginally , shifting the value of @xmath72 by @xmath75 , which is an order of magnitude below the range of @xmath72 values we explore .
we bin our model rotation curves using the same bin sizes and flux weighting as our observations .
the rotation curve parameters of both disks are then adjusted to minimize @xmath10 , subject to the constraints that both rotation curves have the identical central velocity @xmath58 , that @xmath62 is greater than 0.7 , and that the galaxy s kinematic center @xmath59 is fixed to the flux maximum of the galaxy s continuum emission .
the value of @xmath76 is also constrained to be greater than -200 km s@xmath5 to prevent divergent fits .
we therefore have 7 free parameters : @xmath58 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , and @xmath82 .
the best fit values are listed in tables [ rc227_fits ] & [ rc1415_fits ] along with @xmath83 , which is the total reduced chi - squared for the photometric model plus rotation curve fit .
the `` faint thick '' disk models fail to converge to reasonable values , but we include the fits for completeness .
the resulting rotation curve fits are shown in figures [ simple_rcs ] , [ pref_rcs ] and , [ shaded_rcs ] . for both galaxies , _
every _ model has the rotation of the thick disk lagging the thin disk , or even counter - rotating .
we now test the justification for modeling the rotation curve data with two distinct kinematic components .
specifically , we are attempting to rule out the hypothesis that a single underlying rotation curve can explain both the midplane and above plane observations . using @xmath10 as a measure of goodness - of - fit , we expect the minimized values of @xmath10 to be drawn from a chi - squared distribution for @xmath84 degrees of freedom , where @xmath85 is the number of data points and @xmath86 is the number of adjustable parameters in our fit .
we calculate the probability @xmath87 that a value of @xmath10 larger than the one we find should occur by chance . when the midplane and off - midplane are fit with independent rotation curves ( the `` simple '' model ) , we find @xmath87=94% for fgc 1415 and 91% for fgc 227 , suggesting that this is a very good fit and/or we have overestimated our uncertainties .
we then compute the probability that the off - plane observations are described by same simple _ thin _ disk rotation curve that fit the midplane , but binned in the same manner as the thick disk observations .
this tests if both the midplane and the off - plane observations could be sampling the same underlying rotation curve .
we find that the off - plane observations of fgc 1415 have a 24% chance of matching the thin disk rotation curve . for fgc 227
, there is only a 4% chance the off - plane observations are fit by the simple thin disk rotation curve . ] .
although these probabilities do not formally eliminate the possibility that the midplane and off - plane observations are due to the same underlying rotation curve , they do suggest that it is highly unlikely . when coupled with the additional facts that our photometric decompositions strongly support two disk components , that there are different colors of the stellar populations at the two slit positions , and that the kinematics of the thick disk and thin disk kinematics in the milky way are different , we believe that our assumption of two kinematic components is strongly justified .
a final concern is that our observations could be biased by the effects of dust in the galaxies .
obscuration by dust may systematically reduce the velocities measured near the centers of galaxies , where the optical depth due to dust is largest .
these central velocities will be systematically lower due to the failure of observations to penetrate to the center of the disk , allowing only stars with small line - of - sight velocities to contribute to the spectra .
the resulting obscured rotation curve would then appear to be that of a rotating ring , and would thus mimic solid - body rotation at the inner radii that are strongly obscured by dust .
@xcite detect just such a signature in longslit spectroscopy of ngc 891 , a massive galaxy with a prominent dust lane .
the stronger the extinction caused by dust , the more an observed rotation curve would approach solid - body rotation @xcite .
this could be cause for concern as our rc s are slowly rising and fairly linear .
using monte carlo radiative transfer techniques , @xcite and @xcite find that realistic optical depths can not explain the slowly rising rotation curves in edge - on lsb galaxies such as ours , which lack dustlanes and have a very patchy , highly clumped distribution of dust @xcite .
@xcite conclude that dust can be ignored even at inclinations of @xmath88 in lsb galaxies and @xcite find that galaxies with rotational velocities @xmath89 km s@xmath5 are transparent at optical wavelengths .
our galaxies are comparable to those in the @xcite and @xcite studies and thus we are confident that our spectra are a true measure of stellar orbits and not an artifact of dust extinction .
there are three additional pieces of evidence suggesting that our results are largely unaffected by dust .
first , our galaxies are _ bluer _ in their midplanes than above them , indicating that they do not have central dust lanes .
second , our observations show that the off - plane rotation curves are lagging the midplane the _ opposite _ of what we would see if there was a dusty midplane . if dust were affecting the midplane , then the thick disk would be lagging the thin disk by even more than we have reported . finally , our observations , while not in the far infrared , are red enough ( @xmath90 ) that we should be able to penetrate the small amount of patchy dust which undoubtedly exists in the galaxies even in the absence of a concentrated dustlane .
l c c c c c c c c c c c preferred & 0.2 & 0.5 & 5493 .
& 4.6 & 0.9 & 20 .
& 1.4 & 111 . & -51 . & 5.0 + faint thick & 0.1 & 0.1 & 5493 .
& 4.8 & 0.5 & 200 . & 207.5 & 87 . & -37 . & 15.8 + bright thick & 0.5 & 0.9 & 5493 .
& 5.0 & 5.8 & 28 .
& 10.6 & 145 . & 16 . & 8.2 + simple & 0.0 & 1.0 & 5493 . & 5.1 & 7.0 & 176 . & 12.0 & 80 . & 35 . &
+ l c c c c c c c c c c c preferred & 0.3 & 0.7 & 1522 . & 2.8 & 0.8 & 17 . & 0.7 & 102 .
& 28 . & 24 + faint thick & 0.2 & 0.4 & 1522 .
& 2.7 & 21.4 & 100 .
& 15.2 & 97 . & -200 .
& 45 + bright thick & 0.4 & 0.9 & 1522 .
& 2.8 & 2.5 & 21 .
& 5.8 & 101 . & 40 . &
26 + simple & 0.0 & 1.0 & 1522 . & 2.7 & 2.2 & 33 . & 13.4 & 77 . & 41 . &
we now turn to an analysis of the line - of - sight velocity dispersion ( losvd ) of our galaxy spectra .
while direct fitting returns quality fits for recessional velocities , we find that at low s / n this method fails when adding additional free components , and is thus unable to measure the losvd . to measure the losvd we have therefore used the updated cross - correlation ( xc ) method of @xcite .
this method measures the losvd by fitting the peak of a template star - galaxy xc with a broadened stellar autocorrelation function ( axf ) . to avoid possible errors from template mismatch , we use two stellar templates , hd4388 ( spectral type k3iii ) and hd213014 ( spectral type g9iii ) .
both were observed with gmos - n for other programs and were downloaded from the canadian astronomy data centre ( cadc ) archive .
the stars were reduced using standard iraf routines and binned logarithmically in wavelength .
spectra of reference arcs were used to determine the amount of artificial broadening needed to match the resolution of the stellar template to the galaxy spectrum . before performing the xc
, we performed several pre - processing steps .
first , we once again extracted galaxy spectra by summing our reduced 2d image along the spatial direction until adequate s / n was reached ( @xmath91 ) .
next , the galaxy and stellar templates were normalized by dividing by a low - order fit to the continuum , after which the mean was subtracted .
the spectra were padded ( 100 pixels ) to eliminate `` wrap - around '' effects when cross - correlating .
residual low frequency components were removed by tapering the ends of the spectra in the fourier domain .
the faintness of our off - plane spectra requires that we bin or observations spatially to reach adequate signal - to - noise .
however , using such large extraction bins artificially broadens our galaxy spectra by including stars from different parts of the rotation curve . to eliminate this
, we first run the xc fitting code to find the recessional velocity along the galaxy .
we then shift individual rows of the 2d galaxy spectrum to remove the measured rotation and re - extract 1d spectra .
the xc fit is then run again in order to measure the losvd .
we find no major effects from template mismatch and can use our template stars interchangeably . when fitting for rotation , the two templates return nearly identical results with an rms spread of 1.2 km s@xmath5 with a maximum of 4 km s@xmath5 difference .
velocity dispersions show a @xmath92 km s@xmath5 systematic difference which is less than the statistical uncertainties and may result from mismatching the dispersions based on reference arc spectra .
we must emphasize that the velocity dispersion measurements are very uncertain , especially for fgc 227 . in the direct fitting method
, we were able to mask troublesome skylines and use partial sections of the ca triplet .
when using the xc method , masking small sections of spectra can create large changes in the fourier domain , so we were forced to use only 2 of the 3 ca absorption features for fgc 227 .
therefore , the losvd measurements for fgc 227 should be viewed with caution .
the resulting measurements of the line - of - sight velocity and velocity dispersion are shown in figure [ cc_fits ] .
the rotation curves from this method confirm the results of our direct fitting method .
both methods show a possible slight lag for the thick disk of fgc 1415 and zero net rotation for the thick disk of fgc 227 .
the losvd measurements of fgc 1415 are similar for the midplane and off - plane , and show a roughly constant velocity dispersion with radius .
the midplane of fgc 227 shows a possible increase in the losvd with radius .
unfortunately , the off - plane observations of fgc 227 have sufficiently low s / n and high sky - line contamination that the losvd measurements have very large error bars .
we are therefore unable to detect the increase in velocity dispersion expected for a thick disk which is not supported by rotation .
the midplane losvd is qualitatively consistent with the presence of a counter rotating thick disk , but the off - plane observations have too low s / n to make firm conclusions . to understand the radial behavior of the losvd , we modeled the velocity fields of two superimposed rotating disks viewed edge - on , corresponding to a thick and thin disk rotating at two different speeds .
we have assumed each disk has an intrinsic velocity dispersion of @xmath93 km s@xmath5 .
we created a series of artificial spectra from a combination of two broadened template stellar spectra with different flux strengths and recessional velocities , and then recovered their losvd with our xc procedure .
a sample of the resulting losvd are plotted in figure [ lagplot ] .
we find that , at our resolution , the xc method is unable to separate the cross - correlation peaks unless the two stellar components have velocity differences @xmath94 km s@xmath5 . for a case like
fgc 1415 , where the thick disk is only mildly lagging the thin disk , we would therefore expect only a slight increase in the observed losvd to occur with radius , consistent with the constant losvd seen in figure [ cc_fits ] .
however , we find that adding a second component at a different velocity can dramatically increase the observed losvd if the rotational velocity difference is in the range 90 - 150 km s@xmath5 and the second component comprises more than 20% of the total flux . for fgc 227
we observe a large increase in velocity dispersion with radius as expected if there is a counter - rotating thick disk present .
the results of [ rcsec ] and [ losvdsec ] indicate that the thick disks of fgc 227 and fgc 1415 have very different kinematics .
the raw rotation curves in figures [ rcs_227 ] & [ rcs_1415 ] show that both galaxies appear to be rotating more slowly above the midplane than within it . however , when the thick and thin disk kinematics are decomposed ( figure [ shaded_rcs ] ) , even stronger differences emerge . in fgc 1415 , the decomposition strongly favors a thick disk that rotates in the same direction as the embedded thin disk , but with a lag of 50% or more .
the kinematics of fgc 1415 are therefore fully consistent with those seen in the milky way , where the thick disk lags the rotation of the thin disk by up to @xmath95 at large scale heights @xcite .
this suggests that any of the theories that can explain the kinematics of the milky way s thick disk can also explain the kinematics of fgc 1415 .
the kinematics of fgc 227 are quite different .
the very slow rotation above the plane in fgc 227 is due entirely to contamination from the thin disk . unlike fgc 1415 ,
the thick disk in fgc 227 is counter - rotating , if it is even rotating at all . while monolithic collapse scenarios could potentially explain the rotation of fgc 1415 and the milky way s thick disk , they are unable to explain the slow counter - rotation seen in fgc 227 . in such scenarios ,
the rotational lag between the thick and thin disks arises from the spin up of gas as it collapses into the thin disk while preserving its angular momentum .
thus , the thick and thin disks must always rotate in the same direction , in contrast to the behavior of fgc 227 . the data for fgc 227 could marginally accommodate a very slowly co - rotating thick disk , but only if its thick disk formed very early in the collapse process , before the gas had spun up substantially .
this would form a thick disk that is substantially thicker and more extended than the thin disk .
however , the structure of fgc 227 s and fgc 1415 s thick and thin disks are comparable , ruling out monolithic collapse as a possible explanation .
the data for fgc 227 are also incompatible with models where the thick disk forms via vertical heating of a young thin disk .
simulations show that infalling satellites @xcite and dark matter halo substructure @xcite increase the velocity dispersion of the disk stars , but do not substantially alter their angular momenta .
the slowly rotating thick disk of fgc 227 is therefore unlikely to have formed from a rapidly rotating thin disk .
another possibility is that the thick disk formed from a thin disk that was also slowly rotating .
such a disk must have had a much smaller scale length because of its low angular momentum , and thus , the heated thick disk would have a smaller scale length as well .
this is in contrast to observations that consistently find large thick disk scale lengths @xcite .
we therefore rule out vertical heating as a plausible mechanism for forming fgc 227 s thick disk .
our conclusions are consistent with those of @xcite , who performed a very different analysis of thick disk structural parameters in s0 galaxies .
having ruled out monolithic collapse and vertical heating , the only remaining possibility is that the thick disk formed from direct accretion of stars from satellites . in this scenario , two or more galactic fragments merge together to form the final galaxy .
the majority of the stars in the satellites remain in the thick disk , due to their inability to lose kinetic energy and `` cool '' into a thinner disk .
meanwhile , the gas in the merging satellites collapses further into a rotating disk , converts into stars , and forms the thin disk .
the gas in the thin disk may be supplemented at later times by additional gas infall .
these `` shredded satellites '' are becoming popular as explanations for a wide variety of structures in the mw @xcite , as well as for low velocity stars at high latitude in the milky way thick disk @xcite .
this class of models can also reproduce the enrichment patterns seen in milky way halo stars @xcite .
recently , specific satellite accretion models for the formation of the thick disk have been advocated on the basis of numerical simulations . @xcite and
@xcite both identify thick disk components in galaxies formed in cosmological simulations . the thick disk in @xcite is composed of stars accreted from merging galactic `` sub - units '' , while @xcite find that the thick disk stars form during the chaotic merging of gas - rich clumps that produce the disk itself .
these new simulations form thick disks through a series of mergers , in contrast to older models that relied on a single final merging event to heat a thin disk into a thick disk @xcite .
the merging scenario explored by the @xcite and @xcite simulations may sometimes reproduce the behavior of the older disk heating models , for example , when one massive satellite comes in at late times .
however , the newer models provide mechanisms for producing a wider range of thick disk properties .
variations in the orbital properties and gas richness of the merging proto - galactic clumps easily yield variations in the kinematics of the thick disk , as well as its size and its luminosity relative to the thin disk .
for example , satellite merging can readily produce a modest rotational lag for the thick disk .
in general , the same satellites produce both the gas that settles into the thin disk and the stars which are left in the thick disk , assuming little subsequent gas accretion
. the thin disk and thick disk should therefore show similar angular momentum distributions , but the rotation speed of the gas should be slightly higher , because it has contracted further into the halo .
this process naturally leads to a thin disk which rotates more rapidly and has a smaller scale length than the thick disk , as seen in the milky way and fgc 1415 .
a different combination of orbital parameters and gas richnesses in the merging satellites can also produce the slowly rotating thick disk of fgc 227 . if the merging satellites had little net angular momentum ( for example , two equal mass satellites orbiting the center of mass in opposite directions ) , then the stars the satellites deposited in the thick disk would show little net rotation .
however , if one of the merging satellites was particularly gas rich , then it alone could provide enough gas to create the rapidly rotating thin disk .
this pathway would lead to a rapidly rotating thin disk , a stationary or even counter - rotating thick disk , and comparable scale lengths for the thick and thin disks , exactly as seen in fgc 227 .
the resulting thick disk would also have a large velocity dispersion , but unfortunately , our data are not sufficient to conclusively demonstrate this for fgc 227 .
one possible limitation with the stochastic satellite accretion model is its questionable ability to consistently form highly flattened stellar systems .
our past imaging indicates that late - type disks routinely host flattened thick disks with @xmath05:1 axial ratios @xcite .
however , if satellites are accreted from random directions , then there is no _ a priori _ reason why the the satellite debris should necessarily be aligned with the embedded thin disk . one would therefore expect to see a significant fraction of galaxies that lack thick disks , but are instead embedded in round stellar halos .
there are several possible resolutions to this problem .
the first is that much of the satellite debris that is not aligned with the disk instead becomes identified with the bulge
. this may be particularly true when the merging satellites are gas - rich and are on largely radial orbits , leading to centralized bursts of star formation and the formation of a high surface brightness bulge .
any stars formed during the burst , or accreted later , will tend to be identified with the brighter bulge component .
however , this potential solution can not explain why flattened thick disks are seen in bulgeless late - type galaxies .
the second possibility is that the orientation of the angular momentum vector of the accreted thick disk stars may be preferentially aligned with that of the accreted gas that forms the thin disk , even if the amplitude of the angular momentum vectors differ .
@xcite have argued that this alignment naturally arises from the tendency of satellites to travel along filaments , providing a preferred direction for satellite accretion .
the alignment also may occur through dynamical friction .
satellites on circular orbits will have longer lifetimes , and will thus be influenced more by dynamical friction .
simulations by @xcite show how a non - planar satellite orbit can be quickly dragged into the plane by dynamical friction in only @xmath96 orbits .
the relative contribution of stars from satellites with different orbits in the @xcite simulations support this picture .
the third possibility is that the tendency to detect flattened thick disks is a selection bias against detecting rounder distributions of satellite debris .
if a merging satellite does not happen to be in a coplanar orbit with the thin disk , then it will deposit stars over a larger volume throughout the stellar halo .
the resulting distribution of stars will have a much lower surface brightness than if the accreted stars were concentrated near the midplane in a thick disk .
this difference naturally biases against detecting a rounder distribution of stellar debris .
@xcite may have detected this component in deep intermediate - band imaging of ngc 4565 , as may have @xcite using stacked images of edge - on galaxies from the sloan digital sky survey .
however , there is a possibility that some of this light is due to the extended light from bulges embedded in the coadded thin disks , so this latter result remains ambiguous .
a final possibility is that the surface brightness of any round component will be further reduced by orbital instabilities . assuming that the potential of most galaxies are triaxial , orbits around the intermediate axes will be unstable .
these orbits will tend to reorient around the stable short or long axes , pulling a fraction of the stars back into a coplanar orientation .
we have measured the rotation curves and velocity dispersion of stars at and above the midplane of two edge - on galaxies .
comparisons between the two positions indicate that the stars above the plane are significantly lagging the rotation of the stars in the midplane .
we have decomposed the observed rotation into two components , corresponding to the thick and thin disks .
we find that the rotation speed of the thick disk component is less than 50% of the rotation speed of the thin disk , and is counter - rotating in one of the two cases .
we use these observations to conclusively rule out older monolithic collapse or disk heating models for forming all thick disks .
we conclude that the only viable mechanism capable of forming counter - rotating thick disks is one where the _ majority _ of thick disk stars are directly accreted from merging satellites .
simulations suggest that the early merging required by these models is a generic feature of hierarchical galaxy formation . as such ,
the models imply that all disk galaxies should host thick disks .
this suggests that merger models for the thick disk not only account for the variation in kinematics , but also explains the apparent ubiquity of thick disks around disk galaxies , as argued for by @xcite .
we gratefully acknowledge karl gebhardt for providing the k - giant spectra that were used for cross - correlating , and discussions with connie rockosi .
we d also like to the thank the gemini observing staff for their skill in executing these observations .
jjd and py were partially supported through nsf grant career ast-0238683 and the alfred p. sloan foundation . based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) , and conicet ( argentina ) . , f. , bond , i. a. , carter , b. s. , dodd , r. j. , fujimoto , m. , hearnshaw , j. b. , honda , m. , jugaku , j. , kabe , s. , kilmartin , p. m. , koribalski , b. s. , kobayashi , m. , masuda , k. , matsubara , y. , miyamoto , m. , muraki , y. , nakamura , t. , nankivell , g. r. , noda , s. , pennycook , g. s. , pipe , l. z. , rattenbury , n. j. , reid , m. , rumsey , n. j. , saito , t. , sato , h. , sato , s. , sekiguchi , m. , sullivan , d. j. , sumi , t. , watase , y. , yanagisawa , t. , yock , p. c. m. , & yoshizawa , m. 1999 , , 118 , 261 , b. , stoughton , c. , smith , j. a. , uomoto , a. , pier , j. r. , yanny , b. , ivezi ' c , .
, york , d. g. , anderson , j. e. , annis , j. , brinkmann , j. , csabai , i. , fukugita , m. , hindsley , r. , lupton , r. , munn , j. a. , & the sdss collaboration .
2001 , , 553 , 184 , h. , burstein , d. , deng , z. , zhou , x. , shang , z. , zheng , z. , chen , j. , su , h. , windhorst , r. a. , chen , w. , zou , z. , xia , x. , jiang , z. , ma , j. , xue , s. , zhu , j. , cheng , f. , byun , y. , chen , r. , deng , l. , fan , x. , fang , l. , kong , x. , li , y. , lin , w. , lu , p. , sun , w. , tsay , w. , xu , w. , yan , h. , zhao , b. , & zheng , z. 2002 , , 123 , 1364 | we present kinematic measurements of the thick and thin disks in two edge - on galaxies .
we have derived stellar rotation curves at and above the galaxies midplanes using ca ii triplet features measured with the gmos spectrograph on gemini north . in one galaxy ,
fgc 1415 , the kinematics above the plane show clear rotation that lags that of the midplane by @xmath020 - 50% , similar to the behavior seen in the milky way .
however , the kinematics of the second galaxy , fgc 227 , are quite different .
the rotation above the plane is extremely slow , showing @xmath125% of the rotation speed of the stars at the midplane .
we decompose the observed rotation curves into a superposition of thick and thin disk kinematics , using 2-dimensional fits to the galaxy images to determine the fraction of thick disk stars at each position .
we find that the thick disk of fgc 1415 rotates at 30 - 40% of the rotation speed of the thin disk .
in contrast , the thick disk of fgc 227 is very likely counter - rotating , if it is rotating at all .
these observations are consistent with the velocity dispersion profiles we measure for each galaxy .
the detection of counter - rotating thick disks conclusively rules out models where the thick disk forms either during monolithic collapse or from vertical heating of a previous thin disk . instead , the data strongly support models where the thick disk forms from direct accretion of stars from infalling satellites . |
about 20% of the universe is made of cold dark matter .
the origin of the dark matter , however , still remains a mystery .
while the standard model ( sm ) of particle physics has been very successful in explaining the data from almost all of the particle physics experiments to date , none of the sm particles can be a good candidate for the dark matter .
its existence provides unambiguous evidence for new physics beyond the standard model .
a weakly interacting massive particle ( wimp ) is a promising candidate for the dark matter , given that the wimp relic density is naturally around the observed value @xcite @xmath0 for a wimp mass around the electroweak scale @xmath1 gev .
wimps also appear naturally in many beyond the standard model scenarios .
one popular candidate is the lightest neutralino in supersymmetric models , which provides an example of spin-@xmath2 wimp dark matter .
spin-1 wimp dark matter has been studied in the framework of extra dimension models , for example , the lightest kaluza - klein photon in universal extra dimension models @xcite .
spin-0 dark matter has been studied @xcite in the framework of the sm plus an extra scalar .
the minimal model includes an extra real scalar gauge singlet @xmath3 that is charged under a @xmath4 symmetry .
the only renormalizable couplings are the quartic self - coupling @xmath5 and the higgs-@xmath3 coupling @xmath6 .
it was shown that for a sm higgs with mass in the range of 100 gev @xmath7 gev and dark matter with mass in the range of 10 gev @xmath8 100 gev , viable regions of parameter space exists for the dark matter candidate @xmath3 to provide the right amount of relic density in the universe @xcite .
unlike the wimp , the mass and the couplings of such a scalar dark matter particle often needs to be fine tuned to be consistent with the observed dark matter relic density .
wimp - type scalar dark matter could appear in a simple extension of the sm with the addition of a higgs @xmath9 doublet .
unlike the sm higgs doublet , which couples to both fermion and gauge sectors , such an extra scalar doublet is _ inert _ in the sense that it couples to gauge bosons only .
such inertness can be guaranteed by imposing a discrete @xmath4 parity .
the lightest inert particle ( lip ) is therefore stable .
if it is neutral , it can be a good wimp dark matter candidate .
the inert higgs doublet model ( ihdm ) scalar doublet is not a higgs doublet since it does not obtain a vacuum expectation value .
such model is also referred to as inert scalar doublet model " in the literature .
] was first proposed in the late 70 s @xcite .
it has received recent attention @xcite since it could be used to solve the naturalness problem in the standard model . with a splitting in the masses of the neutral and charged components of the inert higgs fields , the resulting positive contribution to the oblique @xmath10 parameter allows this model to accommodate a sm higgs with a much heavier mass .
such an inert higgs sector also appears in the recently proposed left - right twin higgs model @xcite .
in addition , an inert higgs doublet ( or doublets ) could be used to explain the small majorana neutrino mass via the one - loop radiative seesaw mechanism @xcite , electroweak symmetry breaking @xcite , grand unification @xcite and leptogenesis @xcite .
there have been some studies on the dark matter candidate in the ihdm .
@xcite studied the dark matter relic density in the ihdm in certain regions of parameter space .
the right amount of relic density could be obtained for dark matter with mass around 40 gev @xmath11 80 gev or larger than 600 gev .
@xcite studied the neutrino signatures from dark matter annihilation .
the continuous gamma ray spectrum from fragmentation and monochromatic gamma ray lines were studied in ref .
@xcite and @xcite respectively .
positron and antiproton signatures were studied in ref .
direct detection has been studied in @xcite .
there is also a collider analysis based on the lep ii limit @xcite as well as collider signatures of @xmath12 associated production with @xmath13 at the large hadron collider ( lhc ) @xcite . in this work
, we performed a complete analysis of the dark matter relic density in the ihdm over the whole parameter space , taking into account various theoretical and experimental constraints .
the latest results of the collider constraints based on supersymmetric process @xmath14 search at the lep ii are imposed @xcite . unlike ref .
@xcite , in which only a low sm higgs mass @xmath15 gev and 200 gev are considered , our analysis also includes studies with a high sm higgs mass @xmath16 gev . in ref .
@xcite , the mass splittings between @xmath17 , @xmath18 and the dark matter candidate @xmath3 , @xmath19 , are fixed to be ( 50 , 10 ) gev and ( 10 , 5 ) gev for the low and high dark matter mass regions , respectively .
we study cases when the mass splittings between @xmath17 , @xmath18 and @xmath3 are small , in which coannihilations play an important role , as well as cases when the mass splittings are large . in regions that overlap with those analyzed in ref .
@xcite , our results agree with the literature .
we identify additional regions of parameter space , in which the dark matter relic density is also consistent with the wmap result but were overlooked previously .
we also present our results in the parameter spaces of physical higgs masses and higgs couplings , which are convenient to use for studies of collider phenomenology and dark matter detections .
the rest of the paper is organized as follows . in sec .
[ sec : model ] , we briefly present the ihdm . in sec .
[ sec : constraints ] , we discuss the theoretical and experimental constraints on the model parameter space . in sec . [ sec : relic ] , we present our results of the relic density analysis . in sec . [
sec : conclusion ] , we conclude .
the ihdm is an extension of the higgs sector of the sm . besides the usual higgs doublet @xmath20 , an additional higgs doublet @xmath21
is introduced : @xmath22 which is charged under @xmath23 as @xmath24 . unlike the sm higgs boson , which couples to both sm gauge bosons and matter fermions , the extra higgs doublet @xmath21 couples to the gauge sector only .
such couplings can be guaranteed by imposing a @xmath4 symmetry ( sometimes also called matter parity ) under which all particles except @xmath21 are even .
in particular , under the @xmath4 symmetry : @xmath25 while @xmath26 obtains a vacuum expectation value ( vev ) @xmath27 gev as in the sm , @xmath21 does not obtain a vev : @xmath28 .
the @xmath4 symmetry is , therefore , not spontaneously broken .
the lightest particle in @xmath21 is stable and could be a good dark matter candidate .
the most general cp - conserving potential in the higgs sector that respects the @xmath4 symmetry can be written as @xmath29 .
\label{eq : higgspotential}\ ] ] notice that the usual mixing term @xmath30 is forbidden by the @xmath4 symmetry .
after electroweak symmetry breaking , three degrees of freedoms in @xmath26 are eaten by massive gauge bosons @xmath31 and @xmath32 .
we are left with one physical higgs boson @xmath33 , which resembles the sm higgs boson , as well as four inert scalars : the cp even one @xmath3 , the cp odd one @xmath18 and a pair of charged ones @xmath17 .
the mass of @xmath33 is related to @xmath34 via @xmath35 the masses of @xmath3 , @xmath18 and @xmath17 are related to parameters in the higgs potential as @xmath36 we define the mass differences @xmath37 and @xmath38 as @xmath39 it is obvious that @xmath40 and @xmath41 control the mass splitting between the charged and neutral cp even states , while @xmath41 also controls the mass splitting between the cp odd and cp even states . in our analysis
below , we assume that the cp even scalar @xmath3 is the lip , therefore the dark matter candidate , giving @xmath42 .
the numerical results of the relic density analysis are similar if @xmath18 is the lip dark matter .
the higgs potential in eq . ( [ eq : higgspotential ] ) has seven parameters : @xmath43 they could be replaced by the higgs vev @xmath44 , physical higgs masses , mass splittings , and a sum of quartic couplings @xmath45 as @xmath46 in particular , @xmath47 shows up in the couplings of @xmath48 and @xmath49 , which is relevant for dark matter annihilation .
it is therefore convenient to pick @xmath47 as a model parameter .
the quartic coupling @xmath50 only shows up in self - couplings involving @xmath3 , @xmath18 and @xmath17 .
it does not play an important role for the dark matter analysis that we present below .
there are various experimental constraints on the ihdm from direct collider searches , indirect electroweak precision test and dark matter direct detections . *
* @xmath51 and @xmath32 decay widths * + light @xmath17 , @xmath3 and @xmath18 could lead to deviations of @xmath32 and @xmath51 decay widths from the sm value . on the other hand ,
@xmath52 have been measured precisely at the lep and the tevatron @xcite , which agree well with sm predictions
. therefore regions in which the decay processes @xmath53 and @xmath54 are kinematically allowed have already been excluded .
these translate into constraints for @xmath55 and @xmath56 as @xmath57 * * direct collider searches * + light neutral and charged higgses have been searched for at the lep and the tevatron .
limits from conventional searches for @xmath17 , @xmath3 and @xmath18 , however , do not apply for higgses in the ihdm since those searches rely on decays of higgses into fermion pairs and/or higgs production via top quark decay .
neutral and charged higgses in the ihdm , on the other hand , do not couple to fermions .
in particular , the charged higgs has been searched for at the lep and the tevatron @xcite . a lower mass bound of 74 @xmath11 79 gev at 95% c.l .
is obtained at the lep @xcite considering @xmath58 . a more recent search at cdf @xcite studied the charged higgs produced in the top quark decay @xmath59 , with @xmath60 further decaying into a pair of quarks , leptons , or @xmath61 .
the bounds on the charged higgs mass from those searches do not apply to higgses in the ihdm since higgs - fermion - fermion couplings are absent .
+ experimental signatures for @xmath3 , @xmath18 and @xmath17 , however , is similar to those of neutralinos and charginos in the minimal supersymmetric standard model ( mssm ) , as @xmath3 appears as missing energy at colliders , similar to the lightest supersymmetric particle(lsp ) in the mssm . in particular ,
searches of @xmath14 at the lep ii can be interpreted as searches for @xmath62 .
therefore , the null result for neutralino and chargino searches at the lep ii can be used to set limits on @xmath63 , @xmath64 and @xmath55 in the ihdm . a recent analysis
@xcite argued that a direct application of the upper limit on the lep ii @xmath65 cross section to the ihdm is oversimplified due to the difference between the mssm process @xmath65 and ihdm process @xmath12 .
based on delphi analyses of @xmath14 with @xmath66 , ref .
@xcite determined the efficiencies for the corresponding mssm and ihdm processes after cuts using monte - carlo simulations .
the ratio of the efficiencies is then used to rescale the mssm cross section upper limit and applied to the ihdm .
it is shown that regions satisfying @xmath67 gev , @xmath68 gev and @xmath69 gev are excluded by lep ii mssm searches .
this limit is stronger than previous estimations by the direct application of the @xmath65 cross section upper limit to the ihdm @xcite . for @xmath70 gev , however , limits on scalar masses are much weaker , due to the small mass splitting and the resulting soft jets and leptons in the final states . for @xmath70 gev ,
only the lep i limit , @xmath71 , applies .
+ similarly , searches of the supersymmetric channel @xmath72 at the lep ii @xcite can be used to set a bound on @xmath63 .
taking into account the cross section difference of scalars versus fermions , a limit of @xmath73 gev can be derived from the lep ii chargino searches @xcite . * * electroweak precision test ( ewpt ) * + electroweak precision measurements provide strong constraints for any new physics beyond the sm . in the global electroweak precision fit to the sm , a light higgs is preferred : @xmath74 gev , with @xmath75 gev at 95% c.l .
oblique parameters @xcite @xmath3 , @xmath10 and @xmath76 are often used to parameterize radiative correction to gauge boson propagators .
a heavier sm higgs contributes positively to @xmath77 while negatively to @xmath78 .
extra higgs bosons in the ihdm , namely @xmath17 , @xmath3 and @xmath18 , also contribute to the @xmath3 and @xmath10 parameters , which could possibly cancel the effect of a heavy sm higgs boson @xcite .
therefore , a heavy sm higgs can be accommodated in the ihdm . in our analysis , we require the overall contribution to the @xmath3 and @xmath10 parameters from extra scalars in the ihdm to fall within the 68% c.l .
ellipse in @xmath79 plane .
since we considered possible large mass splittings @xmath56 , where the approximate formulas for @xmath80 and @xmath81 in ref .
@xcite might not be valid , we used the full expressions for the contributions to @xmath80 and @xmath81 for a higgs doublet @xcite when imposing this constraint . + fig .
[ fig : pew ] shows the allowed region for mass splittings @xmath37 and @xmath38 given the constraints on the oblique parameters @xmath3 and @xmath10 from electroweak precision measurements , for several different choices of @xmath55 .
the dependence on @xmath55 is small , especially for large @xmath56 . for @xmath15 gev
( left plot ) , the allowed region falls along the diagonal direction of @xmath82 .
the mass splittings @xmath56 could both be large as long as these two mass splittings are close to each other .
for @xmath16 gev ( right plot ) , a large contribution from the inert higgs doublet is needed to cancel the large positive @xmath3 and negative @xmath10 contribution from a heavy sm higgs boson .
therefore , the value for @xmath37 is constrained to be quite large @xmath83 gev , while the value for @xmath38 could still be as small as 0 . * * dark matter detection * + dark matter direct detection excluded spin - independent dark matter@xmath11nuclei scattering cross section up to about @xmath84 at 90% c.l .
@xcite , which is 7@xmath118 orders of magnitude larger than dark matter @xmath11 nucleon scattering via @xmath32 exchange .
this limit , however , can be easily avoided by introducing a small mass splitting between @xmath3 and @xmath18 : @xmath85 a few hundred kev , given the typical kinetic energy of the dark matter and the momentum transfer between the dark matter and the scattering nuclei in such scattering processes .
+ the spin - independent dark matter@xmath11nucleon scattering cross section via sm @xmath33-exchange is given by @xcite @xmath86 where @xmath87 is the nucleon mass , @xmath88 is the reduced mass of the dark matter@xmath11nucleon pair , and @xmath89 is used to parameterize the nucleonic matrix element . the typical range for @xmath89 is taken to be @xmath90 @xcite . in our analysis
, we take @xmath89 to be 0.14 to be on the safe side .
latest results from xenon10 , cdms , cresst , cogent and texono are used in our analysis @xcite .
portions of parameter space in small @xmath55 , @xmath91 and large @xmath92 region are excluded by the upper limits on the spin - independent dark matter@xmath11nucleon cross section .
+ the bounds from indirect dark matter detection ( gamma rays , for example ) is weak in the ihdm .
moreover , those bounds are subject to large astrophysical uncertainties involved in those observations
. therefore , we do nt consider constraints from indirect dark matter detections .
in addition , we impose the following theoretical constraints . * * vacuum stability * + we require the vacuum stability of the higgs potential at tree level , which leads to @xmath93 * * perturbativity * + we require that corrections to the beta function of @xmath34 from non - sm quartic couplings @xmath94 is less than the 50% of the sm term @xmath95 @xcite .
this amounts to the constraint : @xmath96 the evolution of the remaining quartic couplings does not lead to extra constraints .
in addition , we require the quartic coupling @xmath50 to be in the perturbativity region : @xmath97
we analyzed the relic density in the ihdm using the program micromegas @xcite .
this program solves the boltzmann equation numerically , using the program calchep @xcite to calculate all of the relevant cross sections .
when the mass splittings between the dark matter candidate and other particles are small , coannihilation effects are also included . in our analysis , we fixed the sm higgs mass to be 120 gev ( preferred by the ewpt in the sm ) or 500 gev ( preferred by naturalness , and could be consistent with the ewpt with large splittings between @xmath17 , @xmath3 and @xmath18 . ) .
we fixed @xmath98 since it does not enter into the dark matter relic density analysis .
we studied both the cases of small mass splittings and large mass splittings .
when @xmath37(@xmath38 ) is small , coannihilation between @xmath3 and @xmath17(@xmath18 ) is important .
part of our study overlaps with the analysis in ref .
@xcite , which , only considered small higgs masses @xmath15 gev and 200 gev with fixed mass splittings @xmath99 gev for the low mass region and @xmath100 gev for the high mass region .
results in ref .
@xcite were presented in @xmath101 plane .
we present our results in @xmath102 plane instead , which is more transparent since @xmath55 and @xmath47 are the two relevant parameters for the dark matter analysis , as well as studies of dark matter direct and indirect detections .
in particular , the value of @xmath47 is closely related to the neutrino and gamma ray flux from dark matter annihilation , and the cross section for dark matter@xmath11nucleon scattering .
larger @xmath47 typically leads to enhanced flux and increased dark matter@xmath11nucleon scattering cross sections , and the inert higgs dark matter has a better chance to be detected at future dark matter detection experiments .
we discuss below in detail two mass regions of @xmath103 that could provide the amount of dark matter relic density consistent with the wmap result at the 3@xmath104 level : ( a ) low mass region where @xmath105 100 gev , and ( b ) high mass region where 400 gev @xmath106 around a tev . to illustrate the dependence of the relic density @xmath107 on @xmath55 , @xmath47 and mass splittings @xmath56 , we present in fig . [
fig : low_120_relic ] curves of @xmath107 vs. @xmath55 for various choices of @xmath38 ( left plot ) and @xmath47 ( right plot ) .
the horizontal band indicates the 3@xmath104 region that is consistent with relic density measurement from wmap : @xmath108 .
a light sm higgs mass @xmath15 gev is used in both plots .
the red curve in the left plot of fig . [ fig : low_120_relic ] corresponds to @xmath99 gev .
therefore , coannihilation between @xmath3 and @xmath18 is important while coannihilation between @xmath3 and @xmath17 is not . for small @xmath55 , @xmath109 via sm @xmath33 exchange dominates , with cross section proportional to @xmath110 .
the cross section is typically small due to the small bottom yukawa coupling , which leads to relic density too big that overcloses the universe . when @xmath55 gets larger , @xmath111 via @xmath32 exchange becomes more and more important .
the relic density enters the wmap @xmath112 region for @xmath55 around 35 gev .
the coannihilation cross section maximizes at the @xmath32-pole : @xmath113 , corresponding to the dip around @xmath114 gev .
the coannihilation cross section decreases when @xmath55 increases away from the @xmath32-pole region , which makes the relic density falls back to the allowed region .
as @xmath55 gets larger , @xmath109 annihilation via @xmath33 starts to dominate and the relic density enters the @xmath112 region again . for @xmath115 ,
@xmath116 annihilation hits the @xmath33-pole , corresponding to the second dip around @xmath117 gev .
the annihilation cross section gets smaller once @xmath55 leaves the @xmath33-pole region . when @xmath118 gev , @xmath119 dominates .
the annihilation cross section quickly increases and the relic density drops below the wmap observed value . the dashed and dot - dashed curves in the left plot of fig . [ fig : low_120_relic ] shows the relic density dependence for @xmath38=8 gev and 5 gev respectively .
coannihilation effects get stronger for smaller mass splittings . therefore ,
for most of the @xmath55 region between 40 @xmath11 60 gev , the coannihilation cross section is too large and the relic density is too small .
curves in the right plot of fig .
[ fig : low_120_relic ] correspond to @xmath47=0.01 , 0.1 , 0.2 , 0.3 and 0.5 , respectively , while @xmath120 is fixed to be @xmath121 gev .
similar @xmath32-pole and @xmath33-pole features appear .
the relic density is smaller for larger @xmath47 , since @xmath116 annihilation via @xmath33-exchange is increased due to the increased @xmath48 coupling .
[ fig : low_120_1 ] shows the wmap 3@xmath104 relic density allowed region ( enclosed by two blue curves ) in the @xmath122 plane for @xmath123 gev with @xmath15 gev for @xmath124(50 , 10 ) gev ( left plot ) and ( 50 , 8) gev ( right plot ) .
shaded regions are excluded by various theoretical and experimental constraints , as described in sec .
[ sec : constraints ] . for @xmath124(50 , 10 ) gev ( left plot )
, the gap around @xmath114 gev corresponds to the @xmath32-pole .
the gap around @xmath125 gev corresponds to the @xmath33-pole .
the lep constraint ( yellow , light shade region ) is very strong due to the strong constraints on @xmath64 and @xmath55 when @xmath69 gev .
the precision electroweak constraints ( orange , medium shade region ) is weak since @xmath126 is slightly preferred by the fit to the @xmath79 contour . given all the constraints , only a small region around @xmath127 gev survives .
the value for @xmath47 for the allowed region , however , could be as large as @xmath128 .
such a large value of @xmath47 would be important for generating a large signal in the indirect detection of dark matter .
the lep constraints on @xmath55 and @xmath64 , however , are weakened for small mass splitting @xmath129 gev .
for such a small mass splitting , @xmath55 as low as around 40 gev is still allowed . in the right plot of fig .
[ fig : low_120_1 ] , the allowed parameter space is given for @xmath124(50 , 8) gev .
in most of the @xmath55 region between 40 gev and 60 gev , the relic density for the dark matter is too small due to the large coannihilation @xmath12 cross section .
however , there is a viable region for 60 gev @xmath130 gev with @xmath47 in the region of @xmath131 . the left plot of fig .
[ fig : low_120_2 ] shows the allowed relic density region for @xmath124(10 , 50 ) gev and @xmath15 gev .
the dark matter annihilation behaves similarly to the ( 50,10 ) case , only that the previous @xmath12 coannihilation via @xmath32 is replaced by @xmath132 coannihilation via @xmath31 .
the lep constraints ( yellow , light shade region ) becomes weaker comparing to @xmath133 gev case , due to the weaker constraints on @xmath63 .
the precision electroweak constraints ( orange , medium shade region ) , however , becomes stronger .
given all the theoretical and experimental constraints , a region with 60 gev @xmath134 80 gev and @xmath135 survives .
when @xmath37 and @xmath38 are both small , the coannihilations between @xmath3 , @xmath18 and @xmath17 are relevant and the allowed 3@xmath104 regions shrink due to the enhanced coannihilation cross sections . for ( @xmath136=(10 , 10 ) gev , there is no allowed region that survives given all the theoretical and experimental constraints , which is shown in the right plot of fig .
[ fig : low_120_2 ] . for @xmath137 gev
, although the lep ii constraints is weaker , the wmap 3@xmath104 region for @xmath138 gev completely disappears due to strong coannihilation effects .
[ fig : low_120_3 ] shows the allowed relic density region for @xmath15 gev when both @xmath56 are large : @xmath139 gev ( left plot ) and @xmath140 gev ( right plot ) .
the gaps in the low mass region corresponding to the @xmath32-pole or @xmath51-pole disappear , since the mass splitting is too large for the coannihilation process to be important .
the @xmath33-pole , however , still survives due to the low value of @xmath15 gev that we pick .
the lep i+ii constraints ( yellow , light shade region ) are weaker because of the large value of @xmath38 .
the perturbativity constraints ( hatched region ) , however , are stronger , since larger mass splittings @xmath56 correspond to larger values for @xmath141 . for splitting @xmath139 gev , a region with 55 gev
@xmath134 90 gev and @xmath142 is consistent with the wmap 3@xmath104 region .
when the mass splittings get larger , the perturbativity constraint gets stronger ( the hatched region shifts to the left ) while the lep ii constraint gets weaker ( the yellow , light shade region also shifts to the left ) .
for @xmath140 gev , a region with 30 gev @xmath134 50 gev and @xmath143 opens up . for @xmath144
gev the perturbativity constraint is so strong that no mass window survives all of the constraints .
[ fig : low_500 ] shows the allowed relic density region for a high value of the sm higgs mass @xmath16 gev .
electroweak precision constraints require a large value for @xmath37 ( as shown in fig .
[ fig : pew ] ) in order to compensate for the @xmath145 and @xmath146 contributions from a heavy sm higgs .
the left plot of fig .
[ fig : low_500 ] shows the allowed relic density region for @xmath147 gev .
no pole regions appear due to the large mass splittings and large @xmath91 . for @xmath148 gev , @xmath109 dominates , which corresponds to the nearly horizontal band of the 3@xmath104 wmap region .
once @xmath149 75 gev , @xmath119 opens up , which leads to the nearly vertical band of the 3@xmath104 region .
dark matter with mass around 75 gev is allowed given all of the constraints .
note that the perturbativity bounds are much weaker due to the large sm higgs mass ( therefore , large @xmath34 ) . on the other hand , to obtain the right relic density , large @xmath47 is needed to compensate the suppression of the annihilation cross section by the large higgs mass @xmath91 .
although @xmath47 could be as large as 3 , the indirect detection in this region is not promising due to large suppression of the annihilation cross section by @xmath150 .
the numerical results do not change much for smaller values of @xmath38 .
a @xmath32-pole coannihilation region appears for @xmath38 around 10 gev . the right plot of fig .
[ fig : low_500 ] shows the allowed relic density region for @xmath151 gev .
the gap around @xmath152 40 gev is due to the @xmath32-pole .
the lep ii constraints on @xmath55 and @xmath64 do not apply , due to the small mass splitting .
this allows a relatively large @xmath55 region around @xmath153 gev to remain open , with @xmath154 .
this region , however , shrinks for smaller @xmath38 due to the stronger coannihilation effects .
therefore , no wmap 3@xmath104 region survives for @xmath38 less than about 6 gev .
[ fig : high_120 ] shows the relic density in the @xmath122 plane for the high mass region : @xmath155 400 gev , with @xmath91=120 gev .
the lep search bound is irrelevant now since the @xmath55 is much larger than the direct search limit .
the precision electroweak constraint is weak for @xmath91 around 120 gev as long as @xmath37 and @xmath38 do not differ too much .
the dark matter direct detection constraint is also weak due to the heavy dark matter mass . for small mass splittings , @xmath156 gev ( left plot in fig .
[ fig : high_120 ] ) , a region of @xmath157 gev and @xmath158 is allowed .
the dominating annihilation channels are @xmath159 , @xmath160 and @xmath161 . the dependence on @xmath47
is introduced by the @xmath48 coupling . for @xmath47 close to zero ,
a smaller value of @xmath55 is preferred to increase the annihilation cross section .
when @xmath92 gets larger , a larger value of @xmath55 is needed for the relic density to fall into the 3@xmath104 band .
the region to the right of the 3@xmath104 band overclose the universe while the region to the left of the 3@xmath104 band corresponds to the under - abundance region .
regions with large values of @xmath92 are excluded due to the perturbativity constraints .
the annihilation cross section grows for large mass splittings .
therefore , the wmap allowed region shifts to larger @xmath55 for larger @xmath56 , as shown in the right plot of fig .
[ fig : high_120 ] for @xmath162 gev .
@xmath163 650 gev falls into the wmap 3@xmath104 region .
the result for @xmath164 gev is very similar .
the region forbidden by perturbativity constraints ( hatched region ) , however , shifts to the left for larger @xmath56 . therefore , no allowed region is left if at least one of @xmath165 12 gev . for a large sm higgs mass @xmath16 gev ,
a large mass splitting @xmath166 gev is needed to satisfy the precision electroweak constraints .
there is no region in @xmath122 that survives after all of the experimental and theoretical constraints are taken into account .
we studied the simple extension of the sm higgs sector when an extra inert higgs doublet is introduced that couples to the gauge sector only .
the lighter of the neutral components could be a good wimp dark matter candidate .
we explored the parameter spaces of the ihdm , taking into account the relic density constraints from wmap and various theoretical and experimental constraints .
table [ table : summary ] summarizes five distinctive regions that could provide the right amount of cold dark matter in the universe which satisfy all of the constraints . in regions ( i )
@xmath11 ( iv ) , the dark matter candidate @xmath3 , along with at least one of the other scalars @xmath18 , @xmath17 is light . those particles could be pair produced at the large hadron collider ( lhc ) as @xmath12 , @xmath167 , @xmath132 and @xmath168 with cross sections around the fb level .
heavier scalars @xmath18 and @xmath17 could decay into the lightest one @xmath3 via on - shell(or off - shell ) @xmath169 and @xmath170 , which further decay into quarks , leptons and neutrinos .
there are typically large missing @xmath171 in those processes due to the undetectable @xmath3 particles which are at the end of the decay chains .
experimentally , we can search for events with single lepton @xmath172 missing @xmath171 , dilepton @xmath172 missing @xmath171 , trilepton @xmath172 missing @xmath171 or in general , jets @xmath172 leptons @xmath172 missing @xmath171 .
the dominant sm background comes from @xmath159 , @xmath160 and @xmath161 .
the collider analysis on this model is currently under study @xcite .
the scalar mass in the high mass region ( v ) , however , is larger than 500 gev .
since they only have weak interactions , the production cross section at the lhc is typically too small . for the low @xmath91 region ( ii ) , when a relatively large @xmath173 could be accommodated , the indirect detection of the dark matter via its annihilation into neutrinos , photons , electrons and positrons could be very promising .
a recent study on the indirect neutrino signals @xcite showed that in the low @xmath55 region , tens to a hundred of neutrino events per year from dark matter annihilation inside the earth and hundreds of neutrino events per year from dark matter annihilation inside the sun can be expected at future neutrino telescopes .
indirect photon signals including monochromatic photon line , fragmentation photon spectrum and final state radiation photon spectrum in the ihdm is under current study @xcite .
high @xmath91 regions ( iii ) and ( iv ) typically do nt have promising indirect detection signals due to the suppression of the annihilation cross section by large @xmath91 .
the indirect detection possibility for the high @xmath55 region ( v ) is also less comparing to region ( ii ) . in ref .
@xcite , it is shown that neutrino events from the earth is too low to be observed , while a few events per year is expected from neutrinos from the sun . in summary ,
the ihdm is a simple extension of the sm which provides a very promising wimp dark matter candidate .
there are several regions of the parameter space that could provide the right amount of dark matter relic density in the universe .
the collider phenomenology of this model is very rich . for certain regions ,
the indirect detection via neutrinos or photons is possible at future neutrino and gamma ray telescopes or ground based experiments .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) .
g. servant and t. m. p. tait , nucl .
b * 650 * , 391 ( 2003 ) [ arxiv : hep - ph/0206071 ]
. v. silveira and a. zee , phys .
b * 161 * , 136 ( 1985 ) ; l. lopez honorez , e. nezri , j. f. oliver and m. h. g. tytgat , jcap * 0702 * , 028 ( 2007 ) .
p. agrawal , e. m. dolle and c. a. krenke , phys .
d * 79 * , 015015 ( 2009 ) [ arxiv:0811.1798 [ hep - ph ] ] .
s. andreas , m. h. g. tytgat and q. swillens , arxiv:0901.1750 [ hep - ph ] .
m. gustafsson , e. lundstrom , l. bergstrom and j. edsjo , phys .
lett . * 99 * , 041301 ( 2007 ) [ arxiv : astro - ph/0703512 ] .
e. nezri , m. h. g. tytgat and g. vertongen , arxiv:0901.2556 [ hep - ph ] .
d. majumdar and a. ghosal , mod .
a * 23 * , 2011 ( 2008 ) [ arxiv : hep - ph/0607067 ] .
e. lundstrom , m. gustafsson and j. edsjo , arxiv:0810.3924 [ hep - ph ] .
q. h. cao , e. ma and g. rajasekaran , phys .
d * 76 * , 095011 ( 2007 ) [ arxiv:0708.2939 [ hep - ph ] ]
. w. m. yao _ et .
_ , journal of physics g * 33 * , 1 ( 2006 ) and 2007 partial update for edition 2008 available on the pdg www pages ( url : http://pdg.lbl.gov/ ) . a. abulencia _ et al .
_ [ cdf collaboration ] , phys .
* 96 * , 042003 ( 2006 ) .
a. pierce and j. thaler , jhep * 0708 * , 026 ( 2007 ) [ arxiv : hep - ph/0703056 ] . the lep susy working group , http://lepsusy.web.cern.ch/lepsusy/ , lepsusywg/01 - 03.1 .
j. angle _ et al . _ [ xenon collaboration ] , phys .
lett . * 100 * , 021303 ( 2008 ) z. ahmed _ et al . _
[ cdms collaboration ] , phys .
* 102 * , 011301 ( 2009 ) sarah andreas , michel h. g. tytgat , and quentin swillensy , jcap * 0904 * , 004 ( 2009 ) d. s. akerib _ et al . _
[ cdms collaboration ] , phys .
d * 68 * , 082002 ( 2003 ) g. angloher _ et al .
_ [ cresst collaboration ] , astropart .
physics * 18 * , 43 ( 2002 ) c. e. aalseth _
[ cogent collaboration ] , phys .
* 101 * , 251301 ( 2008 ) s. t. lin _ et al . _
[ texono collaboration ] , phys .
d * 79 * , 061101 ( 2009 ) g. blanger , f. boudjema , a. pukhov , a. semenov , comput.phys.commun.176:367-382,2007 ; hep - ph/0607059 ; g. belanger , f. boudjema , a. pukhov and a. semenov , arxiv:0803.2360 [ hep - ph ] . | the lightest neutral scalar in the inert higgs doublet model is a natural candidate for wimp dark matter . in this paper
, we analyzed the dark matter relic density in the inert higgs doublet model .
various theoretical and experimental constraints are taken into account .
we found that there are five distinctive regions that could provide the right amount of the relic density in the universe .
four out of those five regions have a light particle spectrum which could be studied at the large hadron collider . |
the layout of the na61/shine detector @xcite is presented in figure [ fig : detector ] .
it consists of a large acceptance hadron spectrometer with excellent capabilities in charged particle momentum measurements and identification by a set of five time projection chambers as well as time - of - flight detectors .
the high resolution forward calorimeter , the projectile spectator detector , measures energy flow around the beam direction , which in nucleus - nucleus reactions is primarily a measure of the number of spectator ( non - interacted ) nucleons and is thus related to the centrality of the collision .
additionally , detector system consist of three time - of - flight walls used for low momentum particle identification .
an array of beam detectors identifies beam particles , secondary hadrons and ions as well as primary ions , and measures precisely their trajectories .
the na61/shine experiment up to now used secondary hadron beams in the momentum range from 13 gev / c to 350 gev / c , as well as attenuated primary pb@xmath1 ion beams in the momentum range from 13@xmath0 gev / c to 158@xmath0 gev / c . for the start of the strong interactions physics program
secondary @xmath2be ions were produced via fragmentation of the pb@xmath1 ions as primary ions other than pb were not available before 2015 . in 2015 primary @xmath3ar ion beams in the same momentum range were delivered to na61/shine and @xmath4xe beams are expected in 2017 .
in the data taking on @xmath5be+@xmath6be collisions a 2 cm diameter interaction trigger counter s4 was placed on the beam - line between vtpc-1 and vtpc-2 .
this arrangement allowed to measure the total inelastic @xmath5be+@xmath6be cross section @xmath7 @xcite .
the na61/shine measurements at 13@xmath0 , 20@xmath0 and 30@xmath0 gev / c are presented in figure [ fig : inelastic ] .
the na61/shine results are in good agreement with an earlier measurement at lower beam momentum @xcite and a glauber model calculation using the glissando code @xcite .
be+@xmath6be cross section as a function of beam momentum .
results are compared with the measurement of ref .
@xcite , world p+p data and glissando model predictions.,width=321 ] the rapidity spectra of @xmath8 in @xmath5be+@xmath6be collisions at the three beam momenta and four centrality classes together with data for inelastic p+p interactions are presented in figure [ fig : rapidity ] .
one can observe a small asymmetry in the rapidity distribution for @xmath5be+@xmath6be collisions around mid - rapidity .
this asymmetry may come from two effects : * the asymmetry between projectile and target nuclei ( @xmath5be projectile on @xmath6be target ) which is expected to enhance particle production in the backward ( target ) hemisphere , * the selection of central collisions which requires a small number of projectile spectators without any explicit requirement imposed on the number of target spectators ; this selection , when used for collisions of identical nuclei , would enhance particle production in the forward hemisphere . note that the two effects partially compensate leading to a relatively small asymmetry of the measured spectra
rapidity distributions of @xmath8 produced in inelastic p+p and be+be interactions ( 4 centrality classes ) at 20@xmath0 , 30@xmath0 , 40@xmath0 , 75@xmath0 and 150@xmath0 gev / c.,width=396 ] the excellent particle identification based on the combined time - of - flight ( tof ) and energy loss ( de / dx ) method allow us to calculate @xmath10 ratios .
the energy dependence of the @xmath11 ratio at mid - rapidity for inelastic p+p interactions and central pb+pb / au+au collisions is presented in figure [ fig : horn ] .
the na61/shine data suggest that even in inelastic p+p interactions the energy dependence of the @xmath11 ratio exhibits rapid changes in the sps energy range .
a step structure is seen which appears to be a precursor of the horn structure @xcite observed in central pb+pb collisions .
data obtained at rhic and lhc @xcite are also plotted in figure [ fig : horn ] to show the trend beyond the sps energy range .
ratio and inverse slope parameter t of @xmath12 transverse mass spectra at mid - rapidity.,width=472 ] the inverse slope parameter t was fitted to the transverse mass spectra .
figure [ fig : horn ] presents the energy dependence of t for @xmath13 in inelastic p+p reactions from na61/shine compared to world data for p+p and pb+pb / au+au reactions from refs .
results from inelastic p+p collisions exhibit a step structure like the one observed in central pb+pb interactions .
the @xmath14 rapidity distribution measured by na61/shine is compared to previous measurements in figure [ fig : lambda ] . in order to partly compensate for differences resulting from different collision energies
the comparison is done in terms of the scaled rapidity @xmath15 and the spectra are normalized to unity .
the mean @xmath14 multiplicity is calculated by summing up the measured rapidity spectrum and extrapolating it to lower and higher rapidities using the parametrization of the world data from figure [ fig : lambda ] .
the collision energy dependence of the mean @xmath14 multiplicity including the new na61/shine result is plotted in figure [ fig : lambda ] . produced in inelastic p+p collisions at 158 gev / c.,width=472 ] the transverse momentum fluctuations were studied using strongly intensive measure @xmath16 $ ] @xcite for @xmath5be+@xmath6be interactions at several collision centralities and energies .
the results were compared to those obtained from inelastic p+p interactions by na61/shine and are very similar .
transverse momentum fluctuations in @xmath5be+@xmath6be interactions do not present structures which could be related to the critical point .
this work was supported by the national science center of poland ( grants : 2014/12/t / st2/00692 , 2013/11/n / st2/03879 , 2012/04/m / st2/00816 ) . | the main physics goals of the na61/shine programme on strong interactions are the study of the properties of the onset of deconfinement and the search for signatures of the critical point of strongly interacting matter .
these goals are pursued by performing an energy ( beam momentum 13@xmath0
158@xmath0 gev / c ) and system size ( p+p , p+pb , be+be , ar+sc , xe+la ) scan .
+ this publication reviews results and plans of na61/shine .
in particular , recent inclusive spectra and new results on fluctuations and correlations of identified hadrons in inelastic p+p and centrality selected be+be interactions at the sps energies are presented .
the energy dependence of quantities inspired by the statistical model of the early stage ( kink , horn and step ) show unexpected behaviour in p+p collisions .
the na61/shine results are compared with the corresponding data of other experiments and model predictions . |
the interplay between mass loss and accretion in pre - main sequence systems is important in determining both the final stellar mass , the mass in the disk available to form planets , and the local physical conditions ( e.g. incident uv flux , temperature ) affecting planet formation and evolution . while many details of the physical mechanisms involved in mass accretion and mass outflow in the immediate circumstellar environments ( @[email protected] au ) for low mass class ii pre main sequence stars , or classical t tauri stars ( cttss ) , are now understood ( e.g. * ? ? ?
* ; * ? ? ?
* ) , the picture is less clear concerning these processes around the intermediate mass herbig ae / be stars ( haebes ) .
optical spectroscopic investigations into the nature of the circumstellar environments around haebes have been numerous .
a small number of these studies have been performed on medium or large samples of haebes .
@xcite presented medium resolution ( @xmath4v@xmath516 km s@xmath6 ) h@xmath7 and na i d lines for a sample of 43 haebes that focused mainly on the p cygni profiles of a small number ( 8) of objects .
the investigations of @xcite , @xcite , and @xcite all examined samples of @xmath530 haebes and a small or medium number of spectral line diagnostics . @xcite and @xcite compare their samples to known tts characteristics but these studies focus mainly on pure emission line morphologies and not on red and blue shifted absorption .
@xcite , in a spectroscopic haebe study covering the largest number of objects to date , classified the h@xmath7 profiles of 131 stars using low and medium resolution spectra .
@xcite also examined forbidden line emission in their sample .
however , no detailed interpretation of the h@xmath7 profile morphologies or forbidden line strengths are performed , yielding little information about the kinematics of the inner circumstellar material . finally , the haebe variability study of @xcite examined h@xmath7 , [ ] 6300 , he i 5876 , and the na i d lines in a medium sized sample using medium resolution spectra .
high spectral resolution observations of a wide range of accretion and outflow diagnostics in a large sample of haebes is clearly needed . besides the studies listed above , most optical spectroscopic haebe studies have focused on single objects or small sample sizes and a small number of spectral line diagnostics ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . while this is useful for providing details about specific objects , studies of this sort provide little information concerning the physical characteristics of mass flows around haebes as a group , which are important for illuminating the general differences between the early evolution of low and intermediate mass pre
main sequence stars . comparing large sample of haebes and cttss
can also provide insight into the important physical mechanisms responsible for generating mass flows in young stars . while both haebes and cttss host gaseous accretion disks , the lack of strong detected magnetic fields on haebes ( e.g. * ? ? ?
* ; * ? ? ?
* ) and their large @xmath8sin@xmath9 values compared to cttss @xcite provide hints that accretion and outflow processes do not proceed identically in both groups of objects .
magnetospheric accretion , widely accepted to operate in most ctts systems , funnels material from a truncated inner disk onto the surface of the star @xcite .
accretion along magnetic field lines is not possible from outside of a narrow region around the corotation radius in the disk @xcite .
@xcite pointed out that in order for magnetospheric accretion to operate around haebes , the large @xmath8sin@xmath9 values of haebes require the corotation radii , and thus the magnetic truncation radii , to be closer to the stellar surface than in cttss .
this has implications for the ability of haebes to launch accretion
generated outflows in the same way as cttss ( see of * ? ? ?
* for an expanded discussion ) .
spectral line profile morphologies can provide insight into how these mass flows are generated .
in this paper we present a high resolution optical study of a large sample of haebes using a large number of spectral line diagnostics with the goal of investigating differences in the accretion and outflow mechanisms operating around haebes compared to cttss .
this work is the second part of a multi wavelength study of accretion and outflows in haebes .
the first part is presented in ( * ? ? ?
* hereafter paper i ) , and focused on the study of in a sample of 56 haebes .
our comprehensive coverage of optical mass flow indicators enables us to derive reliable estimates and confidence intervals of the incidence of accretion and outflow signatures in haebes .
we describe our observations and data reduction in 2 . in 3 we discuss the calculation of stellar radial velocities and rotational velocities .
section 4 covers the classification of the line profiles .
we present the individual line and overall mass flow statistics in 5 , as well as a comparison to the statistics from .
a comparison of the entire sample statistics to those of cttss is also be given in 5 .
section 6 investigates the physical cause of the differences in morphology statistics .
our conclusions are summarized in 7 .
our sample consists of 87 haebes covering a wide range of spectral types and was chosen from the catalogues of @xcite ; @xcite ; and @xcite .
classification as a haebe in a previous survey was the only selection criteria .
it is important to note that our sample likely contains objects at various phases in their pre main sequence evolution .
however , the small number of haebes currently identified in the literature weakens the significance of a more specific comparison between distinct evolutionary groups .
thus we include all objects in our central analysis regardless of spectral type and estimated age , though presumably all are pre main sequence in nature .
our observations were performed using two cross dispersed echelle spectrographs on two telescopes : the sandiford echelle spectrometer ( ses * ? ? ?
* ) on the mcdonald observatory 2.1 m otto struve telescope and the tull 2dcoud echelle spectrometer ( ts2 * ? ? ?
* ) on the 2.7 m harlan j. smith telescope .
the specifics of each observation are given in .
some objects were observed multiple times during different observing runs . the use of observations from different epochs in the analysis is described in .
lccccc ab aur & ts2 & 2.7 m & 18sep2011 & 120 & 65 + ae lep & ses & 2.1 m & 01nov2012 & 2400 & 35 + bd+61 154 & ts2 & 2.7 m & 20sep2011 & 2500 & 45 + `` & ses & 2.1 m & 30oct2012 & 1200 & 30 + '' & ses & 2.1 m & 31oct2012 & 2000 & 40 + bf ori & ts2 & 2.7 m & 21jan2013 & 3000 & 65 + bh cep & ts2 & 2.7 m & 20sep2011 & 2500 & 45 + `` & ses & 2.1 m & 01nov2012 & 3600 & 30 + '' & ses & 2.1 m & 02nov2012 & 3600 & 30 + cq tau & ts2 & 2.7 m & 19sep2011 & 2100 & 65 + dw cma & ts2 & 2.7 m & 20jan2013 & 3600 & 20 + gsc 0479400827 & ts2 & 2.7 m & 21jan2013 & 3600 & 50 + hd 141569 & ts2 & 2.7 m & 20sep2011 & 180 & 90 + `` & ts2 & 2.7 m & 20jan2013 & 75 & 60 + hd 142666 & ts2 & 2.7 m & 20jan2013 & 600 & 65 + hd 163296 & ts2 & 2.7 m & 18sep2011 & 300 & 55 + hd 169142 & ts2 & 2.7 m & 19sep2011 & 200 & 50 + hd 190073 & ts2 & 2.7 m & 18sep2011 & 180 & 55 + '' & ts2 & 2.7 m & 26sep2012 & 350 & 60 + hd 203024 & ts2 & 2.7 m & 20sep2011 & 390 & 60 + `` & ses & 2.1 m & 30oct2012 & 600 & 30 + '' & ses & 2.1 m & 31oct2012 & 1500 & 50 + hd 244314 & ts2 & 2.7 m & 20sep2011 & 2200 & 65 + `` & ts2 & 2.7 m & 20jan2013 & 3000 & 80 + hd 244604 & ts2 & 2.7 m & 20sep2011 & 1300 & 75 + hd 245185 & ts2 & 2.7 m & 19sep2011 & 2300 & 65 + hd 249879 & ts2 & 2.7 m & 26sep2012 & 3200 & 60 + '' & ses & 2.1 m & 01nov2012 & 3600 & 35 + `` & ses & 2.1 m & 02nov2012 & 3600 & 45 + hd 250550 & ses & 2.1 m & 01nov2012 & 2000 & 40 + '' & ses & 2.1 m & 02nov2012 & 3000 & 100 + hd 287823 & ts2 & 2.7 m & 22sep2011 & 3000 & 55 + hd 290409 & ts2 & 2.7 m & 22sep2011 & 3000 & 70 + hd 290500 & ts2 & 2.7 m & 23sep2011 & 3600 & 60 + hd 290764 & ts2 & 2.7 m & 23sep2011 & 2400 & 70 + hd 290770 & ts2 & 2.7 m & 21sep2011 & 1200 & 90 + hd 34282 & ts2 & 2.7 m & 21sep2011 & 1500 & 70 + hd 35187 & ts2 & 2.7 m & 18sep2011 & 240 & 70 + `` & ses & 2.1 m & 01nov2012 & 1200 & 35 + '' & ses & 2.1 m & 02nov2012 & 2400 & 40 + hd 35929 & ts2 & 2.7 m & 18sep2011 & 400 & 50 + hd 36408 & ts2 & 2.7 m & 18sep2011 & 120 & 40 + hd 37357 & ts2 & 2.7 m & 18sep2011 & 500 & 55 + hd 37411 & ses & 2.1 m & 30oct2012 & 2400 & 30 + hd 38120 & ses & 2.1 m & 30oct2012 & 600 & 30 + `` & ses & 2.1 m & 31oct2012 & 2000 & 55 + hd
50083 & ts2 & 2.7 m & 18sep2011 & 180 & 50 + hk ori & ts2 & 2.7 m & 21sep2011 & 3600 & 30 + il cep & ts2 & 2.7 m & 20sep2011 & 1300 & 80 + '' & ts2 & 2.7 m & 22sep2011 & 1200 & 75 + `` & ts2 & 2.7 m & 23sep2011 & 1300 & 80 + ip per & ts2 & 2.7 m & 19sep2011 & 2400 & 70 + iras 050440325 & ts2 & 2.7 m & 20jan2013 & 3200 & 35 + iras 06071 + 2925 & ts2 & 2.7 m & 20jan2013 & 3600 & 30 + iras 070610414 & ts2 & 2.7 m & 21jan2013 & 3600 & 60 + iras 174811415 & ts2 & 2.7 m & 20sep2011 & 3600 & 30 + iras 183060500 & ts2 & 2.7 m & 20sep2011 & 3600 & 30 + iras 18454 + 0250 & ts2 & 2.7 m & 19sep2011 & 3000 & 20 + iras 19343 + 2926 & ts2 & 2.7 m & 19sep2011 & 3000 & 20 + '' & ts2 & 2.7 m & 26sep2012 & 3000 & 15 + `` & ses & 2.1 m & 01nov2012 & 3600 & 10 + '' & ses & 2.1 m & 02nov2012 & 3600 & 10 + lkh@xmath7 134 & ts2 & 2.7 m & 20sep2011 & 2700 & 60 + `` & ses & 2.1 m & 01nov2012 & 3600 & 35 + '' & ses & 2.1 m & 02nov2012 & 3600 & 40 + lkh@xmath7 208 & ses & 2.1 m & 01nov2012 & 3600 & 35 + `` & ses & 2.1 m & 02nov2012 & 3600 & 35 + '' & ts2 & 2.7 m & 22jan2013 & 3000 & 65 + lkh@xmath7 233 & ts2 & 2.7 m & 22sep2011 & 3600 & 20 + `` & ts2 & 2.7 m & 26sep2012 & 3600 & 20 + lkh@xmath7 257 & ts2 & 2.7 m & 21sep2011 & 3800 & 35 + lkh@xmath7 324 & ts2 & 2.7 m & 20sep2011 & 3600 & 40 + lkh@xmath7 339 & ts2 & 2.7 m & 22jan2013 & 1960 & 20 + mwc 1080 & ts2 & 2.7 m & 21sep2011 & 2500 & 20 + '' & ts2 & 2.7 m & 22sep2011 & 2500 & 20 + `` & ts2 & 2.7 m & 23sep2011 & 2500 & 20 + '' & ses & 2.1 m & 30oct2012 & 2000 & 20 + `` & ses & 2.1 m & 31oct2012 & 3000 & 10 + mwc 120 & ts2 & 2.7 m & 18sep2011 & 400 & 80 + mwc 137 & ts2 & 2.7 m & 20jan2013 & 1800 & 15 + mwc 147 & ts2 & 2.7 m & 26sep2012 & 600 & 20 + mwc 300 & ts2 & 2.7 m & 18sep2011 & 2700 & 10 + mwc 361 & ts2 & 2.7 m & 18sep2011 & 180 & 70 + mwc 480 & ts2 & 2.7 m & 18sep2011 & 180 & 50 + mwc 610 & ts2 & 2.7 m & 18sep2011 & 360 & 50 + '' & ts2 & 2.7 m & 26sep2012 & 900 & 60 + mwc 614 & ts2 & 2.7 m & 18sep2011 & 120 & 75 + mwc 758 & ts2 & 2.7 m & 18sep2011 & 400 & 55 + mwc 778 & ses & 2.1 m & 30oct2012 & 3600 & 10 + `` & ses & 2.1 m & 31oct2012 & 4000 & 10 + mwc 863 & ts2 & 2.7 m & 23sep2011 & 500 & 50 + nz ser & ts2 & 2.7 m & 19sep2011 & 3000 & 40 + r mon & ts2 & 2.7 m & 26sep2012 & 2000 & 20 + rr tau & ses & 2.1 m & 30oct2012 & 2500 & 35 + '' & ses & 2.1 m & 31oct2012 & 3600 & 35 + `` & ts2 & 2.7 m & 20jan2013 & 3600 & 55 + sv cep & ts2 & 2.7 m & 20sep2011 & 2500 & 65 + t ori & ts2 & 2.7 m & 23sep2011 & 3000 & 70 + '' & ses & 2.1 m & 30oct2012 & 1200 & 20 + `` & ses & 2.2 m & 31oct2012 & 3600 & 50 + ux ori & ts2 & 2.7 m & 22sep2011 & 3600 & 30 + uy ori & ses & 2.1 m & 30oct2012 & 3600 & 20 + v1185 tau & ts2 & 2.7 m & 19sep2011 & 2400 & 70 + '' & ses & 2.1 m & 30oct2012 & 2000 & 30 + `` & ses & 2.1 m & 31oct2012 & 3000 & 40 + v1578 cyg & ts2 & 2.7 m & 20sep2011 & 2500 & 90 + '' & ts2 & 2.7 m & 22sep2011 & 2400 & 90 + `` & ts2 & 2.7 m & 23sep2011 & 2400 & 80 + v1685 cyg & ts2 & 2.7 m & 18sep2011 & 2500 & 25 + '' & ts2 & 2.7 m & 26sep2012 & 3000 & 30 + `` & ses & 2.1 m & 01nov2012 & 2000 & 25 + '' & ses & 2.1 m & 02nov2011 & 3000 & 30 + v1686 cyg & ts2 & 2.7 m & 19sep2011 & 3600 & 20 + v1787 ori & ts2 & 2.7 m & 22jan2013 & 3600 & 35 + v1818 ori & ts2 & 2.7 m & 22jan2013 & 3600 & 40 + v346 ori & ts2 & 2.7 m & 21sep2011 & 2400 & 75 + v351 ori & ts2 & 2.7 m & 18sep2011 & 650 & 50 + v361 cep & ts2 & 2.7 m & 18sep2011 & 2200 & 50 + v373 cep & ts2 & 2.7 m & 21sep2011 & 3500 & 40 + v374 cep & ts2 & 2.7 m & 21sep2011 & 2800 & 30 + `` & ts2 & 2.7 m & 26sep2012 & 3000 & 50 + '' & ses & 2.1 m & 30oct2012 & 2000 & 30 + `` & ses & 2.1 m & 31oct2012 & 2000 & 60 + v380 ori & ses & 2.1 m & 30oct2012 & 2400 & 15 + '' & ses & 2.1 m & 31oct2012 & 3600 & 15 + v586 ori & ts2 & 2.7 m & 20sep2011 & 1500 & 75 + v590 mon & ts2 & 2.7 m & 21jan2013 & 3600 & 40 + v599 ori & ts2 & 2.7 m & 22jan2013 & 3200 & 30 + v718 sco & ts2 & 2.7 m & 23sep2011 & 600 & 50 + v791 mon & ses & 2.1 m & 01nov2012 & 2400 & 30 + `` & ses & 2.1 m & 02nov2012 & 3600 & 45 + vv ser & ts2 & 2.7 m & 18sep2011 & 3000 & 45 + '' & ts2 & 2.7 m & 21sep2011 & 3000 & 40 + `` & ts2 & 2.7 m & 22sep2011 & 3000 & 40 + ww vul & ts2 & 2.7 m & 18sep2011 & 1600 & 40 + '' & ts2 & 2.7 m & 26sep2012 & 2000 & 70 + `` & ses & 2.1 m & 30oct2012 & 600 & 25 + '' & ses & 2.1 m & 31oct2012 & 1500 & 45 + xy per & ts2 & 2.7 m &
19sep2011 & 2000 & 65 + z cma & ts2 & 2.7 m & 21jan2013 & 1800 & 20 + the ts2 data were obtained during three separate runs in 2011
september ( 7 nights ) , 2012 september ( 3 nights ) , and 2013 january ( 3 nights ) . for the ts2 observations , we employed the 1.2 slit with the e2 grating to achieve a resolving power of @xmath10@xmath560,000 , or a velocity resolution of @xmath55 km s@xmath6 .
the grating configuration provided wavelength coverage from approximately 370010000 .
exposure times were limited to 1hour in order to maximize the number of objects observed with at least moderate signal to noise .
a s / n of @xmath550 per pixel is typical for the ts2 data , although this varies from s / n@xmath510 up to @xmath590 depending on a combination of the object brightness and weather conditions during the exposure
. thar comparison lamp spectra were taken multiple times each night to be used for wavelength calibrations .
flat field exposures were collected at the beginning of each night .
telluric and spectroscopic standards were observed periodically throughout the course of each observing run .
the ses data were collected during a single 4night observing run in 2012 october .
we used two different grating settings , each one on two separate nights , to record both `` red '' ( 54306620 ) and `` blue '' ( 43754870 ) wavelengths .
this was necessary due to the ses s smaller wavelength coverage ( @xmath55001000 ) per exposure than that of ts2 . even using two wavelength settings , we were not able to observe all of the spectral diagnostics that are available in the ts2 data
. thus objects that are only observed with the ses are lacking data for some diagnostics . the resolving power for each grating
setting is @xmath10@xmath560,000 .
thar lamp spectra were taken throughout the night and flat field exposures were obtained at the beginning of each night .
telluric and spectroscopic standards were observed , although the spectroscopic standards obtained using the ts2 are used interchangeably due to the almost identical velocity resolutions of the two instruments .
all observations were reduced using custom idl routines optimized for extracting echelle spectra .
each object exposure is bias subtracted and divided by a normalized median flat to remove pixel to pixel variations .
the spectral orders are located and then fit using a 7@xmath11 order polynomial .
each order is optimally extracted which also removes most hot pixels and cosmic rays .
wavelength calibrations are performed on the thar exposures using a two dimensional polynomial fit to all spectral orders simultaneously .
hundreds of individual lamp lines in the exposures are matched to known atlas wavelengths .
this procedure results in dispersion solutions that are accurate to @xmath50.10 pixels , or 0.005 , across the entire spectral range .
telluric absorption was subtracted out of the spectral region near 6300 in order to make identification of the o i @xmath126300 line unambiguous . due to the prominence of the h@xmath7 and na i d lines , telluric absorption in these regions does not affect the classification of the line morphologies since these absorption are small perturbations of the true line profile shape .
equivalent width calculations of h@xmath7 lines include the telluric features which results in a maximum additional uncertainty of @xmath50.40 .
some exposures contain strong telluric o i emission near 5577 , 6300 , and 6363 .
we have modified our echelle reduction program to locate the night sky emission above and below the spectral trace of the star and subtract it from the actual object spectrum . however , this routine was not always successful , especially for objects with profiles that were not centered on the slit or for irregularly shaped profiles ( e.g. binaries ) . thus some profiles contain residual night sky emission ( e.g. hd 244314 ) .
this extra emission can be identified by its narrow ( @xmath510 km s@xmath6 ) width and is easily distinguished from the stellar line profile .
approximate continuum normalization in each order was performed by dividing each order by the blaze function of the same order , derived from the reduced spectrum of the median flat for that particular grating setting .
a linear fit is then performed to remove any residual slope .
some objects display strong photospheric balmer features which have wings that often extend to the ends of the spectral order .
the continuum normalization in these cases is likely not exactly the true stellar continuum . however , identification of non
photospheric features in these lines is not problematic with the use of a comparison standard since the chosen continuum level does not affect the line shape .
the fe ii 4352 and n ii 6583 features often lie on the wings of h@xmath13 and h@xmath7 , respectively .
these profiles are normalized to the continuum in the balmer lines .
in order to correctly separate circumstellar from photospheric features , it is important to have estimates of both the stellar rotational velocity ( @xmath8sin@xmath9 ) and the radial velocity ( rv ) of the object relative to the barycenter . for many of the objects in our sample these values
have been estimated in the literature . for objects that do not have previous estimates , we fit rotationally broadened and velocity shifted synthetic spectra to a single spectral order to obtain rv and @xmath8sin@xmath9 estimates .
the order selected is chosen based on the strongest available photospheric lines .
this method works well for a type and late b
type objects but fails for the earliest spectral types .
typically the strong he i @xmath124471 and mg ii @xmath124481 lines are used to measure @xmath8sin@xmath9 for rapidly rotating b type stars @xcite ; for more slowly rotating b type stars , metal lines near 6400 and 4500 are used @xcite .
many of our early type objects , however , show no absorption features at any of these wavelengths . for a typical s / n spectrum in our sample
, this method can also fail for stars with @xmath8sin@xmath14 200 km s@xmath6 because the features become too weak to reliably measure .
lower resolving powers of @xmath10@xmath510,000 - 30,000 are typically used to measure the large @xmath8sin@xmath9 values of b type stars ( e.g. * ? ? ?
we do not obtain @xmath8sin@xmath9 estimates for these objects . for objects that do not have literature or synthetic spectrum
fit rv estimates , we use the interstellar na i d lines as an rv proxy . @xcite were the first to show , using a sample of 10 haebes , that the velocities of the molecular clouds in which haebes are embedded are very similar ( mean residual of -3.1 @xmath153 km s@xmath6 ) to the velocities of observed interstellar ca ii and na i d velocities in the same objects
. they also showed that independently measured stellar rvs were similar to the molecular cloud velocities , although with a larger mean residual of 8.5 @xmath15 8 km s@xmath6 .
thus the na i d velocities can , in general , provide a rough estimate of the stellar rv to within @xmath510 km s@xmath6 . as the mass flow diagnostics we are interested in are typically several hundred wide , this is adequate for our purposes
. for objects in our sample for which we determine rvs based on synthetic spectrum fits , we compare these measurements to the rv determined from the object s interstellar na i d lines .
rvs determined by the synthetic spectral fits from @xcite are also used .
the interstellar na i rv estimate is made by manually pointing at the approximate center of the interstellar absorption .
this is appropriate since the estimates are only accurate to @xmath58 .
an example of the na i interstellar medium ( ism ) absorption is shown in for hd 142666 .
the na i ism absorption is very strong and narrow and is clearly distinguished from the photospheric profile .
this is a particularly good example of the strength of using the na i value as an estimate for the system rv : the na i and synthetic model values differ by only @xmath52.5 .
further examples of the na i ism features can be seen in the individual object plots in the appendix .
the result of the comparison is plotted in .
most objects show a velocity discrepancy of @xmath1610 .
the four objects that lie outside of the 20 boundaries are all confirmed binaries and the large velocity differences may be due in part to their orbital motion .
we find a mean velocity difference of [email protected] for the objects in , where the uncertainty is the 95% confidence interval given by the _ t_distribution @xcite .
the mean difference drops to 7.9 if the four outliers are excluded , a negligible difference .
the standard deviation of the absolute value of the difference between the interstellar rv values and those determined by the model fits is 8.5 .
we take this to be the uncertainty associated with the values determined from the na i lines .
our sample thus confirms the findings of @xcite . for future haebe studies we suggest that the interstellar na i d line velocities be used for objects that , ( 1 ) do not allow for reliable photospheric fits either due to lack of absorption features or very high @xmath8sin@xmath9 , or ( 2 ) only @xmath510 accuracy is needed in the radial velocity .
all other physical information about the objects examined in this study ( e.g. @xmath17 , @xmath1 , @xmath18 ) are taken from the literature .
the relevant references are given in .
lccccccccc ab aur & a0 & 24.7 & 2.50 & 2.62 & 116 & -6.85 & y & 40 & 1,2 + ae lep & b6 & * 28.0 * & & & & & & & 39 + bd+61 154 & b8 & -16.0 & 3.40 & 2.42 & 112 & & y & 70 & 1,6,9 + bf ori & a2 & 22.0 & 2.58 & 3.26 & 39 & @xmath16 - 8.00 & y & & 1,2,6,8 + bh cep & f5 & -5.6 & 1.70 & 2.40 & 98 & @xmath16 - 8.30 & y & 84 & 2,38,40 + cq tau & f2 & 35.7 & 2.93 & 5.10 & 98 &
@xmath16 - 8.30 & y & 29 & 1,2,12,37 + dw cma & b3 & & & & & & y & & 4,13 + gsc 04794 - 00827 & a ? & * 25.0 * & & & & & & & 43 + hd 141569 & a0 & 35.7 & 2.33 & 1.94 & 228 & -6.90 & y & 55 & 1,2,14 + hd 142666 & a5 & * -9.7 * & 2.15 & 2.82 & 65 & -7.22 & y & & 1,3,15 + hd 163296 & a1 & -9.0 & 2.23 & 2.28 & 129 & -7.16 & y & & 1,3,6,18 + hd 169142 & a7 & -0.4 & 1.69 & 1.64 & 48 & & y & 0 & 1,46 + hd 190073 & a1 & 0.2 & 2.85 & 3.60 & 4 & & y & 45 & 1,19 + hd 203024 & a1 & -14.0 & 2.80 & 3.40 & 162 & & y & 74 & 1,44 + hd 244314 & a1 & 22.5 & 2.33 & 2.07 & 52 & -6.90 & & & 1,3 + hd 244604 & a4 & 26.8 & 2.66 & 3.69 & 98 & -7.20 & y & & 1,3,21 + hd 245185 & a1 & 16.0 & 2.19 & 1.85 & 118 & -7.20 & y & 81 & 1,3,6,44 + hd 249879 & a2 & 11.0 & 4.00 & 5.90 & 249 & -8.00 & & & 1,3 + hd 250550 & b8 & -22.0 & 4.80 & 3.50 & 79 & -7.80 & y & & 1,3,13 + hd 287823 & a0 & -0.3 & 2.50 & 2.60 & 10 & & ? & & 1 + hd 290409 & a2 & 80.3 & 2.04 & 1.75 & 250&&&&1 + hd 290500 & a2 & 29.0 & 1.96 & 1.68 & 85 & & & & 1 + hd 290764 & a8 & 26.3 & 1.86 & 2.30 & 60 & & y & 32 & 41 + hd 290770 & b9 & * 24.6 * & 2.86 & 2.49 & 240&&&&1 + hd 34282 & a3 & 16.2 & 1.59 & 1.66 & 105&@xmath16 - 8.30 & y & & 1,2,18,22 + hd 35187 & a2 & 27.0 & 1.93 & 1.58 & 93 & -7.60 & y & & 1,3,24 + hd 35929 & f1 & 21.1 & 4.13 & 8.10 & 61.8&&y & 56 & 1,44 + hd 36408 & b8 & 15.0 & 4.10 & 3.50 & & @xmath16 - 8.00 & ? & & 2 + hd 37357 & a1 & 21.4 & 2.47 & 2.83 & 124 & & y & & 1,47 + hd 37411 & b9 & * 14.1 * & & & & & y & 63 & 33,44 + hd 38120 & b9 & 28.0 & 2.49 & 1.91 & 97&-6.90 & y ?
& @xmath168 & 1,3,35 + hd 50083 & b4 & -0.5 & 12.10 & 10.0 & 233 & & ? & & 1 + hk ori & a3 & * 22.9 * & 3.00 & 4.10 & * 20 * & -5.24 & y & & 2,6 + il cep & b4 & -39.0 & & & 179 & & ? & & 1 + ip per & a3 & 13.7 & 1.86 & 2.10 & 80 & & ? & & 1 + iras 05044 - 0325 & b3 & * 25.2 * & & & & & & & 43 + iras 06071 + 2925 & b9 & * 12.5 * & & & & & & & 43 + iras 07061 - 0414 & b0 & * 43.0 * & & & & & & & 42 + iras 17481 - 1415 & a0 & * -5.1 * & 2.50 & & & & & & 10 + iras 18306 - 0500 & b2 & 33.1 & 12.00 & 6.80 & & & y & 87 & 40,44 + iras 18454 + 0250 & b1 & * @xmath50 * & 19.00 & 8.50 & & & & & 42,13 + iras 19343 + 2926 & b1 & * -10.7 * & & & & & & & 43 + lkh@xmath7 134 & b3 & * -13.0 * & 6.00 & 5.05 & & & y & 58 & 40,44 + lkh@xmath7 208 & a4 & * 13.6 * & 2.40 & 3.20 & & & y & 56 & 40,44 + lkh@xmath7 233 & a3 & * -9.10 * & 3.20 & 4.45 & & & y & 18 & 40,44 + lkh@xmath7 257 & b5 & * -15.2 * & 3.61 & 2.68 & & & y & 57 & 44 + lkh@xmath7 324 & b8 & * -10.5 * & 5.10 & 6.30 & & & y & 42 & 40,44 + lkh@xmath7 339 & a1 & * 26.4 * & 3.18 & 3.92 & & & y & 40 & 40,44 + mwc 1080 & b1 & & 17.4 & 7.3 & & & y & 83 & 1,6,10 + mwc 120 & b9 & 47.0 & 3.94 & 4.60 & 120&-6.85 & y & & 1,3 + mwc 137 & b1 & & & & & & y & 80 & 6,10,13 + mwc 147 & b2 & * 18.6 * & 6.60 & 4.80 & 83 & -6.12 & y & & 1,3,13,48 + mwc 300 & b1 & * -4.5 * & 7.37 & 3.33 & & & y & 62 & 13,44 + mwc 361 & b4 & * -7.0 * & 10.70 & 10.40 & 26 & & y & 26 & 1,44 + mwc 480 & a4 & 12.9 & 1.93 & 1.93 & 98&@xmath16 - 7.23 & y & 37 & 1,2,12 + mwc 610 & b3 & 14.0 & 8.00 & 4.70 & 219 & & ? & & 1 + mwc 614 & a0 & 15.1 & & & 69 & -6.59 & y & & 1,3,28 + mwc 758 & a5 & 17.8 & 2.90 & 4.40 & 54&-6.05 & y & 21 & 1,3,12 + mwc 778 & b2 & * 10.3 * & 10.00 & 3.50 & & & y & 75 & 13,42,49 + mwc 863 & a1 & -5.0 & 2.56 & 2.89 & 108&-6.12 & y & 38 & 1,2,29 + nz ser & b3 & * -17.0 * & 10.10 & 5.80 & & & y & 5 & 10,13 + r mon & b8 & * 22.3 * & 5.10 & 12.00 & & & y & 60 & 10,13 + rr tau & a0 & * 15.3 * & 5.80 & 9.30 & * 225 * & -6.86 & y & 44 & 2,32,44 + sv cep & a1 & * 24.6 * & 2.62 & 3.00 & 180 & & y & 49 & 1,44 + t ori & a3 & 56.1 & 3.13 & 4.47 & 147 &
-6.60 & y & & 1,2,30 + ux ori & a1 & 12.0 & 6.72 & 12.1 & 221&-7.18 & y & @xmath168 & 1,3,6,35 + uy ori & b9 & * 19.8 * & & & & & & & 4 + v1185 tau & a2 & * 20.0 * & & & * 250 * & & & & + v1578 cyg & a1 & * -3.0 * & 5.90 & 9.70 & 199 & & y & & 1,6 + v1685 cyg & b4 & * -16.0 * & & & & & y & 41 ? &
30,34 + v1686 cyg & f8 & * -12.2 * & 3.50 & 14.00 & & & y & & 6,40 + v1787 ori & a5 & * 26.4 * & 1.88 & 1.84 & & -7.43 & & & 45 + v1818 ori & b7 & * 26.4 * & & & 46 & & & & 43 + v346 ori & a7 & 20.0 & 1.72 & 1.96 & 116&-6.90 & n ? & & 1,3,35 + v351 ori & a6 & 15.0 & 2.88 & 4.38 & 100 & & ? & & 1 + v361 cep & b4 & * -20.0 * & 8.11 & 6.70 & 278 & & n ? & & 1,6 + v373 cep & b8 & * -18.6 * & 5.30 & 14.00 & & & y & 45 & 39,40 + v374 cep & b5 & * -13.6 * & & & & & & & 51 + v380 ori & b9 & 27.5 & 2.87 & 3.00 & 7&-5.60 & y & & 1,3,6 + v586 ori & a1 & 31.0 & 2.28 & 1.94 & 200 & & y & 81 & 1,44 + v590 mon & b7 & * 17.7 * & 4.71 & 5.86 & & & & 84 & 40,44 + v599 ori & f0 & * 7.0 * & 1.83 & 5.29 & * 96 * & & y & 42 & 43,44 + v718 sco & a4 & -3.6 & 1.93 & 2.25 & 113 & & y ? & 32 ? & 1,35 + v791 mon & b5 & * -2.6 * & & & & & ? & & 4 + vv ser & b7 & * -2.0 * & 4.00 & 3.10 & 124 & -7.50 & y & 46 & 1,3,44 + ww vul & a2 & -4.0 & 3.70 & 5.40 & 196 & -6.38 & & & 1,2 + xy per & a2 & 2.0 & 1.95 & 1.65 & 224&-7.02 & ? & & 1,3 + z cma & b9 & * -27.0 * & 3.80 & 3.20 & & -6.72 & y & & 3,31 + [ fig : fig1a ] [ fig : fig1 ]
line profiles are classified into six morphological groups : p cygni ( pc ) , inverse p cygni ( ipc ) , double peaked emission ( dp ) , single peaked emission ( e ) , absorption ( a ) , and featureless ( f ) . for the purpose of generating mass flow statistics , pure absorption profiles ( a )
are further grouped as red or blue absorption based on the maximum absorption velocity : @xmath19 @xmath20 0 km s@xmath6 and @xmath21 @xmath16 0 km s@xmath6 .
the objects counted as showing blue(red)shifted absorption either have a pc(ipc ) profile _ or _ an a profile that is blue(red)shifted .
objects are classified as e or dp if the profile is in emission ( single and double peaked , respectively ) and there is no clear evidence of absorption below the local continuum .
profiles are labeled as f if they show no detectable departure from the estimated photospheric line profile .
we note that a small number of the pc profiles , most of them h@xmath7 profiles , show blue shifted absorption that does not extend below the stellar continuum , i.e. , the absorption is superposed on the wing of the emission profile ( see , for example , the h@xmath7 profile in ) . while these are not strictly p
cygni profiles , we have chosen to extend the pc classification to these objects since it is the closest morphology group that describes the profile . in order to accurately classify the morphologies , each profile was compared with a rotationally broadened spectroscopic standard of a similar effective temperature . objects with unknown @xmath8sin@xmath9 values did not have their comparison spectra broadened . for most objects , this allowed weak emission and absorption features to be identified in lines with significant photospheric components .
this comparison is most important for the balmer lines since most of our objects have strong photospheric balmer line absorption .
an example of this effect is shown in the top panel of where the core of the h@xmath13 profile of hd 190073 is clearly filled in by emission .
a second example is shown in the bottom panel of for a more complex line morphology at h@xmath0 for hd 290770 . in this case
the blue shifted absorption at @xmath5 - 100200 can clearly be seen superimposed on the broad emission profile .
the comparison with the standard makes it obvious that this is non photospheric absorption . to be clear
, we note that the profiles shown in and show only the continuum normalized profiles no photospheric subtraction was done to make these plots even though photospheric comparisons were made while classifying the profiles .
the photospheric comparisons are done by eye and , in general , it is obvious when a non photospheric contribution to the line profile is present .
thus no quantitative comparison of the photospheric and circumstellar components is performed .
the approximate comparisons prevent gross misclassification of line profiles , i.e. identifying circumstellar absorption when the absorption actually forms in the photosphere .
thus objects will often show line profiles that appear to show structure but are in fact due entirely to photospheric absorption .
examples of an a0 and b3 spectroscopic standard are shown in and . in some objects
there is significant contamination of the he i 3889 , mg i 5167 , ca ii 3969 , and ca ii 8662 spectral lines by neighboring features .
this is especially severe at mg i 5167 ( e.g. z cma and dw cma ) , where we tentatively identify the nearby fe i 5168.89 line ( + 81.8 km s@xmath6 ) , and ca ii 3969 and he i 3889 which are often overwhelmed by the nearby h@xmath22 and h@xmath23 lines , respectively .
for example , in the he i 3889 line appears to show a pc morphology .
however , this profile more likely due to the nearby h@xmath23 line .
thus this line is listed as `` contaminated '' since the true profile is obscured by the h@xmath23 line .
the n ii 6548 line lies in the far blue wing of the h@xmath7 and often appears to show structure when in fact the line emission and shape are due entirely to h@xmath7 ( e.g. , ) .
if there are other lines of the same element that show a non featureless line profile , e.g. the mg i 5173 or 5184 lines in the case of mg i 5167 , the line is categorized as c for contaminated. if the other lines of the same element are featureless , we assume a featureless spectrum for the feature in question .
the rest velocities of these contaminating lines are marked with a dotted line in the individual profiles plots . with the exception of ca
ii 8662 , the wavelength regions containing these diagnostics are mostly featureless in the whole sample . contaminated profiles comprise 16% of the he i 3889 profiles , 3% of the mg i 5167 profiles , 7% of the ca ii k profiles , and 9% of the ca ii 8662 profiles .
thus the contamination is not problematic concerning the classification of most profile morphologies .
examples of the extracted profiles and their categorization are shown in and .
plots of the individual line profiles for all objects are given in appendix a. the profile classification is shown in the upper right corner of each plot window ; the spectral feature being shown is listed in the upper left .
spectral lines for which there is no data are labeled as such in the plots .
these lines either fell in between orders on the detector or the s / n was too low for analysis .
features with rest wavelengths below 5000 are plotted using a 5 km s@xmath6 ( 2 pixel ) bin window in order to enhance the s / n . the bottom
right panel of each figure shows a bar plot of the number of line profiles in each morphology category .
we note that only the pc and ipc bars are shown in color .
the bar showing the count for pure red or blue shifted absorption ( i.e. , the a profiles ) is shown in black even though the subdivision of the a profiles into red and blue shifted absorption is included in the statistics .
[ fig : fig2 ] [ fig : fig3 ] [ fig : fig4 ] profile of hd 190073 with a spectroscopic standard of similar t@xmath24 ( top panel ) .
the object profile is plotted in black and the standard photospheric profile is plotted in red .
the core of the line is clearly filled in by emission , placing this profile in the e morphology group .
a similar comparison is shown for hd 290770 ( bottom panel ) at h@xmath0 .
the standard , hr4386 , has been rotationally broadened to match the @xmath8sin@xmath9 of hd 290770 .
the blue shifted absorption can be seen superimposed on the emission profile , resulting in a profile classification of pc.,title="fig : " ]
haebes display a wide variety of seds indicating that , as a group , they represent objects at different phases of pre main sequence evolution @xcite .
an accurate calculation of mass flow statistics would only include objects that are currently in the same pre
main sequence phase as one another .
however , this is challenging due to the relatively small number of haebes each specific evolutionary phase would contain .
studies of this sort will necessarily suffer from smaller sample sizes in each group thus increasing the uncertainties associated with the mass flow statistics . in order to eliminate any objects from our sample that no longer show evidence of inner disk material , and
thus should not show evidence of mass accretion and mass loss , we employ the color color classification of @xcite .
the @xmath25 vs @xmath26 classification is a proxy for the full sed analysis presented in @xcite .
objects with @[email protected] and @[email protected] , the group iii objects from @xcite , likely have insufficient material in their inner disks to be actively accreting and ejecting material .
these objects should not be included in any comparisons to cttss since all cttss show strong evidence for accretion and mass loss .
[ fig : fig5a ] [ fig : fig5b ] we have collected @xmath27 , @xmath28 , and @xmath29 data from the literature for our observed sample in order to identify any potential group iii objects .
optical extinctions are also collected for some objects .
the @xmath25 and @xmath26 colors for our sample are plotted in .
objects with de
reddened colors are plotted as green circles while objects without known @xmath30 values are plotted as red squares .
reddening vectors for the standard @xmath10=3.1 extinction law are plotted as solid black lines .
an approximation of the main sequence for dwarf spectral types b8 through m6 is represented by the dashed dotted line .
main sequence objects are required to remain within the boundaries of the reddening vectors .
it can be seen that most of our objects lie well to the right of the main sequence region , confirming the existence of significant excess emission .
the group iii objects from @xcite all exist in region @[email protected] , @[email protected] .
we use the box defined by the intersection of these boundaries as our criteria for identifying group iii objects in our sample .
the box is plotted in using dashed lines .
there are 9 objects ( ae lep , hd 141569 , hd 203024 , hd 35929 , hd 36408 , hd 50083 , il cep , mwc 610 , and v361 cep ) that lie within the group iii box .
these objects are marked with a @xmath31 in and will be excluded from the analysis presented in the rest of the paper .
this reduces the sample to 78 objects .
[ fig : fig5 ] and @xmath26 colors for our sample .
reddening vectors are plotted as solid lines .
the approximate location of the main sequence is indicated by the dashed dotted line .
the area within the dashed lines contains objects that are most likely evolved haebes without any significant ir excess .
objects with de
reddened colors are plotted as green circles ; objects without reddening corrections are plotted as red squares .
there are 9 objects contained in the box that we exclude from further analysis.,title="fig : " ] in order to exclude objects that are not unambiguously experiencing some sort of outflow or accretion , we do not include pure emission lines ( double or single peaked ) as evidence of mass flows .
while emission lines in many species can form in a wind or accretion flow , they can also form in static geometries such as extended stellar and disk atmospheres or remnant circumstellar envelopes . on the other hand ,
doppler shifted absorption is an unambiguous sign of mass motion .
the ctts comparison statistics that are discussed in are calculated using the same criteria .
thus we only count red and blue shifted absorption below the local continuum ( e.g. the blue shifted absorption seen at h@xmath0 in the bottom panel of ) as indications of mass flows when calculating the final statistics in and . for objects with multiple observations ( @xmath520% of the sample ,
see ) we search for absorption signatures in all of the exposures .
thus the computed statistics include objects that , for example , show blue shifted absorption in a line for one exposure but do not show the same morphology in a separate exposure .
this enables us to account for variability , in a limited way , in at least some objects .
the morphology statistics for the mass flow diagnostics are presented in and a summary of each object s profile classifications is given in .
in the line diagnostic with the highest occurrence of each profile type is highlighted in bold font .
for each line the total number of object profiles examined is give in column 2 .
the number of each type of morphology is listed and the fraction of the total number of profiles is given in parentheses . in column 8 of the total number of line diagnostics examined is given for each object . the number of contaminated profiles for each object is not shown although they are included in the total number of lines in column 8 .
it is immediately obvious from that most of the diagnostics show featureless profiles .
in fact , h@xmath7 , h@xmath0 , and ca ii @xmath128498 are the only lines for which a majority of the sample displays some form of activity .
the lines with the highest incidence of circumstellar features displaying pc and ipc morphologies are h@xmath0 ( 30% ) and he i @xmath125876 ( 14% ) , respectively . the na
i d lines show the highest rate ( 12% ) of pure absorption profiles . after the balmer lines ,
the next best spectral line for observing a pc morphology is fe ii @xmath124924 which displays pc characteristics in 11% of the object profiles .
the last row in displays the final mass flow statistics for our sample .
the uncertainties are the 68% confidence intervals calculated using wilson s score test for large samples @xcite .
these intervals are very close to the values obtained assuming poisson statistics in the distribution of observed incidence . however , the wilson score test results in a weighted average towards the center of the distribution .
we note that the given incidence is not the adjusted fraction given by the score test and thus the confidence intervals may be asymmetric around the value .
the occurrence of both inflow and outflow signatures in our sample are fairly low : only 26% of our objects have at least one profile displaying red shifted absorption ; only 37% have at least one profile displaying blue shifted absorption .
a comparison of our mass flow statistics to previous haebe studies is presented in .
we find similar rates of blue absorption in our sample compared to both ( * ? ? ?
* fm84 ) and ( * ? ? ? * hp92 ) .
both ( * ? ? ?
* v03 ) and ( * ? ? ?
* r96 ) find rates of blue shifted absorption that are significantly lower than ours ( as well as fm84 s and hp92 s ) .
we attribute this large discrepancy to the fact that they only observe h@xmath7 which tends to mask sub
continuum absorption due to strong emission components from multiple sources ( e.g. emission from accretion , stellar winds , nebular background ) .
we note , however , that using solely h@xmath7 in our sample results in a blue shifted absorption fraction of 26%@xmath32 , a value in between the low rates of @xcite and @xcite and our final fraction ( 37% ) .
this may be due to our classification of profiles as p
cygni if the blue feature is clearly absorption and not a second peak , even if the absorption does not extend below the local emission continuum ( e.g. ) .
our results , and those of fm84 and hp92 , show that in order to recover accurate outflow statistics diagnostics other than h@xmath7 should be examined .
the line with the largest occurrence of blue absorption in our sample is h@xmath0 ( 30% ) , although the incidence in h@xmath7 is similar at 26% .
in fact , all objects in our sample that show p cygni type profiles in lines other than h@xmath7 and h@xmath0 also show p
cygni morphologies in either h@xmath7 or h@xmath0 .
thus it appears h@xmath0 is a very strong stand
alone diagnostic of outflows in haebes .
lccccccc h@xmath7 & 78 & 20(.26 ) & 5(.06 ) & 34(*.44 * ) & 17(.22 ) & 0(.00 ) & 2(.03 ) + h@xmath0 & 74 & 22(*.30 * ) & 4(.05 ) & 15(.20 ) & 18(.24 ) & 5(.07 ) & 10(.14 ) + h@xmath13 & 61 & 12(.20 ) & 0(.00 ) & 3(.05 ) & 14(.23 ) & 3(.05 ) & 29(.48 ) +
h@xmath33 & 51 & 5(.10 ) & 0(.00 ) & 1(.02 ) & 8(.16 ) & 5(.10 ) & 32(.63 ) + he i @xmath123889 & 49 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 1(.02 ) & 0(.00 ) & 40(.82 ) + he i @xmath125876 & 78 & 1(.01 ) & 11(*.14 * ) & 1(.01 ) & 11(.14 ) & 1(.01 ) & 53(.68 ) +
he i @xmath126678 & 71 & 1(.01 ) & 1(.01 ) & 0(.00 ) & 2(.03 ) & 1(.01 ) & 66(.93 ) + mg i @xmath125167 & 71 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 0(.00 ) & 0(.00 ) & 69(.97 ) + mg i @xmath125173 & 71 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 3(.04 ) & 0(.00 ) & 68(.96 ) + mg i @xmath125184 & 71 & 1(.01 ) & 0(.00 ) & 0(.00 ) & 5(.07 ) & 0(.00 ) & 65(.92 ) + ca ii @xmath123933 & 57 & 2(.03 ) & 1(.01 ) & 0(.00 ) & 1(.01 ) & 4(.05 ) & 65(.89 ) + ca ii @xmath123969 & 42 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 3(.07 ) & 1(.02 ) & 35(.83 ) + ca ii @xmath124227 & 71 & 1(.01 ) & 0(.00 ) & 0(.00 ) & 0(.00 ) & 0(.00 ) & 70(.99 ) + ca ii @xmath128498 & 27 & 2(.07 ) & 1(.04 ) & 2(.07 ) & 9(.33 ) & 0(.00 ) & 13(.48 ) + ca ii @xmath128662 & 70 & 2(.03 ) & 2(.03 ) & 3(.04 ) & 22(.31 ) & 0(.00 ) & 35(.50 ) + na i @xmath125889 & 78 & 4(.05 ) & 1(.01 ) & 0(.00 ) & 19(.24 ) & 9(*.12 * ) & 45(.58 ) + na i @xmath125896 & 78 & 4(.05 ) & 1(.01 ) & 0(.00 ) & 18(.23 ) & 8(.10 ) & 47(.60 ) + @xmath34n ii@xmath35 $ ] @xmath125755 & 78 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 3(.04 ) & 0(.00 ) & 75(.96 ) + @xmath34n ii@xmath35 $ ] @xmath126548 & 78 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 2(.03 ) & 0(.00 ) & 76(.97 ) + @xmath34n ii@xmath35 $ ] @xmath126583 & 40 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 6(.15 ) & 0(.00 ) & 34(.85 ) + @xmath34o i@xmath35 $ ] @xmath125577 & 78 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 6(.08 ) & 0(.00 ) & 72(.92 ) + @xmath34o i@xmath35 $ ] @xmath126300 & 77 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 43(*.56 * ) & 0(.00 ) & 34(.44 ) + @xmath34o i@xmath35 $ ] @xmath126363 & 64 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 20(.31 ) & 0(.00 ) & 44(.69 ) + @xmath34o iii@xmath35 $ ] @xmath125007 & 71 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 3(.04 ) & 0(.00 ) & 68(.96 ) + fe i @xmath125498 & 78 & 0(.00 ) & 0(.00 ) & 0(.00 ) & 1(.01 ) & 0(.00 ) & 77(.99 ) + fe ii @xmath124924 & 64 & 7(.11 ) & 2(.03 ) & 1(.02 ) & 7(.11 ) & 3(.05 ) & 44(.69 ) + fe ii @xmath124352 & 55 & 3(.05 ) & 0(.00 ) & 0(.00 ) & 3(.05 ) & 0(.00 ) & 49(.89 ) + lccccccccc ab aur & 4 & 0 & 0 & 5 & 1 & 16 & 26 & no & yes + ae lep@xmath36 & 0 & 0 & 0 & 5 & 0 & 7 & 12 & no & no + bd+61 154 & 6 & 0 & 0 & 9 & 0 & 10 & 26 & no & yes + bf ori & 0 & 2 & 1 & 0 & 8 & 15 & 27 & yes & no + bh cep & 1 & 0 & 0 & 0 & 3 & 22 & 26 & no & yes + cq tau & 0 & 0 & 1 & 2 & 0 & 24 & 27 & no & no + dw cma & 5 & 0 & 0 & 7 & 0 & 12 & 25 & no & yes + gsc 0479408227 & 0 & 0 & 0 & 6 & 0 & 20 & 26 & no & no + hd 141569@xmath36 & 0 & 0 & 2 & 1 & 0 & 27 & 30 & no & no + hd 142666&0&1&1&1 & 0 & 25 & 28 & yes & no + hd 163296&1&1&0&3 & 1 & 21 & 27 & yes & yes + hd 169142&0&0&0&2 & 0 & 23 & 25 & no & no + hd 190073&2&0&0&12 & 0 & 14 & 29 & no & yes + hd 203024@xmath36&0&0&0&1 & 0 & 26 & 27 & no & no + hd 244314&2&0&0&7 & 0 & 20 & 29 & no & yes + hd 244604&2&0&0&5 & 0 & 19 & 26 & no & yes + hd 245185&0&0&1&3 & 2 & 20 & 26 & yes & no + hd 249879&2&0&0&4 & 0 & 18 & 24 & no & yes + hd 250550&2&0&0&4 & 0 & 6 & 12 & no & yes + hd 287823&1&0&0&1 & 0 & 24 & 26 & no & yes + hd 290409&0&0&1 & 1 & 0 & 24 & 26 & no & no + hd 290500&0&0&1 & 0 & 0 & 25 & 26 & no & no + hd 290764&0&0&0 & 2 & 0 & 24 & 26 & no & no + hd 290770&1&0&0 & 5 & 0 & 20 & 26 & no & yes + hd 34282&0&0&1&0 & 0 & 25 & 26 & no & no + hd 35187&0&2&0&1 & 0 & 24 & 27 & yes & no + hd 35929@xmath36&0&0&1&1 & 0 & 24 & 26 & no & no + hd 36408@xmath36&1&0&0&5 & 0 & 19 & 25 & no & yes
+ hd 37357&2&0&0&1 & 0 & 23 & 26 & no & yes + hd 37411&0&0&1&1 & 0 & 10 & 12 & no & no + hd 38120&0&1&1&2 & 0 & 8 & 12 & yes & no + hd 50083@xmath36&0&0&3&1 & 0 & 21 & 26 & no & no + hk ori & 0&0&2&3 & 0 & 21 & 26 & no & no + il cep@xmath36 & 0&0&2&0 & 0 & 24 & 26 & no & no + ip per & 0&0&0&4 & 0 & 22 & 26 & no & no + iras 05044 - 0325 & 0 & 0 & 2 & 1 & 0&22 & 25 & no & no + iras 06071 + 2925 & 2 & 0 & 0 & 5 & 0 & 18 & 26 & no & yes + iras 07061 - 0414 & 0 & 0 & 0 & 0 & 0 & 25 & 25 & no & no + iras 17481 - 1415 & 0 & 0 & 0 & 2 & 0 & 18 & 21 & no & no + iras 18306 - 0500 & 0 & 0 & 0 & 0 & 0 & 23 & 23 & no & no + iras 18454 + 0250 & 0 & 0 & 0 & 1 & 0 & 19 & 20 & no & no + lkh@xmath7 134 & 2 & 0 & 0 & 6 & 0 & 18 & 26 & no & yes + lkh@xmath7 208 & 0 & 0 & 1 & 2 & 0 & 24 & 27 & no & no + lkh@xmath7 233 & 2 & 0 & 0 & 4 & 0 & 17 & 23 & no & yes + lkh@xmath7 257 & 0 & 0 & 2 & 0 & 0 & 20 & 22 & no & no + lkh@xmath7 324 & 0 & 0 & 2 & 0 & 0 & 21 & 23 & no & no + lkh@xmath7 339 & 0 & 0 & 0 & 3 & 0 & 19 & 22 & no & no + mwc 1080 & 9 & 0 & 0 & 4 & 0 & 10 & 25 & no & yes + mwc 120 & 0 & 3 & 3 & 7 & 0 & 12 & 26 & yes & no + mwc 137 & 0 & 0 & 0 & 13 & 0 & 10 & 24 & no & no + mwc 147 & 0 & 0 & 3 & 3 & 0 & 16 & 22 & no & no + mwc 300 & 0 & 0 & 2 & 12 & 0 & 12 & 26 & no & no + mwc 361 & 0 & 0 & 3 & 1 & 0 & 21 & 26 & no & no + mwc 480 & 6 & 0 & 1 & 1 & 3 & 13 & 25 & no & yes + mwc 610@xmath36 & 0 & 0 & 2 & 0 & 0 & 27 & 29 & no & no + mwc 614 & 0 & 0 & 0 & 2 & 0 & 24 & 26 & no & no + mwc 758 & 2 & 0 & 0 & 4 & 0 & 20 & 26 & no & yes + mwc 778 & 0 & 0 & 2 & 3 & 0 & 8 & 13 & no & no + mwc 863 & 2 & 0 & 0 & 3 & 0 & 20 & 25 & no & yes + nz ser & 0 & 0 & 0 & 7 & 0 & 13 & 21 & no & no + r mon & 1 & 0 & 1 & 7 & 0 & 13 & 22 & no & yes + rr tau & 0 & 1 & 2 & 3 & 0 & 21 & 27 & yes & no + sv cep & 0 & 0 & 1 & 1 & 0 & 23 & 25 & no & no + t ori & 0 & 1 & 2 & 5 & 0 & 19 & 27 & yes & no + ux ori & 0 & 5 & 0 & 3 & 0 & 15 & 23 & yes & no + uy ori & 0 & 0 & 1 & 2 & 2 & 7 & 12 & no & yes + v1185 tau & 1 & 0 & 0 & 2 & 0 & 24 & 27 & no & yes + v1578 cyg & 7 & 0 & 0 & 2 & 2 & 14 & 26 & no & yes + v1685 cyg & 1 & 1 & 4 & 3 & 0 & 20 & 29 & yes & yes + v1686 cyg & 0 & 0 & 2 & 2 & 0 & 16 & 19 & no & no + v1787 ori & 0 & 0 & 1 & 1 & 0 & 21 & 23 & no & no + v1818 ori & 0 & 0 & 5 & 1 & 2 & 16 & 24 & yes & no + v346 ori & 0 & 1 & 0 & 0 & 1 & 24 & 26 & yes & no + v351 ori & 0 & 2 & 0 & 0 & 2 & 22 & 26 & yes & no + v361 cep@xmath36 & 0 & 0 & 1 & 3 & 0 & 21 & 26 & no & no + v373 cep & 3 & 0 & 0 & 3 & 0 & 15 & 22 & no & yes + v374 cep & 3 & 0 & 1 & 0 & 2 & 22 & 29 & no & yes + v380 ori & 0 & 0 & 0 & 8 & 0 & 5 & 14 & no & no + v586 ori & 0 & 1 & 1 & 6 & 2 & 16 & 26 & yes & no + v590 mon & 0 & 0 & 2 & 4 & 0 & 20 & 26 & no & no + v599 ori & 0 & 1 & 0 & 2 & 0 & 15 & 18 & yes & no + v718 sco & 0 & 1 & 0 & 0 & 1 & 21 & 23 & yes & no + v791 mon & 2 & 0 & 0 & 6 & 0 & 5 & 13 & no & yes + vv ser & 0 & 2 & 3 & 3 & 0 & 17 & 25 & yes & no + ww vul & 0 & 2 & 1 & 3 & 2 & 20 & 28 & yes & no + xy per & 0 & 1 & 1 & 0 & 5 & 18 & 26 & yes & no + z cma & 13 & 0 & 0 & 5 & 1 & 7 & 27 & no & yes + + * total fraction@xmath37 : * & 0.36 & 0.23 & 0.45 & 0.83 & 0.22 & & & * 0.26@xmath38*&*0.37@xmath39 * + lccccc present work & 78 & 20 & 26@xmath40 & 29 & 37@xmath41 + fm84 & 46 & 2 & 4@xmath42 & 16 & 35@xmath43 + hp92 & 32 & 2 & 6@xmath44 & 11 & 34@xmath45 + r96 & 18 & 1 & 6@xmath46 & 2 & 11@xmath47 + v03 & 131 & & & 18 & 15@xmath48 + the difference in red absorption statistics between our study and the others listed in is more pronounced .
we find that 26% of our objects show red shifted absorption in at least one line while the studies of fm84 , hp92 , and r96 show roughly a 46% occurrence .
the discrepancy is easily explained for our study versus those of fm84 and r96 : the occurrence of red shifted absorption at h@xmath7 in our study is @xmath56% , comparable to that found by fm84 and r96 both of which primarily examine h@xmath7 .
fm84 also examine the na i d lines but both their sample and ours show a negligible occurrence of red shifted absorption in those lines .
the discrepancy with hp92 is harder to explain . a large percentage ( 15% ) of our objects
show red shifted absorption at he i @xmath125876 .
the rest of the red shifted absorption detections result mainly from h@xmath7 , h@xmath0 , and ca ii irt lines .
hp92 also examine these lines but they detect red shifted absorption in only 2 out of 32 objects , a detection rate of @xmath56% .
this might be explained by the fact that hp92 only observe 6 objects at he i @xmath125876 and thus , to obtain a detection rate at @xmath125876 similar to that of our sample , they would , on average , detect red shifted absorption in @xmath51 object , which they do .
the non detections in the other lines is more puzzling but may be accounted for by the lower resolution of their observations ( 1185 km s@xmath6 ) .
these medium resolution observations may mask weak red shifted absorption signatures .
we note that the red shifted absorptions observed at h@xmath7 and h@xmath0 in our sample tend to appear as deeper absorptions superimposed on a photospheric absorption core . however , when these lines are carefully compared to a rotationally broadened standard it is clear that the red shifted absorptions are non photospheric and that the profile morphology is inverse p cygni .
it is not clear that hp92 compare their objects to photospheric standards .
if not , this may also contribute to the discrepancy .
by comparison to the studies listed above , our analysis demonstrates the necessity of large sample , high spectral resolution studies of outflow and inflow diagnostics in haebes in order to gather accurate statistics of mass flows around these objects .
the h@xmath0 diagnostic shows the largest percentage of p
cygni profiles , although other lines ( e.g. h@xmath7 , h@xmath13 , fe i 4924 ) also show non negligible p
cygni detection fractions .
the he i @xmath125876 line is by far the most sensitive line for detecting inverse p cygni morphologies .
this is not surprising since the line is also a very efficient diagnostic for detecting mass flows @xcite and the @xmath125876 transition lies immediately above the upper energy level of the @xmath1210830 transition , preceding @xmath1210830 in a recombination cascade sequence .
the na i d lines are also relatively sensitive to outflows in haebes , although the lines more frequently display pure emission profiles . the line profile statistics ( ) for both inflows and outflows from our optical data set are similar to those from the data set , although the fractions are higher in the sample .
we use contingency tests to compare the fractions of red and blue shifted absorption between the two studies .
a contingency test of the measured fraction of a characteristic in two groups gives the probability that the null hypothesis is true , i.e. that the intrinsic fraction is the same in both groups @xcite .
this probability is referred to here as the @xmath49value .
a high @xmath49-value indicates a high probability of having obtained the difference in the measured fractions by chance , i.e. the null hypothesis can not be ruled out ; a low @xmath49-value indicates a low probability that the difference in the measured fractions was obtained accidentally , i.e. the difference is real and the null hypothesis can probably be ruled out . a contingency test comparing the sample to the optical sample shows that neither the blue shifted ( @xmath500.85 ) nor the red shifted ( @[email protected] ) absorption incidences are significantly different .
rcc @xmath51 & 78 & 48 + red shifted abs . &
26%@xmath40 & 27%@xmath52 + blue shifted abs . &
37%@xmath41 & 40%@xmath43 + although the overall morphology statistics are similar between the optical and data , the line is much more efficient at tracing mass flows around haebes than any _ single _ optical diagnostic .
in addition , is generally less contaminated by sources other than nearby circumstellar material due to the high excitation temperatures needed to populate the lower energy level ( @xmath520,000 k ; * ? ? ? * ) .
this is not the case for other popular diagnostics such as h@xmath7 , h@xmath0 , and the na i d lines which can frequently display strong emission signatures from chromospheres , nebular background , and interstellar absorption in the case of na i d. the tendency of the line to remain in its lower energy level until a photon is absorbed accounts for the large opacity needed to generate the blue shifted absorption commonly seen in the data from .
compares the maximum absorption velocities of and h@xmath7 or h@xmath0 for objects that display blue shifted absorption in both and h@xmath7 ( red diamonds ) and h@xmath0 ( blue crosses ) .
this figure shows that traces approximately the same outflow velocities as h@xmath7 and h@xmath0 , although the velocities in h@xmath0 are generally smaller when compared to he i @xmath1210830 .
this could be due to the larger optical depth of and h@xmath7 relative to h@xmath0 , allowing photons to be absorbed up to larger fractions of the terminal wind velocity .
the number of objects with both optical and data that also show blue shifted absorption is small and thus it is not possible to make a strong case for choosing one of the optical diagnostics over based on a velocity comparison .
each line appears to trace similar material in the wind .
thus for future studies concerned with detecting the presence of outflows and accretion around haebes , could serve as a primary diagnostic due to its high efficiency in detecting mass flows .
in addition , it is relatively free of photospheric and interstellar contaminants making the circumstellar nature of the line morphologies unambiguous . in terms of mass flow statistics and outflow velocities
, it appears that optical data does not provide any additional information over that provided by the line .
[ fig : fig6 ] and h@xmath0 for objects with blue shifted absorption at both and either h@xmath7 or h@xmath0 .
the and the balmer lines generally trace the same velocity of outflowing material , although there is significant scatter .
the h@xmath0 line velocities match the velocities less well than the h@xmath7 velocities .
this may be an optical depth effect : the optical depth at h@xmath0 may simply be too low to trace the same wind extent as and h@xmath7 , resulting , in general , in smaller observed velocities at h@xmath0 .
note that the data for the optical lines and the he i @xmath1210830 lines are not simultaneous.,title="fig : " ] to our knowledge , the study performed by ( * ? ? ?
* hereafter ab00 ) is the only current high resolution , moderate sample size investigation of cttss that examines many different spectral line diagnostics .
we take the statistics from to be representative of cttss in general .
examine the following accretion and outflow diagnostics : h@xmath7 , h@xmath0 , h@xmath13 , h@xmath33 , he i @xmath125876 , na i d , ca ii k , ca ii @xmath128662 , ca ii @xmath128498 , fe i @xmath126192 , @xmath125497 , and fe ii @xmath124924 and @xmath124352 .
our study is well suited for comparison with since we examine all of these except for fe i @xmath126192 .
the resolving power of their observations ( @xmath5348,000 ) is also similar to ours increasing the chance that no weak features will be missed in one data set but not in the other .
we will also compare certain profile statistics from our sample to those of @xcite and @xcite , studies which analyzed a smaller number of mass flow diagnostics in similar sample sizes of cttss to that of .
find high fractions of both red shifted absorption ( 12 objects , 40%@xmath45 ) and blue shifted absorption ( 24 objects , 80%@xmath54 ) in at least one line for the cttss in their sample .
our sample displays much lower incidences of both red and blue shifted absorption : 26% and 37% , respectively .
contingency tests comparing our study to the of @xcite are shown in .
the red shifted absorption comparison between our final sample and that of suggests a marginal difference ( @xmath49=0.16 ) in the incidence of red shifted absorption .
the blue shifted absorption statistics are significantly different for both sample comparisons , with @xmath550.001 .
llc red absorption & haebes vs cttss & 0.160 + blue absorption & haebes vs cttss & 9x10@xmath56 + in addition to the total mass flow statistics , there are differences in the absorption statistics in the lines that are common to both our study and that of .
this comparison is presented in and is shown graphically in .
the most common line displaying blue shifted absorption in their data is h@xmath7 followed by h@xmath0 , h@xmath13 , h@xmath33 and the na i d lines .
a similar trend is observed in our data except that h@xmath0 shows a slightly higher incidence of blue shifted absorption than h@xmath7 .
the most common red shifted absorptions features appear at he i @xmath125876 and na i d in their sample .
the highest red shifted absorption occurrence in our data is also at he i @xmath125876 .
however , we observe only 1 object with red shifted absorption at na i d. we find a small incidence of inverse p
cygni profiles in h@xmath7 and h@xmath0 ( 6% and 5% , respectively ) while observe a larger incidence of red shifted absorption components at h@xmath7 ( 23% ) and h@xmath0 ( 23% ) .
find fe ii @xmath124924 in emission in 40% of their sample .
our data show fe ii @xmath124924 emission in only 9% of the sample .
we also observe both blue and red - shifted absorption at fe ii @xmath124352 in a non negligible fraction ( 13% ) of our sample .
do not fit for the absorption components at fe ii @xmath124352 , likely due to strong nearby photospheric features . by comparing the ctts @xmath124352 lines in their sample to that of v410 tau ,
a wtts also from their sample , we searched for red and blue shifted absorption in the line .
we find no profiles that are not consistent with purely photospheric absorption .
thus we estimate the mass flow incidence in this line to be @xmath50 in the ctts sample .
[ fig : fig7 ] the largest incidence differences between our sample and , outside of the balmer lines , occur in the na i d lines , he i @xmath125876 , fe i @xmath125498 , and fe ii @xmath124924 . at he
i @xmath125876 we find a higher occurrence of red shfited absorption ( 15% ) than ( 10% ) .
our observed fraction of he i @xmath125876 red shifted absorption is more similar to that found by @xcite ( 16% ) for a sample of 31 cttss .
we find significantly lower fractions of red shifted absorption in both na i d lines ( 8% ) compared to ( 27% ) .
we find a red shifted absorption fraction of 8% at fe ii @xmath124924 .
do not observe any red shifted absorption at fe ii @xmath124924 .
similar discrepancies are observed in the blue shifted absorption statistics .
at fe i @xmath125498 , we find no evidence of blue shifted absorption in _ any _ objects while report a detected fraction of 60% .
our detected blue shifted absorption fraction at na i d ( @xmath510% ) is also much lower than ( 40% ) .
finally , the blue and red shifted absorption fractions at h@xmath7 , h@xmath0 , h@xmath13 , and h@xmath33 are significantly higher in s sample of cttss than in our haebe sample . although we do not focus on the pure emission profiles , we note that our detected fraction of emission at he i @xmath125876 is low ( 14% ) compared to that of @xcite who detect emission in 100% of their 31 cttss . in
6 we will comment on possible causes for this differing line profile statistic .
the possible physical causes of the absorption statistics detailed above will be outlined in the next section .
llccccccccccccc & @xmath57 & 9 & 8 & 8 & 8 & 3 & 8 & 8 & 4 & 1 & 1 & 0 & 0 & 0 + & % ra & 30 & 31 & 31 & 35 & 10 & 27 & 27 & 13 & 3 & 4 & 0 & 0 & 0 + & @xmath58 & 23 & 14 & 8 & 6 & 0 & 12 & 12 & 0 & 1 & 2 & 18 & 0 & 0 + & % ba & 77 & 54 & 31 & 26 & 0 & 40 & 40 & 0 & 3 & 7 & 60 & 0 & 0 + & & & & & & & & & & & & & & + this study & @xmath57 & 5 & 7 & 3 & 3 & 12 & 6 & 6 & 2 & 1 & 0 & 0 & 5 & 0 + & % ra & 6 & 9 & 5 & 6 & 15 & 8 & 8 & 3 & 4 & 0 & 0 & 8 & 0 + & @xmath58 & 21 & 24 & 12 & 7 & 1 & 8 & 7 & 2 & 2 & 1 & 0 & 7 & 3 + & % ba & 24 & 32 & 20 & 14 & 1 & 10 & 9 & 3 & 8 & 2 & 0 & 11 & 5 +
the optical line statistics presented in 5 indicate that there are differences in the characteristics of , and/or the physical mechanisms producing , the mass flows in cttss vs haebes . below
we discuss possible scenarios capable of producing the trends and differences seen in the measured statistics
. shows clear evidence that haebes , in general , show significantly more mass flow activity in the fe ii lines than ctts indicating an absorbing region that is hotter or is subject to higher amounts of ionizing radition .
fe i @xmath125498 shows a 60% incidence of blue shifted absorption in s sample but is not detected in absorption in any of our haebes .
the reverse is true for the fe ii @xmath124924 and @xmath124352 lines : we see non zero fractions of both lines exhibiting blue shifted absorption while do not detect absorption in either of these lines in any objects .
all of the objects in which we detect blue shifted absorption in fe ii are b type or early a type objects . this behavior was also noted by @xcite concerning the relative strengths of fe i and fe ii emission lines in haebes versus cttss .
the first ionization potential of fe is 7.9 ev which corresponds to a wavelength of 1570 . the ionizing radiation flux for a 10,000 k blackbody , a similar temperature to @xmath59 for an a0 star , is @xmath510@xmath60 times higher at 1570 than for a 3,800 k blackbody , a typical @xmath59 for a ctts of spectral type m0 .
in addition , cttss display large uv excess fluxes due to accretion which can be 10@xmath6110@xmath62 times higher than the stellar photosphere at typical temperatures used to model the excess continuum .
thus the actual flux ratio at 1570 is likely much lower than 10@xmath63 but still favors haebes over cttss .
the larger number of ionizing photons emitted by b and early a stars compared to cttss could account for the observed disparities in the fe ii line profile statistics observed here .
this trend also holds for another low ionization potential metal , na i , which is ionized by photons with energies higher than 5.1 ev , or @xmath12@xmath32431 .
the overall mass flow incidence in our sample is much lower at na i d than in s sample .
however , we do observe na i outflow signatures in a majority of our objects that also display blue shifted absorption at fe ii , indicating that significant amounts of neutral material are simultaneously present in these outflows .
although it is unclear to what extent the temperature of the outflow determines the ionization state of the wind compared to the ionizing flux from the star , na i is likely not present in the regions of the outflow being sampled by fe ii . the observed statistics for ca ii are not significant enough to warrant a comparison to those from .
it appears that the differences in fe i and fe ii and na i morphology statistics between haebes and cttss suggest more highly ionized environments around haebes , likely due to the higher number of ionizing photons emitted by haebes . in we showed that , on average , the maximum red shifted absorption velocities in haebes are smaller fractions of the stellar escape velocity than in cttss ( see figures 5 and 6 from ) .
the comparison in was performed using only the line velocities measured from data . here
we present the same comparison in but incorporating all of the optical diagnostics displaying red shifted absorption for each object . for haebes that show red shifted absorption in more than one diagnostic , their average value is used .
we note that , as in , is identical in format to figure 5 from @xcite in order to facilitate easy comparison with the red shifted absorption velocities presented in that study .
the diagonal lines mark the velocity ratios that correspond to the infall distance from the stellar surface , given in stellar radii , marked above each line .
in other words , the dotted line shows the maximum free fall velocity that can be obtained if infall begins at the specified distance from the star .
we estimate the red shifted absorption velocities of 13 cttss from the published optical profiles of @xcite .
@xcite observed balmer line profiles as well as na i d and he @xmath125876 . most of the red shifted absorption profiles in our sample are seen in the same lines so the comparison should not be subject to lines tracing mass flows with significantly different physical conditions . [
fig : fig8 ] , the maximum velocity obtained if infall begins from the distance labeled next to the line ( in stellar radii ) .
the haebes , on average , show maximum red shifted absorption velocities that are smaller fractions of their stellar escape velocities , indicating that the infalling material originates closer to the stellar surface.,title="fig : " ] [ fig : fig9 ] values ( bin width of 0.0005 ) for ten thousand trials ( bottom panel ) and edfs for the haebe and ctts samples ( top panel ) .
the median value of @xmath64 is marked by the vertical dashed line in the bottom panel . in the top panel ,
the haebe edf is shown in blue ; the ctts edf is shown in red .
a majority of the tests ( 74% ) return @[email protected] indicating a significant difference between the two edfs.,title="fig : " ] in the cttss are plotted as solid red diamonds ; the haebes are represented by the filled blue circles . for comparison
we have overplotted the haebes velocities from as light blue unfilled circles and the ctts velocities from @xcite as unfilled light red diamonds .
objects that are common to both the optical and samples have their velocities , as measured at and in the optical diagnostics , connected by a dotted line and @xmath1 estimates from the studies that the spectral line velocities are taken from . ] .
the mean velocity ratio ( @xmath65 ) for the haebe sample is 0.40 ; for the cttss it is 0.78 . a two sided ks
test performed on the sample returns a @xmath49value of 2.2x10@xmath66 indicating a significant difference between the two populations . in order to test the effect of uncertainties in determining @xmath67 on the ks statistic
, we generated ten thousand random velocity distributions and performed a ks test on each one .
we follow the same procedure for assigning uncertainties to the stellar mass and radius ( 50% and 25% , respectively ) as performed in . in addition , we let the measured red shifted absorption velocity vary by 50% .
this provides some accounting of accretion variability .
the result of the ks
tests is shown in the bottom panel of where the median @xmath49value is seen to be @xmath50.014 .
the emipirical distribution function ( edf ) for each sample is shown in the upper panel .
for each of the trials each object is assigned a random value within the limits of the uncertainty in the object s escape velocity and red shifted absoprtion velocity .
the low @xmath49values suggest a real difference in the distributions which in turn indicates that accretion flows in haebes begin deeper in the system s gravitational potential , i.e. closer to the star .
this supports the similar result found in , although the @xmath49-values are , on average , higher in the optical case due to the red - shifted absorption velocity being allowed to vary in the simulation . and
provide additional support for our suggestion in that haebes have smaller magnetospheres than cttss .
the evidence in the optical is perhaps even more convincing due to the larger number of red shifted absorption lines used in the comparison , although there is a large spread in the measured line absorption velocities for some individual objects .
however , the lack of large @xmath49values in our ks test simulation indicates that the velocity ratio difference between haebes and cttss is independent of the particular line velocity chosen for the comparison .
the statistical tests also support the idea that the red shifted absorption in haebes originates deeper within the gravitational potential of the central star .
if the relative distances are similar at which the magnetospheres in haebes and cttss are dominating the kinematics of the accretion flow , the velocity ratios of the haebe and ctts samples should be comparable .
since they are not , there must be some difference in the formation mechanisms of the absorption signatures .
we attribute this difference to smaller magnetospheres in haebes compared to cttss .
we note that smaller corotation radii , and thus smaller magnetospheres , are suggested for most haebes compared to cttss due to their large @xmath8sin@xmath9 values ( see 2 of ; * ? ? ?
thus , this result is expected from magnetospheric accretion physics .
in we demonstrated the lack of red shifted absorption at in hbe stars ( figure 7 of ) . this same trend holds for our larger optical sample .
this is shown in , which plots the difference between the blue ( @xmath68 ) and red ( @xmath69 ) line equivalent widths versus @xmath59 , for the objects with at least one profile displaying evidence of blue or red shifted absorption .
if an object shows more than one profile in absorption , we average the equivalent width difference .
the equivalent width difference is essentially a measure of profile type : red shifted absorption results in a negative equivalent width difference while blue shifted absorption results in a positive equivalent width difference .
objects in each quadrant are color coded for clarity .
there is an absence of objects in the upper - left quadrant .
although v1818 ori shows evidence of red shifted absorption in the na i d lines , we do not calculate an equivalent width for either line due to the strong interstellar absorption features superimposed on the line .
thus there are only 3 hbe objects with red shifted absorption shown in and they all lie very close to the vertical dashed line .
the ipc and red absorption he i @xmath125876 profiles generally have small equivalent width differences ( 0.0 @xmath20@xmath70-@xmath71@xmath20 - 2.0 ) and comprise most of the red shifted absorption objects ( red crosses and magenta right facing triangles ) .
[ fig : fig10 ] vs blue equivalent width minus red equivalent width for all of the objects in our sample that have absorption profiles .
objects with more than one line in a quadrant are shown using the average of the equivalent width differences for those lines .
colors are used for clarity in distinguishing quadrants .
the trend found for the data from holds for the optical sample : there is a lack of red shifted absorption profiles in hbe stars .
the strongest blue shifted absorption profiles are found in the hbe group.,title="fig : " ] this result is also born out in the overall profile statistics : 15 of the 19 red shifted absorption objects are hae stars ; 4 are hbe stars , with 2 of those having spectral types of b9 .
there are 45 hae stars in the final sample and 33 hbe stars .
a contingency test comparing hae and hbe red shifted absorption returns @xmath49=0.04 providing support for a real difference in the morphology statistics .
a test comparing the blue shifted absorption reveals no difference between hae and hbe objects ( @xmath49=0.48 ) .
a similar trend is seen in the data but the larger optical sample has added significance to the red shifted absorption comparison .
the similarity in the optical and hae versus hbe comparisons suggests that the conclusions reached in are not altered by the analysis presented here .
hae stars exhibit red shifted absorption more frequently than hbe stars indicating that mass infall , most likely magnetically channeled , is more common in hae stars .
outflows seem to be present in similar fractions of hae and hbe objects suggesting that the wind mechanism is equally as efficient in both groups of objects .
in we noted the lack of simultaneous blue and red shifted absorption in the profiles . to clarify , simultaneous absorption signatures in the current , optical context refer to detected absorption in _ any _ diagnostics from the same exposure , or exposures taken very close in time to one another .
this includes simultaneous red and blue shifted detections in the same line , such as that seen in for the cttss from @xcite , or detections in two different lines as seen in @xcite .
this lack of simultaneous absorption contrasted sharply with the large detected fraction of simultaneous absorption signatures in cttss presented by @xcite .
the optical data continues this trend : we observe only 1 object ( hd 163296 ) that shows clear red and blue shifted absorptions signatures in at least one line from the same exposure .
based on the published centroid velocities of the detected absorption features from , we count 17 out of 30 cttss , or 57% , showing red and blue shifted absorption in at least one line from nearly simultaneous exposures .
the lack of simultaneous absorption features observed in our optical sample supports the arguments presented in , namely that accretion driven outflows in haebes may be more difficult to produce .
the strong optical outflow signatures observed in both hae and hbe objects that do not also exhibit red shifted absorption , combined with the conclusion from that the pc morphologies are likely accretion driven , suggests that accretion in some haebes may proceed through a non
magnetic mechanism such as a boundary layer .
this would explain the relative lack of simultaneous red and blue shifted absorption signatures in haebes compared to cttss .
it is still unclear whether or not the accretion flows detected in the red shifted absorption objects can drive outflows .
outflow signatures in these objects are not observed ( except for hd 163296 ) suggesting that if the accretion is related to driving outflows , the outflow signatures are too weak to be detected or are launched in a geometry that requires a specific viewing angle to observe ( e.g. a highly collimated jet emerging from the polar region ) .
this seems unlikely since these simultaneous signatures are almost entirely absent from our sample , suggesting that they are present in very few objects .
indeed , jets have been detected in a relatively fewer number of haebes ( e.g. * ? ? ?
* ; * ? ? ?
* ) compared to cttss hinting that they do not form as efficiently in haebes .
one potential explanation is that , in general , the accreting material as traced by red shifted absorption is not energetic enough to provide the energy necessary to drive an outflow .
support for this idea is seen in from this study and figure 5 from where the red shifted absorption velocities of haebes are found to be smaller fractions of their stellar escape velocities .
we also find support for this scenario in figure 7 from which demonstrates the relative lack of objects with low accretion rates that also exhibit blue shifted absorption profiles .
we note that this distinction is not clear in our optical data , i.e. the separation of profile type with accretion rate is not distinct .
we have fewer estimates of @xmath18 in hand , however , for the red shifted absorption objects from the optical sample and thus the comparison is less significant .
this could be improved and further tested by obtaining simultaneous line profile morphologies and accretion rate estimates for a large sample of haebes .
the idea of accretion powered stellar winds was first proposed for cttss by @xcite .
they suggest that a portion of the deposited accretion energy will heat the corona ( via conduction or magnetosonic wave dissipation ) and assist in launching stellar winds .
while it is unclear whether or not this process is occurring in cttss , magnetically channeled accretion , involving smaller magnetospheres in haebes compared to cttss , will provide a smaller fraction of the necessary energy for launching an outflow from the stellar surface of haebes than from the surface of cttss . since the outflows observed in cttss are generally accepted to be driven largely by accretion ( e.g. * ? ? ?
* ; * ? ? ?
* ) , the amount of energy deposited by the accretion flow onto the stellar surface could strongly influence the strength of the outflow . for haebes ,
if we imagine that all of the kinetic energy of the accretion flow is converted into kinetic energy in a wind driven from very near the stellar surface ( i.e. a stellar wind ) , these outflows will be incapable of escaping the gravitational potential of the star due to the small ( @xmath50.5 ; ) fractions of the stellar escape velocity that the accretion flows attain during their infall .
obviously this is an ideal scenario , but the energy budget argument holds if a large portion of the energy in the wind is provided by the accretion flow , which it seems to be in cttss . to summarize , the lower infall velocities relative to the stellar escape velocities in haebes may indicate that these accretion flows can not efficiently drive mass outflows from the stellar surface
. if magnetically controlled mass infall in haebes is incapable of driving outflows , another form of accretion may be partly responsible for driving the outflows from objects with large @xmath18 .
we suggested in that boundary layer accretion in haebes could provide the necessary energy , although it is unclear exactly how the energy is tranferred to the outflow .
one possible scenario discussed in for launching outflows from the boundary layer is the @xcite x celerator mechanism .
the high @xmath8sin@xmath9 values of haebes are ideal for producing this type of outflow .
better @xmath18 estimates for a larger number of haebes will help elucidate the relationship between profile type and the role of the accreting material in producing the outflow signatures .
@xcite examine the he i @xmath125876 in a sample of 31 cttss .
they find an emission component ( broad , narrow , or both ) in all of the objects in their sample .
they interpret the broad component emission as arising in a hot wind as well as in the accretion flow onto the star .
the narrow emission component is interpreted as forming in the post shock gas at the base of an accretion column at the surface of the star .
our sample contains 13 haebes that display emission profiles at he i @xmath125876 .
none of these objects show the narrow component that is commonly seen in cttss , as well as wttss
. this may be due to the much higher average @xmath8sin@xmath9 values of haebes compared to cttss .
however , 11 of the 13 objects show broad emission profiles similar in width to those observed by @xcite and 8 objects show unambiguous blue shifted absorption in a separate but simultaneous diagnostic . of these 8 objects ,
five show unambiguous evidence of mass outflow at he i @xmath1210830 , although these observations are not simultaneous to the he i @xmath125876 observations .
due to the prevalence of blue shifted absorption observed in the he i @xmath125876 emission objects , our data supports the suggestion by @xcite that the broad he i @xmath125876 emission component may partly arise in a hot wind near the star .
we have presented a large sample census of mass flows around herbig ae / be stars using a wide variety of mass flow diagnostics and determined the fraction of objects that display unambiguous ( i.e. red and blue shifted absorption ) mass flow signatures .
we find that the na i d interstellar absorption line centers are good approximations , to within 8.1@xmath72 km s@xmath6 , to the center - of - mass radial velocity in most haebe systems , confirming the conclusion reached by @xcite .
of the 78 haebes included in the final analysis , 26%@xmath40 show direct signs of mass infall and 37%@xmath41 show direct evidence of mass outflow .
these percentages are very similar to those found in using the he i @xmath1210830 line .
overall , the optical data reinforce the conclusions outlined in .
more specifically , we find lower rates of both red and blue shifted absorption for haebes than is observed for cttss , suggesting that , in general , the innermost environments of haebes are not simply scaled analogues of cttss . similar to the data ,
the haebe optical lines show maximum red shifted absorption velocities that are smaller fractions of the stellar escape velocities than is found for cttss .
this adds support to the argument presented in concerning the smaller size of haebe magnetospheres compared to those of cttss .
this has implications for the ability of haebe accretion flows to power strong winds : less energetic funnel flows provide less of the required energy for launching outflows .
this is supported by the lack of simultaneous blue and red shifted absorption signatures seen in both the optical sample and the sample from .
this contrasts with cttss for which the accretion process is believed to be the main energy source for outflows .
we also find a significant difference between the occurrence of red shifted absorption in hae stars compared to hbe stars ; both groups appear to exhibit outflow signatures at a similar rate .
this suggests that hbe stars accrete material from their disks differently than hae stars , perhaps through a boundary layer instead of along magnetic field lines .
the prominence of blue shifted absorption in accreting objects in both subgroups suggests that the outflows in haebes are accretion driven .
however , the lack of simultaneous red and blue shifted absorption in all but one haebe ( hd 163296 ) indicates that whatever geometry is producing the absorption signature is not capable of producing the opposite absorption signature . in other words ,
boundary layer accretion , which should not produce red shifted absorption , may be capable of driving strong outflows which results in observed blue shifted absorption ; magnetically controlled accretion from a small magnetosphere produces red shifted absorption but is perhaps not energetic enough to drive outflows from the stellar surface , resulting in the lack of simultaneous red shifted absorption in these profiles . due to the observed statistics
, this suggests a general transition from magnetically controlled accretion in hae stars to boundary layer accretion in hbe stars .
we thank the mcdonald observatory staff in west texas for their hospitality during the observing runs for this research .
p. w. c. especially thanks david doss for his help , without which the collection of some of this data would not have been possible .
p. w. c. would also like to acknowledge pat hartigan for his comments and discussion throughout this project .
we thank the referee for their helpful comments which improved the quality of this manuscript .
this research has made use of the simbad database and the nasa astrophysics data system .
we acknowledge partial support for this research from the nsf through grant 1212122 , from nasa astrophysics data analysis program through grant nnx13af09 g , and from space telescope science institute through grant hst-60 - 12996.001 , all made to rice university .
acke , b. , & van den ancker , m. e. 2004 , a&a , 426 , 151 acke , b. , van den ancker , m. e. , & dullemond , c. p. 2005 , a&a , 436 , 209 alecian , e. , et al .
2013 , mnras , 429 , 1001 alencar , s. h. p. , & basri , g. 2000 , aj , 119 , 1881 alonso albi , t. , et al .
2009 , a&a , 117 , 136 bagnoli , t. , et al .
2010 , apjl , 724 , l5 batalha , c. c. , stout
batalha , n. m. , basri , g. , & terra , m. a. o. 1996 , apjss , 103 , 211 beristain , g. , edwards , s. , & kwan , j. 2001 , apj , 551 , 1037 350 bhm , t. , & catala , c. 1994 , a&a , 290 , 167 bhm , t. , & catala , c. 1995 , a&a , 301 , 155 bossier , j. , et al . 2011 , a&a , 531 , a50 carmona , a. , van den ancker , m. e. , audard , m. , henning , th . , setiawan , j. , & rodmann , j. 2010 , a&a , 517 , a67 caratti o garatti , a. , et al .
2012 , a&a , 538 a64 catala , c. , bhm , t. , donati , j.f . , & semel , m. 1993 , a&a , 278 , 187 cauley , p. w. , & johns - krull , c. m. 2014 , apj , 797 , 112 chen , l. , et al .
2012 , a&a , 541 , a104 corporon , p. , & lagrange , a.m . 1999 ,
a&ass , 136 , 429 dent , w. r. f. , greaves , j. s. , & coulson , i. m. 2005 , mnras , 359 , 663 donehew , b. , & brittain , s. 2011 , aj , 141 , 46 dupree , a. k. , brickhouse , n. s. , smith , g. h. , & strader , j. 2005 , apj , 625 , l131 edwards , s. , hartigan , p. , ghandour , l. , & andrulis , c. 1994 , aj , 103 , 3 edwards , s. , fischer , w. , hillenbrand , l. , & kwan , j. 2006 , apj , 646 , 319 eisner , j. a. , lane , b. f. , hillenbrand , l. a. , akeson , r. l. , & sargent , a. i. 2004 , apj , 613 , 1049 eisner , j. a. , graham , j. r. , akeson , r. l. , & najita , j. 2009 , apj , 692 , 309 ellerbroek , l. e. , et al .
2014 , a&a , 563 , a87 feigelson , e. d. , & babu , g. j. 2012 , modern statistical methods for astronomy ( new york , ny : cup ) fekel , f. c. 2003 , pasp , 115 , 807 finkenzeller , u. , & jankovics , i. 1984 , a&ass , 57 , 285 finkenzeller , u. , & mundt , r. 1984 , a&ass , 55 , 109 fischer , w. , kwan , j. , edwards , s. , & hillenbrand , l. 2008 , apj , 687 , 1117 fuente , a. , martn
pintado , j. , bachiller , r. , rodrguez franco , & palla , f. 2002 , a&a , 387 , 977 fukagawa , m. , tamura , m. , itoh , y. , hayashi , s. s. , & oasa , y. 2003 , apj , 590 , l49 garcia lopez , r. , natta , a. , testi , l. , & habart , e. 2006 , a&a , 459,837 grady , c. a. , et al .
2004 , apj , 608 , 809 grady , c. a. , et al .
2010 , apj , 719 , 1565 guilloteau , s. , dutrey , a. , pitu , v. , & boehler , y. 2011 , a&a , 529 , a105 guilloteau , s. , di folco , e. , dutrey , a. , simon , m. , grosso , n. , & pitu , v. 2013 , a&a , 549 , a92 halbedel , e. h. 1996 , pasp , 108 , 833 hamann , f. , & persson , s. e. 1992 , apjss , 82 , 285 hartigan , p. , edwards , s. , & ghandour , l. apj , 452 , 736 hernndez , j. , calvet , n. , brice@xmath73o , c. , hartmann , l. , & berlind , p. 2004 , apj , 127 , 1682 hillenbrand , l. a. , strom , s. e. , vrba , f. j. , & keene , j. 1992 , apj , 397 , 613 johns krull , c. m. 2007 , apj , 664 , 975 juhsz , a. , et al .
2010 , apj , 721 , 431 knigl , a. 1991 , apj , 370 , l39 kraus , s. , et al .
2008 , a&a , 489 , 1157 kraus , s. , et al .
2013 , apj , 768 , 80 lee , h. t. , & chen , w. p. 2009 ,
apj , 694 , 1423 liu , w. m. , et al .
2007 , apj , 658 , 1164 liu , t. , zhang , h. , wu , y. , sheng - li , q. , & miller , m. 2011 , apj , 734 , 22 malfait , k. , bogaert , e. , & waelkens , c. 1998 , a&a , 331 , 211 manoj , p. , bhatt , h. c. , maheswar , g. , & muneer , s. 2006 , apj , 653 , 657 marias , n. , telesco , c. m. , fisher , r. s. , & packham , c. , 2011 , apj , 737 , 57 matt , s. , & pudritz , r. e. 2005 , apj , 632 , l135 mccarthy , j. k. , sandiford , brendan , a. , boyd , d. , & booth , j. 1993 , pasp , 105 , 881 meeus , g. , waters , l. b. f. m. , bouwman , j. , van den ancker , m. e. , waelkens , c. , & malfait , k. 2001 , a&a , 365 , 476 mendiguta , i. , et al .
2011 , a&a , 529 , a34 mendiguta , i. , et al .
2011 , a&a , 535 , a99 millan gabet , r. , schloerb , f. p. , & traub , w. a. 2001 , apj , 546 , 358 muzerolle , j. , calvet , n. , & hartmann , l. 2001 , apj , 550 , 944 muzerolle , j. , dalessio , p. , calvet , n. , & hartmann , l. 2004 , apj , 617 natta , a. , testi , l. , neri , r. , shepherd , d. s. , & wilner , d. j. 2004 , a&a , 416 , 179 okamoto , y. k. , et al .
2009 , apj , 706 , 665 oudmaijer , r. d. , van der veen , w. e. c. j. , waters , l. b. f. m. , trams , n. r. , waelkens , c. , & engelsman , e. 1992 , a&ass , 96 , 625 osorio , m. , et al .
2014 , apjl , 791 , l36 perrin , m. d. , duch@xmath74ne , g. , kalas , p. , & graham , j. r. 2006 , apj , 645 , 1272 perrin , m. d. , vacca , w. d. , & graham , j. r. 2009 , aj , 137 , 4468 ragland , s. , et al .
2012 , apj , 746 , 126 reed , b. c. 2003 , aj , 125 , 2531 reipurth , b. , pedrosa , a. , & lago , m. t. v. t. 1996 , a&ass , 120 , 229 romanova , m. m. , ustyugova , g. v. , koldoba , a. v. , & lovelace , r. v. e. 2009 , mnras , 399 , 1802 sartori , m. j. , gregorio
hetem , j. , rodrigues , c. v. , hetem jr . ,
a. , & batalha , c. 2010 , aj , 139 , 27 schegerer , a. a. , ratzka , t. , schuller , p. a. , wolf , s. , mosoni , l. , & leinert , ch .
2013 , a&a , 555 , a103
sch@xmath75tz , o. , meeus , g. , & sterzik , m. f. 2005 , a&a , 431 , 165 shu , f. h. , lizano , s. , ruden , s. p. , & najita , j. 1988 , apj , 328 , l19 shu , f. , et al .
1994 , apj , 429 , 781 tang , y .- w . , guilloteau , s. , pitu , v. , dutrey , a. , ohashi , n. , & ho , p. t. p. 2012 , a&a , 547 , a84 th , p. s. , de winter , d. , & perez , m. r. 1994 , a&ass , 104,315 thi , w.f . ,
2014 , a&a , 561 , a50 tull , r. g. , macqueen , p. j. , sneden , c. , & lambert , d. l. 1995 , pasp , 107 , 251 verhoeff , a. p. , et al .
2012 , a&a , 538 , a101 vieira , s. l. a. , et al .
2003 , aj , 126 , 2971 vink , j. s. , drew , j. e. , harries , t. j. , & oudmaijer , r. d. 2002 , mnras , 337 , 356 wade , g. a. , bagnulo , s. , drouin , d. , landstreet , j. d. , & monin , d. 2007 , mnras , 376 , 1145 wilson , e. b. 1927 , _ journal of the american statistical association _
below are the extracted profiles for a piece of the sample . the profiles for the entire sample will be available with the published version .
the format of each figure is the same as and . | we examine a broad range of mass flow diagnostics in a large sample of herbig ae / be stars ( haebes ) using high resolution optical spectra .
the h@xmath0 and he i 5876 lines show the highest incidence of p cygni ( 30% ) and inverse p cygni ( 14% ) morphologies , respectively .
the fe ii 4924 line also shows a large incidence of p cygni profiles ( 11% ) .
we find support for many of the conclusions reached in a study based on the analysis of the line in a large sample of haebes .
namely , haebes exhibit smaller fractions of both blue shifted absorption ( i.e. mass outflow ) and red shifted absorption ( i.e. mass infall or accretion ) than their lower mass cousins , the classical t tauri stars ( cttss ) .
in particular , the optical data supports the conclusion that haebes displaying red shifted absorption , in general , show maximum red shifted absorption velocities that are smaller fractions of their stellar escape velocities than is found for cttss .
this suggests that haebe accretion flows are originating deeper in the gravitational potentials of their stars than in ctts systems .
in addition , we find a lack of inner disk wind signatures in the blue shifted absorption objects ; only stellar wind signatures are clearly observed .
these findings , along with the lack of detected magnetic fields around haebes , support the idea that large magnetospheres are not prevalent around haebes and that accretion flows are instead mediated by significantly smaller magnetospheres with relatively smaller truncation radii ( e.g. 12 @xmath1 ) .
red shifted absorption is much more common around herbig ae stars than be stars , suggesting that herbig be stars may accrete via a boundary layer rather than along magnetic field lines . |
interstellar magnetic field plays significant roles on the dynamics of molecular clouds .
magnetic field supports molecular cloud against gravitational collapse if it is strong enough .
it is well known that the mass to flux ratio , i.e. , the amount of gas contained in a magnetic flux tube , is the key parameter for the magnetic support .
@xcite pointed out from the viral analysis that magnetic force reduces gravity by a constant factor independent of the size of the cloud .
they evaluated the critical magnetic flux to support a cloud of mass , @xmath1 , to be @xmath4 @xcite .
the virial theorem provides the correct scaling law but not quantitatively correct value for the critical mass .
one needs equilibrium models to evaluate the critical value .
@xcite for the first time obtained a magnetohydrostatic equilibrium for a disk like cloud supported in part by magnetic field .
he evaluated magnetic pressure and tension as well as self - gravity and gas pressure consistently .
he confirmed that the mass to flux ratio is one of the key parameters .
he also noticed that the equilibrium depends also on ratio of the magnetic pressure to the gas pressure and the flux function , i.e. , the amount of gas contained in each magnetic flux .
@xcite obtained a fitting formula for the critical equilibrium from their numerical results .
it reads @xmath5 .
finite electrical conductivity changes the flux function through ambipolar diffusion and induces quasi - static contraction . when the magnetic field is weaker than the critical one , the gravitational collapse continues to form stars .
thus magnetohydrostatic equilibria help our understanding of the gravitational collapse to form stars .
magnetohydrostatic equilibria depend on the configuration .
we need to consider filamentary clouds in magnetohydrostatic equilibria since observed molecular clouds are often filamentary and associated with either longitudinal or perpendicular magnetic field .
the magnetic field is parallel to elongation of the molecular cloud in the ophiuchus star forming region while it is perpendicular in the taurus ( see , e.g. , * ? ? ?
? * ; * ? ? ?
* ) and in the musca @xcite .
@xcite and @xcite obtained equilibrium model having the density distribution , @xmath6 for an isothermal filamentary cloud in the cylindrical coordinate , ( @xmath7 ) , where @xmath8 and @xmath3 denote the isothermal sound speed and gravitational constant , respectively .
interestingly , the line density defined by @xmath9 does not depend on the filament width , @xmath10 .
thus , this model is neutrally stable against radial contraction .
@xcite showed that filamentary clouds having a larger line density can be sustained by purely longitudinal or partially helical magnetic field .
these models are unstable against fragmentation , i.e. , sinusoidal perturbation in the @xmath11-direction ( see , e.g. , * ? ? ?
the fragmentation and its further evolution were studied extensively by @xcite , @xcite and others . on the contrary , filamentary clouds permeated by magnetic field perpendicular to the axis
have been studied little .
its structure was obtained only recently by @xcite referred to paper i in the following .
this is mainly because the structure is obtained only by extensive numerical computation solving the poisson and grad - shafranov equations simultaneously .
the latter is highly nonlinear since it describes force balance in the directions parallel and perpendicular to the magnetic field .
it requires high spatial resolution especially when the filamentary cloud is condensed .
paper i showed that the maximum line density increases in proportion to the magnetic flux . in this paper
we show analytical models which reproduce the main features of the 2d models obtained in paper i. when the line density is higher than @xmath12 and thus supported mainly by magnetic field , the filamentary cloud is flattened like italian pasta fettucine .
the magnetic force is dominated by the magnetic tension .
then we can use the thin disk approximation to analyze the structure and stability of filamentary cloud permeated by perpendicular magnetic field .
this paper is organized as follows . in 2 we derive an analytical model for an isolated filamentary cloud .
the mass to flux ratio is uniform in the model . in
3 we use the fourier series for filamentary clouds arranged periodically .
it is shown that the solution approaches to that obtained in 2 when the filament thickness is much smaller than the distance to the neighboring filaments .
it is also shown that the ratio of the thickness to the distance can be regarded as the model parameter specifying the quasi - static evolution . in
4 we confirm that our 1d models reproduce the main features of the 2d model obtained in paper i. in 5 the model obtained in 2 is proved to be unstable against fragmentation . in 6 we discuss the stability of the models obtained in 3 .
in this paper , we consider an isothermal flattened filamentary cloud having infinite length and infinitesimal thickness .
this simplification is justified for observed filamentary clouds since their lengths are much larger than the widths and the magnetic fields are often perpendicular to cloud elongation on the sky . it is well known that self - gravitating clouds are flattened in the direction perpendicular to the global magnetic field . in the following we use the cartesian coordinates in which the cloud is confined in the plane of @xmath13 and elongated in the @xmath14-direction .
both the gas and magnetic fields are uniform in the @xmath14-direction . from the above assumptions we can express the gas density distribution as @xmath15 where @xmath16 and @xmath17 denote the surface density and the dirac s delta function , respectively .
the poisson equation reduces to @xmath18 we obtain the boundary condition , @xmath19 by integrating equation ( [ poisson ] ) in the vertical direction over the infinitesimal interval around @xmath13 ( see , e.g. , * ? ? ? * for the derivation ) .
the symbol , @xmath20 , denotes an infinitesimally small positive quantity .
the gravity is given by @xmath21 we assume for simplicity that the magnetic field has only the @xmath22- and @xmath11-components in the equilibrium .
the condition for magneto - hydrostatic equilibrium requires that the magnetic field should be force free and hence current free outside the disk , @xmath23 since both the gravity and pressure thereof are negligibly small .
the symbols , @xmath24 and @xmath25 , denote the speed of light and the electric surface current running in the @xmath14-direction inside the cloud per unit length in the @xmath22-direction , respectively .
the magnetic field is kinked on the disk , @xmath26 note that the gravity and magnetic field can be perfectly aligned outside the cloud .
both of them are divergence and rotation free , i.e. , @xmath27 , @xmath28 , @xmath29 , and @xmath30 , outside the cloud , @xmath31 . in the following we focus on the case that the magnetic field is proportional to the gravity , @xmath32 in @xmath33 and @xmath34 in @xmath35 , where @xmath36 denotes a non - dimensional constant .
this assumption is equivalent to the constant mass to flux ratio ( isopedic in the terminology of * ? ? ?
* ) , @xmath37 thanks to equation ( [ g - boundary ] ) .
hence , the cloud is subcritical ( supercritical ) when @xmath38 ( @xmath39 ) .
when the mass to flux ratio is constant , the magnetic tension working on the cloud is proportional to the gravity , @xmath40 in the cloud .
equations ( [ magf ] ) simplifies the equation for magnetohydrostatic balance in the @xmath22-direction , @xmath41 where @xmath8 denotes the isothermal sound speed .
substituting equation ( [ magf ] ) into equation ( [ mhdvertical ] ) integrated over the infinitesimal interval around @xmath13 , we obtain @xmath42 the magnetic field reduces the effective gravity by a factor @xmath43 .
see @xcite for the validity of this approximation .
equation ( [ equilibrium ] ) indicates that the cloud is in equilibrium only when @xmath44 ( supercritical ) . in the following we consider the lorentz profile , @xmath45 as a model for the flattened filamentary cloud .
the line density and fwhm of this model cloud are @xmath46 and @xmath47 , respectively . using the method of images we obtain the corresponding gravity , @xmath48
the pressure force is given by @xmath49 ^2 } .
\label{pressure}\end{aligned}\ ] ] we obtain the condition for the magnetohydrostatic equilibrium , @xmath50 by substituting equations ( [ surface ] ) through ( [ pressure ] ) to equation ( [ equilibrium ] ) . equation ( [ filament ] ) is rewritten as @xmath51 by the help of equation ( [ masstof ] ) .
when equation ( [ lined ] ) holds , the model cloud settles in an equilibrium for any @xmath52 .
in other words , the width of the cloud is not specified by @xmath46 or @xmath2 .
this character is common for isothermal filamentary cloud in equilibrium .
the magnetic field is expressed as @xmath53 } \left ( \begin{array}{c } \displaystyle \frac{zx}{|z| } \\ \displaystyle |z| + a \end{array } \right ) .
\end{aligned}\ ] ] the magnetic field lines are straight and kinked on the plane of the flattened cloud .
they look as if they emanate radially from the lines of @xmath54 .
the strength decreases inversely proportional to the distance from the lines .
we can estimate the density by assuming the hydrostatic balance in the vertical direction , @xmath55 \right\ } , \\
\phi & = & g \lambda \ln \left ( x ^2 + z ^2 + a ^2 \right ) .\end{aligned}\ ] ] here the gravitational potential , @xmath56 , is evaluated by the method of image and the magnetic force is neglected in the hydrostatic balance .
the magnetic force , @xmath57 , vanishes outside the disk since the electric current is confined in the disk .
the magnetic force is perpendicular to the field inside the disk .
hence it compresses the gas disk ( see * ? ? ?
* for the evaluation of the magnetic force acting on the vertical structure ) .
the condition , @xmath58 gives us @xmath59 where @xmath60 denotes the gamma function .
accordingly we have @xmath61 figure [ rho - xz ] denotes the density distribution and the magnetic field for @xmath46 = 5 , 10 , and 20 @xmath62 . here the abscissa and ordinate are denoted in unit of @xmath52 while the density is shown in unit of @xmath63 .
the cloud is more flattened for a larger @xmath46 .
in this section we relax the uniform mass to flux ratio , @xmath64 , assumed in the previous section .
the mass to flux ratio should increase at the cloud center through the ambipolar diffusion if the cloud is supported in part by the magnetic field against gravity . on the contrary , the mass to flux ratio decreases outside the cloud . in order to take account of magnetic field outside clouds
, we consider the situation in which filamentary clouds are arranged periodically on a plane . again
each filamentary cloud is assumed to be geometrically thin
. then the surface density is given by the fourier series , @xmath65 where @xmath66 denotes the spacing between the filaments .
the 0th fourier series coefficient , @xmath67 , denotes line density per each filament , @xmath46 .
we obtain the gravitational potential , @xmath68 by solving the poisson equation .
the gravity is expressed as @xmath69 similarly the force free magnetic field is expressed as @xmath70 the 0th fourier series coefficient , @xmath71 , denotes the magnetic flux permeating each filament , @xmath2 in the model for an isolated filamentary cloud . the condition for the force balance ,
@xmath72 is rewritten as @xmath73 this equation can be solved successively if once @xmath74 and @xmath71 are given .
the fourier series coefficient , @xmath75 , is given by @xmath76 if @xmath77 , @xmath78 , @xmath71 , @xmath79 are known .
in other words we can solve equation ( [ balance ] ) if the mass distribution and the magnetic flux are given . in the following
we consider the case , @xmath80 where @xmath81 denotes the ratio of the filament width to the interval
. then equation ( [ balance2 ] ) reduces to @xmath82 the right hand side of this equation indicates that the gravity overcomes the pressure force when @xmath83 .
after some algebra we find a solution , @xmath84 the mass to flux ratio is constant in this solution . the surface density and magnetic fields are expressed as @xmath85 } , \label{period1 } \\ b _ x ( x , z=+\varepsilon ) & = & \frac{\phi \sin \left ( \displaystyle \frac{2\pi x}{\ell } \right ) } { \ell \left [ \cosh w - \cos \left ( \displaystyle \frac{2 \pi x}{\ell } \right ) \right ] } , \\ b _ z ( x , z=0 ) & = & \frac{\phi \sinh w } { \ell \left [ \cosh w - \cos \left ( \displaystyle \frac{2 \pi x}{\ell } \right ) \right ] } , \label{period3}\end{aligned}\ ] ] respectively . see appendix a for more details on the derivation .
this solution of @xmath86 approaches to that given in 2 for a given @xmath52 in the limit of @xmath87 and hence @xmath88 .
the solution obtained in the previous paragraph is a critical one .
when @xmath71 is larger than the critical value , @xmath89 we obtain solutions in which the mass to flux ratio is not uniform .
equation ( [ criticalb ] ) can be rewritten as @xmath90 the right hand side of which denotes the maximum line density supported by magnetic flux , @xmath71 .
the filamentary clouds are confined by the magnetic pressure of the neighboring clouds .
remember that the filamentary cloud is confined by uniform magnetic field in the 2d model of paper i. thus the model shown in this section is closer to the 2d model than that in the previous section .
figure [ rho - xz2 ] shows the magnetic field for @xmath91 and @xmath92 .
the color denotes the density evaluated to be @xmath93 \right\ } ^{-2 } , \label{rho}\end{aligned}\ ] ] in the logarithmic scale .
equation ( [ rho ] ) denotes the equilibrium density distribution for an isothermal plane parallel disk having the surface density , @xmath94 .
the magnetic field is vertical to the disk plane near @xmath95 .
figure [ lambda5 ] shows the distribution of @xmath96 for models having various magnetic fluxes , @xmath97 , 10 , 12 , 14 , 16 , 18 , and 20 @xmath98 .
all the models have the same line density ( @xmath91 ) and accordingly the same surface density distribution shown by the dashed curve .
the value of @xmath81 is fixed at @xmath99 . the magnetic field and the surface density profiles are very similar near the filament axis ( @xmath100 ) . in other words
almost the same amount of magnetic flux is confined in the filament and the mass to flux ratio is constant , @xmath101 . when the magnetic flux is larger , the magnetic field is stronger outside the filamentary cloud . when the magnetic flux is less than a critical value ( @xmath102 in case of @xmath91 , see eq .
( [ criticalb ] ) ) , we can not construct an equilibrium model .
when it is close to the critical value , the mass to flux ratio is nearly constant also outside the cloud as in the model shown in 2 .
next we compare the models having the same @xmath46 and @xmath2 but different @xmath81 .
decrease in @xmath81 mimics quasi - static contraction since the surface density at the cloud center increases .
figure [ sc_ratio ] denotes the mass to flux ratio at the cloud center , @xmath103 , as a function of @xmath104 , where @xmath104 and @xmath105 denote the surface density and magnetic field at @xmath106 , respectively .
the ordinate is denoted in unit of @xmath107 .
the red curve denotes the models having @xmath108 and @xmath109 while the black one does those @xmath110 and @xmath109 . as the surface density at the cloud center increases , the mass to flux ratio there increases monotonically and is saturated at @xmath111 note that the mass to flux ratio tends to @xmath112 in the limit of @xmath113 since both the surface density and magnetic field are uniform in the limit .
interestingly , however , the mass to flux ratio at the cloud center depends little on @xmath46 for an intermediate @xmath81 . when the line density is smaller than the critical value , @xmath114 , the filamentary cloud is confined not by the self - gravity but by the magnetic field .
this is because the gas pressure dominates over the gravity .
we can construct such a model by assuming a small @xmath46 and a large magnetic flux .
they are similar to models with small @xmath115 in paper i.
the thin disk model shown in 2 and 3 reproduces main features of the 2d model shown in paper i. equation ( [ filament ] ) is essentially the same as equation ( 45 ) of paper i , @xmath116 where @xmath117 and @xmath118 denote the radius of the filament and the gas pressure at cloud surface , respectively .
the symbol , @xmath119 , denotes a half of the magnetic flux permeating the filament per unit length .
equation ( [ tomisaka45 ] ) was derived from the virial analysis and the left hand side denotes the pressure force acting on the cloud surface , which vanishes in our 1d model .
differences in the numerical factors come from the assumptions used . in this paper
we neglected the finite thickness of the cloud and hence pressure force acting in the @xmath11-direction for simplicity .
thus the critical line density is evaluated to be @xmath120 in case of no magnetic field ( @xmath121 ) in this paper while it should be @xmath12 in paper i. another difference comes from the assumed mass to flux ratio ; it is uniform in the model shown in 2 while it is not in paper i. the following equation , @xmath122 was derived from equation ( 45 ) of paper i as well as equation ( [ lined ] ) is derived from equation ( [ filament ] ) .
equation ( [ tomisaka46 ] ) has an asymptotic form , @xmath123 which resembles equation ( 38 ) of paper i , @xmath124 equation ( [ tomisaka38 ] ) is useful since it provides an upper limit on the line density supported against gravity by magnetic field and gas pressure . the upper limit increases in proportion to the magnetic flux contained in the cloud .
thus we examine the coefficient quantitatively . in paper
i the mass to flux ratio is assumed to be the same as that of the filament having uniform density , @xmath125 , and threaded by uniform magnetic field , @xmath126 .
then the line density and magnetic flux are denoted by @xmath127 respectively , where @xmath115 denotes the radius of the uniform filament .
note that the magnetic field surrounding but not permeating the filament is not in the count of @xmath128 .
it should be also noted that @xmath128 is twice as large as @xmath119 used in paper i. hence , the differential mass to flux ratio is expressed as @xmath129 ^{1/2 } , \label{masstof_tomiska } \\
\phi ^\prime & \equiv & \int _ 0 ^x b _ z ( x ^\prime ) dx ^\prime .\end{aligned}\ ] ] here the symbols , @xmath130 and @xmath131 , denote the magnetic field on the plane of @xmath13 and mass per unit magnetic flux , respectively . in order to evaluate the effect of the mass to flux ratio distribution
, we modified the cross section of the initial filament to be @xmath132 where @xmath133 .
the initial magnetic field is again uniform at @xmath126 .
then the mass to flux ratio is expressed as @xmath134 ^{{\cal n}/2 } , \label{masstof_tomiskan } \end{aligned}\ ] ] where @xmath60 denotes the gamma function .
the mass to flux ratio is uniform in the cloud when @xmath135 . on the other hand ,
most of the gas is filled in the magnetic field running at the center in case @xmath136 .
thus the parameter , @xmath137 , specifies the mass to flux ratio distribution as well as @xmath81 in the model shown in 3 .
see table [ nn ] for the ratio of the mass to flux ratio at the center to the average .
.the ratio of the central mass to flux ratio to the average .
[ nn ] [ cols=">,^",options="header " , ] figure [ sc_ratio2d ] shows the mass to flux ratio at the cloud center , @xmath138 as a function of the density at the cloud center . each curve denotes the locus of models having the same @xmath46 and @xmath139 , where @xmath140 denotes the initial plasma beta in the ambient gas surrounding the filamentary cloud . as in paper
i , the model clouds are assumed to be confined by a very tenuous gas of which pressure is @xmath141 .
the density at the cloud center is measured in unit of @xmath142 .
thus the abscissa denotes the ratio of the gas pressure at the cloud center to the magnetic pressure in the region very far from the cloud , since the magnetic field is nearly uniform at the initial value , @xmath126 , in our 2d model .
the line density is specified in unit of @xmath62 on the diagram .
the index , @xmath137 , increases from @xmath143 ( left ) to 10 ( right ) on each locus .
this means that the filamentary cloud is more centrally condensed and the central mass to flux ratio is higher when @xmath137 is larger .
the mass to flux ratio depends only a little on @xmath46 in the models of @xmath144 .
this is because the gas is relatively cold and accordingly the gas pressure has only minor contribution to the cloud support .
note that the mass to flux ratio is slightly larger for a smaller @xmath145 when @xmath146 is given .
the maximum non - dimensional mass to flux ratio is only slightly larger than unity when @xmath147 and @xmath148 .
when @xmath149 , the gas is relatively warm and hence the pressure has a significant contribution to the cloud support .
the non - dimensional mass to flux ratio is significantly higher than unity especially when @xmath150 , although it is still less than three .
this high mass to flux ratio is realized only when the cloud is relatively warm and the line density is relatively small .
the maximum line density depends also on @xmath137 , i.e. , @xmath151 .
equation ( 46 ) of paper is that for @xmath152 . when @xmath119 and @xmath8 are given , the maximum line density is lower for a larger @xmath137 .
increase in @xmath137 mimics the quasi - static evolution of a molecular cloud by the ambipolar diffusion .
as @xmath153 approaches to zero , the maximum line density tends to the value given by equation ( [ lined ] ) .
we consider the stability of the isolated filamentary cloud against fragmentation , i.e. , a sinusoidal perturbation in the @xmath11-direction . the gas is assumed to be still confined in the plane of @xmath13 and we use the method of images to obtain the changes in the gravity and density consistently .
the image density is assumed to have the form , @xmath154 for calculating the change in the gravity in the upper half space of @xmath33 .
the change in the gravity is evaluated as @xmath155^{3/2 } } dx^\prime dy^\prime dz ^\prime \\ & = & - g b \cos k y \int _ { -\infty } ^{+\infty } \int _ 0 ^\infty \frac { \delta \lambda ( b ) \cos k s } { \left ( x ^2 + s ^2 + b ^2 \right ) ^{3/2 } } ds db \\ & = & - 2 g k b \cos k y \int _ 0 ^\infty \frac{\delta \lambda ( b ) k _ 1 \left(k\sqrt{x^2+b^2 } \right ) } { \sqrt{x ^2 + b ^2 } } db , \label{dgz}\end{aligned}\ ] ] where @xmath156 denotes the modified bessel function of the 1st order .
here we used the mathematical formula @xcite , @xmath157 similarly we obtain @xmath158 the change in the surface density is expressed as @xmath159 the magnetic field is assumed to be aligned with the gravity also in the perturbed state .
this assumption is reasonable since the gas density is quite low and hence the inertia is negligibly small outside the cloud ( @xmath31 ) .
the magnetic field is still current free and does not extract any momentum from the cloud .
then the change in the magnetic field are expressed as @xmath160 in @xmath33 . both the gravity and magnetic field change with the time in proportion to @xmath161 , where @xmath162 denotes the angular frequency of the perturbation .
we use the displacement vector , @xmath163 for our stability analysis .
the @xmath14-dependence of the displacement vector is chosen so that the the equation of mass conservation be the ordinary differential equation with respect to @xmath22 , @xmath164 where @xmath165 denotes the surface density in the equilibrium .
also the equation of motion reduces to the ordinary differential equation and algebraic equation , @xmath166 it is difficult to solve equations ( [ per1 ] ) through ( [ per3 ] ) simultaneously since they are associated with equation ( [ dgz ] ) , ( [ dgx ] ) and ( [ dgy ] ) . instead of solving the equations we use the variational principle to evaluate @xmath167 .
first we rewrite equations ( [ per2 ] ) and ( [ per3 ] ) into @xmath168 by using the condition for magnetohydrostatic equilibrium , @xmath169 note that the displacement is irrotational , @xmath170 since angular momentum extraction by magnetic field is not taken into account
. we obtain @xmath171 by taking the sum of the products of @xmath172 and equation ( [ per1 ] ) , @xmath173 and equation ( [ per4 ] ) , @xmath174 and equation ( [ per5 ] )
. we obtain @xmath175 dx } { \displaystyle \int _ { -\infty } ^{+\infty } \sigma _ 0 \left ( \xi _ x ^2 \ , + \ , \xi _ y ^2 \right ) dx } \\ & = & c _ s ^2 \frac{\displaystyle \int _ { -\infty } ^{+\infty } \sigma _ 0 \left [ \left ( \frac{\delta \sigma}{\sigma _ 0 } \right ) ^2 - \frac{1}{g \lambda } \left ( \xi _ x \delta g _ x + \xi _ y \delta g _ y \right)\right ] dx } { \displaystyle \int _ { -\infty } ^{+\infty }
\sigma _ 0 \left ( \xi _ x ^2 \ , + \ , \xi _ y ^2 \right ) dx } \label{var2 } \end{aligned}\ ] ] by integrating equation ( [ var1 ] ) and submitting @xmath176 at @xmath177 .
the right hand side of equation ( [ var2 ] ) gives a lower bound for the eigenvalue , @xmath167 , when an arbitrary perturbation is substituted .
if it is negative for a given perturbation , the filament is unstable .
we evaluate the right hand side of equation ( [ var2 ] ) as a functional of the change in the image density , @xmath178 .
the change in the gravity is evaluated by numerical integral of ( [ dgx ] ) , and ( [ dgy ] ) .
the change in the surface density is evaluated by the numerical integral of ( [ dsigma ] ) .
the displacement , @xmath179 , is obtained by solving the ordinary differential equation , @xmath180 - k ^2 \sigma _ 0 \xi _ x & = & 0 , \label{xix2}\end{aligned}\ ] ] which is obtained by combining equations ( [ per1 ] ) and ( [ irrotational ] ) .
the displacement , @xmath181 is obtained simultaneously when we solve equation ( [ xix2 ] ) .
the above mentioned procedure allows us to evaluate the right hand side of equation ( [ var2 ] ) as a functional of @xmath182 .
see appendix b for more details .
as shown in the previous section , fwhm of our model cloud ( = @xmath183 ) is not specified by the condition for magnetohydrostatic equilibrium .
if the filamentary cloud has proper line density and magnetic flux , it can be settled in equilibrium at any width .
this means that our model cloud is neutrally stable for perturbation having @xmath184 . as in the case of longitudinal or helical magnetic field
, we use the non - dimensional wavenumber , @xmath185 , in our analysis . in this paper
we consider two types of trial functions for the variational principle .
type i trial function is expressed as @xmath186 where @xmath187 is chosen to minimize the value of @xmath188 .
type ii is expressed as @xmath189 where a set of @xmath190 is chosen to minimize the value of @xmath167 for fixed @xmath191 .
the left panel of figure [ type1 ] shows the growth rate obtained by applying type i trial function .
the abscissa is the wavenumber in unit of @xmath192 while the ordinate is the growth rate in unit of @xmath193 .
the right panel of figure [ type1 ] shows that obtained by applying type ii trial function in which the values of @xmath194 are taken to be @xmath195 .
our model filamentary clouds are unstable against fragmentation as well as the filamentary clouds threaded by longitudinal magnetic field .
the most unstable mode has the wave number , @xmath196 , and the growth rate , @xmath197 , irrespectively of @xmath46 .
as shown in the previous sections , our 1d model based on the thin disk approximation reproduces the main features of 2d model shown in paper i. the maximum mass sustained by magnetic field is roughly proportional to the magnetic flux .
the critical mass to flux ratio depends on the mass loading , i.e. , the distribution of @xmath198 . when the mass loading is uniform ( @xmath199 const . ) , it is slightly larger than @xmath200 . the same critical ratio is obtained for disks in equilibrium and slabs stable against fragmentation .
remember that @xcite obtained a similar value from the viral analysis on the magnetohydrostatic equilibrium .
the critical ratio seems to depend little on the cloud geometry .
remember that filamentary clouds with a longitudinal field are unstable against fragmentation even if the magnetic field is very strong .
the instability is due to the fact that the mass to flux ratio is infinitely large since the clouds are highly extended in the direction of magnetic field . given the critical mass to flux ratio depends little on the cloud geometry , then it can be applied to the stability of periodically arranged filamentary clouds .
they are almost isolated each other in the limit of @xmath88 .
thus they are unstable against fragmentation as well as the isolated filamentary clouds since the mass to flux ratio is larger than the critical value at the cloud center ( @xmath201 ) .
when @xmath113 , our 1d model reduces to a slab of which stability was already investigated by @xcite .
the slab is stable as far as @xmath202 . in short
the models shown in figure [ sc_ratio ] are stable / unstable for small / large value of @xmath203 .
we surmise that the models of @xmath204 are stable against fragmentation .
most of the 2d models shown in figure [ sc_ratio2d ] are also likely to be stable against fragmentation since @xmath205 .
the above argument brings us an interesting result .
filamentary clouds supported by longitudinal magnetic fields are unstable against fragmentation while those supported by perpendicular fields can be stable .
the latter can collapse quasi - statically through ambipolar diffusion , while the former can not .
our models are still idealistic since turbulence and other dynamical effects are not taken into account .
nevertheless , they provide physical insights on the dynamics of filamentary clouds .
filamentary clouds with perpendicular magnetic field are likely to be able to sustain their forms as far as they are subcritical at the cloud center .
they may fragment after the mass to flux ratio exceeds the critical value by ambipolar diffusion or by mass accretion along the magnetic field @xcite .
this work was supported in part by jsps kakenhi grant number 24540226 .
we used the mathematical formula , @xmath206 for @xmath207 in order to derive equation ( [ period1 ] ) from equations ( [ psigma ] ) and ( [ seriesa ] ) .
the gravitational potential , @xmath56 , and the @xmath14-component of the vector potential , @xmath208 , are expressed as @xmath209 , \\ a _ y & = & \frac{\phi x}{\ell } + \sum _ { n=1 } ^\infty \frac{\phi}{\pi n } \sin \frac{2 \pi n x}{\ell } \exp \left [ - n \left ( w + \frac{2 \pi}{\ell } |z|
\right ) \right ] , \end{aligned}\ ] ] respectively , for the solution specified by equations ( [ period1 ] ) through ( [ period3 ] ) .
they are also expressed as @xmath210 , \\ a _ y & = & \frac{\phi}{\pi } \tan ^{-1 } \left [ \frac{\tan \left ( \displaystyle \frac{\pi x}{\ell } \right ) } { \tanh \left ( \displaystyle \frac{w}{2 } + \displaystyle \frac{\pi |z|}{\ell } \right ) } \right ] , \end{aligned}\ ] ] respectively , since @xmath211 for @xmath212 .
for a given @xmath213 we obtain @xmath179 and @xmath181 by the following procedure .
first we rewrite equations ( [ xix2 ] ) and ( [ irrotational ] ) as @xmath214 we consider the case in which @xmath213 and @xmath215 are symmetric and antisymmetric with respect to @xmath22 , respectively .
this choice is rational since the unperturbed state is symmetric and we are interested in the gravitational instability .
an eigenmode perturbation should be either symmetric or antisymmetric and the fragmentation of a filamentary cloud is symmetric with respect to @xmath22 .
thus we obtain the boundary condition , @xmath176 at @xmath106 .
we assume that the displacement diminishes in the form , @xmath216 in the limit of @xmath217 .
this choice is based on the fact that the relative change in the surface density is proportional to @xmath218 in the limit .
equations ( [ asympto ] ) and ( [ asympto2 ] ) are valid also when @xmath219 decreases more steeply with increase in @xmath22 .
we integrate equations ( [ per6 ] ) and ( [ per7 ] ) from @xmath106 to @xmath220 with stepsize @xmath221 by the 4th order runge - kutta method to obtain two sets of solutions .
one has initial condition @xmath222 .
the other is the solution of equations ( [ per6 ] ) and ( [ per7 ] ) without the source term , i.e. , the homogenous one and has the initial condition @xmath223 .
we obtain the solution satisfying the boundary conditions at @xmath106 and @xmath224 as a linear combination of them .
the outer boundary is set at @xmath225 .
the @xmath14-component of the displacement , @xmath226 , is derived from equation ( [ irrotational ] ) .
now we can evaluate the right hand side of equation ( [ var2 ] ) for a given @xmath178 .
consequently the growth rate is evaluated to be a function of @xmath227 in type i trial function .
we searched for @xmath227 which maximizes the growth rate in the interval of @xmath228 with the interval @xmath229 .
both the denominator and delimiter of the right hand side of equation ( [ var2 ] ) are the quadratic expressions of @xmath230 , when type ii trial function is used .
we used lapack , a mathematical library for linear algebra installed in idl , the interactive data language , for minimizing @xmath188 .
abramowicz , m. , stegun , i.a .
1965 , handbook of mathematical functions with formulas , graphs and mathematical tables , new york , dover binney , j. , tremaine , s. 2008 , galactic dynamics , 2nd ed .
princeton , princeton univ .
press , sec .
2.6 goodman , a.a . ,
bastien , p. , myers , p.c . ,
mnard , f. , 359 , 363 hanawa , t. , nakamura , f. , matsumoto , t. et al .
1993 , , 404 , l83 heitsch , f. & hartmann , l. 2014 , , 443 , 230 mestel , l. , paris , r.b . 1984 , , 136 , 98 mestel , l. , spitzer , jr . ,
l. 1956 , , 116 , 503 moneti , a. , pipher , j. l. , helfer , h. l. et al .
, 282 , 508 mouschovias , t.c .
1976a , , 206 , 753 mouschovias , t.c .
1976b , , 207 , 141 nakamura , f. , hanawa , t. , nakano , t. 1995 , , 444 , 700 nakano , t. , nakamura , t. 1978 , , 30 , 671 ostriker , j. 1964 , , 140 , 1056 palmeirim , p. , andr , ph . , kirk , j. , et al . , 550 , a38 pereyra , a. , magahes , a.m. 2004 , , 603 , 584 shu , f.h . ,
li , z .- y .
1997 , , 475 , 251 stodkiewicz , j.s . 1963 , aca , 13 , 30 strittmatter , p.a . 1966 , , 132 , 359 tomisaka , k. 1991 , , 376 , 190 tomisaka , k. 1995 , , 438 , 226 tomisaka , k. , ikeuchi , s. , nakamura , t. 1988 , , 335 , 239 2014 , , 785 , 24 | we have constructed two types of analytical models for an isothermal filamentary cloud supported mainly by magnetic tension .
the first one describes an isolated cloud while the second considers filamentary clouds spaced periodically .
both the models assume that the filamentary clouds are highly flattened .
the former is proved to be the asymptotic limit of the latter in which each filamentary cloud is much thinner than the distance to the neighboring filaments .
we show that these models reproduce main features of the 2d equilibrium model of @xcite for filamentary cloud threaded by perpendicular magnetic field .
it is also shown that the critical mass to flux ratio is @xmath0 , where @xmath1 , @xmath2 and @xmath3 denote the cloud mass , the total magnetic flux of the cloud , and the gravitational constant , respectively .
this upper bound coincides with that for an axisymmetric cloud supported by poloidal magnetic fields .
we applied the variational principle for studying the jeans instability of the first model .
our model cloud is unstable against fragmentation as well as the filamentary clouds threaded by longitudinal magnetic field .
the fastest growing mode has a wavelength several times longer than the cloud diameter .
the second model describes quasi - static evolution of filamentary molecular cloud by ambipolar diffusion . |
there has been a considerable interest in the study of the number and nature of limit cycles in a lienard equation@xmath6 recently @xmath7 $ ] .
limit cycles are isolated periodic curves in the phase plane and arise in numerous applications as self - sustained oscillations which exist even in the absence of external periodic forcing .
the equation @xmath8 is usually studied as an autonomous system , called the lienard system , given by@xmath9 where @xmath10 . the phase plane defined by @xmath11
is called the lienard plane .
lienard gave a criterion for the uniqueness of periodic cycles for a general class of equations when @xmath2 is an odd function and satisfies a monotonicity condition as @xmath12 .
an interesting problem for the system @xmath13 is the determination of the number of limit cycles for a given odd degree @xmath14 polynomial @xmath2 .
lins , pugh and de melo @xmath15 conjectured that the system @xmath13 has at most @xmath16 limit cycles if @xmath17 or @xmath18 .
currently this problem is being investigated by many authors in connection with the still unsolved hilbert s 16th problem .
giacomini and neukirch @xmath19 have developed a general procedure for constructing a sequence of polynomials whose roots of odd multiplicity are related to the number and location of the limit cycles of equation @xmath20 when @xmath21 is an even degree polynomial .
they have also given a sequence of algebraic approximations to the equation of each such cycles , although their method is mainly of experimental ( numerical ) in nature and a rigorous justification is still lacking .
holst and sundberg @xmath22 have extended rychkov s theorem @xmath23 for a class of @xmath2 having @xmath24 degree polynomial like behaviour .
the proof of richkov s theorem however requires _
bifurcation theory_. odani gave a proof on the existence of exactly @xmath16 limit cycles of the lienard equation @xmath20 with @xmath25 .
his proof does not make use of the bifurcation theory .
his method also gave an improved estimate of the amplitude of a limit cycle .
recently there have been some progress in elucidating sufficient conditions extending the previous results .
chen and chen @xmath26 , for instance , proved the lins - pugh - de melo conjecture for lienard system with function @xmath4 odd . on the other hand
, it has been shown @xmath27 that for suitable polynomial @xmath4 of degree @xmath28 , the system @xmath13 has @xmath29 limit cycles , contradicting the conjecture in @xmath15 .
chen , llibre and zhang @xmath30 proved a sufficient condition for existence of exactly @xmath16 limit cycles for the system @xmath31 with a general class of @xmath2 functions .
we investigate an equivalent problem covering , however , a different class of functions @xmath4 as compared to @xmath30 . in the study on the number of limit cycles
several authors have studied the equation @xmath20 in the usual phase plane @xmath32viz .
theorem @xmath33 , chapter @xmath29 in @xmath34 whereas some considered the lienard plane @xmath30 . in theorem
@xmath35 , chapter @xmath29 @xmath36 the function @xmath37 is taken as a periodic function .
theorem @xmath38 is a generalization of theorem @xmath35 in which the function @xmath39 is a monotone function in certain regions .
however theorem @xmath40 and theorem @xmath41 in the present paper do not depend upon the monotonicity of @xmath37 .
rather , we have used the monotonicity of @xmath4 . as a consequence , merely the sign of the function
@xmath37 determines the monotonic nature of @xmath4 , and hence determines the number of limit cycles in lienard system @xmath13 .
thus our results cover a different class of functions than those covered by theorems @xmath35 and @xmath38 mentioned above .
theorem @xmath42 , chapter @xmath29 in @xmath36 and the theorem in @xmath30 have been proved on lienard plane .
both of these results have assumed the existence of @xmath43 $ ] , @xmath44 such that @xmath45 where , @xmath46 s are positive roots of @xmath4 and @xmath47 s are unique extremum of @xmath4 in @xmath48 $ ] for @xmath49 .
however , if we do not get any such @xmath50 then these results are not applicable . in such situations
theorem @xmath41 in section @xmath51 of the present paper is still applicable to determine the exact number of limit cycles .
one such example is given in section @xmath52 . in this paper
we first give a simple but , nevertheless , an important extension of the lienard s theorem for the unique limit cycle by removing the unbounded nature of the function @xmath4 as @xmath12 .
next , in theorem @xmath40 we prove that the system @xmath13 has exactly two limit cycles when the odd function @xmath53 undergoes two sign changes in @xmath54 and is monotonic not only as @xmath12 , but also near @xmath32actually at the right of@xmath55 the first zero .
however , @xmath3 @xmath56 for @xmath57 can be any odd continuous function .
example @xmath58 in support of theorem @xmath40 reveals clearly the strength of this theorem over analogous results @xmath32e.g .
theorem @xmath59 of @xmath36 .
the _ new insights _ gained from theorem @xmath40 @xmath32and also from theorem @xmath60 then provide a general approach in obtaining an existence theorem for multiple limit cycles in a systematic manner .
in theorem @xmath41 , we state a set of such conditions for the existence of exactly @xmath16 limit cycles .
although we are dealing with odd functions @xmath4 only , there are certain odd functions as shown in example @xmath61 , which satisfy the conditions of theorem @xmath41 in the current paper but do not satisfy the theorem of @xmath30 .
this establishes our claim that the present theorems cover different classes of functions @xmath4 than those covered in @xmath36 and @xmath30 .
moreover , as stated above , @xmath3 here is an odd function while for the theorem of @xmath30 @xmath25 .
the second important result that we find in section @xmath62 is an efficient upper estimate of the amplitude of the limit cycle for the system @xmath11 .
the values of the amplitudes for the van der pol equation are obtained in example @xmath63 , which are much more accurate compared to those in @xmath64 and @xmath65 .
the paper is organized as follows . in section @xmath66
, we sketch the main steps of the proof of the classical lienard theorem thus introducing our notations . in section @xmath62
, we discuss some special observations leading to an extension of the classical lienard theorem .
our main result , theorem @xmath40 , on the existence of two limit cycles is proved in section @xmath51 . in theorem
@xmath41 we state the sufficient conditions for existence of exactly @xmath16 limit cycles .
an outline of the proof is given in appendix @xmath32for a detailed proof , see @xmath67 .
the proof of this general existence theorem is based on an induction method with _ non - trivial initial hypotheses _ for @xmath68 and @xmath69 .
we present some examples in section @xmath52 highlighting the key features of the above theorems .
section @xmath70 contains some concluding remarks .
here we present an outline of the lienard s theorem for the sake of completeness .
this helps us introducing necessary notations which will be used subsequently .
[ lienard theorem]the equation @xmath20 has a unique periodic solution if [ c]rl@xmath71 & @xmath37 and @xmath72 are continuous ; + @xmath73 & @xmath4 and @xmath3 are odd functions with @xmath74 for @xmath54 ; + @xmath75 & @xmath4 is zero only at @xmath76 , @xmath77 , @xmath78 for some @xmath79 + @xmath80 & @xmath81 as @xmath12 monotonically for @xmath82 . [ a brief sketch of the proof]the general shape of the path can be obtained from the following observations . 1 . because of the symmetry of the system@xmath8 under @xmath83 any periodic orbit is symmetric about the origin .
the slope of a phase path is given by@xmath84 thus , a phase path is horizontal if @xmath85 , i.e. if @xmath86 , i.e. if @xmath76 @xmath87 .
similarly , a phase path is vertical on the curve @xmath88 . above the curve @xmath89 we have @xmath90 and below @xmath91 .
moreover , @xmath92 for @xmath54 and @xmath93 for @xmath94 .
.,height=226 ] a path @xmath95 @xmath32figure [ lienard limit cycle]@xmath55 is closed iff @xmath96 and @xmath97 coincide , which means by symmetry @xmath98@xmath99 this is equivalent to@xmath100 where for a typical path @xmath101 in figure [ typical path]@xmath102 and@xmath103 writing@xmath104 where @xmath105 is a line parallel to the @xmath106 axis and passing through the point @xmath107 when the function @xmath4 changes its sign from negative to positive , one then proves that 1 . as @xmath108 moves out of the point @xmath109 along the curve @xmath110 , the potentials @xmath111 is positive and monotone decreasing .
2 . as @xmath108 moves out of the point @xmath109 along the curve @xmath110 ,
@xmath112 is monotone decreasing .
3 . from @xmath113 and
@xmath114 it follows that @xmath115 is monotone decreasing to the right of the point @xmath116 , @xmath32figure [ typical path]@xmath55 .
the quantity @xmath112 tends to @xmath117 as the paths moves away to infinity .
5 . from @xmath118 and @xmath119
, it follows that the quantity @xmath115 is monotone decreasing to @xmath117 , at the right of the point @xmath116 @xmath32figure [ typical path]@xmath55 .
@xmath120 when the point @xmath108 is at @xmath116 or to the left of the point @xmath116 .
it thus follows from @xmath121 and @xmath122 that @xmath115 is monotone decreasing continuous function which changes its sign from positive to negative as the point @xmath108 moves out of @xmath123 along the curve . as a result
, @xmath115 will vanish once and only once .
thus , there is one and only one closed path and the proof is complete .
[ simple limit cycle]the unique limit cycle in the above theorem is * simple * in the sense that no ( differentiable ) perturbation satisfying the conditions @xmath71-@xmath80 can bifurcate the limit cycle into two or more number of limit cycles .
the condition @xmath121 enables us to conclude that once @xmath115 becomes negative , it can never be positive as @xmath108 moves to infinity through the curve of @xmath2 .
this observation helps us to deduce the existence of a unique limit cycle .
however , we see that the existence of the limit cycle is indeed ensured only if @xmath115 becomes negative from positive i.e. , if there is a change in sign of @xmath115 .
further , the unique value of @xmath124 for which @xmath125 gives the * amplitude * of the limit cycle . accordingly , if @xmath115 becomes negative as @xmath108 moves out from origin through the curve of @xmath2 then we get a limit cycle . _
this observation actually gives one with a possibility of weakening the conditions of the classical theorem , so as to accommodate a larger class of functions _
@xmath2 _ but still having a unique limit cycle .
_ theorem @xmath126 is one such realizations of a stronger version of the classical theorem , which shows that the existence of the @xmath32unique@xmath55 limit cycle actually depends on the * local * monotonicity of @xmath2 on a bounded interval containing the point where @xmath115 vanishes .
a limit cycle can indeed be realized even when @xmath2 is bounded as @xmath127 @xmath32c.f .
example @xmath128if it happens further that @xmath115 becomes positive from negative once more , then also , by an analogous argument as above we can get a point @xmath108 on the curve @xmath2 , through which another limit cycle must pass .
to prove this result we consider a function @xmath2 @xmath32 in section @xmath129 which is monotonically increasing to the right of the point @xmath116 for a sufficiently large value of @xmath124 and then it becomes decreasing for some subsequent values of @xmath124 and ultimately become negative .
the proof depends on an efficient estimate of the amplitude of the first limit cycle .
let , @xmath130 be the coordinate of @xmath108 , as shown in figure [ typical path ] and let @xmath131 be the amplitude of the limit cycle of theorem [ lienard theorem ] .
it is well known that determining the exact value of limit cycle of the lienard system is a relatively difficult problem @xmath132 $ ] .
we now find an estimate of @xmath133 , for which the corresponding @xmath115 just become negative from positive .
since , @xmath134 is a monotone decreasing continuous function as the point moves out of the point @xmath135 along the curve , without any loss of generality we can say @xmath115 can just become negative from positive if at least one of the following two cases hold , viz . , @xmath71 @xmath136 but @xmath137 @xmath73 @xmath138 but @xmath139.the third possibility @xmath139 and @xmath137 can be reduced to either of the above two cases by monotonicity and continuity of @xmath115 , i.e. by taking an @xmath140 closer to @xmath133 , @xmath32i.e .
, @xmath141 @xmath55 either one of @xmath142 and @xmath143 can be made to vanish .
similarly , the possibility that either one of @xmath142 and @xmath143 is positive while their sum is negative , can also be eliminated by choosing @xmath140 far from @xmath133 @xmath144 .
here @xmath145 is possible if@xmath146 in step @xmath113 of the proof of theorem @xmath147 it has been proved that @xmath148 .
we are now going to show that @xmath149 .
on the path @xmath150 , we have @xmath151 and @xmath152 .
therefore , @xmath153 thus , we can say that @xmath154 is true if @xmath155@xmath156 where @xmath157 @xmath32figure [ typical path]@xmath55 .
it is possible if@xmath158 so , we have@xmath159 where @xmath160 .
let @xmath161 be a root of @xmath162 @xmath163 so that@xmath164 here , @xmath138 is possible if@xmath165 in step @xmath113 of the proof of the theorem @xmath147 it is proved that @xmath166 .
proceeding analogous to case @xmath71 one establishes that @xmath167 and consequently @xmath168 is true provided@xmath169 where @xmath170 @xmath32figure [ typical path]@xmath55 . if @xmath171 be a root of @xmath172 we have@xmath173 it now follows that if we take@xmath174 then for any value of @xmath175 @xmath176since , @xmath177 . thus the function @xmath4 should be monotonic increasing in the interval @xmath178 .
notice that the classical lienard theorem already ensures the existence of such an @xmath179 . in the light of the above discussion we can now _
extend _ the classical lienard theorem by weakening the unbounded nature of the function @xmath4 as stated in the following theorem and cover a more large class of functions .
[ lienard theorem extension]the equation @xmath20 has a unique limit cycle if [ c]rl@xmath71 & @xmath37 and @xmath72 are continuous in @xmath180 for sufficiently large @xmath181 ; + @xmath73 & @xmath4 and @xmath72 are odd functions with @xmath182 for @xmath54 ; + @xmath75 & @xmath4 is zero only at @xmath76 , @xmath77 , @xmath78 for some @xmath183 where @xmath184 + @xmath80 & @xmath185 a number @xmath179 defined by @xmath186 such that @xmath4 is monotonic increasing in + & @xmath178 and nondecreasing in @xmath187 .
the existence of @xmath179 ensures a sign change in @xmath115 whereby we get the existence of a unique limit cycle in the finite phase plane .
the remaining part of the proof of this theorem remain same as that of classical lienard theorem . also , from the above discussion and the proof of classical lienard theorem it follows that @xmath115 does not change its sign any more if the function @xmath4 is simply _
monotone nondecreasing _ in @xmath188 . in such cases
the function @xmath4 can _ even be bounded and even attain a constant value as _
@xmath189 , but still we get a unique limit cycle for such a bounded lienard system .
thus , we can indeed cover a larger class of functions than those covered by the lienard theorem @xmath32c.f .
[ ex compare amplitude]@xmath55 . in the beginning of the proof of theorem [ lienard theorem ]
we observed that above the curve @xmath89 we have @xmath90 and below @xmath91 .
so , the @xmath124-coordinate of a point on a limit cycle will achieve its maximum absolute value on the curve @xmath89 .
therefore , the amplitude of a limit cycle for the lienard system is the abscissa of the point @xmath108 lying on the curve @xmath89 . since , for a limit cycle we have @xmath190 , so by the construction of @xmath179 it follows that it is an _ efficient upper estimate of the amplitude _ of the limit cycle . in the following example
we find the values of @xmath179 for the well known van der pol equation against different values of @xmath191 and compare them with the results obtained in @xmath64 and @xmath65 .
this also gives an example of a _ bounded _ van der pol equation having same amplitude as that of the standard van der pol equation .
[ ex compare amplitude]here , in the following table we present estimates of the amplitude of the limit cycle for van der pol equation in which @xmath192 and @xmath193 for different values of @xmath194 it is clear that our estimates are reasonably close to the exact @xmath32numerically computed@xmath55 values @xmath32as reported in @xmath195 .
our values also appear to be much better than the upper bound @xmath196 of @xmath64 @xmath32the estimated values of @xmath65 are valid only for small @xmath197 . from our numerical estimates
it follows that although the estimated values of the amplitude seem to vary irregularly for the moderately large values of @xmath198 $ ] , these are nevertheless bounded above by @xmath199.@xmath200{|c|c|c|}\hline $ \mu$ & $ y_{+}\left ( 0\right ) $ and $ y_{-}\left ( 0\right ) $ & $ \bar { \alpha}$\\\hline $ 0.1 $ & $ 2.00117 $ & $ 2.0000586437166383$\\\hline $ 0.2 $ & $ 2.007076 $ & $ 2.002540101136999$\\\hline $ 0.3 $ & $ 2.015912 $ & $ 2.0054678254782505$\\\hline $ 0.4 $ & $ 2.028253 $ & $ 2.0091503375996034$\\\hline $ 0.5 $ & $ 2.044065 $ & $ 2.013278539452526$\\\hline $ 1 $ & $ 2.1727135 $ & $ 2.0327736318429275$\\\hline $ 1.5 $ & $ 2.3710897 $ & $ 2.0436704679281523$\\\hline $ 2 $ & $ 2.6149725 $ & $ 2.04739132291152$\\\hline $ 2.5 $ & $ 2.8844602 $ & $ 2.047213463900291$\\\hline $ 3 $ & $ 3.1687156 $ & $ 2.045311842105752$\\\hline $ 3.5 $ & $ 3.462322 $ & $ 2.0427848260891426$\\\hline $ 4.5 $ & $ 4.06701715 $ & $ 2.037557405718347$\\\hline $ 5 $ & $ 4.3752293 $ & $ 2.035154629371522$\\\hline $ 10 $ & $ 7.5528123 $ & $ 2.020095969119061$\\\hline \end{tabular } \ \ \ \ \ ] ] we get the same result if we consider the bounded function@xmath201{ll}\mu\left ( \dfrac{x^{3}}{3}-x\right ) & x\in\left ( -2.4,2.4\right )
\\ \mu\left ( \dfrac{(2.4)^{3}}{3}-2.4\right ) -\dfrac{4.76\mu}{\sin\left ( 0.6\right ) } \left ( \cos\left ( 0.6\right ) -\cos\left ( x-3\right ) \right ) & x\in\left ( -3,-2.4\right ] \cup\left [ 2.4,3\right ) \\ \mu\left ( \dfrac{(2.4)^{3}}{3}-2.4\right ) -\dfrac{4.76\mu}{\sin\left ( 0.6\right ) } \left ( \cos\left ( 0.6\right ) -1\right ) & x\in\left ( -\infty,-3\right ] \cup\left [ 3,\infty\right ) \end{array } \right.\ ] ] in the whole phase plane or the following function in finite phase plane.@xmath201{ll}\mu\left ( \dfrac{x^{3}}{3}-x\right ) & x\in\left ( -2.4,2.4\right ) \\ \mu\left ( \dfrac{(2.4)^{3}}{3}-2.4\right ) -\dfrac{4.76\mu}{\sin\left ( 0.6\right ) } \left ( \cos\left ( 0.6\right ) -\cos\left ( x-3\right ) \right ) & x\in\left ( -3,-2.4\right ] \cup\left [ 2.4,3\right ) \end{array } \right.\ ] ] this also tells that the _ value of _
@xmath181 _ need not by too large_. a limit cycle is assured even for a moderately large @xmath181 .
it follows from this example that the amplitude of the unique limit cycle is independent of the asymptotic behaviour of @xmath2 as @xmath12 .
indeed , the amplitude corresponds to the point @xmath202 on the limit cycle for which the integral @xmath203 vanishes and clearly depends on the form of @xmath2 in the finite interval @xmath204 . to the best of author s knowledge ,
this result apparently is not recorded clearly in the literature .
we therefore state this observation as the following corollary .
amplitude of the unique limit cycle of the lienard system @xmath11 is independent of the asymptotic behaviour of @xmath2 as @xmath5 .
by the observations discussed in last section it is clear that , in the interval @xmath205 we obtain the limit cycle of theorems @xmath147 and @xmath126 . moreover , because of the condition @xmath80 we see that the limit cycle remains unique .
however , if the function does not satisfy this condition , then the limit cycle may not be unique .
we now present our new theorem .
[ new theorem]let @xmath37 and @xmath72 be two functions satisfying the following properties .
[ c]rl@xmath71 & @xmath37 and @xmath72 are continuous ; + @xmath73 & @xmath4 and @xmath72 are odd functions and @xmath182 for @xmath54 . ; + @xmath75 & @xmath4 has positive simple zeros only at @xmath206 , @xmath207 for some @xmath208 and + & some @xmath209 , @xmath179 being defined by @xmath186 and @xmath210 , where @xmath211 is the first + & local maxima of @xmath2 in @xmath212 ; $ ] + @xmath80 & @xmath4 is monotonic increasing in @xmath213 and @xmath214 as @xmath215 + & monotonically for @xmath216 then the equation @xmath20 has exactly two limit cycles around the origin . *
proof : * we can get exactly the same observations as we get in observations @xmath98 and @xmath217 in beginning of the proof of theorem @xmath147 . by the observations in section @xmath62 we can ensure the existence of inner limit cycle .
so , we shall now prove the existence of one more limit cycle by showing that@xmath218 once more when @xmath219 . to prove the result
we shall consider the function @xmath220 as in @xmath221 , and write,@xmath222 where , @xmath223 is a line parallel to the @xmath106 axis passing through the point @xmath224 where the function @xmath4 changes its sign from negative to positive and @xmath105 is a line parallel to the @xmath106 axis passing through the point @xmath225 where the function @xmath4 changes its sign from positive to negative . the proof is carried out through the steps @xmath113 to @xmath122 below . here
we refer to the figure @xmath226 .
we choose two points @xmath231 and @xmath232 on the curve of @xmath2 where @xmath233 .
let @xmath101 and @xmath234 be two paths through @xmath108 and @xmath235 respectively . on the segments @xmath236 and @xmath237 we have@xmath238 now,@xmath200{lc } & $ \left ( y - f\left ( x\right ) \right ) _ { yx}<\left ( y - f\left ( x\right ) \right ) _ { y_{1}x_{1}}$\\ $ \implies$ & $ \left ( \dfrac{1}{y - f\left ( x\right ) } \right ) _ { yx}>\left ( \dfrac{1}{y - f\left ( x\right ) } \right ) _ { y_{1}x_{1}}$\end{tabular } \nonumber\ ] ] since @xmath74 for @xmath54 , we have@xmath239 so by @xmath240 we get@xmath241 therefore,@xmath242 using @xmath243 we get@xmath244 since , @xmath4 and @xmath245 are positive on @xmath237 we have@xmath246 next , on the segments @xmath247 and @xmath248 we have@xmath249 now,@xmath200{lcl } & $ \left ( y - f\left ( x\right ) \right ) _ { x^{\prime}y^{\prime}}>\left ( y - f\left ( x\right ) \right ) _ { x_{1}^{\prime}y_{1}^{\prime}}$ & \\ $ \implies$ & $ \left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) } \right ) _ { x^{\prime}y^{\prime}}>\left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) } \right ) _ { x_{1}^{\prime}y_{1}^{\prime}}$ & \end{tabular } \ ] ] so , by @xmath240@xmath250 therefore,@xmath251 using @xmath252 we get@xmath253 since , @xmath4 and @xmath245 are negative on @xmath254 we have@xmath255 from @xmath256 and @xmath257 we have@xmath258 therefore , @xmath230 is positive and monotonic decreasing as the point @xmath108 moves out from @xmath228 along @xmath229 . on the segments @xmath261 and @xmath262 we have@xmath238 now,@xmath200{lcl } & $ \left ( y - f\left ( x\right ) \right ) _ { xb}<\left ( y - f\left ( x\right ) \right ) _ { x_{1}b_{1}}$ & \\ $ \implies$ & $ \left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) }
\right ) _ { xb}<\left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) } \right ) _ { x_{1}b_{1}}$ & \end{tabular}\ ] ] so , by @xmath240 we get@xmath263 therefore,@xmath264 using @xmath265 we get@xmath266 since , @xmath267 and @xmath268 on @xmath261 and @xmath262 . next , on the segments @xmath269 and @xmath270 we have@xmath271 now,@xmath200{lc } & $ \left ( y - f\left ( x\right ) \right ) _ { b^{\prime}x^{\prime}}>\left ( y - f\left ( x\right ) \right ) _ { b_{1}^{\prime}x_{1}^{\prime}}$\\ $ \implies$ & $ \left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) } \right ) _ { b^{\prime}x^{\prime}}>\left ( \dfrac{-g\left ( x\right ) } { y - f\left ( x\right ) } \right ) _ { b_{1}^{\prime}x_{1}^{\prime}}$\end{tabular}\ ] ]
using @xmath240 we get@xmath272 therefore,@xmath273 so , by @xmath274 we have@xmath275 since , @xmath267 and @xmath268 on @xmath269 and @xmath276 .
from @xmath277 and @xmath278 we have@xmath279 therefore , @xmath260 is negative and monotonic increasing as the point @xmath108 moves out from @xmath228 along @xmath229 .
on @xmath282 and @xmath283 , we have @xmath284 .
we draw @xmath285 and @xmath286 parallel to @xmath287 axis .
therefore,@xmath288 since , @xmath289 and @xmath290 for points on @xmath291 .
again since , @xmath292 _ { b^{\prime}qb}\geq\left .
f\left ( x\right ) \right ] _ { h_{1}^{\prime}q_{1}h_{1}}\ ] ] for same value of @xmath293 we get @xmath294{l}$\implies v_{b_{1}q_{1}b_{1}^{\prime}}$\end{tabular } & \geq v_{bqb^{\prime}}. \label{potential comparison5}\ ] ] next , let @xmath295 be a point on the curve of @xmath2 , to the right of @xmath228 , and let @xmath282 be an arbitrary path , with @xmath108 to the right of @xmath295 .
the straight line @xmath296 is parallel to the @xmath106 axis .
then,@xmath297 since , @xmath298 and @xmath299 along @xmath300 .
now by condition @xmath80 of this theorem it follows that @xmath4 is monotonic decreasing for @xmath301 and so we have @xmath302 on @xmath303 and since further @xmath304 on @xmath303 so this implies @xmath305 on @xmath303 .
again , @xmath306 .
thus we get@xmath307 but as @xmath108 goes to infinity towards the right , @xmath308 .
hence , we can conclude that @xmath112 is positive and monotonic increasing and tends to @xmath281 as the paths recede to infinity . from steps
@xmath113 and @xmath114 it follows that the quantities @xmath230 and @xmath310 are bounded quantities .
thus by @xmath311 and by step @xmath118 it follows that @xmath115 is monotonic increasing to @xmath281 to the right of @xmath228 . by the construction of @xmath179
it is clear that @xmath313 in @xmath314 i.e. , to the left of @xmath228 . again from step
@xmath315 we conclude that @xmath115 ultimately becomes positive as @xmath108 moves out of @xmath228 along the curve of @xmath2 .
therefore , by the same reason given in conclusion of the theorem @xmath147 , it follows that there is one and only one path in the region @xmath219 such that@xmath316 also , by @xmath317 and the symmetry of the path it is clear that the path is closed . by the construction of @xmath179 and by step @xmath121 it is clear that equation @xmath20 has exactly two limit cycles around the origin , the second limit cycle surrounds the first one .
this completes the proof of the theorem @xmath40 .
it also follows from the proof that both the limit cycles are _ simple _ @xmath32c.f .
, remark @xmath319 that neither can bifurcate under any small @xmath320 perturbation satisfying the conditions of the theorem one can not assume that @xmath313 if @xmath321 .
we give a counter example below .
it is well known that two consecutive limit cycle can not both be stable @xmath322 . because of our choice of the function @xmath2 @xmath32negative and monotone decreasing at the right of and near the origin and infinity@xmath55 , the inner limit cycle is stable and outer limit cycle is unstable @xmath32 in reverse to those of reference @xmath323 ) $ ] .
the existence of exactly @xmath16 limit cycles is established by an easy extension of the above proof @xmath324 .
we state the theorem as follows .
a brief outline of its proof is given in the appendix .
[ new theorem n]let @xmath37 and @xmath72 be two functions satisfying the following properties .
[ c]rl@xmath71 & @xmath37 and @xmath72 are continuous ; + @xmath73 & @xmath4 and @xmath72 are odd functions and @xmath182 for @xmath54 . ; + @xmath75 & @xmath4 has @xmath16 number of positive simple zeros only at @xmath325 , @xmath326 + & where @xmath327 such that in each interval @xmath328 $ ] , + & @xmath329 , there exists @xmath330 , satisfying properties given by @xmath331 , + & such that @xmath332 where @xmath333 is the unique extremum in @xmath334 , + & @xmath335 and @xmath336 , the first local extremum in @xmath337 $ ] .
+ @xmath80 & @xmath4 is monotonic in @xmath338 @xmath339 @xmath340 and @xmath341 as @xmath12 + & monotonically for @xmath342 .
then the equation @xmath343 has exactly @xmath16 limit cycles around the origin , all are simple .
the conditions @xmath71 and @xmath80 of theorem [ new theorem ] and theorem [ new theorem n ] may be weakened following theorem [ lienard theorem extension ] .
for instance , the condition @xmath344 of theorem [ new theorem ] may be restated as [ c]rl@xmath80 & @xmath4 is monotonic increasing in @xmath345 @xmath346 and @xmath185 a number @xmath347 + & given by @xmath348 such that @xmath4 is monotonic decreasing in @xmath349 and + & nonincreasing in @xmath350 .
it is shown in section 3.3 of @xcite that the limit cycles of the autonomous system@xmath351{c}\dot{x}=y\\ \dot{y}=-x-\mu h\left ( x,\dot{x}\right ) \end{array } \right\ } \label{canonical system}\ ] ] are asymptotic to the circle @xmath352 as @xmath353 where the values of @xmath354 are the roots of the equation@xmath355 we note that this is not the lienard system .
it is the canonical phase plane for lienard equation .
the phase diagram of above system and the lienard system , however , should be similar .
here we take @xmath356 @xmath357 @xmath358 so that@xmath359@xmath360 and@xmath361 therefore , @xmath362 reduces to@xmath363 giving@xmath364 so , @xmath362 has real and distinct roots if @xmath365 i.e. , if @xmath366 , real and repeated if @xmath367 and imaginary if @xmath368 .
therefore it follows that we will get two distinct limit cycles , which are asymptotic to the circles corresponding to the above two distinct values of @xmath354 if @xmath366 .
similarly , we will get only one limit cycle when @xmath367 and no limit cycle when @xmath368 .
it can be verified that the system undergoes a saddle node bifurcation at @xmath367 .
we note that the point @xmath369 on the limit cycle in the lienard plane gets transformed to the point @xmath370 lying on the almost circular limit cycle of radius @xmath371 @xmath32 in the canonical phase plane@xmath55 where @xmath372 @xmath373 corresponds to the first limit cycle@xmath55 under the transformation@xmath374 @xmath375 and @xmath376 being the corresponding points in lienard plane and canonical phase plane with @xmath37 an even function .
we thus have@xmath377
we now present the phase diagrams of the above systems in lienard plane in the following examples for different values of @xmath378 .
these examples justify our new theorem .
we use mathematica 5.1 in constructing the examples .
[ ex1]here we consider the autonomous system @xmath11 with @xmath379 , @xmath380 @xmath25 and @xmath381 .
the phase diagram in lienard plane is shown in figure @xmath382 which does not have any limit cycle .
again we take @xmath383 in the above system .
the corresponding phase diagram is shown in figure @xmath384 .
this phase diagram also does not contain any limit cycle , but we see that the path is concentrating in a certain circular region .
[ ex one limit cycle]here we consider the autonomous system @xmath11 discussed above with @xmath367 so that @xmath385 , @xmath386 , @xmath387 and @xmath388 both are approximately equal to @xmath389 and @xmath390 .
let @xmath391 be the point of minima of @xmath4 in @xmath224 and @xmath392 be the point of maxima of @xmath4 in @xmath393 .
here @xmath2 is increasing in@xmath394 and decreasing in@xmath395 where , @xmath396 and @xmath397 . in this case
we obtain only one limit cycle as shown in figure @xmath398 . here , @xmath4 is not monotone increasing throughout the interval @xmath399 , violating the condition @xmath80 of theorem @xmath40 @xmath32since @xmath400 .
[ ex two limit cycle]we now take @xmath401 . here , @xmath387 and @xmath388 are approximately equal to @xmath402 .
the equations @xmath403 and @xmath404 both reduce to@xmath405 having real roots @xmath406 so that @xmath407 . here , @xmath408 and @xmath409 showing that @xmath410 .
next , @xmath411 , @xmath412 . here
all the conditions of theorem @xmath40 are satisfied except condition @xmath413 .
however we still get two limit cycles as shown in figure @xmath414 drawn in lienard plane .
this example and the above example show that the conditions of theorem @xmath40 are sufficient but not necessary .
[ ex4]finally we take @xmath415 . here ,
@xmath387 and @xmath388 are approximately equal to @xmath416 and @xmath417 . here ,
@xmath418 and @xmath419 showing that @xmath410 .
next , @xmath420 , @xmath421 . here ,
all the conditions of theorem @xmath40 are satisfied and so we get exactly two limit cycles .
the phase diagram in lienard plane is shown in figure @xmath422 . in lienard plane with @xmath415 , and two limit cycles in example @xmath423.,height=226 ] although in the above examples the value of @xmath191 is sufficiently small @xmath32so as to satisfy the amplitude analysis of @xmath424 our theorem should be applicable for large values of @xmath425 .
more detailed bifurcation analysis in the @xmath426 parametric plane will be considered separately .
[ ex compare 2 cycle]we now consider the function@xmath427{ll}-0.1\sin\left ( 10\pi x\right ) & 0\leq x<0.15\\ 0.01\sqrt{1-\left ( \dfrac{x-0.15}{0.01}\right ) ^{2 } } & 0.15\leq x<0.15+\dfrac{1}{\sqrt{101}}\\\begin{array } [ c]{l}0.02099503719021-\\ \hspace{0.5in}2\sqrt{0.1\left ( x-0.2395037190209989\right ) } \end{array } & x\geq0.15+\dfrac{1}{\sqrt{101}}\end{array } \right.\ ] ] then we have @xmath428 , @xmath429 , @xmath430 . here , @xmath431 though @xmath432 showing that the function is not monotone nonincreasing in @xmath433 $ ] and so it does not satisfy the condition @xmath434 of theorem @xmath435 in chapter @xmath29 in the book @xmath36 .
however for the inner limit cycle we have @xmath436 .
so , @xmath437 satisfying the conditions of theorem @xmath40 .
this example clearly shows that the theorem @xmath40 covers a larger class of functions than those covered by theorem @xmath435 in chapter @xmath29 in @xmath438 . the function @xmath4 alongwith two limit cycles are shown in figure @xmath439 . in lienard plane for example
[ ex compare 2 cycle].,height=226 ] [ ex 3 limit cycle]we now consider a different problem . here
we define@xmath427{ll}0.005 - 0.025\sqrt{1-\left ( \dfrac{x-0.048989794}{0.05}\right ) ^{2 } } & 0\leq x < a_{1}\\ -0.0008137888130718 + 0.01\sqrt{1-\left ( \dfrac{x-0.14781375}{0.05}\right ) ^{2 } } & a_{1}\leq x < a_{2}\\ 0.0009168416064002765 - 0.015\sqrt{1-\left ( \dfrac{x-0.29746094}{0.1}\right ) ^{2 } } & a_{2}\leq x <
a_{3}\\ -0.0003265987749816556 + 0.04\sqrt{x-0.3972073012751128 } & x\geq a_{3}\end{array } \right.\ ] ] where@xmath440 and@xmath201{lc}f_{+}\left ( x\right ) & x\geq0\\ -f_{+}\left ( -x\right ) & x<0 \end{array } \right.\ ] ] the function @xmath441 is obtained by matching three ellipses and a parabola successively in the intervals @xmath442 @xmath443 @xmath444 and @xmath445 such that@xmath446 where @xmath447 s are zeros of @xmath448 . the unique extremum of @xmath4 in @xmath449 , @xmath393 , @xmath450 are respectively@xmath451 we obtain three limit cycles which meet the positive @xmath293-axis at the points @xmath452 , @xmath453 , @xmath454 where@xmath455 the matching conditions @xmath456 are used to make @xmath457 with accuracy level @xmath458 .
this function is constructed in a trial and error method and numerical data with large significant digits arise in this fashion
. examples with lower significant digits and lower and higher accuracy are possible in principle . here
we get @xmath459 and @xmath460 .
the function @xmath4 satisfies all the conditions of theorem @xmath41 @xmath32for example @xmath332 etc.@xmath55 and so the existence of the above three limit cycles are ensured by this theorem , the proof of which is presented separately @xmath324 .
however , the function @xmath4 is defined in such a manner that @xmath461 implying that @xmath462 mentioned in theorem @xmath463 of @xmath30 or in theorem @xmath42 , chapter @xmath29 of the book @xmath36 does not exist and hence these theorems are not applicable for the corresponding lienard system .
the limit cycles of the lienard system in lienard plane and the graph of the function @xmath4 have been shown separately in the figure @xmath464 . to conclude , theorem @xmath463 of @xmath30 or theorem @xmath42 , chapter @xmath29 in the book @xmath36 fail to predict the existence of the exact number of limit cycles for the above function @xmath465 .
many interesting new results have been proved on the existence of an exact number of multiple limit cycles @xmath466 $ ] in the recent past .
odani has proved a sufficient condition in @xmath467 using a choice function @xmath468 which can be exploited to obtain better estimates of amplitudes of the limit cycles .
we used a straight forward method depending on the geometry of phase diagram .
we have proved a similar result with more general class of functions @xmath2 by a simpler method . in the present approach a strict monotonicity of @xmath2
is required only in the intervals @xmath469 and @xmath301 .
consequently , @xmath2 can accommodate small scale " oscillations in the interval @xmath470 .
odani , for instance , considered an @xmath465 which is not only @xmath320 but also has a unique extremum in the interval @xmath471 .
further , the theorem is valid for a more general class of the function @xmath3 .
odani s theorem however is valid only for @xmath25 .
an interesting problem will be to establish the relation between @xmath179 of our approach and the function @xmath468 .
we note that @xmath472 corresponds to the amplitude of the limit cycles .
our estimates of amplitude of the limit cycle of the van der pol equation constitute an improvement over those available in the literature @xmath473 $ ] .
examples @xmath58 & @xmath61 , on the other hand , show the difference between the present theorem and those of @xmath474 and @xmath30 .
the calculations are accurate upto the accuracy level @xmath475 .
the existence of limit cycles in a lienard system allowing discontinuity @xmath32see , for instance , @xmath476 is an interesting problem for further study . determining the shape of the limit cycles
is also left for future investigations . before closing
we note that the value @xmath179 , in general , is a function of the parameters of @xmath2 in the parametric space .
for instance , in examples @xmath477 , @xmath179 is a function of the parameters @xmath191 and @xmath378 .
the study of the variation of @xmath179 in the parametric space seems to offer interesting insights into the bifurcation and related issues of the multiple limit cycles in a lienard system .
the relationship with poincare s return map also needs to be studied .
we wish to investigate these problems in future .
here we present a brief outline of the proof of theorem @xmath41 which is given in detail in @xmath324 .
the theorem is proved by an induction method which is dependent on the _ non - trivial _ initial hypotheses that the the result holds for @xmath68 and @xmath69 .
we shall prove the theorem by showing the result that each limit cycle intersects the @xmath287axis at a point lying in the open interval @xmath478 $ ] , @xmath479 , where @xmath480 is the local minima of @xmath2 in @xmath481 $ ] . by lienard theorem and theorem @xmath40
it follows that the result is true for @xmath68 and @xmath69 .
we shall now prove the theorem by induction .
we assume that the theorem is true for @xmath482 and we shall prove that it is true for @xmath483 .
we prove the theorem by taking @xmath484 as an odd @xmath485 integer so that @xmath486 is even .
the case for which @xmath484 is even can similarly be proved and so is omitted .
it can be shown that @xmath487 , @xmath115 changes its sign from @xmath485 to @xmath488 as @xmath108 moves out of @xmath489 along the curve @xmath89 and hence vanishes there due to its continuity and generates the first limit cycle around the origin .
next , in theorem @xmath40 we see @xmath115 again changes its sign from @xmath488 to @xmath485 and generates the second limit cycle around the first .
also , we see that for existence of second limit cycle we need the existence of the point @xmath179 , which we denote here as @xmath490 . since by induction hypothesis
the theorem is true for @xmath482 , so it follows that in each and every interval @xmath491 $ ] , @xmath492 the system @xmath11 has a limit cycle and the outermost limit cycle cuts the @xmath287 axis somewhere in @xmath493 .
also @xmath115 changes its sign alternately as the point @xmath108 moves out of @xmath447 s , @xmath494 .
since @xmath486 is even , it follows that @xmath115 changes its sign from @xmath485 to @xmath488 as @xmath108 moves out of @xmath495 along the curve @xmath89 .
since there is only one limit cycle in the region @xmath496 , so it is clear that @xmath115 must change its sign from @xmath488 to @xmath485 once and only once as @xmath108 moves out of @xmath497 along the curve @xmath89 .
also it follows that once @xmath115 becomes @xmath485 , it can not vanish further , otherwise we would get one more limit cycle , contradicting the hypothesis so that total number of limit cycle becomes @xmath484 .
we now try to find an estimate of @xmath140 for which @xmath115 vanishes for the last time .
we shall now prove that the result is true for @xmath483 and so we assume that all the hypotheses or conditions of this theorem are true for @xmath483 .
so , we get one more point @xmath498 and another root @xmath499 , ensuring the fact that @xmath115 vanishes as @xmath108 moves out of @xmath500 through the curve @xmath89 , thus accommodating a unique limit cycle in the interval @xmath501 $ ] . by the result
discussed so far it follows that @xmath120 when @xmath140 lies in certain suitable small right neighbourhood of @xmath502 .
we shall prove that @xmath115 ultimately becomes @xmath488 and remains @xmath488 as @xmath108 moves out of @xmath503 along the curve @xmath89 generating the unique limit cycle and hence proving the required result for @xmath483 .
we draw straight line segments @xmath504 , @xmath505 , passing through @xmath506 and parallel to @xmath293-axis as shown in figure @xmath507 . for convenience
, we shall call the points @xmath508 as @xmath509 respectively .
we write the curves@xmath510@xmath511@xmath510so that@xmath510@xmath512@xmath513and@xmath514 which is used in place of the function in @xmath311 .
the rest of the proof are analogous to that of theorem @xmath40 and proved separately in @xmath515 . | we present a simpler proof of the existence of an exact number of one or more limit cycles to the lienard system @xmath0 , @xmath1 , under weaker conditions on the odd functions @xmath2 and @xmath3 as compared to those available in literature .
we also give improved estimates of amplitudes of the limit cycle of the van der pol equation for various values of the nonlinearity parameter .
moreover , the amplitude is shown to be independent of the asymptotic nature of @xmath4 as @xmath5 . _
key words and ph__r__ases : _ * * autonomous system , lienard equation , limit cycle . _
2010 mathematics subject classification : _ * * 34c07 , 70k05 |
we begin with a brief review of the dispersionless limits of the toda and pfaff lattice hiearchies ( see for examples @xcite ) . here
the toda hierarchy is the 1-dimensional one with the parameters @xmath5 .
these limits are found from the equations for the @xmath0-functions of the lattice hierarchies by subsituting an asymptotic expansion for the logarithm of the @xmath0-functions in the small @xmath6 limit .
we also describe the connection of the @xmath0-functions of these hierarchies with the partition functions of the unitary , orthogonal , and symplectic ensembles of random matrix models with the matrix size @xmath7 ( or @xmath8 ) related to @xmath9 ( see for examples @xcite ) .
in particular the asymptotic expansions of the free energy , the logarithm of the partition functions , as the size of the matrices approaches infinity satisfy the underlying assumption for the asymptotic expansion of the logarithm of the @xmath0-functions of the toda and pfaff lattice hierarchies .
then the leading order terms of the asymptotic expansions satisfy the dispersionless toda and pfaff lattice hierarchies , which are written in terms of the two - point functions denoted by @xmath1 ( see for examples @xcite ) .
there is a rich literature dealing with dispersionless integrable systems including the kp and toda lattice hierarchies , which mainly involve studying algebraic and complex analytic aspects of the equations and/or discussing applications to 2-dimensional topological field theories ( see , for examples , @xcite ) .
however , it seems that there is no paper which connects directly the dispersionless integrable systems with combinatorial problems associated with random matrix ensembles .
the present paper is to deal with this connection for combinatorial problems such as counting ribbon graphs on a compact surface of genus zero , and give a universality result for the dispersionless integrable systems which is an analogue of that for the random matrix ensembles .
the toda lattice equation ( the first member of the toda lattice hierarchy ) is given by @xmath10 \displaystyle{\frac{\partial b_n}{\partial t_1}}&=a_n - a_{n-1}\ , .
\qquad n=1,2,\ldots.\ ] ] there exists a sequence @xmath11 of @xmath0-functions with @xmath12 which generate the @xmath13 by the formulas @xmath14 we may then write the toda lattice equation in the hirota bilinear form , @xmath15 here @xmath16 is the usual hirota derivative , i.e. for a variable @xmath17 which is a flow - parameter for the @xmath18-th member of the toda lattice hierarchy , @xmath19 is defined by @xmath20 note that the hitota equation ( [ toda - hirota ] ) of the toda lattice gives @xmath21 the toda lattice is expressed in the lax representation with a tridiagonal semi - infinite matrix @xmath22 by @xmath23,\qquad b_1=[l]_{-}\,,\ ] ] where @xmath24_-$ ] is the strictly lower triangular part of the matrix @xmath22 given by @xmath25 the lax matrix is also written in terms of a shift operator @xmath26 , that is , with the eigenvector @xmath27 , we have @xmath28 the hierarchy of the toda lattice is defined by @xmath29,\qquad b_k=[l^k]_-,\qquad k=1,2,3,\ldots\,.\ ] ] the @xmath0-functions are then the functions of infinite variables , i.e. @xmath30 with @xmath31 .
then it is well - known that each @xmath32-function also satisfies the kp hierarchy @xcite : @xmath33 where @xmath34 and @xmath35 with @xmath36 are the elementary symmetric functions of @xmath37 defined by @xmath38 in particular , the first equation with @xmath39 gives @xmath40 which is the kp equation , that is , the function @xmath41 for each @xmath2 satisfies @xmath42 with @xmath43 and @xmath44 .
it is also well - known that the @xmath32-functions of the toda lattice hierarchy satisfy the following set of equations ( see for example @xcite ) : @xmath45 the first equation with @xmath46 gives @xmath47 this equation with ( [ toda - hirota ] ) gives the nonlinear schrdinger equation ( with the change @xmath48 ) , i.e. @xmath49 with @xmath50 and @xmath51 .
thus , the nonlinear schrdinger equation is the second member of the toda lattice hierarchy .
we now briefly summarize the dispersionless limit of the toda hierarchy : the key ingredient in the dispersionless limit is to introduce a free energy , denoted @xmath52 , @xmath53 where @xmath6 is a small parameter , and @xmath54 represents the slow variables with @xmath55 for @xmath56 .
note in particular that the limit @xmath57 gives a continuous limit of the lattice structure , that is , the lattice spacing has the order @xmath58 and the limit introduces the continuous variable @xmath59 , @xmath60 the free energy @xmath61 is calculated from the limit , @xmath62.\ ] ] then the dispersionless toda hierarchy can be written in terms of the second derivatives , @xmath63 the @xmath64 play an important role for the dispersionless integrable systems , and they are sometimes referred to as the _ two point functions _ of the corresponding topological field theory ( see for examples @xcite ) .
in the limit @xmath57 , the @xmath65 variables in the toda lattice become @xmath66 \displaystyle{b_n(t)=\frac{\partial}{\partial t_1}(\log\tau_{n}-\log\tau_{n-1})\longrightarrow f_{01 } } \ , . \end{array}\right.\ ] ] then the hirota form of the toda lattice equation ( [ toda - hirota ] ) becomes @xmath67 which is called the dispersionless toda ( dtoda ) equation .
likewise the kp equation ( [ kp ] ) gives @xmath68 which is the dispersionless kp ( dkp ) equation .
also the dispersionless limit of is @xmath69 one should also note that in the dispersioless limit , the spectral problem @xmath70 with the lax operator and the eigenvector @xmath71 leads to an algebraic equation ( the spectral curve of the dtoda equation ) , @xmath72 where @xmath73 is a quasi - momentum given by @xmath74 .
this is a spectral curve of genus @xmath75 , and the dtoda hierarchy defines a integrable deformation of the curve ( see lemma [ tridiagonality ] ) . here
the eigenvector @xmath76 is assumed to be in the wkb form as @xmath57 , @xmath77 although the function @xmath78 plays a key role for the dispersionless theory , we will not use it in this paper .
we just mention that the @xmath78-function can be expressed in the form ( see for example @xcite ) , @xmath79 the dtoda hierarchy may be formulated by the form , @xmath80 \displaystyle{e^{-d(\lambda ) d(\mu)f}=1-\frac{e^{f_{00}}}{\lambda\mu}e^{(d(\lambda)+d(\mu))f_0 } } \end{array}\right.\ ] ] where @xmath81 this is obtained by the reduction from the 2-dimesional dtoda hierarchy in @xcite with the constraints @xmath82 for all @xmath18 , and the toda hierarchy with those constraints is referred to as the 1-dimensional toda hierarchy
. then one can show that the dtoda hierarchy leads to the curve ( see lemma [ tridiagonality ] below ) . we will further discuss the detailed structure of the hierarchy in section [ s : dtoda ] , where we show that indeed gives the dispersionless limits of and .
a special class of solutions to the toda lattice equations defines the partition functions of the unitary ensembles of random matrices .
explicitly the @xmath0-functions are taken to be @xmath83 where @xmath84 and @xmath85 is a polynomial of even degree ( e.g. @xmath86 for the gaussian ensemble ) .
this function represents integration over the eigenvalues of the random matrices of the unitary ensemble , and its taylor expansion in @xmath87 contains the moments of traces of powers of the matrix ( see for example a common reference @xcite ) .
a ribbon graph is a one complex embedded on an oriented genus @xmath88 surface , such that the complement of the complex is a disjoint collection of sets homeomorphic to discs .
we may think of a ribbon graph as a collection of discs ( vertices ) and ribbons ( edges ) such that the ribbons are glued together preserving the local orientations of the discs in other words the ribbons lay flat on the oriented surface .
our ribbon graphs will be labeled in the following way : the vertices are labeled so that they are distinct , and the edges emnating from each vertex are labeled as distinct as well . we also define the degree of the vertex as the umber of edges attached to the vertex .
we represent ribbon graphs for this paper as graphs drawn on an oriented surface ( we only consider a compact surface of genus zero ( i.e. a sphere ) in this paper , and refer a ribbon graph on the sphere as a ribbon graph of genus zero ) . in figure [ fig : vertex ] , we illustrate one vertex of degree 8 and two vertices of degrees of 3 and 5 on a sphere . ,
width=377 ] a fundamental result of random matrix theory is that for a quadratic potential @xmath85 , @xmath89 posseses an asymptotic expansion in even powers of @xmath90 , whose terms give generating functions partitioning ribbon graphs by genus : [ biz - thm ] for the gaussian unitary ensemble ( gue ) , we have @xmath91 = \sum_{g\geq 0 } e_g(\mathbf{t } ) n^{2 - 2g}\ ] ] where @xmath92 the coefficient @xmath93 gives the number of the connected ribbon graphs with @xmath94 labeled vertices of degree @xmath18 for @xmath95 on a compact surface of genus @xmath88 . note that theorem [ biz - thm ] does not contain our @xmath96 variable in large @xmath7 limit . the @xmath59-variable can be inserted naturally by taking the @xmath0-function in the form with @xmath97 ( or @xmath98 ) , @xmath99 then the free energy is given by @xmath100=\log\left [ z_n^{(2)}\left ( \frac{n}{2 } \lambda^2 ; n \mathbf{t}\right ) \big/ z_n^{(2)}\left(\frac{n}{2 } \lambda^2 ; \mathbf{0}\right ) \right ] + \log\left[z_n^{(2)}\left(\frac{n}{2 } \lambda^2 ; \mathbf{0}\right)\right]\,.\ ] ] we note here that the scaling @xmath101 leads to @xmath102 where @xmath103 ( which is the so called penner scaling ) .
this expression then has the asymptotic expansion including @xmath59 , @xmath104 & = \sum_{g\geq 0 } e_g ( \mathbf{\hat{t } } ) n^{2 - 2 g } + c(t_0;n)\\ & = \sum_{g\geq 0 } t_0^{2 - 2 g } e_g(\mathbf{\hat{t } } ) \ , n^{2 - 2 g } + c(t_0;n)\,,\end{aligned}\ ] ] where @xmath105 $ ] which can be computed directly ( see appendix a ) .
note that this asymptotic expansion agrees ( at leading order ) with the assumption ( [ asymptotic - assumption ] ) ; and thus gives a class of solutions of the toda lattice satisfying this assumption .
then the leading order of this asymptotic expansion gives a solution of the dispersionless toda equation , @xmath106\\ & = t_0 ^ 2 e_0 ( \hat{\mathbf{t } } ) + c_0(t_0)\,,\end{aligned}\ ] ] where @xmath107 is the leading order of @xmath108 for large limit of @xmath7 , and as shown in appendix b , it is given by @xmath109 this formula has also been found directly from the dtoda equations in a connection to the topological field theory ( see ( 8.6 ) of p.209 in @xcite ) .
in particular , theorem [ biz - thm ] implies that the two point function @xmath110 for @xmath111 represents the number of connected ribbon graphs with two vertices of degrees @xmath2 and @xmath3 on a sphere @xmath112 , i.e. @xmath113 where @xmath114 s in @xmath115 are at the @xmath2- and @xmath3-th places .
we will give an explicit formula for @xmath116 in section [ combin - dtoda ] .
we also attach an enumerative meaning to derivatives with respect to @xmath59 : [ comb - f0 ] each derivative @xmath117 for @xmath118 , corresponds to counting the number of connected ribbon graphs with a vertex of degree @xmath119 and a marked face on a sphere .
we start with our relation @xmath120 one then notes that the number of vertices of a ribbon graph with vertices of type @xmath121 is @xmath122 and the number of edges is @xmath123 . if the ribbon graph is of genus 0 , then by the euler characteristic formula , the number of faces is @xmath124 so that by inserting this equation into ( [ t0meaning ] ) , we find @xmath125 from which one sees that differentiating with respect to @xmath59 is related to counting ribbon graphs with marked faces in the enumerative part ( i.e. except for the contributions coming from @xmath126 ) . for example : @xmath127 with @xmath128 is the number of ribbon graphs with a single vertex of degree @xmath119 and one of the @xmath129 faces marked .
these numbers will also be found in theorem [ fformula ] as a consequence of the dtoda hierarchy .
the pfaff lattice hierarchy is normally referred to as the dkp hierarchy ( the d " stands for the d - type lie group which is the symmetry group for the hierarchy @xcite ) .
the first member of the dkp hierarchy , called the dkp equation , is given by the set of equations @xmath130
\displaystyle{2\frac{\partial v^{\pm}}{\partial t_3}+\frac{\partial^3 v^{\pm}}{\partial t_1 ^ 3}+6u\frac{\partial v^{\pm}}{\partial t_1 } \mp 3\left(\frac{\partial^2 v^{\pm}}{\partial t_1\partial t_2}+2 v^{\pm}\int^{t_1}\frac{\partial u}{\partial t_2}\,dt_1\right)=0 } \ , .
\end{array}\right.\ ] ] the left hand side of the first equation is just the kp equation , and the right hand side gives a coupling term with the field @xmath131 . because of this
, the dkp equation is sometimes called a coupled kp equation @xcite . in terms of the @xmath0-functions , @xmath132 and @xmath131
are expressed as @xmath133 the dkp equation is then given by @xmath134 \displaystyle{(2d_3+d_1 ^ 3\mp 3d_1d_2)\tau_{n\pm 1}\cdot\tau_{n}=0}\ , .
\end{array}\right.\ ] ] it is known that the @xmath0-functions are given by pfaffians @xcite .
since the equation with the set of @xmath0-functions has a lattice structure ( i.e. @xmath135 is determined by the previous @xmath0-functions with @xmath12 ) , we call it the _ pfaff lattice _ in this paper . this choice of name is also convenient for this paper as we wish to compare the pfaff lattice , as given in @xcite , and the toda lattice .
the pfaff lattice hierarchy is then given by @xcite @xmath136 \quad h_{k+3}(\mp \tilde{\mathbf{d}})\,\tau_{n\pm 1}\cdot\tau_{n}=0 , \end{array } \qquad k=0,1,2,\ldots.\right.\ ] ] in a similar manner as in the toda lattice case , one can also consider the dispersionless limit of the pfaff lattice equation .
the dispersionless pfaff ( dpfaff ) lattice equation is then given by @xmath137 \displaystyle{2f_{03}+f_{01}^3 + 6f_{01}f_{11}-3f_{01}f_{02}-6f_{12}=0}\ , .
\end{array}\right.\ ] ] here we have used the limits , @xmath138 and @xmath139 as @xmath57 . in @xcite , takasaki formulates the dispersionless dkp ( pfaff lattice ) hierarchy by taking the dispersionless limit in the differential fay identities of the pfaff lattice hierarchy : @xmath140 \displaystyle { e^{-d(\lambda)d(\mu)f}\left ( 1-\frac{(d(\lambda)-d(\mu))f_1}{\lambda-\mu}\right)= 1-\frac{e^{f_{00}}}{\lambda^2\mu^2}e^{(d(\lambda)+d(\mu))f_0}}\ , . \end{array}\right.\ ] ] this equation is somewhat similar to the dtoda hierarchy , and in fact , this similarity is one of our motivations to consider the universality between dpfaff and dtoda hierarchies .
two special classes of solutions of the pfaff lattice hierarchy are given by the partition functions of the orthogonal and symplectic ensembles of random matrices ( see for examples @xcite ) .
we denote those partition functions by @xmath141 with @xmath142 for the orthogonal ensembles and @xmath143 for the symplectic ensembles . with this notation ,
the partition function ( or @xmath0-function ) for the toda hierarchy is denoted by @xmath144 , i.e. @xmath145 . in this case , the partition function is taken to be latexmath:[\[z_{2n}^{(1)}(v_0(\lambda ) ; \mathbf{t } ) = \int_{\mathbb{r}}{d\lambda_1}\cdots\int_{\mathbb{r}}d\lambda_{2n}\,\prod_{i < j } where @xmath84 and @xmath85 is a polynomial of even degree .
this integral represents integration over the eigenvalues of the random matrices in the orthogonal ensemble , and as in the case of unitary ensemble , its taylor expansion in @xmath87 gives the moments of traces of powers of the matrix . this partition function
is also related to the combinatorial problem of counting ribbon graphs on a compact surface . in this case
we will consider mbius graphs , which are defined to be one - complexes embedded in an unoriented surface of euler characteristic @xmath147 such that the complement of the complex is a collection of sets homeomorphic to discs .
we may think of a mbius graph as a ribbon graph where the ribbons are allowed to be glued together with a twist , i.e. reversing the local orientation of the vertices .
our mbius graphs are labeled in the same manner as the ribbon graphs .
we introduce mbius graphs here for completeness sake , however in this paper we consider only those mbius graphs of euler characteristic @xmath148 , such graphs are equivalent , up to a choice of local orientations at the vertices , to ribbon graphs embedded in the sphere ( see lemma [ mobius - is - ribbon ] ) , so we will not encumber our discussion here with examples of mbius graphs .
one may see @xcite for a discussion of mbius graphs with euler characteristics less than 2 . for a quadratic potential @xmath85
, we have an asymptotic expansion for @xmath149 of the same basic character as that of the unitary ensembles : [ ghj ] for the gaussian orthogonal ensemble ( goe ) , we have @xmath150 = \sum_{\chi\leq 2 } e^{(1)}_\chi(\mathbf{t } ) ( 2n)^{\chi}\ ] ] where @xmath151 and @xmath152 is the number of connected mbius graphs with @xmath94 labeled vertices of degree @xmath18 for @xmath153 on a compact surface of euler characteristic @xmath147 .
we then take our @xmath0-function in the form with @xmath154 , @xmath155 where @xmath156 is given by the penner scaling , i.e. @xmath157 with @xmath158 .
this expression then has the asymptotic expansion @xmath159 & = \sum_{\chi \leq 2 } e^{(1)}_\chi ( { \hat{\mathbf{t } } } ) ( 2n)^{\chi } + c^{(1)}(t_0;n ) \\ & = \sum_{\chi \leq 2 } ( 2 t_0)^{\chi } e^{(1)}_\chi ( { \hat{\mathbf{t}}})n^{\chi } + c^{(1)}(t_0;n)\,,\end{aligned}\ ] ] where @xmath160 $ ] .
note that the form of this asymptotic expansion agrees ( at leading order ) with the assumption ( [ asymptotic - assumption ] ) . for future reference , for this initial condition
, the leading order of the expansion gives a solution of the dpfaff hierarchy , @xmath161\\ & = 4 t_0 ^ 2 e_2^{(1 ) } ( \hat{\mathbf{t}})+c^{(1)}_0(t_0 ) \,,\end{aligned}\ ] ] where @xmath162 is the leading order of @xmath163 . in appendix b
, we explicitly calculate @xmath162 , and it gives @xmath164 where @xmath165 given in ( [ c0toda ] ) .
the relation with @xmath166 of the dtoda hierarchy is a consequence of the universality , which will be discussed in section [ dtoda - dpfaff ] . with @xmath167
, we have the expansion formula for the solution of the dpfaff hierarchy corresponding to the gaussian orthogonal ensemble , @xmath168 this formula then gives the combinatorial meaning of the solution of the dpfaff hierarchy ( also notice the similarity with the formula ( [ ufree ] ) for the dtoda hierarchy . in this case
, the partition function is given by @xmath169 where @xmath84 and @xmath85 is a polynomial of even degree .
this function represents integration over the eigenvalues of the random matrices in the symplectic ensemble , and the moments of traces of powers of the matrix are obtained in its taylor expansion in @xmath87 .
then the corresponding combinatorial result is give by the following theorem : [ mw ] for the gaussian symplectic ensemble ( gse ) , we have @xmath170 = \sum_{\chi\leq 2 } e^{(1)}_\chi(\mathbf{t } ) ( -2n)^{\chi}\,,\ ] ] where @xmath171 is defined as in theorem [ ghj ] .
this theorem can also be found from theorem 4.1 of @xcite , in particular the @xmath172 appearing in that theorem leads to the @xmath173 in the above formula of theorem [ mw ] .
there is a duality in the gaussian orthogonal and symplectic ensembles .
theorems [ ghj ] and [ mw ] give the @xmath174 duality between the partition functions of the gaussian orthogonal and symplectic ensembles of random matrices .
in general the even terms of the asymptotic expansions agree , while the odd terms have opposite signs .
see the discussions in @xcite .
the @xmath0-function in this case is defined by @xmath175 where @xmath176 is the penner scaling defined in the same form as in the case of orthogonal ensemble , i.e. @xmath177 with @xmath167 .
then the free energy is given by @xmath178&= \log\left [ z_n^{(4)}\left ( \frac{n}{4 } \lambda^2 ; n \mathbf{t}\right ) \big/ z_n^{(4)}\left ( \frac{n}{4 } \lambda^2 ; \mathbf{0}\right ) \right ] + c^{(4)}(t_0;n ) \\ & = \log\left [ z_n^{(4)}\left ( \frac{n}{2 } \lambda^2 ; 2 n \hat{\mathbf{t}}\right)\big/ z_n^{(4)}\left ( \frac{n}{2 } \lambda^2 ; 2n \hat{\mathbf{t}}\right ) \right ] + c^{(4)}(t_0;n ) \\ & = \sum_{\chi \leq 2 } ( -2t_0)^{\chi } e^{(1)}_\chi(\hat{\mathbf{t } } ) n^{\chi } + c^{(4)}(t_0;n)\,,\end{aligned}\ ] ] where @xmath179 $ ] .
in the limit @xmath180 , the leading order of the free energy is given by @xmath181\\ & = ( 2t_0)^2e_2^{(1)}(\hat{\mathbf{t}})+c^{(4)}_0(t_0)\,,\nonumber \end{aligned}\ ] ] where @xmath182 is the leading order of @xmath183 , and in appendix b it is found to be the same as @xmath162 .
this implies that with ( [ ofree ] ) we have the duality of the free energies , i.e. @xmath184 in section [ dtoda - dpfaff ] , we further discuss the universality for those free energies corresponding to the gue , goe and gse random matrix theories .
the paper is organized as follows : in section [ s : dtoda ] , we first show that the dtoda hierarchy is determined by variables @xmath185 and @xmath186 ( proposition [ f00f01 ] ) .
then we discuss some properties of @xmath64 ( proposition [ parity - proposition ] ) .
in particular , we show that @xmath187 are related to the catalan numbers , which gives a combinatorial meaning of the dtoda hierarchy ( proposition [ catalantoda ] ) . in section [ combin - dtoda ]
, we first derive the explicit formulae for the two point functions @xmath1 for the dtoda hierarchy under the conditions of @xmath188 and @xmath189 ( theorem [ fformula ] ) , and then discuss the combinatorial description of @xmath64 as the enumeration of ribbon graphs on a sphere ( proposition [ fmngraph ] ) .
we also show that the combinatorial meaning of @xmath64 can be directly obtained from the dtoda hierarchy ( theorem [ gfcomb ] ) . in section [ dpfaffhierarchy ]
, we describe the dpfaff hierarchy .
in particular , we show that under the assumption of the special dependency of the free energy @xmath52 in the variables @xmath190 , the dpfaff hierarchy can be reduced to the dtoda hierarchy ( proposition [ sr - todaconnection ] ) .
this corresponds to the dispersionless limit of the pfaff lattice hierarchy restricted to symplectic matrices @xcite . in section [ dtoda - dpfaff ] ,
we first show the universality among the free energies for the gue , goe and gse in the leading order of large @xmath7 expansions ( proposition [ uosrelations ] ) .
then we show in general that the solution of the dtoda hierarchy also satisfies the dpfaff hierarchy under the rescaling of the two point functions @xmath64 ( theorem [ dkp - toda ] ) .
this means that the dpfaff hierarchy contains the dtoda hierarchy , and shows the universality of the dtoda hierarchy at the leading orders among the unitary , orthogonal and symplectic ensembles in random matrix theory . in appendix
a , we give a brief introduction of the dkp theory , and derive the dkp hierarchy ( [ cauchykernel ] ) which plays an important role in this paper .
we also give in appendix b explicit computations of the terms @xmath191 for @xmath192 and @xmath193 .
in this section , we first show that the dtoda hierarchy defines a deformation of the algebraic curve given by , and as a result we show that the two point functions @xmath1 are determined by @xmath186 and @xmath185 .
we also discuss a connection of the dtoda hierarchy with the dkp hierarchy , and define the faber polynomials as the generators of the higher members of the dkp hierarchy .
these results will then be used in section [ combin - dtoda ] to derive explicit formulae of @xmath64 solving the two - vertex problem .
let us first define the function @xmath73 , @xmath194 this corresponds to the equation @xmath195 with the @xmath78-function given by ( [ s ] ) .
then the dtoda hierarchy is defined on the dtoda curve : [ tridiagonality ] the dtoda hierarchy gives a deformation of the algebraic curve of genus 0 given by @xmath196 eliminating @xmath197 from two equations in , we obtain @xmath198 we then separate the variables , @xmath199 with the asymptotic condition , @xmath200 , we obtain the curve . with this lemma [ tridiagonality ] , we define in this paper the dispersionless toda ( dtoda ) hierarchy in the form , @xmath201 \displaystyle{\lambda = p(\lambda)+f_{01}+\frac{e^{f_{00}}}{p(\lambda ) } \quad { \rm with}\quad p(\lambda)=\lambda e^{-d(\lambda)f_0}}\ , .
\end{array}\right.\ ] ] in particular one should note through the paper that the dtoda curve ( the second equation ) on @xmath202-plane plays a central role for the combinatorial description of the dtoda hierarchy .
we now note that the first equation in implies : [ f0k ] each @xmath64 with @xmath203 can be determined by the set @xmath204 .
the first equation with @xmath205 gives @xmath206 the right hand side depends only on @xmath207 for @xmath208 .
also , setting @xmath209 and expanding for large @xmath210 , the first equation gives @xmath211 & = e^{-d(\lambda)^2f}=\sum_{n=0}^{\infty}\frac{1}{n}h_n(-\hat{\mathbf{z}})\,,\end{aligned}\ ] ] where @xmath212 with @xmath213 , and @xmath214 with @xmath215 for @xmath216 defined by @xmath217 comparing the coefficients of the power @xmath218 , we obtain @xmath219 this set of equations is called the dipersionless hirota equation for the dtoda hierarchy , which gives a half set of the dispersionless hirota equations ( see for example @xcite , and the other half is given by below ) .
the first equation with @xmath220 then gives the dispersionless toda equation , @xmath221 for @xmath222 , we obtain @xmath223 thus @xmath224 and @xmath225 are determined by only @xmath185 and @xmath186 .
in fact , we have from the second equation ( dtoda curve ) that [ f00f01 ] the two point functions @xmath1 of the dtoda hierarchy is determined by @xmath185 and @xmath186 only .
the curve gives @xmath226 expanding this with large @xmath210 , we have @xmath227 then one can see that @xmath228 is determined by the previous @xmath207 for @xmath229 . for examples
, we have for @xmath230 and @xmath231 , @xmath232 then from lemma [ f0k ] , we conclude that all @xmath64 for @xmath203 are determined by @xmath185 and @xmath186 ( or @xmath186 and @xmath233 ) . from the dtoda curve
, @xmath73 can be explicitly calculated as follows : [ catalantoda ] the inverse of the dtoda curve for the case @xmath234 as @xmath235 is give by @xmath236 where @xmath237}c_k\binom{n}{2k}f_{01}^{n-2k}f_{11}^{k+1}\ , , \qquad { \rm with}\quad c_k=\frac{1}{k+1}\binom{2k}{k}\,.\ ] ] note here that @xmath238 and @xmath239 is the @xmath18-th catalan number with @xmath240 . from the curve ,
first we have @xmath241 solving this for @xmath242 with the asymptotic condition @xmath243 as @xmath235 , we have @xmath244 & = \lambda - f_{01}-\frac{1}{2}\left(\lambda - f_{01}-\sqrt{(\lambda - f_{01})^2 - 4f_{11}}\,\right)\,.\end{aligned}\ ] ] noting the last term can be expressed as @xmath245 thus we have @xmath246 then the coefficients @xmath247 can be found by @xmath248 using the expansion @xmath249 , we obtain the formula .
the appearance of the catalan numbers indicates some connections of combinatorial problems to the dtoda hierarchy .
this is one of the main motivations of the present study .
here we consider the second equation of , that is , in terms of @xmath73 , @xmath250 this equation is also known as the dkp hierarchy , and @xmath73 in the dkp hierarchy is given by @xmath251 ( see appendix a , and also for examples @xcite ) . for the dtoda hierarchy , @xmath252 can also be expressed in a similar form .
this is obtained from an interesting connection with the faber polynomials : [ faberfg ] let @xmath210 be a laurent series in @xmath242 given by @xmath253 then the faber polynomials @xmath254 of degree @xmath2 in @xmath242 are defined by @xmath255_{\ge 0 } \quad n=0,1,2,\ldots,\ ] ] where @xmath256_{\ge 0}$ ] represents the polynomial part of @xmath257 in @xmath242 .
the polynomial @xmath258 has the laurent series in @xmath210 , @xmath259 where @xmath260 are the functions of @xmath261 .
those @xmath260 are related to the grunsky coefficients @xmath262 . then gives the generating function of the faber polynomials with @xmath263 ( see remark [ fg ] below for a further discussion ) . in the dkp theory , the faber polynomials give the generators of the flows in the hierarchy ( see appendix a ) . then with ( [ cauchykernel ] )
, we have [ faber ] the faber polynomials can be expressed as @xmath264 and @xmath265 .
we first note that @xmath266 taking the derivative with respect to @xmath210 , we have @xmath267 then for @xmath56 , we have @xmath268 & = \oint_{p=\infty}\frac{dp}{2\pi i}\frac{\lambda(p)^n}{p - p(\mu)}= \left[\lambda(p(\mu))^n\right]_{\ge 0}.\end{aligned}\ ] ] this completes a proof .
we now derive the other set of the dispersionless hirota equations : taking the limit @xmath269 of the equation in ( [ cauchykernel ] ) , we have @xmath270 with @xmath252 , we have @xmath271 where @xmath272 with @xmath273 defined by @xmath274
this then gives the other half set of the dispersionless hirota equation for the dtoda hierarchy ( the first half set is given by ( [ dtoda1 ] ) ) , @xmath275 which gives the dispersionless limit of the hierarchy .
the first equation of ( [ dtoda2 ] ) with @xmath230 gives @xmath276 with proposition [ faber ] with the case @xmath222 , i.e. @xmath277 , the function @xmath73 can also be expressed by @xmath278 then note that that the dhirota equation of the dkp hierarchy is derived from the same equation @xmath279 with @xmath280 , @xmath281 which is the dispersionless limit of . here
recall that @xmath214 with @xmath282 is given by @xmath283 .
the first equation with @xmath284 gives the dkp equation , @xmath285 we also mention the positivity of @xmath64 : [ parity - proposition ] the coefficients @xmath64 satisfy the following properties : * if @xmath286 for all @xmath287 , then @xmath288 for all @xmath203 .
* if @xmath289 for all @xmath118 and all others @xmath290 , then @xmath291 for all @xmath292 with @xmath293odd . with @xmath294
, we have @xmath295 then taking the logarithm and the differentiation with respect to @xmath210 of this equation , we have @xmath296 this shows the property ( a ) , i.e. @xmath297 implies @xmath298 . the second property ( b ) can be obtained from @xmath299 .
this equation gives @xmath300 which shows that if all @xmath289 , then @xmath301 then from ( a ) , we conclude @xmath291 if @xmath293odd .
[ fg ] the faber polynomials @xmath254 appear in a classical complex function theory : the bieberbach conjecture sates that if @xmath302 is a univalent function on @xmath303 with the expansion @xmath304 , then the coefficients @xmath305 satisfy @xmath306 .
let @xmath307 be the function defined by @xmath308 which is univalent for @xmath309 .
then the faber polynomials @xmath254 appear in a expansion , @xmath310 with the expansion of @xmath311 for large @xmath312 , @xmath313 grunsky defined the grunsky coefficients @xmath314 .
then grunsky showed that his coefficients should satisfy a sequence of inequalities in order that @xmath88 is univalent on @xmath309 .
in particular , an explicit formula for the gransky coefficient was found in the paper by schur in terms of the coefficients @xmath305 of @xmath315 @xcite . in the section [ explicitfnm ]
, we find an explicit formula of @xmath1 for the case where the cefficients are given by the catalan numbers and discuss their combinatorial meaning ( our derivation is independent from the calculation used in @xcite ) . from ( [ cauchykernel ] ) and ( [ cnm ] )
, one can see that the grunsky coefficients @xmath314 are related to our two point functions , that is , @xmath316 .
the dtoda hierarchy then defines an _ integrable _ deformation with an infinite number of parameters @xmath317 for the univalent function @xmath302 , so that the grunsky coefficients are given by the second derivatives of a function @xmath52 , the free energy , i.e. @xmath318 thus a most important consequence of the dtoda hierarchy is to show the existence of a unique function @xmath52 , the free energy .
the appearance of the catalan numbers in the dtoda hierarchy ( see proposition [ catalantoda ] ) directly demonstrates a combinatorial meaning of the functions @xmath1 . in particular , if @xmath319 and @xmath320 , then we have , from proposition [ catalantoda ] , @xmath321 which gives @xmath322 in general , the two point functions @xmath1 have the following combinatorial interpretation : one sees from the connection to random matrices discussed in section [ unitarytoda ] that for the particular choice of @xmath323 the two point function with @xmath324 , @xmath325 is the number of genus 0 connected ribbon graphs with a vertex of degree @xmath2 and a vertex of degree @xmath3 ( see theorem [ biz - thm ] ) . here
the degree of the vertex represents the number of edges ( ribbons ) attached to the vertex . for example
, the @xmath326 gives the number of connected graphs with a vertex with a single edge and a vertex with @xmath327 edges on a sphere ( recall that the number of ways to connect @xmath119 edges of a single vertex is given by the catalan number @xmath239 ) . in this section
we compute a remarkable closed formula for the @xmath328 in the case with @xmath329 and @xmath330 . also recall that differentiation with respect to @xmath59 has the meaning of counting ribbon graphs with a marked face .
let us first recall proposition [ parity - proposition ] ( b ) : if @xmath331 and @xmath332 , we have @xmath333 also from proosition [ catalantoda ] , if @xmath334 and @xmath189 , we have @xmath289 and @xmath335 .
so we calculate only @xmath1 with @xmath293even and positive
. then we state the following theorem which gives an explicit formula of @xmath1 for those cases of @xmath3 and @xmath2 .
the method to find the formula is based on the dispersionless toda hierarchy .
[ fformula ] the two point functions @xmath336 for the dtoda hierarchy ( [ dtodahierarchy ] ) with @xmath188 and @xmath189 ( i.e. @xmath337 ) are given by @xmath338 \displaystyle{f_{2j+1,2k+1}}&=&{(2j+1)(2k+1)\frac{(j+1)(k+1)}{j+k+1}c_jc_k},\qquad j , k=0,1,2,\ldots,\\[2.0ex ] \displaystyle{f_{2j,2k}}&=&{jk\frac{(j+1)(k+1)}{j+k}c_jc_k},\qquad j , k=1,2,\ldots,\\[2.0ex ] f_{nm}&=&0 , \qquad { \rm otherwise}\ , , \end{array}\right.\ ] ] where @xmath339 is the @xmath340-th catalan number .
we first derive the formula @xmath117 for @xmath128 : we start with the definition of @xmath242 , @xmath341 which implies @xmath342 therefore we may compute @xmath343 by the contour integral @xmath344 using the dtoda curve , @xmath345 we write the integral with respect to @xmath242 , and then integrate by parts , @xmath346 in the last step , we have used the expansion , @xmath347 .
thus we obtain @xmath348 we now derive the formula @xmath1 for @xmath349 : we use the faber polynomials in proposition [ faber ] , i.e. @xmath350_{\ge 0}=\lambda^n - d(\lambda)f_n\,.\ ] ] this implies @xmath351 now changing variables from @xmath210 to @xmath242 , we have @xmath352_{\le 0}\,.\end{aligned}\ ] ] here we have noticed that @xmath254 can be replaced by @xmath257 with @xmath353_{<0}])$ ] for @xmath354 .
the notation @xmath355_{<0}$ ] implies the part of the laurent series @xmath356 with the inverse powers of @xmath242 , i.e. @xmath355_{<0}=f(p)-[f(p)]_{\ge0}$ ] .
thus we have @xmath357}(m-2k)\binom{m}{k}\oint_{p=\infty } \frac{dp}{2\pi i}\,\left(p+\frac{1}{p}\right)^np^{2k - m-1}\,.\ ] ] then from the expansion @xmath358 , we have the residue integral @xmath359 which implies @xmath360 in the non - zero case .
therefore , @xmath361 only if @xmath362 is even ( recall proposition [ parity - proposition ] ) .
so we have two cases : ( a ) @xmath363 , and ( b ) @xmath364 . with the residue calculation for the case ( a )
, we obtain @xmath365 we consider the case ( a ) ( the case ( b ) can be shown in the similar way ) : let us assume @xmath366 ( because of the symmetry in @xmath367 , this case is enough to show ) . also confirm that the case of @xmath368 gives ( [ f1 m ] ) . then from ( [ fmnint ] )
, we have @xmath369 we note the following identity which is a key equation in the proof , @xmath370\,.\ ] ] also we write @xmath371 then we have the following formula with the common denominator @xmath372 , @xmath373 \frac{k(k-1)\cdots(k - j+l+1)}{(k+2)(k+3)\cdots(k+j - l+1)}\,c_k \\ & = \frac{c_k}{(k+2)(k+3)\cdots(k+j+1)}g(j , k)\,,\end{aligned}\ ] ] where @xmath374 is the polynomial of degree @xmath340 in @xmath18 @xmath375 \prod_{\alpha=0}^{j - l-1 } ( k-\alpha ) \times \prod_{\alpha=0}^{l-1 } ( k+j+1-\alpha ) \,.\ ] ] here the products @xmath376 are understood to give 1 when @xmath377 and @xmath378 .
we then claim that the polynomial @xmath374 can be expressed by the simple form , @xmath379 here note that the coefficient @xmath380 is just @xmath381 $ ] .
this then gives the desired formula , that is , this equation for @xmath374 leads to the formula @xmath382 . to prove the claim , we write ( [ g1-virgil ] ) in the sum of the terms with the coefficients @xmath383 ( like a telescoping sum ) : @xmath384 where the last term in ( [ g1-b ] ) is the remainder from the telescoping
we then observe that this reduces the degree of the polynomials of both sides of ( [ g ] ) , that is ( [ g ] ) becomes the equation @xmath385 where the right hand side of this equation is the difference between the remainder in ( [ g1-b ] ) and the right hand side of ( [ g ] ) .
we then note that the coefficients in the sum can be written in the form similar to ( [ key ] ) , @xmath386\,.\ ] ] here we have used the relations ( the second one is the pascal rule ) , @xmath387 then we can write ( [ g1 ] ) in a similar form to ( [ g ] ) , @xmath388 \prod_{\alpha=0}^{j - l-2 } ( k-\alpha ) \times \prod_{\alpha=0}^{l-1 } ( k+j+1-\alpha ) = \binom{2j-1}{j-1}(k+2)(k+3)\cdots(k+j)\,.\ ] ] again we write the sum in the terms with the coefficients @xmath389 and move the remainder to the right hand side to arrive at the equation @xmath390 both sides are now polynomials of degree @xmath391 in @xmath18 . in the similar way with the identity , @xmath392\,,\ ] ] the above equation takes the similar form as before ,
i.e. @xmath393 \prod_{\alpha=0}^{j - l-3 } ( k-\alpha ) \times \prod_{\alpha=0}^{l-1 } ( k+j+1-\alpha ) = \binom{2j-2}{j-2}(k+3)(k+4)\cdots ( k+j).\ ] ] then we can further reduce the degree of the polynomials in the same way . continuing this process
, we complete the proof .
we recall that @xmath1 has a combinatorial meaning of enumeration of connected ( ribbon ) graphs on a sphere .
the presentation of such combinatoric problems in terms of the function @xmath394 arising from the gaussian unitary ensemble example is not a new idea .
however viewing the genus zero solutions as arising purely from the dtoda hierarchy is a new approach to solving the problem , although a similar idea was used in the case of a single even time for the computations of @xcite .
surprisingly we have found a closed form for the numbers @xmath1 with the conditions @xmath188 and @xmath189 for general @xmath2 and @xmath3 .
the solution of the two - vertex problem given by theorem [ fformula ] is new , previously the problem has been solved only in the so called dipole " case when the vertices are of the same degree ( i.e. @xmath395 ) @xcite .
these authors consider the general genus problem of two vertices of the same degree on a genus @xmath88 surface .
similar formulae were also found for other low genus problems in @xcite , but again the vertices must all have the same degree .
let us start by giving a combinatoric argument for the @xmath343 formula of theorem [ fformula ] , that is , @xmath117 gives the number of genus 0 ribbon graphs with a vertex of degree @xmath119 and a marked face : as shown above counting graphs with marked faces in the case of fixed genus is equivalent to just counting the number of graphs , the number of faces being given by the vertex structure .
so the problem is then to count the number of genus 0 ribbon graphs with a single vertex of degree @xmath119 .
it is easy to see that this is equivalent to counting the number of ways to make pair - wise connections with @xmath119 points on a circle without crossing and the number of sections divided by the chords giving the connections of the vertices .
the first number is well - knwon and is given by the @xmath18-th catalan number @xmath239 .
then the sphere is divided up to @xmath396 sections .
so we have @xmath397 which is @xmath117 ( see theorem [ fformula ] ) .
now we consider the two vertices problem , and show that @xmath1 satisfy the following recursion relations as a direct consequence of the enumeration of connected ribbon graphs appearing in the problem : [ fmngraph ] the two point functions @xmath398 with @xmath349 for the two - vertex problem satisfy the recursion relations , @xmath399 and @xmath327 .
the gluing of the edge @xmath114 of @xmath400 to the edge @xmath401 of the same vertex divides into the two sets of edges ; the set of @xmath402 edges marked by @xmath403 is of the one - vertex problem , since those edges can not cross the glued ribbon between the edges @xmath114 and @xmath401 on a sphere .
the gluing of the @xmath404 edges of the vertex @xmath400 to @xmath327 edges of @xmath405 gives the two - vertex problem.[isolate ] , width=340 ] we consider the case of @xmath382 ( the case of @xmath406 can be shown in the same way ) .
we will use @xmath400 and @xmath405 to refer to the vertex of degree @xmath407 and @xmath327 respectively .
we start with the graph , in which a specific edge , say 1 , of @xmath400 is glued to an edge of the vertex @xmath405 .
the number of ways to accomplish this gluing is @xmath327 ( the number of edges at @xmath405 ) .
having glued the edge 1 of @xmath400 with an edge of @xmath405 , we have the situation of one - vertex problem of degree @xmath408 .
so the number of connected graphs for each situation is given by the catalan number @xmath409 .
thus the total number of the edge 1 gluing is @xmath410 , which is the last term in the equation .
now we calculate the other cases where the edge 1 is glued to another edge of @xmath400 .
let the edge 1 be glued with the edge @xmath401 ( @xmath411 ) .
this divides the edges of @xmath400 into two sets one of which is cut off the rest of the graph ( see figure [ isolate ] ) .
this group of edges is marked by @xmath412 , and these @xmath402 edges of @xmath400 must be glued together within this group .
this is the one - vertex problem of degree @xmath402 , and the number of gluings is @xmath413 .
now we are left with the two vertices problem with the vertex of degree @xmath404 for @xmath400 and the vertex of degree @xmath327 .
the number of connected graphs of this problem is , of course , @xmath414 .
this situation with one - vertex problem of degree @xmath402 and two - vertex problem of degrees @xmath404 and @xmath327 appears also the case where the edge 1 is glued with the edge @xmath415 . here
the group of @xmath402 edges in the one - vertex problem is @xmath416 .
so we have the total number of connected graphs is @xmath417 then changing the variables in the sum with @xmath418 , we have the equation .
we also note that these relations in proposition [ fmngraph ] define a recursion relation of the faber polynomials on the dtoda curve : we consider only the first equation ( the second equation can be shown in the same manner ) .
multiplying @xmath419 to the first equation in proposition [ fmngraph ] and summing up for @xmath420 , we obtain , after rearranging the terms , @xmath421 ( recall that @xmath422 and @xmath423 . )
this can be considered as a recursion relation of the faber polynomials defined on the dtoda curve @xmath424 , that is , with , one can define @xmath425 from @xmath426 .
conversely , starting from this recursion , one can derive the first equation of proposition [ fmngraph ] .
this implies that we have an alternative proof of this proposition using only the faber polynomials , that is , the explicit form of two point functions @xmath64 obtained in theorem [ fformula ] satisfy the recursion relations of proposition [ fmngraph ] .
let us now show that the relation can be directly derived from the dtoda curve .
we first note that the left hand side is a polynomial in @xmath242 .
it is indeed easy to see that the projection of the right hand side on the polynomial part gives the polynomial in the right hand side .
this means that the non - polynomial part in the right hand side must vanish , when @xmath427 is on the dtoda curve , i.e. @xmath424 . so we want to prove @xmath428_-=\left[\sum_{l=1}^jc_{j - l}\lambda(p)^{2l-1}\right]_-\,,\ ] ] where @xmath355_-$ ] is the non - polynomial part of the laurent series @xmath356 . with the dtoda curve
, this leads to @xmath429 then from the coefficients of the power @xmath430 and using the pascal relation , @xmath431 , we have the following lemma which proves equation : for @xmath432 , we have @xmath433 let us first note that we may write the second binomial coefficient as @xmath434 where we denoted @xmath435 and @xmath436 .
we also note @xmath437 now let @xmath438 be a generating function of the catalan numbers , @xmath439 which satisfies @xmath440 with @xmath441 at @xmath442
. then one can show that @xmath443 this result can be found by computing the contour integral , using @xmath444 and @xmath445 , @xmath446 then from , we see that @xmath447 \\ & = \oint_{t=0}\frac{dt}{2\pi i } \,\frac{1}{2t^{n+1 } } ( z-1)^{i-1 } \left ( z-1 + t z ' \right ) \\ & = \oint_{t=0}\frac{dt}{2\pi i } \,\frac{1}{2t^{n+1 } } ( z-1)^i \left(1 + \frac{z}{2-z } \right)\\ & = \oint_{t=0}\frac{dt}{2\pi i } \,\frac{1}{t^{n+1 } } \frac{(z-1)^i } { 2-z } \ , .
\end{aligned}\ ] ] this implies that @xmath448 also with @xmath240 , we have @xmath449 , so that we see by setting @xmath436 , @xmath450 then the sum in the lemma is given by the coefficient of @xmath451 , i.e. @xmath452 this lemma proves the relation which leads to the recursion relation of the faber polynomials defined on the dtoda curve .
the point here is that the combnatorial meaning of the two point functions @xmath64 given in proposition [ fmngraph ] can be directly obtained from the structure of the faber polynomials defined on the dtoda curve ( i.e. independent from the random matrix description of the two point functions ) .
we summarize this as the following theorem : [ gfcomb ] the grunsky coefficients @xmath263 of the faber polynomials defined on the dtoda curve @xmath424 satisfy the recursion relations given in proposition [ fmngraph ] , that is , they have the combinatorial meaning of counting the connected ribbon graphs for the two - vertex problem on a sphere .
in this section , we consider some basic properties of the dpfaff hierarchy .
in particular , we show that the dpfaff hierarchy can be reduced to the dtoda hierarchy by choosing a particular section of the solution space which depends only on the even times of the hierarchy ( i.e. the symplectic pfaff lattice hierarchy studied in @xcite ) .
we also discuss a rescaling of the variables based on the structure of the @xmath0-functions of the toda and pfaff lattice hierarchies .
that structure is that one is given by determinants and other by pfaffians .
let us first note that the dpfaff hierarchy reduces to the dtoda hierarchy as a special choice of the variables .
[ sr - todaconnection ] let @xmath453 and @xmath454 .
suppose @xmath455 for all @xmath456 , and let @xmath457 for @xmath56 and @xmath458 . then the dpfaff hierarchy ( [ ddkp - fay ] )
is reduced to the dtoda hierarchy , @xmath459 \displaystyle{e^{-\tilde d(\alpha)\tilde d(\beta)\tilde f}=1-\frac{e^{\tilde f_{00}}}{\alpha\beta}e^{(\tilde
d(\alpha)+\tilde d(\beta))\tilde f_0}}\ , , \end{array}\right.\ ] ] where @xmath460 and @xmath461 . the proof is based on the following observation : with @xmath453 and @xmath462 @xmath463 , @xmath464 it is then immediate to see that with @xmath465 gives .
this can be considered as the dispersionless limit of the pfaff lattice restricted to symplectic matrices ( a continuous version of the sr algorithm @xcite ) .
the dpfaff equation is then given by @xmath466 the second equation in the dpfaff ( ddkp ) equation ( [ ddkpequation ] ) becomes trivial .
we consider the following scalings : @xmath467 the scaling in the time variables may be considered as the size of the matrix models , i.e. the matrices we consider for the orthogonal ensembles are twice the size of matrices for the unitary ensembles .
the scaling in the free energy function is due to the structure of the @xmath0-functions , i.e. the determinant for the toda lattice and the pfaffian for the pfaff lattices ( a pfaffian is a square root of a determinant ) .
we will show in section [ dtoda - dpfaff ] that those scalings appear in the particular choices of our @xmath0-functions corresponding to the gaussian ensembles discussed in the introduction . with the scalings ( [ scalings ] ) ,
@xmath64 is scaled by @xmath468 for all @xmath469 , and we redefine the dpfaff hierarchy in the follwoing form similar to the dtoda hierarchy , @xmath470 & \displaystyle{q(\lambda)^2=p(\lambda)^2+g_2+\frac{e^{2f_{00}}}{p(\lambda)^2}}\ , , \end{array}\right.\ ] ] with @xmath471 , and @xmath472 we call the algebraic equation of @xmath473 in the dpfaff curve .
the first equation is the scaled version of the second one in , and the first one in is derived from this definition : [ 2ndddkp ] from , we have @xmath474 which is the first equation in with the scaling @xmath475 . from the dpfaff curve , we have @xmath476 & = \left(p(\lambda)^2-p(\mu)^2\right)e^{-2d(\lambda)d(\mu)f}\left(\frac{q(\lambda)-q(\mu)}{\lambda-\mu } \right),\end{aligned}\ ] ] which leads to the equation in the lemma .
this shows the equivalence between and .
let us now discuss the functional relations among @xmath64 , i.e. the dispersionless hirota equations for the dpfaff hierarchy : we first note each @xmath64 for @xmath203 can be determined by the set @xmath477 . from the ddkp curve , @xmath478 , we can see that each @xmath479 for @xmath480 can be determined by @xmath481 .
then from the first equation , we have @xmath482 this implies the assertion . in the similar way as in the dtoda hierarchy , the dispersionless hirota equations can be derived as follows : from the dpfaff curve , we have for @xmath56 @xmath483 \displaystyle{=-\frac{4}{n+1}f_{1,n+1}+\frac{4}{n}f_{01}f_{1n}+4\sum_{j+k = n}\frac{f_{1j}}{j}\frac{f_{1k}}{k}\ , , } \end{array}\ ] ] where @xmath212 with @xmath484 .
this gives the first half of the dispersionless hirota equations for the dpfaff hierarchy , which shows that @xmath187 are determined by @xmath485 .
the first nontrivial equation with @xmath222 is given by @xmath486 this is the second equation of the dpfaff lattice equation ( [ ddkpequation ] ) with the scaling @xmath487 .
the second half of the dispersionless hirota equations is given by taking @xmath209 in the first equation , i.e. @xmath488 expanding for large @xmath210 , this gives @xmath489
\displaystyle{=\left(\sum_{n=0}^{\infty}\frac{1}{\lambda^{n}}h_n(-2\hat{\mathbf{z}}))\right)\left(1 + 2\sum_{n=1}^{\infty}\frac{1}{\lambda^{n+1}}f_{1n}\right)}\ , . \end{array}\ ] ] then we have for @xmath490 @xmath491 where @xmath214 with @xmath492 . the first equation with @xmath493
gives @xmath494 this is the first equation of the dispersionless limit of the pfaff lattice equation ( [ ddkpequation ] ) with the scaling @xmath495 .
the dpfaff curve in has the structure on @xmath496 with @xmath497 and @xmath498 , @xmath499 which is the same as the dtoda curve representing a sphere .
the curve has also only two parameters of the deformation .
however , in terms of the coordinates @xmath500 , the curve has an infinite number of parameters @xmath207 , unlike the dtoda case . in @xmath473-coordinates
, the dpfaff curve has an asymptotic property @xmath501 where @xmath502 are determined by @xmath503 and @xmath185
. then one can consider the faber polynomials associated with the function @xmath504 .
however it seems that there is no role for those polynomials in the dpfaff theory .
in this section we first show that the combinatoric descriptions for the functions @xmath505 and @xmath506 in the leading order terms of the free energies ( see theorems [ biz - thm ] and [ ghj ] ) are intimately related .
the fundamental fact underlying this is that the euler characteristic two mbius graphs are equivalent to genus 0 ribbon graphs together with some data specifying local orientations at the vertices .
the relationship between the leading orders of the asymptotic expansions of the free energies @xmath507 in the gue ( @xmath145 ) , goe ( @xmath142 ) and gse ( @xmath143 ) random matrix settings , implies that for the special initial condition corresponding to these examples the solutions of the dtoda and dpfaff hierarchies agree ( after a rescaling with proper choice of initial conditions ) .
we will then show that this is merely an example of a more general fact : the solution of the dtoda hierarchy is in fact always a solution of the rescaled dpfaff hierarchy , or in other words the dtoda hierarchy implies the dpfaff hierarchy .
however the two hierarchies are not equivalent , the family of solutions to the dpfaff hierarchy being much larger ( see @xcite ) in broad terms universality in random matrix theory states that the limiting behavior of eigenvalues of the unitary ensembles of random matrices is independent of the external potential @xmath85 in certain scaling regimes .
there is also a universality for the limiting behavior of the eigenvalues of the random matrices across the orthogonal , unitary , and symplectic ensembles in some regions and choices of scalings , see @xcite and @xcite . in the gaussian case , when @xmath85 is quadratic
, we will illustrate universality explicitly by giving a simple relationship between the leading order expressions of the free energy @xmath508 for the gue and goe random matrix models ( recall that @xmath509 , that is , the goe and gse cases give the same leading order ) .
a related result is that there is a universality between integrals over the unitary and orthogonal groups ( not the same expressions as our @xmath510 and @xmath511 ) after appropriate scalings @xcite .
let us first note the following lemma relating the leading order combinatoric interpretation given by theorems [ biz - thm ] and [ ghj ] .
[ mobius - is - ribbon ] mbius graphs of euler characteristic 2 are in 1 - 1 correspondence with pairs made up of a ribbon graph of genus 0 together with one of the @xmath512 possible configurations for the local orientations of the vertices , where @xmath513 is the number of vertices of the graph it is easy to see that a ribbon graph of genus 0 together with a choice of local orientations of the @xmath513 vertices gives a mbius graph of euler characteristic 2 . for a mbius graph to have euler characteristic 2
there must be a choice of the local orientations of the vertices so that the ribbons are untwisted ( as the graph lays flat on a sphere ) , thus giving the pair of a ribbon graph of genus 0 and a choice of local orientations . from this lemma , we see that ( for leading order only ) there is the following relation between the coefficients @xmath115 and @xmath514 of the expansions of the free energies ( [ ufree ] ) and ( [ ofree ] ) , @xmath515 where @xmath516 is the number of vertices .
we now have : [ uosrelations ] the free energies @xmath508 have the following relations , @xmath517 the duality between @xmath518 and @xmath519 has already been shown in section [ osensemble ] .
the relation between @xmath518 and @xmath520 is a direct consequence of the relation between @xmath115 and @xmath521 above .
namely from ( [ ofree ] ) , we have @xmath522 & = & \displaystyle{\sum_{\mathbf{j}}\kappa_0(\mathbf{j})2^{v-1}(2t_0)^f\frac{\mathbf{t}^{\mathbf{j}}}{\mathbf{j}!}+c_0^{(1)}(t_0)}\\[2.0ex ] & = & \displaystyle{\frac{1}{2}\,\sum_{\mathbf{j}}\kappa_0(\mathbf{j})(2t_0)^f\frac{(2\mathbf{t})^{\mathbf{j}}}{\mathbf{j } ! } + \frac{1}{2}c_0^{(2)}(2t_0)}\\[2.0ex ] & = & \displaystyle{\frac{1}{2}f^{(2)}(2t_0,2\mathbf{t})}\ , .
\end{array}\ ] ] this proposition implies that the free energy of the dtoda hierarchy gives an universality over those gaussian ensembles in the leading order with the scalings given in ( [ scalings ] ) .
also notice that the scalings in ( [ scalings ] ) explicitly appear in this result .
it is interesting to note that the scalings in proposition [ uosrelations ] , which give universality between the dtoda and dpfaff hierarchies , agree with the scalings found in @xcite giving the universality at leading order between the matrix integrals over the orthogonal and unitary groups ( see theorems 7.1 and 9.1 of @xcite ) . in our case
the partition functions are derived from integrals over the spaces of symmetric and hermitian matrices .
we now show that a solution of the dtoda hierarchy gives a solution of the dpfaff hierarchy .
this extends the previous results for special choices of the free energies associated with the gaussian ensembles , and it implies that the universality holds for the general choice of the potential @xmath85 or the measure in the random matrix theory .
[ dkp - toda ] the free energy @xmath52 of the dtoda hierarchy also satisfies the dpfaff hierarchy .
we first calculate the square of the first equation of the dtoda hierarchy , @xmath523\\ & = 2 e^{d(\lambda)d(\mu)f}-e^{2d(\lambda)d(\mu)f}\left(1-\frac{e^{2f_{00}}}{p(\lambda)^2p(\mu)^2}\right)\,.\end{aligned}\ ] ] then from lemma [ cauchykernel ] of the kernel formula of @xmath73 , we have @xmath524 now recall that @xmath73 for the dtoda hierarchy is expressed by @xmath294 . then noting that @xmath525
, we obtain the first equation of the dpfaff hierarchy .
now we show that the dpfaff curve @xmath526 is reduced to the dtoda curve .
if @xmath64 are the solutions of the dtoda hierarchy , then the @xmath527 in the curve is given by @xmath528 here we have used @xmath529 and @xmath238 ( see section [ dtodac ] ) .
this implies @xmath530 with @xmath531 , this equation leads to the dtoda curve , which is equivalent to the choice , @xmath532 .
finally , we remark that this universality can be seen in the dpfaff equation with the rescaling and the proper normalization : the dpfaff equation is given by @xmath533 which can be split to @xmath534 then if @xmath52 satisfies the dtoda hierarchy , then both sides of this equation are zero .
notice that the left hand side is the dkp equation and the other side is the dtoda equation .
this splitting can be also seen in the pfaff lattice equation ( [ dkpequation ] ) , i.e. the first equation in can be written in the form , @xmath535 then one has a kp solution for @xmath132 which also satisfies the toda lattice equation given in the right hand side ( see e.g. @xcite ) .
the purpose of this appendix is to show how one can get the kernel formula ( [ cauchykernel ] ) for the kp hierarchy , which is one of the most important formulae in this paper .
we begin by giving a quick introduction to the dkp hierarchy : let @xmath210 be a laurent series in @xmath242 given by @xmath536 where @xmath537 are the functions of an infinite variables @xmath538 .
then the dkp hierarchy is defined by the set of equations , @xmath539 where @xmath540 is the faber polynomial given by @xmath541_{\ge0}.\ ] ] here @xmath355_{\ge0}$ ] is a polynomial part of the laurent series @xmath356 in @xmath242 . in particular , @xmath542 , and we denote sometime @xmath543 .
the bracket @xmath544 is the poisson bracket defined by @xmath545 the faber polynomial can be written in the laurent series ( see definition [ faberfg ] ) , @xmath546 in particular , we have @xmath547 then , following the computation in @xcite ( see also @xcite ) , we first multiply @xmath548 to @xmath73 , i.e. @xmath549 taking the projection of this equation on the polynomial part , @xmath550 now multiply by @xmath551 and sum over @xmath552 , we have @xmath553 this then gives , @xmath554 thus we have @xmath555 integrating this with respect to @xmath242 , and fixing the boundary condition at @xmath556 , we obtain @xmath557\\ & = \frac{1}{1-\frac{\lambda}{\mu}}\exp\left(-\sum_{n=1}^{\infty}\sum_{m=1}^{\infty}\frac{1}{\lambda^m\mu^n}\frac{q_{nm}}{n}\right)\ ,
. \end{aligned}\ ] ] then expressing the grunsky coefficients in terms of the free energy , i.e. @xmath558 , we obtain the dkp hierarchy ( [ cauchykernel ] ) , @xmath559 one should note here that the dispersionless hirota equation for @xmath1 is completely determined by the function @xmath560 ( spectral curve ) .
then the dkp hierarchy defines an integrable deformation of the function @xmath561 , and as a result , the grunsky coefficients can be expressed by the second derivatives of the free energy with the deformation parameters .
we also remark that in @xcite ( see also @xcite ) , takasaki derives this formula as a classical limit of the differential fay identity , which can be obtained from the bilinear identity @xcite , @xmath562\right ) \tau\left(\mathbf{t}+[z^{-1}]\right)=0\,.\ ] ] where @xmath563 and @xmath564 $ ] is given by @xmath565=\left(\frac{1}{z},\,\frac{1}{2z^2},\,\frac{1}{3z^3},\,\ldots\right)\,.\ ] ] takasaki also derives the dpfaff hierarchy from the classical limit of the differential fay identity obtained from the bilinear identity for the dkp hierarchy @xcite .
this is remarkable , since there seems to be no proper classical limit for the pfaff lattice hierarchy in the matrix representation given in @xcite ( this is contrary to the toda lattice hierarchy ) .
here we give the computation of @xmath191 for @xmath192 and @xmath193 .
we use formula ( 3.3.10 ) given in @xcite , reproduced here : with @xmath566 one finds the following formula , which we call mehta s formula , @xmath567^{-n } \prod_{j=1}^n \gamma\left ( 1 + \frac{\beta j}{2}\right ) \,.\ ] ] for @xmath145 , we have @xmath568.\ ] ] from mehta s formula , we have @xmath569 a straightforward change of variables gives @xmath570 inserting this into our formula for @xmath571 we find : @xmath572 = -\frac{n^2}{2 } \log n+ \frac{n}{2 } \log(2\pi ) + \sum_{j=1}^n \log(j!)\,.\ ] ] so the problem is to find the leading order behavior of @xmath573 .
we compute this in a generalization of the riemann sums approach to deriving sterlings approximation .
we start by rewriting the expression @xmath573 as a double summation which then simplifies : @xmath574 we then approximate this sum by an integral using the trapezoid rule @xmath575 where the error in the approximation can be computed using the euler - maclaurin formula . for our purposes
it is enough to show that @xmath576 in truth it performs much better .
the integral then gives @xmath577 inserting this back into our expression for @xmath578 we find @xmath579 then in the limit , we have with @xmath580 , @xmath581 in the case of @xmath142 , we have @xmath582\ ] ] mehta s formula then gives @xmath583 likewise in this case we find that @xmath584 = \left(n^2 + \frac{n}{2}\right ) \log\frac{2}{n } + \log\left [ z_{2n}^{(1)}\left(\frac{1}{2 } \lambda^2 ; \mathbf{0}\right)\right ] \\ & = n^2 \log \frac{2}{n } + \log(n ! ) - 2 \log 2\left ( \sum_{j=1}^n j\right ) + \sum_{j=1}^{n-1 } \log[(2j+1 ) ! ] + \mathcal{o}(n\log n)\,.\end{aligned}\ ] ] stirlings approximation implies that @xmath585 so we only need to find the behavior of @xmath586 for large @xmath2 : as above we follow the basic method of stirlings approximation and write the sum as the approximation of an integral plus an error term . for large @xmath2
the error term is of order @xmath587 ; @xmath588 \\ \nonumber & = \frac{1}{2 } \log ( ( 2n-2)(2n-1 ) ) + \int_1^{n-1 } ( n - x ) \log ( 2x ( 2x+1 ) ) dx + e \\ \label{gen - stirling } & = -\frac{3}{2 } n^2 + n^2 \log(2n ) + \mathcal{o}(n\log n)\ , .
\end{aligned}\ ] ] then we find @xmath589 which gives @xmath590 for @xmath143 , we have @xmath591\,.\ ] ] mehta s formula with @xmath143 gives @xmath592 which together with a change of variables gives @xmath593 therefore @xmath594 so we must compute the asymptotic behavior of @xmath595 it is not suprising that it is identical to the asymptotic behavior of @xmath596 found in the @xmath142 case .
this is in fact yet more evidence of the duality that exists between @xmath142 and @xmath193 . inserting this back into our expression for @xmath183 we find @xmath597
from which we conclude that @xmath598 i. p. goulden , j. l. harer , and d. m. jackson , a geometric parametrization for the virtual euler characteristics of the moduli spaces of real and complex algebraic curves , _ trans .
_ * 353 * ( 2001 ) 4405 - 4427 . | it is well - known that the partition function of the unitary ensembles of random matrices is given by a @xmath0-function of the toda lattice hierarchy and those of the orthogonal and symplectic ensembles are @xmath0-functions of the pfaff lattice hierarchy . in these cases
the asymptotic expansions of the free energies given by the logarithm of the partition functions lead to the dispersionless ( i.e. continuous ) limits for the toda and pfaff lattice hierarchies .
there is a universality between all three ensembles of random matrices , one consequence of which is that the leading orders of the free energy for large matrices agree . in this paper , this universality , in the case of gaussian ensembles , is explicitly demonstrated by computing the leading orders of the free energies in the expansions .
we also show that the free energy as the solution of the dispersionless toda lattice hierarchy gives a solution of the dispersionless pfaff lattice hierarchy , which implies that this universality holds in general for the leading orders of the unitary , orthogonal , and symplectic ensembles .
we also find an explicit formula for the two point function @xmath1 which represents the number of connected ribbon graphs with two vertices of degrees @xmath2 and @xmath3 on a sphere .
the derivation is based on the faber polynomials defined on the spectral curve of the dispersionless toda lattice hierarchy , and @xmath4 are the grunsky coefficients of the faber polynomials . |
consider a trader who wants to buy or sell a european option @xmath0 on asset @xmath1 with maturity @xmath2 and payoff @xmath3 .
the trader wants to hedge this position , but the underlying asset @xmath1 is illiquid .
however , some liquid proxies of @xmath1 are available in the marketplace .
first , there is a financial index ( or simply an _ index _ ) @xmath4 ( such as e.g. s&p500 or cdx.na ) whose market price is correlated with @xmath1 . in addition , there is another correlated asset @xmath5 which has a liquidly traded option @xmath6 with a payoff @xmath7 similar to that of @xmath8 , and with the same maturity @xmath9 .
the market price @xmath10 of @xmath6 is also known .
our trader realizes that hedging @xmath1-derivative with the index @xmath4 alone may not be sufficient for a number of reasons .
first , she might be faced with a situation where correlation coefficients @xmath11 ( which for simplicity are assumed to be constant ) are such that @xmath12 . in this case
we would intuitively expect a better hedge produced by using @xmath5 or @xmath6 as the hedging instruments .
second , if we bear in mind a stochastic volatility - type dynamics for @xmath1 , the stochastic volatility process may be `` unspanned '' , i.e. the volatility risk of the option may not be traded away by hedging in option s underlying .
if that is the case , one might want to hedge the unspanned stochastic volatility by trading in a `` similar '' option with on the proxy asset @xmath5 .
so our trader is contemplating a hedging strategy that would use both @xmath13 and @xmath14 . to capture an `` unspanned '' stochastic volatility
, the trader wants to use a derivative @xmath15 written on @xmath14 rather than asset @xmath14 directly . as transaction costs are usually substantially higher for options than for underlyings
, our trader sets up a static hedge in @xmath16 and a dynamic hedge in @xmath17 .
the static hedging strategy amounts to selling @xmath18 units of @xmath16 options at time @xmath19
. an _ optimal _ hedging strategy would be composed of a pair @xmath20 where @xmath21 is the optimal static hedge , and @xmath22 ( where @xmath23 ) is an optimal dynamic hedging strategy in index @xmath17 .
the pair @xmath20 should be obtained using a proper model .
the same model should produce the highest / lowest price for which the trader should agree to buy or sell the @xmath1-option . in this paper
we develop a model that formalizes the above scenario by supplementing it with the specific dynamics for asset prices @xmath24 and @xmath25 , and providing criteria of optimality for pricing options @xmath8 . for the former ,
we use a standard correlated log - normal dynamics . for the latter ,
we employ the utility indifference framework with an exponential utility , pioneered by @xcite and others , see e.g. @xcite for a review . as will be shown below , this results in a tractable formulations with analytical ( in quadratures ) expressions for optimal hedges and option prices . as the above setting of pricing and hedging an illiquid option position using a pair of liquid proxies ( e.g. a stock and an option on a different underlying ) is quite general
, one could visualize its potential applications for various asset classes such as equities , commodities , fx etc .
for definiteness , in this paper we concentrate on a problem of practical interest for counterparty credit risk management .
namely , we consider the problem of pricing and hedging an exposure to a counterparty with an illiquid debt , and in the absence of liquidly traded cds referencing this counterparty .
for such situation , no market - implied spreads are available for the counterparty in question . instead
, one should rely on a model to come up with _ theoretical _ credit spreads for the counterparty . to this end
, we use a version of the classical merton equity - credit model @xcite which is set up in a multi - name setting , and under the physical ( i.e. `` real '' , not `` risk - neutral '' ) measure .
most importantly , unlike the classical merton s model , we do not intend to use firm s equity to hedge firm s debt . instead ,
illiquid debt is hedged with a proxy liquid debt , and a proxy credit index . in what follows , to differentiate our framework from that of the classical merton model
, we will refer to it as the hedged incomplete - market merton s dynamics , or himd for short .
our model unifies three strands in the literature on indifference pricing .
the fist strand deals with hedging an option with a proxy asset , as developed in @xcite , and others . in this setting , one typically hedges an option on an illiquid underlying with a liquid proxy asset
. the second strand develops generalizations of the classical merton credit - equity model to an incomplete market setting .
typically , this is achieved by de - correlating asset value and equity price at the level of a single firm , see e.g. @xcite .
as long as we do not use firms equity to hedge firm s debt but instead use a liquid proxy bond as a hedge , such modification of the merton model is not needed in our setting .
the third strand is presented by @xcite who develop a static - dynamic indifference hedging approach for barrier options .
ilhan and sircar considers hedging a barrier option under stochastic volatility using static hedges in vanilla options on the same underlying plus a dynamic hedge in the underlying .
this results in a two - dimensional hamilton - jacobi - bellmann ( hjb ) equation .
our construction is similar but our hedges are a proxy asset and a proxy option , while volatility is taken constant for simplicity .
borrowing from an approach of @xcite for a similar ( but not identical ) setting , we now show how the method of indifference utility pricing can be generalized to incorporate our scenario of a mixed dynamic - static hedge . to this end , let @xmath26 be the final payoff of the portfolio consisting of our option positions , i.e. @xmath27 as long as both european options @xmath28 pay at the same maturity @xmath9 , we can view this as the payoff of a combined ( `` static hedge portfolio '' ) option @xmath29 , which involves payoffs @xmath3 and @xmath30 of both derivatives @xmath8 and @xmath6 .
such option may be priced using the standard utility indifference principle .
the latter states that the derivative price @xmath31 is such that the investor should be indifferent to the choice between two investment strategies . with the first strategy
, the investor adds the derivatives to her portfolio of bonds and stocks ( or indices in the setting of the merton s optimal investment problem . ] ) @xmath4 , thus taking @xmath29 from , and adding @xmath32 to her initial cash @xmath33 . with the second strategy
, the investor stays with the optimal portfolio containing bonds and the stocks / indices .
the value of each investment is measured in terms of the _ value function _ defined as the conditional expectation of utility @xmath34 of the terminal wealth @xmath35 optimized over trading strategies . in this work
, we use an exponential utility function @xmath36 where @xmath37 is a risk - aversion parameter . in our case
, the terminal wealth is given by the following expression : @xmath38 with @xmath39 be the total wealth at time @xmath9 in bonds and index @xmath13 . in turn
, the value function reads @xmath40\ ] ] where @xmath41 is a set of admissible trading strategies that require holding of initial cash @xmath33 .
the expectation in the eq.([value_function ] ) is taken under the `` real - world '' measure @xmath42 . for a portfolio made exclusively of stocks / indices and bonds ,
the value function for the exponential utility is known from the classical merton s work : @xmath43 where @xmath44 , @xmath45 is the risk free interest rate assumed to be constant , and @xmath46 is the stock sharpe ratio . in our setting , in addition to bonds and stocks / indices , we want to long @xmath8 option and short @xmath47 units of @xmath6 option to statically hedge our @xmath8 position , or , equivalently , buy the @xmath48 option .
the value function in our problem of optimal investment in bonds , index and the composite option @xmath49 has the following form : @xmath50 where @xmath39 is a cash equivalent of the total wealth in bonds and the index at time @xmath9 .
we represent it in a form similar to eq.([merton_vv ] ) : @xmath51 where function @xmath52 will be calculated in the next sections .
the indifference pricing equation reads @xmath53 plugging this in eq.([merton_vv ] ) and eq.([vv_him ] ) and re - arranging terms , we obtain @xmath54 the highest price of the @xmath1-derivative is given by choosing the optimal static hedge given by the number @xmath47 of the @xmath5-derivatives , i.e. @xmath55 where we temporarily introduced subscript @xmath47 in @xmath56 to emphasize that the value function depends on @xmath47 through a terminal condition .
to use eq.([indif_price_2 ] ) and thus be able to compute both the option price and optimal static hedge , we need to find the `` reduced '' value function @xmath52 . to this end
, we first derive the hamilton - jacobi - bellman ( hjb ) equation for our model , and then obtain its analytical ( asymptotic ) solution .
let @xmath57 be the dynamic investment strategy in the index @xmath17 at time @xmath58 starting with the initial cash @xmath33 , and @xmath59 be the markov generator of price dynamics corresponding to strategy @xmath60 .
both the optimal dynamic strategy and the value function should be obtained as a solution of the hjb equation @xmath61 we assume that all state variables @xmath62 follow a geometric brownian motion process with constant drifts @xmath63 and volatilities @xmath64 @xmath65 if our total wealth at time @xmath66 is @xmath67 and we invest amount @xmath68 of this wealth into the index and the rest in a risk - free bond , the stochastic differential equation for @xmath69 is obtained as follows : @xmath70 then @xmath59 reads @xmath71 where @xmath72 is defined on the domain @xmath73 $ ] . since @xmath59 is a regular function of @xmath74 , @xmath75
is achieved at @xmath76 plugging this into eq.([hjb ] ) , we obtain @xmath77 this is a nonlinear pde with respect to the dependent variable @xmath78 with standard boundary conditions ( see @xcite ) and the terminal condition determined by a choice of the writer s maximal expected utility ( value function ) of the terminal wealth @xmath35
. note that so far the derivation is valid for a generic utility function . to make further progress we specialize to the case of exponential utility in eq.([exputility ] ) since it gives rise to a natural dimension reduction of the hjb equation .
indeed , the ansatz @xmath79 with @xmath80 is both consistent with terminal condition eq.([valuefunction ] ) and , upon substitution in ( [ pde ] ) , leads to a pde for function @xmath81 which does not contain variable @xmath82 : @xmath83 here @xmath84 equation eq.([gpdews ] ) is defined at the domain @xmath85 $ ] . the initial condition for this equation
is obtained from eq.([valuefunction ] ) . in
what follows , we choose a specific payoff of the form eq.([payoff ] ) with @xmath86 that corresponds to a portfolio of bonds of firms @xmath14 and @xmath87 with notionals @xmath88 within the merton credit - equity model .
then the terminal condition for @xmath89 reads @xmath90\ ] ] where @xmath91 and @xmath92 .
we were not able to find a closed form solution of eq.([gpdews ] ) with the initial condition eq.([initcondws ] ) .
on the other hand , a numerical solution of this equation is expensive , especially when it should be used many times for calibration to market data .
therefore , we proceed with aasymptotic solutions of eq.([gpdews ] ) .
we suggest two approaches to construct asymptotic solutions .
as we want to statically hedge option @xmath8 with options on another underlying , we look for an asset @xmath14 that is strongly correlated with asset @xmath87 .
further , if we have a `` similar '' option @xmath15 on asset @xmath14 ( i.e. similar maturity , type , strike , etc . ) , we expect that such option provides a good static hedge for our option @xmath8 . , but @xmath93
. we may want to hedge an illiquid option with strike @xmath94 ( say , deep otm ) with a liquid option on the same underlying but with a different strike @xmath95 . under this setup
, we have @xmath96 , i.e. a prefect correlation case . ] therefore , a natural assumption would be to consider @xmath97 to be a small parameter under our setup .
utilizing this idea we represent the solution of eq.([gpdews ] ) as a formal perturbative expansion in powers of @xmath98 : @xmath99 where @xmath100 is the small parameter to be precisely defined in the next section . as we shown below , eq.([gpdews ] ) can be solved analytically ( in quadratures ) to any order of this expansion , thus significantly reducing the computation time .
we start with a change of variables @xmath101 defined as follows : @xmath102 simultaneously , we change the dependent variable @xmath103 as follows : @xmath104 using eq.([gpdews ] ) , eq.([uv ] ) and eq.([changeoffun1 ] ) , we obtain the following pde for function @xmath105 : @xmath106 further we will show that @xmath107 is a slow ( `` adiabatic '' ) variable of our asymptotic method , while @xmath108 becomes a `` fast '' variable . in what follows we need an inverse of eq.([uv ] ) at @xmath109 : @xmath110 using this in eq.([initcondws ] ) , we obtain the initial condition in @xmath111 variables for the function @xmath112 : @xmath113\ ] ] where @xmath114 for any real @xmath82 .
recall that a correlation matrix @xmath115 of @xmath116 assets can be represented as a gram matrix with matrix elements @xmath117 where @xmath118 are unit vectors on a @xmath119 dimensional hyper - sphere @xmath120 . using the 3d geometry , it is easy to establish the following _ cosine law _ for correlations between three assets : @xmath121 with @xmath122 being an angle between @xmath123 and its projection on the plane spanned by @xmath124 . as discussed e.g. by @xcite ,
three variables @xmath125 are independent , but @xmath126 are not .
therefore , one of them , e.g. @xmath127 , has to be found using eq.([cosine ] ) given @xmath128 .
further we define @xmath100 as @xmath129 and also define the following constants @xmath130 using eq.([cosine ] ) and definitions in eq.([eps ] ) , eq.([theta_12 ] ) , coefficients at @xmath131 and @xmath132 in eq.([phi_uv ] ) are evaluated as follows : @xmath133 accordingly using this notation eq.([phi_uv ] ) takes the form @xmath134 in the limit @xmath135 this equation does not contain any derivatives wrt @xmath107 , therefore @xmath107 enters the equation only as a parameter ( since @xmath136 is a function of @xmath107 ) .
we call this limit the _ adiabatic limit _ in a sense that will be explained below .
it should be noted that our expansion in powers of @xmath98 can diverge if @xmath137 is very small .
we exclude such situations on the `` financial '' grounds assuming that all pair - wise correlations in the triplet @xmath138 are reasonably high ( of the order of 0.4 or higher in practice ) , for our hedging set - up to make sense in the first place .
thus , parameter @xmath139 is treated as @xmath140 is always @xmath140 and positive , parameter @xmath141 could be both positive and negative for typical values of correlations .
for example , if @xmath142 , then @xmath143 , while for @xmath144 it is @xmath145 .
the cosine law can also be used to find proper values of correlation parameters in the limit @xmath146 .
to this end , we first use eq.([cosine ] ) to convert the estimated triplet @xmath147 into a triplet of independent variables @xmath148 , and then take the limit @xmath146 while keeping @xmath149 and @xmath150 constant . ] .
it turns out that the last equation of the previous section could be further simplified . introducing new independent variables
@xmath151 we can transform eq.([phi_uv_2 ] ) into the following equation @xmath152 it is seen that in new variables the mixed derivative drops from from the equation , as so does @xmath100 .
however , further let us formally introduce a multiplier @xmath153 in the term @xmath154 which transforms eq.([phi_vv_2 ] ) into @xmath155 let us also formally assume that @xmath153 is small under certain conditions .
the idea of this trick is as follows .
one way to construct an asymptotic solution of the eq.([phi_vv_2 ] ) is to assume that all derivatives are of the order @xmath140 , and then estimate all the coefficients . if one manages to find a coefficient which is @xmath156 , then it is possible to build an asymptotic expansion using that coefficient ( or @xmath153 itself ) as a small parameter .
if , however , all the coefficients in the considered pde are of order @xmath140 we need to check if perhaps some of the derivatives in the eq.([phi_vv_2 ] ) are small , e.g. @xmath156 . if this is the case , in order to apply standard asymptotic methods we formally have to add a small parameter @xmath153 as a multiplier to the derivative which is @xmath156 , , where @xmath157 make an asymptotic expansion on @xmath153 , solve the obtained equations in every order on @xmath153 , and at the end in the final solution put @xmath158 .
that is exactly the way we want to proceed with .
this means that instead of the eq.([expan ] ) , we now have the following expansion : @xmath159 to find conditions when @xmath154 could be small as compared with the other terms in the eq.([phi_vv_2 ] ) , we use an inverse map at @xmath160 this is a regular map . ]
@xmath161 and rewrite the payoff function eq.([initcondws ] ) in the form @xmath162 , \\
\omega_1 & = \bar{v}\frac{\theta_1}{\sqrt{\theta_2 } } , \qquad \omega_2 = \bar{v } \frac{\theta_3}{\beta \sqrt{\theta _ 2 } } , \qquad \zeta = \frac{1}{\sqrt{\theta_2 } } \nonumber\end{aligned}\ ] ] suppose that @xmath163 or @xmath164 .
differentiating the payoff twice by @xmath165 and twice by @xmath107 and computing the ratio of the first and second terms in the rhs of the eq.([phi_vv_2 ] ) , one can see that in the limit @xmath135 this ratio becomes @xmath166 .
typical values @xmath167 give rise to @xmath168 , therefore the second term is small as compared with the first one , and @xmath153 is a good small parameter .
this , however changes if @xmath169 is small or / and @xmath170 is close to 1 , and then @xmath171 . still in this case
we have @xmath172 which can be used as a small parameter .
therefore , our approach is as follows : 1 . if @xmath173 we use eq.([phi_vv_2 ] ) and find its asymptotic solutions using eq.([expan1 ] )
this is better than using eq.([phi_uv_2 ] ) because first , @xmath153 is typically smaller then @xmath100 , and second , the term @xmath154 in our first method eq.([expan ] ) appears only in the second order of approximation while in the second method it is taken into account already in the first order on @xmath153 .
if , however , @xmath174 , then @xmath153 is not anymore a small parameter , therefore we use eq.([phi_uv_2 ] ) and solve it asymptotically using eq.([expan ] ) . in general , this argument can not be applied if @xmath175 or @xmath176 because then both derivatives of the payoff vanish .
however , the above argument is intended to provide an intuition as to why @xmath154 could be much smaller that the other terms in the rhs of eq.([phi_vv_2 ] ) .
this intuition can be verified numerically , and our test examples clearly demonstrate that smallness of @xmath153 often takes place .
below we discuss under which conditions this could occur . note that at the first glance , the described method looks similar to the quasi - classical approximation in quantum mechanics ( @xcite ) .
the similarity comes from the observation that transformation eq.([uchange ] ) is singular in @xmath100 which is similar to the quasi - classical limit @xmath177 . if we would construct an asymptotic expansion on @xmath100 we would expand the rhs of the eq.([phi_vv_2 ] ) on @xmath100 , but not the payoff function .
after getting the solution of eq.([phi_vv_2 ] ) in zero - order approximation on @xmath100 as a function of @xmath178 , we would apply the inverse transformation @xmath179 which is non - singular .
therefore , the final result would not contain any singularity .
since @xmath180 is defined via @xmath181 and @xmath182 , it could seem that @xmath183 , and we face a `` quasi - classical '' situation .
however , as explained above , the independent parameters are @xmath184 and @xmath185 . therefore , by definition @xmath186 does nt depend on @xmath187 , or on @xmath100 .
thus , our two methods actually correspond to different assumptions .
the first one utilizes a strong correlation between assets z and y. the second assumes a strong correlation between index @xmath4 and asset @xmath1 while at the same time the vector of correlation @xmath169 in 3d space is not collinear to the vector of correlation @xmath127 . by financial sense
this means ( see eq.([cosine ] ) ) the following . 1 .
either @xmath169 is about 1 and , therefore , @xmath188 . in other words ,
index @xmath4 strongly correlates to asset z , so @xmath4 is almost @xmath1 , therefore correlation of y and z ( @xmath187 ) is close to correlation of y and x ( @xmath127 ) . that , in turn , means that asset z can be dynamically hedged with @xmath4 , and extra static hedge with y does nt bring much value
in contrast , under the former assumption static hedge plays an essential role .
@xmath189 which means that @xmath190 . for instance , @xmath191 .
this is an interesting case , since it differs from two previously considered assumptions on high value of either @xmath187 or @xmath169 .
indeed , all correlations could be relatively moderate while providing a smallness of @xmath141 . in what follows , we describe in detail the asymptotic solutions for zero and first order approximations in @xmath153 , and outline a generalization of our approach to an arbitrary order in @xmath153 .
asymptotic solutions in @xmath100 are constructed in a very similar way and are given in appendix [ a ] . also to make our notation lighter in the next section we
will us @xmath107 instead of @xmath192 since that should not bring any confusion .
in the zero order approximation we set @xmath193 , so that eq.([phi_vv_2 ] ) does not contain derivatives wrt @xmath107 : @xmath194 where @xmath195 .
therefore , dependence of the solution on @xmath107 is determined by the terminal condition . in other words ,
our system changes along variable @xmath165 , but it remains static ( i.e. of the order of @xmath196 slow ) in variable @xmath107 .
using analogy with physics , we call this limit the _ adiabatic limit_. the last equation can be solved by a change of dependent variable ( closely related to the hopf - cole transform , see e.g. @xcite ) : @xmath197^{1/(1- \bar{\rho}_{xz}^2)}\ ] ] which reduces eq.([zeroeq ] ) to the heat equation @xmath198 subject to the initial condition @xmath199 .
the latter can be obtained from the eq.([initconduv1 ] ) if one replaces @xmath37 with the `` correlation - adjusted '' risk aversion parameter @xmath200 .
it can also be written as a piece - wise analytical function having a different form in different intervals of @xmath165-variable . if @xmath201 , we have @xmath202 , & u < - \omega_1 \\ \exp\left [ - \bar{\gamma } \left(k_z - \alpha k_y e^ { \zeta \beta \sigma_y\left(\omega_2 + u\right)}\right ) \right ] , & - \omega_1 \le u < -\omega_2 \\ \exp\left [ - \bar{\gamma } \left(k_z - \alpha k_y\right ) \right ] , & u \ge -\omega_2 , \end{cases}\end{aligned}\ ] ] while for @xmath203 we have @xmath204 , & u < -\omega_2
\\ \exp \left [ - \bar{\gamma } \left(k_z e^{\zeta \sigma_z\left(\omega_1 + u \right ) } - \alpha k_y \right ) \right ] , & -\omega_2 \le u < - \omega_1 \\ \exp\left [ - \bar{\gamma } \left(k_z - \alpha k_y \right ) \right ] , & u \ge - \omega_1 \\ \end{cases}\end{aligned}\ ] ] using the well - known expression for the green s function of our heat equation @xmath205 ( see e.g. @xcite ) , the solution of eq.([linpde2 ] ) is then @xmath206 the explicit zero - order solution thus reads @xmath207^{1/(1-\bar{\rho}_{xz}^2)}\ ] ] note that eq.([phi_0_sol ] ) provides the general zero - order solution for the hjb equation with arbitrary initial conditions at @xmath208 . for our specific initial conditions eq.([init_cond_quand2 ] ) , the solution is readily obtained in closed form in terms of the error ( or normal cdf ) function ( see appendix [ a1 ] ) .
however , for numerical efficiency it might be better to use another method which is based on a simple observations that the expression in square brackets in the eq.([phi_0_sol ] ) is just a gauss transform of the payoff .
this transform can be efficiently computed using a fast gauss transform algorithm which in our case is @xmath209 with @xmath210 being the number of grid points in @xmath165 space . for the first correction in eq.([phi_vv_2 ] ) we obtain the following pde @xmath211 with @xmath212 .
this is an inhomogeneous _ linear _ pde with variable coefficients .
as long as our zero - order solution of eq.([zeroeq ] ) already satisfies the initial condition , this equation has to be solved subject to the zero initial condition .
this considerably simplifies the further construction .
we look for a solution to eq .
( [ first2 ] ) in the form @xmath213^{\bar{\rho}_{xz}^2 } h(u , v , \tau)\ ] ] this gives rise to an inhomogeneous heat equation for function @xmath214 subject to zero initial condition @xmath215 thus , using the duhamel s principle ( @xcite ) we obtain @xmath216 there exists a closed form approximation of the internal integral ( see appendix [ b2 ] ) .
the second order equation has the same form as the eq.([first2 ] ) @xmath217 where @xmath218\ ] ] as this equation has to be solved also subject to zero initial conditions , the solution is obtained in the same way as above : @xmath219 this shows that in higher order approximations in @xmath153 both the type of the equation and boundary conditions stay the same .
therefore , the solution to the @xmath220-th order approximation reads @xmath221 where @xmath222 can be expressed via already found solutions of order @xmath223 and their derivatives on @xmath165 and @xmath107 .
the exact representation for @xmath222 follows combinatorial rules and reads @xmath224 where @xmath225 is a coefficient at @xmath226 in the expansion of @xmath227 , \\
\beta_1 & = \sum_{i=1}^\infty \mu^i \phi_{i , u}/\phi_{0,u } , \qquad \beta_2 = \sum_{i=1}^\infty \mu^i \phi_{i}/\phi_{0}\end{aligned}\ ] ] in series on @xmath153 .
this could be easily determined using any symbolic software , e.g. mathematica .
for instance @xmath228 reads @xmath229 the explicit representation of the solutions of an arbitrary order in quadratures is important because , per our definition of @xmath153 , convergence of eq.([expan ] ) is expected to be relatively slow .
indeed , if one wants the final precision to be about @xmath230 at @xmath231 ( @xmath232 ) , the number of important terms @xmath233 in expansion eq.([expan1 ] ) could be rawly calculated as @xmath234 , which gives @xmath235 , while at precision @xmath236 this yields @xmath237 .
note that all integrals with @xmath238 do not admit a closed form representation and have to be computed numerically .
again , this could be done in an efficient manner using the fast gauss transform .
to verify quality of our asymptotic method we compare two sets of results .
one is obtained using zero and first order approximations ( being computed via a series representation and @xmath239 functions given in appendix [ a1 ] in eq.([jdef ] ) , eq.([omega ] ) and eq.([omega1 ] ) , or using the fast gaussian transform ) .
our tests showed that the number of terms in the double sum that should be kept is small , namely truncating the upper limit in @xmath240 from infinity to @xmath241 produces nearly identical results .
therefore , the total complexity of calculation is about 45 computations of @xmath242 and @xmath243 functions which is very fast . a typical time required for this at a standard pc with
the cpu frequency 2.3 ghz ranges from 0.68 sec ( test 1 ) to 0.35 sec ( test 2 ) ( see below ) .
the other test is performed using a numerical solution of eq.([phi_vv_2 ] ) . in doing
so we use an implicit finite difference scheme built in a spirit of @xcite .
after the original non - linear equation is discretized to obtain the value function at the next time level , we need to solve a 2d algebraic system of equations each of which contains a non - linear term .
this could be done e.g. by applying a fixed point iterative method ( @xcite ) .
in other words , at the first iteration as an initial guess we plug - in into the non - linear term the solution obtained at the previous level of time .
this reduces the equation to a linear one since the non - linear term is explicitly approximated at this iteration .
next we solve the resulting 2d system of equations with a block - band matrix using a 2d lu factorization . at the second iteration
, the solution obtained in such a way is substituted into the non - linear term again , so again it is approximated explicitly .
then the new system of linear equations is solved and the new approximation of the solution of the original non - linear equation is obtained .
we continue this process until it converges . the number of iterations to needed for numerical convergence depends on gradients of the value function , which are considerably influenced by the value of @xmath37 . for small values of @xmath37 ( about 0.03 )
we need about 1 - 2 iterations , while for @xmath244 , 5 - 7 iterations might be necessary . for higher values of @xmath37 , the fixed point iteration scheme could even diverge
, so another method has to be used instead .
it is also important to note that we solve the non - linear equation using the dependent variable @xmath245 , rather than @xmath246 to reduce relative gradients of the solution .
note that for _ linear _
2d parabolic equations with mixed derivatives more efficient splitting schemes exist , see e.g. @xcite . in principle
, such schemes could be adopted to solve eq.([phi_vv_2 ] ) .
though eq.([phi_vv_2 ] ) is a _ non - linear _ equation , the presence of the non - linear term does not change its type ( it is still a parabolic equation ) , and second , does not affect stability of the scheme if we approximate it implicitly and solve the resulting non - linear algebraic equations .
a more detailed description of this modification of the splitting scheme of hout and welfert will be presented elsewhere .
implementation of the numerical algorithm is done similar to @xcite . in typical tests
we use a non - uniform finite difference grid in @xmath165 and @xmath107 of size 50x50 nodes , where @xmath247 .
the number of steps in time depends on maturity @xmath9 , because we use a fixed time step @xmath248 0.1 yrs .
typical computational time for @xmath249 yrs on the same pc at @xmath37 = 0.03 is 17 sec . in fig [ testuv3d ]
a 3d plot of the value function @xmath250 is presented for the initial parameters marked in table [ tab ] as test 1. we also use @xmath251 .
it is seen that @xmath250 quickly goes to constant outside of a narrow region around @xmath252 and @xmath253 .
@xmath254 & @xmath45 & @xmath187 & @xmath94 & @xmath255 & @xmath256 & @xmath169 & @xmath257 & @xmath95 & @xmath258 & @xmath259 & @xmath127 & @xmath260 1 & 0.04 & 0.25 & 0.02 & 0.8 & 110 & 0.05 & 0.2 & 0.4 & 100 & 90 & 0.03 & 0.3 & 0.3 & 100 2 & 0.04 & 0.25 & 0.02 & 0.8 & 110 & 0.05 & 0.3 & 0.3 & 50 & 90 & 0.03 & 0.3 & 0.2 & 100 in fig [ comp1 ] the same quantities are computed for @xmath261 = 1.22 and @xmath26210 yrs . here
the first plot presents comparison of the numerical solution with zero and 0 + 1 approximations .
the second plot compares the zero and first order approximation .
it is seen that the first approximation makes a small correction to the zero one in the region closed to @xmath252 . also both 0 " and 0 + 1 approximations fit the exact numerical solution relatively well .
this proves that our asymptotic closed form solution is robust .
results obtained with a second set of parameters ( test 2 in the table [ tab ] ) are shown in fig [ comp2 ] . in fig [ price ]
we present price @xmath263 computed in tests 1,2 as a function of @xmath264 , where @xmath10 is black - scholes put price with parameters of the corresponding tests .
note that for values of parameters used above the nonlinear term is small , therefore the solution is closed to the solution of the linear 2d heat equation obtained from the eq.([phi_uv_2 ] ) by omitting the nonlinear term . that is because @xmath37 is small in our test 1 . to make testing more interesting , we changed @xmath37 to @xmath265 .
results obtained for @xmath9 = 3 yrs are presented in fig [ compgamma3 ] for test 1 . and in fig [ compgamma4 ] for test 2 .
it is seen that in this case the 0 + 1 analytical approximation still fits the numerical solution .
the computational time in these tests is higher because the matrix root solver converges slower .
the typical time at @xmath9=3yrs and @xmath47 = 1 is 24 secs .
computation of the analytical approximation requires the same time as before which is about 1 sec .
if the exponent of the payoff function is positive , e.g. at @xmath47 = 2 and other parameters as in test 1 , then the solution looks like a delta function . under such conditions any numerical method experiences a problem being unable to resolve very high gradients within just few nodes .
therefore , in this case one has to exploit a non - uniform grid saturated close to the peak of the value function .
this brings extra complexity to the numerical scheme while our analytical approximation is free of such problems . on the other hand
it is natural to statically hedge @xmath8 option with the option @xmath6 with same or close moneyness .
this significantly reduces the exponent and allows one to use the same mesh even at high values of the risk aversion parameter @xmath37 . as the numerical performance of the model depends on the value of @xmath266 , a few comments are due at this point . though we do not address calibration of our model in the present paper ( leaving it for future work ) , the value of @xmath266 should be found by calibrating the model to market data .
it is not entirely obvious how to do this since we use an illiquid asset @xmath1 , and moreover the price of our complex option @xmath263 is not an additive sum of its components for incomplete markets .
it therefore makes sense to calibrate the model to another set of instruments that are both liquid and strongly correlated with the original instruments . in the context of equity options , such calibration of @xmath266
was done in @xcite , giving rise to values of @xmath267 .
while it is not exactly clear how @xmath37 found in this setting is related to @xmath37 of our original problem , we expect the latter to be of the same order of magnitude .
note that the results in fig.[compgamma4 ] clearly demonstrate that the proposed method is just asymptotic .
indeed , the first correction to the solution in fig.[compgamma4 ] is a few times larger then the zero - order solution .
as usual , there exists an optimal number of terms in the asymptotic series that fit the exact solution best .
we do not pursue such analysis in this paper , however due to our definition of @xmath100 via @xmath187 this is not an interesting case . ] . to characterize sharpness of the peak in the value function
it is further convenient to introduce a new parameter @xmath268 .
if @xmath269 is high , e.g. @xmath270 the value function is almost a delta function , so the asymptotic solution as well as its numerical counterpart are not expected to produce correct results unless they are further modified . based on the results of @xcite one can see that the value of @xmath269 calibrated to the market varies from 0.01 to 15 .
therefore , in our test 2 we used @xmath269 = 20 .
both our numerical and asymptotic methods work with no problem for these values of @xmath271 . in fig [ alphastar ] the optimal hedge @xmath272
is computed based on eq.([indif_price_2 ] ) which was solved using brent s method ( @xcite ) .
the initial parameters correspond to test 1 in table [ tab ] in the first plot , and test 2 in the second plot .
note that for @xmath273 ( the first plot ) and @xmath274 ( the second plot ) , eq.([indif_price_2 ] ) does not have a minimum , so the maximum is obtained at the edge of the chosen interval of @xmath47 .
the latter could be defined based on some other preferences of the trader , for instance , the total capital she wants to invest into this strategy etc .
finally , in fig [ pricealphagamma ] price @xmath263 is presented as a function of @xmath47 for various @xmath37 .
it is seen that this function is convex which was first showed in @xcite in a different setting .
note that these results were obtained using a new numerical method mentioned above .
it combines strang s splitting with the fast gaussian transform , and accelerates calculations approximately by factor 40 as compared with a non - linear version of the 2d crank - nicholson scheme .
a detailed description of the method will be given elsewhere .
we thank peter carr and attendees of the `` global derivatives usa 2011 '' conference for useful comments . i.h .
would like to thank andrew abrahams and julia chislenko for support and interest in this work .
we assume full responsibility for any remaining errors .
benth , f.e . ,
groth , m. , & lindberg , c. 2010 .
the implied risk aversion from utility indifference option pricing in a stochastic volatility model .
, * 16*(m10 )
. dash , j. 2004 . .
world scientific .
davis , m. 1997 .
option pricing in incomplete markets .
dempster , m. , & pliaka , stanley ( eds ) , _ mathematics of derivative securities_. cambridge : cambridge university press .
davis , m. 1999 .
option valuation and basis risk .
djaferis , t.e . , & shick , l.c .
( eds ) , _ system theory : modelling , analysis and control_. academic publishers .
henderson , v. , & hobson , d. 2002 . substitute hedging .
, * 15*(5 ) , 7175 .
henderson , v. , & hobson , d. 2009 .
utility indifference pricing : an overview .
carmona , r. ( ed ) , _ indifference pricing_. princeton university press .
hodges , s.d .
, & neuberger , a. 1989 . optimal replication of contingent claims under transaction costs . , * 8 * , 222239 .
ilhan , a. , & sircar .
, r. 2006 .
optimal static - dynamic hedges for barrier options .
, * 16*(2 ) , 359385 .
, k. j. , & welfert , b. d. 2007 .
stability of adi schemes applied to convection - diffusion equations with mixed derivative terms .
, * 57 * , 1935 .
itkin , a. , & carr , p. 2011 .
jumps without tears : a new splitting technology for barrier options . , *
8*(4 ) , 667704 .
jaimungal , s. , & sigloch , g. 2009 . .
working paper .
khiari , n. , & omrani , k. 2011 .
finite difference discretization of the extended fisher - kolmogorov equation in two dimensions . , * 62 * , 41514160 .
landau , l.d . , lifshitz , e.m .
pergamon press , new york .
merton , r. 1974 . on the pricing of corporate debt : the risk structure of interest rates .
, * 29 * , 449470 .
musiela , m. , & zariphopoulou , t. 2004 .
an example of indifference prices under exponential preferences .
, * 8 * , 229239 .
polyanin , a.d .
chapman & hall / crc .
salleh , s. , zomaya , a. , & bakar , s. 2007 . .
wiley - interscience .
t. leung , r. sircar , & zariphopoulou , t. 2008 .
credit derivatives and risk aversion . , * 22 * , 275291 .
trolle , a.b . , & schwartz , e.s . 2009 .
unspanned stochastic volatility and the pricing of commodity derivatives . , * 22*(11 ) , 44234461 .
here we describe in more detail the solutions for the zero and first order approximation in @xmath100 , and outline a generalization of our approach to an arbitrary order in @xmath100 . in the zero order approximation @xmath275 the eq.([phi_uv_2 ] ) does not contain derivatives wrt @xmath107 : @xmath276 the solution of this equation proceeds along similar lines to sect .
[ sect : zeroorder ] using a change of dependent variable @xmath277^{1/(1-{\rho}_{xz}^2)}\ ] ] which reduces eq.([zeroeq2 ] ) to the heat equation @xmath198 subject to the initial condition @xmath278 .
the explicit form for the latter coincides with ( [ init_cond_quand2 ] ) provided we substitute @xmath279 , @xmath280 , @xmath281 and @xmath282 .
the explicit zero - order solution thus reads ( compare with eq.([phi_0_sol ] ) ) @xmath283^{1/(1-{\rho}_{xz}^2)}\ ] ] note that eq.([phi_0_sol2 ] ) provides the general zero - order solution for the hjb equation with arbitrary initial conditions at @xmath208 . for our specific initial conditions eq.([init_cond_quand2 ] ) , the solution is readily obtained in closed form in terms of the error ( or normal cdf ) function ( see appendix [ a1 ] ) . for the first correction in the eq.([phi_uv_2 ] ) we obtain the following pde @xmath284 where @xmath285 .
this equation coincides with eq.([first2 ] ) except that the free term is different , and the correlation parameter is @xmath149 rather than @xmath286 .
the solution proceeds as in sect .
[ sect : firstorder ] , resulting in the following expression : @xmath287 the double integral that enters this expression can be split out in two parts .
one of them could be found in closed form , while the other one requires numerical computation ( see appendix [ a2 ] ) .
our numerical tests show that for many sets of the initial parameters the first integral is much higher than the second one , so the latter could be neglected .
however , we were not able to identify in advance at which particular values of the parameters this could be done . moreover , for some other initial parameters we observe an opposite situation .
the second order equation has the same form as eq.([first ] ) @xmath288 where @xmath289 as this equation has to be solved also subject to zero initial conditions , the solution is obtained in the same way as above : @xmath290 this shows that in higher order approximations in @xmath100 the type of the equation to solve does nt change as well as the initial conditions .
therefore , the solution to the @xmath220-th order approximation reads @xmath291 where @xmath222 can be expressed via already found solutions of order @xmath223 and their derivatives on @xmath165 and @xmath107 .
the exact representation for @xmath222 follows combinatorial rules and reads @xmath292 where @xmath225 is a coefficient at @xmath293 in the following expansion @xmath294 ^ 2 , \qquad \beta_3 = \sum_{i=1}^\infty \varepsilon^i \phi_i/\phi_0 \nonumber\end{aligned}\ ] ] in particular , @xmath228 reads @xmath295 the explicit representation of the solutions of an arbitrary order in quadratures is important because per our definition of @xmath100 the convergence of the eq.([expan ] ) is expected to be slow .
indeed , if one wants the final precision to be about 0.1 at @xmath231 , the number of important terms @xmath233 in the expansion eq.([expan ] ) could be rawly calculated as @xmath296 , which gives @xmath297 .
all the integrals with @xmath238 do not admit a closed form representation and have to be computed numerically .
since our payoff is a piece - wise function , the integral in eq.([phi_0_sol ] ) can be represented as a sum of three integrals .
we denote @xmath298 and represent the zero - order solution as follows : @xmath299^{\frac{1}{1-\rho_{xz}^2 } } , \qquad \zeta = { \rm sign}(v ) \\\ ] ] where sign @xmath300 means that @xmath301 , and sign @xmath302 - that @xmath303 , and @xmath304 \\ & = \omega\left(-\omega , - \bar{\gamma } k_z e^{\sigma_z\frac{\rho_{xz}}{\rho_{xy}}\omega } , \sigma_z \frac{\rho_{xz}}{\rho_{xy } } , \bar{\gamma } \alpha k_y , \sigma_y\right ) , \nonumber \\
j_2^{(+ ) } & = \frac{1}{\sqrt{2 \pi \tau}}\int_{- \omega}^{0}d u ' e^{-\frac{(u ' - u)^2}{2 \tau } } \exp\left
[ - \bar{\gamma } \left(k_z - \alpha k_y e^{\sigma_y u'}\right ) \right ] , \nonumber \\ & = \omega\left(0 , - \bar{\gamma } k_z , 0 , \bar{\gamma } \alpha k_y , \sigma_y\right ) - \omega\left(-\omega , - \bar{\gamma } k_z , 0 , \bar{\gamma } \alpha k_y , \sigma_y\right ) , \nonumber \\
j_3^{(+ ) } & = \frac{e^{- \bar{\gamma } \left(k_z - \alpha k_y\right)}}{\sqrt{2 \pi \tau}}\int_{0}^{\infty}d u ' e^{-\frac{(u ' - u)^2}{2 \tau } } = e^{- \bar{\gamma } \left(k_z - \alpha k_y\right ) } \left[1 -\frac{1}{2 } { \rm erfc}\left(\frac{u}{\sqrt{2\tau}}\right)\right ] , \nonumber \\ j_1^{(- ) } & = \frac{1}{\sqrt{2 \pi \tau}}\int_{- \infty}^{0}d u ' e^{-\frac{(u ' - u)^2}{2 \tau } } \exp \left [ - \bar{\gamma } \left(k_z e^{\sigma_z \frac{\rho_{xz}}{\rho_{xy } } ( \omega + u ' ) } - \alpha k_y e^{\sigma_y u ' } \right ) \right ] \nonumber \\ & = \omega\left(0 , - \bar{\gamma } k_z e^{\sigma_z\frac{\rho_{xz}}{\rho_{xy}}\omega } , \sigma_z \frac{\rho_{xz}}{\rho_{xy } } , \bar{\gamma } \alpha k_y , \sigma_y\right ) , \nonumber \\
j_2^{(- ) } & = \frac{1}{\sqrt{2 \pi \tau}}\int_{0}^{-\omega}d u ' e^{-\frac{(u ' - u)^2}{2 \tau } } \exp\left [ - \bar{\gamma } \left ( k_z e^{\sigma_z \frac{\rho_{xz}}{\rho_{xy}}(\omega + u ' ) } - \alpha k_y \right ) \right ] \nonumber \\ & = \omega\left(-\omega , - \bar{\gamma } k_z e^{\sigma_z\frac{\rho_{xz}}{\rho_{xy}}\omega } , \sigma_z \frac{\rho_{xz}}{\rho_{xy } } , \bar{\gamma } \alpha k_y , 0\right ) - \omega\left(0 , - \bar{\gamma } k_z e^{\sigma_z\frac{\rho_{xz}}{\rho_{xy}}\omega } , \sigma_z \frac{\rho_{xz}}{\rho_{xy } } , \bar{\gamma } \alpha k_y , 0\right ) , \nonumber \\
j_3^{(- ) } & = \frac{e^{- \bar{\gamma } \left(k_z - \alpha k_y \right)}}{\sqrt{2 \pi \tau}}\int_{-\omega}^{\infty}d u ' e^{-\frac{(u ' - u)^2}{2 \tau } } = \frac{1}{2 } e^{- \bar{\gamma } \left(k_z - \alpha k_y \right ) } { \rm erfc}\left(-\frac{u+\omega}{\sqrt{2\tau}}\right ) .
\nonumber\end{aligned}\ ] ] here @xmath305 is the complementary error function , and @xmath306 u ' } e^{-\frac{(u ' - u)^2}{2 \tau } } du ' \nonumber \\ & = \frac{1}{2}\sum _ { i=0}^\infty \sum _ { j=0}^i \frac{\delta^{i - j } \beta^j}{j ! ( i - j ) ! } e^{a\left(u+a
\frac{\tau}{2}\right)}{\rm erfc}\left(\frac{u - a + a \tau}{\sqrt{2\tau } } \right ) , \nonumber \\ a & = i p + j ( q - p ) .
\nonumber\end{aligned}\ ] ] since the complementary error function quickly approaches zero with @xmath307 , or 1 with @xmath308 , the number of terms one has to keep in the above sums should not be high .
we also need the following function @xmath309 when computing the first order approximation . using a general form of the payoff @xmath310 , ) . ]
it can be represented in the form @xmath311 where @xmath312 and @xmath313 , \\
\lambda(b ) & \equiv e^{a(b)\left[u+a(b ) \frac{\tau}{2}\right]}{\rm erfc}\left(\frac{u - a + a(b ) \tau}{\sqrt{2\tau } } \right ) , \qquad a(b ) \equiv i p + j ( q - p ) + b. \nonumber\end{aligned}\ ] ]
the first order approximation is given by eq.([firstordersol ] ) where the zero - order solution @xmath314 has been already computed in appendix [ a1 ] .
we plug in this solution into eq.([firstordersol ] ) to obtain @xmath315 \nonumber\end{aligned}\ ] ] where the integrals @xmath316 are defined in eq.([jdef ] ) .
the internal integral can be simplified .
indeed , since @xmath317 the internal integral in eq.([int1app ] ) can be represented as a sum of two integrals @xmath318 , where @xmath319 as we already mentioned the second integral could be either smaller or larger than the second one depending on the parameters .
this is illustrated in fig .
[ figj2 ] .
the initial parameters are taken from test 1 in table [ tab ] with @xmath47 = 1 and @xmath9 = 3 yrs . in fig .
[ figj1 ] the same calculation is shown for the parameters corresponding to test 2 in table [ tab ] .
the first integral @xmath320 can be rewritten using the integral representation of @xmath321 in eq.([phi_0_sol ] ) @xmath322 substituting this into eq.([int1app ] ) and integrating over @xmath323 , we find @xmath324 thus , the first correction to the solution obtained in the zero order approximation on @xmath100 is approximately @xmath325 , and the full solution in the `` 0 + 1 '' approximation reads @xmath326
the first order approximation is given by eq.([firstordersol ] ) .
it is assumed that the zero - order solution @xmath314 is already computed by using either the method presented in appendix [ a1 ] , or using the fast gauss transform .
we plug in this solution into eq.([firstordersol ] ) to obtain @xmath327 \nonumber\end{aligned}\ ] ] where the integrals @xmath316 are defined in eq.([jdef ] ) .
it can be seen that eq.([int1app2 ] ) is similar to eq.([int1app ] ) if one replaces @xmath169 with @xmath328 and @xmath329 with @xmath330 .
therefore , we can use the same idea as in appendix [ a2 ] to further simplify this integral .
the first integral @xmath320 can be modified using the integral representation of @xmath321 in eq.([phi_0_sol ] ) @xmath333 substituting this into eq.([int1app2 ] ) and integrating over @xmath323 , we find @xmath334 | this work addresses the problem of optimal pricing and hedging of a european option on an illiquid asset z using two proxies : a liquid asset s and a liquid european option on another liquid asset y. we assume that the s - hedge is dynamic while the y - hedge is static . using the indifference pricing approach we derive a hjb equation for the value function , and solve it analytically ( in quadratures ) using an asymptotic expansion around the limit of the perfect correlation between assets y and z. while in this paper we apply our framework to an incomplete market version of the credit - equity merton s model , the same approach can be used for other asset classes ( equity , commodity , fx , etc . ) , e.g. for pricing and hedging options with illiquid strikes or illiquid exotic options . |
the primary source of energy for the earth s atmosphere is the sun , so it is reasonable to explore whether changes in the global climate result from solar variability .
it was first suggested by the astronomer william herschel ( @xcite ) that variations in the solar irradiance caused by sunspots could lead to climatic changes on earth , and he cited the variation of british wheat prices with sunspot number as evidence for this link .
the occurrence of the `` little ice age '' during the 1645 - 1715 maunder sunspot minimum ( @xcite ) , the correlation between the long - term solar cycle variations and tropical sea surface temperatures ( @xcite ) , polar stratospheric temperatures ( @xcite ) , and the width of tree rings ( @xcite ) , along with many other studies also support a link between solar variations and the earth s climate .
a direct link between the sun and these phenomena is tenuous , however , because the magnitude of the solar irradiance variation over the 11-year solar cycle is very small . over the 1979 - 1990 solar cycle ,
for example , the variation in the irradiance was only @xmath0 ( @xcite ) , or @xmath1 w m@xmath2 globally - averaged at the top of the atmosphere .
this is insufficient to power the sea surface temperature changes associated with the solar cycle by a factor of @xmath3 ( @xcite ) , and is significantly smaller than the globally - averaged forcings due to clouds ( @xmath4 w m@xmath2 ; e.g. @xcite ) , anthropogenic greenhouse gases ( @xmath5 w m@xmath2 ; @xcite ) , and anthropogenic aerosols ( @xmath6 w m@xmath2 ; @xcite ; @xcite ) , suggesting that any direct atmospheric forcing from solar irradiance variations would be relatively unimportant .
an indirect link between solar cycle variations and the earth s climate appears more likely , especially given the discovery of a link between the flux of galactic cosmic rays ( gcrs ) and global cloudiness ( @xcite ) in the isccp cloud database ( @xcite ) .
the sun modulates the gcr flux at the earth through the action of the solar wind , which scatters and attenuates the gcrs in times of heightened solar activity ( solar maximum ; e.g. @xcite ) .
using @xmath7 @xmath8 m infrared ( ir ) cloud amounts from the isccp database for the years 1983 - 1993 , marsh and svensmark ( 2000 ) and pall bag and butler ( 2000 ) showed that there is evidence of a positive gcr - cloud correlation only for low ( @xmath9 km ) clouds , and that the effect of the cosmic rays on global cloud amount appears to be greatest at the low to mid latitudes .
the globally - averaged forcing due to the increase in low clouds associated with the solar cycle gcr variations is estimated ( @xcite ) to be approximately @xmath10 w m@xmath2 , which is sufficient to power the sea surface temperature variations ( @xcite ) .
this is also comparable in magnitude ( but opposite in sign ) to the forcing due to anthropogenic co@xmath11 emission over the last century ( @xcite ) . decreasing local cloud amounts correlated with short - term forbush decreases in cosmic - ray rates were observed by pudovkin and veretenko ( 1995 ) .
the reality of the gcr - cloud connection has been questioned by a number of authors ( @xcite ; @xcite ; @xcite ) .
these objection can be distilled into three main points : 1 ) the gcr - cloud correlation should be seen prominently in high ( cirrus ) clouds at high latitudes where the cosmic - ray intensity is highest , 2 ) the increased cloudiness can be more plausibly attributed to other phenomena instead of gcrs , and 3 ) the correlation is an artifact of the isccp analysis .
the first objection is addressed by the theory of ion - mediated nucleation ( imn : @xcite ; yu 2002 ) , in which the efficiency of the cosmic - ray interaction is limited at high altitudes by the lack of aerosol precursor vapors such as h@xmath11so@xmath12 relative to the ion concentration . for the second objection , the temporal profile of the gcr - cloud correlation may be inconsistent with the profiles of the dominant volcanic and el nio / southern oscillation ( enso ) events during the same time period ( @xcite ) , although no quantitative study of the various temporal signatures in the data has been undertaken . finally , the isccp artifacts pointed out by norris ( 2000 ) are troubling , but it is not clear that they are of sufficient magnitude to produce the observed gcr - cloud correlation , and it does nt explain why the correlation exists only for low clouds and not the other cloud types in the isccp database .
the linkage between cosmic rays and cloud formation has been recently investigated by a number of authors ( @xcite ; @xcite ; @xcite and references therein ) . here
we apply a perturbative approach to quantify the effects of variations in the cosmic - ray rate on the optical thicknesses , or opacities , of clouds , and use the observed cloud opacity variations to constrain the microphysical models of ion - mediated ultrafine particle formation .
the paper is organized as follows . in the next section
we discuss how the effect of cosmic rays could alter the optical thickness and emissivity of clouds by affecting the nucleation of condensation nuclei ( cn ) .
the search for variations in cloud optical properties using the isccp database and their correlation with cosmic - ray flux variations are discussed in section [ variations ] a discussion of the results is given in section [ discussion ] , and finally we summarize our results in section [ summary ]
cosmic rays form water droplets in the supersaturated air of a classical cloud chamber ( @xcite ) , and it seems plausible that they could also play a significant role in natural cloud formation .
yu and turco ( 2000 , 2001 ) and yu ( 2002 ) have investigated the formation of ultrafine cn from charged molecular clusters formed from cosmic - ray ionization , and they find that the charged clusters grow more rapidly and are more stable than their neutral counterparts up to a size of @xmath13 nm . although the subsequent growth of the cosmic - ray formed ultrafine cn to viable @xmath14 nm cloud condensation nuclei ( ccn ) has not been explored , the concentration of ccn should also reflect the cn concentration , as well as the direct influence of cosmic rays , if the cosmic - ray ionization rate does not affect other important nucleation efficiency parameters such as condensible vapor concentration , temperature , and pressure .
we will make this assumption here although it may not be strictly true with respect to the condensible vapor concentration ( see e.g. @xcite ; @xcite ) .
althought the formation of ccn and ultimately cloud droplets is a function of many variable factors such as temperature , pressure , vapor concentration , and relative humidity , we can quantify the effects of small variations in the ionization rate ( primarily due to cosmic rays above ocean and at altitudes @xmath15 km above land ; e.g. @xcite ) on the number of ccn through a perturbation approach , i.e. @xmath16 where @xmath17 is the concentration of ccn , @xmath18 is the ionization rate , @xmath19 refers to the set of parameters other than the ionization rate affecting @xmath17 , and the partial derivative is evaluated for fixed @xmath19 ( hereafter this will not be written explicitly ) . along with the assumption discussed previously , this approach assumes that the quantity @xmath20 , which is probably true for solar cycle variations , where @xmath18 typically varies by @xmath21 , but may not be true during periods of large scale changes in the geomagnetic field ( e.g. @xcite ) . to quantify the effect of varying ccn concentrations on cloud optical thicknesses , we envision the two idealized scenarios depicted in figure 1 . in both cloud formation scenarios , changes in the ionizing cosmic - ray flux cause changes in the number of cloud condensation nuclei through the process of ion - mediated nucleation on the formation of ultrafine cn in accordance with the assumptions mentioned above in figure 1 , which need not be valid for all @xmath18 . ] . in the first case
we assume that the nucleation of cloud droplets is limited by the available amount of water in the supersaturated air , so that the liquid water content ( lwc ) , or density of water in droplets , is constant .
therefore the amount of water per droplet and hence the effective radii of cloud droplets will change with the cosmic - ray ionization rate .
this is analogous to the `` twomey effect '' of enhanced aerosol pollution on droplet size distributions and the albedo of clouds ( @xcite ; @xcite ) , and would primarily occur in environments where the amount of water in the air ( and not the number of ccn ) is the limiting factor .
thus , using ( [ perturb ] ) , we would expect that the effective radius @xmath22 of the cloud droplet distribution resulting from a small change in the cosmic - ray ionization rate @xmath23 in any particular volume of air will be @xmath24}^{1/3}r_{eff}^{0}\approx{\left ( 1 + { \delta q\over n_{\rm{ccn}}}{\partial n_{\rm{ccn}}\over \partial q}\right ) } ^{-1/3 } r_{eff}^{0},\ ] ] where @xmath25 is the effective radius of the unperturbed droplet distribution , which we will associate with the solar maximum period of the solar cycle . in the second case in figure 1
, we assume that the change in ccn concentration resulting from change in cosmic - ray ionization causes a proportionate change in the amount of water extracted from the supersaturated air , with the effective radius of the cloud droplet distribution remaining constant .
this is the case where the formation of the cloud is limited by the local availability of ccn and not condensible water .
this effect has been seen in the marine boundary layer in ship track clouds ( @xcite ) , which have higher reflectivities ( @xcite ) and liquid water contents ( @xcite ) due to the formation of additional ultrafine cn from ship exhaust .
the perturbed liquid water content of a cloud in any particular volume of air will then be given by @xmath26 where @xmath27 is the unperturbed cloud liquid water content associated with solar maximum as before .
these two scenarios probably represent extremes of the direct cosmic - ray ionization effect on the clouds . as in the ship track clouds , the effect of the gcrs
will probably be a combination of both lwc changes and @xmath22 changes , with the magnitude of the effect being bounded by the changes given in ( [ reff ] ) and ( [ lwc ] ) .
changes in the cloud liquid water content and droplet effective radius , associated with changes in the ionization rate due to cosmic rays , will result in changes in cloud opacities .
the optical thickness @xmath28 of a uniform cloud layer of thickness @xmath29 is given by ( @xcite ) : @xmath30 where @xmath31 is the concentration of cloud droplets with radii between @xmath32 and @xmath33 , @xmath34 is the mie extinction efficiency , and it is commonly assumed that @xmath35 which is a good approximation when @xmath36 , where @xmath37 is the wavelength ( @xcite ) .
the effective radius of the cloud droplet distribution is given by @xmath38 and the cloud liquid water content is given by @xmath39 where @xmath40 is the density of liquid water .
combining these equations , we see that @xmath41 thus from ( [ tau2 ] ) we would expect that an increase ( decrease ) in the mean @xmath22 and a decrease ( increase ) in the mean @xmath42 , resulting from ionization variations due to cosmic rays , would result in a decrease ( increase ) the mean opacity of clouds . the change in cloud opacity with cosmic - ray rate
can be quantified using the perturbation assumptions discussed in section [ nucleation ] and equations ( [ reff ] ) , ( [ lwc ] ) , and ( [ tau2 ] ) . the fractional change in cloud opacity
is then given by @xmath43 where @xmath44 for ccn ( water ) limited cloud formation , and the fractional change in the perturbed opacity @xmath28 ( relative to the unperturbed opacity @xmath45 ) is defined by @xmath46 . as mentioned previously
, this derivation assumes that the right hand side of ( [ dtaureff ] ) is much less than one , which may not be the case for large changes in @xmath18 and @xmath17 . as before we will assume that the unperturbed ( perturbed ) values of @xmath18 and @xmath17 refer to the values at solar maximum ( minimum ) . at visible wavelengths from space
, the primary consequence of the change in cloud opacity associated with cosmic rays will be an increase in cloud reflectivity , or albedo . to investigate this
, we use the radiative transfer code sbdart ( @xcite ) to calculate the top of the atmosphere broadband ( @xmath47 @xmath48 ) upward flux for three uniform low cloud models : 1 ) a @xmath49 km thick cloud layer extending to a height of @xmath50 km , 2 ) a @xmath50 km thick cloud extending to a height of @xmath51 km , and 3 ) a @xmath52 km cloud layer extending to @xmath53 km .
these simulations were done with a tropical atmosphere profile ( @xcite ) and an ocean surface albedo .
the fractional increases in albedo , resulting from a 10% increase in the number of cloud droplets due to cosmic ray ionization variations , is shown in figure 2 for the @xmath49 km thick cloud case , for a wide range of lwc and @xmath22 in the variable lwc case ( top panel ) and the variable @xmath22 case ( bottom panel ) . in both cases
the contours of changing albedo approximately parallel the change in optical thickness calculated assuming @xmath54 figure 3 shows the fractional change in albedo directly as a function of opacity for all three cloud models .
this figure clearly shows that the change in albedo is largest for clouds with opacities @xmath28 between 1 and 10 , but is roughly independent of cloud geometrical thickness .
figures 2 and 3 indicate that the change in cloud optical thickness can be used to quantify the effects of the cosmic rays on cloud optical properties .
although the fractional change in albedo due to the cosmic rays is only @xmath55 for a @xmath56 variation in the number of cloud droplets , this can produce a significant forcing per cloud of @xmath57 w m@xmath2 at the top of the atmosphere for a solar zenith angle of @xmath58 .
the modulation of cloud opacity due to cosmic rays could therefore produce a similar modulation of the earth s energy budget over the 11 year solar cycle , although the exact amount of forcing due to cosmic rays will depend sensitively on cloud amount variations , cloud opacity variations , and the efficiency at which changes in the cosmic ray rate are reflected in the number of cloud condensation nuclei .
because of the relationship between cloud opacity and emissivity , the cosmic rays should also produce an observable effect on cloud emission at infrared ( ir ) wavelengths .
the effective ir emissivity @xmath59 can be parameterized by a relation of the form ( @xcite ) : @xmath60 where @xmath61 is the mass absorption coefficient .
empirical fits to ir emission from water clouds yield @xmath62 ( @xcite ) .
the exponent in ( [ emiss ] ) is proportional to the cloud optical thickness for a given droplet effective radius , so the infrared emissivity increases with cloud opacity , with the change being most noticeable for optically thin clouds .
therefore one would expect a change in ir emission , along with the primary effect of changes in visible albedo , from clouds at solar minimum relative to clouds at solar maximum if the cosmic rays change the cloud liquid water contents .
interestingly , a correlation between cosmic ray rate and cloud top temperature for low clouds has been reported by marsh and svensmark ( 2000 ) , supporting this hypothesis .
to search for systematic temporal changes in synoptic scale cloud optical properties , we used the international cloud climatology project ( isccp ) monthly gridded cloud products ( `` d2 '' ) datasets , a compilation of cloud properties derived from satellite observations during the period 19831999 ( @xcite ) .
the isccp d2 data used here consists of mean daytime cloud amount fractions and visible optical depths , as a function of time , for @xmath63 `` boxes '' with equal area covering the entire surface of the earth . for a given time
, the cloud amount fraction in each box is defined as the number of cloudy satellite image pixels , as determined by a cloud detection algorithm , divided by the total number of pixels in the box .
the cloud optical thicknesses are derived from the visible satellite cloud albedos by using a radiative transfer model and assuming spherical droplets with droplet sizes characterized by a gamma distribution with variance @xmath64 and @xmath65 @xmath8 m .
isccp cloud top temperatures are simultaneously determined from the @xmath66 ir radiances , allowing for determination of cloud altitude and pressure , and the low , mid - level , and high clouds are defined as having cloud top pressures @xmath67 mb , @xmath68 mb , and @xmath69 mb , respectively . because we require the simultaneous visible and infrared radiances to determine the opacity and cloud height for our analysis , we only use the isccp daytime data .
this is a different dataset than the diurnal 1983 - 1993 ir data used for the cloud amount analyses of marsh and svensmark ( 2000 ) and pall bag and butler ( 2000 ) .
detailed information on the distribution of cloud optical thicknesses is not preserved in the the isccp d2 database , and instead the mean optical thickness @xmath70 is recorded for three broad opacity bands @xmath71 : @xmath72 , @xmath73 , and @xmath74 .
thus a detailed analysis of the change in @xmath28 over the solar cycle is not possible using the d2 data , but a value of the _ weighted _ mean cloud optical thickness @xmath75 can be calculated using @xmath76 where the @xmath77 are the total mean cloud amount fractions within each of the broad isccp optical thickness bins mentioned above .
we calculated @xmath75 separately for the three cloud altitude levels and for two latitude bands with @xmath78 ( low latitude ) and @xmath79 ( high latitude ) .
the error associated with each @xmath70 was estimated by calculating the standard deviation of each isccp data point , from the scatter about the mean , and scaling by the square root of the number of data points .
the mean optical thicknesses @xmath75 as a function of time for the low latitude clouds are shown in figure 4 , and the corresponding result for global high latitude clouds is shown in figure 5 . shaded is the two year period in which the effects of the mt .
pinatubo eruption appear to be most significant .
also shown for comparison are the mean counting rates from the climax , colorado neutron monitor run by the university of chicago ( obtained from http:@xmath80ulysses.uchicago.edu@xmath81neutronmonitor@xmath81neutron_mon.html ) , which is a good measure of the local cosmic - ray ionization rate . in the low latitude case ,
the abrupt and large decrease in @xmath75 during 19911993 is due to the eruption of mt .
pinatubo , and the subsequent plot scaling obscures smaller scale opacity variations . for comparison
, we also plot the total mean cloud amount fractions @xmath82 for the same two latitude bands in figures 6 and 7 .
these plots show evidence for increases in mean cloud amount due to mt .
pinatubo , as well as the smaller - scale temporal variations . to search for subtle variations in the isccp cloud opacities and amounts due to cosmic rays only , it is first necessary to eliminate the opacity variations in the data due to the mt .
pinatubo volcanic eruption in june - september 1991 and strong enso events during the period of the isccp data . to separate out the various temporal signatures in the isccp data , we use a linear temporal model of the form @xmath83 where @xmath84 is the mean isccp quantity of interest for the year @xmath85 , which for our purpose is either the visible cloud opacity @xmath75 or the mean cloud amount / fractional area @xmath86 .
the model consists of four temporal basis vectors @xmath87 , which are functions of time , each scaled by a linear coefficient @xmath88 . for our temporal model
we choose basis vectors corresponding to constant level of the given quantity ( @xmath89 ) and variations due to enso events ( e.g. @xcite ) , the mt .
pinatubo eruption of 1991 , and cosmic rays ( @xmath90-@xmath51 , respectively ) .
given the functional form of the basis vectors , the best - fit values of the linear coefficients can be determined through least squares minimization , and the fractional change in the time - varying isccp quantity over the data stretch is then given by @xmath91 , where @xmath90-@xmath51 . this model assumes a linear correlation between the quantity of interest and the basis vectors and assumes no time delays ; more complicated models are possible but will not be considered here .
the normalized basis vectors used in the temporal analysis of the isccp cloud data are shown in figure 8 .
all of the vectors are scaled to values between zero and one . for the enso term @xmath92 we use the scaled yearly - averaged southern oscillation index ( soi ) from the australian bureau of meteorology ( obtained from http://www.bom.gov.au/climate/current/soihtm1.shtml ) .
the soi is a measure of the size of fluctuations in the sea level pressure difference between tahiti and darwin , australia , and small values of the scaled soi denote el nio conditions and large values la nia both of which affect global weather ( @xcite ) . to parameterize the effect of the mt .
pinatubo eruptions of 1991 , we adopt a simple step function for @xmath93 , with identical non - zero intensities only for years 1991 and 1992 . for the final term in the temporal model , @xmath94 , we use the scaled cosmic - ray rate from the climax neutron monitor .
neutron monitor rates are directly proportional to the ionization rates due to cosmic rays because the neutrons are produced by the same cosmic ray cascade particles that produce the ionization , and the neutrons subsequently diffuse through less than 100 m of air before they are thermalized and captured by n to form @xmath95c ( e.g. lingenfelter , 1963 ) .
neutron counters are thus unsusceptible to background ionizations due to terrestrial radiation from radioactive decays , which dominate the ionization signal from galactic cosmic rays only below @xmath96 km in the atmosphere ( @xcite )
. the results of the temporal fitting of both the isccp visible cloud opacities and amounts are shown in table 1 .
formally most of the fits are not good , with reduced chi - squares ranging from @xmath97 for twelve degrees of freedom .
there are a number of possible factors that could be contributing to this .
for example the error bars on the data may have been underestimated , leading to an artificially large values of chi - squared .
another possibility is that our fitting model is missing other significant temporal drivers , or perhaps a non - linear model or different basis vectors may be required to fit the data .
we tried to fit the isccp data with linear models composed of different combinations of our four basis vectors , and models with the cosmic - ray term provided a better fit to the data in general .
nevertheless it is possible that un - modeled phenomena mimic the temporal signature of cosmic rays in the data ; more robust calculations of isccp error bars , inclusion of more isccp data , and exploration of more complicated temporal models in future work will help resolve this issue .
the fractional variation in visible opacity @xmath98 associated with the cosmic rays ranges from @xmath99 for high clouds to @xmath100 for low clouds . for the mean visible cloud amounts
the variation due to cosmic rays is just the opposite becoming greater in magnitude as the cloud height
_ decreases _ qualitatively consistent with the positive correlation seen in the isccp ir data between cosmic ray rate and low clouds ( @xcite ; marsh and svensmark 2000 ; pall bag and butler 2000 ) .
therefore the high clouds appear to become thicker but smaller in response to increasing cosmic ray flux , while for the low clouds the response is the opposite .
the observed variation of cloud optical thicknesses with cosmic - ray rate can be used to constrain microphysical models of the cloud condensation nuclei concentration @xmath17 using ( [ dtaureff ] ) . of crucial importance
is the partial derivative @xmath101 , which determines the sign of the change in opacity with cosmic - ray rate . recently
yu ( 2002 ) calculated @xmath17 as a function of altitude and ionization rate using an ion - mediated nucleation code .
given this model and the vertical profiles of sulfuric acid vapor concentration , ionization rate , temperature , relative humidity , pressure , and surface area of pre - existing particles assumed therein ( yu 2002 ) , the value of @xmath102 peaks at values of @xmath103 , @xmath104 , and @xmath105 ion pairs @xmath106 for low , mid - level , and high clouds , respectively , such that @xmath107 for @xmath108 and @xmath109 for @xmath110 . using the cosmic - ray ionization rates found by neher ( 1961,1967 ) interpolated to geomagnetic latitude @xmath58 , we find typical ionization rates of , respectively , @xmath111 , @xmath104 , and @xmath112 ion pairs @xmath106 for the low , mid - level , and high isccp clouds .
therefore from ( [ dtaureff ] ) we would expect a positive or zero correlation between opacity and cosmic - ray rate only for low clouds , and negative correlations for higher clouds for this model .
we observe just the opposite , but the precision of the temporal model fits to the isccp data is not sufficient for us to rule out the yu ( 2002 ) model based on the data .
all three of the time - varying parameters in our temporal model show inverse correlations between mean visible cloud opacity and amount , suggesting a common origin for this behavior .
these inverse correlations are illustrated in figure 9 .
these are probably not artifacts of the averaging process because the quantities in figure 9 have been normalized by their constant model terms in their temporal fits .
one possible explanation for the inverse opacity - amount correlation is via a feedback mechanism . for the case of positive opacity variations , an increase in mean cloud opacity and albedo would result in increased energy loss to space and eventually less surface heating and subsequent water evaporation . hence clouds
would tend to be smaller and have smaller areas than they would otherwise .
conversely , for negative opacity variations clouds would tend to be larger .
global climate simulations ( @xcite ) indicate that global cloud albedo - increasing perturbations similar to the changes induced by cosmic rays decrease the global transport of moisture from the tropics , which then could conceivably produce fewer or smaller global clouds on average by this mechanism .
dynamical simulations of the response of global cloudiness to synoptic changes in the opacity are needed to investigate this .
here we consider a model in which galactic cosmic rays alter the optical properties of clouds by changing the number of available cloud condensation nuclei .
the main observational consequence of our model is a change in mean cloud opacity , with a secondary effect being a change in infrared emittance for optically thin clouds due to the relationship between cloud emissivity and opacity .
we use the global isccp cloud database to search for variations in cloud properties due to cosmic rays , and after subtracting the background signals in the data due to mt .
pinatubo and ensos , we find systematic variations in both opacity and cloud amount associated with changes in the cosmic - ray rate .
the fractional variation in opacity attains a maximum positive value for high clouds and decreases with height , becoming negative or zero for low clouds .
the fractional variation of the cloud amounts with cosmic - ray rate , however , show the opposite trend increasing from a negative correlation at high altitudes to a positive correlation at low altitudes , which is consistent with the positive correlation between global low clouds clouds and cosmic - ray rate seen in the infrared ( @xcite ; marsh and svensmark 2000 ; pall bag and butler 2000 ) clearly more work is needed to model the opacity and cloud amount variations seen in the isccp data . using our simple temporal model and perturbative approach ,
we have outlined a framework on which the variations in the data due to cosmic rays can be isolated and compared to model predictions . as the time span of the isccp data increases in length , more complicated models with additional components and nonlinear dependencies can be used , and the analysis can then be more robust .
the isccp data requires the culling together and normalizing of many disparate satellite datasets ( @xcite ) , and although this approach is necessary at the present time it is not ideal .
one complement to the isccp global cloud data would be provided by the nasa deep space mission _ triana _ , which would be able to retrieve cloud optical thicknesses simultaneously over the entire sunlit earth from the l1 lagrangian point between the earth and the sun .
continuous deep space observing of earth s clouds would be ideal for detecting not only the solar cycle variations seen here but also the shorter duration but possibly more frequent variations in global cloud cover associated with forbush decreases of galactic cosmic rays and high energy solar proton events from the sun .
we thank the avanti article service of the scripps institution of oceanography library , and acknowledge the use of cosmic - ray data from the university of chicago ( national science foundation grant atm-9912341 ) and southern oscillation index data from the australian bureau of meteorology .
we also would like to thank the anonymous referees for very helpful comments .
herschel , w. 1801 : observations tending to investigate the nature of the sun , in order to find the causes or symptoms of its variable emission of light and heat ; with remarks on the use that may possible be drawn from solar observations . _ phil .
royal soc .
, _ * 91 , * 265 - 318 .
jrgensen , t. s. , and a. w. hansen 2000 : comment on `` variation of cosmic - ray flux and global cloud coverage a missing link in solar - climate relationship '' by henrik svensmark and eigil friis - christensen [ journal of atmospheric and solar - terrestrial physics 59 ( 1997 ) 1225 - 1232 ] .
_ j. atmos .
phys . , _ * 62 , * 73 - 77
. lcccccr @xmath75 & @xmath113 & high & @xmath114 & @xmath115 & @xmath116 & @xmath117@xmath75 & @xmath113 & mid - level & @xmath118 & @xmath119 & @xmath120 & @xmath121@xmath75 & @xmath113 & low & @xmath122 & @xmath123 & @xmath124 & @xmath125@xmath75 & @xmath126 & high & @xmath127 & @xmath128 & @xmath129 & @xmath130@xmath75 & @xmath126 & mid - level & @xmath131 & @xmath132 & @xmath133 & @xmath134@xmath75 & @xmath126 & low & @xmath135 & @xmath136 & @xmath137 & @xmath138@xmath86 & @xmath113 & high & @xmath139 & @xmath140 & @xmath141 & @xmath142 @xmath86 & @xmath113 & mid - level & @xmath143 & @xmath144 & @xmath145 & @xmath146@xmath86 & @xmath113 & low & @xmath147 & @xmath148 & @xmath149 & @xmath150@xmath86 & @xmath126 & high & @xmath151 & @xmath152 & @xmath153 & @xmath154@xmath86 & @xmath126 & mid - level & @xmath155 & @xmath156 & @xmath157 & @xmath158@xmath86 & @xmath126 & low & @xmath159 & @xmath160 & @xmath161 & @xmath162 | the observed correlation between global low cloud amount and the flux of high energy cosmic rays supports the idea that ionization plays a crucial role in tropospheric cloud formation .
we explore this idea quantitatively with a simple model linking the concentration of cloud condensation nuclei to the varying ionization rate due to cosmic rays . among the predictions of the model
is a variation in global cloud optical thickness , or opacity , with cosmic - ray rate . using the international satellite cloud climatology project database ( 19831999 )
, we search for variations in the yearly mean visible cloud opacity and visible cloud amount due to cosmic rays . after separating out temporal variations in the data due to the mt .
pinatubo eruption and el nio / southern oscillation , we identify systematic variations in opacity and cloud amount due to cosmic rays .
we find that the fractional amplitude of the opacity variations due to cosmic rays increases with cloud altitude , becoming approximately zero or negative ( inverse correlation ) for low clouds .
conversely , the fractional changes in visible cloud amount due to cosmic rays are only positively - correlated for low clouds and become negative or zero for the higher clouds .
the opacity trends suggest behavior contrary to the current predictions of ion - mediated nucleation ( imn ) models , but more accurate temporal modeling of the isccp data is needed before definitive conclusions can be drawn . |
we gratefully acknowledge the effort of the cesr staff in providing us with excellent luminosity and running conditions .
this work was supported by the national science foundation , the u.s .
department of energy , the research corporation , and the texas advanced research program .
et al . _ , z. phys .
c * 34 * , 103 ( 1987 ) .
p. bedaque _
et al . _ ,
d * 49 * , 269 ( 1994 ) .
chau and h .- y .
cheng , phys .
d * 36 * , 137 ( 1987 ) .
k. terasaki , int .
phys . a * 10 * , 3207 ( 1995 ) .
f. buccella _
et al . _ ,
b * 379 * , 249 ( 1996 ) .
e. golowich and a. petrov , phys.lett .
b * 427 * , 172 - 178 ( 1998 ) . | in @xmath0 collisions using the cleo detector we have studied the decay of the @xmath1 to the final state @xmath2 with the initial flavor of the @xmath1 tagged in charged @xmath3 decay .
we use the dalitz technique to measure the resonant substructure in this final state and clearly observe ten different contributions by fitting for their amplitudes and relative phases .
we observe a @xmath4 component which arises from doubly cabibbo suppressed decays or @xmath5 mixing . 6.5 in 9.0 in -0.50 in 0.00 in 0.00 in h. muramatsu,@xmath6 s. j. richichi,@xmath6 h. severini,@xmath6 p. skubic,@xmath6 s.a .
dytman,@xmath7 j.a .
mueller,@xmath7 s. nam,@xmath7 v. savinov,@xmath7 s. chen,@xmath8 j. w. hinson,@xmath8 j. lee,@xmath8 d. h. miller,@xmath8 v. pavlunin,@xmath8 e. i. shibata,@xmath8 i. p. j. shipsey,@xmath8 d. cronin - hennessy,@xmath9 a.l .
lyon,@xmath9 c. s. park,@xmath9 w. park,@xmath9 e. h. thorndike,@xmath9 t. e. coan,@xmath10 y. s. gao,@xmath10 f. liu,@xmath10 y. maravin,@xmath10 i. narsky,@xmath10 r. stroynowski,@xmath10 m. artuso,@xmath11 c. boulahouache,@xmath11 k. bukin,@xmath11 e. dambasuren,@xmath11 k. khroustalev,@xmath11 r. mountain,@xmath11 r. nandakumar,@xmath11 t. skwarnicki,@xmath11 s. stone,@xmath11 j.c .
wang,@xmath11 a. h. mahmood,@xmath12 s. e. csorna,@xmath13 i. danko,@xmath13 g. bonvicini,@xmath14 d. cinabro,@xmath14 m. dubrovin,@xmath14 s. mcgee,@xmath14 a. bornheim,@xmath15 e. lipeles,@xmath15 s. p. pappas,@xmath15 a. shapiro,@xmath15 w. m. sun,@xmath15 a. j. weinstein,@xmath15 d. m. asner,@xmath16 r. mahapatra,@xmath17 h. n. nelson,@xmath17 r. a. briere,@xmath18 g. p. chen,@xmath18 t. ferguson,@xmath18 g. tatishvili,@xmath18 h. vogel,@xmath18 n. e. adam,@xmath19 j. p. alexander,@xmath19 k. berkelman,@xmath19 f. blanc,@xmath19 v. boisvert,@xmath19 d. g. cassel,@xmath19 p. s. drell,@xmath19 j. e. duboscq,@xmath19 k. m. ecklund,@xmath19 r. ehrlich,@xmath19 l. gibbons,@xmath19 b. gittelman,@xmath19 s. w. gray,@xmath19 d. l. hartill,@xmath19 b. k. heltsley,@xmath19 l. hsu,@xmath19 c. d. jones,@xmath19 j. kandaswamy,@xmath19 d. l. kreinick,@xmath19 a. magerkurth,@xmath19 h. mahlke - krger,@xmath19 t. o. meyer,@xmath19 n. b. mistry,@xmath19 e. nordberg,@xmath19 j. r. patterson,@xmath19 d. peterson,@xmath19 j. pivarski,@xmath19 d. riley,@xmath19 a. j. sadoff,@xmath19 h. schwarthoff,@xmath19 m. r. shepherd,@xmath19 j. g. thayer,@xmath19 d. urner,@xmath19 b. valant - spaight,@xmath19 g. viehhauser,@xmath19 a. warburton,@xmath19 m. weinberger,@xmath19 s. b. athar,@xmath20 p.
avery,@xmath20 l. breva - newell,@xmath20 v. potlia,@xmath20 h. stoeck,@xmath20 j. yelton,@xmath20 g. brandenburg,@xmath21 d. y .- j .
kim,@xmath21 r. wilson,@xmath21 k. benslama,@xmath22 b. i. eisenstein,@xmath22 j. ernst,@xmath22 g. d. gollin,@xmath22 r. m. hans,@xmath22 i. karliner,@xmath22 n. lowrey,@xmath22 m. a. marsh,@xmath22 c. plager,@xmath22 c. sedlack,@xmath22 m. selen,@xmath22 j. j. thaler,@xmath22 j. williams,@xmath22 k. w. edwards,@xmath23 r. ammar,@xmath24 d. besson,@xmath24 x. zhao,@xmath24 s. anderson,@xmath25 v. v. frolov,@xmath25 y. kubota,@xmath25 s. j. lee,@xmath25 s. z. li,@xmath25 r. poling,@xmath25 a. smith,@xmath25 c. j. stepaniak,@xmath25 j. urheim,@xmath25 z. metreveli,@xmath26 k.k .
seth,@xmath26 a. tomaradze,@xmath26 p. zweber,@xmath26 s.
ahmed,@xmath27 m. s. alam,@xmath27 l. jian,@xmath27 m. saleem,@xmath27 f. wappler,@xmath27 e. eckhart,@xmath28 k. k. gan,@xmath28 c. gwon,@xmath28 t. hart,@xmath28 k. honscheid,@xmath28 d. hufnagel,@xmath28 h. kagan,@xmath28 r. kass,@xmath28 t. k. pedlar,@xmath28 j. b. thayer,@xmath28 e. von toerne,@xmath28
t. wilksen,@xmath28 and m. m. zoeller@xmath28 @xmath6university of oklahoma , norman , oklahoma 73019 + @xmath7university of pittsburgh , pittsburgh , pennsylvania 15260 + @xmath8purdue university , west lafayette , indiana 47907 + @xmath9university of rochester , rochester , new york 14627 + @xmath10southern methodist university , dallas , texas 75275 + @xmath11syracuse university , syracuse , new york 13244 + @xmath12university of texas - pan american , edinburg , texas 78539 + @xmath13vanderbilt university , nashville , tennessee 37235 + @xmath14wayne state university , detroit , michigan 48202 + @xmath15california institute of technology , pasadena , california 91125 + @xmath17university of california , santa barbara , california 93106 + @xmath18carnegie mellon university , pittsburgh , pennsylvania 15213 + @xmath19cornell university , ithaca , new york 14853 + @xmath20university of florida , gainesville , florida 32611 + @xmath21harvard university , cambridge , massachusetts 02138 + @xmath22university of illinois , urbana - champaign , illinois 61801 + @xmath23carleton university , ottawa , ontario , canada k1s
5b6 + and the institute of particle physics , canada m5s 1a7 + @xmath24university of kansas , lawrence , kansas 66045 + @xmath25university of minnesota , minneapolis , minnesota 55455 + @xmath26northwestern university , evanston , illinois 60208 + @xmath27state university of new york at albany , albany , new york 12222 + @xmath28ohio state university , columbus , ohio 43210 weak hadronic decays of charmed mesons are expected to proceed dominantly by resonant two - body decays in several theoretical models @xcite .
a clearer understanding of final state interactions in exclusive weak decays is an important ingredient for our ability to model decay rates as well as for our understanding of interesting phenomena such as mixing @xcite . in this context an interesting final state
is @xmath29 which can proceed through a number of two - body states .
previous investigations @xcite of the substructure in this channel were limited by statistics to the cabibbo favored decays . a key motivation is to observe one or more of the @xmath30 resonant submodes that proceed via mixing or double cabibbo suppression , such as @xmath31 or @xmath32 , and to measure their phase relative to the corresponding unsuppressed @xmath33 submodes .
this analysis uses an integrated luminosity of 9.0 fb@xmath34 of @xmath0 collisions at @xmath35gev provided by the cornell electron storage ring ( cesr ) .
the data were taken with the cleo ii.v configuration of the cleo ii multipurpose detector @xcite .
a silicon vertex detector ( svx ) was installed in the upgraded configuration @xcite .
the event selection is similar to that used in our search for @xmath1-@xmath36 mixing via the process @xmath37
@xcite .
we reconstruct candidates for the decay sequence @xmath38 , @xmath39 .
consideration of charge conjugated modes is implied throughout this letter .
the charge of the slow pion ( @xmath40 or @xmath41 ) identifies the charm state at @xmath42 as either @xmath1 or @xmath36 .
we require the @xmath43 momentum @xmath44 to exceed 2.0 gev@xmath45 , and we require the @xmath1 to produce the final state @xmath2 .
we reconstruct @xmath46 with the requirement that the daughter pion tracks form a common vertex , in three dimensions , with a confidence level @xmath47 .
signal candidates pass the vertex requirement with @xmath48 relative efficiency . throughout this letter
, relative efficiency is defined as the number of events passing all requirements relative to the number of events when only the requirement under study is relaxed .
our silicon vertex detector provides precise measurement of the charged tracks in three dimensions @xcite .
we exploit the precision tracking of the svx by refitting the @xmath49 and @xmath50 tracks with a requirement that they form a common vertex in three dimensions .
we use the trajectory of the @xmath51 system and the position of the cesr luminous region to obtain the @xmath1 production point .
we refit the @xmath52 track with a requirement that the trajectory intersect the @xmath1 production point .
we require that the confidence level of each refit exceed @xmath53 .
the signal candidates pass the @xmath1 mass and decay vertex requirement with @xmath54 and @xmath55 relative efficiency , respectively .
we reconstruct the energy released in the @xmath56 decay as @xmath57 , where @xmath58 is the reconstructed mass of the @xmath59 system , @xmath60 is the reconstructed mass of the @xmath61 system , and @xmath62 is the charged pion mass .
the addition of the @xmath1 production point to the @xmath52 trajectory yields the resolution @xmath634 kev , where @xmath64 is the core value from a fit to a bifurcated student s @xmath65 distribution .
we obtain a resolution on @xmath60 of @xmath66mev and a resolution on @xmath67 of @xmath68mev .
we apply a set of ` prophylactic ' requirements to exclude candidates with a poorly determined @xmath69 , @xmath60 , or @xmath1 flight time , and @xmath70 candidates that are likely to be background .
the typical computed @xmath71 for @xmath69 , @xmath60 and flight time error is @xmath72 kev , @xmath73 mev and @xmath74 , respectively .
these are computed from the reconstruction covariance matrix of the daughters of the @xmath1 candidate .
we reject candidates where @xmath64 , @xmath75 or flight time error exceeds 300 kev , 10 mev or 2.0@xmath76 , respectively .
the relative efficiencies for the signal candidates to pass these cuts is 98% , 98% and 89% , respectively .
we exclude @xmath70 candidates with a vertical flight distance less then @xmath77 microns to remove combinatoric background with zero lifetime .
the signal candidates survive this requirement with @xmath78 relative efficiency .
the distributions of @xmath69 and @xmath60 for our data are shown in figure [ fig : qandm ] .
we select 5299 candidates within three standard deviations of the expected @xmath69 , @xmath60 , and @xmath79 .
the efficiency for the selection described above is nearly uniform across the dalitz distribution . in our simulation
we generate the @xmath80 uniformly populating the allowed phase space .
we study our efficiency with a geant @xcite based simulation of @xmath81 events in our detector with a luminosity corresponding to more than three times our data sample .
we observe that our selection introduces distortions due to inefficiencies near the edge of phase space , and fit the efficiency to a two dimensional cubic polynomial in ( @xmath82,@xmath83 ) requiring that the efficiency does not change if the @xmath84 and @xmath85 are interchanged .
our standard result uses this efficiency parameterization to interpret the dalitz distribution . to take into account a systematic uncertainty in our selection efficiency
we compare the standard result with an efficiency that is uniform across the allowed dalitz distribution .
figure [ fig : qandm ] shows that the background is small , but non - negligible . fitting the @xmath60 distribution to a signal shape as described above plus a quadratic background shape
we find a background fraction of @xmath86% .
we use this fraction as a constraint when fitting the dalitz distribution . to model the background contribution in the dalitz distribution we consider those events in the data that are in sidebands five to ten standard deviations from the signal in @xmath69 and @xmath60 and within three in @xmath67 .
there are 445 candidates in this selection , about four times the amount of background we estimate from the signal region .
we compare these with background from our simulation also including @xmath0 annihilations producing the lighter quarks .
we note that the background from the simulation is dominated by random combinations of unrelated tracks , and the shape of the background in the simulation and the sideband data sample agree well .
the simulation predicts that the background uniformly populates the allowed phase space , and we model this contribution to the dalitz distribution by fitting the data sideband sample to a two dimensional cubic polynomial in ( @xmath82,@xmath83 ) .
all parameters except the constant are consistent with zero .
other possible contributions to the background , resonances combined with random tracks and real @xmath1 decays combined with random soft pions of the wrong charge , are negligible in the simulation .
the latter , called mistags , are especially dangerous to our search for `` wrong sign '' @xmath1 decays .
mistags populate the dalitz distribution in a known way that depends on the shape of signal .
when we analyze the dalitz distribution we allow a mistag fraction with an unconstrained contribution .
we have looked for the contribution of an anomalous resonance , such as @xmath87 or @xmath88 , plus random tracks to the background in the data by studying the sidebands in @xmath69 , @xmath60 , and @xmath67 , and conclude that any such contributions are negligible .
figure [ fig : proj ] shows the dalitz distribution for the @xmath80 candidates .
a rich structure is evident .
contributions from @xmath89 and @xmath90 are apparent .
depopulated regions exist suggesting destructive interference between some resonances and the dominant decay modes .
we parameterize the @xmath51 dalitz distribution following the methodology described in ref . @xcite using the same sign convention used in previous investigations of this decay channel @xcite .
we consider nineteen resonant subcomponents , @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , and the `` wrong sign '' @xmath108 and @xmath109 , as well as a non - resonant contribution .
the parameters of the resonances are taken from ref .
@xcite except for the @xmath110 s .
we use ref .
@xcite for the @xmath111 and the coupled channel analysis of ref .
@xcite for the @xmath112 , @xmath113 and @xmath114 .
we consider that each of the resonances has its own amplitude and relative phase .
the non - resonant contribution is modeled as a uniform distribution across the allowed phase space with a fixed relative phase .
the phases and widths of the resonance contributions vary as given by the spin of the resonance as described in ref .
@xcite .
this study is sensitive only to relative phases and amplitudes .
thus we fix one phase and one amplitude . to minimize correlated errors on the phases and amplitudes we choose the largest color suppressed mode , @xmath98 , which should be out of phase with the unsuppressed modes to have a fixed zero phase and an amplitude of one . since the choice of normalization , phase convention , and amplitude formalism may not always be identical for different experiments , fit fractions
are reported in addition to amplitudes to allow for more meaningful comparisons between results .
the fit fraction is defined as the integral of a single component divided by the coherent sum of all components .
the sum of the fit fractions for all components will in general not be unity because of the effect of interference .
backgrounds , combinatorics and mistags , are considered as described above .
they do not interfere with the signal , but the mistag background shape depends on the signal shape as noted earlier .
one must also consider the statistical errors on the fit fractions .
we have chosen to use the full covariance matrix from the fits to determine the errors on fit fractions so that the assigned errors will properly include the correlated components of the errors on the amplitudes and phases .
after each fit , the covariance matrix and final parameter values are used to generate 500 sample parameter sets . for each set ,
the fit fractions are calculated and recorded in histograms .
each histogram is fit with a single gaussian to extract its width , which is used as a measure of the statistical error on the fit fraction .
we perform an initial fit , using the unbinned maximum likelihood technique including the resonances observed by e687 @xcite .
we then consider each of the intermediates states listed above retaining those that are more than three standard deviations significant .
we do not find the @xmath92 , @xmath93 , @xmath97 , @xmath100 , @xmath101 , @xmath106 , @xmath107 , and the``wrong sign '' @xmath109 to be significant .
the @xmath102 is a special case .
it is excluded in our standard fit , and its possible contribution is discussed further below .
the remaining ten resonances , a non - resonant contribution , and backgrounds as described above are included in our standard fit and give our central results .
table [ tab : fit ] gives the results of our standard fit .
figure [ fig : proj ] shows the three projections of the fit . we note that there is a significant `` wrong sign '' @xmath115 amplitude .
mistags are not significant , having a rate of @xmath116% .
when we compare the likelihood of our standard fit to one where the @xmath108 amplitude fixed to zero we see that the statistical significance of the @xmath108 amplitude is 5.5 standard deviations .
also we note that the phase difference between the @xmath91 and @xmath108 contributions is consistent with @xmath117 , as expected from cabibbo factors .
we consider systematic uncertainties that arise from our model of the background , the efficiency , and biases due to experimental resolution .
our general procedure is to change some aspect of our standard fit and interpret the change in the values of the amplitudes and phases as an estimate of the systematic uncertainty .
the background is modeled with a two dimensional cubic polynomial and the covariance matrix of the polynomial coefficents determined from a sideband .
our standard fit fixes the coefficients of the background polynomial , and to estimate the systematic uncertainty on this background shape we perform a fit with the coefficients allowed to float constrained by the covariance matrix .
similarly we perform a fit with a uniform efficiency rather than the efficiency shape determined from the simulation as an estimate of the systematic uncertainty due to the efficiency .
we change selection criteria in the analysis to test whether our simulation properly models the efficiency .
these variations to the standard fit are the largest contribution to our experimental systematic errors . to study the effect of the finite resolution our experiment has on the variables in the dalitz plots we vary the size of the bins used to compute the overall normalization .
another class of systematic uncertainties arise from our choices for the decay model for @xmath80 .
we consider the zemach formalism @xcite , rather than the standard helicity model , which enforces the transversality of intermediate resonances and we vary the radius parameter @xcite for the intermediate resonances and for the @xmath1 between zero and twice their standard value of @xmath118 and @xmath119 , respectively .
these variations to the standard fit are the largest contribution to our modeling systematic errors .
additionally , we allow the masses and widths for the intermediate resonances to vary within their known errors @xcite .
we also consider uncertainty arising from which resonances we choose to include in our fit to the dalitz plot .
we compared the result of our standard fit to a series of fits where each of the possible resonances were included one at a time .
we also considered a fit including all possible resonances .
we take the maximum variation of the amplitudes and phases from the standard result compared to the results in this series of fits as a measure of the uncertainty due to our choice of included resonances .
the @xmath120 has been reported by e791 in @xmath121 decays @xcite .
the parameters of the @xmath120 are sensitive to the choice of decay model , discussed previously . replacing the non - resonant contribution in our standard fit with a @xmath122 component yields an amplitude of @xmath123 and phase of @xmath124 with a mass of @xmath125 and width @xmath126 mev for the @xmath120 consistent with e791 results @xcite ( @xmath127 and @xmath128 mev ) . while we find this suggestive that there is a @xmath102 contribution , we are unable to definitively confirm this because of the known shortcomings of our description of the scalar resonances .
the systematic uncertainty does include the difference between the standard fit which does not include the @xmath120 , and the fits allowing it .
on the other hand , we find no evidence for a scalar @xmath129 as also suggested by e791 @xcite in charm decays .
we also do separate standard fits for @xmath1 and @xmath36 tags to search for @xmath130 violating effects .
we see no statistically significant difference between these two fits .
a more general study would consider a @xmath130 violating amplitude for each component observed in our standard fit . in conclusion
, we have analyzed the resonant substructure of the decay @xmath80 using the dalitz technique .
we observe ten contributions including a `` wrong sign '' @xmath115 amplitude with a significance of 5.5 standard deviations .
this decay arises from a double cabibbo suppressed decay or @xmath5 mixing .
we measure @xmath131 , and the relative phase between the two decays to be @xmath132 , consistent with @xmath117 .
we consider @xmath1 and @xmath36 tags separately , and see no @xmath130 violating effects .
the time dependence of the dalitz distribution of this decay mode in our data is able to discern the source of the `` wrong sign '' signal , doubly cabibbo suppressed decays or @xmath5 mixing , and have sensitivity to the mixing parameters @xmath133 , explicitly to its sign , and @xmath134 at the few percent level @xcite . |
because of their old ages , relative homogeneity of their populations , and high luminosities , globular clusters ( gcs ) can be used to study the chemical evolution history of galaxies .
the most luminous gcs presumably only trace the major star forming events , including mergers .
even basic metallicity provides interesting information for comparison with the galaxy .
however , the detailed chemical composition of gcs could potentially provide much more information on galaxy evolution , because the chemical elements are produced by a variety of stars , with varying sensitivity to stellar mass and metallicity .
high resolution spectroscopic chemical abundance analysis of individual stars in galactic gcs has long been pursued @xcite .
such abundance studies are a useful tool for probing the chemical evolution of the galaxy @xcite .
unfortunately , similar studies have not been possible outside of the milky way , as individual stars can not be resolved in distant galaxies .
[ [ section ] ] for extragalactic gcs , cluster metallicities have been estimated using broadband photometric colors @xcite and low - resolution integrated light ( il ) spectra @xcite .
results from such studies include the discovery of bimodal gc metallicity and color distributions in extragalactic systems @xcite reminiscent of the bimodal gcs in the milky way @xcite .
[ [ section-1 ] ] in recent years , stellar population synthesis ( sps ) of gc populations has provided a new avenue of investigating the chemical composition of either spatially resolved or unresolved gcs , provided high spectral - resolution il spectra can be obtained @xcite . using both broadband photometric colors and equivalent widths ( ews ) of spectroscopic absorption lines as diagnostics , detailed information on the chemical composition of a cluster as a whole can be derived .
[ [ section-2 ] ] non - local thermodynamic equilibrium effects have been demonstrated to be present in the spectra of asymptotic - giant branch ( agb ) stars in the galactic gcs 47 tuc @xcite , affecting the analysis of fe i and ii abundances . while the same effects have yet to be demonstrated in red - giant branch ( rgb ) gc stars , there is a possibility that the effects will be reflected in the il spectra of a cluster if it is present in its brightest members .
our primary goal is to investigate nlte modeling effects on the il spectra of synthetic gcs , as well as ages and metallicities subsequently derived from them , using the population synthesis methodology presented by @xcite .
we test two degrees of nlte treatment , one where we model only the component of the population more evolved than the sub - giant branch in nlte , and the other where we model the entire population in nlte .
we also investigate the uncertainty associated with the color - magnitude diagram ( cmd ) discretization used in this method .
we compare ages for synthetic clusters derived from photometric colors of both lte and nlte spectra , and we assess the significance of deviations caused by nlte effects by comparing them to those caused by photometric uncertainty .
we used phoenix v.15 to compute spherically symmetric model atmospheres and high resolution synthetic spectra ( r @xmath0 300000 ) for stars that cover the cmd parameter spaces of gcs at various ages and metallicities spanning galactic values .
we produced a library of stellar atmospheres covering the ranges of @xmath1 = 3000 to 15000 k ( in steps of 200 k below @xmath1 = 4000 k , 1000 k above @xmath1 = 10000 k , and 250 k otherwise ) and -0.5 to 5.0 dex in log @xmath2 ( in steps of 0.5 dex ) @xcite .
figure [ fig : library ] shows the extent of the coverage in @xmath1 vs log @xmath2 space .
this coverage was reproduced for three values of metallicity , [ m / h ] = @xmath3 , @xmath4 , and @xmath5 , and two values of stellar mass , m = 0.5 and 1.0 @xmath6 .
[ [ section-3 ] ] our library was built in two halves , `` warm '' stars and `` cool '' stars , using a separate pipeline for each . for cool star models ,
we consider 47 different molecules , with a combined total of 119 isotopologues and isotopomers , in both the equation of state ( e.o.s . ) and opacity calculations .
the molecules taken into consideration are listed in table [ tab : molecules list ] .
the atmospheres are also left to naturally form convection zones . for warm star models
, we do not consider molecules in the e.o.s . or opacity calculations .
molecules are fully dissociated in stars earlier than f0 , which @xcite lists as @xmath77300 k ; we choose to err on the side of including molecules unnecessarily and include them in models with @xmath87500 k , to ensure they are present in all models where they are significant .
we also take this as the @xmath1 value above which the possibility of convection is no longer considered in our models , treating the whole extent of the atmosphere as if it were in radiative equilibrium .
the one exception to this division of warm and cool stars is modeling the linear stark broadening of h i lines .
we include the broadening in the spectra of atmospheres with @xmath95000 k ; spectra cooler than this do not have significant stark wings on h i features . [
[ section-4 ] ] all of our atmospheric models have alpha enhanced abundances , with @xmath10 = + 0.4 dex . for our initial solar composition ,
we take our abundances up to o from @xcite , and take the revised abundances of @xcite ( f to ca ) , @xcite ( sc to ni ) , and @xcite ( cu to cs ) .
we assume values for microturbulent velocities of @xmath11 for stars of log @xmath2 = 3.0 and lower , and @xmath12 for stars of log @xmath2 = 3.5 and higher .
two distinct values are used here instead of a more realistic continuous variation to artificially enhance the distinction between spectral lines dominated by evolved and unevolved populations in the il spectrum .
[ [ section-5 ] ] the synthetic spectral output of phoenix is the monochromatic flux spectral energy distribution ( sed ) of a model atmosphere , measured at an optical depth surface of @xmath13 = 0 .
each synthetic spectrum in our library needs to be scaled by a factor of @xmath14 , where the radii are obtained from the corresponding structural models , to convert to the flux spectrum at the @xmath13 = 1 surface .
this provides consistency between models of equivalent @xmath1 that have varying values of log @xmath2 .
we sample our spectra over the wavelength range @xmath15 = 2000 to 27000 at a spectral resolution of r = 300000 , allowing us to compare values of cluster parameters derived from uv @xcite and ir @xcite photometry , to those derived from more traditional optical photometry and spectroscopy . in this work ,
we explore the effects of nlte atmospheric modeling on synthetic il spectra and cluster ages and metallicities derived from them .
the atmospheric structures and synthetic spectra are self - consistently modeled in nlte .
we focus on nlte modeling because fe i line strengths and ews in an il spectrum are the main diagnostic features for deriving cluster parameters @xcite , and fe i is one of the atomic species most heavily affected by nlte in synthetic spectra .
because we study nlte effects on gc parameters derived from il spectra , the accuracy and completeness of our nlte treatment is an important concern .
it was shown by @xcite that using a more complete fe i atomic model will reduce nlte overionization effects by providing more high energy excited states to facilitate recombination from fe ii .
they found that the greater the number of energy levels within @xmath16 of the ground state ionization energy @xmath17 , the more accurate the nlte ionization equilibrium solution .
[ [ section-6 ] ] to this end , we have adopted a set of new and updated nlte model atoms , that are improved with respect to @xcite .
these new atoms generally add to the numbers and refine the exact atomic data values of the energy levels and transitions over the old model atoms .
table [ tab : nlte species list ] shows a comparison of the numbers of levels and lines between the old and new model atoms .
in addition to the species listed in the table , h i , he i , and ne i are also treated in nlte , but are handled internally by phoenix and have not been updated . however , in the cases of he i and ne i , these species do not dominate massive line blanketing , while the old h i treatment is sufficiently complete to model the asymptotic convergence of the balmer series lines for rounding out the balmer jump . for details on the nlte treatment of these species , see @xcite .
[ [ section-7 ] ] these new model atoms provide a significant improvement for some species , such as fe i and ii .
specifically , for fe i , the new model atom has nearly double the number of energy levels , and more than triple the number of b - b transitions than the old .
additionally , the difference between the highest energy level and @xmath18 in the old fe i model atom was @xmath19 , which meant that stars cooler than @xmath20 would not have any energy levels within @xmath21 of @xmath18
. with the new model atom , there are now 45 energy levels closer to @xmath18 , and for the coolest stars in our library at @xmath22 , 17 energy levels are within @xmath21 of @xmath18 .
for this work , we have employed the teramo theoretical isochrones , from the basti group @xcite .
the teramo isochrones are offered with a variety of theoretical assumptions made in calculating the stellar evolutionary tracks , including alpha - enhanced or scaled solar compositions , with or without convective core overshooting , two different mass - loss rates following the reimers law , and normal or extended agbs .
the isochrones cover ranges of -3.27 to 0.51 dex in metallicity and 30 myr to 19 gyr in age .
each one is sampled at 2000 mass points with a variable sampling frequency to ensure that each area of the cmd is critically sampled .
the 30 myr isochrones cover a range of initial masses from 0.5 to 8.5 m@xmath23 , with the upper mass limit being reduced for isochrones with greater ages as stars evolve beyond the modeled tip of the agb .
the mass sampling frequency is increased for these isochrones to maintain the 2000 sampling points .
[ [ section-8 ] ] we select a subset of isochrones to investigate , with alpha enhancement of @xmath24 , mass loss parameter @xmath25 , normal agbs , and without core overshoot , similar to the selection of @xcite .
we focus our investigation on those isochrones with ages and metallicities covering the range of observed galactic gc values , spanning 9 to 15 gyr and [ m / h ] = -1.49 to -0.66 dex .
these values were chosen as the isochrone metallicities closest to the peak values of the metal - rich and metal - poor galactic gc populations @xcite .
we extend our ages beyond the range of average galactic values ( 10 - 13 gyr ) , to investigate the possible size of the effect nlte modeling can have on derived ages .
[ [ section-9 ] ] the isochrones include stars mapping out the transition from the tip of the rgb to the red end of the hb .
we remove these stars before populating our cmd for two reasons : 1 ) this transition is not well understood theoretically ; and 2 ) including these stars would interfere with our cmd discretization procedure , outlined below in section [ boxing ] .
when included in a population , these stars contribute @xmath26 of the total cluster luminosity , and have a negligible impact on the il sed . to expand the isochrones into full populations , we use kroupa s initial mass function ( imf ) @xcite , normalized as a probability density function of the form @xmath27 with @xmath28 , @xmath29 , @xmath30 , and @xmath31 ensuring a continuous function , where @xmath32 , @xmath33 , and @xmath34 @xcite .
@xmath35 is a global normalization constant .
this form gives information about the relative frequencies of stars of various masses as opposed to the number of stars of different masses in a unit spatial volume .
[ [ section-10 ] ] the normalization constant , @xmath35 , is determined for each isochrone individually , relative to their respective mass sampling ranges . to get the relative frequencies of the stars in each isochrone independently , we take a continuous mass range and divide it into bins centered on the isochrone points , with bin divisions halfway between adjacent points .
the imf is then numerically integrated over these bins using the extended trapezoid rule to get the relative frequency for each . to build our synthetic populations , we take a given target luminosity for the population , @xmath36 , representative of the luminosity of a real cluster , and analytically allocate fractions of @xmath36 to the isochrone points according to the relative frequencies .
this determines the total luminosity of each point and , when divided by a point s individual luminosity , the number of stars representing that point in the population .
this can result in non - whole numbers of stars for each isochrone point .
[ [ section-11 ] ] there is no intrinsic spread of cmd features or observational scatter from any source inherent in the synthetic population of our cmds ; they are built as simple stellar populations out of single isochrones .
we choose not to introduce any spread or scatter in the population artificially .
this is unnecessary , as any random variations introduced will be averaged away in the creation of the il spectrum , outlined in section [ boxstars ] .
the integrated light spectrum of a globular cluster is the combined light of every individual star within the cluster .
since it is not feasible to model hundreds of thousands of stars for each cluster that is to be studied , even when interpolating within a library of stellar models , a method by which an il spectrum can be approximated is necessary . we choose to represent groups of parametrically similar stars in a gc cmd by a single stellar spectrum per group and weight their contributions to the il spectrum , following the method of @xcite .
the method involves discretizing a cmd by binning parametrically similar stars areas bounded by lines of constant @xmath1 and @xmath37 .
these areas , or boxes , are established such that an approximately equal fraction of the total cluster luminosity is emitted by the stars contained in each .
@xcite limit the luminosity contained in any box to @xmath38 3 to 4 % of the total luminosity for a given cluster ( @xmath38 25 to 30 boxes total ) ; we choose to increase the total number of boxes to 50 ( @xmath38 2 % of @xmath36 each ) , effectively doubling the discretization resolution in our cmds .
this prevents the boxes from covering too large a range of values of the stellar parameters , which ensures that a single representative stellar spectrum will be an accurate approximation of the integrated spectrum for a box .
[ [ section-12 ] ] starting at the low - mass end of the main sequence , stars of increasing mass are added to a box until the sum of their luminosities matches the allotted percentage of the total cluster luminosity for a box .
the box is then considered full and subsequent stars are added to a new box . the process is repeated for increasing stellar mass until everything up to the tip of the rgb is enclosed in a box .
we repeat the process starting at the low - mass end of the hb and proceed up the agb , until the entire population has been placed into boxes .
the process is explicitly broken and restarted here to avoid too large of a box from encompassing both the rgb tip and hb by blindly adding stars of increasing masses to boxes .
this effectively splits our cmd into two components by ignoring short lived and poorly understood stars in the transition from the rgb to the hb .
the first component is comprised of the main sequence , sub - giant branch , and red giant branch ( the mr component ) , while the second encompasses the asymptotic giant and horizontal branches ( the ha component ) . figure [ fig : samplecmd ] shows an example of a theoretical population similar in age and metallicity to 47 tuc ( age = 13.06 gyr , [ m / h ] = -0.78 ) @xcite , including the discretization boxes . [ [ section-13 ] ] to determine the number of boxes necessary to properly sample the cmd , while maintaining approximately equal luminosity for each box , we calculate the ratio of the luminosity of the mr component ( @xmath39 ) to that of the ha component ( @xmath40 ) .
we then iterate through a total number of boxes , as well as the numbers of boxes allotted to each component , finding the optimal combination to be the one where the ratio of mr boxes ( @xmath41 ) to ha boxes ( @xmath42 ) most closely matches the ratio of luminosities . by limiting the total number of boxes to between 25 and 50
, this combination gives the closest agreement between @xmath43 maintaining approximately equal luminosity in each box , while still limiting each box to 2 to 4 % of the total cluster luminosity .
once the cmd discretization is completed , we proceed to calculate the atmospheric modeling parameters for a representative star in each box . to produce a synthetic spectrum for this box star , we require @xmath1 , log @xmath2 , [ m / h ] , and either @xmath44 or @xmath45 .
metallicity is constant along each isochrone , and prescribes the value for a box star directly .
as we are interested in representing the combined light of every star in a box with a single stellar spectrum , we take the average of the parameters over a whole box , weighted by @xmath46 the fractional number of stars of each isochrone mass sampling bin per box .
[ [ section-14 ] ] the @xmath1 of our box star is found from @xmath47 where the values for individual stars , @xmath48 , are taken directly from the isochrones .
we take the average of the fourth power , rather than a linear average , to include the relative contribution each star makes to the total luminosity of a box .
similarly , the average mass and luminosity can be found from @xmath49 @xmath50 where once again , the individual quantities are taken directly from the isochrones .
[ [ section-15 ] ] there is some ambiguity in calculating the average log @xmath2 for a box .
because log @xmath2 is not a stellar interior modeling parameter , it is not included directly in the isochrones , so a direct average is unavailable in this case .
we choose to take the averages of the relevant isochrone quantities and calculate a single @xmath2 from those average values , @xmath51 without calculating individual @xmath2 values for the isochrone sampling points .
this method has the benefit of being consistent with the reverse process of observers inferring parameters from the il spectrum of a group of spatially unresolved stars . now that the full cmd distribution has been reduced to the representative stars , we generate stellar spectra for each box by interpolating among synthetic spectra in our library .
we linearly interpolate our library spectra weighted by three or four atmospheric modeling parameters ; @xmath1 , @xmath52 , m , and , in cases where the isochrone value is not a direct match to one of the values in our library , [ m / h ] .
this interpolation scheme results in each box spectrum being formed by interpolating among either 8 ( matching library [ m / h ] ) or 16 ( interpolating [ m / h ] ) individual spectra from our library .
we chose to interpolate using a linear method because our library is already pushing the lower boundaries of atmospheric structure convergence in @xmath1 and @xmath52 , and higher order methods would require additional synthetic spectra with even lower values for these parameters .
we test the accuracy of this interpolation by comparing two il spectra generated using this procedure ( one interpolating linear flux spectra and one interpolating log flux spectra ) to one generated from synthetic spectra with the exact parameters of the representative stars .
figure [ fig : interpolation ] shows that there is relatively little difference between our interpolated and exact il spectra , except for the shortest wavelengths that we model , and that there is little appreciable difference between interpolating linear or log fluxes .
representative populations for the youngest and oldest isochrones in this study are plotted in figure [ fig : library ] as a visual indication of where the interpolation between library spectra will occur for @xmath53 and @xmath52 .
[ [ section-16 ] ] in nature , il spectra are combinations of the luminosity spectra of individual stars , not the flux spectra .
because of this , we must convert our box representative spectra from phoenix surface fluxes to luminosities .
the most direct method of doing so takes advantage of @xmath54 where @xmath55 for a box is the average value as calculated above , and we calculate the @xmath56 by numerically integrating low resolution phoenix spectra from 10 to 10,000,000 using the extended trapezoid rule , ensuring sufficient coverage of both the wien side and rayleigh - jeans tail of the spectra . [ [ section-17 ] ] we now combine the box spectra into a synthetic il spectrum .
each spectrum is scaled by @xmath57 to account for the total luminosity of the box , and then added together .
three il spectra per cmd are created this way ; one composed of lte stellar spectra , one of nlte stellar spectra , and one composed of both lte and nlte stellar spectra where only the evolved population were nlte . for our purposes , we consider any star more evolved than the sub - giant branch to be `` evolved '' .
these hybrid il spectra allow us to isolate and study the impact nlte modeling of the evolved population has on cluster parameters inferred from il spectra .
the lte and nlte spectra for our 47 tuc population are overplotted for comparison in figure [ fig : ilspec ] , with the absolute and relative nlte - lte differences , to highlight the most disparate spectral features .
[ [ section-18 ] ] identifying the atomic and molecular species responsible for the large differences between the lte and nlte spectra is not a straightforward task when dealing with il spectra .
because an il spectrum is the co - added light of multiple spectra of different spectral types , what appears to be a single feature in the il spectrum may be caused by multiple sources of opacity .
additionally , at the modest spectral resolution of figure [ fig : ilspec ] , blending of features in crowded regions can confuse the issue even further . to proceed with il feature identification ,
we take five sample stars from our library models ( @xmath53 = 6500 k and @xmath52 = 4.5 , @xmath53 = 5750 k and @xmath52 = 3.5 , @xmath53 = 5000 k and @xmath52 = 3.0 , @xmath53 = 4250 k and @xmath52 = 2.0 , @xmath53 = 3600 k and @xmath52 = 0.5 ) , and identify the sources of any large discrepancies between the sample nlte and lte spectra .
we weight a nlte - lte difference in the spectrum of a sample star by that star s relative contribution to the il luminosity in the photometric band corresponding to the wavelength of that difference .
figure [ fig : bubbles ] displays an example of these relative contributions for the 15.0 gyr population .
if the discrepancies in the sample stars are also found to be present in the il spectrum , the sources in the individual spectra are considered to be responsible .
the species responsible for the majority of large differences were found to be primarily light metals in their ground states , fe i being the most prominent among these , with a few exceptions .
the large deviation observed in the range from @xmath58 = 4000 to 8500 was identified as tio molecular bands from our cool giant stars .
the clusters of lines seen in the range @xmath58 = 10000 to 12000 were ti i , and the strong lines near @xmath58 = 19000 to 20000 were found to be ca i. a full set of high resolution diagnostics will be included in a forthcoming paper in this series .
to take an initial estimate of the magnitude of the effect nlte modeling has on parameters derived from il spectra , we examine four photometric color indices ( @xmath59 , @xmath60 , @xmath61 , and @xmath62 ) produced using bessel s updated johnson - cousins ubvri photometric system @xcite as well as bessel and brett s vjhkllm photometric system @xcite .
we analyze a collection of il spectra for synthetic cmds approximating 47 tuc , with [ m / h ] = -0.66 and substituting @xmath63 = -8.64 for @xmath64 , except ranging in age from 9 to 15 gyr .
two sets of these spectra are generated , each using the lte , nlte , and hybrid prescriptions , limiting the number of boxes to 25 to 35 ( set 1 ) and 40 to 50 ( set 2 ) .
synthetic photometric colors for these il spectra are single - point calibrated to both a nlte synthetic spectrum approximating vega ( @xmath1 = 9600 k , @xmath52 = 4.1 , [ m / h ] = -0.5 ) , and the library spectrum that most closely approximates arcturus ( @xmath1 = 4250 k , @xmath52 = 2.0 , [ m / h ] = -0.66 ) .
arcturus was chosen for this second calibration to compare the il colors to that of a standard star that is representative of the populations .
we denote color values calibrated to each star as @xmath65 and @xmath66 respectively .
figures [ fig : boxes25 ] and [ fig : boxes50 ] present the values of the color indices as a function of population age for sets 1 and 2 respectively , highlighting the nlte - lte differences .
qualitatively , there is little difference between the two sets .
all four color indices display a reddening of il color as the population ages , as more of the population evolves into older , cool stars .
the nlte colors for the first three color indices are bluer than lte at all ages , because nlte overionization of fe i weakens the myriad weak fe i lines that have the character of a pseudo - continuous opacity in the blue and near uv bands .
conversely , the @xmath62 nlte colors are redder than lte , where the surplus of free electrons produced by the overionization increases h@xmath67 opacity in the j band ( @xmath68 1.6 @xmath69 ) . the nlte - lte color differences for the indices in both sets are consistent with a constant value as a function of age , with the exception of @xmath59 , where the difference is seen to increase as a function of age .
[ [ section-19 ] ] the hybrid color values fall midway between the lte and nlte colors for the uv and optical indices , and converge with the nlte values at ir wavelengths .
there are two possible explanations for this convergence : 1 ) evolved stars dominate the il spectra in these filters wavebands ; and 2 ) nlte effects in main sequence stars negligible in these wavebands . to determine which of the two effects is responsible for the convergence
, we refer back to the representative stars relative luminosity contributions in figure [ fig : bubbles ] . as can be seen for @xmath61 , evolved stars do not dominate the il spectrum , with all representative stars making approximately equal contributions to the il spectrum in the @xmath70 band , and the main sequence stars near the turnoff are the strongest contributors in the @xmath71 band . in this case
, it would suggest that negligible nlte effects in main sequence stars at these wavelengths are responsible for the observed convergence of hybrid to nlte colors .
for @xmath62 , the same explanation is likely responsible , but a combination of the two effects is also possible .
for both the @xmath72 and @xmath73 bands , evolved stars on the red giant branch are the strongest contributors to the il spectrum ( although they do not dominate it ) .
[ [ section-20 ] ] to give a quantitative estimate for how much of an impact nlte effects in il spectra have on derived ages , we first define the quantity @xmath74 , the uncertainty in derived age from the uncertainty in measured color index value , such that @xmath75 where @xmath76 is the numerical uncertainty of a given color value , and @xmath77 describes the derived cluster age as a function of `` observed '' color value , found by parameterizing the il color vs age relation with a low order polynomial .
we find that a linear function does not provide a good match to the relation , but a parabola provides an excellent match for all color indices , with coefficients of determinations of at least @xmath78 .
we contrast these @xmath74 values with the difference between the nlte and lte derived ages for a given color , @xmath79 .
the full list of @xmath74 values and the differences in nlte and lte derived ages for all color indices are presented in table [ tab : age errors ] . for comparison , we also present the uncertainty in derived ages found from setting @xmath76 equal to 0.01 mag , the limiting precision of the harris catalog @xcite .
[ [ section-21 ] ] the @xmath59 color index returned the largest differences , but these results were considered to be unrealistic for a number of reasons , including keeping the metallicity fixed throughout this experiment and the difficulty associated with modeling the near - uv region of stellar spectra .
the other three indices , @xmath60 , @xmath61 , and @xmath62 , produced comparable age differences , with those derived from @xmath61 generally being the largest by @xmath80 40 % .
age differences for our @xmath60 index ranged from 0.61 to 1.58 gyr for set 1 , and 0.55 to 2.53 gyr for set 2 .
the smaller age difference for each set corresponds to the bluest color measured for the lte il spectra , and the larger difference corresponds to the reddest measured lte color . for comparison ,
the range of @xmath74 values for @xmath60 are 0.22 to 0.82 gyr and 0.13 to 0.53 gyr , for sets 1 and 2 respectively . assuming an observational limiting precision of 0.01 mag for @xmath60 , the limiting precision of the harris catalog , results in @xmath74 values of 1.12 to 2.41 gyr and 1.05 to 2.84 gyr .
[ [ section-22 ] ] we note that all of our il spectra were produced from isochrones of fixed metallicity , and that age estimates derived in this fashion may vary greatly with changing metallicity . it should also be noted that both lte and nlte model atmospheres have been shown to overpredict near - uv flux in the spectra of cool giant stars @xcite , and this will likely be reflected in the il spectra .
any results found from fitting synthetic il spectra to observed spectra at uv wavelengths would be impacted by this overprediction . either or
both of these effects may help explain the large differences in ages derived from our @xmath59 colors .
additionally , this work is only concerned with a differential analysis between lte and nlte il spectra , and in turn is only affected by any difference in the overpredictions .
nlte models are worse in the overprediction than their lte counterparts as a result of nlte fe i overionization , but the updated fe i nlte model atom we have implemented should minimize this for the range of stellar parameters with which we are concerned .
we also expect that , to first order , the nlte - lte difference in the overprediction of uv flux to be constant as a function of isochrone age .
figure [ fig : uncertainty ] shows the @xmath60 colors for two isochrones , ages 9 and 15 gyr , as a function of the number of boxes used to discretize the cmd .
the value of a color index for a given population varies at the millimagnitude level as a function of the cmd discretization resolution .
as might be expected , the variations are larger for lower discretization resolution , and are reduced for higher resolution .
this trend is qualitatively repeated for the other color indices .
[ [ section-23 ] ] to isolate which regions of the cmd are being over- or under - sampled , we plot in figure [ fig : histograms ] the @xmath53 of the representative stars in three histograms , comparing the cmds on either side of the largest change in color ( ie .
that between 30 and 32 boxes ) , and two control cases to either side of the largest change where there is relatively little change ( 29 to 30 boxes , and 32 to 33 boxes ) .
the bin size for each histogram was set to 250 k , the temperature resolution of our library of spectra , and the bar heights for the four sets of representative stars ( 29 , 30 , 32 , and 33 boxes ) were each weighted by a factor of @xmath81 , the flux in the b band emitted by stars in that bin divided by the flux in the v band , the influence a given bin has on the il @xmath60 value , and independently normalized to sum to 1 .
we use a reduced @xmath82 statistic , listed with each panel in figure [ fig : histograms ] , to confirm that there is a greater difference between the populations in the 30 to 32 boxes histogram than in either the 29 to 30 boxes or the 32 to 33 boxes histograms . inspection of this histogram reveals the most significant differences between the two populations occur for @xmath53 values @xmath83 4750 k. for our populations , this corresponds to the upper main sequence ( including the turnoff ) , the base of the red giant branch , and the horizontal branch .
special care must be paid to these regions when discretizing the cmd to ensure they are not under - sampled .
[ [ section-24 ] ] as a measure of the numerical uncertainty in the il spectrum resulting from cmd discretization , we evaluate the @xmath84 deviations in the computed integrated b - v index as a function of the number of boxes used to discretize isochrones at several ages spanning the age range .
uncertainties for the other color indices are obtained in a similar fashion .
these uncertainties are represented as the error bars in figures [ fig : boxes25 ] and [ fig : boxes50 ] .
we have investigated a number of aspects of the method presented by @xcite , refined from that of @xcite , for synthesizing gc populations and il from a library of stellar atmospheric models and spectra . following this method ,
a collection of 98 il spectra for clusters approximating 47 tuc were generated with different cmd discretization resolutions and different degrees of nlte treatment .
for these clusters , age estimates that may be derived by fitting observed photometric colors with synthetic lte colors were shown to differ from those similarly obtained from nlte modeling by up to 2.54 gyr .
these age differences , while larger than the numerical uncertainties inherent in our methodology , are comparable with the limiting observational precision of current catalogs . our investigation of cmd discretization resolution has shown that the il spectrum is resolution - dependent , and that the effects on the spectrum are stronger at lower resolution . these effects are more prominent at shorter wavelengths .
we also find that the 25 to 35 boxes recommended in the literature do not provide enough resolution to critically sample the upper main sequence and horizontal branch .
at least 40 - 50 boxes are necessary to minimize the resolution dependency .
initial analysis suggests that nlte effects in ms stars have approximately equivalent influence on il spectra as do those in evolved stars for uv and optical wavelengths , but negligible influence for ir wavelengths .
this effect appears to be independent of cmd discretization resolution for those resolutions investigated here .
future papers in this series will be based on an expanded spectral library with a metallicity dimension that spans the range between the peaks of the bimodal distribution of galactic gcs .
we will expand the photometric analysis to include the hubble photometric system for additional uv colors , motivated by the fact that spectral features in the uv are highly sensitive to changes in metallicity .
we will investigate equivalent widths and line profiles for nearly 600 spectral lines as well as lick indices , all identified in the literature as useful features in determining cluster metallicities , for multiple atomic species , including fe i and ii . we also plan on expanding the methodology to include identifying multiple populations within single clusters and deriving their parameters .
we would like to thank the nserc discovery grant program and saint mary s university s faculty of graduate studies and research for funding this work .
we would also like to acknowledge compute canada member acenet for providing us with all computational resources and cpu time .
bellini , a. , renzini , a. , anderson , j. , et al . , 2015 , , 805 , 178 - 197 bessel , m. s. , 1988 , , 100 , 1134 bessel , m. s. , 1990 , , 102 , 1181 briley , m. m. , smith , v. v. , & lambert , d. l. , 1994 , , 424 , l119-l122 brodie , j. p. , & huchra , j. p. , 1990
, , 362 , 503 - 521 cohen , j. g. , 1978 , , 223 , 487 - 508 cohen , r. e. , hempel , m. , mauro , f. , et al . , 2015 , , in press ( arxiv:1509.01470v2 ) coelho , p. r. t. , 2014 , , 440 , 1027 - 1043 colucci , j. e. , bernstein , r. a. , cameron , s. , et al . , 2009 , , 704 , 385 - 414 colucci , j. e. , bernstein , r. a. , cameron , s. , et al .
, 2009 , , 735 , 55 cox , a. n. , 2000 , allen s astrophysical quantities ( 4th ed . ; new york , ny : springer ) elson , r. a. w. , & santiago , b. x. , 1996 , , 278 , 617 forbes , d. a. , & bridges , t. , 2010 , , 404 , 1203 forte , j. c. , strom , s. w. , & strom , k. m. , 1981 , , 245 , l9 geisler , d. , lee , m. g. , & kim , e. , 1996 , , 111 , 1529 grevesse , n. , noels , a. , & sauval , a. j. , 1996 , asp conf .
ser . , 99 , 117 - 126 grevesse , n. , scott , p. , asplund , m. _ et al . _ , 2015 , , 573 , a27 harris , w.e . , 1996 , , 112 , 1487 kroupa , p. , 2001
, , 322 , 231 - 246 lapenna , e. , mucciarelli , a. , lanzoni , b. et al , 2014 , , 797 , 124 maschberger , th . , 2012 , , 429 , 1725 - 1733 mashonkina , l. , gehren , t. , shi , j .-
, et al . , 2011 , , 528 , a87 mcwilliam , a. , & bernstein , r. , 2008 , , 684 , 326 - 347 pietrinferni , a. , cassisi , s. , maurizio , s. _ et al .
_ , 2006 , , 642 , 797 - 812 pilachowski , c. a. , canterna , r. , & wallerstein , g. , 1980 , , 235 , l21 piotto , g. , milone , a. p. , bedin , l. r. , et al . , 2015 , , 149 , 91 - 118 racine , r. , oke , j. b. , & searle , l. 1978 , , 223 , 82 scott , p. , grevesse , n. , asplund , m. _ et al
_ , 2015 , , 573 , a25 scott , p. , asplund , m. , grevesse , n. _ et al .
_ , 2015 , , 573 , a26 short , c. i. , hauschildt , p. h. , 2009 , , 691 , 16341647 sneden , c. , kraft , r. p. , prosser , c. f. , et al . , 1991 , , 102 , 2001 valcheva , a. t. , ovcharov , e. p. , lalova , a. d. , et al . , 2015 , , 446 , 730 - 736 whitmore , b. c. , sparks , w .
b. , lucas , r. , et al .
, 1995 , , 454 , l73 young , m. e. & short , c. i. , 2014 , , 787 , 43 - 55 zinn , r. , 1985 , , 293 , 424 - 444 .
the species responsible for the majority of discrepancies between the spectra are light metals , primarily fe i , with a few exceptions .
the large deviation observed in the range from @xmath58 = 4000 to 8500 is caused by tio molecular bands .
the clusters of lines seen in the range @xmath58 = 10000 to 12000 are those of ti i , and the strong lines near @xmath58 = 19000 to 20000 are those of ca i. all three panels have been convolved from our high resolution spectra to a spectral resolution of r @xmath38 5000 for ease of viewing .
@xmath85 - synthetic il spectra for the population of figure [ fig : samplecmd ] .
@xmath86 - absolute difference between nlte and lte synthetic il spectra .
the nlte spectrum is more luminous in the uv than lte , while showing stronger absorption features in the ir .
@xmath87 - relative difference between nlte and lte synthetic il spectra .
[ fig : ilspec ] ] - photometric colors for synthetic il spectra of populations with constant bolometric luminosity , and [ m / h ] = -0.66 .
all populations are discretized with 25 to 35 boxes .
the error bars are the cmd discretization uncertainty , outlined in section [ uncertainty ] .
solid lines are second order polynomials used for parameterizing the data .
@xmath88 - difference between nlte and lte photometric colors .
[ fig : boxes25 ] ] resolution of our library .
each bin is weighted by a factor of @xmath81 , the flux the bin contributes to the il spectrum in the b band divided by the flux in the v band , representing the influence each bin has on the il @xmath60 value .
the @xmath89 statistics , calculated in each case as @xmath90 , where n and n are the bin values for the sets with the greater and fewer number of boxes respectively , are used as a confirmation that the histograms in the middle panel exhibit greater differences than those in either the top or bottom panels . the middle panel , which compares discretization resolutions on either side of the large jump observed in figure [ fig : uncertainty ] , shows noticeable differences in bins with @xmath53 values @xmath83 4750 k. [ fig : histograms ] ] c@xmath91 & 3 & h@xmath91o & 4 & no & 3 + c@xmath91h@xmath91 & 2 & h@xmath91o@xmath91 & 1 & no@xmath91 & 1 + c@xmath91h@xmath92 & 1 & h@xmath91s & 3 & o@xmath91 & 3 + ch & 2 & h@xmath93 & 1 & o@xmath94 & 3 + ch@xmath94cl & 2 & hbr & 2 & ocs & 4 + ch@xmath95 & 5 & hcn & 3 & oh & 4 + cn & 4 & hcl & 2 & ph@xmath94 & 1 + co & 7 & hf & 1 & sf@xmath92 & 1 + co@xmath91 & 8 & hi & 1 & so@xmath91 & 2 + cof@xmath91 & 1 & hno@xmath94 & 1 & sih & 3 + cah & 2 & hocl & 2 & sio & 4 + clo & 2 & mgh & 3 & tio & 5 + crh & 1 & n@xmath91 & 1 & vo & 1 + feh & 1 & n@xmath91o & 5 & yo & 1 + h@xmath91 & 1 & nh & 2 & zro & 7 + h@xmath91co & 3 & nh@xmath94 & 2 & & + [ tab : molecules list ] li i & 57/333 & 60/394 & 5.392 & 5.296 & 5.296 + li ii & 55/124 & 55/135 & 75.640 & 74.128 & 74.128 + c i & 228/1387 & 230/3262 & 11.260 & 11.155 & 11.155 +
n i & 252/2313 & 254/3704 & 14.534 & 14.460 & 14.460 + o i & 36/66 & 146/855 & 13.618 & 12.728 & 13.482 + na i & 53/142 & 58/334 & 5.139 & 5.044 & 5.044 + na ii & 35/171 & 35/171 & 47.287 & 45.260 & 45.257 + mg
i & 273/835 & 179/1584 & 7.646 & 7.644 & 7.634 + mg ii & 72/340 & 74/513 & 15.035 & 14.585 & 14.585 + al i & 111/250 & 115/482 & 5.986 & 5.977 & 5.977 + al ii & 188/1674 & 191/2608 & 18.828 & 18.665 & 18.665 +
p i & 229/903 & 230/945 & 10.487 & 10.266 & 10.266 +
p ii & 89/760 & 90/882 & 19.726 & 17.542 & 17.542 + s
i & 146/349 & 152/1995 & 10.360 & 10.146 & 10.284 + s ii & 84/444 & 84/501 & 23.334 & 20.375 & 20.375 + k i & 73/210 & 80/576 & 4.341 & 4.300 & 4.300 + k ii & 22/66 & 22/66 & 31.625 & 27.177 & 27.177 + ca i & 194/1029 & 196/2893 & 6.113 & 6.054 & 6.054 + ca ii & 87/455 & 89/760 & 11.872 & 11.641 & 11.641 + ti
i & 395/5279 & 555/13304 & 6.820 & 6.653 & 6.653 + ti ii & 204/2399 & 204/2586 & 13.577 & 10.504 & 10.504 + mn i & 316/3096 & 297/3067 & 7.435 & 7.418 & 7.418 +
mn ii & 546/7767 & 512/8299 & 15.640 & 14.968 & 14.968 + fe i & 494/6903 & 902/24395 & 7.871 & 7.539 & 7.815 + fe ii & 617/13675 & 894/22453 & 16.183 & 14.665 & 14.814 + co i & 316/4428 & 364/6447 & 7.864 & 7.363 & 7.472 +
co ii & 255/2725 & 255/2853 & 17.057 & 15.618 & 15.618 + ni
i & 153/1690 & 180/2671 & 7.635 & 7.422 & 7.422 + ni ii & 429/7445 & 670/17935 & 18.169 & 17.359 & 17.359 + + total & 6134/70492 & 8632/130728 & & & + [ tab : nlte species list ] @xmath96 & 5.07 & 11.50 & 0.24 & 0.27 & 0.83 & 1.26 + @xmath97 & 4.97 & 10.29 & 0.06 & 0.21 & 0.83 & 1.20 + @xmath98 & 0.61 & 1.58 & 0.22 & 0.82 & 1.12 & 2.41 + @xmath99 & 0.55 & 2.53 & 0.13 & 0.53 & 1.05 & 2.84 + @xmath100 & 0.98 & 2.28 & 0.40 & 1.06 & 1.06 & 1.88 + @xmath101 & 0.92 & 2.54 & 0.19 & 0.50 & 0.99 & 2.39 + @xmath102 & 0.79 & 1.19 & 0.56 & 1.18 & 1.65 & 2.66 + @xmath103 & 0.74 & 1.46 & 0.23 & 0.53 & 1.54 & 3.52 + [ tab : age errors ] | we present an investigation of the globular cluster population synthesis method of @xcite , focusing on the impact of nlte modeling effects and cmd discretization .
johnson - cousins - bessel u - b , b - v , v - i , and j - k colors are produced for 96 synthetic integrated light spectra with two different discretization prescriptions and three degrees of nlte treatment .
these color values are used to compare nlte and lte derived population ages .
relative contributions of different spectral types to the integrated light spectra for different wavebands are measured .
integrated light nlte spectra are shown to be more luminous in the uv and optical than lte spectra , but show stronger absorption features in the ir .
the main features showing discrepancies between nlte and lte integrated light spectra may be attributed to light metals , primarily fe i , ca i , and ti i , as well as tio molecular bands .
main sequence stars are shown to have negligible nlte effects at ir wavelengths compared to more evolved stars .
photometric color values are shown to vary at the millimagnitude level as a function of cmd discretization .
finer cmd sampling for the upper main sequence and turnoff , base of the red giant branch , and the horizontal branch minimizes this variation .
differences in ages derived from lte and nlte il spectra are found to range from 0.55 to 2.54 gyr , comparable to the uncertainty in gc ages derived from color indices with observational uncertainties of 0.01 magnitudes , the limiting precision of the harris catalog @xcite . |
the physics of one - dimensional ( 1d ) bose gases in many aspects differs from the physics encountered in higher dimensional systems . for example
, the lieb - liniger ( ll ) gas of @xmath1-interacting bosons in one spatial dimension becomes less ideal as its density decreases @xcite , and eventually approaches the tonks - girardeau ( tg ) limit of a gas of `` impenetrable - core '' bosons @xcite as it becomes sufficiently diluted .
the interest in these 1d systems is greatly stimulated by their experimental realization with atoms confined in tight 1d atomic wave guides @xcite .
the special features of effectively 1d atomic gases @xcite are reflected by properties of nonequilibrium dynamics in these systems , which have become accessible experimentally @xcite .
the possibility of finding exact time - dependent solutions for ll @xcite and tg @xcite evolution is of particular theoretical interest , as they have the potential to provide insight beyond various approximation schemes .
exact solutions for a homogeneous bose gas with ( repulsive ) point - like interactions of arbitrary strength @xmath2 , and periodic boundary conditions , were presented by lieb and liniger in 1963 @xcite . for attractive interactions , @xmath3 ,
exact ll wave functions were analyzed in ref .
the case of box confinement for @xmath4 was studied in ref .
@xcite . in the light of recent experiments
@xcite exact studies of the ll model are even more attractive today @xcite .
besides providing insight into the physics of 1d bose gases , exact solutions can serve as a benchmark for various approximations as well as for numerical approaches ( see , e.g. , refs .
@xcite ) . the calculation of correlation functions of a ll gas from the wave functions is a difficult task ; these functions furnish observables like the momentum distribution of particles in the gas , and were studied by using various approaches ( e.g. , see refs .
time - dependent phenomena in the context of ll gases with finite - strength interactions have been addressed by using both analytical @xcite and numerical methods ( see , e.g. , refs .
irregular dynamics of a ll gas was studied numerically in a mesoscopic system in ref .
@xcite . in ref .
@xcite , it was shown that phase imprinting by light pulses conserves the so - called cusp condition for the ll wave function imposed by the interactions .
exact solutions for 1d bose gases are conveniently constructed by using the fermi - bose mapping techniques @xcite . in 1960
girardeau discovered that the wave function of a spinless noninteracting 1d fermi gas can be symmetrized such that it describes an impenetrable - core 1d bose gas @xcite .
this mapping is valid for arbitrary external potentials @xcite , for time - dependent problems @xcite , and in the context of statistical mechanics @xcite .
in fact , fermion - boson duality in 1d exists for arbitrary interaction strengths @xcite .
furthermore , a time - dependent antisymmetric wave function describing a 1d system of noninteracting fermions can be transformed , by using a differential fermi - bose mapping operator , to an exact time - dependent solution for a ll gas , as outlined by gaudin @xcite .
this method is applicable in the absence of external potentials and other boundary conditions .
therefore , it is particularly useful to study free expansion of ll gases from an initially localized state .
free expansion of interacting bose gases has recently attracted considerable attention .
it has been utilized in experiments to deduce information on the initial state ( see , e.g. , ref .
@xcite and references therein ) , and can be considered as a quantum - quench - type problem which provides insight into the relaxation of quantum systems ( see , e.g. , refs .
@xcite and references therein ) .
free expansion of a ll gas has been analyzed in ref .
@xcite by employing the hydrodynamic formalism @xcite ; it was shown that the density of the gas does not follow self - similar evolution @xcite . however , in 1d bose systems , most exact many - body solutions are given for the tg gas @xcite .
an important result is that the momentum distribution of the freely expanding tg gas asymptotically approaches the momentum distribution of free fermions @xcite .
recently , we have constructed a particular family of exact solutions describing a ll gas freely expanding from a localized initial density distribution @xcite .
it was shown that for any interaction strength , the wave functions asymptotically ( as @xmath5 ) assume tg form .
even though it is generally accepted that 1d bose gases become less ideal with decreasing density , this intuition is mainly based on the studies of a ll gas in equilibrium ground states @xcite .
thus , a more rigorous analysis of the expanding ll gas , which leads to more dilute system , but out of equilibrium , is desirable . in particular , it is interesting to study the dependence of the asymptotic wave functions on the initial state , and to see how are the initial conditions imprinted in the asymptotic states
. here we study the asymptotic form of the wave function describing a freely expanding lieb - liniger gas , which can be constructed via the fermi - bose transformation and the stationary phase approximation . in section [ sec : ll ] we describe the ll model and the fermi - bose transformation . in section [ sec : expasymp ] we demonstrate that the asymptotic wave functions have tonks - girardeau structure , that is , they vanish when any of the two particle coordinates coincide . the dependence of the asymptotic state on the initial state is discussed .
we illustrate that the properties of the asymptotic wave functions can significantly differ from the properties of a tg gas in the ground state of some external potential .
this study generalizes and adds upon our previous result from ref .
@xcite , as the initial conditions studied here encompass ground states for generic external potentials and various interaction strengths . from the next - to - leading order term in the asymptotic regime
, we deduce that the interaction energy of the ll gas decays as a universal power law in time @xmath0 .
this is illustrated on a particular example in section [ sec : check ] , where we provide further analysis of the particular family of time - dependent ll wave functions studied in ref .
explicit expressions for the asymptotic form of the single - particle density are provided in section [ sec : spden ] . in section [ sec : hydro ]
we calculate the asymptotic single - particle density for free expansion of a ll gas from an infinitely deep box potential .
we compare our exact calculation with the hydrodynamic approximation introduced in ref .
@xcite , and employed in ref .
@xcite in the context of free expansion , obtaining good agreement for all values of the interaction strength .
a system of @xmath6 identical @xmath1-interacting bosons in one spatial dimension is described by the many - body schrdinger equation @xcite @xmath7 here , @xmath8 is the time - dependent wave function , and @xmath2 is the strength of the interaction .
it is assumed that the initial wave function is localized , e.g. , by the system being trapped within some external potential , before , at @xmath9 , the trap is suddenly switched off and the gas starts expanding .
we are interested in the behavior of @xmath10 when @xmath11 . here
the spatial dimension is infinite @xmath12 , i.e. , we do not impose any boundary conditions . due to the bose symmetry of the wave function ,
it is sufficient to express it in the fundamental sector of the configuration space , @xmath13 , where @xmath10 obeys @xmath14 the @xmath1-interactions create a cusp in the wave function when two particles touch .
this can be expressed as a boundary condition at the borders of @xmath15 @xcite : @xmath16_{x_{j+1}=x_j}\psi_b=0 . \label{interactions}\ ] ] these boundary conditions can easily be rewritten for any permutation sector . in the tg limit ,
i.e. , for @xmath17 , the cusp condition implies that the wave function vanishes when two particles are in contact : @xmath18 @xcite .
exact solutions of the time - dependent schrdinger equation ( [ llmodel ] ) can be obtained by using a fermi - bose mapping operator @xcite acting on fermionic wave functions : if @xmath19 is an antisymmetric ( fermionic ) wave function , which obeys the schr " odinger equation for a noninteracting fermi gas , @xmath20 then the wave function @xmath21 where @xmath22 , \label{oo}\ ] ] is the differential fermi - bose mapping operator , and @xmath23 is a normalization constant , obeys eq .
( [ llmodel ] ) @xcite . for the purpose of completeness we outline , in appendix [ app : fb ] ,
the proof that the wave function ( [ ansatz ] ) obeys both the cusp condition imposed by the interactions and the schrdinger equation ( [ free ] ) .
in this section we study the asymptotic form of time - dependent ll wave functions @xmath24 which are obtained by the fermi - bose transformation ( [ ansatz ] ) .
all information on the initial condition @xmath25 is contained in the initial fermionic wave function @xmath26 : @xmath27 the initial bosonic wave function , which can be expressed in this way , is assumed to describe a ll gas in its ground state when trapped in some external potential @xmath28 , e.g. , in a harmonic oscillator potential , or some other trapping potential used in experiments .
we consider the evolution from this initial state after the trapping potential has been suddenly turned off , as studied in experiments to deduce information on the initial state @xcite .
the time - dependent fermionic wave function @xmath19 , which freely expands from the initial condition @xmath29 , can be expressed in terms of its fourier transform , @xmath30 } , \label{psift}\end{aligned}\ ] ] where @xmath31 , and @xmath32 by using the fermi - bose transformation , the time - dependent bosonic wave function describing the freely expanding ll gas can be expressed as @xmath33 } , \label{psibt}\end{aligned}\ ] ] where the function @xmath34 is defined as @xmath35 .
\label{gk}\end{aligned}\ ] ] it should be noted that @xmath34 is _ not _ the fourier transform of @xmath24 because it depends on @xmath36 through the @xmath37 terms .
the asymptotic form of the wave function ( [ psibt ] ) can be obtained by evaluating the integral with the stationary phase approximation .
the phase @xmath38 $ ] is stationary when @xmath39 .
let @xmath40 denote the @xmath41-values for which @xmath42 that is , @xmath43 .
the phase can be rewritten as @xmath44 the leading term of the integral in eq .
( [ psibt ] ) , as well as the next - to - leading term , can be evaluated by expanding @xmath45 in a taylor series around the stationary phase point @xmath46 : @xmath47 . \label{psibtseries}\end{aligned}\ ] ] the remaining integrals in the three terms written out in this expansion can be calculated analytically .
the second term involving the first derivatives of @xmath48 vanishes .
the third term is nonvanishing only for @xmath49 . thus eq .
( [ psibtseries ] ) reduces to @xmath50 . \label{psibtseries2}\end{aligned}\ ] ] from eq .
( [ psibtseries2 ] ) we obtain in leading order the asymptotic wave function @xmath51 \nonumber \\ & \quad \times\ \tilde \psi_f(k^{'}_1,\ldots , k^{'}_n ) e^{i \sum_{j=1}^{n } [ k^{'}_j x_j - \omega(k^{'}_j ) t ] } , \label{psiasym1}\end{aligned}\ ] ] which is written in a more convenient form in terms of the variables @xmath52 : @xmath53 \nonumber \\ &
\quad\times\ \tilde \psi_f(\xi_1/2,\ldots,\xi_n/2)\ , e^{\frac{i}{4}\sum_{j=1}^{n } \xi_j^2 t}. \label{psiasym}\end{aligned}\ ] ] equation ( [ psiasym ] ) is the main result of this paper .
evidently the asymptotic form of the ll wave function @xmath54 has tg form .
namely , the fourier transform of a fermionic wave function @xmath55 is antisymmetric , which implies that @xmath54 is zero whenever @xmath56 @xmath57 .
furthermore , @xmath54 is symmetric under the exchange of any two coordinates @xmath58 and @xmath59 .
this clearly shows that a localized ll wave function during free expansion asymptotically approaches a wave function with the tg structure .
however , it should be emphasized that the properties of the asymptotic state are not necessarily similar to the wave function describing tg gas in equilibrium , in the ground state of some external potential .
the connection between the initial and the asymptotic state is illustrated below . in the derivation of eq .
( [ psiasym ] ) we have analyzed ll wave functions which are obtained through the fermi - bose transformation ( [ ansatz ] ) .
this class of wave functions is quite general and corresponds to numerous situations of practical relevance .
let us discuss the case in which the initial bosonic wave function @xmath60 is a ground state of a repulsive ll gas in an experimentally realistic external potential @xmath28 , e.g. , a harmonic oscillator potential .
the eigenstates of the ll system in free space are of the form @xmath61_{m , j=1}^{n } , \label{expansionll}\ ] ] where the set of @xmath6 real values @xmath62 uniquely determines the eigenstate ; the normalization constant is given by @xmath63}},\ ] ] see ref .
@xcite . in free space
, there are no restrictions on the numbers @xmath64 . if periodic boundary conditions are imposed as in ref .
@xcite ( i.e. , the system is a ring of length @xmath65 ) , the wave numbers @xmath41 must obey a set of coupled transcendental equations @xcite which depend on the strength of the interaction ( see , e.g. , ref .
the ll eigenstates @xmath66 possess the closure property @xcite and they are complete @xcite .
thus , our initial state @xmath67 can be expressed as a superposition of ll eigenstates , @xmath68_{m , j=1}^{n } , \label{llcomplete}\end{aligned}\ ] ] where the coefficients @xmath69 can be obtained by projecting the initial condition @xmath67 onto the ll eigenstates . by comparing eqs .
( [ initial ] ) and ( [ llcomplete ] ) we find that the initial fermionic wave function is @xmath70_{m , j=1}^{n}. \label{psif0viab}\ ] ] since we have assumed that @xmath28 is an experimentally realistic smooth function , also @xmath71 is smooth and differentiable such that the operator @xmath72 can be applied .
the connection between the asymptotic state ( [ psiasym ] ) and the initial state @xmath67 is made through the fourier transform of the initial fermionic wave function @xmath73 .
more insight into the connection between the initial state and the asymptotic state can be made by expressing @xmath73 through the coefficients @xmath69 utilized in the expansion ( [ llcomplete ] ) .
first , let us note that the coefficients @xmath74 are antisymmetric with respect to the interchange of any two arguments @xmath75 and @xmath41 @xmath57 .
this follows from the fact that the ll eigenstates @xmath66 possess the same property , see ref .
@xcite . by using this property of @xmath69 , eq .
( [ psif0viab ] ) can be rewritten as @xmath76 by comparing eqs .
( [ connection ] ) and ( [ psift ] ) we obtain @xmath77 evidently , the fourier transform of the initial fermionic wave function @xmath78 is directly proportional to the projections @xmath69 of the initial bosonic wave function onto the ll eigenstates .
from this relation we can conclude that the asymptotic wave function ( [ psiasym ] ) has tg structure as a consequence of the antisymmetry of the coefficients @xmath69 , which originates from the antisymmetry of the ll eigenstates with respect to @xmath41 arguments @xcite .
it is also worthy to note that eq .
( [ psibt ] ) , and therefore our main result , can be obtained without explicit use of the fermi - bose transformation ; by writing the time dependent ll states as @xmath79 , and after employing the antisymmetry of @xmath69 [ equivalently as in eq .
( [ connection ] ) ] one obtains eq .
( [ psibt ] ) .
formulae ( [ psiasym ] ) and ( [ connection1 ] ) provide , under general conditions , the asymptotic form of the wave functions for the freely expanding ll gas , and the connection between these asymptotic states and the initial states . for
the sake of the clarity of the paper , let us illustrate the asymptotic state of the ll gas on a particular example .
suppose that initially the ll gas is in the ground state , enclosed in an infinitely deep box of length @xmath65 .
the ground state @xmath67 for this potential was found by employing the superposition of the bethe ansatz wave functions in ref .
the coefficients @xmath69 can be relatively easily found for a few particles by employing a computer program for algebraic manipulation ( mathematica ) . in fig .
[ figdensity ] we illustrate the initial state and the asymptotic state for the case of @xmath80 particles , and for values of @xmath81 , and @xmath82 , by showing the contour plots of the probabilities @xmath83 ( left column ) and @xmath84 ( right column ) . thus , one particle is fixed in the center of the system , while the plots illustrate the probability of finding the other two particles in space .
the left column illustrating the initial states shows that the system becomes more correlated with increasing interaction strenght @xmath2 and it enters the tg regime for sufficiently large @xmath2 ( e.g. , for @xmath85 depicted in fig .
[ figdensity ] ( g ) the ground state of the system is in the tg regime ) .
the right column illustrating the asymptotic state shows that the wave function is zero whenever two of the coordinates coincide
. however , it is important to note that the properties of the asymptotic wave functions , even though they possess the tg structure , can significantly differ from the properties of the tg gas in the equilibrium ground state .
this can be seen by comparing the asymptotic state in fig .
[ figdensity ] ( b ) , and the tg ground state shown in fig .
[ figdensity ] ( g ) .
the asymptotics of fig .
[ figdensity ] ( b ) is obtained after free expansion from a weakly interacting ground state @xmath86 ; from fig .
[ figdensity ] ( b ) we observe that when one particle is fixed at zero , there is still a relatively large probability of finding the other two particles to the left and to the right of the fixed one . in contrast , for the tg ground state shown in fig .
[ figdensity ] ( g ) , if one particle is fixed in the center of the system , the other two are on the opposite sides of that one . furthermore , by comparing the asymptotic states in figs .
[ figdensity ] ( b ) , ( d ) , ( f ) , and ( h ) , we see that their properties depend on the interaction strength @xmath2 .
it is worthy to mention again that free expansion can be utilized to deduce information on the initial state ( see , e.g. , refs .
@xcite and references therein ) ; free 1d expansion can distinguish between different initial regimes of the ll gas @xcite .
let us now address the case of attractive interactions . for @xmath3 ,
the cusp condition assumes a form that is identical to that for @xmath4 ( see , e.g. , ref .
therefore , by acting on some fermionic time - dependent wave function obeying eq .
( [ schf ] ) with the fermi - bose transformation operator @xmath87 , one obtains an exact solution for the attractive time - dependent ll gas in the form @xmath88 ( see appendix [ app : fb ] ) ; our derivation holds for this family of wave functions .
experiments where the attractive quasi-1d bose gas is suddenly released from a trapping potential were used to study solitons made of attractively interacting bec @xcite .
exact studies of such a system within the framework of the ll model are expected to provide deeper insight into nonequilibrium phenomena beyond the gross - pitaevskii mean - field regime , where interesting dynamical effects can occur @xcite .
it should be noted that the time scale it takes for the ll system to reach the tg regime depends on the initial condition .
the next - to - leading term of the asymptotic wave function is suppressed relative to the leading term by a factor @xmath89 , as obtained by the stationary phase expansion in eq .
( [ psibtseries2 ] ) . from this
we can deduce the scaling of the interaction energy , defined as @xmath90 as @xmath11 . since the interaction strength @xmath2 is finite , and
since the asymptotic density @xmath91 equals zero for any pair of arguments being equal , @xmath56 , one concludes that asymptotically the leading term of the interaction energy vanishes . since the first correction to the leading tg term of the wave function is of order @xmath92 , and since @xmath93 , the interaction energy asymptotically decays to zero as @xmath0 .
this power law decay of the interaction energy is illustrated in the following section .
@xcite , we have constructed a particular family of time - dependent wave functions describing a freely expanding ll gas .
the wave functions were obtained by acting with the fermi - bose mapping operator onto a specific time - dependent fermionic wave function , @xmath94 ^ 2 \big\ } \nonumber \\ & & \times\ b(t)^{-{n^2}/{2 } } \prod_{1\leq i < j \leq n}(x_j - x_i ) , \label{tdf}\end{aligned}\ ] ] which describes free expansion of noninteracting fermions in one spatial dimension .
the initial fermionic wave function at @xmath9 corresponds to a fermionic ground state in a harmonic trap @xmath95 ( see , e.g. , ref .
@xcite ) . here
, @xmath96 corresponds to the trapping frequency , @xmath97 , and @xmath98 .
the limiting form of the ll wave function , @xmath99 , for @xmath11 , was shown to have the following form characteristic for a tg gas : @xmath100 where @xmath101 .
equation ( [ psiasym ] ) is a generalization of this result given first in ref .
@xcite . since eq .
( [ psiasym ] ) was obtained with the help of the stationary phase approximation , whereas ( [ asymho ] ) is obtained straightforwardly from the exact form of the specific ll wave function ( see ref .
@xcite ) , it is worthy to verify that eq .
( [ psiasym ] ) reproduces eq .
( [ asymho ] ) as a special case . in order to do so
, we calculate the fourier transform of the initial fermionic wave function , i.e. , @xmath29 from eq .
( [ tdf ] ) .
interestingly , the fourier transform has exactly the same functional form as the initial condition in @xmath102-space : @xmath103 by plugging this form into eq .
( [ psiasym ] ) we obtain : @xmath104 .
\label{xitdf}\end{aligned}\ ] ] after replacing @xmath52 with @xmath105 which asymptotically approaches @xmath106 , we obtain the functional form identical to eq .
( [ asymho ] ) .
this verifies the validity of eq .
( [ psiasym ] ) in the special case studied in ref .
@xcite .
in order to verify the asymptotic power law decay of the interaction energy @xmath107 obtained in the previous section , let us calculate the time - evolution of @xmath107 for the specific family of ll wave functions discussed in this section .
we calculate integral ( [ inte ] ) for @xmath80 particles , and @xmath108 . given these parameters , @xmath107 depends on the strength of the interaction @xmath2 and time @xmath109 .
figure [ eintt ] illustrates time - evolution of the interaction energy for three values of @xmath2 ; displayed curves depict the ratio @xmath110 , where @xmath111 denotes the total energy , which is a constant of motion .
evidently , after some initial transient period the interaction energy starts its asymptotic power law decay @xmath112 .
it should be noted that the contribution of the interaction energy to the total energy depends on the interaction strength @xmath2 .
this is illustrated in fig .
[ eintc ] which shows @xmath110 as a function of @xmath2 at three points in time . at @xmath9 ,
the contribution of the interaction energy to the total energy is non - monotonous with the increase of @xmath2 ; it is zero at @xmath113 and in the tg limit @xmath114 , with a specific maximal value in between .
the form of the curve is preserved for finite values of @xmath109 , with the evident decay of the interaction energy to zero as @xmath11 .
note that an equivalent non - monotonous behavior of the interaction energy as a function of @xmath2 was found for the lieb - liniger gas in the ground state for @xmath4 and with periodic boundary conditions @xcite .
given the asymptotic form of the wave function , we finally consider the asymptotic form of the single - particle density which is of considerable interest for experiment .
the single - particle density is defined as @xmath115 . for studying asymptotics ,
it is convenient to define the asymptotic form in terms of the rescaled coordinates @xmath116 : @xmath117 here the normalization constant @xmath118 is chosen such that @xmath119 , the total number of particles , while the factor @xmath120 cancels the trivial time - scaling of the asymptotic single - particle density . for the specific asymptotic form of the wave function ( [ xitdf ] ) we can analytically calculate the asymptotic form of the density for a few particles .
as an example , for @xmath80 , the normalization constant is @xmath121 while the single - particle density has the following structure : @xmath122 . \nonumber\end{aligned}\ ] ] this expression shows that the gaussian shape of the single - particle density is modulated with the @xmath6-hump structure characteristic for the single - particle density of a tg gas in the ground state of some external potential . the corresponding density ( [ rhoinfty3 ] ) , in terms of @xmath123
is shown in fig . 2 of ref .
it should be noted that such an asymptotic form of the single - particle density corresponds to a particular family of time - dependent wave functions obtained in ref .
@xcite . for different initial conditions one can obtain a different shape of the asymptotic single - particle density as follows from eqs .
( [ psiasym ] ) and ( [ connection1 ] ) ; the asymptotic single - particle density depends on @xmath73 , that is @xmath69 .
besides providing insight into the physics of interacting time - dependent many - body systems , our motivation to study exact solutions of such systems is to utilize those solutions as a benchmark against various approximations .
free expansion of a lieb - liniger gas has been studied in ref .
@xcite by employing the formalism introduced in ref .
@xcite , referred to as the hydrodynamic approximation .
this formalism can be written in a form of a nonlinear evolution equation for a single - particle wave function @xmath124 [ see eq .
( 9 ) in ref .
@xcite ] , @xmath125 where @xmath126 denotes the single - particle density normalized to @xmath127 , while the function @xmath128 which appears in the nonlinear term is defined in ref .
@xcite , and also tabulated in ref .
[ 19 ] of ref .
the potential is @xmath129 during free expansion .
the hydrodynamic approximation was used to obtain eq .
( [ hydro ] ) , which is written in units corresponding to the lieb - liniger model of eq .
( [ llmodel ] ) .
the nonlinear equation above reduces to the standard gross - pitaevskii equation for small interactions , and to the nonlinear equation from ref .
@xcite for strong interactions @xcite .
the hydrodynamic approximation overestimates the coherence in the system , and therefore it may not be accurate for analyzing observables strongly connected to coherence .
however , it is reasonable to compare the exact asymptotic form of the single - particle density after free expansion with the asymptotic form obtained from the hydrodynamic approximation .
let us follow upon our example from section [ sec : expasymp ] , that is , let us consider the asymptotic form of the single particle density @xmath130 of a ll gas which is initially in the ground state of a box with infinitely high walls ; the length of the box is @xmath131 .
the calculation of the exact sp density demands performing multi - dimensional integration over @xmath132 variables which is not a simple task . for this reason ,
the number of particles in our calculation of the exact sp density is @xmath80 . for the initial condition of the hydrodynamic approach
@xmath133 we could choose @xmath134 within the box , and zero otherwise .
this would be a good initial condition in the thermodynamic limit ( large @xmath6 , @xmath135 ) .
however , since for our exact calculation we used @xmath80 , we have chosen , in order to be able to compare between the two approaches , the hydrodynamic initial field @xmath136 , where @xmath137 is the exact sp density of the initial ground state ( this can be calculated by employing ref .
@xcite ) . figure [ comparison ] displays the exact asymptotic form of the sp density , and the hydrodynamic asymptotic sp density .
the latter is obtained numerically by solving eq .
( [ hydro ] ) with the standard split - step fourier technique ; the nonlinear term in eq .
( [ hydro ] ) , that is , the function @xmath138 is calculated by using values tabulated in ref . [ 19 ] of ref .
the asymptotic dynamics in the hydrodynamic approach occurs after sufficiently long propagation , when the sp density starts exhibiting self - similar propagation ( see also @xcite ) .
the agreement is qualitatively excellent for all values of the interaction strength , and quantitatively excellent for @xmath139 .
the width of the sp density as a function of @xmath116 indicates the velocity of the expansion of the cloud .
the asymptotic fwhm ( full - width at half maximum ) expansion velocity is in good agreement for all values of @xmath2 . the hydrodynamic approximation does not reproduce small humps in the sp density , characteristic in the tg regime after expansion from the ground state ; this discrepancy is expected to be smaller if we had calculated expansion from the ground state with large @xmath6 , where the hydrodynamic approximation is expected to work even better . another possible comparison that can be made with the hydrodynamic approximation is the following .
the ll wave function which is utilized as the initial condition in sec .
[ sec : check ] and ref .
@xcite is obtained by acting with the operator @xmath72 onto the fermionic ground state @xmath71 in the harmonic trapping potential @xmath140 .
this wave function can approximate the ground state only when the commutator @xmath141 $ ] can be neglected @xcite .
the sp density of this state can be compared with the static hydrodynamic density obtained in ref .
@xcite for the ll gas in a harmonic trap . due to the properties of the operator @xmath72 @xcite and the fermionic ground state in the harmonic trap @xmath71 , it is straightforward to verify that the shape of the sp density corresponding to the state @xmath142 scales as @xmath143 under the transformation @xmath144 , @xmath145 , that is , the shape of the sp density does not change under this transformation .
the same is true for the shape of the ( ground - state ) sp density obtained with the hydrodynamic approach , which has been shown @xcite to depend on a single parameter @xmath146 that is invariant under the transformation @xmath144 , @xmath145 .
this is fully analogous to the case of a homogeneous ll gas where the only governing parameter @xmath147 is invariant under a simultaneous rescaling of the interaction strength @xmath2 and the linear particle density @xmath148 @xcite .
the shape of the sp density of the state @xmath142 ( calculated for @xmath80 ) agrees with the shape obtained in ref .
@xcite only in the tonks - girardeau limit ( @xmath149 ) where @xmath142 is a good approximation for the ground state .
if we reduce the interaction strength @xmath2 by keeping @xmath96 fixed , thereby increasing @xmath150 , the two sp densities will no longer have a similar shape ; this stems from a simple fact that @xmath142 is an excited state for sufficiently small values of @xmath2 , because the commutator @xmath141 $ ] can not be neglected , whereas the hydrodynamic solution approximates the ground state .
we have derived the asymptotic form of the wave function describing a freely expanding lieb - liniger gas .
it is shown to have tonks - girardeau structure [ see eq .
( [ psiasym ] ) ] , that is , the wave functions vanish when any two of the particle coordinates coincide .
we have pointed out that the properties of these asymptotic states can significantly differ from the properties of a tg gas in a ground state of an external potential [ see fig .
[ figdensity ] ] .
the dependence of the asymptotic state on the initial state is discussed [ see eq .
( [ connection1 ] ) ] .
the analysis was performed for time - dependent lieb - liniger wave functions which can be obtained through the fermi - bose transformation ( [ ansatz ] ) .
this encompasses initial conditions which correspond to the ground state of a repulsive lieb - liniger gas in physically realistic external potentials .
thus , our analysis characterizes the free expansion from such a ground state , after the potential is suddenly switched off . in deriving our main result , eq .
( [ psiasym ] ) , we have used the stationary phase approximation .
this generalizes and adds upon the result from ref .
@xcite which was derived for a particular family of time - dependent lieb - liniger wave functions .
we have demonstrated that the interaction energy of the freely expanding ll gas asymptotically decays according to a power law , @xmath0 .
furthermore , we have calculated the asymptotic single - particle density for free expansion of a ll gas from an infinitely deep box potential .
we have compared our exact calculation with the hydrodynamic approximation introduced in ref .
@xcite , and employed in ref .
@xcite in the context of free expansion , obtaining good agreement for all values of the interaction strength . as a possible future avenue of research ,
we point out that the methodology employed here for the analysis of asymptotic wave functions has the potential to be exploited further to study the evolution of various observables ( e.g. , the momentum distribution which was studied for a tg gas @xcite ) and correlations ( e.g. , see @xcite and refs . therein ) during free expansion .
we are grateful to m. fleischhauer and v. dunjko for very useful comments and suggestions .
h.b . and r.p
acknowledge support by the croatian ministry of science ( mzo ) ( grant no .
119 - 0000000 - 1015 ) .
t.g . acknowledges support by the deutsche forschungsgemeinschaft .
this work is also supported by the croatian - german scientific collaboration funded by daad and mzo , and in part by the national science foundation under grant no .
phy05 - 51164 .
in this appendix we outline the proof that the wave function ( [ ansatz ] ) obeys both the cusp condition imposed by the interactions and eq .
( [ free ] ) , i.e. , that it obeys eq .
( [ llmodel ] ) . without loss of generality
we restrict our discussion to the fundamental permutation sector @xmath15 .
let us write the differential operator as @xmath151 , where @xmath152 \label{bij}.\ ] ] we first show that the wave function ( [ ansatz ] ) obeys the cusp condition ( [ interactions ] ) ( see ref .
@xcite ) . consider an auxiliary wave function @xmath153 where the primed operator @xmath154 omits the factor @xmath155 as compared to @xmath72 .
the auxiliary function can be written as @xmath156 \hat o'_{j , j+1 } \psi_f .
\label{auxasym}\ ] ] it is straightforward to verify that the operator @xmath157 in front of @xmath158 is invariant under the exchange of @xmath36 and @xmath159 . on the other hand ,
the fermionic wave function @xmath158 is fully antisymmetric with respect to the interchange of @xmath36 and @xmath159 .
thus , @xmath160 is antisymmetric with respect to the interchange of @xmath36 and @xmath159 , which leads to @xmath161 this is fully equivalent to the cusp condition ( [ interactions ] ) , @xmath162 .
thus , the wave function ( [ ansatz ] ) obeys constraint ( [ interactions ] ) by construction .
if we use the expression @xmath165 , \label{bijsgn}\ ] ] we obtain @xmath151 as in eq .
( [ oo ] ) , which is valid inside any sector of the configuration space ( see @xcite ) .
note that for @xmath114 , one recovers girardeau s fermi - bose mapping @xcite , where the operator @xmath166 maps a noninteracting fermionic to a bosonic tonks - girardeau wave function .
f. schreck , l. khaykovich , k.l .
corwin , g. ferrari , t. bourdel , j. cubizolles , and c. salomon , phys .
lett . * 87 * , 080403 ( 2001 ) ; a. g " orlitz , j.m .
vogels , a.e .
leanhardt , c. raman , t.l .
gustavson , j.r .
abo - shaeer , a.p .
chikkatur , s. gupta , s. inouye , t. rosenband , and w. ketterle , _ ibid . _ * 87 * , 130402 ( 2001 ) ; m. greiner , i. bloch , o. mandel , t.w .
hansch , and t. esslinger , _ ibid . _ * 87 * , 160405 ( 2001 ) ; h. moritz , t. st " oferle , m. kohl , and t. esslinger , _ ibid . _ * 91 * , 250402 ( 2003 ) ; b. laburthe - tolra , k.m .
ohara , j.h .
huckans , w.d .
phillips , s.l .
rolston , and j.v .
porto , _ ibid .
_ * 92 * , 190401 ( 2004 ) ; t. st " oferle , h. moritz , c. schori , m. kohl , and t. esslinger , _ ibid . _ * 92 * , 130403 ( 2004 ) .
t. kinoshita , t. wenger , and d.s .
weiss , science * 305 * , 1125 ( 2004 ) ; b. paredes , a. widera , v. murg , o. mandel , s. f " olling , i. cirac , g. v. shlyapnikov , t. w. h " ansch , and i. bloch , nature ( london ) * 429 * , 277 ( 2004 ) . | the asymptotic form of the wave functions describing a freely expanding lieb - liniger gas is derived by using a fermi - bose transformation for time - dependent states , and the stationary phase approximation .
we find that asymptotically the wave functions approach the tonks - girardeau ( tg ) structure as they vanish when any two of the particle coordinates coincide .
we point out that the properties of these asymptotic states can significantly differ from the properties of a tg gas in a ground state of an external potential .
the dependence of the asymptotic wave function on the initial state is discussed .
the analysis encompasses a large class of initial conditions , including the ground states of a lieb - liniger gas in physically realistic external potentials .
it is also demonstrated that the interaction energy asymptotically decays as a universal power law with time , @xmath0 .
hd
thep0810 + |
ever since the original deciphering of clathrate hydrate structures by claussen , pauling , and others , @xcite researchers have been fascinated by the interactions of hydrophobic gases with water producing these beautiful crystal structures .
natural gas hydrates are a potential source of energy @xcite and problem in fossil fuel transport ; @xcite clathrates could play a role in global climate change @xcite and be used in co~2~ sequestration . @xcite
clathrate hydrates were famously proposed as a molecular explanation of general anaesthesia .
@xcite xenon , a noble gas that readily forms a clathrate hydrate , manifests distinct physiological actions despite its chemical inertness .
owing to its strong anesthetic effect , @xcite xenon is considered to be the best ( though expensive ) inhalation anesthetic to date and is officially approved for application as such in germany and russia .
xenon has an antidepressant effect ( presumably due to its anti - nmda @xcite action ) and is used to treat alcoholism and drug addiction .
@xcite xenon shows a protective effect on various organs subject to ischaemia
reperfusion injury @xcite ( presumably due to its anti - nmda effect and interference with many aspects of intracellular ca^2+^ homeostasis @xcite ) . an important potential application of xenon is its cryoprotective action .
@xcite in order to understand this phenomenon , the physicochemical basis of interaction of xenon with liquid water , ice and biological molecules must be understood at the microscopic level .
the mechanisms that xenons antifreeze effect could be ascribed to include : 1 .
trapping of water in clathrate hydrate in competition with ice crystals formation ( akin to traditional cryoprotectants such as glycerol and dmso ) .
vitrification @xcite of xenon solution .
ice - blocker action via kinetic suppression of ice growth ( similar to biological antifreeze proteins @xcite ) by clathrate hydrate forming on ice crystal surface .
@xcite 4 .
accumulation of xenon in lipid membranes .
@xcite 5 .
protection against protein thermal denaturation .
@xcite in the present molecular dynamics study we focus on the `` nonbiochemical '' mechanisms ( _ i__iii _ ) while the interaction of xenon with biological membranes and proteins is left for follow - up studies .
first we investigate the structure and properties of homogeneous xenon solution and its behavior under cooling below the freezing temperature of water .
we observe the formation of clathrate phase , with a complex dependence on the concentration and temperature of solution , and a signature of another metastable amorphous phase upon deep supercooling .
then we study the process of ice growth in the presence of dissolved xenon , observing preferential adsorption of xenon on the ice surface but no preferential nucleation that could have caused an ice - blocking effect .
for xenon , we used lennard jones interaction parameters @xcite that were previously successfully used to study hydrophobic hydration of xenon by paschek : @xcite @xmath1 , @xmath2 . to simulate water
, we used the tip5p five - site rigid model .
@xcite cross - species interaction parameters were derived using the geometrical mean combination rule .
tip5p was chosen to represent water because it has pronounced directional hydrogen bonding thanks to its lone - pair virtual sites , leading to stronger tetrahedral structuring of water , which is important here because of our interest in the processes of hydrophobic hydration that is crucially dependent on the local ordering of water molecules @xcite ( especially at low temperatures @xcite ) , and crystal formation .
the parametrization of the tip5p model was done over a wide range of temperatures and pressures , @xcite and it correctly reproduces both the anomalous density maximum of water at 4 c as well as the melting temperature of ice ih ( @xmath3 = 275 k as observed in the present work and in earlier studies @xcite ) . although the thermodynamically most stable ice polymorph is erroneously predicted by tip5p to be ice ii , @xcite and the proton - ordered ice ih structure has lower predicted free energy than the ( correct ) disordered ice i , @xcite we still believe tip5p to be the most reliable choice of water model for our purposes .
molecular dynamics simulations were carried out using the gromacs package .
@xcite the simulation time step was 2 fs .
coulomb and lennard - jones interactions were truncated at a distance of 0.9 nm ( as in the original tip5p parametrization ) .
long - range coulomb interactions were calculated using the particle - mesh ewald method , @xcite and a standard long - range dispersion correction term was applied .
@xcite for liquid systems , preequilibration was done using the berendsen thermostat , @xcite and production runs were performed using the nos hoover thermostat @xcite with a 0.5 ps time constant and the nos parrinello rahman barostat @xcite with a 2.0 ps time constant . for two - phase systems , production runs used the velocity rescaling thermostat @xcite and the berendsen barostat @xcite with a 0.5 ps time constant for both , to accommodate the total energy and volume change during phase transition .
l * 6r @xmath4 ( k ) [ table ] & 280 & 270 & 265 & 260 & 255 & 250 + 1% & 80 & 80 & 50 & 100 & 100 & > 100 + 2% & 50 & 50 & 50 & 100 & 100 & > 100 + 3% & 30 & 30 & 50 & 100 & 100 & > 100 + 4% & 30 & 30 & 50 & 100 & 100 & > 100 + 5% & 30 & 50 & 100 & 100 & 100 & > 100 + liquid systems were prepared by randomly inserting xenon molecules into a preequilibrated 500-molecule water box to produce a desired xenon concentration of 1 , 2 , 3 , 4 , or 5% mol.xe/mol.h~2~o ( meaning 100 , 80 , 60 , 40 , and 20 water molecules per xenon , respectively ) .
the 5% concentration was chosen as the upper boundary since it still , in principle , allows each xenon molecule to have its separate hydration shell ( which contains about 20 molecules ) ; at higher concentrations , there would necessarily be insufficient water to solvate each xenon molecule , casting doubt on our treatment of the system as a liquid solution .
the systems were annealed at fixed cell dimensions for a short period ( 200 ps ) starting at a temperature of 600 k and cooling down to the target temperature of 250280 k. after that , they were replicated to produce @xmath5 supercells ( 4000 water molecules , 40200 xenon ) and incubated at the target temperature until equilibrium was reached .
the corresponding run lengths ranged from 30 to 100 + ns ( see table [ table ] ) .
such large systems and long simulation times were necessary to collect good statistics for xenon molecules , and because of the long equilibration times at low temperatures due to very slow diffusion . at the lowest temperatures ,
simulation times even this long sometimes would nt produce full convergence of radial distribution function ( rdf ) and mean - square displacement ( msd ) plots ; at the lowest concentration of 1% , though the spatial configurations of xenon atoms were written out as frequently as once every 100 fs ( more than sufficient , given the typical timescale of xenon motion of @xmath61 ps , as observed below ) , rdf plots still contain noticeable statistical noise .
two - phase solid - liquid equilibrium simulations were performed with a @xmath7 ice ih slab , since the prism plane of ice grows faster than basal and hence is the most relevant from the viewpoint of both pure physics @xcite and biological applications .
@xcite the size of the slab was @xmath8 nm , and the thickness was 3.07 nm , corresponding to four nonprimitive crystal cells , or eight water molecule bilayers ( see figures below ) ; the total number of water molecules in the slab was 3072 .
initial coordinates for ice molecules were taken from the work of buch _
@xcite the liquid layers were taken approximately twice as thick as the slab ( 6.1 nm ) to accommodate enough xenon molecules and to allow us to study longer - range effects of ice on the structure of the solution .
liquid and solid phases preequilibrated at 275 k were brought in contact and relaxed using steepest descent energy minimization with a maximum force criterion of 100 ( kj / mol)/nm , which was found to be sufficient to remove any bad molecular contacts at the interface while still ensuring that atomic coordinates still correspond well to the 270275 k temperature of subsequent production runs , thus providing smooth continuation of molecular dynamics .
the results reported herein were mostly produced at a simulated pressure of 1 atm irrespective of xenon concentration in the systems . while xenon clathrate hydrate is stable under these conditions or close ( its experimental dissociation pressure value is 2.5 bar at 5 c , 1.5 bar at 0 c , and 1 bar at -3.6 c@xcite ) , these concentrations are considerably above equilibrium at a pressure of 1 atm , which may create an excessive driving force for clathrate formation .
extrapolation of empirical xenon solubility data@xcite to our temperature range suggests that our concentrations should correspond to `` equilibrium '' ( barring formation of solid phases ) pressures on the order of tens to few hundred atmospheres , which is , in fact , a much milder driving force than used in most if not all recent md studies on clathrate formation .
we never observed separation of xenon gas phase in our simulations ; in contrast , such separation was observed to occur spontaneously in molecular dynamics simulations of methane hydrate at 100 mpa ; @xcite note that the dissociation pressure of methane hydrate is only @xmath65 mpa , and the model system was smaller than ours ( 1472 water molecules ) , although the gas concentration was higher ( 256 molecules , or 17% mol.me/mol.h~2~o ) .
further , our tests on two - phase systems at 10 atm did not produce any noticeable change in the equilibrium characteristics that would suggest the need for computationally expensive pressure dependence studies , and the decomposition temperature of xenon clathrate hydrate did not depend on the concentration ( see discussion below ) . in summary
, this leads us to believe that this choice of simulation conditions does not significantly affect the quality of our results ( and , in fact , allows us to draw additional important conclusions ) .
physically , this corresponds to the approximation that the pressure dependence of the balance between gas - phase and dissolved xenon molecules , which determines the concentration of xenon in solution , is primarily governed by the change in the chemical potential in the gas phase , while the structure and thermodynamics of the dissolved phase are , to a reasonable degree , unaffected by pressure on the relevant magnitude scale .
to study the effects of xenon on the freezing of water , we performed simulations of five model systems with different xenon concentration at the temperatures of 280 , 270 , 265 , 260 , 255 , and 250 k ( production run durations are listed in table [ table ] ) .
the initial motivation had been to study the dependence of xenon - xenon interactions , as manifested in radial distribution function ( rdf ) plots , and xenon diffusion rate determined from the slope of plots of mean square displacement ( msd ) over time , on the temperature and concentration of xenon in water solution below the freezing point . ] * figure [ fig:280k ] * shows the plots of xe xe rdf and xe msd time dependence for the concentration range studied in the present work at a temperature of 280 k. it can be seen that the structure of all five systems is identical to a good precision ( the corresponding xe o rdf are also indistinguishable and hence not shown ) . by inspecting the structures , we observe that the first three peaks in the rdf correspond to the following three cases of relative position of xenon molecules : 1 . _ contact solute pair _ ( csp ) : direct contact of xenon molecules ( hydrophobic association in a common hydration shell ; this peak was previously shown to be enhanced with increasing temperature @xcite)the maximum of the peak roughly corresponds to twice the radius of xenon atom ( @xmath60.4 nm ) ; 2 . _
solvent - separated solute pair _ ( sssp ) : face - sharing of water cages containing two xenon molecules ( water molecules `` sandwiched '' between xenon molecules)the maximum corresponds to the clathrate hydrate structure ( see below ) ; 3 .
contact of water cages : hydration shells of the two xenon molecules have no common molecules ; a hydrogen bond connects a vertex of one with a vertex of the other like , for example , the two 5 ^ 12^ cages in the clathrate unit cell .
geometrically , this corresponds to @xmath9 [ 1.1 nm from xe - xe rdf ] @xmath10 [ 0.76 nm from xe o rdf ] + @xmath11 [ 0.27 nm from o o rdf in pristine water ] .
we propose the term _ contact cluster pair _
( ccp ) to describe this configuration .
however , the dynamics of the systems shows a more complex two - timescale behavior .
the first part of the msd time dependence is completely identical for all systems , while the second part demonstrates a slow - down of xenon diffusion with increasing concentration .
this indicates that there two modes of xenon atom motion in the solution : 1 .
fast confined oscillatory motion of xenon inside the hydration cage with a characteristic time of about 1 ps and a squared amplitude of about ( 0.112 nm)^2^ , showing no visible dependence on the structure of the medium surrounding the hydration shell , and 2 .
slow drift of the cluster ( xenon molecule plus its hydration shell ) as a whole in the surrounding solution with occasional hopping of xenon from shell to shell ( see movie s1@xcite ) , on the timescale of over 1 ps .
the dependence of the second type of motion on the concentration can be attributed to their collisions decreasing the effective `` mean free path '' , or by formation of more massive cluster aggregates ( potential clathrate hydrate nuclei ) . for a representative sample of 10 atoms .
the background shows the full @xmath12 distribution for all atoms .
the histograms on the left and right correspond to @xmath13 ns and @xmath14 ns , respectively .
( * d * ) a close - up image of two crystallites around @xmath15 ns .
the green atoms have an instantaneous @xmath12 below the dashed horizontal line in ( * c * ) .
see * * movie s2**@xcite for an animation of the entire run .
[ fig:270k ] ] at temperatures of 270 k and below , in systems with highest xenon concentrations , a phase transition was observed , leading to a structure radically different from liquid solution .
* figure [ fig:270k ] * shows the rdf ( * a * ) and msd ( * b * ) plots for the 5% mol.xe/mol.h~2~o system at 270 k over five sequential 10-ns simulation intervals .
it can be seen that during the initial 20 ns , the system was in a liquid state same as above : the rdf peaks have the same positions ; the suppression of the leftmost csp peak compared to 280 k is a manifestation of the hydrophobic dissociation with decreasing temperature ; it results from the negative entropy of hydration , @xcite which makes it favorable for water molecules at low temperatures to become a part of a hydration shell , forcing the neighboring xenon molecules apart to increase the `` surface area '' of hydration .
the dynamics during the first 20 ns also corresponds to that seen in * figure [ fig:280k ] * , with fast motion on a 1 ps time scale with a ( 0.122 nm)^2^ squared amplitude and a slower mode over longer time scales , but with a slope smaller than at 280 k , obviously , due to the lower temperature .
however , after the initial 20 ns , the system underwent an abrupt phase transition , manifesting itself both in the structure and dynamics of the system .
in the rdf plot ( * figure [ fig:280k](a ) * ) , it can be seen that the height of the first peak dropped sharply ( below 1 , indicating effective repulsion of xenon molecules ) , while the second peak became much more intense and shifted slightly to the left ; the third peak moved to the right , and some previously unseen ordering appeared at larger distances . to analyze the changes in the system we calculated the diffusion coefficient @xmath12 every 20 ps for each atom using 200-ps msd fits .
we then computed the running average over 100 consecutive values to smoothen the fluctuations .
the results are plotted in * figure [ fig:270k](c)*. the density and contour maps in the background represent the distribution of atomic mobilities over time .
the colored solid lines show the @xmath16 trajectories of a representative random sample of 10 atoms .
it can be seen that the system starts in a relatively homogeneous state with most atoms having @xmath17 cm^2^/s but after about 15 ns a number of them make a transition to a state with an order - of - magnitude lower mobility , @xmath18 cm^2^/s .
the left and right histograms in * figure [ fig:270k](a ) * further illustrate the qualitative changes in atomic mobility distributions before this event and at the end of our md trajectory , with the striking bimodal distribution seen in the right chart .
to better understand the nature of the phase transition we visualized the trajectory coloring the atoms according to their instantaneous diffusion coefficient . in *
* movie s2**@xcite , atoms having a diffusion coefficient lower than an arbitrary cutoff shown by the dashed line in * figure [ fig:270k](c ) * are highlighted in green . in the first @xmath19 ns of the trajectory ,
two events of short - lived cluster formation are observed .
these are seen in * figure [ fig:270k](c ) * as two small `` chunks '' below the dashed line on the left .
this suggests that our conditions are very close to reversibility .
then , after 15 ns , a supercritical cluster forms and continues to grow , followed shortly by a second one , until the end of the run .
the final structure of the two crystallites , shown close - up in * figure [ fig:270k](d ) * , visibly corresponds to the cubic si clathrate lattice of xenon hydrate ( * * movie s2**@xcite ) , and agrees well with the experimental lattice constant value of 1.182 nm .
even though a strict assignment of crystal structure is ill - posed for such a small cluster , and clathrate nucleation may involve `` amorphous '' clathrate - like precursors,@xcite the rdf for ideal si lattice ( * figure [ fig:270k](a ) * , dashed line , with gaussian broadening applied to match the peak height ) clearly shows that the trends are consistent with clathrate formation . in this context ,
the `` migration '' of xenon molecules from the first rdf peak ( csp ) to the second ( sssp ) is a result of all guest molecules in the clathrate hydrate sharing cage walls with their neighbors ( no immediate xenon
xenon contacts ) .
the shift of the third ( ccp ) peak to the right may be due to the fact that the t - cage in si clathrate hydrate ( tetrakaidecahedron6 ^ 2 ^ 5 ^ 12^ ) is actually somewhat larger than the dodecahedral cage , which is the preferred shape of hydration shell in the liquid state ( as indicated by , e.g. , nmr studies of xenon clathrate hydrate growth ; @xcite integration of the first peak of xe o rdf [ not shown ] in the liquid systems also produces an average number of molecules in the hydration shell of 21.5 , consistent with the predominance of dodecahedral cages and in contrast to the 23.0 value of ideal clathrate hydrate that has 8 xenon molecules per 2 5 ^ 12^ and 6 6 ^ 2 ^ 5 ^ 12^ cages in the unit cell ) .
it should be noted that , although only a minor fraction of xenon molecules get converted to the clathrate phase , the stability of crystalline structure with respect to thermal motion of atoms results in the fact that this structure is very markedly seen in the rdf . at the same time , no detectable changes are observed in the water o
o nor xe o rdf .
this is understandable as only a small fraction of water converts into clathrate , and even in solution , xe is always surrounded by a polyhedral cage of water molecules . as for the diffusion rate of xenon ,
the quantitative differences are much less pronounced because of the low fraction of encaged xenon molecules , but the transition is also similarly sharp , and the parameters of the system reached the new equilibrium almost immediately ( on the order of 1 ns ) . ] the change in the dynamical characteristics of the systems upon clathrate hydrate formation is more clearly seen in * figure [ fig : msd - c - t ] * , where msd plots are shown for different system compositions at the temperatures of ( * a * ) 265 k and ( * b * ) 260 k , for which the transition from the liquid - like to clathrate state is the most pronounced ( as demonstrated by the relative height of the second rdf peaks in * figure [ fig : rdf - c - t ] * ) .
it can be seen that at 265 k , only in the 5% mol.xe/mol.h~2~o system clathrate hydrate formation did occur , while at 260 k , it formed in both the 5% and the 4% systems .
the following observations can be made regarding the dynamics of xenon in the systems : in the `` fast '' part of the msd curves , there are virtually no differences , meaning that a xenon atom inside its hydration shell is completely unaffected by the environment outside the water cage . upon formation of the crystalline phase
, the amplitude of this oscillation decreases from ( 0.122 nm)^2^ at 280 k ( * figure [ fig:270k ] * ) to ( 0.114 nm)^2^ at 270 k and further down to ( 0.1050.100 nm)^2^ at even lower temperatures , where the number of xenon molecules in the clathrate phase is larger , and hence , their statistical contribution is greater .
this reduction is probably linked to the greater rigidity of the clathrate lattice compared to liquid water , which prevents the hydration shell from deformations that could accommodate larger xenon displacements ( xenon molecules `` bounce off '' cage walls without deforming them ) . in the `` slow '' part
, the slope of msd curves is substantially decreased compared to the liquid phase . in the first approximation
, this decrease can be considered proportional to the fraction of xenon in the clathrate phase , since the motion of the latter is that of the crystallite as a whole orders of magnitude slower ( at least , in the absence of vitrification of solution ) . for systems where no clathrate formed ,
extrapolation of the `` slow '' parts to @xmath20 results in almost similar values of ( 0.1050.110 nm)^2^ , which is close to the above - reported amplitude of xenon molecule motion in the cages of clathrate hydrate . at low temperatures , traces of fast motion can be seen against the background of the slow drift . these indicate the former indeed has the character of oscillation .
the period of these oscillations in the crystalline phase is about 2.0 ps ; in the liquid phase , it is somewhat greater about 2.2 ps which is probably again linked to the greater flexibility of the hydrogen bond network in the liquid compared to the clathrate . as seen in * figure [ fig : rdf - c - t ] * , crystalline clathrate hydrate formation is only observed in the most concentrated 5% ( 270 k and below ) and 4% mol.xe/mol.h~2~o ( 260 k and below ) systems , while at lower concentrations , no hints of crystalline phase are seen in the rdf ( except maybe for a small increase of second peak intensity for the 3% system at 255 and 250 k ; visual inspection revealed some signs of emerging ordering but no well - identifiable crystallites ) .
however , it is not entirely clear if this absence of clathrate hydrate is due to the concentration being insufficient for homogeneous nucleation , or simply because of kinetic restrictions not allowing for a detectable amount of crystalline phase to grow over the limited duration of the simulation . as for the 13% systems that remain liquid over the entire temperature range , a steady monotonous behavior
is observed in rdf as a function of the decreasing temperature .
the intensity of the first csp peak gradually decreases , while the sssp and ccp peaks grow .
this is a manifestation of the hydrophobic dissociation , as described above .
in addition , the rdf minima become deeper , and higher - order maxima ( 4th , 5th , and 6th order , up to @xmath62.2 nm ) become clearly distinguishable .
this pronounced structuring of the systems without any detectable crystallinity , as well as the associated collective dynamics in these and similar systems , deserve separate attention outside the boundaries of this work .
it is also interesting to note that in the 4% and 5% systems , there is actually less clathrate hydrate formed in the end of simulations at temperatures below the onset of homogeneous nucleation ( 255 and 250 k for 4% , 260 k and below for 5% , see * figure [ fig : rdf - c - t ] * ) .
while this could possibly be explained by slower diffusion rate not allowing enough xenon molecules to drift towards the growing nuclei of clathrate phase to reach its equilibrium mass during the limited time of simulation , this explanation appears unlikely since we did assure full convergence in the rdf for these systems .
we therefore find it more plausible that this effect is a consequence of low - temperature hydrophobic dissociation : there exists an optimal temperature at which xenon clathrate hydrate is most stable ( e.g. 265 k for the 5% system , * figure [ fig : rdf - c - t](e ) * ) , above which it becomes less stable due to thermal decomposition , and below which it is `` pulled apart '' as a result of entropy - driven effective repulsion of xenon molecules ( see the rdf plots for the 13% systems , * figure [ fig : rdf - c - t](a c ) * ) .
although the entropy of the molecules that compose a crystal is much lower than in the liquid state , there is only 5.75 water molecules per each xenon in the clathrate versus an average of 21.5 in the hydration shell of xenon in the liquid phase , meaning that despite that the water molecules in the latter case are thermodynamically of `` poorer quality '' ( higher entropy though , importantly , still lower than in bulk water ) , ccp configuration could still be more favorable because of the sheer quantity of low - entropy water molecules . to put it in other terms , the hydrophobic solubility of xenon in low - temperature water increases with cooling faster than the thermal stability of clathrate hydrate .
the fact that in the 5% mol.xe/mol.h~2~o system at 270 k , the first two nucleation events were unsuccessful and it took about 23 ns for the first stable crystal nuclei to form , as well as the low amount of xenon that converted to the clathrate phase , suggests that these system parameters ( composition and temperature ) are exceptionally close to the balance between clathrate growth and dissociation . in all other cases , clathrate hydrate formation proceeded right from the start of the simulation even at the preequilibration stage , and
the growth of the crystal occurred very rapidly , despite the fact that even the highest xenon concentrations used in the present work are still several times smaller than that in pure clathrate hydrate ( @xmath21 5% versus 17.4% mol.xe/mol.h~2~o , respectively ) .
it is also important that clathrate nucleation proceeds much faster than the homogeneous nucleation of ice ( which was never observed in our simulations , and is generally a notoriously difficult process to capture with molecular dynamics simulations @xcite ) .
this is because xenon molecules with their well - defined hydration shells enforce a preferential orientation for the surrounding water molecules via sandwiching @xcite and `` inverse solvation '' @xcite , which radically decreases the variety of available equiprobable amorphous local structures that allows pristine water to stay in a supercooled state for macroscopic periods of time .
finally , * figure [ fig : dandvoft ] * shows the plots of xenon diffusion coefficients ( * a * ) and specific volume ( * b * ) of the systems as a function of the temperature ( not shown for @xmath4 = 250 k because the duration of our simulations did not allow us to reach full convergence of this parameter ) . in both plots ,
clathrate formation manifests itself in the form of discontinuities at 5%/265(270 ) k and 4%/260 k ; in particular , the conversion of xenon to the clathrate phase is associated with an appreciable volume decrease ( a point to be recalled below ) .
another interesting observation is that , even though water shows anomalous thermal behavior of density in the studied range of temperatures ( density decreases with temperature ) , the presence of xenon appears to oppose this tendency , and for the 3% system , the volume is nearly constant throughout the entire 250280 k range , while at higher concentrations , when a large fraction of water molecules are part of hydration shells ( above the onset of clathrate formation ) , this effect even dominates over the liquid water behavior . while the practical consequences of this effect are unclear due to the unusual system composition , it is a further interesting manifestation of the negative entropy change associated with hydrophobic hydration .
@xcite ] two - phase simulations of ice crystals in xenon - rich water solution were undertaken in order to investigate how xenon affects the growth of ice , and to assess how our above findings relate to heterogeneous crystal growth .
the simulations were carried out for three systems with ice slabs embedded in ( _ i _ ) neat water , ( _ ii _ ) 1% mol.xe/mol.h~2~o liquid solution ( low xenon concentration ) and ( _ iii _ ) 4% mol.xe/mol.h~2~o ( high xenon concentration ) .
the latter concentration was chosen so as to ensure that no clathrate hydrate forms at our target temperatures ( above the homogeneous nucleation point of 260 k*figures [ fig : rdf - c - t](d ) and [ fig : dandvoft ] * ) .
snapshots of the systems are shown in * figure [ fig:2phase - eq]*. equilibrium coexistence temperature @xmath22 values were obtained via temperature bracketing .
growth / melting of ice was monitored by the change in the potential energy .
the resulting values of @xmath22 were : 275 k for pristine water ; 274 k for the 1% xe system ; 271 k for the 4% xe system .
it can be seen that the deviation of @xmath22 from 275 k is directly proportional to the amount of xenon in the liquid phase , i.e. , conforming to the normal cryoscopic law .
curiously , this suggests that our homogeneous 5%/270 k run ( * figure [ fig:270k ] * ) was close to equilibrium with _ both _ clathrate and ice . increasing the pressure to 10
atm did not shift the equilibrium temperatures , confirming the physical soundness of our choice of @xmath23 atm for the bulk of our calculations reported above .
visual inspection showed that the presence of xenon did affect the structure of ice surface .
in pure water runs , the surface of ice slab was generally smooth down to the length of one unit cell in the corresponding dimension .
however , the heavy xenon molecules were frequently seen to plunge into the surface layers of the slab , sometimes up to two crystal cells deep .
this created a roughened surface with disordered liquid - like local structure of water molecules .
the residence times of xenon molecules were up to about several ns , after which they would be ejected back into the surrounding solution , and the protrusions would be healed , restoring the perfect crystalline structure of ice .
* figure [ fig:2phase - eq](d ) * shows xe and o atom density profiles for the ice/4% mol.xe/mol.h~2~o system at its equilibrium temperature of 271 k ( the xenon plot has been multiplied by 40 to compensate for the difference in total atom number to facilitate visual comparison ) .
the abscissa corresponds to the distance along the _ z _ direction perpendicular to the ice slab surface .
the statistics were accumulated over 10 ns ( 1000 configurations ) .
it can be seen that water molecules are in a crystalline state at 03 nm ( forming pairs of crystal planes ) , are uniformly distributed in the liquid region of the system ( 48 nm ) , and form quasi - liquid ( or quasi - crystalline ) interface layers at 34 nm and 89 nm .
the distribution of xenon is also more or less uniform to the extent of confidence allowed by the poorer statistical sampling in the liquid layer ( absence of clathrate hydrate was confirmed visually and by rdf plots ) , and exhibits 3 or 4 pronounced peaks in the interface layer with intensities of up to @xmath61.52 times that in the liquid phase .
the xe peak closest to the surface is located between two o crystal plane peaks , and the liquid - phase peaks of xe and o coincide .
inspection of trajectories reveals that the ice / solution interface contains some adsorbed xenon molecules that stay there for a duration of about 35 ns and then drift away back into the liquid part of the system ; at any instant , only on the order of 10 molecules are adsorbed on the @xmath631 nm^2^ surface of the simulation slab cell a concentration that is insufficient to cause clathrate hydrate nucleation , even though it is enough to depress the melting temperature by 4 .
the origin of xe concentration peaks can be understood basing on the difference between ice and liquid water densities .
since the density of interfacial water is lower than the bulk value , it can more easily accommodate guest xenon molecules , similarly to what has been previously observed in simulations of hydrophobic hydration in deeply supercooled water .
@xcite on the other hand , formation of xenon clathrate hydrate is associated with the opposite sign of density change ( densification ) , which suggests that nucleation of xenon clathrate hydrate should not be more favorable on the ice / water interface than in the bulk @xcite ( see also the next section ) .
finally , comparison of o atom density profiles in * figure [ fig:2phase - eq](d ) * for the cases of pristine water and water with xenon shows an unperturbed structure , except for a slight enhancement of the third peak in the quasi - liquid layer , @xmath24 nm ( the differences in the topmost part of the plot are artifacts coming from the different periodic boundary conditions ) .
this peak should be attributed to water molecules that solvate xenon adsorbed at the interface .
the present findings for the prism face of ice can , with reasonable caution , be expected to generalize to other ice surfaces , e.g. , basal .
because of the lower relative density of ice , the quasi - liquid layer at its surface is expanded , and provides a favorable environment for relatively large xenon molecules , thus causing them to preferentially adsorb in this layer .
however , additional simulations will be needed to fully confirm this .
after equilibration at the corresponding @xmath22 values , the temperature of our systems was decreased to 270 k for 1% and to 268 k for 4% mol.xe/mol.h~2~o . although the computational expense of the simulations prevented us from doing an extensive systematic investigation , the following behavior was observed : _ 1% , 270 k. _ the ice slab smoothly grows , expelling xenon into the remaining liquid region .
the growth completes in a state when most of the water is in the form of ice , with a thin amorphous ( no crystalline clathrate hydrate ) layer containing all the xenon ( * figure [ fig : freezing](a ) * , @xmath25 ns ) . _ 4% , 270 k , 1 atm .
_ after @xmath625 ns of ice front propagation , the concentration of xenon in the liquid layer built up sufficiently to result in the nucleation of a clathrate hydrate seed ( @xmath610 ns ) , which then proceeded to grow forming a crystal parallel to the ice slab surface .
the resulting structure ( @xmath26 ns ) is shown in * figure [ fig : freezing](b ) * , and the corresponding potential energy plot is shown in the inset .
the clathrate hydrate appears as a relative increase of xenon molecule density slightly above the middle of the liquid layer ; it is more clearly visible in the molecular dynamics trajectory available in * movie s3*.@xcite it must be stressed that nucleation started in the middle of the liquid layer and not at the surface of the ice slab , consistent with our reasoning above based on the density of the interfacial quasi - liquid layer and of clathrate hydrate .
ice surface propagation stopped at @xmath645 ns , although the potential energy continued to decrease , most likely , because of clathrate growth or rearrangement . _ 4% ,
270 k , 10 atm_. the growth rate of ice was similar to the 1 atm simulation , and the growth stopped at @xmath645 ns with the same ice slab thickness .
however , no detectable clathrate hydrate fragments were formed even after @xmath26 ns ( * figure [ fig : freezing](c ) * ; the distribution of xenon in the liquid layer is uniform ) .
this behavior appears to be in disagreement with what should be expected basing on the negative volume change upon clathrate hydrate formation , but may be simply due to the stochastic nature of crystal nucleation .
statistics over multiple freezing trajectories would be needed to clarify this issue . in summary , we have observed that during ice growth in xenon water solution , xenon alters the structure of ice surface making it rougher than in pure water , and that this effect reduces the melting temperature of ice in agreement with the cryoscopic law .
in all three systems , the final volume containing the xenon ( as estimated by measuring the advancement of ice surface ) corresponds to concentrations of about 66.5% which is consistent with the simulation temperatures of 57 k below ice freezing .
however , no clathrate hydrate is formed on the growing prismatic ice surface ; instead , clathrate hydrate forms via homogeneous nucleation when the concentration of xenon in the bulk liquid reaches a threshold value .
first , we would like to concisely recapitulate our key findings .
( _ i _ ) hydration of xenon proceeds via the formation of clusters : a xenon molecule plus its hydration shell ( 21.5 water molecules on average ) .
the cluster experiences diffusion as a whole , and the xenon molecule in its center librates with a characteristic time of @xmath62.2 ps and an amplitude ( squared ) of @xmath21 ( 0.122 nm)^2^ , with occasional hopping from cage to cage . in the crystalline clathrate hydrate phase ,
the period and amplitude are slightly smaller ( 2.0 ps , ( 0.1 nm)^2^ ) .
( _ ii _ ) with increasing xenon concentration , the diffusion rate slightly decreases .
possible explanations include ( 1 ) collisions of clusters and ( 2 ) aggregation of clusters in potential clathrate hydrate nuclei ( similar to methane `` blobs '' observed in a recent study @xcite ) .
( _ iii _ ) homogeneous nucleation of crystalline clathrate hydrate is observed already at concentrations several times smaller than in bulk hydrate .
the structure of the phase seen in simulations agrees well with the si structure typical for real xenon hydrate , and the lattice parameter is in good agreement with the experimental figure of 1.182 nm .
( _ iv _ ) formation of crystalline clathrate hydrate in all cases is observed to happen radically faster than homogeneous nucleation of ice crystals .
the excess ( compared to the equilibrium concentration of dissolved gas at the temperature of simulation ) is released via clathrate growth and not via gas bubble formation .
( _ v _ ) upon clathrate formation , the volume of the systems experiences a sharp drop .
( _ vi _ ) at the simulation pressure of 1 atm , heating the clathrate - containing systems shows no strong dependence of clathrate hydrate decomposition temperature ( between 275 and 280 k ) on the concentration of xenon in the surrounding solution .
( _ vii _ ) although xenon does accumulate in the low - density interfacial quasi - liquid layer on the surface of ice , clathrate hydrate formation is not observed at ice surface in the equilibrium .
however , xenon does affect the morphology of the surface , making it more rough , and decreasing the melting temperature of ice according to the cryoscopic law .
( _ viii _ ) under nonequilibrium conditions , the presence of xenon in the interfacial layer does not prevent ice growth .
clathrate hydrate is formed when the concentration of xenon in the not - frozen - yet part of the system builds up sufficiently for homogeneous nucleation to occur , and this happens in the thickness of the liquid layer .
( _ ix _ ) from ( 1 ) the independence of clathrate hydrate melting temperature ( unlike nucleation temperature ) of the concentration of xenon , ( 2 ) the experimentally observed dependence of decomposition pressure on the temperature,@xcite and ( 3 ) localization of clathrate hydrate nucleation not in the low - density / high - concentration surface layer but in the bulk of the liquid , we conclude that the main thermodynamical driving force behind clathrate hydrate stability is the negative volume change upon addition of solvated molecules to the clathrate crystal ( via the associated negative enthalpy change ) .
( _ x _ ) at temperatures below @xmath6255250 k , the viscosity of xenon solution becomes so high that it has been impossible to reach full convergence in the dynamic properties of the systems during our long ( 100 ns ) simulation runs , implying possible vitrification .
( _ xi _ ) the interaction of xenon molecules in solution can be viewed as the balance between hydrophobic attraction of xenon molecules and entropy - driven `` hydrational '' interaction between xenon and water .
the latter pulls apart contact ( and even solvent - separated , including those forming the clathrate lattice ) pairs of xenon molecules . with decreasing temperature , its relative role increases , resulting in an exotic amorphous metastable structure with a short - range order distance of at least @xmath62 nm resembling an `` atomic gel '' .
supersaturation of water with xenon can produce a phase that is metastable with respect to both homogeneous nucleation of xenon clathrate hydrate and the nucleation of ice .
although it is poorly understood how this phase could be realized in practice , the high solubility of xenon in lipid membranes @xcite could produce such an effect during cooling , when the surplus xenon dissolved in the membranes gets ejected into the surrounding water via an entropy - driven mechanism similar to hydrophobic dissociation .
quantitative assessment of this effect requires separate confirmation by simulations and experiment .
if true , this may be one of the explanations behind the protective effect of xenon observed experimentally on biological samples .
@xcite at the same time , the present study provides evidence against the possibility of `` ice - blocking '' behavior of xenon , where clathrate hydrate could form on the surface of ice , stopping its growth .
this is because xenon hydrate formation favors higher densities , while the density of quasi - liquid layer on the surface of ice is lower than the density of bulk water .
however , this also suggests that other hydrophobic gases might be more useful in this respect ( for example , recent molecular dynamics simulations observed nucleation of _ methane _ clathrate hydrate at ice
water interfaces @xcite ) .
finally , any cryoprotective effects specific to the interaction of xenon with biological objects , especially those containing large hydrophobic regions ( proteins , lipid membranes ) , still remain to be investigated .
given xenon s drastic effects on living organisms , including anaesthesia and even drug addiction therapy , such interactions may turn out crucial for the explanation of the cryoprotective effects of xenon .
we hope that the physicochemical findings of the present work may assist future investigations of the underlying causes of the biological action of xenon .
in the present work , we have carried out molecular dynamics simulations of xenon in water under both homogeneous and heterogeneous ( _ i.e. _ , on ice surface ) conditions .
we observed the homogeneous nucleation of xenon clathrate hydrate , as well as an exotic metastable amorphous phase at high xenon concentrations and low temperatures . despite the fact that xenon adsorbs at the ice surface preferentially to being dissolved in the liquid water , no clathrate hydrate nucleation
is seen at the interface .
we trace this result to the lower density of the interfacial quasi - liquid layer , while clathrate hydrate formation is associated with an increase of density . together with the observed independence of xenon clathrate hydrate decomposition temperature on the concentration , on one hand , and the known temperature dependence of its dissociation pressure , on the other , this leads to the conclusion that the stability of the hydrate is governed not by the concentration of xenon in the liquid phase , but by the interplay of thermal instability of the clathrate lattice , the entropy - driven hydrophobic association / dissociation force , and the negative contribution to the enthalpy resulting from the negative change in volume associated with xenon molecules going from the liquid to clathrate phase .
our results also hint at an optimal pressure - temperature combination , above which the clathrate dissociates due to thermal motion ( translational entropy of xenon molecules , hydrophobic association into contact solute pairs ) , and below which it is `` pulled apart '' by hydrophobic dissociation forces ( entropy difference between `` bulk '' water molecules and those in xenon hydration shells ) .
these findings are interesting both for understanding the effects of xenon on the freezing of water , as well as the broader aspects of clathrate hydrate formation mechanisms .
our study lays the groundwork for future investigations of the cryobiological effects of xenon .
58ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1063/1.1748187 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1748327 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1748038 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1748079 [ _ _ _ _ * * * * , ( ) ] http://www.ncbi.nlm.nih.gov/pmc/articles/pmc1063513/ [ _ _ * * , ( ) ] link:\doibase 10.1126/science.1175074 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1038/nature02135 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1029/96pa01021 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1073/pnas.0602251103 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1126/science.134.3471.15 [ _ _ _ _ * * * * , ( ) ] http://www.ncbi.nlm.nih.gov/pubmed/15181400 [ _ _ _ _ * * * * , ( ) ] http://www.ncbi.nlm.nih.gov/pubmed/18475253 [ _ _ _ _ * * * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1038/sj.bjp.0706063 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1196/annals.1344.025 [ _ _ _ _ * * * * , ( ) ] @noop _ _ _ _ * * * * , ( ) in @noop _ _ ( , ) pp .
@noop _ _ _ _ * * * * , ( ) @noop `` _ method for organs and tissues cryoconservation in situ _ '' , ru patent 2268590 c1 ( ) http://www.ncbi.nlm.nih.gov/pmc/articles/pmc2480575/ [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1016/j.cryobiol.2004.02.002 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1088/0953 - 8984/19/41/412101 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1002/jmr.1130 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1039/tf9625802262 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/jp303330c [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1016/j.bbamem.2013.01.016 [ _ _ * * , ( ) ] link:\doibase 10.1063/1.465634 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1652015 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.481505 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1103/physrevlett.94.217802
[ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1862245 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/jp064021c [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1950647 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1562610 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1002/jcc.20291 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.464397 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1002/andp.19213690304 [ _ _ _ _ * * * * , ( ) ] in @noop _ _ ( , ) pp .
link:\doibase 10.1063/1.448118 [ * * , ( ) ] link:\doibase 10.1080/00268978400101201 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1103/physreva.31.1695 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.328693 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1080/00268978300102851 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.2408420 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1103/physrevlett.96.255502 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/jp980866f [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/je60063a006 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/jp062182a [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/ja1051445 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/ja00133a025 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1038/416409a [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1126/science.1174010 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/jz100340e [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1063/1.1737294 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1038/nmat2508 [ _ _ _ _ * * * * , ( ) ] link:\doibase 10.1021/ja400521e [ _ _ _ _ * * * * , ( ) ] | motivated by recent experiments showing the promise of noble gases as cryoprotectants , we perform molecular dynamics modeling of phase transitions in water with xenon under cooling .
we follow the structure and dynamics of xenon water solution as a function of temperature .
homogeneous nucleation of clathrate hydrate phase is observed and characterized . as the temperature
is further reduced we observe hints of dissociation of clathrate due to stronger hydrophobic hydration , pointing towards a possible instability of clathrate at cryogenic temperatures and conversion to an amorphous phase comprised of `` xenon + hydration shell '' xe@xmath0(h~2~o)~21.5~ clusters .
simulations of ice xenon solution interface in equilibrium and during ice growth reveal the effects of xenon on the ice liquid interface , where adsorbed xenon causes roughening of ice surface but does not preferentially form clathrate .
these results provide evidence against the ice - blocker mechanism of xenon cryoprotection . |
hensel lifting techniques are at the basis of several polynomial factoring algorithms that are fast in practice .
the classical algorithms are designed for generic bivariate polynomials over finite fields without reference to sparsity ( e.g. @xcite ) .
the polytope method of @xcite is intended to factor sparse polynomials more efficiently , by exploiting the structure of their newton polygon .
it promises to be significantly fast when the polygon has a few decompositions , and can help factor families of polynomials which possess the same newton polytope . while the pre - processing stages of the polytope method benefit from the sparsity of the input in reference to its newton polygon , the hensel lifting phase that pursues the boundary factorisations does not do so .
our chain of work in @xcite reveals that the inner workings of hensel lifting remain oblivious to the sparsity of the input as well as fluctuations in the sparsity of intermediary output , so long as one is designing the hensel lifting phase using the dense model for polynomial representation .
in contrast , the sparse distributed representation considers the problem size to be a function of the number of non - zero terms of the polynomails treated , which captures the fluctuation in sparsity throughout the factorisation process . in @xcite
, we revised the analysis of the hensel lifting phase when polynomials are in sparse distributed representation .
we derived that the asymptotic performance in work , space , and cache complexity is critically affected not only by the degree of the input polynomial , but also by the following factors : ( i ) the sparsity of each polynomial multiplication , and ( ii ) the sparsity of the resulting polynomial products to be merged into a final summand .
we further showed that even with advanced additive ( merging ) data structures like the cache aware tournament tree or the cache oblivious @xmath0-merger , the asymptotic performance of the serialised version in all three metrics is still poor .
this was a result of the straightforward implementation which performed polynomial products first off , to be followed by sums of those products , a process that we dubbed _
serialised_. we remedied this by re - engineering the hensel lifting phase such that sums of polynomial products are computed simultaneously using a max priority queue .
this generalises the approach of @xcite for a single polynomial multiplication .
we derived orders of magnitude reduction in work , space , and cache complexity even against a serialised version that employs many possible enhancements , and succeeded in evading expression swell .
hereafter , we label the serialised and the priority queue versions of hensel lifting as ser - hl and pq - hl respectively . more specifically and with regard s to the latter algorithm , we will denote by pq - hl@xmath1 the version that uses binary heap as a priority queue , and by pq - hl@xmath2 the version that uses funnel heap instead .
our experiments in @xcite demonstrate that the polytope method is now able to adapt significantly more efficiently to sparse input when its newton polygon consists of a few edges , something not to have been observed when employing ser - hl . in @xcite , we shifted to enhancing the overlapping algorithm pq - hl@xmath2 .
the motivation lies in the fact that binary heap is not scalable , which , on a serial machine , is interpreted to say that its performance will deteriorate once data no longer fits in in - core memory , thus restricting the number of non - zero terms that input and intermediary output polynomials are permitted to possess . by performing priority queue operations using optimal cache complexity and in a cache oblivious fashion , funnel heap beats binary heap at large scale .
the fact that funnel heap assumes no knowledge of the underlying parameters such as memory level , memory level size , or word length , makes it ideal for applications where polynomial arithmetic is susceptible to fluctuations in sparsity .
however , all of those features can also be observed when adopting an alternate cache oblivious priority queue ( see for example , @xcite ) . as such
, we pursued funnel heap for further attributes that can improve on its asymptotic performance , as well as exploit it at small scale , specifically for hensel lifting . in @xcite , we addressed the chaining optimisation , and how funnel heap can be tailored to implement it in a highly efficient manner .
we exploited that funnel heap is able to identify equal order monomials `` for free '' as part of its inner workings whilst it re - organises itself over sufficiently many updates during one of its special operations known as the `` sweep '' . by this
we were able to eliminate entirely the requirement for searching from the chaining process .
we designed a batched mode for chaining that gets overlapped with funnel heap s mechanism for emptying its in - core components .
in addition to also managing expression swell and irregularity in sparsity , batched chaining is sensitive to the number of distinct monomials residing in funnel heap , as opposed to the number of replicas chained .
this allows the overhead due to batched chaining to decrease with increasing replicas . for sufficiently large input size with respect to the cache - line length , and also sufficiently sparse input and intermediary polynomials ,
batched chaining that is `` search free '' leads to an implementation of hensel lifting that exhibits optimal cache complexity in the number of replicas found in funnel heap , and one that achieves an order of magnitude reduction in space , as well as a reduction in the logarithmic factor in work and cache complexity , when comparing against pq - hl@xmath1 of @xcite .
we label as fh - hl the enhancement of hensel lifting using funnel heap and batched chaining .
this paper extends all of the above work in garnering the prowess of funnel heap . to this end , we incorporate analytical as well as experimental algorithmics techniques as follows : * in section [ proofs ] , we provide proofs of results introduced in @xcite pertaining to properties of funnel heap , several of which are of independent worth extending beyond hensel lifting . for example , we provide complete proofs for the following : * * we establish where the replicas will reside immediately after each insertion into funnel heap .
* * we determine the number of times one is expected to call sweep on each link of funnel heap throughout a given sequence of insertions .
* * given an upper bound on the maximum constituency of funnel heap at any one point in time across a sequence of operations , we compute the total number of links required by funnel heap . * * we establish that the cache complexity by which one performs batched chaining within fh - hl is optimal . * in section [ rank ] , we exploit that the work and cache complexity of an insert operation using funnel heap can be refined to depend on the rank of the inserted monomial product , where rank corresponds to its lifetime in funnel heap . by optimising on the pattern by which insertions and extractions occur during the hensel lifting phase of the polytope method , we are able to obtain an adaptive funnel heap that minimises all of the work , cache , and space complexity of this phase .
this , in turn , maximises the chances of having all polynomial arithmetic performed in the innermost levels of the memory hierarchy , and observes _ nearly optimal _ spatial locality .
we show that the asymptotic costs of such preprocessing can be embedded in the overall costs to perform hensel lifting with batched chaining ( fh - hl ) , independently of the amount of minimisation taking place .
we call the resulting algorithm fh - rank . * in section [ experimental ] ,
we develop the experimental algorithmics component to our work addressing various facets : * * we conduct a detailed empirical study confirming the scalability of funnel heap over the generic binary heap . by simulating out of core behaviour ,
funnel heap is superior once swaps to external memory begin to take place , despite that it performs considerably more work than binary heap .
this supports the notion that funnel heap should be employed even when performing a single polynomial multiplication or division once data grows out of core . * * we support the theoretical analysis of the cache and space complexity in @xcite using accounts of cache misses and memory consumption of fh - hl .
this can be seen as an extension of @xcite , as the performance measures presented there capture only the real execution time . *
* we benchmark fh - rank against several other variants of hensel lifting , which include pq - hl@xmath1 , pq - hl@xmath1 with the chaining method akin to @xcite , pq - hl@xmath2 , and fh - hl .
our empirical account of time , space and cache complexity of fh - rank confirm the predicted asymptotic analysis in all three metrics . * * we demonstrate that funnel heap is a more efficient merger than the cache oblivious @xmath0-merger , which fails to achieve its optimal ( and amortised ) cache complexity when used for performing sums of products . we attribute this to the fact that the polynomial streams to be merged during hensel lifting can not be guaranteed to be of equal size ( as a result of fluctuating sparsity ) .
this provides an empirical proof of concept that the overlapping approach for performing sums of products using one global funnel heap is more suited than the serialised approach , even when the latter uses the best merging structures available .
we now begin with the following section on background literature and results .
in the remainder of this paper , we will consider that in - core memory is of size @xmath3 .
it is organised using cache lines ( disk blocks ) , respectively , each consisting of @xmath4 consecutive words .
all words in a single line are transferred together between in - core and out - of - core memory in one round ( i / o operation ) referred to as a cache miss ( disk block transfer ) .
funnel heap implements insert and extract - max operations in a cache oblivious fashion .
for @xmath5 elements , funnel heap can perform these operations using amortised ( and optimal ) @xmath6 cache misses @xcite . at the innermost level ,
funnel heap is first constructed using simple binary mergers .
each binary merger processes two input sorted streams and produces their final merge .
the heads of the input streams and the tail of the output stream reside in buffers of a limited size .
a binary merger is _ invoked _ using a fill function when merge steps are repetitively performed until its output buffer is full or both its input streams are exhausted .
one can construct binary merge trees by letting the output buffer of one merger be an input buffer of another merger . now let @xmath7 for @xmath8 .
a @xmath0-merger is a binary merge tree with exactly @xmath0 input streams .
the size of the output buffer is @xmath9 , and the sizes of the remaining buffers are defined recursively in a van emde boas fashion ( see @xcite ) .
funnel heap consists of a sequence @xmath10 of @xmath0-mergers , where @xmath0 increases doubly exponentially across the sequence .
the @xmath11 s are linked together in a list , with the help of extra binary mergers and buffers at each juncture of the list . in fig .
[ funnelheap ] , the circles are binary mergers , rectangles are buffers , and triangles are @xmath0-mergers .
link @xmath12 in the linked list consists of a binary merger @xmath13 , two buffers @xmath14 and @xmath15 , and a merger @xmath11 with @xmath16 input buffers labeled as @xmath17 .
link @xmath12 has an associated counter @xmath18 for which @xmath19 .
initially , @xmath20 .
it will be an invariant that @xmath21 are empty .
the first structure in funnel heap is a buffer @xmath22 of extremely small size @xmath23 , dedicated for insertion .
this buffer occupies in - core memory at all times .
funnel heap is now laid out in memory in the order @xmath22 , link @xmath24 , link @xmath25 , etc . within link
@xmath12 the layout order is @xmath18 , @xmath14 , @xmath13 , @xmath15 , @xmath11 , @xmath26 , @xmath27 , @xmath28 .
+ the linked list of buffers and mergers constitute one binary tree @xmath29 with root @xmath30 and with sorted sequences of elements on the edges .
this tree is heap - ordered : when traversing any path towards the root , elements will be passed in increasing order .
if buffer @xmath31 is non - empty , the maximum element will reside in @xmath31 or in @xmath22 .
the smaller mergers in funnel heap are meant to occupy primary memory , and can process sufficiently many insertions and extractions in - core before an expensive operation is encountered .
in contrast , the larger mergers tend to be out of core , and contain elements that are least likely to be accessed in the near future . to perform an extract - max
, we call fill on @xmath30 if buffer @xmath31 is empty . we return the largest element residing in both @xmath22 and @xmath31 . to insert into funnel heap , an element has to be inserted into @xmath22 .
if @xmath22 is full , a sweep function is called .
its purpose is to free the insertion buffer @xmath22 together with all the heavily occupied links in funnel heap which are closer to in - core memory . during a sweep , all elements residing in those dense links are extracted then merged into one single stream . this stream
is then copied sufficiently downwards in funnel heap , towards the first link which has at least one empty input buffer . as a result of sweep ,
the dense links are now free and funnel heap operations are resumed within in - core memory .
the sweep kernel is considerably expensive , yet , sufficiently many insertions and all the extractions can be accounted for between any two sweeps .
let @xmath32 denote a finite field of characteristic @xmath33 , and consider a polynomial @xmath34 $ ] with total degree @xmath35 .
we wish to obtain a polynomial factorisation of @xmath36 into two factors @xmath37 and @xmath38 such that @xmath39 and @xmath37 , @xmath40 $ ] .
let @xmath41 denote the newton polygon @xmath42 of @xmath36 defined as the convex hull of the support vector of @xmath36 .
one identifies suitable subsets @xmath43 of edges belonging to @xmath41 , such that all lattice points can be accounted for by a proper translation of this set of edges .
one then specialises terms of @xmath36 along each edge @xmath44 .
those specialisations are derived from the nonzero terms of @xmath36 whose exponents make up integral points on each @xmath45 , and we label them as @xmath46 .
these can be transformed into laurent polynomials in one variable . for at least one @xmath47 , the associated edge polynomials
@xmath46 ought to be squarefree , for all @xmath48 .
one then begins lifting using the boundary factorisations given by @xmath49 , for all @xmath48 .
for each boundary factorisation , we determine the associated @xmath50 s and @xmath51 s that satisfy the hensel lifting equation @xmath52 for @xmath53 . in @xcite we revised the analysis associated with the bottleneck in computation arising in eq .
( [ maineq2 ] ) , using the sparse distributed representation . in this model of representation ,
a polynomial is exclusively represented as the sum of its non - zero terms , sorted upon some decreasing monomial ordering .
( [ maineq2 ] ) can be modeled using the input and output requirements shown in alg . 1 : [ local - iter ] an integer @xmath0 designating one iterative step in the hensel lifting process .
two sets of univariate polynomials over @xmath32 , @xmath54 , @xmath55 , in sparse distributed monomial order representation .
the polynomial @xmath56 , where @xmath57 .
compute @xmath58 .
compute @xmath59 .
we distinguish between the serialised approach ( ser - hl ) and the overlapping approach ( pq - hl ) for performing the required arithmetic . in the serialised version ,
one performs all polynomial multiplications first , and then merges all the resulting polynomial products . in the overlapping approach
, one handles all arithmetic simultaneously using a single max priority queue . in @xcite , we analysed the work , space , and cache complexity , when polynomials are in sparse distributed representation .
we derived that the performance of the serialised version in all three metrics is critically affected not only by the degree of the input polynomial , but also by the following factors : ( i ) the sparsity of each polynomial multiplication , and ( ii ) the sparsity of the resulting polynomial products to be merged into a final summand .
we further showed that this remains the case even with advanced additive ( merging ) data structures like the cache aware tournament tree or the cache oblivious @xmath0-merger , for performing the sums of resulting polynomial products , and that the serialised approach is not able to fully exploit the cache efficiency of these structures . in the overlapping approach
, the priority queue is initialised using the highest order monomial products generated from each product @xmath60 .
then , terms of @xmath61 are produced in decreasing order of degree , via successive invocations of extract - max upon the priority queue . in @xcite
, we pursued funnel heap as an alternative to the generic binary heap for implementing the overlapping approach .
beyond its cache oblivious nature and optimal cache complexity , we showed that funnel heap allows for a mechanism of chaining that significantly improves its overall performance .
chaining replicas outside the priority queue following insertions is a well known technique ( e.g. see @xcite ) for the case of single polynomial multiplication using binary heap ) .
it helps reduce several parameters tied to performance , such as the total number of extractions required to perform a single polynomial multiplication and the size of the priority queue . in turn , the latter results in reducing the number of monomial comparisons as well as the cache complexity required to perform each priority queue operation . in the straight - forward implementation
, one has to search for a replica immediately after an insertion and then chain the newly inserted element to the end of a linked list tied to that replica in the priority queue .
when using binary heap , chaining hinders performance critically .
each insertion into the linked list denoting the chain incurs a random miss , whereas a single search query may require traversing the entire heap .
it follows that the work and cache complexity of a single insertion amounts to that of traversal of @xmath5 elements for a heap of size @xmath5 .
when employed in the priority queue that is implementing sums of products arising in hensel lifting , chaining becomes daunting as the size of the queue and the amount of replication change irregularly from one iteration to the other .
in @xcite we showed how to exploit the expensive sweep kernel of funnel heap in order to develop a cache friendly batched chaining mechanism ( batched - chain ) that gets intertwined with the sweep s internal operations .
the crux behind our approach lies in delaying chaining and performing it in batches , somehow at the `` right time '' . in the interim ,
a prescribed amount of replication is tolerated , whose effect is shown to be insignificant at scale . here , we restrict chaining to only two specific phases in funnel heap s operations . if one is inserting a monomial product into the ( sorted ) insertion buffer @xmath22 , a replica that resides in @xmath22 is immediately identified and chaining can take place .
one does not attempt to find a replica outside of @xmath22 .
if such a replica exists , chaining will be deferred until @xmath22 is full .
that is when sweep is invoked upon some link @xmath12 as well as one of its input buffers @xmath62 . in the duration of sweep
, one is forming the stream @xmath63 which contains the merged output of all elements in the buffers leading from @xmath14 to @xmath62 together with all elements in links @xmath64 . during the merge
, the replicas residing in those specified regions of funnel heap will be aligned consecutively and thus identified .
one can then chain them all and at once outside of funnel heap .
batched - chain eliminates entirely the need for searching for replicas , and lesser links would be allocated to funnel heap , which reduces garbage collection .
batched - chain is further sensitive to the number of distinct monomials in funnel heap , and not the number of replicas chained .
this can be understood to mean that the overhead due to chaining decreases with increasing replicas , which is intuitively appealing , since chaining is likely to be disabled once the number of replicas is lower than an acceptable threshold .
when incorporating funnel heap and batched - chain into the priority queue algorithm for sums of products , alg .
fh - hl was shown to be significantly fast .
the timings reported in @xcite correspond to overall run - time , with the following percentages of improvement recorded , attained with increasing input size : about 90%-98% ( fh - hl to magma 2.18 - 7 ) , about 90%- 99% ( fh - hl to ser - hl ) , about 10%-60% ( fh - hl to pq - hl@xmath1 ) .
the dramatic reduction in run - time over ser - hl is largely attributed to substantial expression swell , and that over pq - hl@xmath1 is attributed to batched - chain .
in this section we revisit several claims made in @xcite and provide their complete proofs .
those results pertain to the behaviour of funnel heap in general and not necessarily only in relation to hensel lifting , and thus are of independent worth .
unless otherwise stated , all lemmas and corollaries in this specific section are stated in @xcite .
we begin by the following invariant which identifies where the replicas will reside immediately after each insert into funnel heap : let @xmath65 denote the index of the last link in funnel heap . using batched - chain , and immediately following each insertion , there will be no replication within the constituency of any buffer @xmath66 . as a result , a given element in some buffer @xmath67
may only be replicated at most once in each of the preceding buffers @xmath68 in its own link or in each of the buffers @xmath69 in the larger links .
[ noreplicas ] consider the case when one is inserting immediately into the insertion buffer @xmath22 .
batched - chain ensures that chaining is happening immediately , and so there will be no replicas in this particular buffer .
now consider a random @xmath67 for @xmath70 .
we know that one can only write elements to @xmath67 upon a call onto @xmath71 .
this call produces the stream @xmath63 which merges the content of all links @xmath72 together with the content of the path @xmath33 leading from @xmath31 down to @xmath67 .
since batched - chain employs chaining during the formation of @xmath63 , buffer @xmath67 will not contain any replicas .
now , by the first claim above , each buffer @xmath67 in funnel heap contains distinct elements . when @xmath73 , it is straightforward to see that since @xmath22 has no buffers which precede it , each of its elements is replicated at most once in each of the following buffers .
now take @xmath70 .
we know that once sweep is called onto @xmath67 , each buffer @xmath74 in the @xmath12th link must be empty .
also , as we form @xmath63 the end of which is written to @xmath67 we exclude the elements residing in each buffer that is also in the same link as @xmath67 but which precede it in that link .
it follows that the only possible replicas of each element in @xmath67 will be in each of the buffers @xmath68 preceding it in its own link , as well as each of the buffers @xmath75 in the larger links .
the following result captures the number of times one is expected to call sweep on each link of funnel hap throughout a given sequence of insertions and extractions : let @xmath65 denote the index of the last link in funnel heap and let @xmath76 denote the total number of times sweep@xmath77 is called , across a given sequence of insertions and extractions .
then @xmath78 [ totalsweeps ] we proceed by backward induction on @xmath79 .
take @xmath80 .
link @xmath65 has @xmath81 input buffers .
since this is the last link , not all of its input buffers @xmath67 may be written onto using sweep .
in fact , exaclty @xmath82 of them will be so .
we thus have @xmath83 .
we now show that @xmath84 assuming the property holds for @xmath85 .
observe that before any sweep on link @xmath86 has occurred , there should have preceded it exactly @xmath87 sweeps , in order to fill each of the input buffers in link @xmath79 . also , by the inductive hypothesis , the total number of sweeps on link @xmath86 is given by @xmath88 .
combining , we get that there are @xmath89 sweeps on link @xmath79 .
given an upper bound on the maximum constituency of funnel heap at any one point in time across a sequence of operations , we now determine the total number of links the heap requires : let @xmath65 denote the index of the last link in funnel heap .
then @xmath90 where @xmath91 designates the maximum number of elements residing in funnel heap at any point in time .
[ last - versus - total ] from @xcite we invoke the following proven results which we require for our proof : 1 .
the space usage @xmath92 of each input buffer in link @xmath12 satisfies @xmath93 , where @xmath16 is the number of input buffers in link @xmath12 .
the space usage of link @xmath12 is @xmath94 , i.e. it is dominated by the space usage of all of its @xmath16 input buffers .
@xmath95 since link @xmath65 is the last link required by funnel heap to host all elements of its elements , those elements will consume at least one path leading to the first input buffer of link @xmath65 , and at most all @xmath81 such possible paths . by ( 2 )
above , the space usage of each such path is dominated by the size of the input buffer itself and we thus have @xmath96 and @xmath97 .
by @xmath98 we have : @xmath99 where the last equality follows by ( 3 ) above and by unrolling the recursive relation down to the base case . using @xmath100 and composing the logarithm function on the two bases 2 and 4/3 respectively ,
we get @xmath101 .
taking @xmath102 one can proceed analogously as above and obtain @xmath103 .
this concludes the proof . as in @xcite ,
reasoning in the sparse distributed representation produces worst - case versus best case polynomial multiplication , depending on the structure of the output . in the worst case ,
a given multiplication @xmath104 is sparse as it yields a product with @xmath105 non - zero terms , an incidence of a memory bound computation . at best , the multiplication is dense as it yields a product with @xmath106 terms .
when the product has significantly fewer terms due to cancelation of terms , the operation is said to suffer from expression swell .
we now establish that the cache complexity by which one performs batched - chain within fh - hl is optimal . for this
, we require a few notations from @xcite that will be helpful in the forthcoming sections as well . let @xmath107 and @xmath108 , which denote the maximum number of non - zero monomials comprising each @xmath109 and @xmath110 respectively .
let @xmath111 denote the fraction of reduction in the size of the heap during chaining , such that the largest size the priority queue attains during the @xmath0th lifting step is @xmath112 .
let @xmath113 denote the fraction of replication in the total number of monomial products such that the total number of replicas chained during the @xmath0th hensel lifting step is @xmath114 .
the two parameters @xmath111 and @xmath113 reflect , in an asymptotic sense , the changes in the size of the queue as a function of the amount of replicas . particularly , the bounds on @xmath111 and @xmath113 are as follows .
when no replicas are encountered at all during any one lifting step , we have that @xmath115 and @xmath116 . in contrast , when each polynomial in the pair @xmath117 is totally dense and all resulting products in one lifting step are of the same degree , the heap will contain only one element , leading to @xmath118 and @xmath119 .
we now have the following : assume the sparse distributed representation for polynomials .
assume further that @xmath120 .
in the worst case analysis when each polynomial multiplication @xmath121 is sparse , the cache complexity by which one performs batched - chain within fh - hl is optimal .
[ cor1 ] following the analysis in prop .
3.6 of @xcite , the cache complexity of fh - hl is split into two major parts .
the first part accounts for all the insertions into funnel heap using @xmath122 cache misses .
the second part accounts for the cost to perform batched - chain using @xmath123 cache misses .
when @xmath120 , we get that the second summand in the cache complexity incurred by batched - chain is dominated by the cost to perform all the insertions into funnel heap , or that the cost for batched - chain is dominated by @xmath124 , where @xmath125 denotes the total number of replicas chained .
it follows that the cache complexity of batched - chain corresponds to that of traversal , and hence is optimal . in the following ,
we provide a detailed proof that fh - hl , and thanks to batched - chain , outperforms pq - hl@xmath2 ( and thus by transitivity , also pq - hl@xmath1 ) .
in other words , performing sums of products using funnel heap with batched - chain is provably more efficient in work , space , and cache complexity than if we were to resort to a standalone funnel heap implementation .
assume the sparse distributed representation for polynomials , and assume further the conditions in cor .
[ cor1 ] . in the worst case analysis
when each polynomial multiplication @xmath121 is sparse , fh - hl achieves an order of magnitude reduction in space , as well as a reduction in the logarithmic factor in work and cache complexity , over pq - hl@xmath2 .
[ cor2 ] from @xcite , alg .
pq - hl@xmath2 requires the following costs : [ cols="^,^,^",options="header " , ] [ k - ksq ]
in this paper we presented a comprehensive design and analysis that extends the work in @xcite and @xcite .
fh - rank exploits all the features of funnel heap for implementing sums of products arising in hensel lifting of the polytope method , when polynomials are in sparse distributed representation .
those features involve a batched mechanism for chaining replicas as well as optimising on the sequence of insertions and extractions in order to minimise the size of the priority queue as well as the work and cache complexity .
the competitive asymptotics are validated by empirical results , which , in addition to asserting the high efficieny of fh - rank whether or not data fits in in - core memory , help us derive two other main conclusions .
firstly , we confirm that at a large scale , all polynomial arithmetic employing a priority queue will benefit substantially from using funnel heap over binary heap , even without the proposed mechanisms for chaining and/or optimising the sequence of insertions / extractions .
secondly , funnel heap is confirmed to be superior in practice as a merger when tested against the provably optimal @xmath0-merger structure , despite having a higher work complexity .
this is attributed to its ability to adapt to merging input streams of fluctuating density , which in turn , makes funnel heap ideal for performing polynomial arithmetic in the sparse distributed representation , where such fluctuation affects overall performance .
this supports our argument that one should resort to the overlapping approach using a single priority queue , as opposed to handling each of the the local multiplications separately using a local priority queue , to be followed by additive merging of all polynomial streams .
this conclusion remains valid whether or not expression swell is taking place .
we thank the lebanese national council for scientific research and the university research board
american university of beirut , for supporting this work .
50 natexlab#1#1url # 1`#1`urlprefix abu salem , f. k. , el - harake , k. , gemayel , k. , 2015 . cache oblivious sparse polynomial factoring using the funnel heap . in : proc .
pasco 15 . vol .
8660 of lecture notes in computer science .
springer , pp .
715 .
brodal , g. s. , fagerberg , r. , meyer , u. , zeh , n. , 2004 .
cache - oblivious data structures and algorithms for undirected breadth - first search and shortest paths . in : proc . of swat 04 . vol .
3111 of lecture notes in computer science .
springer , pp . 480492 .
monagan , m. , pearce , r. , 2007 .
polynomial division using dynamic arrays , heaps , and packed exponent vectors . in : proc . of casc 07 .
4770 of lecture notes in computer science .
springer , pp . | this work is a comprehensive extension of @xcite that investigates the prowess of the funnel heap for implementing sums of products in the polytope method for factoring polynomials , when the polynomials are in sparse distributed representation .
we exploit that the work and cache complexity of an insert operation using funnel heap can be refined to depend on the rank of the inserted monomial product , where rank corresponds to its lifetime in funnel heap . by optimising on the pattern by which insertions and extractions occur during the hensel lifting phase of the polytope method , we are able to obtain an adaptive funnel heap that minimises all of the work , cache , and space complexity of this phase .
this , in turn , maximises the chances of having all polynomial arithmetic performed in the innermost levels of the memory hierarchy , and observes _ nearly optimal _ spatial locality .
we provide proofs of results introduced in @xcite pertaining to properties of funnel heap , several of which are of independent worth extending beyond hensel lifting . additionally , we conduct a detailed empirical study confirming the superiority of funnel heap over the generic binary heap once swaps to external memory begin to take place .
we support the theoretical analysis of the cache and space complexity in @xcite using accounts of cache misses and memory consumption , and compare the run - time results appearing there against adaptive funnel heap .
we further demonstrate that funnel heap is a more efficient merger than the cache oblivious @xmath0-merger , which fails to achieve its optimal ( and amortised ) cache complexity when used for performing sums of products .
this provides an empirical proof of concept that the overlapping approach for performing sums of products using one global funnel heap is more suited than the serialised approach , even when the latter uses the best merging structures available .
our main conclusion is that funnel heap will outperform binary heap for performing sums of products , whether data fits in in - core memory or not .
hensel lifting , newton polytopes , polynomial factorisation , cache oblivious algorithms and data structures , cache complexity , priority queues , funnel heap |
in the past three decades plasma processing such as surface activation , deposition , or etching has become of great importance for a large number of technological applications . with the goal of optimizing plasma processes ,
several types of plasma sources have been successfully introduced and studied , e.g. , hybrid plasma sources , capacitively coupled radio frequency discharges ( ccrfds ) driven by more than one frequencies , or magnetized ccrfds which are used for magnetically enhanced reactive ion etching ( merie ) .
it has been recently shown that dual frequency ccrfds which exploit the electrical asymmetry effect ( eae ) are most suitable for a separate control of the ion flux and energy at substrates or electrodes @xcite .
it is possible to control the ion energy distribution function over a wide range of parameters .
in fact , this is one of the main goals in the context of plasma processing @xcite . by utilizing the eae
it is even possible to invert the electrode functionality , i.e. , to change the dc self bias voltage from a large negative value to a large positive value even for geometrically symmetric ccrfds with the driven electrode area equal to the grounded area .
it has been further shown that geometrically asymmetric ccrfds ( small driven electrode and large grounded area , or vice versa ) , which produce a natural dc self bias voltage , can be completely symmetrized , in the sense that the natural dc self bias voltage can be completely compensated @xcite . in this work
we study the effect of an externally imposed static magnetic field in combination with the eae on the characteristics of dual frequency ccrfds .
we focus on the sheath dynamics and the iedfs at the electrodes .
similar to the merie concept where usually a static and spatially inhomogeneous magnetic field is applied @xcite , we investigate the effect of a spatially inhomogeneous magnetic field on the iedf .
it will be shown , that the applied magnetic field is able to ( over-)compensate or even to amplify the eae .
the manuscript is organized as follows . after a short introduction of the discharge configuration and the operation regime , a brief review of the employed numerical schemes ,
namely a one - dimensional particle - in - cell algorithm using monte - carlo collisions ( pic ) and the ensemble - in - space time scheme ( est ) , is presented @xcite .
we discuss simulation results for both the magnetized and the non - magnetized case , focussing on the sheath dynamics and the iedfs .
we conclude that the iedfs are completely controlled by the sheath dynamics , even in magnetized ccrfds .
having argued that , we show that est can be used as an efficient postprocessing tool to obtain the iedfs behind the sheaths in non - magnetized and magnetized discharges @xcite .
magnetized discharges used for processes such as merie employ a static and usually inhomogeneous magnetic field to obtain a high electron density at very low gas pressures ( pascal or sub - pascal range ) .
the spatial dimensions of such discharges are typically a few centimeters , and the magnetic field strength can vary up to tens of mt . under such conditions the electrons are magnetized and their effective mobility is reduced ; as will be discussed , later this leads to a larger voltage drop across the plasma bulk and thus to a larger plasma density and corresponding ion flux
. the ions , in contrast , are _ not _ magnetized ( particularly not in the sheath ) , as their larmor radius @xmath0 is large compared with the typical length scales of the system .
for instance , the ratio of @xmath0 and the average sheath width @xmath1 is @xmath2 . here
@xmath3 and @xmath4 are the ion plasma frequency and the ion cyclotron frequency , respectively .
in this work we investigate a magnetized argon discharge at @xmath5 pa .
the chamber configuration is depicted in figure [ fig : geometry ] ; it is a flat , geometrically symmetric cylinder of height @xmath6 = 5 cm .
the chamber walls are dielectric , except for the two opposite electrodes of area 10@xmath7 m@xmath8 ( which is small compared to the diameter of the chamber ) .
the geometric symmetry of the chamber is broken by an external magnetic field of `` half - magnetron '' shape which is applied near the left electrode ( cf . the magnetic field configuration depicted in figure [ fig : geometry ] ) , and by an electrically asymmetric discharge voltage @xmath9 applied to the left electrode via a blocking capacitor of @xmath10 nf .
we assume the following explicit form , where the relative phase @xmath11 the control parameter for the electrical asymmetry effect ( eae ) varies between @xmath12 and @xmath13 : @xmath14 this discharge is simulated under the assumption of a one - dimensional geometry .
clearly , this assumption is rather drastic and neglects the inherent inhomogeneity of the magnetic field and many phenomena related to it ( e.g. , particle drifts and magnetic mirror effects ) .
however , for the purpose of calculating iedfs behind the sheaths the assumption can be justified , at least for the small corridor ( indicated by the dashed lines ) where the magnetic field is approximately parallel to the electrodes . in a magnetized discharge , current flow along the field lines is not inhibited . however , in our case , all field lines end at dielectric sections of the wall ; this suppresses the rf current and turns the field lines into equipotential lines . in other words , the current can only flow into the electrodes , roughly along the dashed corridor . in our one - dimensional model
we assume that the magnetic field is given as follows : @xmath15 where the maximum magnetic field amplitude is @xmath16 = 6 mt and the decay length is @xmath17 = 5 mm . the profile of @xmath18 within the corridor is shown in figure [ fig : magneticfield ] .
preliminary results of a self - consistent two - dimensional pic simulation of the discussed discharge configuration indicate that the presented reasoning is essentially valid : a one - dimensional approximation can capture the essential physics relevant to our topic .
to analyze the magnetized ( and non - magnetized ) ccrfd described above , we employ two conceptually different kinetic approaches .
the first is the explicit electrostatic 1d3v pic code _
yapic _ which has been benchmarked against a number of different pic codes @xcite . within a single pic cycle
the poisson solver , the particle pusher and the monte - carlo collision module are successively iterated until convergence of the overall simulation is achieved . to couple the different modules , a first order field interpolation and charge assignment
is performed .
( see e.g. , @xcite ) .
since the discharge is assumed to be magnetized both particle species , electrons and ions , are moved according to the full lorentz force law @xmath19 , although the ions are mainly non - magnetized due to their large mass ) .
we apply an explicit push scheme based on boris approach @xcite .
collisions of electrons and ions with the fixed background gas are performed in the frame of a slightly modified null - collisions method reported by mertmann et al .
@xcite , with respect to the classical method proposed by skullerud @xcite .
since argon is used as the background gas the collision processes include elastic scattering of both electrons and ions , ionization and excitation due to electron collisions , and backward scattering ( i.e. , charge exchange collisions ) of ions .
cross - section data are taken from phelps et al . and the lxcat database @xcite .
the second approach which is used in the end of this paper relies on the ensemble - in - space time ( est ) scheme which resolves only the sheath , unlike pic which treats the entire discharge .
est is an iterative algorithm based on the solution of a set of kinetic equations for the ions , boltzmann s relation for the electrons , and poisson s equation for the electric field .
similar to pic , est employs the null - collision method to account for elastic and charge exchange collisions of ions , but does not take into account ionization and excitation processes .
est is fed by two input parameters : the first is the ion flux through the sheath .
the second is the sheath voltage which can have an arbitrary but periodic waveform . because of its efficient convergence behavior
, est allows for fast and kinetically self - consistent simulations of sheaths of ccrfds .
more details concerning the mathematical description and the validation of the model can be found in @xcite . in order to calculate the sheath voltage as one input parameter for est from the pic simulation results ,
we define the time - dependent sheath width @xmath20 using @xmath21 this definition is proposed by brinkmann @xcite and can be substituted by other definitions of the sheath edge .
the sheath voltage is then the potential difference between the potential at the position @xmath20 and the electrode potential .
the ion flux as the second input parameter is taken from the pic simulation somewhere in the quasineutral zone near the sheaths .
it is clear that the input parameters for est can also be obtained from other plasma simulation approaches , e.g. , fluid simulations or global plasma models .
it is important to note that est does not include any magnetic field forces .
we start our analysis by comparing two different discharge scenarios using pic simulations .
case i is without and case ii is with an applied static magnetic field at the left electrode .
the simulation parameters are specified in section [ sec : dischargesetup ] .
the relative phase between the two consecutive harmonics of the driving voltage is set to @xmath22 .
figure [ fig : electrondensity_ab ] shows the spacetime dynamics of the electron densities for the two different cases . for case i we obtain , as expected , a symmetric density profile and symmetric sheath dynamics . in contrast , for case ii the discharge shows a significant asymmetry due to the applied magnetic field
. this asymmetry can also be observed in figure [ fig : iondensity_ab ] where the time - averaged ion density profiles are shown .
it is clearly visible that the maximum is shifted and the slope is steeper towards the magnetized region in front of the left electrode .
there are two competing mechanisms governing the electrical conductivity in the magnetized region . at first sight electron transport
is suppressed by the magnetic field .
this leads to a decrease of the conductivity . on the other hand , due to the confinement of the electrons the plasma heating and
thus the ionization processes become more efficient . as a result ,
the plasma density slightly increases locally , followed by corresponding enhancement in the conductivity .
the cumulative effect of these two different mechanisms is that locally the net conductivity decreases , while the plasma density increases . due to the magnetically induced asymmetry
, one can also observe that the sheath in front of the left driven electrode ( which is in the magnetized region ) is much smaller than the sheath in front of the right grounded electrode .
the asymmetry of the electron density profile ( and the ion flux ) is the reason for the different sheath widths .
figure [ fig : sheathwidth_ab ] compares the temporal behavior of the sheath widths for the two different cases of both the driven ( left ) and grounded ( right ) sheath .
while for case i both sheaths show very similar dynamics , for case ii the left magnetized sheath is on average significantly smaller than the right non - magnetized sheath .
this magnetically induced asymmetry results in a substantial dc self bias voltage ( of about 130 v ) .
this can be seen in figure [ fig : potential_ab ] , where the time - averaged potential profiles for the two different cases are compared .
one can also find that the plasma potential is reduced from about 170 v ( for the non - magnetized case ) to about 100 v ( for the magnetized case ) , which is expected .
figure [ fig : iedf_ab ] shows energy distribution functions of ions impinging the electrodes for both cases , without ( top , case i ) and with ( bottom , case ii ) an external magnetic field . with a magnetic field the discharge is strongly asymmetric .
one obtains very high ion energies at the right grounded electrode while the ion energy is smaller at the left driven electrode .
this is clearly due to the dc self bias voltage mentioned before . for the non - magnetized symmetric case
the ion energies at the left and the right electrode are almost equal .
one can observe a small asymmetry in the iedfs which is due to a small asymmetry in the sheath potentials .
the reason for this is the difference of the sheath expansion and the resulting beams of energetic electrons .
this effect has been previously reported by schulze et al .
@xcite .
the parameter which controls the eae and therefore the iedfs is the relative phase between the two consecutive harmonics of the driving voltage @xmath22 .
using a spatially inhomogeneous magnetic field , which itself produces an asymmetry ( as discussed above ) , one is able to modify the asymmetry using the eae .
one can argue , that the eae can be used for the compensation or amplification of the asymmetry induced by the inhomogeneous magnetic field .
however , in any case the two effects compete against each other . here
, we briefly discuss numerical results for case ii ( the magnetized case ) for different relative phases , i.e. , @xmath23 , @xmath24 , and @xmath25 .
figure [ fig : iedf_bc ] shows the iedfs for this scenario .
one can observe that for @xmath23 the eae can be used to amplify the magnetically induced asymmetry ( in terms of increasing the ion bombardment energy at the right grounded electrode and its decreasing at the same time at the left driven electrode ) .
for a relative phase of @xmath26 one obtains an over - compensation of the magnetically induced asymmetry .
the role of the left and the right electrode changes .
it is interesting to note that another knob for controlling the iedfs could be the change of the absolute value of the magnetic field in conjunction with the eae .
however , this should not be discussed in the frame of a simple one - dimensional model , but should be studied by means of a more complete , at least two - dimensional model .
in contrast to the electrons , the ions are not confined by the magnetic field .
this is due to their large mass and a relatively low ion velocity .
the main effect introduced by the external magnetic field is the efficient electron heating in the region where the magnetic field is high .
when comparing the energy distribution functions of ions impinging the electrodes for the magnetized and the non - magnetized case , we can attribute the differences mainly to the different plasma density profiles ( thus the ion flux towards the electrodes ) and to the sheath voltages accelerating the ions .
the iedfs are mainly controlled by the sheath dynamics . to justify this hypothesis
we analyze the sheath dynamics using est , which does _ not _ include any magnetic field effects . for this purpose
we first compare the spatiotemporal sheath dynamics obtained from the pic simulation for the magnetized case ( case ii ) with the results from est . as described in section [ sec : numericalanalysis ] we extract the ion flux and the sheath voltage from the pic simulation and use then as input parameters for est . figure [ fig : electrondensity_b_picest ] shows the spacetime dynamics of the electron density for both approaches .
the sheath dynamics are in very good qualitative agreement .
deviations can be observed particularly near the sheath edge .
this is due to the different collision models used in pic and est .
while the pic scheme allows for ionization collisions of electrons with the background gas , est does not include plasma generation .
therefore one can observe a higher electron density in the pic results .
a second reason ( which has to be proved ) is nonlinear electron resonance heating due to the excitation of higher harmonics in the plasma current and thus the plasma series resonance @xcite .
the higher harmonics are clearly visible in the sheath dynamics . of course , this phenomenon due to nonlinear sheath bulk interaction
is not included in est results , since est resolves the sheath only .
figure [ fig : iedf_a_picest ] compares the iedfs for the non - magnetized case i obtained using both pic and est .
it is evident that both models are in excellent agreement , which is expected .
figure [ fig : iedf_b_picest ] shows the iedf for the magnetized case ii .
the results are again obtained using both pic and est .
also for this case the results are in excellent agreement .
only minor deviations can be observed when comparing the magnitudes of the double peak structure .
it is interesting to note that est yields the same iedfs as pic , regardless of whether the modeled sheath is magnetized or not .
the reason is , that both the sheath dynamics and the ion motion in the sheath are mainly driven by the electric field , and only weakly affected by the magnetic field .
we therefore argue that est can be used for calculating iedfs regardless of the nature of the discharge .
in this work we analyze the effect of a magnetic field on the characteristics of capacitively coupled radio frequency discharges , with a focus on the energy distribution function ( iedf ) of ions impinging the electrodes and the sheath dynamics . using a well - justified one - dimensional particle - in - cell approach with monte - carlo collisions ( pic ) , we show that geometrically symmetric capacitively coupled radio frequency discharges can be asymmetrized by applying a spatially inhomogeneous static magnetic field .
this magnetically induced effect is similar to the electrical asymmetry effect ( eae ) .
we further show that the eae itself can be ( over-)compensated and even amplified .
it this context it could be interesting to study the effect of the magnitude and the shape of the external magnetic field on the electrical asymmetry effect in terms of the symmetry parameter introduced by heil et al . @xcite .
the symmetry parameter is defined as the absolute value of the ratio of the two sheath voltages .
it can be expressed in terms of an algebraic model .
however , it should be discussed in detail in the frame of an at least two - dimensional particle - in - cell code which is able to capture the whole discharge dynamics including the inherent inhomogeneity of the magnetic field and the phenomena related to it ( e.g. , particle drifts and magnetic mirror effects ) .
finally , we find that the novel ensemble - in - space time scheme ( est ) which resolves the sheath only , is able to provide iedfs almost equal to the ones obtained from the fully self - consistent pic simulation , regardless of whether the related discharge is magnetized or non - magnetized .
as such est may be used as an efficient postprocessing tool to obtain the iedfs in many scenarios including non - magnetized and magnetized situations .
this work is supported by the deutsche forschungsgemeinschaft dfg in the frame of collaborative research centre trr 87 . | the effect of a magnetic field on the characteristics of capacitively coupled radio frequency discharges is investigated and found to be substantial . a one - dimensional particle - in - cell simulation shows that geometrically symmetric discharges can be asymmetrized by applying a spatially inhomogeneous magnetic field .
this effect is similar to the recently discovered electrical asymmetry effect .
both effects act independently , they can work in the same direction or compensate each other .
also the ion energy distribution functions at the electrodes are strongly affected by the magnetic field , although only indirectly .
the field influences not the dynamics of the sheath itself but rather its operating conditions , i.e. , the ion flux through it and voltage drop across it . to support this interpretation ,
the particle - in - cell results are compared with the outcome of the recently proposed ensemble - in - spacetime algorithm .
although that scheme resolves only the sheath and neglects magnetization , it is able to reproduce the ion energy distribution functions with very good accuracy , regardless of whether the discharge is magnetized or not . |
binary mass transfer ( mt ) is thought to be responsible for the production of most observed exotic stellar populations , including blue straggler stars , low - mass x - ray binaries , millisecond pulsars , cataclysmic variables , etc . , and
the process is thought to contribute significantly to the rate of type ia supernovae , gravitational waves , gamma - ray bursts , etc .
a complete understanding of binary evolution is a prerequisite for predicting the rates of these high - energy phenomena .
many are thought to occur in star clusters , facilitated by their high densities that stimulate direct gravitational encounters leading to e.g , exchange interactions , fly - bys and/or collisions .
such strong encounters can modify the binary orbital parameters , and hence directly affect the course of binary evolution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . by necessity , full star cluster dynamical simulations , including both direct @xmath4-body ( * ?
* e.g. , ` nbody6 ` ) and monte carlo methods ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , rely on a number of simplifying assumptions in order to treat binary star evolution .
in particular , binary mt is often parameterized , following a pre - calculated grid of more detailed models that were evolved in isolation .
these models are based on a sophisticated treatment of stellar evolution , with some basic assumptions for the actual mt process ( e.g. energy conservation , mass conservation , etc . ) .
the hydrodynamics involved in the mt process is traditionally not modeled directly within full star cluster simulations .
moreover , even if the parameterizations are able to approximate the binary evolution accurately , perturbations during mt are rarely treated properly ( and are often ignored ) .
for example , a direct encounter involving a binary undergoing mt could disrupt the binary or exchange a different star into the system , thus halting the mt process .
scenarios such as these are not accounted for in many star cluster dynamical models .
the assumption that mt occurs in relative isolation ( even in star clusters ) is often justified by comparing time - scales ; in many clusters the duration of mt , particularly for high - mass stars , can be quite short when compared to the single - binary ( 1 + 2 ) encounter time .
however , for lower - mass stars , the typical duration of mt can greatly increase ( e.g. , due to the longer thermal and nuclear time - scales within the stars ) .
furthermore , the binary - binary ( 2 + 2 ) encounter time - scale can be much shorter than the 1 + 2 encounter time in clusters with large binary frequencies .
( e.g. * ? ? ?
* ; * ? ? ?
] in this paper , we begin to quantify the validity of the assumption that mt can be treated as an isolated process within star clusters .
due to mass segregation and dynamical friction , most binaries live in the dense cluster core , where they may not be allowed to evolve fully before being interrupted .
we use the binary evolution code ` seba ` @xcite to simulate the time evolution of the binary orbital parameters of individual binaries undergoing stable mt , as described in section [ method ] .
we then compare the mt time - scales to analytic estimates for the time - scales over which other single and binary stars in the cluster are expected to encounter , or `` interrupt '' , the ongoing mt , as calculated in section [ times ] .
we show that the fraction of interrupted binaries can reach a few tens of percent , for certain cluster parameters .
finally , we discuss in sections [ discussion ] and [ conc ] the implications of our results and offer some concluding remarks .
we present results from @xmath5 individual binary evolution simulations performed using the ` seba ` code . `
seba ` is a fast stellar and binary evolution code , that parameterizes stars by their mass , radius , luminosity , core mass , etc . as functions of time and initial mass .
the binary evolution includes mass loss , mt , angular momentum loss , common envelope evolution , magnetic braking and gravitational radiation . we consider only stable mt between hydrogen - rich non - degenerate stars .
the stability of mt and the mt rate depend mainly on the adiabatic and thermal response of the donor star s radius , and the response of the roche lobe , to the re - arrangement of mass and angular momentum within the binary .
for an overview of the method and a comparison with other methods , see @xcite , appendix b. we follow a nearly identical method to define the masses and orbital parameters of our initial binary population as in @xcite . briefly
, primary masses are chosen from a @xcite initial mass function between 0.1 m@xmath1and 100 m@xmath1 .
we draw mass ratios ( @xmath6 ) from a uniform distribution such that @xmath7 and @xmath8 m@xmath1 .
we choose binary orbital periods and eccentricities from the log - normal and flat distributions , respectively , as observed by @xcite for binaries with solar - type primary stars in the galactic field , which are also consistent with observations of solar - type binaries in open clusters ( ocs ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we initially performed a set of simulations allowing binaries to occupy nearly the entire log - normal period distribution and determined that the longest - period binary that will undergo mt in ` seba ` within a hubble time ( over all primary masses ) is @xmath9 } ) = 4.16 $ ] .
we then imposed this maximum orbital period , along with a minimum orbital period chosen such that the binaries are initially detached , and drew @xmath10 random binaries for ` seba ` .
binaries in the contact stage are excluded in our analysis .
we performed two sets of experiments , one at @xmath11}=-1.5 $ ] , typical for an old globular cluster ( gc ) , and the other at @xmath11}=0 $ ] , appropriate for an oc .
we use the ` seba ` code to evolve these binaries for a hubble time and produce time - averaged orbital parameters ( i.e. , mean separation , mean binary mass ) calculated over the course of stable mt , as well as the cluster age at the onset of mt and the total duration of the mt phase , @xmath12 , for each binary .
we use the output from ` seba ` to calculate the expected time - scales for another single or binary star in the cluster to ( directly ) encounter , or `` interrupt '' , each of the binaries while mt is ongoing , which we will call @xmath13 .
we then compare this time - scale directly to the mt duration , @xmath12 , which is calculated and outputted by ` seba ` . to calculate the encounter time - scale , @xmath13 , we must first find the time for a single ( @xmath14 ) or binary ( @xmath15 ) star to interrupt an ongoing episode of stable mt , which we will refer to as , respectively , @xmath16 and @xmath17 .
as in , @xcite , we define these timescales as follows @xcite : @xmath18 where @xmath19 is the star cluster central number density , @xmath20 is the central root - mean - square velocity ( and we assume @xmath21 ) , @xmath22 is the mean mass of a star in the cluster ( for which we adopt @xmath22 @xmath23 0.5 m@xmath1 ) , @xmath24 is the binary frequency , and @xmath25@xmath26 is the mean gravitationally - focused cross section , where @xmath27 is either the time - averaged separation of the binary undergoing stable mt ( for 1 + 2 encounters ) or the orbital separation corresponding to the cluster hard - soft boundary . ] ( for 2 + 2 encounters ) . finally , the total time until a subsequent encounter is given by , @xmath28 where @xmath29 .
to calculate the parameters in equation [ eq : tau_enc1 ] , we consider total cluster masses @xmath30 ranging from @xmath31 to @xmath10 m@xmath1 in steps of 1 in @xmath32 [ m@xmath1 ] @xmath34 , and assume a plummer model @xcite with a half - mass radius .
] of 3 pc .
see @xcite for specific details .
in general @xmath19 increases with increasing cluster mass , while @xmath24 decreases with increasing cluster mass ( as is consistent with observations , e.g. , @xcite ) .
we find from these binary evolution simulations that , the probability that a given binary will be interrupted during mt in a star cluster core ranges from @xmath2 1% to well over unity .
the main results are plotted in figures [ fig : fig1 ] and [ fig : fig2 ] .
as shown in figure [ fig : fig1 ] , both @xmath35 and @xmath36 are sensitive to the total cluster mass , whereas the mt durations @xmath12 are not .
the sensitivity of @xmath35 and @xmath36 to @xmath30 is due to the dependence of the semi - major axis at the hard - soft boundary on @xmath30 , which results ( on average ) in tighter binaries and lower binary fractions in more massive clusters . in low - mass clusters ,
the mean binary separation and binary fraction are both large , causing @xmath36 to be short . as @xmath30 increases , both the mean binary separation and binary fraction decrease along with the separation at the hard - soft boundary , causing @xmath35 and @xmath36 to decrease and increase , respectively .
importantly , most binaries undergoing stable mt have time - averaged separations that are on the order of only a few stellar radii , causing @xmath35 to exceed a hubble time nearly independent of @xmath30 . on the other hand , @xmath36 is always much shorter than a hubble time , due to the larger mean separation of the entire cluster binary population . in figure
[ fig : fig1 ] , a bimodality is apparent in the distribution for @xmath35 at low metallicity ( i.e. z @xmath23 0.0006 ) that is not apparent in either the @xmath35 distribution at high metallicity ( i.e. z @xmath23 0.02 ) or in either of the @xmath36 distributions .
this is due to a bimodality that appears in the mean binary separation during mass transfer , which dominates the 1 + 2 cross - section but not the 2 + 2 cross - section , since the mean binary separation in the cluster is much larger than the mean binary separation during mass transfer .
the first peak in this bimodal distribution ( at small orbital separations ) is due to systems for which stable mass transfer is initiated in a relatively tight binary , involving relatively unevolved stars .
the second peak is due to systems with asymptotic giant branch ( agb ) donors that have lost so much mass that they are less massive than their companion .
consequently , the resulting mass transfer is stable . for low metallicities ,
this second group is bigger , and a second peak becomes clearly visible ( relative to high metallicities ) .
this is because the lifetimes of low metallicity stars are shorter relative to more metal - rich stars .
therefore , low mass stars are able to evolve off the main sequence , become an agb star , and hence fill their roche lobe all within a hubble time . in figure
[ fig : fig2 ] , we compare @xmath12 and @xmath13 for each individual encounter directly , by calculating the cumulative poisson probability or the probability that a given binary will undergo at least one encounter with another star or binary within a time interval @xmath12 : @xmath37 equation [ eqn : poisson ] approximates the probability that a given binary undergoes a _ direct _ encounter with another star or binary in the cluster before mt stops ( shown on the x - axis in figure [ fig : fig2 ] ) .
importantly , fly - by encounters will also perturb binaries undergoing mt , and this is not accounted for in figures [ fig : fig1 ] and [ fig : fig2 ] . to quantify this effect , we ran simulations of 1 + 2 encounters using the ` fewbody ` code @xcite .
we fix the binary orbital parameters and relative velocity at infinity for the encounters , but sample the impact parameter in multiples of 0.5@xmath38 , where @xmath38 is the impact parameter corresponding to a distance of closest approach equal to the binary semi - major axis @xmath39 , or @xcite : @xmath40 here , @xmath22 is the mean single star mass in the cluster and @xmath41 is the relative velocity at infinity .
we perform @xmath42 numerical scattering experiments for every choice of the impact parameter .
all stars are assumed to be point particles with masses of 1m@xmath1 .
the initial binary semi - major axis and eccentricity are 1 au and 0 , respectively .
the incoming velocity at infinity is set to @xmath43 km / s .
the results of these scattering experiments are illustrated in figure [ fig : fig3 ] , which shows the mean ( absolute ) change in the binary semi - major axis and eccentricity for all simulations , as a function of the impact parameter .
we assume here that a fly - by encounter `` interrupts '' a given binary undergoing mt if either of the two following criteria are met : 1 ) the change in semi - major axis @xmath44 is greater than or equal to the typical stellar radius at the onset of roche lobe overflow ( @xmath45 1 r@xmath1 for most binaries ) , and/or 2 ) the change in eccentricity @xmath46 corresponds to a change in the quantity @xmath47 that is greater than or equal to the typical stellar radius at the onset of mt , where @xmath48 and @xmath49 are the pericenter and apocenter distances , respectively , from the binary center of mass .
we assume that the magnitude of such a perturbation is sufficiently large to at least temporarily halt the mt process ( provided @xmath50 and/or @xmath51 ) .
as shown in figure [ fig : fig3 ] , the perturbative effects of fly - bys remain significant for b @xmath52 1.5@xmath38 .
moreover , both @xmath53 and @xmath54 undergo a steep drop - off between b @xmath23 1@xmath38 and b @xmath23 2@xmath38 , such that neither of our criteria for a `` significant '' perturbation are satisfied for b @xmath55 1.5@xmath38 .
our analytic estimates for the 1 + 2 and 2 + 2 encounter rates are directly proportional to the gravitationally - focused cross - section @xmath56 , for some impact parameter b. hence , if fly - bys are included , the time - scales for 1 + 2 and 2 + 2 encounters should each decrease by a factor @xmath57 , as should @xmath13 .
working from equation [ eqn : poisson ] , we find that the probabilities typically increase by a corresponding factor of @xmath45 2 .
that is , the ratio between the probabilities calculated with and without this additional factor of @xmath45 2.1 in our estimate for @xmath13 peak at a value of @xmath45 2 with very little scatter .
the fraction of binaries expected to be interrupted during mt can be estimated from the probability distributions presented in figure [ fig : fig2 ] .
to do so , we integrate the probability distributions in figure [ fig : fig2 ] and divide by the total number of binaries .
the resulting fractions are @xmath58 0.18 , 0.10 , 0.03 and 0.01 for our model clusters with @xmath30[m@xmath1 ] @xmath23 10@xmath59 , 10@xmath60 , 10@xmath61 and 10@xmath62 , respectively , and a metallicity z @xmath23 0.02 .
these fractions remain the same to within @xmath63% for a metallicity z @xmath23 0.0006 .
as shown in figure [ fig : fig4 ] , the mean mass transfer duration is @xmath45 10@xmath64 years during the first 1 gyr of cluster evolution , then quickly increases by roughly an order of magnitude .
after this , the mean mt duration slightly and slowly decreases over the next @xmath45 12 gyr , but remains much larger than during the initial 1 gyr .
hence , if the initial 1 gyr of the cluster lifetime is excluded from our calculation for the fraction of binaries expected to be interrupted during mt , these estimates increase by a factor @xmath45 2 .
13 gyr seen for the z @xmath23 0.0006 case should be regarded with caution .
this is due in part to small number statistics , and the presence of a handful of binaries with unusually long mt durations at late times .
this is not visible in the error bars , since we are showing the standard error of the mean here , and the number of binaries included in this calculation is large .
perhaps more importantly , these error bars do not reflect any inherent uncertainties in the ` seba ` code and binary evolution in general , which are certainly non - negligible . ]
these fractions correspond approximately to upper limits , since we adopt the mean orbital separation at the hard - soft boundary when calculating the 2 + 2 encounter time . if we adopt instead the mean orbital separation _ below _ the hard - soft boundary ( which does not include the presence of soft binaries ) , these fractions decrease by a factor of a few ( of the same order as the increase from including fly - bys ) . from these fractions
, we can also estimate the number of binaries that are expected to be directly interrupted during ongoing stable mt : @xmath65 where @xmath66 is the total number of initial binaries in the cluster ( estimated as @xmath67 ) , @xmath68 is the fraction of binaries expected to undergo mt ( calculated by ` seba ` for our assumed binary orbital parameter distributions ) and @xmath69 is the fraction of binaries undergoing mt that are expected to be interrupted over a hubble time of cluster evolution ( as calculated above for our four cluster masses ) .
these parameters are shown in table [ table : one ] , for all four assumed values of the total cluster mass .
.parameters adopted in equation 5 to calculate the numbers of interrupted mt binaries .
[ cols="^,^,^,^,^ " , ] [ table : one ] the resulting total numbers of binaries expected to be interrupted over the lifetime of the clusters is then @xmath70 6 , 27 , 59 and 116 for our model clusters with @xmath30[m@xmath1 ] @xmath23 10@xmath59 , 10@xmath60 , 10@xmath61 and 10@xmath62 , respectively .
if perturbative fly - bys are included , these values each increase by a factor @xmath45 2 .
_ the calculated rate of mt interruption for clusters is @xmath45 1 - 10 binary per gyr , with only a weak tendency for this rate to increase with increasing cluster mass . _
we find that a relatively large fraction @xmath69 of binaries undergoing stable mt may encounter another single or binary star in the core of a star cluster before the mt proceeds to completion .
we calculate @xmath58 0.18 , 0.10 , 0.03 and 0.01 for our model clusters with @xmath30[m@xmath1 ] @xmath23 10@xmath59 , 10@xmath60 , 10@xmath61 and 10@xmath62 , respectively , nearly independent of metallicity .
these fractions are only slightly lower than we calculated in @xcite for the fraction of interrupted stellar encounters in star clusters , which ranges from @xmath71 0.01 in massive gcs to @xmath72 0.4 in low - mass ocs .
the origin of this analogous dependence between the fraction of interrupted `` events '' and the total cluster mass is ultimately the same ( see below ) . as a binary undergoing
mt generally has a relatively small geometric cross section , the interloper in these cases are most often binary stars themselves , resulting in 4-body encounters that may dramatically alter the course of the binary evolution .
interrupted mt is expected to occur most often in massive ocs and low - mass gcs ( @xmath73 m@xmath1 ) , where in general the binary frequency is larger , and ( detached ) binaries can have larger orbital separations and therefore larger geometric cross sections .
these larger binary frequencies and geometric cross sections present in the low - mass cluster regime also result in a larger fraction of interrupted stellar encounters .
no publicly available code for star cluster dynamics accounts directly for the hydrodynamics of binary mt needed to accurately model such interrupted / perturbed mt binaries . as discussed in @xcite for interrupted stellar encounters ,
the dependence of @xmath69 on @xmath74 is encouraging for numerical models of star clusters , since low - mass clusters such as ocs are most often modeled with direct @xmath4-body simulations .
direct @xmath4-body codes , such as ` nbody6 ` , can in principle apply perturbations to binaries undergoing ( parameterized ) mt , if the time steps are chosen appropriately . given our results , a more detailed study of
how often such cases of interrupted mt occur and how accurately they are treated in direct @xmath4-body codes is warranted .
massive clusters , on the other hand , are more often modeled with monte carlo codes ( e.g. , * ? ? ?
* ; * ? ? ?
, binary evolution is assumed to run to completion in isolation . in massive gcs
, we find that this assumption is much more valid than in ocs ( albeit not perfect , see figure [ fig : fig2 ] ) , similar to what was found in @xcite for interrupted stellar encounters .
the time - averaged separations of binaries undergoing stable mt are typically on the order of a few solar radii .
however , stable and conservative mt leads to an increase in the binary orbital separation ( if the donor is less massive than the accretor ) . generally toward the end of the mt phase , after a mass - ratio inversion , the orbits expand significantly , increasing the binary cross - section and decreasing the 1 + 2 and 2 + 2 encounter times .
hence , the time - scale for interruption decreases markedly at the end of the mt phase and beyond , which is not properly accounted for in our models ( since we only use the mean binary separation during mt ) .
interestingly , the final separation is on the order of @xmath55 1 au in our models , which is comparable to the hard - soft boundary in old massive ( @xmath75 m@xmath1 ) gcs .
hence , these binaries become subject to ionization due to dynamical encounters with other single and binary stars after mt . assuming a mt origin ,
this could be of particular relevance for blue straggler formation in gcs .
in particular , this naively predicts a rough correlation whereby the fraction of blue stragglers in binaries increases with decreasing @xmath30 .
even the very intimate dance of binary mt can be interrupted by an interloping single or , more often , binary star in a star cluster . up to about a quarter of all binaries undergoing stable mt in the cores of massive ocs and low - mass gcs
( @xmath73 m@xmath1 ) may encounter another star ( or stars ) before completing this phase of evolution ( see figure [ fig : fig2 ] ) .
this fraction decreases towards @xmath76% for more massive gcs , primarily because the binary frequency in these clusters is much lower .
cases of interrupted mt may lead to significantly different outcomes than are assumed for isolated evolution , which in turn may impact the formation rates of mt products in star clusters , from exotic stars like blue stragglers , to sn ia and gamma - ray burst progenitors .
our results highlight the need for a more detailed treatment of binary evolution in star cluster simulations .
the authors would like to thank an anonymous reviewer for useful suggestions for improvement .
n.w.c.l . is grateful for the generous support of an nserc postdoctoral fellowship . a.m.g .
is funded by a national science foundation astronomy and astrophysics postdoctoral fellowship under award no . | binary mass transfer is at the forefront of some of the most exciting puzzles of modern astrophysics , including type ia supernovae , gamma - ray bursts , and the formation of most observed exotic stellar populations .
typically , the evolution is assumed to proceed in isolation , even in dense stellar environments such as star clusters . in this paper , we test the validity of this assumption via the analysis of a large grid of binary evolution models simulated with the ` seba ` code . for every binary ,
we calculate analytically the mean time until another single or binary star comes within the mean separation of the mass - transferring binary , and compare this time - scale to the mean time for stable mass transfer to occur .
we then derive the probability for each respective binary to experience a direct dynamical interruption .
the resulting probability distribution can be integrated to give an estimate for the fraction of binaries undergoing mass transfer that are expected to be disrupted as a function of the host cluster properties .
we find that for lower - mass clusters ( @xmath0 m@xmath1 ) , on the order of a few to a few tens of percent of binaries undergoing mass - transfer are expected to be interrupted by an interloping single , or more often binary , star , over the course of the cluster lifetime , whereas in more massive globular clusters we expect @xmath2 1% to be interrupted .
furthermore , using numerical scattering experiments performed with the ` fewbody ` code , we show that the probability of interruption increases if perturbative fly - bys are considered as well , by a factor @xmath3 .
nsf astronomy and astrophysics postdoctoral fellow ] |
the study of neutron - proton pairing especially in the t=0 channel is a particularly prominent topic these days . while the number of journal articles are far too numerous to reference one might begin to make some headway into the varied approaches by starting from the references found in refs .
@xcite . in
so doing one will find a field of study filled with disagreement and occasionally strife .
the approach here is to study the effects of the t=0 portion of the neutron - proton interaction by removing it altogether . in this way it is hoped that we can understand where in nuclear structure
this portion of the interaction is found to play a large role .
this would in turn suggest where t=0 pairing might most clearly reveal its presence or absence . in this work
we will examine the yrast spectra of the even - even @xmath4 nuclei @xmath0ti,@xmath1ti , @xmath2cr , and @xmath5 with two sets of interaction parameters .
first we have the full fpd6 interaction @xcite , and then we use the same interaction for the two - body t=1 matrix elements while setting all the t=0 two - body interaction matrix elements to zero . we shall denote this interaction as t0fpd6 .
( this modification of an effective interaction is along the same lines of that used by satula et al . to examine wigner energies a few years ago @xcite . )
we have used this modification of fpd6 in the past to study a variety of things and in particular the full fp spectrum of @xmath0ti @xcite .
we found to our surprise and to the surprise of many others that one could obtain a fairly decent spectrum for the levels of @xmath0ti with the second interaction here labeled t0fpd6 .
this is surprising because the t=0 interaction is by no means small .
the only bound two nucleon system is the t=0 combination of two nucleons , the deuteron .
moreover the largest valued matrix elements in the fpd6 interaction tend to be those in the t=0 channel . with the one nucleus @xmath0ti shown in a full fp space calculation in ref .
@xcite we have the first clue that the spectra of even - even nuclei are relatively insensitive to the t=0 two - body interaction matrix elements . to make this more conclusive however we need to examine more nuclei .
we have expanded the examination of the even - even nuclei to those listed above .
the sample we have chosen consists of nuclei with the same collective properties - ground state bands which have some rotational properties but are not extremely rotational . also we now look at transition rates - perhaps these will prove more sensitive to the presence of t=0 two - body matrix elements than the spectra
. this will be our interest in a few sections .
it should be noted that in refs .
@xcite a wide range of topics is addressed beyond the spectra of even - even nuclei .
these topics include a partial dynamical symmetry that arises when one uses the t0fpd6 interaction in a single j shell for @xmath6sc and @xmath0ti . also while using the t0fpd6 interaction a subtle relationship between the t=@xmath7 states in @xmath6sc and t=@xmath8 states in @xmath6ca likewise between the the t=@xmath9 states in @xmath0ti and t=@xmath10 states in @xmath0ca .
we also considered even - odd nuclei and addressed the topic of how the t=0 two - body matrix elements affect b(m1 ) transitions - both spin and orbital components , and gamow - teller transitions . in many cases the transition rates were very sensitive to the presence or absence of the t=0 matrix elements .
this was especially the case for some orbital b(m1 ) s and the gamow - teller transitions . here
things will be kept simple and we focus on the spectra and b(e2 ) s of the yrast levels of selected even - even nuclei .
we will examine the sensitivity of these observables on the t=0 two - body interaction matrix elements by setting them to zero and comparing the results thus obtained with those when the t=0 matrix elements are reintroduced .
as mentioned in the introductions we perform calculations of even - even nuclei with and without the t=0 two - body matrix elements of the fpd6 interaction .
thus each of the figures [ fig : first ] to [ fig : last ] corresponding to the nuclei @xmath0ti , @xmath1ti , @xmath2cr , and @xmath3cr will consist of three columns .
the first column is the yrast spectra calculated with the full fpd6 interaction , the second column with the t0fpd6 interaction , and the third column shows the levels from experiment .
we have previously discussed @xmath0ti so we will start with @xmath1ti , but before dissecting the details of the spectra , note the exceptional results in column 2 of figure [ fig : second ] . there
we see that the results found when all the t=0 two - body interaction matrix elements are set to zero ( t0fpd6 ) in this complete fp calculation that the resulting spectra looks quite reasonable in comparison with both the full fpd6 and the known experimental levels .
indeed it is difficult to choose between t0fpd6 and fpd6 as to which yields a better fit to experiment .
a closer look shows some differences .
the odd spin excitation energies come down by about 1 mev when the t=0 matrix elements are set to zero .
experimentally some odd spin excitation energies are known ( j=1@xmath11 and j=11@xmath11 ) but not as many as the even spins .
the even spin spectrum is slightly more spread out in the full fpd6 calculation that is to say there is more of a tendency towards a rotational spectrum .
the low lying spectrum is fit somewhat better with the full fpd6 interaction but there is a better fit to the j=14@xmath12 state using the t0fpd6 interaction .
there are similar stories for @xmath2cr ( figures [ fig : third],[fig : fourth ] ) and @xmath3cr ( figures [ fig : fifth ] and [ fig : last ] ) .
( these nuclei are of some interest as they both display backbends @xcite- @xcite . ) in @xmath0ti there is a bigger difference between fpd6 and t0fpd6 perhaps results from the greater deformation in @xmath0ti .
as is the case for @xmath1ti and indeed many of the even - even nuclei in the region very few odd spin states are known .
if these were to be made available experimentally it might be easier to demonstrate more clearly the preference between the full fpd6 or the modification t0fpd6 .
the fact that one can even think of offering into competition an interaction in which all the t=0 two - body interaction matrix elements are set to zero is quite remarkable and perhaps a bit disturbing .
the b(e2 ) rates for @xmath0ti , @xmath1ti , @xmath2cr , and @xmath3cr are listed in tables [ tab : first ] to [ tab : last ] .
we allow up to t nucleons to be excited from the f@xmath13 shell to the rest of the fp shell .
the values of t used are 4,3,2 , and 2 respectively .
the effective charges used are the standard @xmath14 for the proton and @xmath15 for the neutron .
the difference in the effective charges from 1 and 0 is intended to take care of the fact that the @xmath16 and higher excitations are not present in this model space .
the results for fpd6 and t0fpd6 are shown .
we also display the ratios of the results for the 2 interactions .
for @xmath1ti the reintroduction of the t=0 two - body matrix elements causes an increase ( relative to t0fpd6 ) of a factor of two or more for all the transitions considered .
so there is evidence here that the t=0 matrix elements contribute to the collectivity .
this is not seen by looking at just the excitation energies in @xmath1ti where fpd6 and t0fpd6 give very similar results . in @xmath2cr and
@xmath3cr a more interesting behavior evolves . in t=2 calculations , the b(e2 ) s for low spin transitions
get and enhancement with the reintroduction of the t=0 two - body matrix elements but for some of the higher spins we get a suppression .
the suppression occurs when backbending occurs as in @xmath3cr in the j=8 @xmath17 10 transition . in summary ,
in studying the problem of the t=0 neutron - proton interaction in a nucleus , it may prove more fruitful to begin by removing this channel altogether as was done here by setting all the t=0 two - body matrix elements elements to zero and then reintroducing them , rather than adopting the more common approach of investigating the effects of a pairing interaction separated from the rest of the interaction .
this may be especially true in the shell model as the suggestion has been made by satula and wyss that it may not be appropriate to separate out a pairing interaction from the rest of the hamiltonian in a shell model context @xcite . | calculations of the spectra of various even - even nuclei in the fp shell ( @xmath0ti , @xmath1ti , @xmath2cr , and @xmath3cr ) are performed with two sets of two - body interaction matrix elements .
the first set consists of the matrix elements of the fpd6 interaction .
the second set have the same t=1 two - body matrix elements as the fpd6 interaction , but all the t=0 two - body matrix elements are set equal to zero . despite the drastic differences between the two interactions , the spectra they yield are very similar and indeed it is difficult to say which set gives a better fit to experiment . that the results for the yrast spectra are insensitive to the presence or absence of t=0 two - body matrix elements is surprising because the only bound two nucleon system has t=0 , namely the deuteron .
also there is the general folklore that t=0 matrix elements are responsible for nuclear collectivity .
electric quadrupole transition rates are also examined .
it is found that the reintroduction of t=0 matrix elements leads to an enhancement of b(e2 ) s for lower spin transitions but in some cases for higher spin transitions one gets another surprising result that there is a small suppression .
insensitivity of the yrast spectra of even - even nuclei to the t=0 two - body interaction matrix elements + shadow j.q .
robinson and larry zamick department of physics and astronomy , rutgers university , piscataway , + new jersey 08855 |
spin glasses ( sg ) are disordered magnetic alloys.@xcite their microscopic modelization includes several interactions , such as the rkky interaction that is invariant over rotations,@xcite and the dzyaloshinsky - moriya ( dm ) interaction that breaks the rotational symmetry.@xcite therefore , in theoretical physics sgs are often studied with simplified models that take in account only a few essential characteristics ( in particular , quenched disorder and symmetries).@xcite the dm interaction , through a spin - orbit coupling with a third spin , causes the interactions between spins in any sg to have a certain degree of random anisotropy .
this implies that real sgs are never fully isotropic ( this theoretical limit is named heisenberg sg ) .
in fact , materials are classified according to the degree of anisotropy in their interactions,@xcite which turns out to be relevant in their non - equilibrium magnetic response.@xcite on one end of the materials spectrum we find the extremely anisotropic fe@xmath1mn@xmath1tio@xmath2 , which is maybe the best realization of the ideal limit of an ising sg ( ising sgs correspond to the idealization of uniaxial spins ) . on the other end
, we have very isotropic alloys such as agmn or cumn ( whose modelization is notoriously difficult,@xcite due to the presence of short range spin - density wave ordering@xcite ) .
despite the variety of interactions , already in the early 90s there was general experimental agreement on that sgs undergo a phase transition at sufficiently low temperature .
@xcite on the other hand , theoretical work was less advanced , even though one works with extremely simple models . for the ising sg
there were arguments supporting the existence of a phase transition,@xcite that were later confirmed numerically.@xcite in the heisenberg case , instead , all the attempts carried out during the 80s and 90s failed in finding a phase transition at a finite temperature @xmath3.@xcite in fact , matsubara et al . argued in 1991 that once a small anisotropic term is added to the heisenberg hamiltonian the phase transition becomes visible.@xcite this was in agreement with a later domain - wall computation.@xcite the accepted picture at the time was that the lower critical dimension ( i.e. the spatial dimension below which there is no phase transition ) lies somewhere between 3d and 4d.@xcite however , the story was slightly more complicated .
villain and coworkers made a provocative suggestion hypothesizing that , although maybe there was no spin glass transition , a different order parameter called chirality ( or vorticity ) could be critical.@xcite chirality is a scalar observable that describes vorticity and alignment between neighboring spins ( see below the precise definition in sect .
[ sec : obs ] ) .
this idea was elaborated by kawamura in his 1992 _ spin - chirality decoupling scenario : _ in the ideal case of a purely isotropic system the spin and chiral glass order parameters would be decoupled , but the introduction of any small anisotropy would couple them.@xcite kawamura s scenario was apparently consistent with all the observations until 2003 , when lee and young employed more efficient simulation algorithms and finite - size scaling techniques to show that the spin glass channel is critical also in the fully isotropic model ( i.e. the heisenberg limit).@xcite both order parameters seemed to become positive at the same temperature .
further simulations confirmed the existence of a spin glass phase transition , although uncertainty remains on whether the transition is unique @xcite or chiralities order at a slightly higher temperature @xmath4.@xcite a parallel issue is measuring the chiral order parameter in experiments .
kawamura proposed in 2003 that the extraordinary hall resistivity is a simple function of the linear and non - linear chiral - glass ( cg ) susceptibilities.@xcite experiments based on this proposal observed the chiral transition and measured , for instance , the critical exponent @xmath5.@xcite interestingly enough , the value of @xmath5 turned out to be compatible between spin and chiral glass sector
. nonetheless , it was impossible to identify a universality class despite the critical exponents of these systems had been extensively measured ( at least in the sg sector):@xcite the impression was that they change in a continuous way from the heisenberg to the ising limit , @xcite as we increased the anisotropy . however , analogy with ferromagnetic materials suggests a different interpretation .
anisotropy would be a relevant parameter in the sense of the renormalization group.@xcite there should be a new dominant fixed point , and symmetry considerations lead to think it should belong to the ising - edwards - anderson universality class . yet , when we add a relevant parameter to the hamiltonian , there should be some _ cross - over _ effects .
in other words , one expects that while the correlation length @xmath6 is small , the critical exponents are closer to the heisenberg - edwards - anderson universality class , and that only for large enough @xmath6 the universality class reveals its nature .
notwithstanding , it is very hard , both numerically and experimentally , to prepare a sg with a large correlation length , since one should wait very long times ( it has been argued that the waiting time @xmath7 required to reach a certain coherence length is proportional to almost its seventh power , see e.g. refs . and ) . to our knowledge , for this reason , the largest measured correlation lengths are of the order of only one hundred lattice spacings .
@xcite that is a rather small distance to reveal the true universality class , so it is plausible that experiments will find critical exponents between the two universality classes . to further complicate things , in experiments one has to take in account at least two relevant crossovers .
the first is the competition , that we just pointed out , between the isotropic and the anisotropic fixed points .
it is the one we treat in this paper .
the second crossover , that we will not address , is about short versus long range interactions .
in fact , the hamiltonian we treat is short range , but the dm interaction has been shown to be quasi - long - range , in the sense that the interactions are long range , but only until a cut - off distance of the order of some tens of atomic spacings .
@xcite aiming to untangle these questions , one of the authors undertook a numerical study of heisenberg sgs with very weak random anisotropies , @xcite but the scenario remained even more foggy , since it was observed that : * the chiral glass critical temperature @xmath8 was significantly higher than @xmath9 , in disagreement with experiments and expectations .
* apparently , the chiral susceptibility was _ not _ divergent at @xmath4 .
this is surprising and , apparently , in contrast with experiments .
@xcite technically , this lacking divergence appeared as a very large anomalous dimension @xmath10 . @xcite * introducing very weak anisotropies changed dramatically @xmath9 .
for example , the @xmath9 found by comparing systems of size @xmath11 was about twice its equivalent on the fully isotropic model .
this is surprising , since one expects that the critical temperature would change very little from the isotropic case when @xmath12 is as small as in ref . .
in this paper we will focus on the uniqueness of the phase transition and on the universality class , proposing that there is a unique transition , belonging to the ising - edwards - anderson ( iea ) universality class.@xcite we will also give an interpretation to the results of ref .
, showing that the apparent inconsistencies are due to scaling corrections , that we will try to characterize , since we believe them to be fundamental both in the interpretation of numerical simulations and of experiments . to do all this , we will study numerically the heisenberg spin glass model with strong random anisotropies , in order to suppress both finite - size effects , and traces of the cross - over from the isotropic limit .
we simulated on the largest lattices to present ( up to @xmath13 ) , over a wide temperature range.@xcite this has been possible thanks to an intense use of graphic accelerators ( gpus ) for the computations .
we made use of the _ tianhe-1a _ gpu cluster in tianjin , china,@xcite and of the _ minotauro _ gpu cluster in barcelona.@xcite the rest of the paper is organized as follows . in section
[ sec : model ] we give an explicit definition of the model , and we introduce the observables we extracted from simulations and analysis .
section [ sec : simulationdetails ] contains details on how we practically conducted the simulations , although much information is relegated to appendix [ sec : numericaldetails ] , where we also discuss the use of gpus for spin glasses . on section [ sec : fss ]
we recall some finite - size scaling concepts we used in our analysis , to find the critical temperatures and exponents ( some technical details are given in section [ sec : methodology ] ) .
finally , in section [ sec : results ] we refer the results obtained in this work , and give our conclusions in section [ sec : conclusion ] .
we study the model introduced by matsubara et al.,@xcite which is particularly convenient because of its simplicity .
we consider @xmath14 3-dimensional unitary vectors @xmath15 on a cubic lattice of linear size @xmath16 , with periodic boundary conditions .
the hamiltonian is @xmath17 where @xmath18 means the sum goes only over nearest neighbors , and the indices @xmath19 indicate the component of the spins .
@xmath20 is the isotropic coupling between sites @xmath21 and @xmath22 .
@xmath23 is the anisotropy operator : a @xmath24 symmetric matrix , where the six matrix elements @xmath25 are independent random variables .
there is quenched disorder , this means that the time scales of the couplings @xmath26 are infinitely larger than those of our dynamic variables , so we represent them as constant in time random variables , with @xmath27 , @xmath28 and @xmath29 the overline @xmath30 denotes the averages over the instances of the disorder , while for thermal averages we will use @xmath31 .
each different realization of the couplings @xmath26 is called _
sample_. independent systems with the same couplings are _ replicas _ of the same sample .
we use two replicas per sample .
notice that if all the matrix elements @xmath32 are zero we recover the fully isotropic heisenberg model , with @xmath33 symmetry .
however , if the @xmath32 are non - vanishing , the only remaining symmetry is time - reversal : @xmath34 for all the spins in the lattice .
time reversal is an instance of the @xmath35 symmetry .
this is the symmetry group of the iea model.@xcite hence , we expect that the @xmath35 symmetry will be spontaneously broken in a unique phase transition belonging to the iea universality class ( see e.g. ref . ) .
of course , underlying this expectation is the assumption that the anisotropic coupling is a relevant perturbation in the renormalization group sense ( as it is the case in ferromagnets @xcite ) . in fact , the infinite - anisotropy limit can be explicitly worked out for a problem with _ site _
anisotropy [ rather than link anisotropy as in eq . ]
: one finds an iea - like behavior.@xcite it is widely accepted that the universality class does not change with the probability distribution of the couplings.@xcite we take advantage of this , and choose a bimodal distribution for @xmath20 and @xmath36 , @xmath37 and @xmath38 .
these couplings can be stored in a single bit , which is important because we are using gpus , special hardware devices where memory read / write should be minimized ( appendix [ sec : numericaldetails ] ) .
we chose the two different values @xmath39 . we want to compare our results with those in ref . , where simulations were done on samples with weak random anisotropies . in that work
the @xmath36 did not follow a bimodal distribution , but were uniformly distributed between @xmath40 and @xmath41 . to make proper comparisons we consider the standard deviation of the distribution . for bimodal distributions
it is exactly @xmath12 , in ref .
it is @xmath42 . to define the sg and cg order parameters we use two replicas .
the overlap field is @xmath43 , where @xmath44 and @xmath45 are replica indices .
its fourier transform at wave vector @xmath46 is @xmath47 .
the chirality represents the oriented volume of the parallelepiped we can construct on 3 consecutive spins : @xmath48 where @xmath49 is the unitary vector in the @xmath50 direction .
the cg overlap is defined similarly to the sg one , as @xmath51 . again
@xmath44 and @xmath45 indicate the replica .
the fourier transform of the cg overlap field is @xmath52 .
we define the wave - vector dependent susceptibilities on the two overlap fields as @xmath53 and from each of them we can compute the correlation length of the related field @xcite @xmath54 being @xmath55 or permutations .
when computing @xmath56 , one can choose @xmath50 parallel or orthogonal to the wave vector @xmath57 . as it was already observed in ref . , there is no apparent difference between the two options , so we averaged over all the values of @xmath50 to enhance our statistics .
we used monte carlo dynamics throughout all the work .
previous experience advises to mix several monte carlo dynamics.@xcite in fact , our single monte carlo step ( mcs ) consisted of ( in successive order ) :
( i ) one full lattice sweep with the heat - bath algorithm , ( ii ) @xmath16 lattice sweeps of microcanonical overrelaxation algorithm,@xcite and ( iii ) one single parallel tempering sweep.@xcite the combination of the first two , which update one spin at a time , has been shown to be effective in the case of isotropic sgs @xcite and other models with frustration .
@xcite both heat - bath and overrelaxation are directly generalized to the anisotropic case.@xcite all the simulations were run on nvidia tesla gpus . except @xmath13 , @xmath58 , where we parallelized 45 gpus , each sample was simulated on a single gpu .
the interested reader can find in appendix [ sec : numericaldetails ] details on how they were performed .
table [ tab : nsamples ] depicts the relevant simulation parameters . for given @xmath16 and @xmath12 ,
the simulations were all equally long , except for @xmath13 , @xmath58 , where we extended the simulation of the samples with the longest relaxation times .
.details of the simulations .
we show the simulation parameters for each anisotropy @xmath12 , and lattice size @xmath16 .
@xmath59 is the number of simulated samples .
@xmath60 is the number of temperatures that were used in parallel tempering .
the temperatures followed a geometric sequence between @xmath61 and @xmath62 , and @xmath60 was chosen so that the parallel tempering s acceptance was around @xmath63 .
@xmath64 is the minimum number of mcs for each simulation .
the simulation for @xmath13 , @xmath65 was intended only to locate @xmath4 . [ cols="^,^,^,^,^,^,^",options="header " , ] although the sg and cg transitions do not coincide yet with our values of @xmath16 and @xmath12 , the critical temperatures , as well as @xmath66 , become more and more similar as the linear size of the system increases . moreover ,
the decrease of @xmath67 as a function of @xmath16 has not yet stabilized , so it is likely that the chiral quantities will keep changing with bigger lattice sizes . as explained in sect .
[ sec : model - symmetries ] , we expect that the transition should belong to the iea universality class . to confirm this expectation , we make the ansatz of a unique transition , of the iea universality class , to seek if the two critical temperatures join for @xmath68 .
figure [ fig : deltatc ] ( upper half ) shows the difference between the critical temperatures as a function of the natural scale for first order corrections to scaling , @xmath69 [ eq . ] .
again , @xmath70 and @xmath71 are taken from ref . .
not only fig .
[ fig : deltatc ] ( top ) reveals a marked increase of the speed of the convergence for @xmath13 ( to which corresponds the smallest anomalous exponent @xmath67 ) , but also , a linear interpolation to infinite volume , taking that point and the previous , extrapolates @xmath72 within the error . and the spin glass transition temperature @xmath73 , in units of @xmath73 ( see table [ tab : tsg ] for the extrapolations of @xmath73 ) . the exponents @xmath70 and @xmath71 are taken from ref . .
in the * upper plot * we represent our data , for @xmath74 .
the two transitions get closer when we increase @xmath16 , and the approach appears faster when the lattice size increases .
notice that a linear interpolation between the two largest lattice sizes intercepts the @xmath75 axis compatibly with a coupling between the two transitions ( i.e. @xmath72 ) . on the * bottom plot
* we show data from ref .
, where much lower anisotropies were considered .
here the scenario is completely different , since the critical temperatures drift apart for large enough @xmath16 .
the horizontal dashed line corresponds to @xmath76 . ]
[ fig : tc_vs_l ] shows how the sg and cg critical temperatures approach each other with @xmath16 .
again , @xmath4 gets closer to @xmath9 , and the speed of the approach increases with the lattice size .
the points in the intercept represent extrapolations to the thermodynamic limit of the @xmath9 .
since the observations are compatible with the ansatz of a unique phase transition , belonging to the iea universality class , we used the infinite - size limit of @xmath9 to plot the model s phase diagram ( fig .
[ fig : tc_vs_l ] , inset).@xcite ( * large plot * ) .
the points on the intercept are the @xmath68 extrapolations from table [ tab : tsg ] .
the * inset * shows the phase diagram of the model with these same points , as the most economic interpretation of our data is that in the thermodynamic limit @xmath72 .
the @xmath77 point is borrowed from ref . .
] both plots of fig .
[ fig : deltatc ] show the same observable , for different anisotropies .
the top plot depicts our data , in the case of strong anisotropies @xmath74 .
the bottom one represents the case of weak anisotropies ( @xmath78),@xcite coming from ref . .
the behavior is very different between the two cases .
for strong anisotropies , the critical temperatures tend to meet as we increase @xmath16 .
that is qualitatively very different from the weak anisotropy case , where their distance increases .
we can ask ourselves where this qualitative difference of behavior comes from .
if we compare same system sizes and different @xmath12 in table [ tab : tcg ] , we notice that finite - size effects are larger ( and @xmath79 closer to two ) the smaller the anisotropy .
these differences in the finite - size effects are appreciable with a factor 2 change in the anisotropy ( from @xmath65 to @xmath58 ) , so it is reasonable that suppressing the anisotropy by a factor 17 or 35 will increase drastically the finite - size effects .
the most economic explanation is then that there is a non - asymptotic effect that disappears with much larger systems or , as we have seen , with larger anisotropies .
in other words there is a @xmath80 after which @xmath9 and @xmath4 start joining . for @xmath78
, @xmath81 is so large that we observe a growing @xmath82 , while for @xmath83 we find @xmath84 . another peculiarity outcoming from ref
. arises from the sg transition alone .
it had been observed that a very weak perturbation on the symmetry of the isotropic system implied huge changes in the critical temperature , while one would expect that the transition line is smooth . to solve this dilemma , we take advantage of having strong evidence for the universality class of the transition .
so , we take the data from ref . , and use once again the exponents @xmath71 and @xmath70 in ref . to extrapolate the infinite volume limit with second order corrections to scaling ( eq . [ eq : scalingcorrections ] ) .
the fit is good ( @xmath85 ) , and , as we show in fig .
[ fig : tsg_d003 ] , its @xmath68 extrapolation for the critical temperature is compatible with @xmath86 within one standard deviation .
thus , taming the finite - size effects was enough to make the scenario consistent , and the issue reduces to the fact that finite - size effects are extremely strong when the anisotropy is smaller .
, @xcite with extrapolations to the thermodynamic limit assuming the ising - edwards - anderson universality class .
the data is the same in both plots .
the dashed line is a fit of the scaling in @xmath16 , considering corrections up to the second order ( eq . [ eq : scalingcorrections ] ) .
the * large * figure displays the trend of the scaling variable @xmath87 as a function of @xmath88 .
the * inset * shows the same data set , plotting @xmath89 as a function of @xmath90 , see eq . .
the extrapolation to large-@xmath16 ( the point in the intercept ) is compared with @xmath9 of @xmath77 from .
the full horizontal line is the central value of @xmath91 , and the dashed lines define the error . ]
we performed a numerical study of the critical behavior of heisenberg spin glasses with strong bimodal anisotropies .
our aim was to clarify the role of scaling - corrections , as well as the crossover effects between the heisenberg and ising universality classes , to be expected when the anisotropic interactions are present .
in fact , we show that anisotropic interactions are a relevant perturbation in the renormalization group sense : no matter how small the anisotropy , the asymptotic critical exponents are those of the ising - edwards - anderson model . however , a fairly large correlation length maybe needed to reach the asymptotic regime .
this observation is relevant for the interpretation of both numerical simulations,@xcite and experiments.@xcite it is then clear that large system sizes are needed to make progress , something that calls for extraordinary simulation methods .
therefore , we performed single - gpu and multi - gpu simulations to thermalize lattices up to @xmath13 at low temperatures .
as side benefit , our work provides a proof - of - concept for gpu and multi - gpu massive simulation of spin - glasses with continuous degrees of freedom .
this topic is elaborated further in appendix [ sec : numericaldetails ] .
we performed a finite - size scaling analysis based on phenomenological renormalization.@xcite we imposed scale - invariance on the second - moment correlation length in units of the system size , @xmath92 .
we followed this approach for both the chiral and spin glass order parameters .
our results for the spin - glass sector were crystal clear : all the indicators of the universality class were compatible with their counterparts in the ising - edwards - anderson model . on the other hand , in the chiral sector scaling - corrections were annoyingly large , despite they decrease upon increasing the magnitude of the anisotropic interactions . regarding the coupling of chiral and spin glass transition , our numerical results seem to indicate that the two phase - transitions take place at the same temperature ( i.e. @xmath93 )
however , it is important to stress that we need our very largest lattices to observe this trend .
nevertheless , what we see is in agreement with both kawamura s prediction and experiments , where the phase transitions are apparently coupled , and the chiral glass susceptibility is divergent.@xcite moreover , we were able to rationalize the numerical results in ref . with corrections to scaling , by assuming the ising - edwards - anderson universality class .
we remark that there are strong analogies between the interpretation of numerical and experimental data . in both cases
, there is a relevant length - scale ( the correlation length for experiments , the system size for simulations ) .
if that length is large enough , the asymptotic ising - edwards - anderson universality class should be observed . otherwise , intermediate results between heisenberg and ising are to be expected , and indeed appear.@xcite the difficulty in reaching the asymptotic regime lies on time : the time growth of the correlation length is remarkably slow ( @xmath94 with @xmath95,@xcite where @xmath7 is the waiting time ) .
indeed , the current experimental record is around @xmath96 100 lattice spacings,@xcite pretty far from the thermodynamic limit.@xcite hence attention should shift to the study of the intermediate cross - over regime .
an intriguing possibility appears : one could envisage an experimental study of the crossover effects as a function of the _ waiting time_. in fact , @xmath97 varies some four orders of magnitude in current experiments,@xcite which should result in a factor 4 variation of @xmath98 .
we thank j. j. ruiz - lorenzo and e. marinari for useful discussions .
we were partly supported by mineco , spain , through the research contract no .
fis2012 - 35719-c02 .
mbj was supported by the fpu program ( mecd , spain ) .
the research leading to these results has received funding from the european research council under the european union s seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement n [ 247328 ] .
the computations were carried out in the gpu - accelerated clusters tianhe-1a ( tianjin , china ) and minotauro ( barcelona , spain ) .
the total amount of time devoted to this project was @xmath99 gpu hours in tianhe-1a and @xmath100 gpu hours in minotauro .
access to tianhe-1a was granted through research contract no .
287746 by the eu - fp7 .
the authors thankfully acknowledge the computer resources , technical expertise and assistance provided by the staff at the _ national supercomputing center - tianjin _ and at the _ red espaola de supercomputacin barcelona supercomputing center_.
the appendix is structured as follows .
the specific algorithms that we have used are explained in sect .
[ sec - app : algorithms ] with no reference to their implementation
. however , implementation _ is _ crucial : our simulations are so demanding that we have used special hardware described in sect .
[ sec - app : hardware ] .
this special hardware speeds up the simulations thanks to parallelization , so in sect .
[ sec - app : parallelization ] we give some brief details about it . finally , we address in sect .
[ sec - app : prng ] some issues regarding the generation of pseudo - random numbers .
heat - bath by itself would provide correct ( but inefficient ) dynamics .
it actually mimics the natural evolution followed by real spin glasses ( that never reach equilibrium near or below the critical temperature ) . for this reason we enhance it with two more algorithms .
however , heat - bath does play a crucial role , since it is irreducible ( i.e. the full configuration space is reachable , at least in principle ) , at variance with overrelaxation , which keeps the total energy constant , and parallel - tempering , which changes the temperature but not the spin configuration .
crucial to perform the heat - bath and overrelaxation dynamics is a factorization property of the boltzmann weight for the hamiltonian .
the conditional probability - density for spin @xmath101 , given the rest of the spins of the lattice is @xmath102 where @xmath103 is the _ local field _
produced by the lattice nearest - neighbors of spin @xmath101 ( its precise definition is given in footnote ) .
the overrelaxation update is deterministic . given a spin @xmath101 and its local field , we change the spin as much as possible while keeping the energy constant : @xmath104 contrarily to heat - bath , the order in which the spins are updated is important in overrelaxation . accessing the lattice randomly increases the autocorrelation time in a substantial way .
on the other hand , a sequential update generates a microcanonic wave that sweeps the lattice .
the resulting change in the configuration space is significantly larger .
a similar microcanonic wave is generated with other types of deterministic lattice sweeps .
for instance , one could partition the lattice in a checker - board way and first update all spins in the black sublattice , updating the white spins only afterwards .
the combination of heat - bath and overrelaxation has been shown to be effective in the case of isotropic spin glasses @xcite and other models with frustration .
@xcite however , if one is interested on very low temperatures or large systems , parallel tempering is often useful . for each sample
we simulate @xmath105 different copies of the system , each of them at one of the temperatures @xmath106 .
a parallel tempering update consists in proposing , as configuration change , a swap between configurations at neighboring temperatures .
the exchange has the metropolis acceptance .
evidently , the acceptance is higher if the temperatures @xmath107 are closer to each other , since the energy of the configurations will be similar .
notice that exchanging configurations is equivalent to exchange temperatures , so the data transfer is reduced to a single number .
the gpus we used were of the tesla generation , produced by nvidia , with an simd architecture ( single instruction , multiple data ) , @xcite optimized for the parallel processing of large amounts of double precision data
. we had access to tesla m2050 gpus in the _
tianhe-1a _ supercomputer in tianjin , @xcite china , and tesla m2090 gpus on the _ minotauro _ cluster @xcite in barcelona , spain . despite the extremely high performances claimed by nvidia ( e.g. 665 gflops in double precision in the case of the m2090 gpus ) , it is practically impossible to reach that limit , because the major bottleneck does not reside in the computing speed , but in the memory access . yet gpus keep being a valid tool to simulate on spin glasses , as they typically allow the same function to be launched concurrently on thousands of threads .
this is exactly what we need , since we can update simultaneously different replicas , and also non - neighboring spins within the same replica , because the interactions are only between nearest neighbors ( see sect .
[ sec - app : parallelization ] ) .
our update - schemes support two levels of parallelism .
heat - bath and overrelaxation are parallelized within a single lattice . on the other hand ,
parallel tempering concerns @xmath108 independent lattices ( two replicas , see sect . [
sec : obs ] , at @xmath105 temperatures ) .
clearly , spins in different lattices can be updated simultaneously ( between temperature swaps ) . for small system sizes ,
the @xmath109 lattices can be updated efficiently within a single gpu . yet , for @xmath13 we have found it convenient to speed up by employing @xmath105 gpus , each of them simulating two lattices .
our parallelization scheme was not very different from the one described extensively in previous works such as refs . and ,
so we limit ourselves to remark that we used binary couplings in order to be able to store a full coupling in a single byte .
also , due to the fact that the lattice positions were evaluated with bitwise operations , and to our coalesced memory - reading scheme , @xcite our program was mostly efficient when the size of the lattice was a power of two , so we favored simulations on those sizes . for @xmath13 and @xmath58
the relaxation times were too long to be able to thermalize on a single gpu .
therefore , we prepared a code that mixed cuda and mpi , in order to be able to concentrate a major computing capability on a single sample .
we took advantage of the two levels of parallelization that our update algorithms allow .
we used @xmath110 gpus , each updating only two independent lattices with the same couplings , but not necessarily with the same temperature . at the level of the single gpu ,
the way we swept the lattice with heat - bath and overrelaxation was similar to the single - gpu version .
yet , we had to arrange it in order to get the same thread occupancy as in the single - gpu version .
our choice has been to divide the lattice in rows of 8 spins along the @xmath111 axis .
non - neighboring rows were updated at the same time .
a side advantage of this scheme was that we could use for it the same type of coalesced memory reading that we developed for the single - gpu lattice sweeping .
this arrangement resulted in an extremely small overhead when passing from the single to the multiple - gpu algorithm .
we were also favored by other factors .
parallel tempering only requires the exchange with the master of a double precision number .
also , the long correlation times allow to take measurements with low frequency . as a consequence of all this
, we obtained a linear scaling of the computing time with the number of gpus , @xmath105 ( fig .
[ fig : scalinggpu ] ) .
pseudo - random number generators ( prngs ) are a critical issue in the implementation of stochastic algorithms , @xcite but even more in cases like ours , where each of the @xmath112 threads had to carry its own prng , and we had a large number of them acting in parallel on the same lattice .
this became a major problem especially in the simulations with mpi , where a huge number of prngs was concentrated on only two lattices .
it was crucial to guarantee the statistical independence of the @xmath112 pseudo - random sequences .
we consider three different aspects : ( @xmath44 ) the prng that each thread uses , ( @xmath45 ) the initialization of the generators and ( @xmath113 ) our tests on the generators . with the parisi - rapuano sequence , @xcite the @xmath114 pseudo - random number @xmath115
is generated through the following relations : @xmath116 where xor is the exclusive or logic operator , and @xmath117 are 64-bit unsigned integers .
although some pathologies have been found in the 32-bit parisi - rapuano prng,@xcite it looks like its 64-bit version is solid .
@xcite on the other side , we used a 64-bit congruential generator , where the @xmath114 element of the sequence , @xmath118 , was given by:@xcite @xmath119 also this generator is not reliable when used alone.@xcite we have found that problems arise if special care is not devoted to the initialization of the random numbers .
this is particularly important in the case of multiple gpus where @xmath122 threads concurrently update the spins in only two lattices .
we decided to use one seed per node .
this seed was used to initialize a 64-bit congruential prng , eq . .
we employed it to initialize the state vector of 24-bit luescher prng.@xcite the 24-bits words were obtained from three consecutive congruential calls ( we kept the most significant byte from each call ) . as for the luescher generator , we employed the _ full luxury _ version , which is fireproof but slow .
we took the 8 most significant bits from each luescher call to fill up the state vector of the 64-bit prngs in eq .
we were probably excessively cautious , given the high quality of the full - luxury generator , but initialization takes only a small fraction of the total computing time .
we tested with success our random sequences through the whole battery of tests proposed by marsaglia in ref . .
to be sure the sequences were reliable also with concurrent threads , we also generated @xmath112 sequences and tested them _ horizontally _ , i.e. taking first the first number of each sequence , then the second , and so on . finally , it has been pointed out that local schwinger - dyson relations ( see e.g. ref . ) can be useful to assess the quality of prngs.@xcite the relevant identity here is @xmath124 we averaged it over all the sites in the lattice , in order to obtain a more stringent test for the simulations .
of course the limiting factor is in the wide range of _ relaxation times , _ rather than temperatures .
however , relaxation times depend on a variety of implementation - dependent factors ( such as the temperature spacing in the parallel tempering , or the number of overrelaxation sweeps ) .
hence , comparison with other work will be easier in terms of temperatures . independence from microscopic details such as the disorder distribution has been found for spin glasses,@xcite as well as for other disordered systems like the random field ising model,@xcite or disordered ferromagnets ( either site @xcite or bond @xcite diluted ) . a more recent yet unpublished article from the janus collaboration ( ref . ) gives a more precise determination of the critical exponents .
using one or the other does not change significantly our results and conclusions .
janus collaboration : m. baity - jesi , r. a. baos , a. cruz , l. a. fernandez , j. m. gil - narvion , a. gordillo - guerrero , d. iiguez , a. maiorano , f. mantovani , e. marinari , v. martin - mayor , j. monforte - garcia , a. muoz sudupe , d. navarro , g. parisi , s. perez - gaviro , m. pivanti , f. ricci - tersenghi , j. j. ruiz - lorenzo , s. f. schifano , b. seoane , a. tarancon , r. tripiccione , d. yllanes , phys . rev .
b ( in press , see arxiv:1310.2910 ) . some of the points we used for those extrapolations shared some of the data .
for example , the crossing of @xmath6 for @xmath128 , had in common the points from size @xmath129 with the pair @xmath130 .
hence , in the fits we have taken in account the covariance matrix that gave a measure of the anticorrelation between measures that share data . in the phase diagram
we show , the @xmath77 point comes from ref . , where chiral and spin glass transition are assumed to be coupled .
there is disagreement on whether @xmath72 also in the isotropic case .
yet , we do plot it as a single transition because although @xmath9 might be lower than @xmath4 , their best estimates are compatible ( and not distinguishable in the plot ) . | we characterize the phase diagram of anisotropic heisenberg spin glasses , finding both the spin and the chiral glass transition .
we remark the presence of strong finite - size effects in the chiral sector . on the spin glass sector , we find that the universality class is that of ising spin glasses .
our data is compatible with a unique phase transition for the chiral and spin glass sector .
we focus on keeping finite - size effects under control , and we stress that they are important to understand experiments . thanks to large gpu clusters we have been able to thermalize cubic lattices with up to @xmath0 spins , over a vast range of temperatures ( hence , of relaxation times ) . |
the studies related to plasma response at fast electron time scales are becoming a topic of considerable research interest lately .
the topics of laser plasma interaction @xcite , reconnection in electron current layers @xcite , fast z pinches @xcite , fast plasma based switching devices @xcite and also fast ignition physics @xcite primarily involve a study of plasma phenomena occuring at electron response regime @xcite .
one particular simplified fluid model description characterizing the electron dynamics against the background stationary charge neutralizing ions is the electron magnetohydrodynamics ( emhd ) @xcite .
the model has so far been primarily studied in the context of plasma with uniform homogeneous density . in realistic situations
, however , the plasma typically has an inhomogeneous density .
the role of density inhomogeneity was outlined briefly earlier in a paper by kingsep @xcite .
the paper , however , discussed the effect in the absence of any electron inertia related terms . in the present manuscript the emhd model has been generalized for the case of a plasma with inhomogeneous density profile including effects associated with finite electron inertia .
the governing equations for the generalized emhd are shown to reduce to two coupled scalar field evolution in the 2-d limit .
for homogeneous density these equations reduce to the well known form studied extensively earlier @xcite .
the emhd equations for the homogeneous plasma permits a variety of coherent localized solutions which are of a stationary monopolar form or a travelling ( with constant velocity ) dipolar form @xcite .
these solutions are extremely robust and have been observed to form spontaneuosly in simulations @xcite . here , we study the propagation and evolution of such coherent solutions when they encounter an inhomogeneous plasma .
this is done by employing our generalized emhd ( g - emhd ) equations for the numerical evolution of these solutions specified as an initial condition . in the next section
we outline the details of the derivation of the governing equations for the g - emhd model corresponding to the spatially inhomogeneous plasma density .
the 2-d limit of the equation is then obtained and it is shown that the governing equations are a coupled evolution equation for two scalar fields .
these scalar fields represent the component of the magnetic field and the vector potential along the symmetry direction . in section iii
we briefly recapitulate the coherent solutions of the emhd equations and their well known propagation characteristics in the context of a homogeneous plasma . the physical similarity of the dipole structure to a forward moving current pulse accompanied by spatially separated reverse shielding currents is also shown in this section . in section
iv we then show with the help of our numerical simulations how the density inhomogeneity influences the propagation of these structures .
the restrictions on total grid points constrains us to consider a ratio of maximum to minimum plasma density in our simulation domain to be at most a factor of @xmath0 only .
this constraint arises from the need of adequately resolving the plasma skin depth scales in the high density regime . in section
v we present a novel observation illustrating the trapping of dipolar emhd structures in the high density plasma region . the consequence of this novel observation on the physics of fast ignition is also pointed out . finally , in section vi we summarize the salient points of the paper .
in this section we obtain the governing equations for describing dynamical phenomena occuring at fast electron time scales in the presence of plasma density inhomogeneity .
the time scales are fast such that the ion response can be ignored .
however , it is considered to be slower compared to the minimum value of the local plasma frequency @xmath1 and/or @xmath2 ( here @xmath3 is the electron gyrofrequency ) whichever is smaller .
this ensures that the displacement current can be ignored and hence electron density fluctuations are considered to be negligible .
this essentially implies that the electron continuity equation drops out from the evolution .
thus the system of equations which we derive here can be looked upon as the generalized electron magnetohydrodynamics ( g - emhd ) model for inhomogeneous plasma .
the curl of electron momentum equation can be written as @xmath4 \label{curl_emom}\ ] ] where @xmath5 .
the electron velocity @xmath6 from the ampere s law , as the displacement current can be ignored and ions being stationary the current is determined by the electron flow velocity alone .
also , here @xmath7 and @xmath8 and @xmath9 denote magnetic vector potential , magnetic field vector and the plasma density respectively .
we choose to normalize the electron density with the minimum value of the plasma density @xmath10 in the region of interest , and the length by the corresponding skin depth @xmath11 ( @xmath12 ) corresponding to this density .
the magnetic field is normalized by @xmath13 , the typical magnitude of the magnetic field and the time by the corresponding electron gyrofrequency @xmath14 .
the normalized equation can then be written as @xmath15 \label{ncurl_emom1}\ ] ] @xmath16 equation([ncurl_emom1],[ncurl_emom2 ] ) represents the g - emhd model written in normalized fields and variables . for the case
when the variations of the fields are confined in two dimensional @xmath17 plane , the g - emhd model ( eq.([ncurl_emom1],[ncurl_emom2 ] ) can be represented by a coupled set of evolution equations for two scalar fields @xmath18 and @xmath19 .
these two scalar fields define the magnetic field of the system through the following relationship : @xmath20 thus @xmath18 represents the magnetic field component along the symmetry direction @xmath21 and the magnetic field along @xmath22 and @xmath23 directions are @xmath24 and @xmath25 respectively .
the normalized electron velocity , the expression for @xmath26 , etc .
, in terms of these two scalar fields can then be written as @xmath27 + \nabla \left(\frac{1}{n } \right ) \times \left[\hat{z } \times \nabla b - \nabla^2 \psi \hat{z } \right ] - \left[b \hat{z } + \hat{z } \times \nabla \psi \right ] \label{vgfld}\end{aligned}\ ] ] equations ( [ ncurl_emom1],[ncurl_emom2 ] ) ( which describe the evolution of @xmath28 ) may be separated in terms of components @xmath29 and @xmath30 from which we may derive the following coupled set of evolution equations for @xmath18 and @xmath19 fields .
@xmath31 + \hat{z } \times \nabla \psi \cdot \nabla \left ( \frac{\nabla^2 \psi}{n}\right ) = 0 \label{2db}\end{aligned}\ ] ] @xmath32 equations([2db],[2dpsi ] ) represent the generalized emhd equations in two dimensions .
when the plasma density @xmath33 is a constant , the above coupled set of equations reduce to the emhd equations in 2-d .
the sum of magnetic and the kinetic energy of the system is given by the expression @xmath34 dx dy\ ] ] and is an invariant for the above coupled set of g - emhd equations .
however , when the in plane component of the magnetic field is zero , i.e. when @xmath35 , an additional square invariant @xmath36 ^ 2 dx dy\ ] ] also exists for the system . note that for the homogeneous case @xmath37 represents the enstrophy ( space integral of mean square vorticity ) , which is the second invariant for the system .
in the subsequent sections we will discuss the results of the numerical simulation of eqs.([2db],[2dpsi ] ) for several distinct choices of plasma inhomogeneity . equations ( [ 2db ] ) and ( [ 2dpsi ] ) are evolved for the generalized vorticities @xmath38 and @xmath39 using the flux corrected algorithm developed by boris _ et . al _ @xcite .
( here @xmath40 and @xmath41 are used to denote the operators , which upon operating on @xmath18 and @xmath19 produce @xmath42 and @xmath43 respectively . )
the value of the @xmath18 and @xmath19 at each time step are then extracted from the generalized vorticities @xmath42 and @xmath43 respectively by constructing matrices corresponding to the inverse operators @xmath44 and @xmath45 for the chosen spatial inhomogeneity of the plasma density .
it should be noticed here that the dimension of this matrix being extremely large ( @xmath46 , where @xmath47 , where @xmath48 and @xmath49 represent the number of grid points along @xmath22 and @xmath23 directions respectively ) it puts a severe constraint on the resolution . to adequately resolve the skin depth at the maximum plasma density and to simultaneously also have the box dimension longer than several electron skin depths ( corresponding to the minimum value of plasma density ) we were constrained to choose the ratio of maximum to minimum plasma density to be at most a factor of @xmath0 only .
alternatively , to achieve higher resolutions a relaxation algorithm to evaluate @xmath50 and @xmath51 , needs to be implemented @xcite which utilizes the standard solvers for helmholtz equation where the ram requirements would not be so extensive .
we are in the process of implementing this for our future studies on g - emhd model .
the g - emhd equations ( eqs.([2db],[2dpsi ] ) ) reduce to the following form for a homogeneous density @xmath52 .
@xmath53 = [ \psi , \omega_{\psi } ] \label{2db_h}\end{aligned}\ ] ] @xmath54 = 0 \label{2dpsi_h}\end{aligned}\ ] ] here , the uniform plasma density has been chosen for the density normalization .
the symbol @xmath55 $ ] used in the equations represents a poisson bracket between the field @xmath56 and @xmath57 .
it is then clearly evident from the above equations that for the homogeneous system all radially symmetric ( monopolar ) structures are exact stationary solutions of the equations .
furthermore , it has been observed that monopolar structures localized within a spatial extent of electron skin depth are fairly robust and stable and they are spontaneously created during simulations with arbitrary initial configurations .
a collection of monopoles which are spatially separated by a distance larger than the electron skin depth also constitute a stationary configuration of the above set of equation .
this is so as in this case the self term of the poisson bracket vanishes on account of radial symmetry and the cross term between two structures vanishes as there is no spatial overlap amidst them .
it is then interesting to see how these structures behave in the presence of density inhomogeneity .
this is explored by us by numerically simulating the evolution of these structures in a given inhomogeneous density profile .
the results of such studies are presented in the next section . the other interesting class of solutions permitted by eqs.([2db_h],[2dpsi_h ] )
have dipolar form @xcite .
these solutions are not stationary but have a steady translational velocity with respect to the static ion frame of the emhd model .
the dipolar solutions can be obtained by seeking stationarity in a moving frame @xmath58 ( assuming the translational velocity to be along the @xmath23 coordinate ) .
equation([2dpsi_h ] ) can then be written as the vanishing of the poisson bracket @xmath59 = 0 $ ] .
this implies that @xmath60 , where @xmath61 is a function of @xmath62 . using this expression for @xmath43 as well as the stationarity criteria in the @xmath63 frame , eq.([2db_h ] ) can be expressed as @xmath64 = 0;\ ] ] here @xmath65 indicates a differential with respect to the argument @xmath66 of the function @xmath61 .
this suggests that @xmath67 ( @xmath68 being another function of @xmath62 ) .
thus a travelling solution can be obtained by seeking solutions of the following coupled set : @xmath69 the general solutions would correspond to any choice of the functions @xmath68 and @xmath61 . however , analytical form of the dipole solutions are typically obtained by choosing the vorticity source functions @xmath68 and @xmath61 as linear functions of their argument @xmath70 in the inner spatial region of radii @xmath71 around the centre of the structure .
the differential equation can be separated in @xmath72 and @xmath73 coordinates in the two dimensional @xmath74 plane .
it can be shown that an explicit appearance of @xmath75 ( as a coefficient of @xmath76 ) produces a @xmath77 dependence and hence a dipolar form of the solution .
furthermore for the radial part in this inner region , the solutions can be represented in terms of a bessel function of the first kind denoted by @xmath78 . to achieve localization the vorticity source functions @xmath68 and @xmath61
are chosen to be zero in the outer spatial region @xmath79 .
the solutions for the fields @xmath18 and @xmath19 in the outer region are thus modified bessel function @xmath80 of the radial coordinate .
the matching of the field and its derivative in the inner and outer region yields a localized solution .
it might appear that the specific linear choice made for the inner region is too restrictive and may in general not correspond to reality .
however , it has been shown recently ( for the system with @xmath35 ) that a linear choice of @xmath68 is consistent with a self organization paradigm based on the enstrophy minimization subject to the constancy of the energy of the system @xcite .
the approach of the system to such a self organized state has been confirmed by numerical simulations which show the spontaneous formations of structures from an arbitrary initial condition satisfying eq.([dipole ] ) for a linear functional form of @xmath68 @xcite .
similar studies in the presence of finite @xmath19 field are underway and the results will be reported elsewhere . in the present manuscript
we report the propagation characteristics of the dipolar solutions for both cases ( viz . @xmath35 and @xmath19 finite , with @xmath68 and @xmath61 both as linear functions of the argument @xmath62 ) as they encounter the inhomogeneous plasma density profile . it should be noted that a monopolar structure is like a circularly rotating current similar to that of a solenoid .
the propagating dipoles on the other hand can be viewed as an electron current pulse translating in space .
the central spatial region within the positive and the negative peaks of the magnetic structure carries electron current in the direction of the propagation of the dipole , whereas the outer edge region carries a reverse current as shown in the schematic plot of fig.1 .
thus the dipole structure can be viewed as a current pulse containing a spatially separated forward and the reverse shielding electron current pulse .
the current configuration of a dipole is thus quite similar to the one encountered during the ignition phase of the fast ignition experiments .
the forward energetic electrons generated at the critical density surface moves towards the dense target core .
the background plasma in this case provides for the return shielding current .
the two currents get spatially separated through weibel instability and form current channels in which the inner region carries the forward current and the outer region the return shielding current .
the numerical study of the propagation of the dipole magnetic structure through an inhomogeneous plasma and its behaviour in the high density region is therefore an important issue for investigation . in the subsequent sections
the result of such a simulation will be presented .
in this section we study the propagation of both monopolar and dipolar solutions through an inhomogeneous plasma .
the plasma density inhomogeneity is chosen to have the following spatial profile @xmath81 the parameters @xmath82 and @xmath18 were chosen appropriately to define either a density hump in the region @xmath83 or a density cavity .
for the above chosen density profile the constant density contours form straight lines parallel to the @xmath22 axis .
the monopolar structures are the exact stationary equilibrium solutions of the homogeneous emhd equations . in order to study the influence of plasma inhomogeneity on their evolution we place a monopolar structure initially at the boundary region ( around @xmath84 ) , where the plasma density inhomogeneity is typically high .
the subplots of the fig.2 show the evolution of this structure . for this case
we had chosen the simulation box of size @xmath85 and @xmath22 and @xmath23 coordinates range from @xmath86 to @xmath87 .
for the plasma density we had chosen @xmath88 , @xmath89 , @xmath90 and @xmath91 .
the maximum and minimum value of density is therefore @xmath92 and @xmath93 respectively .
the local electron skin depth therefore ranges from @xmath94 .
the high density plasma region here is confined within @xmath95 for all @xmath22 .
the density falls sharply within a length @xmath96 from @xmath0 to unity beyond @xmath97 .
the density scale length at the boundary is typically of the order of the electron skin depth as @xmath91 . from the figure it is clear that the monopolar structure moves transverse to the density gradient .
the monopoles being stationary in the homogeneous plasma the propagation velocity is clearly an artifact of the presence of plasma density inhomogeneity .
the direction as well as the magnitude of the propagation velocity is observed to match with the following drift velocity . @xmath98 which can be obtained from the eq.([2db ] ) upon ignoring electron inertia related terms . for the density profile of eq.([profilea ] ) @xmath33 is a function of @xmath23 alone and we have the propagation along @xmath22 and the magnitude of the velocity is @xmath99 from the subplots of the fig.2 , the value of @xmath100 evaluated by observing the distance propagated by the structure along @xmath22 is @xmath101 which is close to that estimated from the above expression for the electron drift velocity , as @xmath18 typically ranges from @xmath102 to @xmath103 in the monopolar structure and @xmath104 ranges from @xmath105 to @xmath106 over the structure .
this implies that the value of @xmath100 from the expression can be about @xmath107 to @xmath108 .
the observed value lies within this range .
in fact the average of @xmath100 evaluated over the @xmath23 extent of the structure ( through which the structure would translate ) turns out to be very close @xmath109 to the observed velocity .
this clearly indicates that the monopole is essentially propagating with the drift velocity of @xmath110 .
the other density gradient dependent terms arising through the finite electron inertia related terms are typically smaller in magnitude and they generally contribute as a source causing modification of the spatial profile of the magnetic structure .
we next study the behaviour of dipoles as they encounter the region of plasma density inhomogeneity .
since the dipolar structures are known to propagate along their axis , we start our simulation by placing an exact nonlinear dipole structure at some distance away from the density gradient region , i.e. at a location of a uniform low density plasma where @xmath111 at a positive @xmath112 .
the sign of the lobes in this case are chosen in such a way that the dipole propagates towards the high density region .
we observe that the dipolar structure crosses past the inhomogeneous density region to enter the high plasma density region .
the subplots of fig.3 clearly illustrate the penetration of the dipolar structure in the high density plasma region .
( for the plots of this figure we again have @xmath113 , @xmath89 , @xmath114 and @xmath115 .
the box length in this case is @xmath116 . ) in fact we observe that at the inhomogeneous density region the axial translational velocity of the dipole increases considerably .
the two lobes get squeezed towards each other forming a shock like structure in the direction transverse to the density gradient .
this behaviour appears to be in stark contrast to the propagation characteristics of the monopolar structures , which merely show a movement transverse to the density gradient direction .
the observations on dipole propagation can , however , be understood readily . for the dipole structure approaching the high density plasma region ( along decreasing @xmath23 , in fig.3 )
the left lobe corresponds to positive values of @xmath18 whereas the right lobe has negative @xmath18 values .
clearly , when the two lobes of the dipole encounter the density inhomogeneity the left lobe has a drift velocity due to the density inhomogeneity towards right ( positive @xmath22 direction ) whereas the right one drifts towards the negative @xmath22 direction .
this sqeezes the two lobes of the dipoles closer in @xmath22 . as a result
the size of the lobes as well as their separation gets significantly reduced .
this also causes an enhancement in the magnitude of @xmath117 of the two lobes .
the reduced distance between the lobes as well as the enhanced amplitude of @xmath117 results in an increased axial propagation velocity of the dipole .
this accelerates the penetration of the dipolar structure in the high density plasma region .
it should be noted that for the case where the dipole approaches a density cavity the effect is entirely different .
the sign of @xmath118 being opposite , in this case the lobes separate with the drift speed .
the separation results in a reduced axial velocity of the dipole , which ultimately dimishes to zero as the separation distance between the two lobes exceeds the electron skin depth distance .
the two lobes then separately move as monopolar structures transverse to the density gradient direction .
thus , the dipole is unable to penetrate the region of lower plasma density .
our simulations indeed show this effect as the subplots of fig.4 clearly illustrate . here
the central region @xmath119 corresponds to a low density plasma region with @xmath120 .
here @xmath121 , @xmath122 and other parameters are same as that of fig.3 .
let us now study in detail the behaviour of the dipole as it enters the high density plasma region . though the shape of the dipole is considerably distorted while it traverses the inhomogeneous plasma region , but once it is inside the high density homogeneous plasma region it regains the familiar dipolar form .
the scale length of the dipole , in the high density region changes by the same factor as the ratio of the skin depth of the high and low density regions .
for instance the initial dipole was chosen to have @xmath123 and at @xmath124 when it is completely inside the high density region the value of @xmath125 ( a reduction by a factor of approximately @xmath126 ) .
we thus observe that the dipolar structures are fairly robust . even after encountering a strong density inhomogeneity , once in the region of homogeneous plasma
they adjust smoothly to the new value of the density . in the next section
we illustrate an interesting consequence of the above observations , namely that of novel trapping behaviour of emhd dipoles in a high density plasma region .
the propagation characteristics of the dipolar structure discussed in the preceding section has an interesting consequence .
it suggests that the emhd magnetic structures of dipolar form can enter a high density plasma region .
however , once inside a high density plasma region they would remain trapped there .
we choose a density profile @xmath127 which has a high density region with a finite transverse extent as well .
a circular region in the @xmath17 plane is chosen to have a high density of the plasma .
the functional form of the plasma density is @xmath128 the choice of parameters for simulation with this density profile is @xmath129 . for this particular density profile a dipole
is placed with its centre on the line @xmath130 at the positive value of @xmath131 .
the axis of the dipole is parallel to the @xmath23 axis as can be seen from the subplot at @xmath132 of fig.5 .
the dipole velocity is directed along the negative @xmath23 axis so that it approaches the high density plasma region .
it can be seen from the subsequent subplots that due to the individual density inhomogeneity related drift velocity of the two lobes , the two lobes of the dipole approach each other .
this enhances the axial dipolar velocity and the dipole structure enters the high density region .
once inside the homogeneous high density region it translates along its axis which is along the diameter of the circular high density region . upon reaching the other end
the dipole again encounters the inhomogeneous plasma density region .
however , the direction of the density gradient is now opposite to the one that it encountered while entering the high density region .
thus in this region the two lobes of the dipole separate from each other . as the separation between the lobes exceeds the skin depth distance the lobes act like individual monopolar structures and move transverse to the density gradient . in this case
the density gradient being along the radial direction , the two structures move along the perimeter of the circle .
they thus again come in close contact at the topmost point of the circle from where they had entered the high density region . at this place they again form a dipolar structure and translate along the diameter of the high density region .
the simulations clearly show the repetition of this cycle .
it is thus clear that the dipole structure remains trapped inside the high density plasma region .
these numerical results showing penetration of the dipolar structures inside the high density region and its trapping within the high density region suggests an important implication for the problem of fast ignition @xcite . in this scheme one
relies on energetic electrons generated at the critical density surface by the ignitor laser pulse to penetrate the high density pre - compressed target core and deposit its energy there for the creation of hot spot . here
, we have shown that a current pulse in the form of a dipolar structure ( similar in form to a forward moving electron current and a spatially separated reverse shielding current ) can easily penetrate a high density plasma region .
it has also been observed that once inside the compressed target the structure is unable to come out of the compressed high density region .
this suggests that there will be ample time for them to dump their energy in the high density compressed region of the core of the fusion target plasma .
another interesting thing to note is the fact that as the dipole enters the high density region the squeezing of its lobes results in a shock formation along the direction transverse to the density gradient .
such a shock formation is an artifact of the thermoelectric like source term @xmath133 in the magnetic field evolution equation where the self consistent magnetic field @xmath134 , itself acts as a pressure term .
this nonlinearity is responsible for the shock formation .
the sheared current layer in the magnetic shock structure would lead to the kelvin helmholtz ( kh ) like instability @xcite .
the instability converts the electron flow energy into fine scale vortices . in the three dimensional case such vortex flows cascade energy towards finer scales @xcite and
eventually dissipate into heat by electron landau damping effects .
we would show through simulations as well as analytical modelling in a subsequent publication that the shoch width adjusts in such a fashion that the energy dissipation in such a shock structure is essentially independent of the value of the dissipation mechanism .
this again is very important in the context of fast ignition scheme as it suggests that the energy dissipation would occur no matter how small is the value of the classical collisional damping process in the dense target core .
furthermore , in some recent fast ignition experiments it has been observed that if a metallic wire is attached to the gold cone where the ignitor laser pulse is incident , the energetic electrons generated at the critical density surface choose a preferential path provided by the ionized wire to traverse from the low density plasma corona towards the compressed target core @xcite .
this result is important as it indicates that the energetic electrons can be guided in the low density plasma corona region to reach the dense target core .
an understanding of this pheneomena has , however , been lacking .
we feel that this can again be understood on the basis of the novel trapping mechanism outlined above .
the ionized wire in this case forms a high density plasma path amidst the low density coronal plasma .
the high density region can then trap the current pulse structure formed by the combination of the forward energetic electrons and the background reverse shielding current .
a generalized electron magnetohydrodynamic ( g - emhd ) model has been introduced here which accounts for the fluid model description of fast electron time scale phenomena in an inhomogeneous plasma .
the electron density fluctuations have , however , been ignored by restricting to times scales which are slower than electron plasma period @xmath135 and/or @xmath136 whichever is smaller .
this is the same condition which is invoked for the derivation of the emhd model and justfies ignoring the displacement current . a reduced g - emhd model for 2-d case has also been obtained .
the model is represented in terms of a coupled set of evolution equation for two scalar fields .
the two scalar fields correspond to the component of the magnetic field and the vector potential along the symmetry direction .
the g - emhd equations in 2-d were used to study numerically the evolution and propagation of nonlinear coherent solutions of the emhd equations in the presence of density inhomogeneity .
the two varieties of coherent solutions ( viz . , the stationary monopolar solutions and the travelling dipolar solutions ) show interesting novel aspect of propagation as they encounter a region of plasma density inhomogeneity .
the monopolar structures are no longer stationary but are observed to propagate with a diamagnetic drift velocity .
the dipoles on the other hand display a very novel behaviour .
they show penetration inside the high density plasma region from a lower density side . however , once inside the higher density region they get trapped inside this region . it has been shown that this behaviour of trapping can have interesting favourable implications in the context of the energetic electron propagation and their stopping inside the dense target core in the context of fast ignition experiments .
* acknowledgement : * this work was financially supported by dae - brns sanction no .
: 2005/21/7-brns/2454 .
sky and ad would like to thank sarveshwar sharma for his help in the preparation of some of the figures .
99 w. l. kruer , _ physics of laser plasma interaction _ ( addison - wesley , 1988 ) .
s. v. bulanov , f. pegoraro , a. s. sakharov , phys , fluids * b4 * , 2499 ( 1992 ) .
j. f. drake , d. biskamp , a zieler , geophysical research letters * 24 * , 2921 ( 1997 ) .
d. d. ryutov , m. s. derzon and m. k. matzen , revs .
phys . * 72 * , 167 ( 2000 ) and _ references therein _ r. z. sagdeev , in _ reviews of plasma physics _ , edited by m. a. leontovich ( consultants bureau , new york , 1966 ) , vol 4 .
a. s. kingsep , k. v. chukbar and v. v. yankov , in _ reviews of plasma physics _ ( consultants bureau , new york , 1990 ) , vol .
16 and references therein ; m. tabak ,
i hammer , m. e. glinsky , _ et .
plasmas * 1 * , 1626 ( 1994 ) . k. a. tanaka , r. kodama , k. mima _
plasmas * 10 * , 1925 ( 2003 ) .
r. kodama _ et al _ nature * 412 * , 798 ( 2001 ) ; r. kodama _ et al .
_ nature * 418 * , 933 ( 2002 ) , m. h. key , nature * 412 * , 775 ( 2001 ) .
a. das and p. h. diamond , phys
. plasmas * 7 * , 170 ( 2000 ) .
d. biskamp , e. schwarz , j. f. drake , phys .
lett . * 76 * , 1264 ( 1996 ) .
m. b. isichenko and a. m. marnachev , sov .
jetp * 66 * , 702 ( 1987 ) .
a. das , plasma phys .
fusion * 41 * a531 ( 1999 ) .
a. das , phys .
plasmas * 15 * , 022308 ( 2008 ) .
j. p. boris and d. l. book , methods comput .
* 16 * , 76 ( 1976 ) .
b. storey from olin college of engineering , u.s.a .
a. das and p. kaw , phys .
plasmas * 8 * 4518 ( 2001 ) .
n. jain , a. das , p. kaw and s. sengupta , phys .
letts * a 363 * , 125 ( 2007 ) .
r. kodama _ et al .
_ nature * 432 * 1005 ( 2004 ) .
* a schematic plot showing the dipole structure ( subplot(a ) ) represented by constant contour lines in the @xmath17 plane for the magnetic field @xmath18 along the symmetry direction .
the subplot(b ) shows a plot of the magnetic field amplitude as a function of @xmath22 at the mid @xmath23 plane of the dipolar structure . in subplot ( c )
the corresponding electron current is shown .
the central region represents the forward electron current ( electron current along the propagation direction of the dipole ) and the reverse electron current flows at the edges . *
( color online ) the propagation of the monopolar structure ( color contours ) in a inhomogeneous plasma density is depicted by showing the location of the structure at various times in the different subplots of the figure .
the thick black lines represent the plasma density contour . in this case
the plasma density is chosen to be a function of @xmath23 only .
the central @xmath23 region of width @xmath137 corresponds to a high density ( @xmath0 times the density at the edge ( @xmath138 ) .
the monopole structure is seen to propagate transverse to the density gradient . *
( color online ) various stages of the propagation of a dipolar structure through an inhomogeneous desnity plasma has been shown .
the inhomogeneity in plasma density is similar to that of fig.2 in this case .
the figure clearly shows the penetration of the dipole through the plasma density inhomogeneity to enter the high density region .
the lobes of teh dipole structure are squeezed towards each other as they pass thorugh the inhomogeneous region . however
, once inside the high density homogeneous region they again aquire a balanced dipolar form . *
( color online ) in this figure the dipole is shown to approach a density cavity ( lower density plasma region ) .
it can be observed that the dipole is unable to penetrate the lower density plasma .
the two lobes of the dipole get separated transverse to the density gradient direction and subsequently they evolve as separate monopolar structures . *
( color online ) the trapping of the dipolar structure in a high density plasma has been illustrated in this figure .
a high density plasma with a circular profile in the @xmath139 plane represented by the thick black contour lines are depicted on the various subplots .
a dipole structure can be seen to penetrate the high density region . however , once inside the high density region it continues to remain rapped in this region . | the fully three dimensional governing equations in the electron magnetohydrodynamic ( emhd ) regime for a plasma with inhomogeneous density is obtained . these equations in the two dimensional ( 2-d ) limit can be cast in terms of the evolution of two coupled scalar fields . the nonlinear simulations for the two dimensional case
are carried out to understand the propagation of emhd magnetic structures in the presence of inhomogeneity . a novel effect related to trapping of dipolar magnetic structures in the high density plasma region in the emhd regime is observed .
the interpretation of this phenomena as well as its relevance to the problem of hot spot generation in the context of fast ignition is presented . |
among the earliest calculations of astrophysical sources of gravitational radiation is the `` drpp '' calculation @xcite@xcite of radiation emitted when a particle , starting from rest at infinity , falls into a nonspinning black hole . in this work ,
the mass of the particle is treated as a perturbation of the schwarzschild spacetime and the einstein equations , to first order in this perturbation , are organized into the zerilli equation@xcite , a single linear wave equation .
this work was later elaborated by ruffini and ferrari@xcite who considered the head - on plunge of a particle which starts with nonzero inward velocity at spatial infinity .
there is now a renewed interest in such calculations .
the advent of laser interferometric gravity wave detectors@xcite has directed attention at black hole collisions as one of the most plausible , and surely the most fascinating , sources of detectable waves .
but the generation of strong waves is a process without simplifying symmetries and one which involves strong field interactions .
the challenge of finding energies and waveforms generated in these processes helped to spur the recent effort in numerical relativity and , in particular , in the binary black hole supercomputing grand challenge@xcite . because of the difficulty of direct supercomputer solutions , approximate methods , as checks and guides , are very valuable .
the `` particle limit '' in which the mass of one of the holes is very small , is in this category , and models of holes in binary orbits are a fundamental tool .
we are interested here in using a particle model to help in the understanding of numerical relativity results for collisions of holes . to date
those computations have been limited to head on , axisymmetric collisions@xcite .
the codes have also been limited in the time for which they can evolve solutions .
infall from very large distances can not at present be evolved ; computations start from finite separation of the holes , so direct comparison with the drpp results can not be made .
we present here an extension of the drpp calculation to the case of infall from finite radius .
the extension is not at all straightforward .
if the particle starts at infinity , the initial value data can be taken as zero , so that the spacetime is unperturbed schwarzschild until the gravitational influence of the particle is felt .
if the particle starts from finite radius , the specification of compatible initial value data can not be avoided .
understanding the influence of the initial data is , in fact , a major motivation for the present work .
supercomputer evolution will be limited , for the foreseeable future , to evolving the interaction of holes for short times .
the starting point for these evolutions will be initial data from some approximation method ( e.g. , a post - newtonian expansion ) .
the way to make the bridge from the approximation to the strong field realm of supercomputing requires that we understand what the crucial features are of the initial value data .
it seems to us that the particle limit provides a very useful probe of this issue .
the remainder of the paper is organized as follows . in sec .
ii we start by presenting the basic mathematical approach . like the drpp calculation ,
our approach uses frequency components .
a laplace transform converts the linearized einstein equation for each @xmath6-pole , from a partial differential equation in schwarzschild coordinates @xmath7 to a set of ordinary differential equations in @xmath8 .
the laplace transform treats the initial data mathematically as if it were a source , and helps with insight into the relation of the initial value data and the stress - energy source .
two points in particular should be noted in sec .
ii . the first concerns the linearized wave equation .
the equation presented by zerilli was formulated in the regge - wheeler gauge@xcite , a specific first order coordinate choice .
the zerilli equation was presented through a fourier transform and has no direct equivalent as a partial differential equation .
this is not a fundamental obstacle , but it does complicate the relationship of the zerilli wave function and the initial data . to avoid these problems we use the moncrief formulation of the problem@xcite .
moncrief s wave function is constructed explicitly from initial data and is equivalent to the zerilli wave function except for source - like terms . the second point to notice
has to do with conventions .
a variety of conventions has been used for normalizing the zerilli ( or moncrief ) functions and their transforms . in sec .
ii , we present a definition of all our conventions and their relationship to those used elsewhere .
aspects of the numerical approach , a green function integral , are discussed in sec .
a major issue is the need for highly accurate evaluation of the green function integrals in order to preclude numerical errors from giving rise to an erroneous initial burst of radiation .
numerical results are presented in sec .
iv for energy , spectra , and waveforms .
plots are given for infall starting from rest at radii ranging from @xmath9 to @xmath10 .
these are combined with results for infall from infinity to give a complete picture of the dependence of the radiation on the initial physical conditions .
the numerical results are discussed in sec .
v and are compared with predictions based on other results and approximations , especially those of the close - limit approximation . throughout the rest of the paper we use @xmath11 units , the signature @xmath12 , and other conventions of the misner , thorne and wheeler textbook@xcite .
for the straight - line plunge of a particle into a nonspinning hole the perturbations of the schwarzschild spacetime are all even - parity .
we take the particle to be moving along the @xmath13 axis , so that the perturbations are not dependent on the azimuthal angle @xmath14 , to describe the perturbations we use the notation of regge and wheeler@xcite , here specialized to the axisymmetric @xmath6-pole case ( @xmath15 ) : @xmath16 where @xmath17 is the unperturbed line element for a schwarzschild spacetime of mass @xmath2 , where @xmath18 and @xmath19 are functions of @xmath7 , and where @xmath20 are the @xmath15 spherical harmonics .
we then follow the prescription given by moncrief@xcite in his eqs .
( 5.8 ) ( 5.12 ) and ( 5.27 ) to arrive at a wave function . rather than choose the normalization of moncrief s `` @xmath21 '' we choose a normalization closely related to zerillis@xcite wave function .
in the regge - wheeler@xcite gauge ( @xmath22 ) this `` moncrief - zerilli '' function is @xmath23\ ] ] where we have used zerilli s notation @xmath24 we now give the relationship of this variable , which we shall use throughout the paper , to the following wavefunctions : ( i ) @xmath21 appearing in moncriefs@xcite eq .
( 5.27 ) ( ii ) @xmath25 and @xmath26 appearing in ref .
@xcite . for @xmath27 the wave function appearing as @xmath28 in eq .
( ii-31 ) of cunningham , price and moncrief@xcite , and @xmath29 defined in eq .
( 13 ) of price and pullin@xcite , agree with @xmath25 .
the relations are : @xmath30\nonumber\\ & = 2[(\ell-2)!/(\ell+2)!]\,\psi_{\rm pert}\\ & = \sqrt{2[(\ell-2)!/(\ell+2)!]}\,\psi_{\rm num}\ .\nonumber\end{aligned}\ ] ] zerilli@xcite derives his equation only in the frequency domain , so no direct comparison can be made with functions of @xmath7 .
it is , however , straightforward to use the same steps as zerilli in the regge - wheeler gauge . but confined to the @xmath7 domain .
it turns out that to decouple the even - parity variables in the perturbed einstein equations , one needs to take an extra time derivative .
one can then define the variable @xmath31 by @xmath32 where the dot indicates a derivative with respect to @xmath33 .
the fourier transform of this variable , divided by @xmath34 is the function defined in zerilli s papers . for vacuum perturbations @xmath35 , so the zerilli wave function , in vacuum , can be chosen to agree with our @xmath29 . by computing the stress - energy pseudotensor in a radiation gauge
, we find that radiated power per solid angle is @xmath36 ^ 2 \left ( \cot\theta\frac{\partial}{\partial\theta}y_{\ell0 } -\frac{\partial^2}{\partial\theta^2}y_{\ell0 } \right)^2\ ] ] ( where @xmath37 is retarded time ) , so that the gravitational wave power , integrated over all angles , is @xmath38 ^ 2\ .\ ] ] our wave function @xmath29 satisfies a second - order wave equation with a source term , @xmath39 here @xmath40 is the regge - wheeler@xcite `` tortoise '' coordinate , @xmath41 is the zerilli potential @xmath42 and @xmath43 is the source term .
we are interested in the case in which the source is a point particle , of mass @xmath0 , following a spacetime trajectory given by @xmath44 , and @xmath45 , where @xmath46 is proper time along the particle world line , @xmath47 here @xmath48 is the particle 4-velocity , and the 0-component is @xmath49 , where @xmath50 , the `` energy - at - infinity per unit particle mass , '' is a constant of motion that takes the value @xmath51 for infall from rest at @xmath1 .
the two dimensional delta function @xmath52 gives the angular location of the particle trajectory @xmath53 with the last expression applying for infall along the positive @xmath13 axis .
we now need to write the the einstein ( @xmath54 ) and ricci ( @xmath55 ) tensors , in a tensor spherical harmonic decomposition like that in ( [ rwform ] ) .
the components we shall need are : @xmath56 ( here , and below , we omit @xmath6 indices where the meaning of the symbols such as @xmath57 is clear from context . ) by rearranging the expressions for the perturbed einstein and ricci components , we find that the source term in ( [ rtzerilli ] ) , for an axisymmetric problem , is @xmath58\ .\end{aligned}\ ] ] the einstein equations can now be used to replace the perturbed ricci and einstein components with components of the particle stress energy @xmath59\delta(r - r)/r^2\\ & r22k=\kappa\left[\delta(r - r)/2 u^0 \right]\ , \nonumber\end{aligned}\ ] ] where @xmath60 when these are used in ( [ srt ] ) the result is @xmath61\ .\end{aligned}\ ] ] when the source term is included , the relation between our wave function @xmath29 , and the zerilli wave function of ( [ zerillidef ] ) becomes @xmath62 for comparison with established results of numerical relativity we want initial data representing an initially stationary spacetime , so we take @xmath63 , the initial time derivative of @xmath29 , to be zero . to determine our @xmath64 , the initial value of @xmath29 , we choose an initial 3-geometry with the conformally flat form that is used in numerical relativity@xcite , @xmath65\ , \ ] ] where @xmath66 satisfies a flat - space laplace equation @xmath67 .
we choose @xmath66 to represent a throat of mass @xmath68 on the @xmath13 axis at @xmath69 and a throat of mass @xmath70 at @xmath71 .
the simplest such solution has the form @xmath72 we now identify @xmath73 , and choose @xmath74 so that the dipole moment vanishes for @xmath75 .
we treat the mass ratio @xmath76 as small and keep terms in ( [ blsoln ] ) only to first order in this ratio .
we ignore the nonradiative @xmath77 perturbations and write @xmath66 as @xmath78 where @xmath79 we next change radial variables , from isotropic - like @xmath80 to a schwarzschild - like coordinate @xmath8 , with the transformation @xmath81 when this is used in ( [ conflfrm ] ) , and terms higher order in @xmath70 are omitted , the result is @xmath82\ .\ ] ] from this we can infer that the initial value perturbtations in the regge - wheeler notation of ( [ rwform ] ) are @xmath83 and @xmath84 we must now complete the identification of this conformally flat solution with the point particle solution . clearly the location @xmath85 of the perturbative throat must be set to @xmath86 the mass @xmath70 of the throat must _ not _ be set to the mass @xmath0 of the particle .
the particle mass @xmath0 can be viewed as the `` bare '' mass of the particle , the mass that is measured very close to the location of the particle . on the other hand
, @xmath70 is a formal parameter of the initial value solution ; since the total adm mass for ( [ blsoln ] ) is @xmath87 , gravitational binding energy is included in @xmath70 .
we give here two arguments for the correct relationship .
the first is based on the definition , given by brill and lindquist@xcite , of `` bare '' mass for a solution with the form ( [ blsoln ] ) , a mass exclusive of binding energy . to find an expression for this bare mass they look at the extension of the geometry through the perturbative throat into an asymptotically flat universe ; the bare mass of the perturbative throat is the mass measured at infinity in that universe . from eq .
( 13 ) of ref .
@xcite the bare mass of the perturbative throat ( to first order in @xmath76 ) is @xmath88 we identify the bare mass with the particle mass @xmath0 and conclude @xmath89 an independent way of finding the relationship is to look at the perturbed hamiltonian constraint .
for a conformally flat ( @xmath90 ) initial 3-geometry [ equivalent to eq .
( c7a ) of ref .
@xcite ] this gives us @xmath91 by integrating across the discontinuity , and by using ( [ fform]),([eq : rdef ] ) , and ( [ initk ] ) , we find @xmath92 which gives us the same relationship as in ( [ massreln ] ) .
this second derivation of the mass relationship is instructive .
if we choose a conformally flat initial geometry then all the information about the perturbations of that geometry is contained in the single function @xmath93 . but
( [ constraint ] ) completely fixes @xmath93 , so the initial geometry has no freedom .
( alternatively : to choose different initial data we would have to introduce nonconformally - flat terms . )
this is to be contrasted with the case of nonperturbative throats , where a variety of choices is possible for conformally flat initial data .
we now define the laplace transform @xmath94 of @xmath29 to be @xmath95 we take @xmath29 to vanish for @xmath96 , which means that @xmath97 must be analytic in the upper half of the complex @xmath98 plane . at large @xmath8 , for outgoing waves , @xmath99 is a function only of @xmath100 so @xmath94 takes the form @xmath101 we call @xmath102 the amplitude of the outgoing radiation .
since the outgoing radiation @xmath99 is pure real , the amplitude satisfies the crossing relation @xmath103 .
the waveform for outgoing radiation , as a function of retarded time @xmath104 can be found by the inverse transform @xmath105\ .\ ] ] for a particle falling in from infinity the waveform extends to @xmath106 , and there is no initial data to deal with . in this case
@xmath107 is to be interpreted as a fourier , rather than laplace , transform .
if the particle has @xmath108 , then the waveform does not vanish at @xmath106 , and the transform exists only for @xmath98 in the lower half plane , and will have a pole at @xmath109 . for the inverse transform ( the first integral in ( [ inverse ] ) the contour should be interpreted as going below the real @xmath98 axis , so the integral is the equivalent of the cauchy principal value plus half the contribution of the residue of the pole at @xmath109 , @xmath110+\frac{i}{2}\lim_{\omega\rightarrow0}\left\ { \omega a(\omega ) \right\}\ .\ ] ] from parseval s theorem and ( [ power ] ) , we have that the radiated energy at scri+ is @xmath111 the energy spectrum is therefore given by @xmath112 we now multiply ( [ rtzerilli ] ) by @xmath113 , integrate from @xmath114 to @xmath115 , and integrate by parts in the integral involving @xmath116 .
the result is an ordinary differential equation for @xmath94 : @xmath117\psi=-\dot{\psi}_0(r ) + i\omega\psi_0(r)+s(r,\omega)\ , \ ] ] where @xmath118 is the initial value of @xmath119 , and @xmath120 is the initial value of @xmath121 , and the source term @xmath122 is defined as @xmath123 for the particle source we substitute ( [ rtsource ] ) into ( [ srctrans ] ) , to get @xmath124_{t(r)}\ .\end{aligned}\ ] ]
the subscript `` @xmath125 '' indicates that functions of time , such as @xmath126 are to be evaluated at the value of @xmath127 where @xmath128 with @xmath45 .
for infall from rest at radius @xmath1 , this gives @xmath33 as a function of @xmath8 through @xmath129\nonumber\\ & & + 2\,{\rm arctanh}\left[\varepsilon_0^{-1 } \sqrt{{2m\over r}-{2m\over r_0}}\right]\ .\end{aligned}\ ] ] the result in ( [ wrsource ] ) for @xmath130 , however , is not valid for @xmath131 . when @xmath131 the delta functions in ( [ rtsource ] ) , as functions of @xmath33 , have vanishing arguments at @xmath114 , the endpoint of @xmath33 integration in ( [ srctrans ] ) , so the integration is not well defined .
the way to deal with this will be explained in the next section .
we start by rewriting ( [ basiceq ] ) in the form @xmath132\psi = s_{\rm tot}(r,\omega)\ , \ ] ] where @xmath133 is the complete right hand side of ( [ basiceq ] ) , including both the stress energy term and the initial value terms , @xmath134 this equation is to be solved for the boundary conditions of ingoing waves at the horizon , and outgoing waves at spatial infinity : @xmath135 for @xmath136 , and @xmath137 for @xmath138 .
the green function solution is found in the usual way .
we define @xmath139 and @xmath140 as the homogeneous solutions of ( [ totaleq ] ) with asymptotic forms @xmath141 we define the wronskian of the homogeneous solutions , an @xmath142-independent constant , to be @xmath143 with the above definitions , the green function solution is written @xmath144\ .\ ] ] in the limit of large @xmath142 this gives us @xmath145 the first two terms in ( [ termsagain ] ) are straightforward to integrate in ( [ ampgreen ] ) , but the stress energy source term @xmath130 can not be evaluated at @xmath1 . [
the expression in ( [ wrsource ] ) formally diverges as @xmath146 and hence can not be used in ( [ ampgreen ] ) . ] to make sense of this we start by writing the source in ( [ rtsource ] ) as @xmath147))+g(r , t)\delta(r - r[t]))\ , \ ] ] where @xmath148 and @xmath19 contain no delta functions .
the troublesome part of the green function integral can then be written as @xmath149 where @xmath150 is the integral involving @xmath19 and @xmath151 involves @xmath148 .
the first of these is @xmath152 in the final integral , the factor @xmath153 diverges but is integrable . a similar set of transformations is now applied to @xmath151 , @xmath154_{r = r(t ) } \\ & = & \int_{-\infty}^{r_0^*}e^{i\omega t(r)}\left[\left(1-\frac{2m}{r}\right ) \frac{\partial}{\partial r}\left(\frac{f}{1 - 2m / r } \right)y_l(r^*,\omega)+\left(\frac{f}{1 - 2m /
r } \right ) \frac{\partial}{\partial r}\left(y_l(r^*,\omega)\right ) \right]\frac{dr^*}{\dot{r}}\ .\nonumber\end{aligned}\ ] ] to evaluate these explicitly we need the fact that for free fall from rest at @xmath1 @xmath155 we can now use the explicit expressions for @xmath156 from ( [ rtsource ] ) to write @xmath157\ .\end{aligned}\ ] ] from ( [ ampgreen ] ) , ( [ termsagain ] ) , and the definitions of @xmath158 , we have that @xmath159\ .\ ] ] since we are considering an initially stationary problem we have @xmath160 .
the initial data for @xmath29 comes from putting ( [ initk ] ) in ( [ psidef ] ) : @xmath161\ , \hspace*{.4in}\nonumber\end{aligned}\ ] ] where the upper expressions apply in the case @xmath162 , and the lower for @xmath163 .
it should be noted that @xmath64 approaches a nonzero constant , and @xmath164 , as @xmath165 , so the integral over @xmath64 , in ( [ sofar ] ) is improper at large negative @xmath142 .
we must recall that we are really computing @xmath102 in the upper half of the complex @xmath98 plane . to deal with this
computationally we recast the integral in ( [ sofar ] ) into the form @xmath166 where @xmath167 is the limit of @xmath64 at @xmath168 .
the value of @xmath169 must be chosen such that there is negligible variation of @xmath29 between @xmath169 and the horizon .
the first step in the solution is to determine the wronskian in ( [ wronsk ] ) .
we denote the form of @xmath170 at large @xmath8 by @xmath171 we find @xmath172 by solving ( [ totaleq ] ) with the righthand side set to zero , and with the starting condition @xmath173 imposed at a large negative value of @xmath142 .
a fourth - order runge - kutta routine is used to integrate @xmath170 out to large values of @xmath142 where it is matched to approximate forms of the asymptotic solution . in practice ,
good accuracy was difficult to achieve with the asymptotic form in ( [ asympt ] ) and asymptotic solutions one order higher in @xmath174 were used . from @xmath175 ,
the wronskian follows immediately : @xmath176 with @xmath175 in hand , with @xmath160 , and with the substitution in ( [ trick ] ) , the problem consists of computing @xmath177\ , \ ] ] where @xmath158 is the integral given in ( [ i1andi2 ] ) and @xmath64 is given in ( [ psi0 ] ) .
a numerical solution for @xmath107 is found by using a fourth - order runge - kutta routine to solve for @xmath170 and @xmath178 , and the integral in ( [ finalcomp ] ) is done by simpson s rule . from the solution for @xmath102 the energy spectrum is computed with ( [ spectrum ] ) , and the waveform from ( [ inverse ] ) .
the numerical solution used a routine to find the `` particle contribution , '' @xmath158 , and one for the `` initial value contribution , '' the integral over @xmath64 in ( [ finalcomp ] ) . for both routines ,
second - order convergence was found and richardson extrapolation was used .
the step size in the runge - kutta and integration routines were halved until the richardson extrapolate agreed with that from the next larger grid within a preset error limit .
the initial value contribution could be usually be found within an error of 0.2% , while the particle contributions required an error preset of 0.5% . for most values of @xmath179 , these precision requirements were easily met .
the exception was a relatively small number of points at which the real and imaginary parts of a contribution differed by more than an order of magnitude . in this case
it was difficult to get high accuracy in the smaller part .
an estimate of the error in our results is complicated by the fact that the physically important results are a superposition of the particle and initial value contributions , and significant cancellations occur in this superposition .
these cancellations , in principle , mean that the error may be much larger than the small relative error in each contribution . to arrive at an estimate of the error in our determinations of the radiated energy we have recomputed the energy for four trials , in which @xmath180 was added to the particle contribution , and then to the initial value contribution .
the results for energies and errors are given in table i. we see that for many cases the computed energy is not highly sensitive to the cancellation ; the estimated error of around 1% is just what we would expect in the square ( energy ) of a quantity ( amplitude ) with an error of 0.5% .
for some cases , however , especially those with higher @xmath6 , there is a significant magnification of error .
we emphasize that the error estimates given in table i , are extremely conservative . in arriving at them
we have used the maximum 0.5% error applied to all values of @xmath98 , whereas this maximum error actually applies only to a small subset of the points . the smoothness of ( most ) waveforms reported in the next section , and the consistent variation of results with changing @xmath1 , is evidence that the actual errors are rather smaller than those reported in table i. in addition to the infall from finite @xmath1 we also have computed spectra and waveforms for infall from infinity .
we characterize these cases with the same parameter @xmath181 we use for infall from @xmath1 . here
it has the value of the lorentz @xmath182-factor @xmath183 for a particle with velocity @xmath184 at infinity .
for computation of infall from infinity the above computational scheme is modified only in the following ways : ( i ) the form of @xmath125 in ( [ time ] ) must be changed to @xmath185\nonumber\\ & & + \ln\left[{(2 \varepsilon_0 ^ 2 - 1)\left ( { r/2 m } \right ) + 1 + 2 \varepsilon_0\left ( { r/2 m } \right ) \sqrt { \varepsilon_0 ^ 2 - 1 + { 2m/ r } } \over\left ( { r/2 m } \right)-1}\right]\ .\end{aligned}\ ] ] ( ii ) the initial value contributions in ( [ finalcomp ] ) must be omitted .
( iii ) the limit of integration in ( [ i1andi2 ] ) must be changed from @xmath186 to @xmath115 .
the computed energy for these @xmath187 cases are given in table ii .
since there is only a particle contribution in these cases , there is no issue of cancellation affecting the errors .
the errors in the energy are , in fact , primarily due to the cutoff in the solution at a finite radius ( with an analytic addition to represent the source contribution to infinity ) .
error estimates were made by varying the cutoff radius , and were found to be around 1% for all @xmath187 cases .
results for quadrupole waveforms and spectra naturally divide themselves into three ranges , small @xmath188 ( less than @xmath189 ) , moderate @xmath188 ( from @xmath189 to @xmath190 ) and large @xmath188 .
waveforms are given as functions of retarded time @xmath104 .
for small @xmath1 , as shown in fig .
[ smallr0]a , the shape of the waveform is that of simple quasinormal ringing .
this shape is the same for all @xmath1 and the single example shown suffices for all small @xmath1 . since the waveforms have the same shape , the energy spectra , shown in fig .
[ smallr0]b , c , also have the same shape , changing only in magnitude as @xmath1 increases .
as @xmath1 increases further , the early negative excursion of the waveform begins to broaden , and the spectrum shifts slightly .
[ specshapes ] shows the @xmath191 spectrum along with the spectra for the two limiting cases , the drpp infall from rest at @xmath192 , and the close - limit spectrum @xmath193 .
( the latter is normalized to have the same energy as the @xmath194 spectrum . )
the @xmath191 spectrum has an appearance that interpolates between the two limits , as might be expected . as @xmath1 increases further
, however , changes in the spectrum develop that might not be expected . as shown in fig .
[ medr0 ] , the simple spectrum for small @xmath1 develops a secondary peak and the secondary peak grows with @xmath1 . as @xmath1 continues to increase , the initial shape of the waveform becomes a very broad depression extending from the moment infall begins to the beginning of quasinormal ringing .
this is illustrated in the waveform in fig .
[ hir0]a , for @xmath195 .
the start of infall , at @xmath196 corresponds to @xmath197 , and it is at this value of @xmath37 that the waveform begins to take on nonzero values . the initial part of the waveform , then , represents the gravitational bremsstrahlung from the early nonrelativistic part of the particle motion .
the small wiggles around @xmath198 are a numerical artifact due to imperfect cancellation of contributions from the initial value and particle parts of the source .
( to verify this we changed the initial value contribution by @xmath199 and found the change in the initial wiggles to be much greater than in other features of the waveform ) .
for comparison , a drpp waveform , for infall from infinity , is also shown in fig .
( for infall from infinity , of course , the zero of time can not be set to the beginning of infall . the time was arbitrarily shifted for the drpp curve . )
this waveform has a similar ringing pattern as the infall from finite radius , but lacks the initial waveform depression .
the end of the initial waveform depression is , roughly , the beginning of of quasinormal ringing , as can be seen in fig .
the generation of quasinormal ringing@xcite is associated with the peak of the potential in ( [ zpotential ] ) .
the time @xmath200 at which the particle reaches the peak , at around @xmath201 is at retarded time @xmath202 .
this is consistent with fig.[hir0]a , which shows ringing beginning somewhere around this value of @xmath37 .
it is interesting to compare the waveforms for a particle falling from rest , to the waveform of a particle on a time - symmetric geodesic trajectory , a particle that long in the past was moving radially outward just outside a hole , that reaches a certain maximum radius @xmath1 at time @xmath114 , and that subsequently falls into the hole .
the analysis of this case requires only a simple modification of ( [ sofar ] ) ; the initial value terms are omitted and ( due to time symmetry of the source ) the complex conjugate of @xmath203 is added to @xmath158 .
the resulting waveform is shown in fig .
[ poisson]a , and has also been given in ref .
that waveform shows two periods of quasinormal ringing , an early one excited when the particle goes outward through the region around the potential peak , and a later one due to motion inward through the peak .
this later period of ringing , and in fact all features of the waveform generated after @xmath204 agree very closely with the waveform for the particle falling from rest . because the earlier ringing has a higher amplitude , the energy spectrum for the symmetric trajectory , shown in fig .
[ poisson]b , is dominated by this early ringing , and has very large amplitude . in the figure
it is seen to be much larger than the drpp spectrum , which ( see fig .
[ hir0 ] ) approximates the spectrum for infall from @xmath205 .
since the spectra for infall from both small @xmath1 and from @xmath115 have a single peak ( see , e.g. , fig .
[ specshapes ] ) it is interesting that for intermediate values of @xmath1 the spectra are characterized by a row of more - or - less evenly spaced bumps .
as @xmath1 grows , the number of bumps increases , and the bumps decrease in spacing and in height .
the origin of these modulations can be understood by considering a simple example : suppose a waveform consists of nothing but a single period of quasinormal ringing with the transform @xmath107 .
the transform of the same waveform shifted later in time by @xmath206 would be @xmath207 .
a wave consisting of _ two _ periods of ringing , the original and the later one , would then have a transform @xmath208 , and hence an energy spectrum @xmath209 the combined spectrum would have the shape of the single waveform spectrum modulated on a frequency scale @xmath210 . if the two waveforms were not identical , we would expect modulation of the spectrum , but less than 100% modulation .
this explanation can be tested on time - symmetric motion , where there are two well separated periods of ringing in in fig .
[ poisson]a .
the amplitudes of the ringing are different , of course , so we should nt expect 100% modulation of the spectrum and , indeed , the modulations in fig .
[ poisson]b , are not 100% .
the time shift between the first and second ringing periods is on the order of @xmath211 ( roughly the time it takes for the particle to rise up from @xmath212 to @xmath205 and fall back to @xmath212 ) .
this suggests that the spacing of the spectral bumps should be @xmath213=0.84/2m$ ] , which is in good agreement with what is seen in fig .
[ poisson]b .
the application to a nonsymmetric waveform , like that in fig .
[ hir0]a , is less obvious .
the spacing of bumps @xmath214 in fig .
[ hir0]c suggests that the time shift is on the order of @xmath215 .
this , presumably , represents the time between the descent of the waveform at @xmath216 and the start of ringing at @xmath217 .
the interpretation works as well as a predictor of the bump spacing for other values of @xmath1 , and seems clearly to be qualitatively correct , and to explain the progression of the spectra .
as @xmath1 approaches the horizon , the time between the initial moment and the onset of ringing goes to zero , so the spacing of bumps is infinite and there are no modulations of the spectrum .
we get the single - humped close - limit spectrum . as @xmath1 grows ,
the time between the initial motion and the start of ringing increases , so the spacing of the bumps gets smaller , and hence more bumps appear .
as @xmath1 is becoming larger , however , the early waveform is becoming less dramatic ( the initial depression is decreasing in amplitude ) so in interacting with the later ringing it is producing smaller modulation ; the height of the bumps is decreasing . finally , as @xmath218 , we can think of the spectrum approaching one that has an infinite number of infinitesimally spaced zero - height bumps , bumps that are invisible in the drpp spectrum .
results for higher @xmath6-poles are shown in figs.[diffells ] .
the waveforms for @xmath219 show a more complicated structure of the pre - ringing radiation , resulting in more complex spectra .
the total energy radiated in different @xmath6-poles is shown in fig.[evl ] .
results are given for several different values of @xmath1 , and for particles falling in from infinity , both with no initial velocity ( the drpp case ) and with @xmath220 .
the distribution of energy among the multipoles is dominated by the quadrupole for small @xmath1 .
as @xmath188 increases to around 1.5 , the higher multipoles become more important , but with further increase of @xmath188 to @xmath189 the trend is reversed , and the ratio of multipole energies take on values that remain constant for further increases in @xmath1 . for infall from infinity , the ultrarelativistic cases , with high @xmath221 ,
radiate more heavily in higher multipoles , as would be expected .
to find the total energy , given in fig .
[ evdist ] , we computed energy radiated in @xmath222 modes and assumed that the energies from each @xmath6 decreased as a geometric series .
this allowed us to add an estimate of the contributions from @xmath223 .
the addition was typically around 2% of the energy .
the energy is plotted as a function of proper distance @xmath224 , rather than @xmath1 , to show more clearly the details for small separation .
aspects of infall from infinity are given in fig .
[ gammas ] , which shows waveforms and spectra .
the waveforms are characterized by a large early amplitude that is an increasing function of @xmath225 , and the spectra show strong low frequency radiation , extending to @xmath109 , due to this early phase of the radiation .
the total energy radiated in the lowest three multipoles is also shown .
( since the relative importance of higher multipoles increases with @xmath221 , extrapolation to total radiated energy is not immediate . )
the energy results are extended to include infall from finite radius where @xmath181 takes the value @xmath51 .
one of the interesting questions that can be clarified with the above results is the validity of the close approximation .
this approximation , for two holes , assumes that the holes are initially close enough so that the structure of the initial data at small radius is inside an initial all - encompassing horizon .
only the large @xmath8 features of the initial data therefore are relevant to the production of outgoing radiation .
we can immediately apply this method to the particle problem by comparing our initial geometry , in ( [ ivmetric ] ) and ( [ fform ] ) , with the initial geometry in equation ( 4.25 ) of ref .
we see that the results of that reference can be applied to the particle problem by the replacement @xmath226 with ( [ z2eq ] ) and ( [ massreln ] ) this can be rewritten as @xmath227 by the methods of refs .
@xcite , @xmath228 , the radiated energy in units of @xmath229 , for each multipole , is shown to be @xmath230 , respectively for @xmath231 .
if we replace @xmath232 with ( [ replace ] ) we find the close limit predictions for energy . in refs .
@xcite the outgoing radiation is computed from an evolution of @xmath233 , using a finite difference representation of ( [ rtzerilli ] ) ( with no source term ) .
here we have also computed waveforms and energies in the close limit directly by the transform methods of secs .
ii and iii , with only the following changes : the integrals @xmath158 are omitted from ( [ sofar ] ) , and only the @xmath75 form of @xmath234 is used in ( [ fform ] ) .
the results of the two methods are energy values that agree to better than 1% and waveforms that are almost indistinguishable .
( the numerically evolved waveforms are rather smoother than the waveforms from the transform method .
they lack the small amplitude wiggles that can be seen , e.g. , in the waveform in fig .
[ smallr0]a at early times . ) in fig .
[ closelim ] , we plot the @xmath27 close - limit energy prediction and compare it to the full computation for the particle infall .
it is clear that the close limit method is acceptable out to @xmath235 , and fails by a large factor at @xmath236 .
this is in accord with the general picture that the close limit should be a reasonably good approximation when the particle starts inside the peak , around @xmath237 , of the potential ( [ zpotential ] ) .
the waveforms in the close limit all have precisely the same shape ; only the amplitudes differ .
the computed waveforms are in excellent agreement with the small @xmath1 waveforms ( see , e.g. , the waveforms for @xmath212 in fig . [
smallr0]a ) .
the close - limit predictions of waveform shape are even better than the energy predictions . only for @xmath188 larger than around 1.5
does the waveform start to change from the close - limit shape .
the dependence of radiated energy on @xmath1 , shown in fig.[evdist ] , is perhaps the most interesting result of our computations .
one has an intuitive instinct that the radiated energy should decrease with decreasing @xmath1 . at the crudest level this intuition
is based on the idea that infall from a larger radius results in a particle which `` strikes '' the black hole harder and excites more quasinormal ringing . since a decrease of energy with smaller @xmath1 is expected , a natural first guess for approximating the decrease is to multiply the drpp energy by a reduction factor @xmath238 , where @xmath239 is some fitting parameter of order unity .
this approximation , with @xmath240 is shown in fig.[farapprox ] and is compared with the computed results , and with the drpp limit .
the choice @xmath240 was made to give good agreement at @xmath10 , and presumably at larger values of @xmath1 .
( it is difficult to compute energies at much larger values of @xmath1 , due to the rapid modulation of the spectrum . )
a more physical justification for decrease of radiation with decreasing @xmath1 can be constructed starting with the quadrupole formula .
a faster moving particle from a larger @xmath1 implies a larger value of the time derivatives of the quadrupole moment .
this argument has been used@xcite as the basis of a simple quantitative model for the effect of varying @xmath1 .
the energy for infall from infinity , in that model , is reduced by a reduction factor @xmath241 based on the quadrupole formula .
( 23 ) of ref .
( for large @xmath1 that reduction factor reduces to @xmath242 . ) in fig .
[ farapprox ] , we show the result of that simple model .
it is clear that the @xmath241 factor captures the correct qualitative feature of a decrease of radiation with decrease of @xmath1 but implies too dramatic a decrease . as @xmath1 continues to decrease
, a rather unexpected effect appears .
below @xmath243 the energy begins to increase with decreasing separation . at
yet smaller radii ( @xmath1 less than around 4.5 m ) the energy again decreases with decreasing @xmath1 , as the close limit dictates it must .
thus the relationship of radiated energy and @xmath1 has the expected nature in the two regimes where simple arguments apply : large separations and small separations . the anomalous behavior in the range @xmath244@xmath245 underscores the fact that the generation of outgoing radiation is tied closely to the nature of the potential ( [ zpotential ] ) , which peaks around @xmath236 , and can not be understood in terms of close or far approximations .
this anomaly , it should be noted , appears to have no equivalent feature in the case of the head - on collision of two equal mass holes@xcite .
presumably this is because the replacement of the particle by a hole means that the infalling hole is not localized at a particular value of the potential of the other hole .
as the mass ratio of the infalling holes becomes smaller and smaller there must come a point at which an anomalous bump develops in the dependence of radiation on initial separation .
we have seen that the particle limit provides a relatively easy tool for understanding some aspects of the generation of radiation , and of the collisions of holes .
we intend next to use this formalism to study what features on initial data are important to determining how much energy is radiated for particle infall , and presumably for black hole collisions .
a. a. abramovici _ et al .
_ , science * 256 * , 325 ( 1992 ) ; k. s. thorne , submitted to _ proceedings of snowmass 94 summer study on particle and nuclear astrophysics and cosmology _ , eds .
w. w. kolb and r. peccei ( world scientific , singapore ) .
lccc @xmath188&@xmath246&@xmath247&error + + 15&2&@xmath248&1% + & 3&@xmath249&1% + & 4&@xmath250&5% + + 5&2&@xmath251&1% + & 3&@xmath252&2% + & 4&@xmath253&6% + + 3&2&@xmath254&2% + & 3&@xmath255&4% + & 4&@xmath256&11% + + 2&2&@xmath257&3% + & 3&@xmath258&5% + & 4&@xmath259&11% + + 1.5&2&@xmath260&3% + & 3&@xmath261&5% + & 4&@xmath262&10% + + 1.1&2&@xmath263&@xmath264 + & 3&@xmath265&1% + & 4&@xmath266&4%
+ lccc @xmath181&@xmath267&@xmath268&@xmath269 + + 1&@xmath270&@xmath271&@xmath272 + 1.1&@xmath273&@xmath274&@xmath275 + 1.3&@xmath276&@xmath277&@xmath278 + 1.5&@xmath279&@xmath280&@xmath281 + 1.8&@xmath282&@xmath283&@xmath284 + 3&@xmath285&@xmath286&@xmath287 + | we compute gravitational radiation waveforms , spectra and energies for a point particle of mass @xmath0 falling from rest at radius @xmath1 into a schwarzschild hole of mass @xmath2 .
this radiation is found to lowest order in @xmath3 with the use of a laplace transform .
in contrast with numerical relativity results for head - on collisions of equal - mass holes , the radiated energy is found not to be a monotonically increasing function of initial separation ; there is a local radiated - energy maximum at @xmath4 .
the present results , along with results for infall from infinity , provide a complete catalog of waveforms and spectra for particle infall .
we give a representative sample from that catalog and an interesting observation : unlike the simple spectra for other head - on collisions ( either of particle and hole , or of equal mass holes ) the spectra for @xmath5 show a series of evenly spaced bumps . a simple explanation is given for this .
lastly , our energy vs. @xmath1 results are compared with approximation methods used elsewhere , for small and for large initial separation . |
the ads / cft correspondence @xcite between string states in anti de sitter ( ads ) space and conformal field theories in physical space - time , modified for color confinement , has led to semiclassical models for strongly - coupled qcd which provide analytical insight into its inherently non - perturbative nature , as well as predictions for hadronic spectra , decay constants , form factors and wavefunctions .
we have recently shown @xcite that there is a remarkable correspondence between the ads description of hadrons and the hamiltonian formulation of qcd in physical space - time quantized on the light front .
a key feature is `` light - front holography '' which allows one to precisely map the @xmath2 solutions @xmath3 for hadronic amplitudes in the fifth dimensional variable @xmath0 to light - front hadron wavefunctions @xmath4 in physical space - time evaluated at fixed light - front time @xmath5 @xcite . for two particles @xmath6 which is conjugate to the invariant mass squared @xmath7 light - front holography
thus allows hadronic amplitudes in the ads / qcd fifth dimension to be mapped to amplitudes in physical space - time .
one can derive this connection by showing that one obtains the identical holographic mapping for matrix elements of the electromagnetic current and the energy - momentum tensor @xcite .
the mathematical consistency of light - front holography for both the electromagnetic and gravitational @xcite hadronic transition matrix elements demonstrates that the mapping between the ads holographic variable @xmath0 and the light - front variable @xmath8 is a general principle .
some of the important features of light - front ads / qcd include : ( a ) effective frame - independent single - particle schrdinger and dirac equations for meson and baryon wave equations in both @xmath0 and @xmath9 for example , the meson eigenvalue equation is @xmath10\phi(z ) = { \cal m}^2\phi(z),\ ] ] where the soft - wall potential has the form of a harmonic oscillator @xmath11 ( b ) the mass spectra formula for mesons at zero quark mass in the soft - wall model is @xmath12 which agrees with conventional regge phenomenology . as in the nambu string model based on a rotating flux tube , the regge slope is the same for both the principal quantum number @xmath13 and the orbital angular momentum @xmath14 .
the ads / qcd correspondence thus builds in a remarkable connection between the string mass @xmath15 in the string theory underlying an effective gravity theory in the fifth dimension with the orbital angular momentum of hadrons in physical space - time .
( c ) the pion is massless at zero quark mass in agreement with general arguments based on chiral symmetry .
( d ) the predicted form factors for the pion and nucleons agree well with experiment .
the nucleon lfwfs have both @xmath16 and @xmath17 wave components , allowing one to compute both dirac and pauli form factors . in general the ads
/ qcd form factors fall off in momentum transfer squared @xmath18 with a leading power predicted by dimensional counting rules and the leading twist of the hadron s dominant interpolating operator at short distances . under conformal transformations
the interpolating operators transform according to their twist , and consequently the ads isometries map the twist scaling dimensions into the ads modes @xcite .
light - front holography thus provides a simple semiclassical approximation to qcd which has both constituent counting rule behavior @xcite at short distances and confinement at large distances @xcite .
( e ) the timelike form factors of hadrons exhibit poles in the @xmath19 vector meson channels , an analytic feature which arises from the dressed electromagnetic current in ads / qcd @xcite .
( f ) the form of the nonperturbative pion distribution amplitude @xmath20 obtained from integrating the @xmath21 valence lfwf @xmath22 over @xmath23 , has a quite different @xmath24-behavior than the asymptotic distribution amplitude predicted from the erbl pqcd evolution @xcite of the pion distribution amplitude .
the ads prediction @xmath25 has a broader distribution than expected from solving the erbl evolution equation in perturbative qcd .
this observation appears to be consistent with the results of the fermilab diffractive dijet experiment @xcite , the moments obtained from lattice qcd @xcite , and pion form factor data @xcite .
the conversion of quark and gluon partons is usually discussed in terms of on - shell hard - scattering cross sections convoluted with _ ad hoc _ probability distributions .
the lf hamiltonian formulation of quantum field theory provides a natural formalism to compute hadronization at the amplitude level @xcite . in this case one
uses light - front time - ordered perturbation theory for the qcd light - front hamiltonian to generate the off - shell quark and gluon t - matrix helicity amplitude using the lf generalization of the lippmann - schwinger formalism : @xmath26 here @xmath27 is the invariant mass squared of the intermediate state and @xmath28 is the set of interactions of the qcd lf hamiltonian in the ghost - free light - cone gauge @xcite .
the @xmath29 matrix elements are evaluated between the out and in eigenstates of @xmath30 .
the lfwfs of ads / qcd can be used as the interpolating amplitudes between the off - shell quark and gluons and the bound - state hadrons . specifically , if at any stage a set of color - singlet partons has light - front kinetic energy @xmath31 , then one coalesces the virtual partons into a hadron state using the ads / qcd lfwfs .
this provides a specific scheme for determining the factorization scale which matches perturbative and nonperturbative physics .
the event amplitude generator is illustrated for @xmath32 in fig .
[ fig1 ] .
\(a ) since propagation in lf hamiltonian theory only proceeds as @xmath35 increases , all particles propagate as forward - moving partons with @xmath36 .
there are thus relatively few contributing @xmath37ordered diagrams .
\(b ) the computer implementation can be highly efficient : an amplitude of order @xmath38 for a given process only needs to be computed once .
in fact , each non - interacting cluster within @xmath29 has a numerator which is process independent ; only the lf denominators depend on the context of the process .
\(e ) the @xmath29 matrix computation allows for the effects of initial and final state interactions of the active and spectator partons .
this allows for novel leading - twist phenomena such as diffractive dis , the sivers spin asymmetry and the breakdown of the pqcd lam - tung relation in drell - yan processes .
this method retains the quantum mechanical information in hadronic production amplitudes which underlie bose - einstein correlations and other aspects of the spin - statistics theorem .
thus einstein - podolsky - rosen quantum theory correlations are maintained , even between far - separated hadrons and clusters @xcite .
a similar off - shell t - matrix approach was used to predict antihydrogen formation from virtual positron
antiproton states produced in @xmath40 collisions @xcite .
j. m. maldacena , adv .
theor . math .
* 2 * , 231 ( 1998 ) [ int .
j. theor .
phys . * 38 * , 1113 ( 1999 ) ] [ arxiv : hep - th/9711200 ] .
s. j. brodsky and g. f. de teramond , phys .
lett . *
96 * , 201601 ( 2006 ) [ arxiv : hep - ph/0602252 ] .
s. j. brodsky and g. f. de teramond , phys .
d * 77 * , 056007 ( 2008 ) [ arxiv:0707.3859 [ hep - ph ] ] .
s. j. brodsky and g. f. de teramond , phys .
d * 78 * , 025032 ( 2008 ) [ arxiv:0804.0452 [ hep - ph ] ] . s. j. brodsky and g. f. de teramond , phys . lett .
b * 582 * , 211 ( 2004 ) [ arxiv : hep - th/0310227 ] .
s. j. brodsky and g. r. farrar , phys .
lett . * 31 * , 1153 ( 1973 ) .
v. a. matveev , r. m. muradian and a. n. tavkhelidze , lett .
nuovo cim .
* 7 * , 719 ( 1973 ) .
s. j. brodsky and g. f. de teramond , arxiv:0802.0514 [ hep - ph ] .
j. polchinski and m. j. strassler , phys .
rev . lett .
* 88 * , 031601 ( 2002 ) [ arxiv : hep - th/0109174 ] .
h. r. grigoryan and a. v. radyushkin , phys .
d * 76 * , 095007 ( 2007 ) [ arxiv:0706.1543 [ hep - ph ] ] .
g. p. lepage and s. j. brodsky , phys .
b * 87 * , 359 ( 1979 ) .
a. v. efremov and a. v. radyushkin , phys .
b * 94 * , 245 ( 1980 ) .
e. m. aitala _ et al .
_ [ e791 collaboration ] , phys . rev . lett . * 86 * , 4768 ( 2001 ) [ arxiv : hep - ex/0010043 ] . h. m. choi and c. r. ji , phys
. rev .
d * 74 * , 093010 ( 2006 ) [ arxiv : hep - ph/0608148 ] .
s. j. brodsky , g. de teramond and r. shrock , aip conf .
proc . * 1056 * , 3 ( 2008 ) [ arxiv:0807.2484 [ hep - ph ] ] .
s. j. brodsky , h. c. pauli and s. s. pinsky , phys .
rept . * 301 * , 299 ( 1998 ) [ arxiv : hep - ph/9705477 ] .
s. j. brodsky , r. roskies and r. suaya , phys .
d * 8 * , 4574 ( 1973 ) . b. i. abelev , et al , arxiv:0812.1063 [ nucl - ex ] . | light - front holography allows hadronic amplitudes in the ads / qcd fifth dimension to be mapped to frame - independent light - front wavefunctions of hadrons in physical space - time , thus providing a relativistic description of hadrons at the amplitude level .
the ads coordinate @xmath0 is identified with an invariant light - front coordinate @xmath1 which separates the dynamics of quark and gluon binding from the kinematics of constituent spin and internal orbital angular momentum .
the result is a single - variable light - front schrdinger equation for qcd which determines the eigenspectrum and the light - front wavefunctions of hadrons for general spin and orbital angular momentum . a new method for computing the hadronization of quark and gluon jets at the amplitude level using ads / qcd light - front wavefunctions is outlined .
ads / cft , qcd , holography , light - front wavefunctions , hadronization 11.15.tk , 11.25tq , 12.38aw , 12.40yx |
seyfert ( 1943 ) included ngc2782 in his famous list of galaxies which `` show spectra having many high - excitation emission lines localized in the nuclei '' .
however , balzano s ( 1983 ) observations showed a more hii - region like spectrum and , hence , she classified ngc2782 as a starburst galaxy . nevertheless , early high - resolution spectroscopy revealed line - width differences ( sakka et al .
1973 ; kennicut et al .
1989 ) that were attributed to an additional high - excitation component .
boer et al .
( 1992 ) identified the latter component as an expanding shell driven by a series of supernova explosions in the starburst nucleus .
kinney et al .
( 1984 ) measured an einstein ipc - band ( 0.24 kev ) luminosity of 1.210@xmath10 or 810@xmath10 erg s@xmath4 for assumed power - law flux distribution indices of @xmath11 and @xmath12 , respectively .
boer et al .
found the observed x - ray luminosity too high as compared to other indicators of the strength of the nuclear starburst .
therefore a pointed observation was proposed by b.b . and h.s . , which will be discussed in the present work .
ngc6240 is a morphologically peculiar and probably interacting galaxy with two apparent nuclei ( fried & schulz 1983 ) .
a more than 8 kpc extended central region reveals a pronounced liner - like spectrum and unusually broad ( fwhm 200900 km s@xmath4 ) emission lines .
( fosbury & wall 1979 ; fried & schulz 1983 ; morris & ward 1988 , keel 1990 , heckman et al . 1990 , veilleux et al . 1995 , schmitt et al
. 1996 ) .
after detection of its huge far - infrared luminosity of @xmath13 ( wright et al . 1984 )
ngc6240 has been considered as a super starburst galaxy ( joseph & wright 1985 ; rieke et al .
1985 ) . in this vein , heckman et al .
( 1987 , 1990 ) suggested the presence of superwind activity .
recently , iso observations confirmed the presence of a young starburst ( lutz et al .
1996 ) . however , there are suspicions of alternative or additional power sources like an obscured agn ( de poy et al .
1986 ) or even ordinary starlight ( thronson et al .
1990 ) . while the latter hypothesis has been ruled out ( shier et al .
1996 ) there is recent evidence from _ hubble space telescope _ data for a high - excitation component indicative of a hidden agn or a young mini starburst ( barbieri et al .
1993 , rafanelli et al .
also , preliminary x - ray data from suggest the presence of an agn ( mitsuda 1995 ) . to date
, no further x - ray data have been published for ngc6240 .
rieke ( 1988 ) inspected heao-1 data and gave an upper limit of @xmath14 erg s@xmath4 @xmath15 corresponding to @xmath16 erg s@xmath4 in the ( 210 ) kev range .
the -pspc data discussed below were taken from the archive .
both objects have characteristics of a nuclear starburst and a possible agn in common .
an agn could be important for the energy budget , in particular for the tremendous amount of fir radiation from an ultraluminous infrared galaxy ( ulirg ) .
perhaps even more interesting , such an object might be the birthplace of a newly forming agn as is suggested by the similar space densities of ulirgs and qsos ( sanders et al .
1988 ) .
x - rays ought to be helpful in distinguishing between these two types of nuclear sources .
sources powered by a starburst wind shocking the dense shell and other ambient gas are usually expected to emit a soft thermal x - ray spectrum with a ( 0.12.4 kev ) luminosity @xmath17 erg s@xmath4 while objects exceeding this limit commonly show agn tracers ( e.g. moran et al .
1994 ; wisotzki & bade 1997 ) . a power - law x - ray spectrum with a photon spectral index
@xmath18 would be indicative of an agn ( e.g. , pounds et al .
1994 , nandra et al . 1997 ; svensson 1994 ) ( although agns may have thermal components or starburst components added ) .
the luminosity of a typical type-1 agn in the band is about 3% of its bolometric luminosity .
luminosities given below are calculated via plain application of the hubble law with @xmath19 km s@xmath4 mpc@xmath4 yielding @xmath20 mpc and 144 mpc for ngc2782 and ngc6240 , respectively ( when comparing luminosities or scales note that several recent papers on ngc6240 assumed @xmath21 mpc which would halve the luminosities given here ) .
the data presented here were taken with the high resolution imager ( hri ) and the position sensitive proportional counter ( pspc ) on board of the x - ray satellite . a detailed description of the project , the satellite and its instrumentation can be found in trmper ( 1983 ) , trmper et al .
( 1991 ) , pfeffermann et al .
( 1987 ) and zombeck et al .
( 1990 ) . in brief
, the satellite observes in the soft x - ray range between 0.1 and 2.4 kev . the spatial resolution of frames obtained with the hri and pspc is about @xmath22 and @xmath23 , respectively .
unlike the hri , the pspc has sufficient spectral resolution for obtaining information about the spectral shape .
on april 1820 , 1992 , a deep hri x - ray image of ngc2782 was taken with a total effective exposure time of @xmath24 21.7 ksec . via a standard source - detection procedure carried out with the exsas x - ray analysis software package ( zimmermann et al .
1994 ) we detected seven x - ray sources ( @xmath25 ) in the field of view . in table 1
the detections are summarized .
the positions , the background subtracted hri count rates , and the optical identifications of the sources are given . [ cols="<,^,^,^,^,^,^,^,^",options="header " , ] in fig . 1 we plotted the x - ray source positions over the optical image of ngc2782 taken from the digitized palomar sky survey plates .
the source numbers are identified in table 1 .
the central source no .
1 is associated with the center of ngc2782 .
a four times fainter source ( no . 2 ) appears 88 north of the center of the galaxy .
the association of no .
2 with ngc2782 is unclear .
it could be a foreground or background object .
the x - ray light curves of both sources do not show any significant evidence for variability .
in addition to the hri observation , ngc2782 appears on an archived pspc frame ( observation number : wg701555p ) , taken on nov . 8 , 1993 .
during this 7 ksec exposure ngc2782 was far off - axis in the field of view . due to the reduced spatial resolution of the telescope away from the axis , sources no .
1 and 2 could not be separated . however , since no .
1 dominates the integrated x - ray flux crude information about its spectral shape can be retrieved from the pspc frame . to this end
we extracted the source photons within a radius of 300 , corrected the data for telescope vignetting and detector dead - time , and binned the pspc spectrum according to a constant signal - to - noise .
for the background correction of the source spectrum we selected the counts of a source free region near ngc2782 .
we then fitted the background subtracted pspc spectrum with a single powerlaw taking into account the galactic interstellar absorption towards ngc2782 of @xmath26 @xmath15 ( dickey & lockman 1990 ) .
the best fit yields a photon index @xmath27 and an x - ray luminosity ( 0.1 - 2.4 kev ) of log(@xmath28)[email protected] .
this does not mean that the spectrum actually _ is _ a powerlaw , but a spectral shape has to be assumed to obtain a flux .
ngc6240 was observed twice with the pspc , on sept . 24 , 1992 and feb .
1314 , 1993 .
the source was located at the center of the field of view .
the total exposure time is about 8 ksec . for the data analysis ,
source photons were extracted within a circle of radius 150 . after removing all detected sources within the inner field of view the background was determined and subtracted .
the data were corrected for vignetting and dead - time following standard prescriptions of the exsas software ( zimmermann et al .
the source - countrate is about 0.06 cts s@xmath4 . for the spectral analysis source photons in the amplitude channels
11 - 240 were binned so that a constant signal / noise ratio of @xmath30 was secured .
for the temporal analysis we chose time bins of 400 s which are the smallest ones to smear out the variations caused by the satellite s wobble motion .
the x - ray isophotes for the 1992 exposure overlayed on the digitized poss are shown in fig .
the k5 star hd152140 is the strongest x - ray source in both observations ( mean countrate @xmath31 cts s@xmath4 ) .
all other sources in the field are significantly weaker than ngc6240 .
the k star was used to positionally align both frames .
approximating the source photon distribution by a gaussian and representing the extent by the fwhm of this distribution , we find a value of 37 .
this exceeds the width of the pspc point spread functions for different channels .
e.g. , for a point source we determined fwhms of 23 and 27 at channel 110 and 60 , respectively .
a notable extension beyond the psf is present in all energy intervalls ( soft as well as hard ) . to be sure
, we additionally verified that the extent is not due to the electronic ghost imaging that widens the point spread function below 0.2 kev ( nousek & lesser 1993 ) .
consequently , the x - ray image obtained with the pspc is extended in both observations .
the x - ray positions of both , ngc6240 and the k star , agree with their optical positions ( as given in simbad ) within less than one arcsec for the 1993 observation . in the 1992 data , _ both _ x - ray positions are shifted by @xmath24 10 relative to their optical positions .
this shift can be traced back to the known boresight error of the telescope ( briel et al .
1994 ) and has been corrected for in the overlay of the x - ray contours on the optical image ( fig .
figure 2 shows the individual countrates versus time for the 1992 and 1993 observations . given the statistical errors no significant variations can be claimed . in the all - sky survey , ngc6240 was detected with a countrate of @xmath32 ( voges et al .
1997 ) which is consistent with the individual pointed observations .
as shown in fig . 3 there is no evidence for spectral changes between the 1992 and 1993 observations so that the data were merged to ensure better statistics .
however , the @xmath33 photons detected in total limit any meaningful fits by spectral models to no more than two spectral components .
even with this restriction , the measured total soft x - ray luminosity is highly model dependent as is illustrated by the numerous statistically acceptable fits in table 2 .
this is due to the large bin widths which causes individual fitted photon numbers in each bin to depend on the spectral shape .
in addition to the individual model determining parameters like temperature @xmath34 ( one parameter of the `` thermal '' models as bb = black body , thb = thermal bremsstrahlung , rs = thin thermal plasma in collisional equilibrium as computed by raymond & smith ( 1977 ) ) or powerlaw index @xmath35 ( for `` nonthermal '' pl = powerlaw models according to a photon flux distribution @xmath36 with @xmath37 = photon energy ) and the necessary normalization constants , the fits depend on the column density @xmath38 of `` cold '' gas that is represented by the summed photoelectric cross sections for gas with cosmic chemical abundances .
table 2 shows that individually fitted @xmath38 values can grossly vary among the fitted models because @xmath38 influences the slope of the soft component .
we adopt the galactic hydrogen column @xmath39 @xmath15 ( dickey & lockman 1990 ) in the direction of ngc6240 as a lower limit .
however , a spectral emission model fitted with this column would have to encounter no absorption _ within _ ngc6240 which appears somehow unlikely in view of the abundant molecular and ionized material observed in the central region of this object .
the emission would have to arise in an `` outer shell '' or penetrate through optically thin windows in the ism .
indeed , a pl - model with @xmath40 ( model e in table 2 ) leads to a bad fit while a bb - model ( model a ) fits well .
however , such a huge bb - shell , optically thick in soft x - rays and extending over more than 30 or , in linear scales , over more than 21 kpc can not be reasonably imagined in physical terms .
the physically expected thin x - ray emitting gas will emit lines with a significant contribution so that the thb fit ( model b ) is excluded as well . our main goal to showing these ( from the beginning ) unrealistic models
is to find out the range of luminosities ( last column in table 2 ) obtained by different spectral shapes .
a powerlaw with free column ( model d ) is statistically feasible but has a steep slope with @xmath41 .
only the subgroup of so - called narrow - line seyfert-1 galaxies ( nlsy1 ; e.g. puchnarewicz et al .
1992 ; boller et al . 1996
; greiner et al . 1996 ) is known to have comparatively steep slopes .
komossa & greiner ( 1995 ) proposed to explain these objects by warm absorption of an intrinsically flat ( @xmath42 ) powerlaw .
the steep spectrum is then mimicked by the presence of matter - bounded , completely h - ionized absorbing gas ( so - called `` warm absorbers '' ) , which absorb by metal ions . applying a standard warm - absorber model ( as e.g. described in komossa & fink 1997 )
we found no acceptable fit .
hence , the steep slope , far off the canonical value @xmath18 , has to be intrinsic which , if persisting in the uv , would cause an ultraluminous uv component never seen in any unobscured object . finally , the luminosity of model d ( last column in table 2 ) of @xmath43 erg s@xmath4 is so huge that , for common agn @xmath44 ratios , a hidden qso of @xmath45 erg s@xmath4 would be required that would heat the central dust in ngc6240 to more than three times the observed fir luminosity unless the bulk of its radiation escapes elsewhere .
we consider model d as unlikely .
we are left with model c as a purely thermal solution and f as a hybrid model consisting of thin thermal gas plus a nonthermal component .
the added pl in f leads to a comfortably small luminosity compared to c because its assumed ` flat ' slope helped to minimize @xmath38 .
tuning the pl index to flatter slopes does not lead to a significant further luminosity reduction ( cf .
model e ) .
possible origins of the components will be discussed in sect .
lllcllccc model & @xmath38 & @xmath46 & norm@xmath47 & @xmath34 & norm@xmath48 & @xmath49 & @xmath50 & @xmath51 + & 10@xmath52 @xmath15 & & ph/@xmath53/s / kev@xmath54 & kev & @xmath55 & erg/@xmath53/s & & erg / s + a ( bb ) & 0.549@xmath56 & & & 0.28 & @xmath57 & @xmath58 & 0.98 ( 16 ) & @xmath59 + b ( thb ) & 1.96 & & & 0.73 & @xmath60 & @xmath61 & 0.89 ( 15 ) & @xmath62 + c ( rs ) & 9.63 & & & 0.44 & & @xmath63 & 1.11 ( 15 ) & @xmath3 + d ( pl ) & 3.5 & @xmath64 & @xmath65 & & & @xmath66 & 0.86 ( 15 ) & @xmath67 + e ( pl ) & 0.549@xmath56 & @xmath68 & @xmath69 & & & @xmath70 & 2.55 ( 16 ) & @xmath71 + f ( rs+pl ) & 1.61 & @xmath72 & @xmath73 & 0.63 & & @xmath74 & 0.77 ( 14)&@xmath5 + @xmath54 at 1 kev ; @xmath55 ph/@xmath53/s for bb , ph/@xmath53/s / kev at 1 kev for thb ; @xmath56 fixed to the galactic value ; @xmath75 fixed +
due its more complete database we first discuss ngc6240 . from the discussion in sect . 5 only model c ( thermal rs with luminosity @xmath3 erg s@xmath4 ) and the pl+rs model f ( rs luminosity @xmath76 erg s@xmath4 , pl luminosity @xmath7 erg s@xmath4 ) of table 2 are acceptable . given the abundant evidence for superwind activity and a possible agn these appear to be the most likely sources for the x - ray emission .
a notable contribution by direct emission from ob stars present in the supposed starburst is ruled out : each star only emits @xmath77 erg s@xmath4 of a rather soft x - ray energy distribution ( e.g. vaiana et al . 1981 , berghfer et al .
a significant fraction of @xmath78 unobscured o stars would have shown up in the optical .
a large number of massive x - ray binaries and a few supernova remnants with luminosities in the range 10@xmath79 erg s@xmath4 do collectively play a significant role in ` normal ' spiral galaxies .
ngc6240 might be the product of the collision of such objects which in their normal state have a x - ray luminosity of @xmath80 erg s@xmath4 ( fabbiano 1989 ) . since there is no known mechanism to boost the number of x - ray binaries by two orders of magnitudes in the interaction process it appears practically ruled out that the @xmath81 erg s@xmath4 soft x - rays come from binaries .
supernova remnant activity can indeed be boosted in a starburst triggered by the interaction , but this leads us directly to the superwind model .
sophisticated two - dimensional hydrodynamical simulations of galactic superwinds are available ( e.g. suchkov et al .
however , such studies mainly address the `` prototype '' m82 so that their particular examples hardly reach the temperatures and luminosities found in our fits .
lacking precise constraints our goal is limited to the estimate whether the huge thermal rs - component luminosities from the fits are feasible .
hence , we use a scalable theory rather than a hydrodynamical model with specific initial conditions . we utilize the most sophisticated models of this type which were built by mac low & mccray ( 1988 ; for an application see also schulz 1990 ) who had extended the seminal work by weaver et al .
( 1977 ) .
densities and distribution functions of the ism in the center of ngc6240 are not known .
the extent of the supershell can be estimated from the five - finger structure on available h@xmath82 images ( schulz 1984 , keel 1990 , thronson et al .
1990 ) that yield @xmath83 kpc including a correction for a disk inclined by up to 40 from edge - on . the strong dust lane on b photographs ( best visible in the masked prints made by fosbury & wall 1979 ) suggests a close - to - edge - on view of the main disk .
the large line widths ( fwhm ) cited in sect . 1 which most frequently lie in the range between 480 and 620 km s@xmath4 , but occasionally reach 900 km s@xmath4 ( keel 1990 ) , are due to the summed effect of outflow , rotation and ` turbulence ' ( the filamentary shell in m82 shows linewidths of up to 300 km s@xmath4 ; cf .
bland & tully 1988 ) . as pointed out by van der werf et al .
( 1993 ) , shock velocities exceeding 200 km s@xmath4 would boost the [ oi ] and [ sii ] emission - line intensities relative to h@xmath82 and lower [ oiii]/[oi ] thereby deteriorating the explanation of the optical spectra by a large contribution of shockheated gas .
since @xmath84 , an outflow velocity of the shell of 200 km s@xmath4 is consistent with the data and turbulence components with @xmath85 km s@xmath4 ( fwhm @xmath86 km s@xmath4 ) . as mean density of the swept up ism we have chosen @xmath87 @xmath88 , higher than the canonical values of @xmath89 @xmath88 from our galaxy or m82 ( strickland et al .
the most prominent extended bright parts of the supershell are on the eastern side of the double nucleus which might be due to a slightly asymmetric location of the sources of the wind . the mechanical input power @xmath90 is given by the supernova rate ( maclow & mccray 1988 ) for which we adopt 1 sn yr@xmath4 ( van der werf et al .
1993 estimate 2 yr@xmath4 ; colbert et al .
1994 obtain 1 yr@xmath4 from the diffuse radio emission but are uncertain about the extra power released in the compact radio knots ) . with an energy per sn of 10@xmath91 erg
, we obtain @xmath92 erg s@xmath4 . according to heckman et al .
( 1996 ) shock models by binette et al .
( 1985 ) predict @xmath93(h@xmath94 , yielding @xmath95 erg s@xmath4 with @xmath93(h@xmath96 erg s@xmath4 from fosbury & wall ( 1979 ) . largely due to a larger reddening correction , veilleux et al .
( 1995 ) obtain @xmath93(h@xmath97 erg s@xmath4 which would lead to @xmath98 erg s@xmath4 or 14 supernovae yr@xmath4 which is not compatible with other evidence .
we take 3 sn yr@xmath4 or @xmath95 erg s@xmath4 as an upper limit and assume that h@xmath82 has to be partially attributed to ionizing sources other than shocks .
scaling of the formula of maclow & mccray ( 1988 ) leads to the following expressions ( for @xmath8 we followed heckman et al.(1996 ) who had carried out the integration over the superbubble model in the band ) : @xmath99 @xmath100 @xmath101 where @xmath102 , @xmath103 and @xmath104 .
these relations show that , with the observational constraints so far available , it is quite natural to account for the thermal @xmath105 erg s@xmath4 obtained for the rs - component of the rs+pl hybrid model .
however , how can we reach the @xmath3 erg s@xmath4 required for the single - rs fit ? we need a factor of 43 in eq .
a factor - of-3 increase of @xmath106 would fit to our upper limit of @xmath90 that is in accordance with shocked h@xmath82 ( see above ) but already strain the conditions for @xmath107 and @xmath108 ( eqs . 1 and 2 ) .
enhancing the density to @xmath109 ( a rather high value in the halo into which the bubble expands ) would boost @xmath8 by a factor 3.5 and comfortably decrease @xmath107 and @xmath108 to still acceptable values .
however , we still need another factor 4.3 requiring a factor of 20 in the expansion - time parameter .
however , a @xmath110 yr old bubble would have already commenced other starbursts for which no trace is visible and , more seriously , our theory would no more be applicable .
this exercise of boosting the parameters to high but not excludable values demonstrates the basic difficulty of getting a soft - x ray luminosity of several @xmath111 erg s@xmath4 from a single superbubble .
some concern about the single - rs fit ( model c in table 2 ) also arises by the high cold absorbing column , which corresponds to @xmath112 mag ( using bohlin s et al.(1978 ) relation ) .
thronson et al . (
1990 ) find @xmath113 in the double nuclei but elsewhere @xmath114 .
the latter value should be more typical for most of the volume occupied by the extended supershell .
inverse - compton ( ic ) upscattered photons from the fir photons in the radio sources of ngc6240 would generate a flux of @xmath115 erg s@xmath4 @xmath15 ( colbert et al .
the powerlaw component of model f ( table 2 ) emits a flux of @xmath116 erg s@xmath4 and is therefore likely to be attributed to an agn - like source .
it may , however , appear striking that the expected ic scattered flux fits our measured powerlaw if the compton x - ray emission disk - halo model devised by goldshmidt & rephaeli ( 1995 , gr95 ) is upscaled from the starburst galaxy ngc253 to ngc6240 via the fir ratio .
the gr95 model predicts an unattenuated compton flux of @xmath117 erg @xmath15 s@xmath4 in the band if ngc253 were placed at 144 mpc , the distance of ngc6240 adopted here . scaling this by the ratio 175 of the fir luminosities ( the flux formula from helou et al .
( 1985 ) yields @xmath118 for ngc6240 , gr95 use @xmath119 for ngc253 ) leads to @xmath120 erg @xmath15 s@xmath4 which is , indeed , nearly the same as the power - law flux of ngc6240 given in the last paragraph . however , this simple scaling may be misleading because the magnetic fields in the radio sources of ngc6240 are higher than in ngc253 ( colbert et al .
1994 ) . in the flux - dominating two
radio nuclei the _ lower limit _ of the magnetic field is fifty times ( @xmath121 g ) larger than that in the central synchrotron source of ngc253 ( @xmath122 g ) while in the western extended regions @xmath123 g which is 2.2 times larger .
granted that other conditions are scalable , the boost in the magnetic energy densities ( @xmath124 ) should diminish the above ngc253-scaled ic flux estimate ( roughly @xmath125 ) by factors between @xmath126 ( compact components ) and @xmath127 ( extended regions ) .
hence , within this scheme it appears that at least the extended radio sources may contribute a significant comptonized power law .
adopting this scaled gr95 model for ngc6240 would , however , mean to scale the radiation energy density distribution from ngc253 by luminosity which requires an identical size of the fir emitting region in objects that differ in luminosity by a factor of hundred .
even if this might be the case the electron distribution in the extended halo components of ngc253 and ngc6240 would have to be similar as well , which is doubtful because of the hundred times smaller propagation lengths for the disk - halo connection via electron convection and diffusion in the latter galaxy .
this is due to the increased radiative loss in the central disk ( see , e.g. , eq . 8 and the subsequent discussion in gr95 ) . in their more direct approach , colbert et al .
related the compton upscattering to the _ observed _ synchrotron sources and tested two complementary types of fir distributions ( distributed like the radio sources and uniform distribution ) .
neither case , regardless of compact or extended sources , yielded any flux above a few percent of that of our pl component . finding no direct clue from available observations that ic upscattering of fir photons plays a significant role
we continue to interpret the powerlaw by an agn - like source .
notwithstanding , gr95-type ic models for extended components might become relevant if some of the presently only crudely known parameters will attain more favorable values for such processes in the future . the supershell model in sect .
6.1.1 easily accounts for the rs - component of model f which makes 20% of the total flux .
we now have to explain the remaining @xmath7 erg s@xmath4 in the powerlaw . aside from the x - rays ,
the only other hint for high - energy radiation of an agn is given by the high - excitation core in the southern nucleus ( rafanelli et al .
1997 ) that accounts for @xmath128% of h@xmath129 .
this component is probably due to a ` window ' in the ism and does not allow a direct view of the continuum source .
consequently , the supposed agn is likely to be seen in scattered light .
this is in agreement with our non - detection of variability of the supposed agn because scattering the x - rays over several kpc would smear out any short - term variability .
the electrons in the extended ionized regions of the superwind and supershell itself are one possible source of scatterers .
a simple geometry with only one constant - density population of electrons defines a fraction @xmath130 of the agn luminosity @xmath8 to be seen as scattered luminosity @xmath131 by the condition @xmath132 where @xmath133 is the thomson cross - section @xmath134 @xmath53 , @xmath135 our measured powerlaw luminosity of @xmath7 erg s@xmath4 . from the central biconical h@xmath82 structure one
estimates that a fraction @xmath136 can escape from the central @xmath8 ( for simplicity assumed to be isotropic in the first instance ) , corresponding to an ideal bicone with opening angle @xmath137 for each side . in practice
, the structure has a broader base and has a shape something in between a biconical and a bycilindrical configuration , but for a crude estimate this distinction is unimportant
. basically two types of scatterers are available : firstly , the dense material close to the baseline and in the broken shells . for isothermal shocks we would expect densities of the order of @xmath138 @xmath88 . for the bright filamentary material within the innermost 2 kpc
an electron density of 500 @xmath88 was measured by heckman et al .
secondly , low - density , distributed in the bicone , could reach significant electron - scattering depths due to its spatial extent of up to 10 kpc .
the h@xmath82 image from keel ( 1990 ) shows a strong surface - brightness enhancement in the southern nucleus suggesting that luminous high - density material is concentrated there .
adopting from the above @xmath139 @xmath88 and @xmath140 pc leads to @xmath141 or a predicted ( isotropic ) @xmath142 erg s@xmath4 .
the mass of this ionized gas in the inner bicone would be @xmath143 , compatible with @xmath144 estimated from hydrogen - line luminosities ( depoy et al .
1986 ; veilleux et al . 1995
; the lower limit applies for pure recombination ) .
the amount of dense ionized gas is only minute compared to the @xmath145 of available cold molecular gas in the center of ngc6240 ( solomon et al .
its filling factor is likely to be @xmath146 , but the covering fraction must be large to ensure an _ effective _ scattering length of 200 pc . in this scenario the power - law component should be unresolved with current x - ray telescopes .
the second possibility an extended low - density region would be much less efficient in scattering and involve a huge mass with unclear origin .
e.g. @xmath147 kpc , @xmath148 @xmath88 would lead to @xmath149 erg s@xmath4 and @xmath150 ; in case of @xmath151 kpc , @xmath152 @xmath88 , we would need @xmath153 , but @xmath154 erg s@xmath4 that would exceed @xmath155 of ngc6240 ! we conclude that scattering is feasible but , even in a simple high - density scenario , it requires a rather luminous agn in the range @xmath156 erg s@xmath4 , which corresponds to @xmath157 erg s@xmath4 for a typical agn continuum , which is an appreciable fraction or nearly all of the fir luminosity ( @xmath158 erg s@xmath4 ) of ngc6240 .
the luminosity demand for the agn is lowered in case of anisotropy of the agn radiation because in the unified scenario more luminosity is expected along the bicone than in the equatorial plane of the molecular torus .
an alternative possibility would be a component like a ` warm scatterer ' , i.e. near - central high density gas that scatters the x - rays into an escape window of the ism but shows itself relatively little intrinsic emission , is possible .
the need for an agn in ngc6240 is also supported by preliminarily reported data ( mitsuda 1995 ) that show a power law in the 210 kev range plus a strong emission - line complex around fek@xmath82 . in sect .
2.1 we obtained @xmath159 erg s@xmath4 ( within a factor of 4 ) .
this can be compared with @xmath8 expected from data by boer et al .
( 1992 ) who found optical evidence for an expanding supershell with @xmath160 kpc and d@xmath107/d@xmath161 km s@xmath4 leading to an age of @xmath162 yrs .
they measured a total h@xmath129 luminosity of @xmath163 erg s@xmath4 yielding @xmath93(h@xmath82)@xmath164 erg s@xmath4 in case b. however , they estimate ( on p. 74 ) that the outflowing shell contributes 1% of the h - recombination lines , but caution on p. 75
that this value might be ten times higher due to incomplete coverage . since the mechanical power should be hundred times that of h@xmath82 ( see sect .
6.1.1 ) we obtain @xmath165 erg s@xmath4 or a sn rate of 0.026 to 0.26 yr@xmath4 . plugging the lower sn rate into eqs .
( 1 ) to ( 3 ) yields ( for @xmath166 ) @xmath167 kpc , @xmath168 km s@xmath4 and @xmath169 erg s@xmath4
. while @xmath107 and @xmath170 are in good agreement with the measured data , @xmath8 is a factor 46 too small ! increasing the uncertain parameter @xmath171 by a factor 10 , would boost the luminosity by a factor of only 3 and lower size and velocity of the shell to 60% the values they should have .
the latter effect can be canceled by taking the upper limit of @xmath90 by a factor of 10 , leading to @xmath172 erg s@xmath4 which is consistent with the measured @xmath8 .
the sn rate is comparable to that of m82 and ngc253 ( rieke et al .
1980 ; kronberg et al .
1985 ) , starburst galaxies with an order of magnitude smaller @xmath8 from their centers ( strickland et al . 1997 ) .
the difference is explained by the hundred times smaller ambient density in these objects ( @xmath173 ) than here assumed for ngc2782 ( @xmath174 ) .
if this latter conjecture will one day turn out to be wrong then seyfert ( 1943 ) might have been correct in considering ngc2782 as a ` seyfert galaxy ' of which we now see some x - rays scattered in the outflow region
. however , presently the evidence for an agn in ngc2782 does not appear compelling .
we analyzed data of the ultraluminous ir galaxy ngc6240 and of the starburst galaxy ngc2782 . with our best fit of the soft - x ray spectrum of ngc6240 we found a luminosity ( 0.12.4 kev ) of @xmath5 erg s@xmath4 that , by at least an order of magnitude , exceeds that of its ultraluminous rival arp220 ( see heckman et al .
( 1996 ) : @xmath175 to @xmath176 erg s@xmath4 ) .
the difference is partly due to a luminous power - law component not present in arp220 .
we interpret this power - law component as scattered light from a hidden agn with a luminosity that , depending on the poorly known scattering geometry , provides a major fraction of the bolometric luminosity of ngc6240 .
about 20% of the total @xmath8 in the band are due to thermal superwind - shocked shell material .
lacking other components the scattered light of the hidden agn is easiest to see in x rays .
the scattered optical continuum is hard to decompose from the strong stellar background that was analyzed by schmitt et al .
( 1996 ) and shier et al .
veilleux et al .
( 1995 ) noticed ` featureless ' continuum components in optical spectra of most ulirgs .
polarization vectors presented in radovich ( 1993 ) reveal a scattering geometry around the center but more detailed work in this regard could be rewarding . in sect .
6.1.1 we saw that the hydrogen recombination - line spectrum is too strong to be completely attributed to shock heating . as shown by schulz & fritsch ( 1994 ) ,
attenuating the ionizing continuum of an agn by hidden ionized absorbing gaseous columns with , e.g. @xmath177 , leads automatically to a liner like spectrum as observed in ngc6240 .
this thin absorber is nearly transparent to x - rays .
hence , the emerging picture of the core of ngc6240 is that of a buried agn component , which unlike a seyfert with a fully developed high - excitation spectrum , sees in no direction a completely clear sky .
ngc2782 has an order - of - magnitude smaller x - ray luminosity than ngc6240 and can , in excellent agreement with optical data , be consistently explained by superbubble activity if the ambient density is relatively high ( average of 1 @xmath88 in the inner 2 kpc of the halo ) .
this does not completely exclude a hidden low - luminosity agn , but so far , there is no positive evidence for it .
we thank an anonymous referee for pointing our attention to the paper by goldshmidt & rephaeli ( 1995 ) .
gratefully acknowledges the hospitality of the physics and astronomy department of dartmouth college while finishing this work .
t.w.b . acknowledges support from the alexander - von - humboldt - stiftung by a feodor - lynen fellowship .
the project is supported by the german bundesministerium fr bildung , wissenschaft , forschung und technologie ( bmbf / dara ) and the max - planck - society .
the optical images shown are based on photographic data of the national geographic society palomar observatory sky survey ( ngs - poss ) obtained using the oschin telescope on palomar mountain .
the ngs - poss was funded by a grant from the national geographic society to the california institute of technology .
the plates were processed into the present compressed digital form with their permission .
the digitized sky survey was produced at the space telescope science institute under us government grant nag w-2166 .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , caltech , under contract with the national aeronautics and space administration .
balzano v.a . , 1983 , apj 268 , 602 barbieri c. , rafanelli p. , schulz h. et al . , 1993 , a&a 273 , 1 berghfer t.w . , schmitt j.h.m.m . , danner r. , cassinelli j. , 1997 , a&a 322 , 167 binette l. , dopita m.a .
, tuohy i.r . , 1985 , apj 297 , 476 bland j. , tully b. , 1988 , nat 334 , 43 boer b. , schulz h. , keel w.c . , 1992 , a&a 260 , 67 bohlin r.c . , savage b.d . , drake j.f . , 1978 , apj 224 , 132 boller t. , brandt w.n .
, fink h.h . , 1996 , a&a 305 , 53 briel u. , aschenbach b. , hasinger g. et al . , 1994 , _ rosat user s handbook _ colbert e.j.m .
, wilson a.s . , bland - hawthorn j. , 1994 , apj 436 , 89 depoy d.l . , becklin e.e . ,
wynn - williams c.g . , 1986 ,
apj 307 , 116 dickey j.m .
, lockman f.j . , 1990 , ara&a 24 , 499 fabbiano g. , 1989 , ara&a 27 , 87 fosbury r.a.e . , wall j.v . , 1979 ,
mnras 189 , 79 ( fw79 ) fried j. , schulz h. , 1983 , a&a 118 , 166 goldshmidt o. , rephaeli y. , 1995 , apj 444 , 113 ( gr95 ) greiner j. , danner r. , bade n. , richter g.a .
, kroll p. , s. , 1996 , a&a 310 , 384 heckman t , m .
, armus l. , miley g.k . , 1987 , aj 93 , 276 heckman t.m . , armus l. , miley g.k . , 1990 ,
apjs 74,833 heckman d.m .
, dahlem m. , eales s.a . , fabbiano g. , weaver k. , 1996 , apj 457 , 616 helou , g. , soifer b.t . , rowan - robinson m. , 1985 , apj 298 , l7 joseph r.d . ,
wright , g.s .
, 1985 , mnras 209 , 111 keel w.c . , 1990 ,
aj 100 , 356 kennicut jr .
keel w.c . ,
blaha c.a . , 1988
, aj 97 , 1022 kinney a.l . , bregman j.n . , huggins p.j
, glassgold a.e . , 1984 ,
pasp 96 , 398 komossa s. , greiner j. , 1995 , ag abstract series 11 , 217 komossa s. , fink h. , 1997 , a&a 322 , 719 kronberg p.p . , biermann p. , schwab f.r . , 1985 , apj 291 , 693 lutz d. , genzel r. , sternberg a. et al .
, 1996 , a&a 315 , l137 mac low m - m .
, mccray r. , 1988 , apj 324,776 mitsuda k. , 1995 , ann.n.y.acad.sc . 759 , proc .
17th texas symp .
ap . and cosm .
h.bhringer , g.e.morfill , j.e.trmper ( new york : the new york acad . of sc . ) , p. 213
moran e.c . ,
halpern j.p . , helfand d.j . , 1994 , apj 433 , l65 morris s.l . , ward m.j . , 1988 , mnras 230 , 639 nandra k.,george i.m . , mushotzky r.f . , turner t.j . , yaqoob t. , 1997 , astro - ph/9606169 , to appear in apj nousek j. , lesser a. , 1993 , newsletter 8 , 13 pfeffermann e. , briel u.g . , hippmann h. et al . , 1987 , spie 733 , 519 pounds k.a . , nandra k. , fink h. , makino f. , 1994 , mnras 267 , 193 puchnarewicz e.m . , mason k.o . , cordova f.a .
et al . , 1992 ,
mnras 256 , 589 radovich m. , 1993 , doctoral thesis , university of padova rafanelli p. , schulz h. , barbieri c. et al .
, 1997 , a&a , in press raymond j.c . , smith b.w .
1977 , apjs 35 , 419 rieke g.h . ,
lebofsky m.j .
, thompson r.i .
, low f.j .
, tokunaga a.t . , 1980 , apj 238 , 24 rieke g.h . , cutrie r.m . , black j.h .
et al . , 1985 ,
apj 290 , 116 rieke g.h . , 1988 ,
apj 331 , l5 sakka k. , oka s. , wakamatsu k. , 1973 , pasj 25 , 153 sanders d.b . ,
soifer b.t . , elias j.h .
et al . , 1988 ,
apj 325 , 74 schmitt h.r . ,
bica , e. , pastoriza , m.g . , 1996 , mnras 278 , 965 schulz h. , 1984 , sterne und weltraum 23 , 311 schulz h. , 1990 , aj 99 , 1442 schulz h. , fried j , rser s. , keel w.c . , 1993 , a&a 277 , 416 schulz h. , fritsch ch . , 1994 , a&a 291 , 713 seyfert c.k . 1943 ,
apj 97 , 28 shier l.m .
, rieke m.j .
, rieke g.h . , 1996 ,
apj 470 , 222 solomon p.m. , downes d. , radford s.j.e . , barrett j.w .
1997 , apj 478 , 144 strickland d.k .
, ponman t.j .
, stevens i.r . , 1997 , a&a 320 , 378 suchkov a.a . , balsara d.s . ,
heckman t.m .
, leitherer c. , 1994 , apj 430 , 511 svensson r. , 1994 , apjs 92 , 585 thronson h.a .
, majewski s. , descartes l. , hereld m. , 1990 , apj 364 , 456 trmper j. , 1983 , adv . space res . 2 , 241 trmper j. , hasinger g. , aschenbach b. et al . , 1991 , nat 349 , 579 vaiana g.s .
et al . , 1981 ,
apj 245 , 163 van der werf p.p . , genzel r. , krabbe a. et al .
, 1993 , apj , 405 , 522 veilleux s. , kim d .- c .
, sanders d.b . , mazzarella m. , soifer b.t . , 1995 , apjs 98 , 171 voges w. et al . , 1997 ,
a&a to be submitted weaver r. , mccray r. , castor j. , shapiro p. , moore r. , 1977 , apj 218 , 377 ; 220 , 742 wisotzki l. , bade n. 1997 , a&a 320 , 395 wright g.s . , joseph r.d .
, meikle w.p.s . , 1984 ,
nat 309 , 430 zimmermann h.u . , becker w. , belloni t. , dbereiner s. , izzo c. , kahabka p. , schwentker o. , 1994 , mpe report 257 zombeck m.v .
et al . , 1990 , in proc .
spie conf .
euv , x - ray , and gamma - ray instrumentation for astronomy , calibration of the rosat high resolution imager , 1344 , 267 | we present observations of the starburst galaxy ngc2782 ( hri plus a weak pspc frame ) and the ultraluminous infrared galaxy ngc6240 ( pspc ) .
the ( 0.12.4 ) kev spectra of both objects appear similar .
however , due to better sampling spectral modeling is only warranted in case of ngc6240 for which both a single thermal raymond - smith model ( @xmath0 ) or a hybrid model consisting of 80% power - law with the canonical photon index @xmath1 plus 20% raymond - smith contribution ( @xmath2 ) lead to good fits . however , the single thermal model turns out to be unlikely because it yields a luminosity of @xmath3 erg s@xmath4 , which is hard to reach in a starburst superwind scenario .
the hybrid model leads to a more moderate luminosity of @xmath5 erg s@xmath4 , of which @xmath6 erg s@xmath4 can be attributed to shocked superwind gas .
we link the remaining @xmath7 erg s@xmath4 powerlaw to an agn component because the alternative of inverse - compton scattering of the fir radiation leads to a too low flux when estimated with available data .
the result appears to be consistent with preliminarily announced asca observations . for ngc2782
we find @xmath8(0.12.4 kev ) = @xmath9 erg s@xmath4 ( within a factor of four ) which can be explained by emission from a shocked superwind region with a high preshock density in agreement with earlier optical evidence for an outflowing supershell .
2.3 cm |
in the book @xcite helgason uses sgeirsson s mean value theorem ( see theorem ii.5.28 ) to solve the wave equation @xmath1 on euclidean spaces .25mm@xmath2 ( see exercise ii.f.1 and its solution pp .
574575 ) and the shifted wave equation @xmath3 ^ 2\bigr){\hskip.25mm}u(x , t){\hskip.25mm},\\ \,u(x,0)=f(x){\hskip.25mm},\ , \partial_{{\hskip.25mm}t}|_{{\hskip.25mm}t=0}\,u(x , t)=g(x){\hskip.25mm},\\ \end{cases}\ ] ] on real hyperbolic spaces @xmath4 ( see exercise ii.f.2 and its solution pp .
575577 ) . in this work
we extend this approach both to damek ricci spaces and to homogeneous trees . along the way we clarify the role of the inverse dual abel transform in solving the shifted wave equation .
recall that damek
ricci spaces are riemannian manifolds , which contain all hyperbolic spaces @xmath4 , @xmath5 , @xmath6 , @xmath7 as a small subclass and share nevertheless several features with these spaces . before @xcite the shifted wave equation on @xmath4
was solved explicitly in ( * ? ? ?
* section 7 ) .
other hyperbolic spaces were dealt with in @xcite and damek
ricci spaces in @xcite .
all these approaches are awkward in our opinion . on one hand , @xcite , @xcite and @xcite
rely on the method of descent i.e. on shift operators , which reduce the problem to checking formulae in low dimensions .
moreover @xcite involves classical compact dual symmetric spaces and does nt cover the exceptional case .
on the other hand , @xcite involves complicated computations and follows two different methods : helgason s approach for hyperbolic spaces and heat kernel expressions @xcite for general damek ricci spaces . in comparison
we believe that our presentation is simpler and more conceptual .
several other works deal with the shifted wave equation without using explicit solutions .
let us mention @xcite ( see also ( * ? ? ?
* section v.5 ) ) for huygens principle and the energy equipartition on riemannian symmetric spaces of the noncompact type .
this work was extended to damek
ricci spaces in @xcite , to chbli trimche hypergroups in @xcite and to the trigonometric dunkl setting in @xcite .
the nonlinear shifted wave equation was studied in @xcite , first on real hyperbolic spaces and next on damek ricci spaces .
these works involve sharp dispersive and strichartz estimates for the linear equation .
related @xmath8 estimates were obtained in @xcite on hyperbolic spaces .
our paper is organized as follows . in section 2 ,
we review damek ricci spaces and spherical analysis thereon .
we give in particular explicit expressions for the abel transform , its dual and the inverse transforms . in section [ section3 ] we extend sgeirsson s mean value theorem to damek
ricci spaces , apply it to solutions to the shifted wave equation and deduce explicit expressions , using the inverse dual abel transform . as an application
, we investigate huygens principle .
section [ section4 ] deals with the shifted wave equation on homogeneous trees , which are discrete analogs of hyperbolic spaces .
most of this work was done several years ago .
the results on damek - ricci spaces were cited in @xcite and we take this opportunity to thank franois rouvire for mentioning them and for encouraging us to publish details .
we are also grateful to nalini anantharaman for pointing out to us the connection between our discrete wave equation on trees and recent works @xcite of brooks and lindenstrauss .
we shall be content with a brief review about damek ricci spaces and we refer to the lecture notes @xcite for more information . damek ricci spaces are solvable lie groups [email protected] mm , which are extensions of heisenberg type groups .25mm@xmath10 by @xmath11 and which are equipped with a left - invariant riemannian structure . at the lie algebra level , @xmath12 with lie bracket @xmath13 = ( \tfrac z2{\hskip.25mm}x'\ ! -\tfrac{z^{{\hskip.25mm}\prime}}2x , { \hskip.25mm}z\,y'\!-z^{{\hskip.25mm}\prime}{\hskip.25mm}y+[{\hskip.25mm}x , x ' ] , 0)\ ] ] and inner product @xmath14 at the lie group level , @xmath15 with multiplication @xmath16 , { \hskip.25mm}z{\hskip-.25mm}+{\hskip-.25mm}z^{{\hskip.25mm}\prime}){\hskip.25mm}.\ ] ] so far .25mm@xmath10 could be any simply connected nilpotent lie group of step @xmath17.25 mm .
heisenberg type groups are characterized by conditions involving the lie bracket and the inner product on @xmath18 , that we shall not need explicitly . in particular
.25mm@xmath19 is the center of .25mm@xmath10 and .25mm@xmath20 .25mmis even .
one denotes by @xmath21 the ( manifold ) dimension of .25mm@xmath22 and by @xmath23 the so called homogeneous dimension of .25mm@xmath10 . via the iwasawa decomposition , all hyperbolic spaces @xmath4 ,
@xmath5 , @xmath6 , @xmath7 can be realized as damek
ricci spaces , real hyperbolic spaces corresponding to the degenerate case where .25mm@xmath10 is abelian . but most damek
ricci spaces are not symmetric , although harmonic , and thus provide numerous counterexamples to the lichnerowicz conjecture @xcite .
despite the lack of symmetry , radial analysis on .25mm@xmath22 is similar to the hyperbolic space case and fits into jacobi function theory @xcite . in polar coordinates , the riemannian volume on .25mm@xmath22
may be written as @xmath24 , where @xmath25 is the common surface measure of all spheres of radius .25mm@xmath26 in @xmath22 and .25mm@xmath27 denotes the normalized surface measure on the unit sphere in [email protected] mm .
we shall not need the full expression of the laplace - beltrami operator @xmath29 on @xmath22 but only its radial part @xmath30 on radial functions and its horocyclic part @xmath31 on [email protected] functions i.e. on functions .25mm@xmath32 depending only on @xmath33.25 mm .
the laplacian .25mm@xmath29 .25mmcommutes both with left translations and with the averaging projector @xmath34 hence with all spherical means @xmath35 thus @xmath36 finally @xmath29 has a spectral gap .
more precisely its [email protected] is equal to the half
line .25mm@xmath38 $ ] . radial fourier analysis on .25mm@xmath22 may be summarized by the following commutative diagram in the schwartz space setting @xcite : @xmath39 here @xmath40 denotes the spherical fourier transform on .25mm@xmath22 , @xmath41 the abel transform , @xmath42 the classical fourier transform on .25mm@xmath43 .25mmand @xmath44 .25mmthe space of smooth radial functions .25mm@xmath45 on .25mm@xmath22 .25mmsuch that @xmath46 for every .25mm@xmath47 .
recall that the abel transform and its inverse can be expressed explicitly in terms of weyl fractional transforms , which are defined by @xmath48 for .25mm@xmath49 .25mmand for [email protected] mm , .25mm@xmath51 .25mmsuch that .25mm@xmath52 .
specifically , @xmath53 where @xmath54 .
more precisely , @xmath55 if .25mm@xmath56 .25mmis odd i.e. @xmath57 .25mmis even , and @xmath58 if .25mm@xmath56 .25mmis even i.e. @xmath57 .25mmis odd .
similarly , the dual abel transform @xmath59 and its inverse can be expressed explicitly in terms of riemann - liouville fractional transforms @xmath60.25 mm , which are defined by @xmath61 for .25mm@xmath49 .25mmand for [email protected] mm , .25mm@xmath62 .25mmsuch that .25mm@xmath52 .
[ theoremdualabeltransform ] the dual abel transform is a topological isomorphism between @xmath63 and @xmath64 .
explicitly , @xmath65\bigr\}(r)\ ] ] and @xmath66 where @xmath67.25 mm .
more precisely ,
@xmath68 if .25mm@xmath56 .25mmis odd i.e. @xmath57 .25mmis even , and @xmath69 if .25mm@xmath56 .25mmis even i.e. @xmath57 .25mmis odd .
everything follows from the duality formulae @xmath70 and from the properties of the riemann - liouville transforms , in particular @xmath71 for every integer [email protected] mm . in the degenerate case .25mm@xmath73 , we recover the classical expressions for real hyperbolic spaces [email protected] mm : @xmath75 where @xmath76.25 mm , @xmath77 if @xmath56 .25mmis odd and @xmath78 if @xmath56 .25mmis even .
[ asgeirssontheorem1dr ] assume that @xmath79 .25mmsatisfies @xmath80 then @xmath81 for every @xmath82 .25mmand @xmath83.25 mm .
the _ proof _ is similar to the real hyperbolic space case ( * ? ? ?
* section ii.5.6 ) once one has introduced the double spherical means @xmath84 and transformed into @xmath85 sgeirsson s theorem is the following limit case of theorem [ asgeirssontheorem1dr ] , which is obtained by dividing by @xmath86 and by letting @xmath87.25 mm .
[ asgeirssontheorem2dr ] under the same assumptions , @xmath88 given a solution .25mm@xmath89 to the shifted wave equation @xmath90 on .25mm@xmath22 with initial data .25mm@xmath91 and .25mm@xmath92 , .25mmset @xmath93 where .25mm@xmath94 .25mmis the .25mm@xmath33 .25mmcoordinate of .25mm@xmath95
. then satisfies , according to . by applying corollary [ asgeirssontheorem2dr ] to with .25mm@xmath96
.25mmand .25mm@xmath97 , we deduce that the dual abel transform of .25mm@xmath98 , as defined in , is equal to the spherical mean .25mm@xmath99 of the initial datum .25mm@xmath100 .
hence @xmath101 by integrating with respect to time , we obtain the solutions @xmath102 to with initial data .25mm@xmath103 and .25mm@xmath104 . in conclusion ,
general solutions to the shifted wave equation @xmath105 on .25mm@xmath22 are given by @xmath106 by using theorem [ theoremdualabeltransform ] , we deduce the following explicit expressions .
[ solutionwaveequationdr ] ( a ) when .25mm@xmath56 .25mmis odd , the solution to is given by @xmath107 with @xmath108.25 mm .
\(b ) when .25mm@xmath56 .25mmis even , the solution to is given by @xmath109 with @xmath110.25 mm .
these formulae extend to the degenerate case .25mm@xmath73 , which corresponds to real hyperbolic spaces @xmath74 : \(a ) @xmath56 .25mmodd : @xmath111 with @xmath112.25 mm .
\(b ) @xmath56 .25mmeven : @xmath113 with @xmath114.25 mm . as an application ,
let us investigate the propagation of solutions .25mm@xmath115 .25mmto the shifted wave equation with initial data .25mm@xmath116 .25mmsupported in a ball @xmath117 .
the following two statements are immediate consequences of theorem [ solutionwaveequationdr ] .
firstly , waves propagate at unit speed .
[ finitepropagationspeeddr ] under the above assumptions , @xmath118 secondly , huygens principle holds in odd dimension , as in the euclidean setting .
this phenomenon was already observed in @xcite .
[ huygensprincipledr ] assume that .25mm@xmath56 .25mmis odd .
then , under the above assumptions , @xmath119 may not vanish when [email protected] mm , but it tends asymptotically to .25mm@xmath121 .
this phenomenon was observed in several settings , for instance on euclidean spaces in @xcite , on riemannian symmetric spaces of the noncompact type @xcite , on damek
ricci spaces @xcite , for chbli trimche hypergroups @xcite , ... our next result differs from @xcite in two ways . on one hand
, we use explicit expressions instead of the fourier transform . on the other hand
, we aim at energy estimates as in @xcite , which are arguably more appropriate than pointwise estimates .
recall indeed that the total energy @xmath122 is time independent , where @xmath123 is the kinetic energy and @xmath124 the potential energy . by the way ,
let us mention that the equipartition of into kinetic and potential energies was investigated in @xcite and in the subsequent works @xcite ( see also ( * ? ? ?
* section v.5.5 ) and the references cited therein ) .
[ pointwiseestimatedr ] let .25mm@xmath115 .25mmbe a solution to .25mmwith smooth initial data .25mm@xmath125 .25mmsupported in a ball @xmath117 .
then @xmath126 for every .25mm@xmath127 .25mmand @xmath128 .25mmsuch that @xmath129.25 mm .
assume @xmath130 .25mmand consider the second part @xmath131 of the solution .25mm@xmath132 in theorem [ solutionwaveequationdr].b .
the case @xmath133 .25mmand the first part are handled similarly .
as @xmath134 , we have @xmath135 and thus it remains to apply the differential operator @xmath136 to @xmath137.25 mm .
firstly @xmath138 and secondly @xmath139 for some constants .25mm@xmath140 . as @xmath141 we conclude that @xmath142 .25mmand hence .25mm@xmath143 are @xmath144 . the derivatives .25mm@xmath145 and .25mm@xmath146 are estimated similarly . as far as .25mm@xmath147 is concerned , we use in addition that @xmath148 this concludes the proof of lemma [ pointwiseestimatedr ] .
[ asymptotichuygensdr ] let .25mm@xmath115 .25mmbe a solution to with initial data @xmath149 and let .25mm@xmath150 be a positive function such that @xmath151 then @xmath152 tend to .25mm@xmath121
.25mmas @xmath153 . in other words
, the energy of @xmath115 .25mmconcentrates asymptotically inside the spherical shell @xmath154 _ proof of theorem [ asymptotichuygensdr]_. .25mmby combining lemma [ pointwiseestimatedr ] with the volume estimate @xmath155 we deduce that the three integrals @xmath156 are .25mm@xmath157 .25mmand hence tend to @xmath121 as @xmath158.25 mm .
this section is devoted to a discrete setting , which is similar to the continuous setting considered so far .
a homogeneous tree .25mm@xmath159 of degree .25mm@xmath160 is a connected graph with no loops and with the same number .25mm@xmath161 of edges at each vertex
. we shall be content with a brief review and we refer to the expository paper @xcite for more information ( see also the monographs @xcite ) . for the counting measure , the volume of any sphere .25mm@xmath162 in .25mm@xmath163 .25mmis given by @xmath164 once we have chosen an origin .25mm@xmath165 and a geodesic .25mm@xmath166
.25mmthrough .25mm@xmath121 , let us denote by @xmath167 .25mmthe distance of a vertex .25mm@xmath168 to the origin and by .25mm@xmath169 .25mmits horocyclic height ( see figure [ uhspt ] ) .
the combinatorial laplacian is defined on .25mm@xmath175
.25mmby @xmath176 and similarly on .25mm@xmath163 .25mmby @xmath177 the [email protected] of .25mm@xmath178 is equal to the interval .25mm@xmath179 $ ] , .25mmwhere @xmath180 we have @xmath181 on radial functions and @xmath182 on horocyclic functions .
again , radial fourier analysis on .25mm@xmath163 may be summarized by the following commutative diagram @xmath183 here @xmath184 denotes the spherical fourier transform on .25mm@xmath163 , @xmath185 the abel transform and @xmath186 a variant of the classical fourier transform on [email protected] mm .
moreover @xmath187 , @xmath188 .25mmdenotes the space of even functions on .25mm@xmath175 .25mmsuch that @xmath189 and @xmath190 .25mmthe space of radial functions on .25mm@xmath163 .25mmsuch that @xmath191 consider finally the dual abel transform @xmath192 the following expressions are obtained by elementary computations .
we are interested in the following shifted wave equation on .25mm@xmath163 : @xmath203 as was pointed out to us by nalini anantharaman , this equation occurs in the recent works @xcite .
the unshifted wave equation with discrete time was studied in @xcite and the shifted wave equation with continuous time in @xcite .
_ proof of theorem [ asgeirssontheoremtree]_. .25mmfix .25mm@xmath207 and consider the double spherical means @xmath214 that we shall denote by .25mm@xmath215 for simplicity . according to lemma [ lemmaasgeirssontree ] ,
our assumption may be rewritten as @xmath216 let us prove the symmetry @xmath217 by induction on @xmath218.25 mm .
first of all , is trivial if .25mm@xmath219 .25mmand with .25mm@xmath220 is equivalent to with [email protected] mm .
assume next that .25mm@xmath222 and that holds for .25mm@xmath223 .
on one hand , let .25mm@xmath224 .25mmwith @xmath225 and let @xmath226.25 mm .
we deduce from at the point @xmath227 that @xmath228 by adding up over [email protected] mm , we obtain @xmath229 which vanishes by induction . on the other hand ,
we deduce from at the points @xmath230 and @xmath231 that @xmath232 hence @xmath233 .25mmby using and by induction .
this concludes the proof of theorem [ asgeirssontheoremtree ] .
let us now solve explicitly the shifted wave equation on @xmath163 as we did in section [ section3 ] for the shifted wave equation on damek ricci spaces .
consider first a solution .25mm@xmath115 .25mmto with initial data .25mm@xmath91
.25mmand [email protected] mm . on one hand ,
as @xmath235 satisfies the same cauchy problem , we have .25mm@xmath236 by uniqueness . on the other hand , according to , the function @xmath237 satisfies .
thus , by applying to .25mm@xmath204 with [email protected] mm , we deduce that the dual abel transform of @xmath239 .25mmis equal to the spherical mean .25mm@xmath240 of the initial datum @xmath100 .
hence @xmath241 consider next a solution .25mm@xmath115 .25mmto with initial data .25mm@xmath103 .25mmand
.25mm@xmath242 @xmath243.25 mm .
.25mmis an odd function of .25mm@xmath56 .25mmand @xmath245 is a solution to with initial data .25mm@xmath246
.25mmand [email protected] mm .
hence @xmath248 with [email protected] mm .
by using lemma [ abeltree].b , we deduce the following explicit expressions . as we did in section [ section3 ] , let us next deduce propagation properties of solutions .25mm@xmath115 .25mmto the shifted wave equation with initial data .25mm@xmath116 .25mmsupported in a ball @xmath260 .
obviously huygens principle does nt hold for , strictly speaking .
let us show that it holds asymptotically , as for even dimensional damek
ricci spaces . for this purpose , define as follows the kinetic energy @xmath265 and the potential energy @xmath266 for solutions .25mm@xmath115 .25mmto . here
@xmath267 is the @xmath171step laplacian on .25mm@xmath163 .25mmand @xmath268
\(b ) notice that the shifted wave equation @xmath275 amounts to @xmath276 as @xmath277 we have on one hand @xmath278 on the other hand , @xmath279 by adding up and , we obtain @xmath280 and we deduce from this expression that @xmath281 this concludes the proof of lemma [ energytree ] .
[ energyconservationtree ] alternatively , lemma [ energytree].b can be proved by expressing the energies .25mm@xmath282 , @xmath283 , @xmath284 in terms of the initial data .25mm@xmath100 , @xmath285 .25mmand by using spectral calculus .
specifically , @xmath286 and @xmath287 here we have used the fact that @xmath288 hence @xmath289 where @xmath290 by considering the corresponding multipliers , we obtain @xmath291 and we conclude that @xmath292 [ asymptoticequipartitiontree ] let .25mm@xmath115 .25mmbe a solution to with finitely supported initial data .25mm@xmath100 and [email protected] mm .
then the kinetic energy .25mm@xmath282 and the potential energy .25mm@xmath283 tend both to @xmath294 as @xmath295.25 mm .
let us show that the difference @xmath296 .25mmtends to [email protected] mm . by resuming the computations in remark [ energyconservationtree ]
, we obtain @xmath297 with @xmath298 as @xmath299 the expression @xmath300 tends to .25mm@xmath121 .
the expressions @xmath301 and @xmath302 are handled in the same way .
this concludes the proof of theorem [ asymptoticequipartitiontree ] .
[ asymptotichuygenstree ] let .25mm@xmath115 .25mmbe a solution to with finitely supported initial data and let .25mm@xmath303 .25mmbe a sequence of positive integers such that @xmath304 then the expressions latexmath:[\[\sum\nolimits_{\substack { \vphantom{o}\\\hspace{-4mm}x\in{\mathbb{t}}\\ \hspace{-6mm}|x|{\hskip.25mm}<{\hskip.25mm}|n|-{\hskip.25mm}n_n}}\hspace{-4 mm } u. bunke & m. olbrich : _ the wave kernel for the laplacian on the classical locally symmetric spaces of rank one _ ( _ theta functions , trace formulas and the selberg zeta function _ ) , ann .
glob . anal . geom .
* 12 * ( 1994 ) , 357405 a. intissar & m. vall ould moustapha : _ solution explicite de lquation des ondes dans un espace symtrique de type non compact de rang 1 _ , smin .
* 12 * , univ .
grenoble 1 ( 1994 ) , 2528 t.h .
koornwinder : _ jacobi functions and analysis on noncompact semisimple lie groups _ , in _ special functions _
( _ group theoretical aspects and applications _ ) , r.a .
askey , t.h .
koornwinder & w. schempp ( eds . ) , reidel , dordrecht ( 1984 ) , 185 f. rouvire : _ espaces de damek
( _ gomtrie et analyse _ ) , in _ analyse sur les groupes de lie et thorie des reprsentations _ ( _ knitra , 1999 _ ) , j. faraut , f. rouvire & m. vergne ( eds . ) , smin . congr . * 7 * , soc . math .
france , paris ( 2003 ) , 45 - 100 | we solve explicitly the shifted wave equation @xmath0 on damek
ricci spaces , using sgeirsson s theorem and the inverse dual abel transform . as an application , we investigate huygens principle .
a similar analysis is carried out in the discrete setting of homogeneous trees . to appear in _ trends in harmonic analysis _ ( _ dedicated to a. fig
talamanca on the occasion of his retirement _ ) m.a .
picardello ( ed . ) , springer indam ser . , springer
verlag , pp . |
the age and chemical composition of the stellar populations in galaxies , together with galaxy mass , are key ingredients to uncover galaxy formation and evolutionary paths .
estimates of stellar populations parameters are derived by interpreting detailed spectral information , such as absorption features , on the basis of stellar population synthesis ( sps ) models .
our ability of interpreting galaxy spectra has greatly improved with the development of sps models that i ) predict the full spectrum of simple stellar populations ( ssps ) at medium / high resolution , allowing to adjust the models to the data quality rather than the other way round , and ii ) have a better coverage of the stellar parameters space , allowing the interpretation of a broader range of stellar populations @xcite .
together with the development of sps models and spectral fitting techniques , the statistical power of large spectroscopic surveys has allowed to put on a firm ground our understanding of the stellar populations in nearby galaxies . in this contribution
i will briefly review results on the dependence of stellar populations properties on galaxy mass , as obtained from the analysis of sloan digital sky survey ( sdss ) galaxy spectra .
i will then discuss to what extent these relations are shaped by the environment in which galaxies reside , as recently analysed in pasquali et al ( 2009 , submitted ) combining the sdss dr4 catalogue of stellar ages and metallicities with the @xcite group catalogue .
i will conclude with a brief outlook on new population synthesis models that allow accurate estimates of @xmath0-element abundance ratios .
estimates of stellar metallicity , luminosity - weighted age and stellar mass have been derived for the spectroscopic sample of sdss dr4 by interpreting the strength of a set of absorption features including the 4000 break , the balmer lines and [ @xmath0/fe]-insensitive mg and fe composite indices .
a library of galaxy spectral features is generated by convolving @xcite ssps with a comprehensive set of monte carlo star formation histories ( sfhs ) . following a bayesian statistical approach ,
the likelihood distribution of each parameter has been computed by comparing the observed absorption features with those predicted by the full library .
the uncertainties on physical parameters estimates depend on the spectral signal - to - noise ( s / n ) and on the galaxy spectral type or sfh . a careful analysis of the statistical uncertainties on stellar
m / l , arguably the easiest parameter to constrain , indicates two interesting points for model development and spectral analysis : i ) for galaxies with old stellar populations and spectra of good s / n statistical uncertainties of 0.03 dex could in principle be reached , indicating the need for continuous model development and control of systematics ; ii ) on the other hand for galaxies with recent bursts of star formation the uncertainties are limited by the large intrinsic scatter in physical parameters at fixed index strength : it is thus crucial to properly model the complexity of sfhs in the universe @xcite . in our previous studies ,
based on a sample of sdss dr2 galaxies at @xmath1 , we have analysed the distribution of galaxies in stellar metallicity age stellar mass . in the mean ,
low - mass galaxies are young and metal - poor while high - mass galaxies are old and metal - rich .
there is a rapid transition between these two regimes at stellar masses around @xmath2 @xcite .
this corresponds to the transition regime in other galaxy properties ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and it is where the scatter in stellar populations parameters is largest .
early - type galaxies , which contain the majority of baryons and metals in stars today @xcite , dominate above the transition mass where the relations between metallicity / age and mass flattens .
they follow a tight color - magnitude relation ( cmr ) which is driven by an increase with stellar mass of both stellar age , total metallicity and element abundance ratios @xcite , suggesting that the stars in massive early - type galaxies have formed earlier and on shorter timescales than less massive ones .
the bottom line of the above results is that galaxy sfh is primarily determined by galaxy mass .
the scatter in physical parameters at fixed mass , however , indicates underlying second - order dependences .
the star formation activity in galaxies is also sensitive to the environment in which galaxies reside , as witnessed by the sfr - density relation ( e.g. , * ? ? ?
* ) and by the differences in scatter and zero - point of the cmr of early - type galaxies in different environments ( e.g. , * ? ? ?
* ; * ? ? ?
these environmental dependences are however subtle , once the primary dependence on stellar mass is removed .
moreover the signal may depend on the environmental estimator adopted , such as e.g. galaxy number densities in apertures which have the disadvantage of being sensitive to environment itself and not directly comparable to the more physical description in terms of host dark matter halos of galaxy formation models . :
1415 ( solid ) , 1314 ( long dashed ) , 12.513 ( dashed ) , 1212.5 ( dot - dashed ) , 1112 ( dotted ) ) .
figure adapted from pasquali et al .
( 2009 , submitted).,title="fig : " ] : 1415 ( solid ) , 1314 ( long dashed ) , 12.513 ( dashed ) , 1212.5 ( dot - dashed ) , 1112 ( dotted ) ) .
figure adapted from pasquali et al .
( 2009 , submitted).,title="fig : " ] in pasquali et al .
( 2009 ) we take advantage of the group catalogue by @xcite which assign a dark matter mass to galaxy groups and classify their members into centrals and satellites .
we complement the galaxy group catalogue with the stellar metallicity and age estimates computed for sdss dr4 .
we then analyse the dependence of stellar age and metallicity on stellar mass , distinguishing centrals and satellites and as a function of host halo mass .
the left panel of fig .
[ fig1 ] shows the relations between luminosity - weighted age and stellar mass for central ( hatched region ) and satellite galaxies ( grey shaded region ) .
the relations between stellar metallicity and mass are shown in the right - hand panel . at stellar masses
@xmath3 centrals and satellites have the same mean ages and metallicities . at lower stellar masses instead , satellite galaxies are older and metal - richer than equally - massive centrals .
moreover , we note in the satellite population an excess of old , metal - rich galaxies at low m@xmath4 and a lack of young metal - poor galaxies at nearly all masses .
we reproduce the result found in other studies that at all scales , centrals are older and metal - richer than satellites that live in equally - massive halos .
this likely reflects the general trends of age / metallicity with mass and the fact that in a given halo centrals are more massive than satellites . we also find that , while the stellar populations of central galaxies are nearly independent of halo mass , this is not the case for satellite galaxies .
the white lines of different style in fig .
[ fig1 ] show the age / metallicity mass relations for satellite galaxies in halos of different mass .
below @xmath5 , at fixed stellar mass the stellar populations of satellite galaxies depend on halo mass : satellites in more massive halos are older and metal - richer than those in less massive halos . at high masses ,
the sfh of galaxies is likely regulated by internal feedback processes regardless of environment . at low masses instead the sfh of satellite galaxies is affected by environmental processes , whereby their stellar populations differ from those of equally - massive centrals by an amount that increases with halo mass .
the results discussed above indicate that satellite galaxies are affected by the environment onto which they are accreted through mechanisms that quench star formation and prevent further replenishment of gas via inflows .
it remains to be determined on which timescale the sf in satellite galaxies is quenched and any possible dependence on halo mass .
important clues about sf timescales can be gained from the abundance of @xmath0-elements with respect to fe ( [ @xmath0/fe ] ) .
previous results suggest that the [ @xmath0/fe ] of isolated early - type galaxies is the same as for cluster galaxies @xcite .
a careful definition of environment ( such as the one introduced in the previous section ) and accurate measures of [ @xmath0/fe ] are necessary to confirm or disprove these findings .
theoretical sps models that predict the full spectrum of stellar populations with different [ @xmath0/fe ] are now available @xcite .
recently , @xcite have developed ` differential ' stellar populations models that exploit the predictive power of fully theoretical models , calibrated onto semi - empirical solar metallicity ssps . in this way it is possible to predict the pixel - by - pixel absolute flux ( as opposed to absorption indices only ) as a function of both [ fe / h ] and [ @xmath0/fe ] .
we are now in a position to predict spectra for stellar populations with variable [ @xmath0/fe ] ( in addition to metallicity ) and with complex sfhs , and exploit the high resolution of the models to : i ) convolve them to match the effective resolution of the data ( including velocity dispersion broadening ) , ii ) use the redundancy of the full spectral information for low - quality data , in addition to individual absorption indices .
we have generated a monte carlo library of sfhs based on the new differential models .
adopting the same bayesian approach used before ( see section [ sec1 ] ) we can derive consistent estimates of @xmath0-element abundance ratio together with luminosity - weighted age , metallicity and mass - to - light ratios , and their associated uncertainties . the analysis of a well - defined sample of early - type galaxies will enable us to validate the model predictions and test the sensitivity of total metallicity and stellar
m / l to [ @xmath0/fe ] variations .
estimates of element abundance ratios for large samples of both early - type and star - forming galaxies ( exploiting the better coverage of the models to high temperatures ) will allow us to address several key questions on sf timescale in galaxies . | the age and chemical composition of the stars in present - day galaxies carry important clues about their star formation processes .
the latest generation of population synthesis models have allowed to derive age and stellar metallicity estimates for large samples of low - redshift galaxies . after reviewing the main results about the distribution in ages and metallicities as a function of galaxy mass ,
i will concentrate on recent analysis that aims at disentangling the dependences of stellar populations properties on environment and on galaxy stellar mass .
finally , new models that predict the response of the full spectrum to variations in [ @xmath0/fe ] will allow us to derive accurate estimates of element abundance ratios and gain deeper insight into the timescales of star formation cessation . |
numerical calculation of spherically symmetric gravitational collapse of a massless scalar field by choptuik @xcite and subsequent results for different matter models and symmetries ( see for example refs . ) spectacularly demonstrate that critical phenomena occur in the gravitational collapse , and that the near - critical behavior is universal in some important aspects .
there are two possible late - time outcomes of gravitational collapse , distinguished by whether or not a black hole is formed in the course of evolution . which end state is realized depends on initial conditions , specified by a control parameter @xmath1 that characterizes the gravitational interaction strength in the ensuing evolution . for @xmath2 ( subcritical solution )
the gravitational field is too weak to form a black hole , while for @xmath3 ( supercritical solution ) a black hole is produced .
the critical solution , corresponding to a control parameter value @xmath4 just at the threshold of black hole formation , acts as an intermediate attractor for nearby initial conditions , and has an additional symmetry : discrete or continuous self - similarity , also referred to as echoing . for supercritical evolution
there is usually no mass gap , and the mass of the resultant black hole is described by the mass - scaling relation @xmath5 where @xmath6 is mass - scaling exponent .
general gravitational collapse is a very difficult problem , even if restricted to the case of spherical symmetry . as a result ,
most studies of critical behavior resort to numerical calculations at some stage , with few fully analytical solutions known . finding critical solutions
is made easier by the fact that they have additional symmetry ; requiring solution to be continuously self - similar can simplify the problem enough to make it solvable for simpler matter models .
both the stability of the critical solution and the exact mass - scaling exponent can then be determined by linear perturbation analysis , as suggested by evans and coleman and carried out in refs . .
in this paper we consider an analytical continuously self - similar solution of gravitational collapse of a massless scalar field constructed by roberts @xcite .
originally it was intended as a counter - example to cosmic censorship conjecture , but it was later rediscovered in context of the critical gravitational collapse by brady @xcite and oshiro , nakamura and tomimatsu . although this solution does exhibit critical behavior , there are two problems .
first , as any self - similar solution , it is not asymptotically flat , and the mass of the black hole grows infinitely .
this can be prevented if we cut the self - similar solution away from the black hole and glue it to some other solution to make the spacetime asymptotically flat and of finite mass , which was in fact done in ref . .
a second , and bigger , problem is that growing non - self - similar modes will dominate the mass - scaling exponent if they exist , but they are not accounted for by the self - similar solution
. this paper addresses the second problem by doing linear perturbation analysis of the roberts solution .
we calculate the mass - scaling exponent @xmath6 and analyze the stability of the critical solution .
the remarkable feature of this model is that it allows exact analytical treatment .
we begin by presenting a continuously self - similar spherically symmetric solution of the gravitational collapse of a massless scalar field ( roberts solution ) .
it is most easily derived in null coordinates .
therefore we write the metric as @xmath7 where @xmath8 and @xmath9 are functions of both @xmath10 and @xmath11 .
the coordinates @xmath10 , @xmath11 can be reparametrized as @xmath12 , @xmath13 , so coordinate choice freedom is given by two functions of one variable each ; it can be fixed by setting @xmath14 at two null hypersurfaces @xmath15 and @xmath16 . with this choice of metric
the einstein - scalar field equations become [ eq : es ] @xmath17 @xmath18 @xmath19 @xmath20 @xmath21 the first four equations ( [ eq : es:3]-[eq : es:6 ] ) are the einstein equation for metric ( [ eq : metric ] ) , and equation ( [ eq : es:7 ] ) is the scalar wave equation @xmath22 . equations ( [ eq : es:5],[eq : es:6 ] ) are constraints .
note that in the case of a massless scalar field the wave equation follows from the einstein equations , so one of the equations ( [ eq : es:3],[eq : es:4],[eq : es:7 ] ) is redundant .
we turn on the influx of the scalar field at the advanced time @xmath23 , so that the spacetime is minkowskian to the past of this surface .
the initial conditions for system ( [ eq : es ] ) are specified there by the continuity of the solution .
we use coordinate freedom to set @xmath24 .
assumption that the collapse is continuously self - similar , i.e. that there exists a vector field @xmath25 such that @xmath26 , allows the einstein - scalar equations ( [ eq : es ] ) to be solved analytically .
the critical solution is given by @xmath27,\end{aligned}\ ] ] and is a member of a one - parameter family of self - similar solutions @xmath28 + \varphi(p).\end{aligned}\ ] ] above @xmath29 is constant component of a scalar field , chosen so that @xmath30 , and the solution converges to the critical solution as @xmath1 goes to zero : @xmath31 = - \frac{1}{2 } \ln p + p + o(p^2).\ ] ] for values @xmath32 of control parameter , a black hole is formed ; for @xmath33 the field disperses , and the spacetime is flat again in region @xmath34 .
( our choice of sign of @xmath1 is opposite to the one in @xcite .
also note that the constant field term is omitted there . )
the existence and position of the apparent horizon is given by condition @xmath35 , where @xmath36 due to the spherical symmetry of the problem it is possible to introduce a local mass @xmath37 by @xmath38 the mass contained inside the apparent horizon is the mass of the black hole @xmath39 where @xmath40 is the radius of apparent horizon given by equation ( [ eq : ah ] ) .
we now apply the perturbation formalism .
it is convenient to analyze perturbations to the self - similar solution in new variables natural to the problem . for this purpose
we introduce a `` spatial '' coordinate @xmath41 and a `` scaling variable '' @xmath42 @xmath43 , \ \
s = - \ln(-u),\ ] ] with the inverse transformation @xmath44 the signs are chosen to make the arguments of the logarithm positive in the region of interest ( @xmath45 , @xmath46 ) , where the field evolution occurs . in these
coordinates the metric ( [ eq : metric ] ) becomes @xmath47 + r^2\ , d\omega^2,\ ] ] and the critical solution ( [ eq : crit ] ) is simply @xmath48 we now consider small perturbations of critical solution ( [ eq : crit : xs ] ) .
quite generally , the perturbation modes can be written as @xmath49 which amounts to doing laplace transformation on general perturbations dependent on both @xmath41 and @xmath42 .
it is understood that there can be many perturbation modes , with distinct @xmath50 , @xmath51 and @xmath52 for each ( possibly complex ) @xmath53 in the perturbation spectrum . to recover a general perturbation
we must take the sum of these modes .
modes with @xmath54 ( called relevant modes ) grow and eventually lead to black hole formation , while modes with @xmath55 decay and are irrelevant .
let us discuss how the mass - scaling exponent @xmath6 is related to the perturbation spectrum .
the function @xmath56 , defined by equation ( [ eq : ah ] ) , takes the form @xmath57 where @xmath58 is the self - similar background term .
recall that @xmath56 vanishes at the apparent horizon .
even thought there is no apparent horizon in critical solution , i.e. @xmath59 , the exponentially growing perturbation will eventually make @xmath56 zero , leading to black hole formation .
the mode with the largest eigenvalue , @xmath60 , will clearly dominate , giving the position of the apparent horizon @xmath61 by @xmath62 here we made use of the fact that @xmath63 to the zeroth order in @xmath1 .
therefore the dependence of the black hole mass on @xmath1 , taken to be the control parameter , is @xmath64 this is precisely the mass - scaling relation ( [ eq : scaling ] ) with exponent @xmath65 . for a more detailed discussion of this approach and its validity see ref . .
perturbing the einstein - scalar equations ( [ eq : es ] ) by a mode ( [ eq : pert ] ) , we obtain a system of linear partial differential equations .
after a change of variables @xmath66 to @xmath67 , the @xmath42-dependence separates , and we are left with a following system of linear ordinary differential equations for @xmath50 , @xmath51 , and @xmath52 in the independent variable @xmath41 : [ eq : p ] @xmath68 @xmath69 @xmath70 @xmath71 @xmath72 here a prime denotes the derivative with respect to @xmath41 .
boundary conditions for the system ( [ eq : p ] ) are specified at @xmath73 and @xmath74 .
we require that perturbations grow slower than background at large @xmath41 ( i.e. that @xmath50 and @xmath51 are bounded , and @xmath52 grows slower than @xmath41 ) for the perturbation expansion to be valid .
because the solution must be continuous at the hypersurface @xmath23 , we have @xmath75 .
since we used coordinate freedom in @xmath10 to set @xmath8 to zero on that surface , we also have @xmath76 .
thus , the boundary conditions are @xmath77 equations ( [ eq : p ] ) together with boundary conditions ( [ eq : bc ] ) constitute our eigenvalue problem . the infinitesimal coordinate transformation @xmath78 corresponding to the remaining coordinate freedom in @xmath11 will give rise to unphysical gauge modes @xmath79 , \nonumber\\ \delta r & = & \frac{\zeta(v)}{(1-v / u)^{1/2 } } = e^{-x } \zeta[e^{-s}(e^{2x}-1 ) ] , \nonumber\\ \delta\phi & = & \frac{\zeta(v)}{(u - v ) } = - e^{s-2x } \zeta[e^{-s}(e^{2x}-1)].\end{aligned}\ ] ] since on a @xmath23 hypersurface @xmath80 , @xmath81 , and @xmath82 must vanish identically , @xmath83 , so a taylor expansion of @xmath84 around zero starts from @xmath85 term .
therefore these gauge modes correspond to negative eigenvalues @xmath53 , and are irrelevant .
observe that there is an obvious solution of perturbation equations ( [ eq : p ] ) , which can be obtained from self - similar solution ( [ eq : ss_soln ] ) by expanding around @xmath86 @xmath87 + \frac{p}{2}\ , \frac{v^2 - 2uv}{u^2 - uv } = x + p \sinh 2x.\end{aligned}\ ] ] clearly , the self - similar mode has eigenvalue @xmath88 , hence any relevant mode will dominate the calculation of mass - scaling exponent .
the perturbation equations ( [ eq : p ] ) inherit their structure from general einstein - scalar equations ( [ eq : es ] ) ; equations ( [ eq : p:3],[eq : p:4],[eq : p:7 ] ) are dynamical equations , while equations ( [ eq : p:5],[eq : p:6 ] ) are constraints .
equation ( [ eq : p:4 ] ) is redundant @xmath89/2,\ ] ] so it will be discarded .
the simple integrable form of equation ( [ eq : p:5 ] ) allows elimination of one of the unknowns , say @xmath50 , from the other equations by @xmath90 thus we are left with two second order ode s for @xmath51 and @xmath52 : [ eq : pe ] @xmath91 @xmath92 with boundary conditions ( [ eq : bc ] ) , and one constraint ( [ eq : p:6 ] ) , which we equivalently rewrite as first order ode with the use of equation ( [ eq : p:3 ] ) @xmath93 equations ( [ eq : pe:1],[eq : pe:2 ] ) can be separated with introduction of the auxiliary function @xmath94 , reducing to [ eq : se ] @xmath95 @xmath96 we convert this into algebraic form using change of variable @xmath97 finally , equations ( [ eq : se ] ) and constraint ( [ eq : pe : c ] ) become [ eq : ae ] @xmath98\ , \frac{dz}{dy } - ( k/2 - 1 ) z = c,\ ] ] @xmath99\ , \frac{dr}{dy } - k r = c - z,\ ] ] @xmath100 = 0,\ ] ] and boundary conditions ( [ eq : bc ] ) are specified at @xmath101 and @xmath102 by @xmath103 `` grow slowly '' means that @xmath104 must be bounded , and @xmath105 grows at most logarithmically , with @xmath106 bounded , as @xmath107 .
the last condition follows from @xmath50 being bounded at infinity .
equation ( [ eq : ae : c ] ) is indeed a constraint , since @xmath108 , and is automatically satisfied for all @xmath109 if @xmath110 = 0\ ] ] is satisfied initially at @xmath111 . imposing initial conditions ( [ eq : abc ] ) , the constraint ( [ eq : constr ] ) yields @xmath112 i.e. @xmath113 unless @xmath114 . observe that equations ( [ eq : ae:1 ] ) and ( [ eq : ae:2 ] ) are inhomogeneous hypergeometric equations .
their properties are well - known and described in a number of books on differential equations ( see , for example , @xcite ) , so the system ( [ eq : ae ] ) can be analytically solved and the perturbation spectrum exactly determined .
equation ( [ eq : ae:1 ] ) is the homogeneous ( except when @xmath114 ) hypergeometric equation @xmath115\ , \frac{dz}{dy } - ab z = 0\ ] ] with coefficients @xmath116 it has singular points at @xmath117 , and its general solution is a linear combination of any two different solutions from the set @xmath118 where @xmath119 is the hypergeometric function ; @xmath120 .
any three of the functions ( [ eq : gen : z ] ) are linearly dependent with constant coefficients , for example @xmath121 functions @xmath122 and @xmath123 have the asymptotic behavior @xmath124 the case when @xmath114 is special . in this case
the solution ( [ eq : gen : z ] ) degenerates and @xmath125 identically . in order to deal with this situation
we continue to use @xmath126 as earlier , but denote by @xmath127 a solution independent of @xmath126 .
it is easy to verify that @xmath127 is logarithmically divergent at @xmath101 .
note also that for @xmath114 we can have @xmath128 , so that the equation ( [ eq : ae:1 ] ) is inhomogeneous , and we must add the particular solution @xmath129 to the general solution above .
observe that at @xmath101 , @xmath127 diverges as a power for @xmath130 , logarithmically for @xmath114 , does not have a limit for @xmath131 , and converges ( to zero ) only for @xmath132
. the solutions that satisfy the initial condition @xmath133 are @xmath134 and @xmath135 . from equation ( [ eq : hyper : connect ] ) we see that @xmath127 ( and @xmath126 ) is connected to @xmath136 , @xmath137 by a linear relation with non - zero coefficients , so the boundary conditions at infinity will only be satisfied if both @xmath136 , @xmath137 do not blow up , i.e. if @xmath138 , @xmath139 .
the curve @xmath140 , written in terms the real and imaginary parts of @xmath53 , has the form @xmath141 which divides the complex @xmath53-plane into two regions shown in fig .
[ fig : spectrum ] ; to the left of the curve @xmath142 , and to the right @xmath138 .
so to satisfy all boundary conditions ( [ eq : abc ] ) , @xmath53 must lie in the shaded region @xmath143 of the complex plane , and for any @xmath144 , @xmath145 is a solution of equation ( [ eq : ae:1 ] ) with boundary conditions ( [ eq : abc ] )
. equation ( [ eq : ae:2 ] ) is also hypergeometric , with coefficients @xmath146 the general solution of the homogeneous equation is @xmath147 and the solution of inhomogeneous equation is easily constructed from ( [ eq : gen : r ] ) by @xmath148.\ ] ] the limits of integration are chosen so that the boundary conditions at @xmath101 are satisfied .
asymptotic ( [ eq : asympt : z ] ) and equation ( [ eq : soln : r ] ) give the behavior @xmath149 clearly , @xmath150 . also , @xmath51 is bounded at infinity since @xmath151 is , for the constraint ( [ eq : ae : c ] ) requires @xmath152 .
therefore , ( [ eq : soln : z ] ) and ( [ eq : soln : r ] ) for @xmath144 is a solution of perturbation equations ( [ eq : ae ] ) with boundary conditions ( [ eq : abc ] ) , and the region @xmath143 is the perturbation spectrum , which turns out to be continuous .
we have perturbed the continuously self - similar critical solution of the gravitational collapse of a massless scalar field ( roberts solution ) , and solved the perturbation equations exactly .
the perturbation spectrum was found to be not discrete , but continuous , and occupying the region of the complex plane @xmath153 which , according to , suggests non - universality of critical behavior for different ingoing wavepackets .
the complex oscillatory modes might lead to decay of a continuously self - similar solution ( [ eq : crit ] ) into discrete self - similar choptuon observed in @xcite .
the eigenvalues @xmath53 approach @xmath154 , which corresponds to the mass - scaling exponent @xmath0 .
this is different from the exponent @xmath155 found in using the self - similar solution .
this research was supported by natural sciences and engineering research council of canada .
i am grateful to d. n. page for interest in this work and stimulating discussions . | this paper studies the perturbations of the continuously self - similar critical solution of the gravitational collapse of a massless scalar field ( roberts solution ) .
the perturbation equations are derived and solved exactly .
the perturbation spectrum is found to be not discrete , but occupying continuous region of the complex plane .
the renormalization group calculation gives the value of the mass - scaling exponent @xmath0 . |
a meromorphic map @xmath0 of a compact khler manifold @xmath1 induces a well - defined pullback action @xmath2 for each @xmath3 .
the @xmath4-th dynamical degree @xmath5 describes the asymptotic growth rate of the action of iterates of @xmath6 on @xmath7 .
originally , the dynamical degrees were introduced by friedland @xcite and later by russakovskii and shiffman @xcite and shown to be invariant under birational conjugacy by dinh and sibony @xcite .
note that dynamical degrees were originally defined with the limit in ( [ eqn : def_dyn_deg ] ) replaced by @xmath8 .
however , it was shown in @xcite that the limit always exists .
one says that @xmath6 is _ cohomologically hyperbolic _ if one of the dynamical degrees is strictly larger than all of the others . in this case
, there is a conjecture @xcite that describes the ergodic properties of @xmath6 .
this conjecture has been proved in several particular sub - cases @xcite .
what happens in the non - cohomologically hyperbolic case ?
if a meromorphic map preserves a fibration , then there are nice formulae relating the dynamical degrees of the map @xcite .
this is the case in the following two examples : 1 .
it was shown in @xcite that bimeromorphic maps of surfaces that are not cohomologically hyperbolic ( @xmath9 ) always preserve an invariant fibration .
meromorphic maps that are not cohomologically hyperbolic arise naturally when studying the spectral theory of operators on self - similar spaces @xcite .
all the examples studied in that context preserve an invariant fibration . in several cases ,
this fibration made it significantly easier to compute the limiting spectrum of the action .
based on this evidence , guedj asked in @xcite whether every non - cohomologically hyperbolic map preserves a fibration .
we will prove : [ thm : main ] the rational map @xmath10 given by @xmath11\ = \ [ -y^2:x(x - z):-(x+z)(x - z)]\ ] ] is not cohomologically hyperbolic ( @xmath12 ) , and no iterate of @xmath6 preserves a singular holomorphic foliation .
moreover , for a baire generic set of automorphisms @xmath13 , the composition @xmath14 has the same properties . since preservation of a fibration is a stronger condition than preservation of a singular foliation
, @xmath6 provides an answer to the question posed by guedj . after reading a preliminary version of this paper ,
charles favre asked is the same behavior can be found for a polynomial map of @xmath15 . in @xcite , there is a list of seven types of non - cohomologically hyperbolic polynomial mappings .
our method of proof does not apply in most of these examples for trivial reasons , but we found a map of type ( 3 ) for which the same result holds : [ thm2 ] the polynomial map @xmath16 extends as a rational map @xmath17 that is not cohomologically hyperbolic ( @xmath18 ) , and no iterate of @xmath19 preserves a singular holomorphic foliation on @xmath20 .
non - cohomologically hyperbolic maps of @xmath21-dimensional manifolds @xmath1 arise naturally as certain pseudo - automorphisms that are `` reversible '' on @xmath22 @xcite . for these mappings
it follows from poincar duality that @xmath23 .
recently , bedford , cantat , and kim @xcite have found a reversible pseudo - automorphism of an iterated blow - up of @xmath24 which does not preserve any invariant foliation .
it also answers the question posed by guedj .
many authors have studied meromorphic ( and rational ) maps that preserve foliations and algebraic webs , including @xcite .
since the proof of theorem [ thm : main ] is self - contained and provides insight on the mechanism that prevents @xmath6 from preserving a foliation , we will provide a direct proof rather than appealing to results from these previous papers .
let us give a brief idea of the proof of theorem [ thm : main ] .
the fourth iterate @xmath25 has an indeterminate point @xmath4 that is blown - up by @xmath25 to a singular curve @xmath26 .
any foliation @xmath27 must be either generically transverse to @xmath26 or have @xmath26 as a leaf .
this allows us to show that @xmath28 must be singular at either at @xmath4 or at some preimage @xmath29 of the singular point .
both of these points have infinite @xmath25 pre - orbits , at each point of which @xmath30 is a finite map .
this generates a sequence of distinct points @xmath31 at such that @xmath32 is singular at @xmath33 . if @xmath34 for some @xmath35 this implies that @xmath27 is singular at infinitely many points , providing a contradiction .
the same method of proof applies for the polynomial map in theorem [ thm2 ] .
our proof of theorem [ thm2 ] relies on the compactness of @xmath20 .
does @xmath19 preserve a foliation on @xmath15 ?
if the foliation is algebraic , then it extends to a foliation of @xmath36 , and the answer follows from theorem [ thm2 ] .
if the foliation does not extend because it has an infinite set of singularities in @xmath37 that accumulate to the line at infinity , then we do not yet know how to address the situation
. it would also be interesting to place the maps @xmath38 in the context of the conjecture from @xcite : do these maps have topological entropy @xmath39 ?
can one find a ( unique ) measure of maximal entropy ?
as @xmath40 , what is the asymptotic behavior of the periodic points of period @xmath41 , and are the periodic points predominantly saddle - type , repelling , or degenerate ? in [ sec : background ] we provide background on the transformation of foliations and fibrations by rational maps . in [ sec : structure ] we describe some basic properties of @xmath6 and we prove that @xmath42 , showing that @xmath6 is not cohomologically hyperbolic . in [ sec : proof1 ] we prove that no iterate of @xmath6 preserves a foliation , and conclude the section with a summary of the properties of @xmath6 that make the proof of theorem [ thm : main ] work . in [ sec : proof2 ] we prove theorem [ thm2 ] by computing that @xmath43 and verifying that @xmath19 has the properties listed in [ sec : proof1 ] .
the last section , [ sec : proof3 ] , deals with the case of generic rotations of @xmath6 , completing the proof of theorem [ thm : main ] .
* acknowledgements .
* we have benefited greatly from discussions with eric bedford , serge cantat , laura demarco , jeffrey diller , tien - cuong dinh , and charles favre .
this work was supported in part by the national science foundation grants dge-0742475 ( to s.r.k ) , dms-1102597 ( to r.k.w.r . ) , dms-1348589 ( to r.k.w.r . ) , and iupui startup funds ( to r.a.p . and r.k.w.r . ) .
in this section we will present some background about singular holomorphic foliations on complex surfaces and their pullback under rational maps .
while these results are well - known to experts , we include them here to make this paper self - contained . more details can be found in @xcite .
the _ standard holomorphic foliation on @xmath44 _ is the representation of the bidisk @xmath44 as the disjoint union of complex one - dimensional disks @xmath45 let @xmath1 be a ( potentially non - compact ) complex surface .
a _ holomorphic foliation _
@xmath27 on @xmath1 is the partition @xmath46 of @xmath1 into the disjoint union of connected subsets @xmath47 , which is locally biholomorphically equivalent to the standard holomorphic foliation on @xmath44 .
each @xmath47 is a riemann surface called a _ leaf _ of @xmath27 .
a _ singular holomorphic foliation _ on @xmath1 is a foliation @xmath27 on @xmath48 , where @xmath49 is a discrete set of points through which @xmath27 does not extend .
the points of @xmath49 are called the _ singular points _ of @xmath27 .
we will abuse notation and denote the corresponding singular foliation by @xmath27 , as well .
it is easy to write @xmath27 as the integral curves of some holomorphic vector field in a neighborhood of any @xmath50 .
this can also be done in a neighborhood of any singular point : [ thm : extension ] in a neighborhood of each singular point @xmath51 , @xmath27 is generated as the integral curves of some holomorphic vector field .
an alternate proof of this result using sheaf theory and hartog s theorem can be found in @xcite . for the remainder of the article we will use the word `` foliation '' to mean singular holomorphic foliation .
theorem [ thm : extension ] allows for any foliation @xmath27 on a surface @xmath1 to be described in the following two equivalent ways : * by an open cover @xmath52 of @xmath1 and a system of holomorphic vector fields @xmath53 on @xmath54 with isolated zeros that satisfy the compatibility condition that @xmath55 for some non - vanishing holomorphic functions @xmath56 * by an open cover @xmath52 of @xmath1 and a system of holomorphic one - forms @xmath57 on @xmath54 with isolated zeros that satisfy the compatibility condition that @xmath58 for some non - vanishing holomorphic functions @xmath59 if @xmath60 satisfies the compatibility condition in ( i ) we will call it a _ compatible system of vector fields _ , and if @xmath61 satisfies the compatibility condition in ( ii ) we will call it a _ compatible system of one - forms_. within a given @xmath54 the tangent direction to a leaf is described by either @xmath53 or the kernel of @xmath57 .
the zeros of @xmath53 or @xmath57 , correspond to the singular points of @xmath27 .
[ lem : equal ] suppose @xmath27 and @xmath62 are foliations on a surface @xmath1 that are equal outside of some analytic curve @xmath26
. then , @xmath63 .
it suffices to prove the statement locally , so we can suppose that @xmath1 is an open subset of @xmath15 .
the tangent lines to leaves of @xmath27 define a meromorphic function @xmath64 , whose indeterminate points are precisely the singular points @xmath49 of @xmath27 . similarly , @xmath62 defines a meromorphic function @xmath65 , whose indeterminate points are precisely the singular points @xmath66 of @xmath62 . both @xmath67 and @xmath68 are holomorphic on @xmath69 and equal on the connected set @xmath70 .
thus , by uniqueness properties of analytic functions they are equal on @xmath69 .
moreover , @xmath71 , since otherwise , one of the functions would define an extension of the other function through a point of its indeterminacy .
let @xmath72 be a dominant holomorphic map between two surfaces and let @xmath27 be a foliation on @xmath73 given by @xmath61 .
the _ pullback _
@xmath74 is defined by @xmath75 where each form @xmath76 is obtained by rescaling @xmath77 , i.e. , dividing it by a suitable holomorphic function in order to eliminate any non - isolated zeros . as an illustration ,
consider @xmath78 given by @xmath79 then , @xmath80 so that @xmath81 .
one can easily have a singular foliation @xmath27 whose pullback under a holomorphic map is a non - singular foliation .
for example , the radial foliation , given by the choice of @xmath82 in ( [ pullbackexample ] ) pulls back under the blow - down @xmath83 to the horizontal foliation given by @xmath84 .
this is impossible if the map is finite : [ preimage_singular ] let @xmath72 be a dominant holomorphic map between complex surfaces that is finite at @xmath85 and let @xmath27 be a foliation on @xmath73 . if @xmath86 is a singular point for @xmath27 then @xmath4 is a singular point for @xmath87 .
suppose that @xmath74 is non - singular at @xmath4 .
then , we can choose local coordinates @xmath88 centered at @xmath4 so that @xmath74 is given by @xmath89 .
let @xmath90 be local coordinates centered at @xmath86 . in these coordinates
, we have @xmath91 where @xmath92 .
suppose @xmath93 is given by @xmath94 with @xmath95 and @xmath96 having an isolated common zero at @xmath97 .
since @xmath89 , the @xmath98 coefficient of @xmath99 must vanish identically : @xmath100 we will show that this is impossible . since @xmath6 is dominant , neither @xmath101 nor @xmath102 can be identically zero . thus , @xmath103 if and only if @xmath104 , and hence , neither can vanish identically since @xmath6 is dominant . moreover , since @xmath6 is finite , @xmath101 and @xmath102 have an isolated common zero at @xmath105 .
this implies that @xmath106 vanishes along the same curves as @xmath107 with at least the same multiplicity ( and similarly for @xmath108 along the same curves as @xmath109 ) .
therefore , if @xmath110 denotes the order of vanishing of a holomorphic function @xmath67 at point @xmath111 , we have : @xmath112 and a similar inequality holds for @xmath113 . note
that @xmath114 and @xmath115 do not both vanish on @xmath116 because @xmath6 does not collapse curves .
therefore , at least one of @xmath114 and @xmath115 has a monomial consisting of a positive power of @xmath117 in its power series .
let @xmath118 be the minimal exponent of an @xmath119 monomial in either series , and without loss of generality assume it is present in the power series of @xmath114 .
let @xmath120 and consider @xmath121 given by @xmath122 since the vertical foliation given by @xmath123 is preserved under @xmath124 , we also have that @xmath125 is the vertical foliation .
moreover , since @xmath19 is a composition of finite , dominant maps , ( [ vanishing1 ] ) holds for @xmath19 . on the other hand ,
@xmath126 was chosen so that @xmath127 with @xmath119 the only term of degree @xmath35 in the expansion for @xmath128 .
in particular , @xmath129 which contradicts that ( [ vanishing1 ] ) applies to @xmath19 . for the remainder of the paper , we will restrict our attention to complex projective algebraic surfaces , so that we can discuss rational maps .
we will refer to them simply as `` surfaces '' .
let us make the convention that all rational maps are _ dominant _ , meaning that the image is not contained within a proper subvariety of the codomain .
let @xmath130 be a rational map of surfaces and let @xmath4 be a point from the indeterminacy set @xmath131 .
a sequence of point blow - ups @xmath132 _ resolves the indeterminacy at @xmath4 _ if @xmath6 lifts to a rational map @xmath133 which is holomorphic in a open neighborhood of @xmath134 and makes the following diagram commute @xmath135^\pi \ar @{-->}[dr]^{\widetilde{\varphi } } & \\ x \ar @{-->}[r]^\varphi & y , } \end{aligned}\ ] ] wherever @xmath136 and @xmath137 are both defined .
it is a well - known fact that one can do a sequence of point blow - ups @xmath138 over @xmath131 so that @xmath6 lifts to a holomorphic map @xmath139 resolving all of the indeterminate points of @xmath131 .
see , for example , ( * ? ? ?
iv , 3.3 ) . let @xmath140 be a dominant rational map and let @xmath27 be a foliation on @xmath73 .
the _ pullback of @xmath27 under @xmath6 _ is defined by @xmath141 , where @xmath142 is a sequence of blow - ups over @xmath131 and @xmath143 is a holomorphic map resolving all of the indeterminacy of @xmath6 .
the push - forward @xmath144 is defined as follows .
the inverse @xmath145 is a birational map whose restriction @xmath146 is a biholomorphism onto its image .
any foliation@xmath62 on @xmath147 has finitely many singular points since @xmath147 is compact .
thus , @xmath148 is a foliation on @xmath149 having finitely many singular points .
since @xmath131 is also a finite set , the result is a singular holomorphic foliation on @xmath1 .
more specifically , at any @xmath150 , the pull - back @xmath151 will either extend through @xmath152 as a regular holomorphic foliation or , by definition , @xmath152 will be a singular point .
the result is called @xmath153 . using theorem [ thm : extension ] , one can re - express @xmath153 using a compatible system of vector fields @xmath60 or a compatible system of one - forms @xmath61 . [ rem : pullback ] the definition of @xmath74 given above coincides with the foliation obtained by extending @xmath154 through @xmath131 , so it is independent of the choice of blow - up .
the blow - up is used to make it clear that the singular points of @xmath155 do not accumulate on @xmath131 .
we will use the following three well - known lemmas .
[ lem_composition ] let @xmath130 and @xmath156 be any dominant rational maps .
for any foliation @xmath27 on @xmath157 we have @xmath158 .
the foliations @xmath159 and @xmath160 agree on @xmath161 .
since @xmath162 is at most an analytic curve , the result then follows from lemma [ lem : equal ] .
[ lem_leaf_or_transv ] let @xmath27 be a foliation on a surface @xmath1 and suppose @xmath163 is an irreducible algebraic curve .
then , either @xmath26 is a leaf of @xmath27 or @xmath26 is transverse to @xmath27 away from finitely many points .
let @xmath164 be the normalization of @xmath26 .
( for background on normalization , see ( * ? ? ?
ii , 5 ) . ) since @xmath26 is irreducible , @xmath165 is a connected riemann surface .
let @xmath61 be a compatible system of one - forms describing @xmath27 .
then , the zeros of @xmath166 are a well - defined analytic subset of @xmath165 .
any point of @xmath167 where @xmath26 is parallel to @xmath27 corresponds to the image under @xmath19 of such a zero , so the result follows .
[ curve_trans_foliation]let @xmath130 be a dominant rational map of surfaces , let @xmath168 , and suppose @xmath132 and @xmath169 give a resolution of the indeterminacy at @xmath4 .
if @xmath27 is a foliation on @xmath73 and one of the irreducible components of @xmath170 is generically transverse to @xmath27 , then @xmath171 is singular at @xmath4 .
let @xmath26 be an irreducible component of @xmath170 that is generically transverse to @xmath27 and suppose @xmath27 is represented by a compatible system of one - forms @xmath61 . then , the tangent spaces to generic points of @xmath26 are not in the kernels of any of the one - forms based at these points .
let @xmath172 be an irreducible component that is mapped by @xmath136 onto @xmath26 .
then , at generic @xmath173 we have that @xmath152 and @xmath174 are smooth points of @xmath175 and @xmath26 , respectively , and @xmath176 is an isomorphism between their tangent spaces .
this implies that @xmath177 are non - vanishing at generic points of @xmath175 and have kernel transverse to @xmath175 at these points .
thus , @xmath178 is generically transverse to @xmath175 .
now we can choose two disjoint holomorphic disks @xmath179 and @xmath180 within leaves of @xmath178 that intersect @xmath134 transversally at two distinct points of @xmath181 .
then , @xmath182 and @xmath183 are distinct integral curves of @xmath184 going through @xmath4 .
this implies that @xmath4 is a singular point for @xmath74 .
a _ fibration _ is a dominant rational map @xmath185 to a non - singular algebraic curve @xmath186 with the property that each fiber is connected .
the level curves of @xmath187 define a foliation on @xmath188 having finitely many singularities .
it thus extends as a singular foliation @xmath27 on @xmath1 called the _ foliation induced by @xmath187_. [ fibration ] let @xmath185 be a fibration inducing a foliation @xmath27 on @xmath1 . a rational map @xmath0 preserves @xmath27 if and only if @xmath187 semi - conjugates @xmath6 to a holomorphic map of @xmath186 : @xmath189^\varphi \ar @{-->}[d]^\rho & x \ar @{-->}[d]^\rho \\ s \ar[r]^\eta & s. } \end{aligned}\ ] ] suppose that @xmath187 semi - conjugates @xmath6 to a holomorphic map @xmath190 .
we will show that @xmath191 . by lemma [ lem : equal ] , it suffices to prove equality in the complement of a finite set of curves and points . if @xmath192 is any system of coordinates on @xmath186 , then @xmath27 is defined on @xmath188 by the compatible system of one - forms @xmath193 . for any @xmath194 we can express @xmath27 by @xmath195 , where @xmath196 is some chart defined in the neighborhood of @xmath197 , since then @xmath198 serves as a chart in the neighborhood of @xmath199 .
meanwhile , commutativity of diagram ( [ eqn : cd ] ) implies that in a neighborhood of @xmath152 we have @xmath200 thus , @xmath27 and @xmath74 agree in a neighborhood of @xmath152 .
now suppose that @xmath0 preserves the foliation @xmath27 .
it is possible that @xmath6 collapses @xmath201 into @xmath202 for some finite set @xmath203 . for any @xmath204
we define @xmath205 where @xmath152 is any element of @xmath206 . since the fibers of @xmath187 are connected , @xmath6 maps fibers to fibers .
therefore , this results in a well - defined function @xmath207 which makes ( [ eqn : cd ] ) commute wherever @xmath208 and @xmath209 are both defined . for any @xmath210
there is a holomorphic disc @xmath211 in @xmath1 intersecting @xmath212 transversally , that is small enough so that @xmath213 is one - to - one .
( we are again using that each fiber is connected . )
let @xmath214 be its inverse .
then , @xmath215 is a meromorphic ( hence holomorphic ) function that agrees with @xmath216 on @xmath217 .
we conclude that @xmath216 extends as a holomorphic map @xmath190 .
using lemma [ fibration ] , it follows immediately from theorem [ thm : main ] that no iterate of @xmath6 preserves a fibration .
recall that @xmath218 is given by @xmath11\ = \ [ -y^2:x(x - z):-(x+z)(x - z)].\ ] ] solving for the points where all three homogeneous coordinates of @xmath6 are zero , we find that the only indeterminate point of @xmath6 is @xmath219.\ ] ] we have @xmath220 , so the curves @xmath221 are critical .
let @xmath222 , and observe that @xmath223 & [ 1:0:0 ] \ar[r ] & [ 0:1:-1 ] \ar[r ] & [ -1:0:1 ] \ar[r ] & [ 0:1:0 ] \ar@(dl , dr)[lll ] } \ ] ] + we will refer to the points in this four cycle as @xmath224 , respectively .
meanwhile , @xmath225 & l_x \ar@(dr , dl)[l ] , } \ ] ] + as illustrated in figure [ lines ] .
note that the orbit of the collapsed curve @xmath226 lies in the four cycle ( [ fourcycle ] ) , which is disjoint from @xmath4 .
it follows that there is no curve @xmath227 such that @xmath228 , so @xmath6 is algebraically stable by ( * ? ? ?
therefore , @xmath229 .
we will show in the proof of lemma [ c_singularity ] that @xmath4 can be resolved by two blow - ups with the image of the exceptional divisors being the line @xmath230 .
in particular , @xmath231 $ ] is neither a critical value nor in the image of the indeterminacy , so it is a generic point for @xmath6 .
its two preimages under @xmath6 are @xmath232 $ ] , so @xmath233 .
this establishes the following resonance of dynamical degrees : [ lem : resonant_degrees ] @xmath234 so that @xmath6 is not cohomologically hyperbolic .
note also that @xmath235 fixes both @xmath236 and @xmath237 , with each line transversally superattracting under @xmath235 . in the local coordinate @xmath238
, @xmath239 is @xmath240 , the well - known basilica map , with period two superattracting cycle @xmath241 .
similarly , in a suitable local coordinate , @xmath242 is also the basilica map , with period two superattracting cycle @xmath243 .
finally , @xmath6 ( and hence @xmath235 ) has @xmath244 $ ] as a superattracting fixed point .
we obtain five superattracting fixed points for @xmath25 .
their basins are shown in figure [ real_slice ] .
is the fatou set of @xmath6 the union of the basin of the superattracting 4-cycle @xmath245 and the basin of the superattracting fixed point @xmath246 ?
proof of theorem [ thm : main ] relies on careful analysis of @xmath25 , the fourth iterate of @xmath6 .
the indeterminacy set of @xmath25 is @xmath247\ } \cup \{[0:\pm i : 1]\ } \cup \{[1:0:-1\pm i]\ } \cup \{[0:i:\pm \sqrt{1 \pm i}]\}.\end{gathered}\ ] ] however , we will only need to resolve the indeterminacy of @xmath25 at @xmath4 .
let @xmath248 be the blow - up @xmath249 at @xmath4 with exceptional divisor @xmath250 , and let @xmath251 denote the proper transform of @xmath226 .
then let @xmath252 be the blow - up of @xmath248 at the point @xmath253 , resulting in a new exceptional divisor @xmath254 .
we will abuse notation by denoting the proper transform of @xmath250 in @xmath252 also by @xmath250 .
[ c_singularity ] the map @xmath255 lifts to a rational map @xmath256 that resolves the indeterminacy of @xmath25 at @xmath4 .
we have that @xmath257 = : s$ ] and that @xmath258 is an irreducible algebraic curve @xmath259 of degree @xmath260 that is singular at @xmath261 .
we will first show that @xmath6 lifts to a rational map @xmath262 that resolves the indeterminacy of @xmath6 at @xmath4 , satisfying @xmath263 $ ] and @xmath264 .
we will do calculations in several systems of local coordinates in a neighborhood of @xmath250 and @xmath254 .
a summary is shown in figure [ blowup ] .
consider the two systems of affine coordinates on @xmath249 given by @xmath265 and @xmath266 .
let @xmath267 , so @xmath268 are coordinates that place @xmath4 locally at the origin .
writing @xmath269 , we have @xmath270 there are two systems of coordinates in a neighborhood of @xmath250 within @xmath271 .
they are @xmath272 , where @xmath273 , and @xmath274 , where @xmath275 . then , @xmath6 lifts to a rational map @xmath276 which is expressed in these coordinates by @xmath277 the exceptional divisor @xmath250 is given in the first set of coordinates by @xmath278 and in the second set of coordinates by @xmath279 .
it is clear from ( [ eqn : blowup1 ] ) that @xmath280 extends holomorphically to all points of @xmath250 other than @xmath281 , sending each of these points to @xmath282 . in the @xmath268
coordinates , @xmath226 is given by @xmath283 .
thus , in the @xmath272 coordinates , the proper transform @xmath251 is given by @xmath284 so that @xmath253 is given by @xmath97 , the indeterminate point for @xmath280 on @xmath250 .
we now blow this point up using two new systems of coordinates in a neighborhood of the new exceptional divisor @xmath254 .
they are @xmath285 , where @xmath286 , and @xmath287 , where @xmath288 .
the map @xmath289 lifts to a map @xmath136 , which can be expressed in local coordinates as @xmath290 this shows that @xmath291 is holomorphic in a neighborhood of @xmath254 .
thus , it is holomorphic in a neighborhood of @xmath292 .
since @xmath254 is given in these systems of coordinates by @xmath278 and @xmath281 , respectively , one can see from ( [ eqn : blowup2 ] ) that @xmath264 .
notice that the line @xmath293 passes through no points of @xmath294,[0:\pm i : 1],[1:0:-1\pm i]\right\}.\end{aligned}\ ] ] this implies that @xmath295 is holomorphic in a neighborhood of @xmath292 ; i.e. , that @xmath296 resolves the indeterminacy of @xmath25 at @xmath4 .
since @xmath263 $ ] it is easy to check that @xmath297 : = s$ ] .
meanwhile , @xmath298 is the projective line @xmath299 .
its forward image under @xmath300 is an algebraic curve @xmath301 of degree @xmath302 . since @xmath299 does not intersect @xmath303 , @xmath259 is irreducible .
it remains to show that @xmath259 is singular at @xmath261 .
notice that @xmath6 maps a neighborhood of @xmath304 $ ] biholomorphically to a neighborhood of @xmath261 .
thus , it will be sufficient to show that @xmath305 is singular at @xmath304 $ ] .
see figure [ curves ] for plots of the real slices of @xmath306 and @xmath259 .
let us work in the @xmath307 local coordinates .
we have @xmath308 if we parameterize @xmath299 by @xmath309 and then compose with @xmath235 , we obtain the following parameterization of @xmath306 : @xmath310 to see that @xmath306 is singular at @xmath311 , it suffices to check that @xmath306 intersects any line through @xmath311 with intersection number @xmath312 .
such a generic line is given by @xmath313 . substituting ( [ param ] ) into this equation
, we obtain @xmath314 which clearly has a double zero at @xmath315 , implying that the intersection number of @xmath306 with the line given by @xmath313 is at least two . applying @xmath6 to the parameterization of @xmath306 given in ( [ param ] )
one gets a parameterization of @xmath259 .
one can then use grbner bases ( * ? ? ?
* ch . 3 ,
3 ) to generate the following implicit equation for of @xmath259 , which is expressed in the local coordinates @xmath316 centered at @xmath261 : @xmath317 since the lowest degree terms of ( [ bigeqn ] ) are of degree @xmath318 , this gives an alternative computational proof that @xmath259 is singular at @xmath261 .
implicitly throughout the proof we assume that @xmath319 for @xmath320 , which follows from lemma [ lem_composition ] .
also note that the whole proof will take place in the coordinates @xmath321 .
the @xmath6-preimage of @xmath226 is the circle @xmath322 which intersects @xmath237 at the indeterminate point @xmath323 and at @xmath324 .
it follows that @xmath235 is a finite holomorphic map at each point of @xmath237 other than @xmath4 and @xmath325 , and consequently , @xmath25 is a finite holomorphic map at every point of @xmath237 except at the finite set of points @xmath326 consisting of @xmath4 , @xmath325 , and their @xmath235-preimages .
one can check that the point @xmath327 is a @xmath25-preimage of the singular point @xmath261 of the blow - up curve @xmath26 .
we will first show for any foliation @xmath27 that @xmath328 is singular either at the indeterminate point @xmath4 or at @xmath329 . by lemma [ lem_leaf_or_transv ] , either @xmath26 is a leaf of @xmath27 or generically transverse to @xmath27 . in the second case , lemma [ curve_trans_foliation ]
immediately implies that @xmath28 is singular at @xmath4 .
on the other hand , if @xmath26 is a leaf of @xmath27 , @xmath261 must be a singular point for @xmath27 , since @xmath261 is a singular point of @xmath26 by lemma [ c_singularity ] . since @xmath25 is a finite holomorphic map at @xmath329 and @xmath330 , by lemma [ preimage_singular ] , @xmath28 is singular at @xmath329 as well .
note that neither @xmath4 nor @xmath329 are critical points of @xmath331 , so they are not exceptional points .
one can check that any preorbit of @xmath4 or @xmath329 under @xmath332 is disjoint from @xmath333 .
let @xmath334 be the point ( @xmath4 or @xmath329 ) where @xmath28 is singular .
if we denote some @xmath332-preorbit of @xmath334 by @xmath335 , then for each @xmath336 , @xmath337 will be singular at @xmath338 .
now suppose the foliation @xmath27 is preserved by @xmath339 for any positive integer @xmath35 .
then @xmath340 is singular at @xmath341 for all @xmath342 .
this implies @xmath27 has infinitely many singular points , giving a contradiction .
* observation : * it was more convenient to explain the proof with a specific map ; however the proof holds in greater generality . in fact , we have established [ prop : conditions ] assume the map @xmath343 satisfies the following conditions : 1 .
there exists @xmath344 and some iterate @xmath126 so that @xmath345 blows up @xmath4 to a singular curve @xmath26 .
@xmath4 has an infinite preorbit along which @xmath216 is a finite holomorphic map .
3 . there is a singular point @xmath346 having an infinite preorbit along which @xmath216 is a finite holomorphic map .
then no iterate of @xmath216 preserves a foliation .
note that one can replace conditions ( 2 ) and ( 3 ) with the equivalent condition that @xmath4 and @xmath261 have infinite @xmath345-preorbits along which @xmath345 is a finite holomorphic map .
the extension of @xmath347 to @xmath20 is @xmath348 = [ x(x - y)+2z^2:(x+y)(x - y)+z^2:z^2].\end{aligned}\ ] ] let us first verify that @xmath19 is algebraically stable .
the indeterminate set consists of the point @xmath349,\end{aligned}\ ] ] and the critical set is @xmath350 one can check that @xmath351 satisfies that @xmath352 $ ] , which is periodic of period @xmath318 : @xmath353 \leftrightarrow [ 4:4:1].\ ] ] the line @xmath354 is not collapsed by @xmath19 ; using the coordinate @xmath355 , the restriction @xmath356 is given by @xmath357 . since no iterate of a collapsing line lands on the indeterminate point @xmath4 , it follows from ( * ? ? ?
* prop.1.4.3 ) that @xmath19 is algebraically stable ; thus , @xmath358 .
any point @xmath359 with @xmath360 has two preimages in @xmath15 given by @xmath361 since generic points of @xmath15 do not have a preimage on @xmath362 , @xmath363 ; thus @xmath19 is not cohomologically hyperbolic .
it suffices to verify that @xmath19 satisfies the conditions of proposition [ prop : conditions ] .
a sequence of two blow - ups at @xmath4 resolves the indeterminacy of @xmath19 .
the first blow - up results in a lift @xmath364 with exceptional divisor @xmath250 .
the point @xmath365 with coordinate @xmath366 is indeterminate for @xmath367 , and @xmath368 $ ] .
the second blow - up ( at @xmath111 ) resolves the indeterminacy , and the lifted map @xmath369 sends the new exceptional divisor @xmath254 to the line @xmath370 .
we omit the calculations as they are very similar to those in lemma [ c_singularity ] .
the indeterminacy set of @xmath371 is @xmath372\ } \cup \{[1:0:0]\ } \cup \{[1:-1:0]\}.\end{aligned}\ ] ] since @xmath373 is disjoint from @xmath299 , the map @xmath374 resolves the indeterminacy of @xmath375 at @xmath4 .
moreover , @xmath376 blows up @xmath4 to the curve @xmath377 we will now check that @xmath378 $ ] is a singular point on @xmath259 .
let us work in the affine coordinates @xmath379 .
if we parameterize @xmath299 by @xmath380 , then a parameterization of @xmath381 is obtained by substituting into the expression for @xmath382 : @xmath383 one can see that @xmath306 is singular at @xmath384 by using grbner bases to convert ( [ eqn : c2param ] ) into the following implicit equation for @xmath306 @xmath385 which is expressed in local coordinates @xmath386 where @xmath387 . since the lowest order terms are of degree @xmath318 , the point @xmath388 is singular for @xmath306 .
( this can also be shown using the parameterization of @xmath306 , like in the proof of lemma [ c_singularity ] . )
one can check that @xmath389 is invertible at @xmath384 so that @xmath390 is singular at @xmath391
. therefore , condition ( 1 ) from proposition [ prop : conditions ] holds .
since the action of @xmath19 on @xmath362 is given by @xmath357 , and @xmath4 is given in this coordinate by @xmath392 , we find that @xmath4 has an infinite pre - orbit under @xmath19 .
moreover , these preimages are disjoint from @xmath4 ( and hence from the collapsing line @xmath226 ) , so condition ( 2 ) of the proposition holds .
it remains to check condition ( 3 ) .
notice that @xmath393 .
therefore , it suffices to show that @xmath394 has an infinite pre - orbit in @xmath15 consisting of points not on @xmath299 .
for any @xmath395 , the two preimages given by ( [ eqn : preimages ] ) can not both be on @xmath299 , since in that case @xmath396 summing these equations yields @xmath397 , a contradiction .
therefore , any point of @xmath398 has an infinite pre - orbit disjoint from @xmath299 .
since all three conditions of proposition [ prop : conditions ] hold , we conclude that no iterate of @xmath19 preserves a foliation .
note that for any @xmath13 , we have @xmath399 . for any @xmath400 ,
let here , @xmath401 denotes the total transform of @xmath402 , i.e. @xmath403 , where @xmath25 blows up @xmath402 to @xmath299 .
are all algebraic . in order to conclude that @xmath408 is a dense open subset of @xmath409 we only need to show that this intersection is not empty .
notice that @xmath410 follows from our study of @xmath25 . for even @xmath411 , property [ three ] holds for @xmath412 because @xmath4 is in the basin of @xmath413 for @xmath332 .
similarly , for odd @xmath411 by taking @xmath414 , which is in the basin of @xmath413 for @xmath415 . for @xmath417 , @xmath14 is algebraically stable , since @xmath418 for any @xmath41 .
. meanwhile , for any @xmath13 , @xmath420 . thus , for any @xmath417
, @xmath14 is non - cohomologically hyperbolic .
we will now verify that for @xmath417 the composition @xmath421 satisfies the hypotheses of proposition [ prop : conditions ] . condition ( 1 ) follows with @xmath422 from lemma [ c_singularity ] since @xmath25 blows - up @xmath4 to @xmath26 , which is singular at @xmath261 .
thus , @xmath421 blows - up @xmath4 to @xmath423 , which is singular at @xmath424 .
since @xmath425 for any @xmath41 , the point @xmath4 has an infinite pre - orbit under @xmath14 along which @xmath14 is finite .
thus , condition ( 2 ) holds . finally , since @xmath25 is finite at @xmath29 with @xmath426 , the composition @xmath14 is finite at @xmath29 and maps @xmath29 to @xmath424 .
condition ( 3 ) then follows from the fact that @xmath427 .
since the hypotheses of proposition [ prop : conditions ] are satisfied , no iterate of @xmath14 preserves a foliation . + * observation : * all the arguments above hold with @xmath6 substituting @xmath25 , except for the verification of condition ( 3 ) , because it is not clear how to ensure that @xmath428 remains singular and has an infinite pre - orbit along which @xmath429 is finite . for the sake of simplicity we decided not to address this technicality , but we expect that for a generic rotation of @xmath6 the same result holds .
favre ) let @xmath430 be an algebraically stable map that is not cohomologically hyperbolic . then generic rotations @xmath431 will have the same dynamical degrees and hence also be non - cohomologically hyperbolic . under what conditions on @xmath216 will generic rotations of @xmath216 not preserve a foliation ?
dominique cerveau .
feuilletages holomorphes de codimension 1 .
rduction des singularits en petites dimensions et applications . in _
dynamique et gomtrie complexes ( lyon , 1997 ) _ , volume 8 of _ panor . synthses _ , pages ix , xi , 1147 . soc . math .
france , paris , 1999 .
vincent guedj .
proprits ergodiques des applications rationnelles . in _
quelques aspects des systmes dynamiques polynomiaux _ , volume 30 of _ panor .
synthses _ , pages 97202 .
france , paris , 2010 .
nessim sibony .
dynamique des applications rationnelles de @xmath433 . in _
dynamique et gomtrie complexes ( lyon , 1997 ) _ , volume 8 of _ panor . synthses _ ,
pages ix x , xi xii , 97185 .
france , paris , 1999 . | we present simple examples of rational maps of the complex projective plane with equal first and second dynamical degrees and no invariant foliation . |
cosmological measurements usually rely on the use of integrated ( luminosity or angular ) distances , which unavoidably introduces a model dependence in the data themselves .
for example , the use of type ia supernovae as ` standard candles ' to measure the distance scale relies on finding the ` correct ' lightcurve shape needed to determine their absolute luminosity .
the parameters defining this estimator are optimized along with the parameters of the assumed cosmological model , necessitating a recalibration of the data for each and every model being tested ( perlmutter et al .
1998 ; riess et al . 1998 ; schmidt et al . 1998 ; melia 2012 ; wei et al . 2013 ) .
galaxies evolving passively on a time scale much larger than their age difference can instead be used to measure the expansion rate @xmath0 using solely the local redshift - time derivative @xmath1 ( jimenez & loeb 2002 ) .
as we shall explain in greater detail below , these are typically massive ( with a stellar content @xmath6 ) early - type galaxies that formed over @xmath7 of their stellar mass at high - redshifts ( @xmath8 , i.e. , before the universe was @xmath9 gyr old ) very rapidly ( in only @xmath10 gyr ) and have experienced only minor subsequent episodes of star formation .
they are believed to be the oldest objects at all redshifts ( treu et al .
2005 ) , and if we observe them at @xmath11 ( i.e. , when the universe was @xmath12 gyr old ) , we see a stellar population that mostly formed during the first @xmath13 of the galaxies evolution .
these so - called cosmic chronometers therefore avoid the need of pre - assuming a cosmological model in order to extract the data , and can therefore form a powerful discriminant to test different expansion scenarios . in previous applications , we ( melia maier 2013 ; melia & mcclintock 2015 ) used the then available catalog of sources to compare two specific models : the @xmath14 universe ( melia 2007 , 2016a , 2017 ; melia & shevchuk 2012 ) and the standard ( concordance ) @xmath2cdm model . in these
one - on - one comparative tests , the use of information criteria showed that the cosmic chronometers strongly preferred @xmath14 over @xmath2cdm .
the significance of this result is that , whereas the standard model requires a transition from deceleration at large redshifts to an accelerated expansion today , the @xmath14 universe predicts a constant expansion rate at all redshifts .
recently , however , 5 new , valuable data points were added to the compilation of @xmath0 measurements at that critical redshift ( @xmath15 ) where the transition is thought to have occurred ( moresco et al .
2016 ) , thereby improving the statistical significance of the fits in the redshift range @xmath16 .
based on a study of this expanded sample , direct evidence for the existence of an epoch of cosmic re - acceleration was claimed to have been seen . to reach this result , however , the locally measured
, high value of @xmath4 had to be included in the data set .
yet this determination of the hubble constant is completely distinct from all the other measurements based on the cosmic chronometers themselves . in this paper
, we argue that this step has unduly biased this recent analysis and that contrary to its conclusion the expanded sample in fact strengthens the case for a constant expansion rate in the redshift range @xmath17 .
when one ignores the effects of the local gravitational potential at the position of the observer , the value of @xmath4 measured directly from the redshift - distance relation of local sources is discrepant at a level of @xmath18 ( or roughly @xmath19 ) relative to the hubble constant inferred from fitting anisotropies in the cosmic microwave background ( cmb ) using @xmath2cdm ( marra et al .
2013 ; planck collaboration 2014 ) . generally speaking
, our observations of the universe are made from a vantage point whose spacetime differs from the mean by a degree that is very difficult to probe with any precision ( valkenburg et al .
estimates of the cosmic variance created by local inhomogeneities are often based on the hubble bubble picture , in which a sphere of matter is carved out of the friedmann - robertson - walker ( frw ) background , and is compressed or diluted in order to obtain a simple distribution of the ensuing inhomogeneity with a slightly different frw spacetime .
detailed analysis of the dependence of the measurement of @xmath4 on the local gravitational potential has shown that the hubble bubble effect can at least partially mitigate the @xmath20 tension between the cmb and local hubble constants ( marra et al . 2013 ) .
compelling observational support for this interpretation of the disparity is provided by the fact that local measurements of @xmath4 using supernovae within @xmath21 mpc ( corresponding roughly to @xmath22 ) is @xmath23 larger than the value of @xmath4 measured using supernovae outside of this region .
consequently , one can largely alleviate the hubble bubble effect by adopting a minimum redshift of @xmath24 in the analysis of the expansion rate .
the idea that we may be living in a local underdense hubble bubble has actually been considered since the 1990 s ( turner et al .
1992 ; suto et al . 1995
; shi et al . 1996 , 1998 ; zehavi et al . 1998 ; giovanelli et al . 1999 ) . in every case ,
the implied variation on the local gravitational potential was shown to generate variance of the cosmological parameters , including the local hubble constant . the general consensus from all
this work is that the locally measured value of @xmath4 , though obtained in a model - independent fashion , is nonetheless high ( compared to what one might expect in the context of other cosmological measurements ) with systematic uncertainties that are difficult to ascertain with any precision . for this principal reason ,
we believe that a test of cosmological models using measurements of @xmath0 is more robust when only the truly model - independent data obtained with cosmic chronometers are used , without the contamination introduced through the inclusion of a poorly understood measurement of @xmath4 .
our approach is bolstered by the fact that the results of the model comparisons could not be more different when the locally measured @xmath4 is included versus when it is not , even with exactly the same sample of 30 cosmic - chronometer measurements .
but if the expansion rate measured with this technique firmly points to an epoch of cosmic re - acceleration ( moresco et al .
2016 ) , then this should be seen regardless of whether or not @xmath4 is added to the data set .
to demonstrate this point compellingly , we will here use the most recent sample of 30 cosmic - chronometer measurements to compare seven different models , including @xmath2cdm . in so doing , we shall re - affirm , and strengthen , the conclusions drawn from our previous two studies ( melia & maier 2013 ; melia & mcclintock 2015)that measurements of @xmath0 using cosmic chronometers strongly favour the constant expansion rate predicted by the @xmath14 universe over other models , including @xmath2cdm . the models we will compare are as follows : 1 . the standard flat @xmath2cdm model , in which the matter and dark - energy densities are fixed by the condition @xmath25 . throughout this paper , @xmath26 is the energy density @xmath27 of species @xmath28 , scaled to today s critical density , @xmath29 . in this model ,
the two free parameters are @xmath4 and @xmath30 , and @xmath31^{1/2}\;.\ ] ] 2 . the concordance o@xmath2cdm model , with free parameters @xmath4 , @xmath30 , and @xmath32 .
in this model , @xmath33^{1/2}\;.\ ] ] 3 . the flat @xmath34cdm model , in which the matter and dark - energy densities are fixed by the condition @xmath35 , but with an unconstrained dark - energy equation - of - state , @xmath36 , where @xmath37 is the pressure .
the three free parameters here are @xmath4 , @xmath30 , and @xmath38 , with an expansion rate give by @xmath39^{1/2}\;.\ ] ] 4 .
einstein de sitter space , in which the cosmic fluid contains only matter .
@xmath4 is the sole free parameter , and @xmath40 5 . a friedmann model ( that we shall call friedmann i ) with negative curvature . here
, @xmath30 is fixed to be @xmath41 and @xmath42 , implying a curvature term with @xmath43 ( baryshev & teerikorpi 2012 ) . in this case
, @xmath4 is the sole free parameter , and @xmath44^{1/2 } \;.\ ] ] 6 . a friedmann model ( that we shall call friedmann ii ) with negative curvature , but with @xmath30 free and @xmath42 , implying a curvature term with @xmath45 . in this case , the two free parameters are @xmath4 and @xmath30 , and @xmath46 7 . the @xmath14 universe ( a friedmann - robertson - walker cosmology with zero active mass ) . in this model ,
the total equation - of - state is @xmath47 , with @xmath48 and @xmath49 the total energy density and pressure of the cosmic fluid ( melia 2007 , 2016a , 2017 ; melia & shevchuk 2012 ) . in this cosmology
, @xmath4 is the sole free parameter , and @xmath50 we test these models collectively against measurements of the hubble constant using the passively evolving galaxies introduced above , based on the observed 4,000 break in their spectra . for old stellar populations ,
this break is a discontinuity in the spectral continuum due to metal absorption lines whose amplitude correlates linearly with the age and metal abundance of the stars ( moresco et al .
it is weakly dependent on the star formation history , and is basically unaffected by dust reddening . when the metallicity of these stars is known , it is possible to measure the age difference @xmath51 of two nearby galaxies as proportional to the difference in their 4,000 amplitudes .
the slope of this proportionality depends on the metallicity .
then , together with the measured redshift difference @xmath52 of these galaxies , one may find the hubble constant at the average of their redshifts using the simple relation @xmath53 of course , there are several factors that may inhibit the accuracy of this procedure , possibly mitigating the value of using differential measurements of age in these systems .
fortunately , extensive tests ( moresco et al . 2016 ) have shown that the 4,000 feature depends principally on the age and metallicity of the host galaxies , but only weakly on the assumed star formation history , the initial mass function , a possible progenitor bias , and the so - called @xmath54-enhancement .
the latter refers to the observation that these early , passive galaxies have higher ratios of @xmath54 elements to iron than milky - way type galaxies . and
the progenitor bias may arise due to a possible evolution of the mean redshift of formation as a function of redshift of the galaxy samples . of these effects
, it turns out that only an uncertainty in the metallicity contributes noticeably to a systematic error @xmath55 comparable to the statistical errors in the sample .
simulations have shown that the progenitor bias contributes at most only a few percentage points to @xmath55 , while the impact of the initial mass function is insignificant .
the difference between the 4,000 amplitudes estimated in a single stellar population using a chabrier or salpeter initial mass function is less than @xmath56 for all reasonable metallicities and less than @xmath57 for the solar metallicity ( moresco et al .
the difference in the 4,000 amplitudes due to the @xmath54-enhancement is likewise extremely small , on average only @xmath58 .
the only factor other than metallicity that contributes noticeably to @xmath55 is the star formation history .
again , simulations have shown that variations in the assumed star forming rate can lead to @xmath59 errors in the inferred value of @xmath52 from measurements of the 4,000 amplitudes ( moresco et al .
together , the combination of uncertainties in the star formation history and the metallicity contribute an overall error of about @xmath60 to @xmath61 , and hence the inferred value of @xmath0 . versus @xmath62 , with error bars , and the best - fit curves from four models ( see text ) .
the corresponding likelihood of each model being the best choice is shown in table 1 . ]
the data are shown in figure 1 ( moresco et al .
2016 ; simon et al . 2005 ; stern et al . 2010 ; moresco et al . 2012 ; zhang et al . 2014
; moresco 2015 ) . for each model
, we find the set of parameters that optimize the fit and minimize the @xmath63 .
the best - fit curves of four cosmologies are also shown in this figure , superimposed on the data . as we have discussed in earlier papers ( melia & maier 2013 ; melia & mcclintock 2015 ) , building on the sound arguments developed in liddle ( 2004 , 2007 ) and liddle et al .
( 2006 ) , among others , a fair statistical comparison between models with different formulations and numbers of free parameters must be based on the use of information criteria ( takeuchi 2000 ; tan & biswas 2012 ) . in this context , the likelihood of model @xmath54 being the best choice is given by the expression @xmath64 where ic@xmath65 is one of aic@xmath65 , kic@xmath65 , or bic@xmath65 , and the sum in the denominator runs over all the models being tested simultaneously .
the akaike information criterion is defined by @xmath66 , while the kullback implementation has @xmath67 , and @xmath68 is the bayes information criterion . in these expressions , @xmath69 is the number of free parameters and @xmath70 is the number of data points .
all of these criteria punish models with a large number of free parameters because these are deemed to be fitting the noise . also , as discussed in these earlier works , the bic is the most appropriate criterion to use when @xmath70 is large , as we have here ( with 30 source measurements ) .
but for completeness , we tabulate the results of all three criteria ( see table 1 ) .
this tabulation includes the individual ic s and each model s likelihood , weighed against all the seven cosmologies being tested simultaneously , of being the best choice .
table 1 also includes the optimized parameters for models that have them , along with their @xmath71 uncertainties .
lcccccccc & & & + model & @xmath4 & @xmath30 & @xmath32 & @xmath38 & @xmath72 & bayes ic&kullback ic&akaike ic + & ( km / s / mpc ) & & & & & bicprob & kicprob & aicprob + & & & & & & & & + & & & & & & & & + @xmath14 & @xmath73 & & & & 0.57 & 20.020.50 & 19.620.44 & 18.620.30 + & & & & & & & & + @xmath2cdm & @xmath74 & @xmath75 & @xmath76 & & 0.52 & 21.300.26 & 20.500.28 & 18.500.32 + & & & & & & & & + friedmann ii & @xmath77 & @xmath78 & & & 0.59 & 23.310.10 & 22.510.10 & 20.510.12 + & & & & & & & & + @xmath34cdm & @xmath79 & @xmath80 & & @xmath81 & 0.53 & 24.560.05 & 23.360.07 & 20.360.12 + & & & & & & & & + o@xmath2cdm & @xmath82 & @xmath83 & @xmath84 & & 0.54 & 24.660.05 & 23.460.06 & 20.460.12 + & & & & & & & & + friedmann i & @xmath85 & @xmath86 ( fixed)@xmath87 & & & 0.74 & 24.870.04 & 24.470.04 & 23.470.03
+ & & & & & & & & + einstein - de sitter & @xmath88 & & & & 1.74 & 53.902e-8 & 53.502e-8 & 52.501e-8 + & & & & & & & & + it is clear from this comparison that when the high , locally measured value of @xmath4 is excluded from the data compilation , the cosmic chronometers favour the @xmath14 cosmology , which predicts a constant expansion rate over the redshift range @xmath3 .
flat @xmath2cdm is second on the list , but only because we assumed prior values for @xmath89 and @xmath90 . in principle , all of the free parameters in each model should be optimized using solely the cosmic chronometers for a statistically fair comparison . and
we can see that when fewer prior values are assumed , e.g. , as for @xmath34cdm and o@xmath2cdm , their probabilities drop considerably . from the compilation in table 1
, we can also see how sensitive each of the model fits is to the choice of @xmath4 .
notice , for example , that in the case of @xmath14 , which has only one free parameter ( the hubble constant itself ) , the optimized value of @xmath4 is very tightly constrained . a @xmath91 variation may be generated with a mere change of only @xmath92 km s@xmath93 mpc@xmath93 .
on the other hand , in a model such as @xmath34cdm , which has three free parameters and therefore more flexibility
a @xmath91 variation requires a change in @xmath4 of about @xmath94 km s@xmath93 mpc@xmath93 from its optimum value . in spite of this clear separation in the model outcomes , however , a possible concern with
this analysis is the fact that all of the @xmath72 values listed in table 1 , with the exception of einstein - de sitter , are significantly smaller than @xmath95 , suggesting that the errors in figure 1 may be over - estimated .
this may mitigate the tension between certain models and the data , perhaps even producing a biased likelihood of some cosmologies relative to the others . to test whether our conclusions are affected in this way ,
we have carried out a parallel comparative analysis of these measurements by artificially reducing the errors by a factor ( equal to @xmath96 , as it turns out ) that makes these @xmath72 values approximately equal to @xmath95 for all but the einstein - de sitter universe . the results are shown in table 2 .
the likelihoods have indeed changed somewhat , and @xmath34cdm and @xmath2cdm are disfavoured less , though the information criteria still prefer @xmath14 . on the other hand ,
the bic probabilities for @xmath14 and standard flat @xmath2cdm are now indistinguishable .
but on the basis of these results , with artifically reduced errors , one still can not conclude that the cosmic chronometers favour an accelerating universe .
this outcome is qualitatively similar to that based on the analysis of the published measurements and their errors .
lcccccccc & & & + model & @xmath4 & @xmath30 & @xmath32 & @xmath38 & @xmath72 & bayes ic&kullback ic&akaike ic + & ( km / s / mpc ) & & & & & bicprob & kicprob & aicprob + & & & & & & & & + & & & & & & & & + @xmath2cdm & @xmath97 & @xmath98 & @xmath76 & & 0.92 & 32.580.42 & 31.780.43 & 29.780.42 + & & & & & & & & + @xmath14 & @xmath99 & & & & 1.02 & 32.950.35 & 32.550.29 & 31.550.17 + & & & & & & & & + @xmath34cdm & @xmath100 & @xmath101 & & @xmath102 & 0.95 & 35.730.09 & 34.530.11 & 31.530.17 + & & & & & & & & + o@xmath2cdm & @xmath103 & @xmath104 & @xmath105 & & 0.95 & 35.910.08 & 34.710.10 & 31.710.16 + & & & & & & & & + friedmann ii & @xmath106 & @xmath107 & & & 1.05 & 36.160.07 & 35.360.07 & 33.360.07 + & & & & & & & & + friedmann i & @xmath108 & @xmath86 ( fixed)@xmath87 & & & 1.32 & @xmath109 41.570.005 & @xmath109 41.170.004 & @xmath109 40.170.002 + & & & & & & & & + einstein - de sitter & @xmath110 & & & & 3.10 & @xmath109 93.183e-14 & @xmath109 92.782e-14 & @xmath109 91.781e-14 + & & & & & & & & +
adding five new measurements of the hubble parameter @xmath0 with the cosmic - chronometer approach , moresco et al .
( 2016 ) claimed to obtain the first cosmology - independent constraint on the transition redshift , showing the existence of an epoch of cosmic re - acceleration .
this result , however , relies heavily on the use of a gaussian prior on the hubble constant , @xmath111 km s@xmath93 mpc@xmath93 .
it is generally recognized that this @xmath4 is not an accurate representation of the smoothed background @xmath4 if the local expansion rate is influenced by a hubble bubble .
the inclusion of this high , locally measured value of @xmath4 unfairly biases the results by contaminating the cosmic - chronometer data with a measurement whose systematics are unknown . in this regard , we point out that all of the optimized values of @xmath4 in table 1 are smaller than the locally measured hubble constant , even for @xmath2cdm .
in fact , were one to use the cosmic chronometers to infer @xmath4 in the context of the standard model , one would actually find a remarkable consistency with the value ( @xmath112 km s@xmath93 mpc@xmath93 ) measured by _ planck _
( planck collaboration 2014 ) .
since the same cosmology is being referenced for the interpretation of these two disparate sets of data , this consistency with the value of @xmath4 seen at intermediate and very high redshifts adds some support to our thesis in this paper that the locally measured hubble constant is not a true representation of its large - scale , smoothed value . still , the fact that the optimized value of @xmath4 in @xmath14 is notably smaller than that in the standard model merits some attention .
the hubble constant can not be measured directly using type ia sne ( since @xmath4 is degenerate with the absolute sn magnitude , which is also free ) .
the most accurate determination of @xmath4 we have today is from the anisotropies in the cmb ( planck collaboration 2014 ) .
the chronometer - measured value of @xmath4 in @xmath14 is about @xmath113 smaller than the _
planck _ measurement .
but the cmb value of @xmath4 is itself model dependent , and the temperature fluctuation spectrum has not yet been analyzed using @xmath14 .
all we have at the moment is the hubble constant optimized for @xmath2cdm , so it is too early to tell whether the chronometer and cmb measurements of @xmath4 are consistent with each other in the @xmath14 model .
what we do know from simulations is that a local hubble bubble can probably modify the large - scale value of @xmath4 by possibly @xmath114 km s@xmath93 mpc@xmath93 ( marra et al .
2013 ; ben - dayan et al .
2014 ) , so to be compatible with the low value of @xmath4 reported here for @xmath14 , the future measurement of the hubble constant in this model based on cmb anisotropies ought to be @xmath115 km s@xmath93 mpc@xmath93 .
it is also worth pointing out that some local measurements of @xmath4 do actually suggest a value smaller than that ( i.e. , @xmath116 km s@xmath93 mpc@xmath93 ) employed by moresco et al .
for example , tammann & reindl ( 2013 ) used red - giant branch stars in the haloes of local galaxies to calibrate the type ia sn luminosity and inferred a hubble constant @xmath117 km s@xmath93 mpc@xmath93 , while the ( now somewhat dated ) sn hst project yielded a value @xmath118 ( sandage et al . 2006 ) .
still , the majority of local measurements of @xmath4 suggest a higher value , so the key question for @xmath14 remains how the cosmic chronometer measurements will compare with those based on model fits to the cmb temperature anisotropies . with these caveats in mind ,
we have demonstrated in this paper that the inference of re - acceleration is reversed when the additional @xmath4 datum is excluded .
we have shown that , used on their own , without the prior on @xmath4 , the latest compilation of 30 model - independent cosmic - chronometer measurements prefer a constant expansion rate over an accelerated one out to @xmath5 .
this result may seem surprising at first , but it merely confirms the outcome of many previous comparative tests between @xmath2cdm and the constant expansion rate cosmology @xmath14 .
this type of analysis has been carried out using diverse sets of data , at low , intermediate , and high redshifts .
the @xmath14 model appears to be favoured over @xmath2cdm by all the observations used so far .
a brief survey of the results includes the following sample : ( 1 ) when the supernovae in the supernova legacy survey sample are correctly recalibrated for each model being tested , these favour @xmath14 over @xmath2cdm with a bic likelihood of @xmath7 versus @xmath119 ( wei et al . 2015 ) ; ( 2 ) according to the quasar hubble diagram and the alcock - pacynski test ( lpez - corredoira et al .
2016 ; melia & lpez - corredoira 2016 ) , @xmath14 is more likely to be correct than @xmath2cdm ; ( 3 ) based on the presumed constancy of the gas mass fraction in clusters , the bic favours @xmath14 over @xmath2cdm with a likelihood of @xmath120 versus only @xmath121 ( melia 2016b ) ; ( 4 ) while @xmath2cdm can not account for the appearance of high-@xmath62 quasars without some anomalous seed formation or greatly super - eddington accretion , none of which have ever been seen , their formation and growth are fully consistent with the timeline predicted in @xmath14 ( melia 2013 ) ; ( 5 ) a similar time - compression problem arises with the emergence of high-@xmath62 galaxies in @xmath2cdm , though not in @xmath14 ( melia 2014 ) ; ( 6 ) based on the age - redshift relationship of old passive galaxies , the bic favours @xmath14 over @xmath2cdm with a likelihood of @xmath122 versus @xmath123 ( wei et al . 2015 ) .
( 7 ) whereas the inferred probability of @xmath2cdm accounting for the cosmic microwave background ( cmb ) angular correlation function is @xmath124 , @xmath14 fits it much better , including the absence of correlation at angles greater than @xmath125 ( melia 2014 ) .
this tension between @xmath2cdm and observations of the cmb may be quite serious because the lack of any large - angle correlation is inconsistent with inflationary scenarios . yet
the standard model would not survive without inflation to fix the horizon problem .
in contrast , the @xmath14 universe does not have or need inflation .
this is only a small sample of the many tests completed thus far , but already it spans a broad range of data , some at low redshifts , others at high redshifts . in every case
, the @xmath14 cosmology has been favoured over @xmath2cdm .
even so , the challenge of establishing the viability of @xmath14 is ongoing . in this paper
, we have added to the observational evidence in its favour , but several important issues need to resolved .
it remains to be seen whether big bang nucleosynthesis ( bbn ) in this model can correctly reproduce the light elements .
initial attempts at simulating bbn with a constant expansion rate have been very promising , showing that the well - known @xmath126li anomaly plaguing the standard model may be solved by the slower burning taking place with such an expansion scenario ( benoit - lvy & chardin 2012 ) .
of course , this wo nt be known for sure until the bbn calculations will have been carried out correctly for the conditions in @xmath14 . also ,
although the anisotropies in the cmb have been analyzed for @xmath14 in terms of their angular correlation function , they have yet to be used to calculate a power spectrum in this cosmology .
as is well known , the ability of @xmath2cdm to accurately account for the cmb power spectrum provides strong support in its favour .
this analysis , however , is beyond the scope of the present work , and its results will be reported elsewhere .
we are grateful to the anonymous referee for his insightful comments that have led to an improvement in the manuscript .
fm is grateful to pmo in nanjing , china , for its hospitality while this work was being carried out .
we acknoledge partial support from the national basic research program ( 973 " program ) of china ( grant nos 2014cb845800 ) , the national natural science foundation of china ( grants nos . 11322328 , 11433009 , 11673068 , and 11603076 ) , the youth innovation promotion association ( 2011231 ) , the key research program of frontier sciences ( qyzdb - ssw - sys005 ) , the strategic priority research program the emergence of cosmological structures " ( grant no .
xdb09000000 ) of the chinese academy of sciences , and the natural science foundation of jiangsu province ( grant no .
bk20161096 ) .
this work was also partially supported by grant 2012t1j0011 from the chinese academy of sciences visiting professorships for senior international scientists , and grant gdj20120491013 from the chinese state administration of foreign experts affairs .
baryshev , y. & teerikorpi , p. 2012 , fundamental questions of practical cosmology , springer , dordrecht ben - dayan , i. , durrer , r. , marozzi , g. and schwarz , d. j. 2014 , prl , 112 , 221301 benoit - lvy , a. & chardin , g. 2012 , a&a , 537 , id.a78 giovanelli , r. , dale , d. , haynes , m. , hardy , e. & campusano , l. 1999 , apj , 525 , 25 jimenez , r. & loeb , a. 2002 , apj , 573 , 37 liddle , a. r. 2004 , mnras , 351 , l49 liddle , a. r. 2007 , mnras , 377 , l74 liddle , a. , mukherjee , p. & parkinson , d. 2006 , astron . & geophys . , 47 , id.040000 lpez - corredoira , m. , melia , f. , lusso , e. & risaliti , g. 2016 , ijmp - d , 25 , id.1650060 marra , v. , amendola , l. , sawicki , i. & valkenburg , w. 2013 , prl , 110 , id.241305 melia , f. 2007 , mnras , 382 , 1917 melia , f. 2012 , aj , 144 , id.110 melia , f. 2013 , apj , 764 , id.72 melia , f. 2014 , aj , 147 , id.120 melia , f. 2014 , a&a , 561 , id.a80 melia , f. 2016a , front .
, 11 ( 4 ) , id.119801 melia , f. 2016b , proc .
r. soc . a , 472 , id.20150765 melia , f. 2017 , front .
, 12 ( 1 ) , id.129802 melia , f. & lpez - corredoira , m. 2016 , ijmp - d , submitted ( arxiv:1503.05052 ) melia , f. & maier , r. s. 2013 , mnras , 432 , 2669 melia , f. & mcclintock , t. m. 2015 , aj , 150 , id.119 melia , f. & shevchuk , a.s.h .
2012 , mnras , 419 , 2579 moresco , m. 2015 , mnras , 450 , l16 moresco , m. , cimatti , a. , jimenez , r. et al .
2012 , jcap , 8 , id.006 moresco , m. , pozzetti , l. , cimatti , a. , jimenez , r. , maraston , c. , verde , l. , thomas , d. , citro , a. , tojeiro , r. & wilkinson , d. 2016 , jcap , 05 , id.014 perlmutter , s. , aldering , g. , della valle , m. et al .
1998 , nature , 391 , 51 planck collaboration .
2014 , a&a , 571 , id.a23 riess , a. g. , filippenko , a. v. , challis , p. et al .
1998 , aj , 116 , 1009 sandage , a. , tammann , g. a. , saha , a. , reindl , b. , macchetto , f. d. , & panagia , n. 2006 , apj , 653 , 843 schmidt , b. p. , suntzeff , n. b. , phillips , m. m. et al .
1998 , apj , 507 , 46 shi , x .- d . & turner , e. l. 1998 , aj , 493 , 519 shi , x. , widrow , l. m. & dursi , l. j. 1996 , mnras , 281 , 565 simon , j. , verde , l. & jimenez , r. 2005 , prd , 71 , id.123001 stern , d. , jimenez , r. , verde , l. , stanford , s. a. & kamionkowski , m. 2010 , apjs , 188 , id.280 suto , y. , suginohara , t. & inagaki , y. 1995 , prog .
, 93 , 839 takeuchi , t. t. 2000 , astrophys .
, 271 , 213 tammann , g. a. & reindl , b. 2013 , a&a , 549 , i d a136 tan , m.y.j . & biswas , r. 2012 , mnras , 419 , 3292 treu , t. , ellis , r. s. , liao , t. x. , van dokkum , p. g. , tozzi , p. , coil , a. , newman , j. , cooper , m. c. and davis , m. 2005 , apj , 633 , 174 turner , e. l. , cen , r. & ostriker , j. p. 1992
, aj , 103 , 1427 valkenburg , w. , marra , v. & clarkson , c. 2014 , mnras , 438 , l6 wang , y. , spergel , d. n. & turner , e. l. 1998 , apj , 498 , 1 wei , j .- j . , wu , x. & melia , f. 2013 , apj , 772 , id.43 wei , j .- j .
, wu , x. , melia , f. & maier , f. 2015 , aj , 149 , id.10 wei , j .-
, wu , x. , melia , f. , wang , f. y. & yu , h. 2015 , aj , 150 , id.35 zehavi , i. , riess , a. g. , kirshner , r. p. & dekel , a. a 1998 , apj , 503 , 483 zhang , c. , zhang , h. , yuan , s. et al .
2014 , res . astron . and astrophys .
, 14 , 1221 | whereas many measurements in cosmology depend on the use of integrated distances or time , galaxies evolving passively on a time scale much longer than their age difference allow us to determine the expansion rate @xmath0 solely as a function of the redshift - time derivative @xmath1 .
these model - independent ` cosmic chronometers ' can therefore be powerful discriminators for testing different cosmologies . in previous applications , the available sources strongly disfavoured models ( such as @xmath2cdm ) predicting a variable acceleration , preferring instead a steady expansion rate over the redshift range @xmath3 .
a more recent catalog of 30 objects appears to suggest non - steady expansion . in this paper
, we show that such a result is entirely due to the inclusion of a high , locally - inferred value of the hubble constant @xmath4 as an additional datum in a set of otherwise pure cosmic - chronometer measurements .
this @xmath4 , however , is not the same as the background hubble constant if the local expansion rate is influenced by a hubble bubble .
used on their own , the cosmic chronometers completely reverse this conclusion , favouring instead a constant expansion rate out to @xmath5 . |
we consider data sets describing contacts between individuals , collected by the sociopatterns collaboration ( http://www.sociopatterns.org ) in three different settings : a workplace ( office building , invs ) @xcite , a high school ( thiers13 ) @xcite and a scientific conference ( sfhh ) @xcite .
these data correspond to the close face - to - face proximity of individuals equipped with wearable sensors , at a temporal resolution of @xmath0 seconds @xcite .
table [ tab : data ] summarises the characteristics of each data set .
the contact data are represented by temporal networks , in which nodes represent the participating individuals and a link between two nodes @xmath1 and @xmath2 at time @xmath3 indicates that the two corresponding persons were in contact at that time .
these three data sets were chosen as representative of different types of day - to - day contexts and of different contact network structures : the sfhh data correspond to a rather homogeneous contact network ; the invs and thiers13 populations were instead structured in departments and classes , respectively . moreover ,
high school classes ( thiers13 ) are of similar sizes while the invs department sizes are unequal .
finally , the high school contact patterns ( thiers13 ) are constrained by strict and repetitive school schedules , while contacts in offices are less regular across days . to quantify how the incompleteness of data , assumed to stem from a uniformly random participation of individuals to the data collection , affects the outcome of simulations of dynamical processes , we consider as ground truth the available data and perform population resampling experiments by removing a fraction @xmath4 of the nodes uniformly at random .
( note that the full data sets are also samples of all the contacts that occurred in the populations , as the participation rate was lower than @xmath5 in each case . in the thiers13 case
however , the participation rate was quite high . )
we then simulate on the resampled data the paradigmatic susceptible - infectious - recovered ( sir ) and the susceptible - infectious - susceptible ( sis ) models of epidemic propagation . in these models ,
a susceptible ( s ) node becomes infectious ( i ) at rate @xmath6 when in contact with an infectious node .
infectious nodes recover spontaneously at rate @xmath7 . in the sir model , nodes then enter an immune recovered ( r ) state , while in the sis model , nodes become susceptible again and can be reinfected .
the quantities of interest are for the sir model the distribution of epidemic sizes , defined as the final fraction of recovered nodes , and for the sis model the average fraction of infectious nodes @xmath8 in the stationary state .
we also calculate for the sir model the fraction of epidemics that infect more than 20% of the population and the average size of these epidemics . for the sis model
, we determine the epidemic threshold @xmath9 for different values of @xmath7 : it corresponds to the value of @xmath6 that separates an epidemic - free state ( @xmath10 ) for @xmath11 from an endemic state ( @xmath12 ) for @xmath13 , and is thus an important indicator of the epidemic risk .
we refer to the methods section for further details on the simulations .
we then present several methods for constructing surrogate data using only information contained in the resampled data .
we compare for each data set the outcomes of simulations performed on the whole data set , on resampled data sets with a varying fraction of nodes removed , @xmath4 , and on the reconstructed data sets built using these various methods .
missing data are known to affect the various properties of contact networks in different ways . in particular
, the number of neighbours ( degree ) of a node decreases as the fraction @xmath4 of removed nodes increases , since removing nodes also removes links to these nodes . under the hypothesis of uniform sampling ,
the average degree @xmath14 becomes @xmath15 for the resampled network @xcite . as a result ,
the density of the resampled aggregated contact network , defined as the number of links divided by the total number of possible links between the nodes , does not depend on @xmath4 .
the same reasoning applies to the density @xmath16 of links between groups of nodes @xmath17 and @xmath18 , defined as the number of links @xmath19 between nodes of group @xmath17 and nodes of group @xmath18 , normalised by the maximum possible number of such links , @xmath20 , where @xmath21 is the number of nodes of group @xmath17 ( for @xmath22 , the maximum possible number of links is @xmath23 ) : both the expected number of neighbours of group @xmath18 for nodes of group @xmath17 ( given by @xmath24 ) and the number @xmath25 of nodes in group @xmath18 are indeed reduced by a factor @xmath26 , so that @xmath16 remains constant .
this means that the link density contact matrix , which gathers these densities and gives a measure of the interaction between groups ( here classes or departments ) , is stable under uniform resampling .
we illustrate these results on our empirical data sets in supplementary figures 1 , 2 , 4 and 5 .
table [ tab : sim_cm ] and supplementary figure 2 show in particular that the similarities between the original and resampled matrices are high for all data sets ( see supplementary figures 45 for the contact matrices themselves ) .
finally , the temporal statistics of the contact network are not affected by population sampling , as already noted in @xcite for other data sets : the distributions of contact and inter - contact durations ( the inter - contact durations are the times between consecutive contacts on a link ) , of number of contacts per link and of cumulated contact durations ( i.e. , of the link weights in the aggregated network ) do not change when the network is sampled uniformly ( supplementary figure 1 ) . in the case of structured population ,
an interesting property is moreover illustrated in supplementary figures 67 : although the distributions of contact durations occurring between members of the same group or between individuals belonging to different groups are indistinguishable , this is not the case for the distributions of the numbers of contacts per link nor , as a consequence , for the distributions of cumulated contact durations .
in fact , both cumulated contact durations and numbers of contacts per link are more broadly distributed for links joining members of the same group .
the figures show that this property is stable under uniform resampling . despite the robustness of these properties , the outcome of simulations of epidemic spread is strongly affected by the resampling . as fig .
[ sampling ] illustrates for instance , the probability of large outbreaks in the sir model decreases strongly as @xmath4 increases and even vanishes at large @xmath4 .
as mentioned above , such a result is expected , since the removed nodes act as if they were immunised : sampling hinders the propagation in simulations by removing transmission routes between the remaining nodes . as a consequence , the prevalence and the final size of the outbreaks
are systematically underestimated by simulations of the sir model on the resampled network with respect to simulations on the whole data set ( for the sis model , the epidemic threshold is overestimated ) : resampling leads overall to a systematic underestimation of the epidemic risk , and fig .
[ sampling ] illustrates the extent of this underestimation for the data at hand .
we now present a series of methods to improve the estimation of the epidemic risk in simulations of epidemic spread on temporal network data sets in which nodes ( individuals ) are missing uniformly at random .
note that we do not address here the problem of link prediction @xcite as our aim is not to infer the missing contacts .
the hierarchy of methods we put forward uses increasing amounts of information corresponding to increasing amounts of detail on the group and temporal structure of the contact patterns , as measured in the resampled network .
we moreover assume that the timelines of scheduled activity are known ( i.e. , nights and weekends , during which no contact occurs ) . for each data set , considered as ground truth ,
we create resampled data sets by removing at random a fraction @xmath4 of the @xmath27 nodes .
we then measure on each resampled data set a series of statistics of the resulting contact network and construct stochastic , surrogate versions of the missing part of the network by creating for each missing node a surrogate instance of its links and a synthetic timeline of contacts on each surrogate link , in the different ways described below ( see supplementary information and methods section for more details on their practical implementation ) .
method 0 . as discussed above , the first effect of missing data is to decrease the average degree of the aggregate contact network , while keeping its density constant .
hence , the simplest approach is to merely compensate this decrease .
we therefore measure the density of the resampled contact network @xmath28 , as well as the average aggregate duration of the contacts , @xmath29 .
we then add back the missing nodes and create surrogate links between these nodes and between these nodes and the nodes of the resampled data set at random , with the only constraint to keep the overall link density fixed to @xmath28 .
we then attribute to each surrogate link the same weight @xmath29 and create for each link a timeline of randomly chosen contact events of equal length @xmath30 ( the temporal resolution of the data set ) whose total duration gives back @xmath29 .
method w. the heterogeneity of aggregated contact durations is known to play a role in the spreading patterns of model diseases @xcite . we therefore refine method 0 by collecting in the resampled data the list @xmath31 of aggregate contact durations , or weights ( w ) .
we build the surrogate links and surrogate timelines of contacts on each link as in method 0 , except that each surrogate link carries a weight extracted at random from @xmath31 , instead of the average @xmath29 .
method ws . the fact that the population is divided into groups of individuals such as classes or departments can have a strong impact on the structure of the contact network @xcite and on spreading processes @xcite .
we thus measure here the link density contact matrix of the resampled data , and construct surrogate links in a way to keep this matrix fixed ( equal to the value measured in the resampled data ) , in the spirit of stochastic block models with fixed numbers of edges between blocks @xcite .
we collect in the resampled data two separate lists of aggregate contact durations : @xmath32 gathers the weights of links between individuals belonging to the same group , and @xmath33 is built with the weights of links joining individuals of different groups . for each surrogate link ,
its weight is extracted at random either from @xmath32 if it joins individuals of the same group or from @xmath33 if it associates individuals of different groups .
timelines are then attributed to links as in w. this method assumes that the number of missing nodes in each group is known , and preserves the group structure ( s ) of the population .
method wt .
several works have investigated how the temporal characteristics of networks ( such as burstiness ) can slow down or accelerate spreading @xcite . in order
to take these characteristics into account , we measure in the resampled data the distributions of number of contacts per link and of contact and inter - contact durations , in addition to the global network density .
we build surrogate links as in method w , and construct on each link a synthetic timeline in a way to respect the measured temporal statistics ( t ) of contacts .
more precisely , we attribute at random a number of contacts ( taken from the measured distribution ) to each surrogate link , and then alternate contact and inter - contact durations taken at random from the respective empirical distributions .
method wst .
this method conserves the distribution of link weights ( w ) , the group structure ( s ) , and the temporal characteristics of contacts ( t ) : surrogate links are built and weights assigned as in method ws , and contact timelines on each link as in method wt .
each of these methods uses a different amount of information gathered from the resampled data .
methods 0 , w and wt include an increasing amount of detail on the temporal structure of contacts : method 0 assumes homogeneity of aggregated contact durations , while w takes into account their heterogeneity , and wt reproduces heterogeneities of contact and inter - contact durations . on the other hand ,
neither of these three methods assume any knowledge of the population group structure .
this can be due either to an effective complete lack of knowledge about the population structure , as in the sfhh data , or also to the lack of data on the repartition of the missing nodes in the groups .
methods ws and wst on the other hand reproduce the group structure as in a stochastic block model with fixed number of links within and between groups , and take into account the difference between the distributions of numbers of contacts and aggregate durations between individuals of the same or of different groups .
indeed , links within groups correspond on average to larger weights , as found empirically in @xcite and discussed above ( supplementary figures 56 ) .
overall , method wst is the one that uses most information measured in the resampled data .
( additional properties such as the transitivity -which is also stable under resampling procedure , see supplementary figure 3- can also be measured in the resampled data and imposed in the construction of surrogate links , as detailed in the supplementary information .
this comes however at a strong computational cost and we have verified that it does not impact significantly our results , as shown in the supplementary figure 20 . ) we check in table [ tab : sim_cm ] and supplementary figures 813 that the statistical properties of the resulting reconstructed ( surrogate ) networks , obtained by the union of the resampled data and of the surrogate links , are similar to the ones of the original data for the wst method .
we emphasise again that our aim is not to infer the true missing contacts , so that we do not compare the detailed structures of the surrogate and original contact networks .
figures [ fig : inferinvs ] , [ fig : inferthiers ] , [ fig : infersfhh ] and supplementary figures 1619 display the outcome of sir spreading simulations performed on surrogate networks obtained using the various reconstruction methods , compared with the outcome of simulations on the resampled data sets , for various values of @xmath4 .
method 0 leads to a clear overestimation of the outcome and does not capture well the shape of the distribution of outbreak sizes .
method w gives only slightly better results .
the overall shape of the distribution is better captured for the three reconstruction methods using more information : ws , wt and wst ( note that for the sfhh case the population is not structured , so that w and ws are equivalent , as are wt and wst ) .
the wst method matches best the shape of the distributions and yields distributions much more similar to those obtained by simulating on the whole data set than the simulations performed on the resampled networks .
we also show in fig .
[ fig : dist ] the fraction of outbreaks that reach at least @xmath34 of the population and the average epidemic size for these outbreaks . in the case of simulations performed on resampled data ,
we rapidly lose information about the size and even the existence of large outbreaks as @xmath4 increases .
simulations using data reconstructed with methods 0 and w , on the contrary , largely overestimate these quantities , which is expected as infections spread easier on random graphs than on structured graphs @xcite , especially if the heterogeneity of the aggregated contact durations is not considered @xcite . taking into account the population structure or using contact sequences that respect the temporal heterogeneities ( broad distributions of contact and inter - contact durations ) yield better results ( ws and wt cases , respectively ) .
overall , the wst method , for which the surrogate networks respect all these constraints , yields the best results .
we show in the supplementary information that similar results are obtained for different values of the spreading parameters .
moreover , as shown in fig .
[ fig : sis_invs ] and supplementary figures 1415 , the phase diagram obtained for the sis model when using reconstructed networks is much closer to the original than for resampled networks .
overall , simulations on networks reconstructed using the wst method yield a much better estimation of the epidemic risk than simulations using resampled network data , for both sis and sir models . even when simulations are performed on reconstructed contact patterns built with the wst method , the maximal outbreak sizes are systematically overestimated ( figs .
[ fig : inferinvs ] - [ fig : infersfhh ] ) , as well as , in most cases , the probability and average size of large outbreaks , especially for the sfhh case ( figs .
[ fig : infersfhh ] - [ fig : dist ] ) .
these discrepancies might stem from structural and/or temporal correlations present in empirical contact data that are not taken into account in our reconstruction methods . in order to test this hypothesis
, we construct several reshuffled data sets and use them as initial data in our resampling and reconstruction procedure .
we use both structural and temporal reshuffling as described in the methods section , in order to remove either structural correlations , temporal correlations , or both , from the original data sets .
we then proceed to a resampling and reconstruction procedure ( using the wst method ) as for the original data , and perform numerical simulations of sir processes . as for the original data , simulations on resampled data
lead to a strong underestimation of the process outcome , and simulations using the reconstructed data gives much better results .
we show in the supplementary figures 21 - 22 that we still obtain discrepancies , and in particular overestimations of the largest epidemic sizes , when we use temporally reshuffled data in which the link structure of the contact network is maintained .
if on the other hand we use data in which the network structure has been reshuffled in a way to cancel structural correlations within each group , the reconstruction procedure gives a very good agreement between the distributions of epidemic sizes of original and reconstructed data , as shown in fig .
[ fig : infercmshuffled ] .
more precisely we consider here `` cm - shuffled '' data , i.e. , contact networks in which the links have been reshuffled randomly but separately for each pair of groups , i.e. , a link between an individual of group @xmath17 and an individual of group @xmath18 is still between groups @xmath17 and @xmath18 in the reshuffled network .
the difference with the case of non - reshuffled empirical data is particularly clear for the sfhh case .
this indicates that the overestimation observed in figs .
[ fig : inferinvs ] - [ fig : infersfhh ] is mostly due to the fact that the reconstructed data does not reproduce small scale structures of the contact networks : such structures might be due to e.g. groups of colleagues or friends , whose composition is neither available as metadata nor detectable in the resampled data sets . when the fraction @xmath4 of nodes excluded by the resampling procedure becomes large , the properties of the resampled data may start to differ substantially from those of the whole data set ( figs . s1 & s2 ) . as a result ,
the distributions of epidemic sizes of sir simulations show stronger deviations from those obtained on the whole data set ( fig .
[ fig : infer_high ] ) , even if the epidemic risk evaluation is still better than for simulations on the resampled networks ( fig .
[ fig : dist ] ) .
most importantly however , the information remaining in the resampled data at large @xmath4 can be insufficient to construct surrogate contacts .
this happens in particular if an entire class or department is absent from the resampled data or if all the resampled nodes of a class / department are disconnected ( see methods for details ) .
we show in the bottom plots of fig .
[ fig : dist ] the failure rate , i.e. , the fraction of cases in which we are not able to construct surrogate networks from the resampled data . the failure rate increases
gradually with @xmath4 for the invs data since the groups ( departments ) are of different sizes . for the thiers13 data , all classes are of similar sizes so that the failure rate reaches abruptly a large value at a given value of @xmath4 . for the sfhh data
, we can always construct surrogate networks as the population is not structured .
another limitation of the reconstruction method lies in the need to know the number of individuals missing in each department or class .
if these numbers are completely unknown , giving an estimation of outbreak sizes is impossible as adding arbitrary numbers of nodes and links to the resampled data can lead to arbitrarily large epidemics .
the methods are however still usable if only partial information is available . for instance , if only the overall missing number of individuals is available , it is possible to use the wt method , which still gives sensible results .
moreover , if @xmath4 is only approximately known , e.g. , @xmath4 is known to be within an interval of possible values @xmath35 $ ] , it is possible to perform two reconstructions using the respective hypothesis @xmath36 and @xmath37 and to give an interval of estimates .
we provide an example of such procedure in supplementary figure 23 .
the understanding of epidemic spreading phenomena has been vastly improved thanks to the use of data - driven models at different scales .
high resolution contact data in particular have been used to evaluate epidemic risk or containment policies in specific populations or to perform contact tracing @xcite . in such studies ,
missing data due to population sampling might represent however a serious issue : individuals absent from a data set are indeed equivalent to immunised individuals when epidemic processes are simulated . feeding sampled data into data - driven models
can therefore lead to severe underestimations of the epidemic risk and might even a priori affect the evaluation of mitigation strategies if for instance some at - risk groups are particularly undersampled . here
we have put forward a set of methods to obtain a better evaluation of the outcome of spreading simulations for data - driven models using contact data from a uniformly sampled population . to this aim
, we have shown how it is possible , starting from a data set describing the contacts of only a fraction of the population of interest ( uniformly sampled from the whole population ) , to construct surrogate data sets using various amounts of accessible information , i.e. , quantities measured in the sampled data .
we have shown that the simplest method , which consists in simply compensating for the decrease in the average number of neighbours due to sampling , yields a strong overestimation of the epidemic risk . when additional information describing the group structure and the temporal properties of the data is added in the construction of surrogate data sets , simulations of epidemic spreading on such surrogate data yield results similar to those obtained on the complete data set .
( we note that the issue of how much information should be included when constructing the surrogate data is linked to the general issue of how much information is needed to get an accurate picture of spreading processes on temporal networks @xcite . ) some discrepancies in the epidemic risk estimation are however still observed , due in particular to small scale structural correlations of the contact network that are difficult or even impossible to measure in the resampled data : these discrepancies are indeed largely suppressed if we use as original data a reshuffled contact network in which such correlations are absent .
the methods presented here yield much better results than simulations using resampled data , even when a substantial part of the population is excluded , in particular in estimating the probability of large outbreaks .
it suffers however from limitations , especially when the fraction @xmath4 of excluded individuals is too large .
first , the construction of the surrogate contacts relies on the stability of a set of quantities with respect to resampling , but the measured quantities start to deviate from the original ones at large @xmath4 .
the shape of the distribution of epidemic sizes may then differ substantially from the original one .
second , large values of @xmath4 might even render the construction of the surrogate data impossible due to the loss of information on whole categories of nodes .
finally , at least an estimate of the number of missing individuals in the population is needed in order to create a surrogate data set .
an interesting avenue for future work concerns possible improvements of the reconstruction methods , in particular by integrating into the surrogate data additional information and complex correlation patterns measured in the sampled data .
for instance , the number of contacts varies significantly with the time of day in most contexts : the corresponding activity timeline might be measured in the sampled data ( overall or even for each group of individuals ) , assumed to be robust to sampling and used in the reconstruction of contact timelines .
more systematically , it might also be possible to use the temporal network decomposition technique put forward in @xcite on the sampled data , in order to extract mesostructures such as temporally - localized mixing patterns .
the surrogate contacts could then be built in a way to preserve such patterns .
indeed , correlations between structure and activity in the temporal contact network are known to influence spreading processes @xcite but are notoriously difficult to measure .
if the group structure of the population is unknown , recent approaches based on stochastic block models @xcite might be used to extract groups from the resampled data ; this extracted group structure could then be used to build the corresponding contact matrix and surrogate data sets .
we finally recall that we have assumed an uniform sampling of nodes , corresponding to an independent random choice of each individual of the population to take part or not to the data collection .
other types of sampling or data losses can however also be present in data collected by wearable sensors , such as partial coverage of the premises of interest by the measuring infrastructure , non - uniform sampling depending on individual activity ( too busy persons or , on the contrary , asocial individuals , might not want to wear sensors ) , on group membership , or due to clusters of non - participating individuals ( e.g. , groups of friends ) .
in addition , other types of data sets such as the ones obtained from surveys or diaries correspond to different types of sampling , as each respondent provides then information in the form of an ego - network @xcite .
such data sets involve potentially additional types of biases such as underreporting of the number of contacts and overestimation of contact durations @xcite : how to adapt the methods presented here is an important issue that we will examine in future work .
finally , the population under study is ( usually ) not isolated from the external world , and it would be important to devise ways to include contacts with outsiders in the data and simulations , for instance by using other data sources such as surveys .
the present work is partially supported by the french anr project harms - flu ( anr-12-monu-0018 ) to m.g . and
a.b , by the eu fet project multiplex 317532 to a.b . , c.c . and c.l.v . , by the a*midex project ( anr-11-idex-0001 - 02 ) funded by the `` investissements davenir '' french government program , managed by the french national research agency ( anr ) to a.b . , by the lagrange project of the isi foundation funded by the crt foundation to c.c . , and by the q - aracne project funded by the fondazione compagnia di san paolo to c.c .
a.b . and c.c .
designed and supervised the study . m.g .
, c.l.v . , c.c . , and a.b .
collected and post - processed the data , analyzed the data , carried out computer simulations and prepared the figures .
, c.l.v . ,
c.c . , and a.b .
wrote the manuscript .
the authors declare that no competing financial interests exist .
we consider data sets collected using the sociopatterns proximity sensing platform ( http://www.sociopatterns.org ) based on wearable sensors that detect close face - to - face proximity of individuals wearing them .
informed consent was obtained from all participants and the french national bodies responsible for ethics and privacy , the commission nationale de linformatique et des liberts ( cnil , http://www.cnil.fr ) , approved the data collections .
the high school ( thiers13 ) data set @xcite is structured in 9 classes , forming three subgroups of three classes corresponding to their specialisation in mathematics - physics ( mp , mp@xmath381 , mp@xmath382 with respectively @xmath39 , @xmath40 and @xmath41 students ) , physics ( pc , pc@xmath38 , psi with respectively @xmath42 , @xmath43 and @xmath44 students ) , or biology ( 2bio1 , 2bio2 , 2bio3 with respectively @xmath45 , @xmath46 and @xmath43 students ) .
the workplace ( invs ) data set @xcite is structured in @xmath47 departments : disq ( scientific direction , @xmath48 persons ) , dmct ( department of chronic diseases and traumatisms , @xmath49 persons ) , dse ( department of health and environment , @xmath44 persons ) , srh ( human resources , @xmath50 persons ) and sfle ( logistics , @xmath51 persons ) . for the conference data ( sfhh )
, we do not have metadata on the participants , and the aggregated network structure was found to be homogeneous @xcite .
simulations of sir and sis processes on the temporal networks of contacts ( original , resampled or reconstructed ) are performed using the temporal gillespie algorithm described in @xcite . for each run of the simulations ,
all nodes are initially susceptible ; a node is chosen at random as the seed of the epidemic and put in the infectious state at a point in time chosen at random over the duration of the contact data .
a susceptible node in contact with an infectious node becomes infectious at rate @xmath6 .
infectious nodes recover at rate @xmath7 : in the sir model they then enter the recovered state and can not become infectious again , while in the sis model they enter the susceptible state again .
if needed , the sequence of contacts is repeated in the simulation @xcite . for sir processes
, we run each simulation , with the seed node chosen at random , until no infectious individual remains ( nodes are thus either still susceptible or have been infected and then recovered ) .
we consider values of @xmath6 and @xmath7 yielding a non - negligible epidemic risk , i.e. , such that a rather large fraction of simulations lead to a final size larger than @xmath52 of the population ( see figs .
[ sampling][fig : infersfhh ] ) : @xmath53 , @xmath54 ( invs ) or @xmath55 ( sfhh and thiers13 ) .
other parameter values are explored in the supplementary information . for each set of parameters ,
the distribution of epidemic sizes is obtained by performing @xmath56 simulations . for sis processes
, simulations are performed using the quasi - stationary approach of @xcite .
they are run until the system enters a stationary state as witnessed by the mean number of infected nodes being constant over time .
simulations are then continued for 50,000 time - steps while recording the number of infected nodes . for each set of parameters ,
the simulations are performed once with each node of the network chosen as the seed node .
we consider a population @xmath57 of @xmath27 individuals , potentially organised in groups .
we assume that all the contacts occurring among a subpopulation @xmath58 of these individuals , of size @xmath59 , are known .
this constitutes our resampled data from which we need to construct a surrogate set of contacts concerning the remaining @xmath60 individuals for which no contact information is available : these contacts can occur among these individuals and between them and the members of @xmath58 .
we assume that we know the group of each member of @xmath61 , and the overall activity timeline , i.e. , the intervals during which contacts take place , separated by nights and weekends .
to construct the surrogate data ( wst method ) , we first compute from the activity timeline the total duration @xmath62 of the periods during which contacts can occur .
then , we measure in the sampled data : * the density @xmath63 of links in the aggregated contact network ; * a row - normalised contact matrix @xmath64 , in which the element @xmath65 gives the probability for a node of group @xmath17 to have a link to a node of group @xmath18 ; * the list @xmath66 of contact durations ; * the lists @xmath67 and @xmath68 of inter - contact durations for internal and external links , _
i.e. _ , for links between nodes of the same group and links between nodes that belong to two different groups , respectively ; * the lists @xmath69 and @xmath70 of numbers of contacts per link , respectively for internal ( within groups ) and external ( between groups ) links ; * the list @xmath71 of initial times between the start of the data set and the first contact between two nodes
. given @xmath63 , we compute the number @xmath72 of additional links needed to keep the network density constant when we add the @xmath73 excluded nodes . we then construct each link according to the following procedure : * a node @xmath1 is randomly chosen from the set @xmath61 of excluded nodes ; * knowing the group @xmath17 that @xmath1 belongs to , we extract at random a target group @xmath18 with probability given by @xmath65 ; * we draw a target node @xmath2 at random from @xmath18 ( if @xmath74 , we take care that @xmath75 ) such that @xmath1 and @xmath2 are not linked ; * depending on whether @xmath1 and @xmath2 belong to the same group or not , we draw from @xmath69 or @xmath70 the number of contact events @xmath76 taking place over the link @xmath77 ; * from @xmath71 , we draw the initial waiting time before the first contact ; * from @xmath66 , we draw @xmath76 contact durations @xmath78 , @xmath79 ; * from @xmath67 or @xmath68 , we draw @xmath80 inter - contact durations @xmath81 , @xmath82 ; * if @xmath83 , we repeat steps ( d ) to ( g ) until we obtain a set of values such that @xmath84 ; * from @xmath85 and the @xmath78 and @xmath81 , we build the contact timeline of the link @xmath77 ; * finally , we insert in the contact timeline the breaks defined by the global activity timeline .
the construction of the surrogate version of the missing links uses as an input the group structure of the subgraph that remains after sampling , as given by the contact matrix of the link densities between the different groups of nodes that are present in the subpopulation @xmath58 . depending on the characteristics of @xmath58 and of the corresponding contacts ,
the construction method can fail in several cases : ( i ) if an entire group ( class / department ) of nodes in the population is absent from @xmath58 ; ( ii ) if the remaining nodes of a specific group ( class / department ) are all isolated in @xmath58 s contact network ; ( iii ) if , during the algorithm , a node of @xmath61 is selected in a certain group @xmath17 but can not create any more links because it already has links to all nodes in the groups @xmath18 such that @xmath86 ; ( iv ) if there are either no internal ( within groups ) or external ( between groups ) links in the contact network of @xmath58 : in this case one of the lists of link temporal characteristics is empty and the corresponding structures can not be reconstructed
. cases ( i ) and ( ii ) correspond to a complete loss of information about the connectivity of a group ( class / department ) of the population , due to sampling .
it is then impossible to reconstruct a sensible connectivity pattern for these nodes .
case ( iii ) is more subtle and occurs in situations of very low connectivity between groups .
for instance , within the contact network of @xmath57 , a group @xmath17 has links only with another specific group @xmath18 , and both @xmath17 and @xmath18 are small ; it is then possible that the nodes of @xmath87 exhaust the set of possible links to nodes of @xmath18 during the reconstruction algorithm .
if a node of @xmath88 is again chosen to create a link , such a creation is not possible and the construction fails .
case ( iv ) usually corresponds to situations in which the links between individuals of different groups which remain in the resampled data set correspond to pairs of individuals who have had only one contact event : in such cases , @xmath68 is empty and external links with more than one contact can not be built . in order to test the effect of correlations in the temporal network , we use four shuffling methods , based on the ones defined in @xcite . link shuffling .
the contact timelines associated with each link are randomly redistributed among the links .
correlations between timelines of links adjacent to a given node are destroyed , as well as correlations between weights and topology .
the structure of the network is kept , as well as the global activity timeline .
time shuffling . from the contact data we build the lists @xmath66 , @xmath89 and @xmath90 of , respectively , contact durations , inter - contact durations and number of contacts per link .
we also measure the list @xmath71 of initial times between the start of the data set and the first contact between two nodes . for each link , we draw randomly a starting time @xmath85 , a number @xmath76 of contacts from @xmath90 , @xmath76 contact durations from @xmath66 and @xmath80 inter - contact durations from @xmath89 , so that the total duration of the timeline does not exceed the total available time @xmath62 .
we then construct the contact timelines , thus destroying the temporal correlations among contacts .
the structure of the network is instead kept fixed .
cm shuffling .
we perform a link rewiring separately on each compartment of the contact matrix , _
i.e. _ , we randomly redistribute links with their contact timelines within each group , and within each pair of groups .
we thus destroy the structural correlations inside each compartment of the contact matrix , while preserving the group structure of the network as given by the link density contact matrix and the contact matrix of total contact times between groups .
cm - time shuffling .
we perform both a cm shuffling and a time shuffling .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , & _
_ ( , ) . , , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . in _ _
( , , ) .
_ et al . _ . in _ _ , vol . , ( ) . , & . _
_ * * , ( ) .
, , & . in _ _ , mswim 10 , ( , , ) .
_ et al . _ . _
_ * * , ( ) . , & . in _ _ , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al .
_ _ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) . .
_ _ * * , ( ) . , , & .
in _ _ , stoc 05 , ( , , ) . .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & ( ) . , &
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , &
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
& _ _ * * , ( ) .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , ,
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , & ( ) .
( ) . , & .
_ _ * * , ( ) .
, , & ( ) . & ( ) . , & _ _ * * , ( ) .
.[tab : data]*data sets . * for each data set we specify the type of social situation , the number @xmath27 of individuals whose contacts were measured , the corresponding participation rate @xmath91 , the duration @xmath92 and the dates of the data collection . [
cols="^,^,^,^,^,^",options="header " , ] * sir epidemic simulations on resampled contact networks .
* we plot the distributions of epidemic sizes ( fraction of recovered individuals ) at the end of sir processes simulated on top of resampled contact networks , for different values of the fraction @xmath4 of nodes removed .
the plot shows the progressive disparition of large epidemic outbreaks as @xmath4 increases .
the parameters of the sir models are @xmath93 and @xmath94 ( invs ) or @xmath95 ( thiers13 and sfhh ) .
the case @xmath96 corresponds to simulations using the whole data set , i.e. , the reference case . for each value of @xmath4 , @xmath56 independent simulations were performed . ]
* sir simulations for the invs ( workplace ) case .
* we compare of the outcome of sir epidemic simulations performed on resampled and reconstructed contact networks , for different methods of reconstruction .
we plot the distribution of epidemic sizes ( fraction of recovered individuals ) at the end of sir processes simulated on top of resampled ( sample ) and reconstructed contact networks , for different values of the fraction @xmath4 of nodes removed , and for the 5 reconstruction methods described in the text ( 0 , w , ws , wt , wst ) .
the parameters of the sir models are @xmath93 and @xmath94 .
the case @xmath96 corresponds to simulations using the whole data set , _
i.e. _ , the reference case . for each value of @xmath4 ,
@xmath56 independent simulations were performed.,scaledwidth=90.0% ] * sir simulations for the thiers13 ( high school ) case .
* we compare of the outcome of sir epidemic simulations performed on resampled ( top left ) and reconstructed contact networks , for different methods of reconstruction .
we plot the distribution of epidemic sizes ( fraction of recovered individuals ) at the end of sir processes simulated on top of resampled ( sample ) and reconstructed contact networks , for different values of the fraction @xmath4 of nodes removed , and for the 5 reconstruction methods described in the text ( 0 , w , ws , wt , wst ) .
the parameters of the sir models are @xmath93 and @xmath95 .
the case @xmath96 corresponds to simulations using the whole data set , i.e. , the reference case . for each value of @xmath4 ,
@xmath56 independent simulations were performed.,scaledwidth=90.0% ] * sir simulations for the sfhh ( conference ) case .
* we compare of the outcome of sir epidemic simulations performed on resampled and reconstructed contact networks , for different methods of reconstruction .
we plot the distribution of epidemic sizes ( fraction of recovered individuals ) at the end of sir processes simulated on top of resampled ( sample ) and reconstructed contact networks , for different values of the fraction @xmath4 of nodes removed , and for three reconstruction methods described in the text ( 0 , w , wt ) . in this case , as the population is not structured , methods w and ws on the one hand , wt and wst on the other hand , are equivalent .
the parameters of the sir models are @xmath93 and @xmath95 .
the case @xmath96 corresponds to simulations using the whole data set , i.e. , the reference case . for each value of @xmath4 ,
@xmath56 independent simulations were performed.,scaledwidth=90.0% ] * accuracy of the different reconstruction methods . *
we perform sir epidemic simulations for each case , for different values of the fraction @xmath4 of missing nodes , for both sampled networks and networks reconstructed with the different methods .
we compare in each case , and as a function of @xmath4 , the fraction of outbreaks that lead to a final fraction of recovered individuals @xmath97 larger than @xmath34 of the population ( a , b ,
c ) , and the average size of these large outbreaks ( d , e , f ) .
the dashed lines give the corresponding values for simulations performed on the complete data sets .
the different methods are : reconstruction conserving only the link density and the average weight of the resampled data ( 0 ) ; reconstruction conserving only the link density and the distribution of weights of the resampled data ( w ) ; reconstruction preserving , in addition to the w method , the group structure of the resampled data ( ws ) ; reconstruction conserving link density , distribution of weights and distributions of contact times , of inter - contact times and of numbers of contacts per link measured in the resampled data ( wt ) ; full method conserving all these properties ( wst ) .
we also plot as a function of @xmath4 the failure rate of the wst algorithm , _
i.e. _ , the percentage of failed reconstructions ( g , h , i ) . for the sfhh case ,
as the population is not structured into groups , methods w and ws are equivalent , as well as methods wt and wst ; moreover , reconstruction is always possible .
the sir parameters are @xmath93 and @xmath94 ( invs ) or @xmath95 ( thiers13 and sfhh ) and each point is averaged over @xmath56 independent simulations . ] * sis simulations for the invs ( workplace ) case .
* we perform sis epidemic simulations and report the phase diagram of the sis model for the original , resampled and reconstructed contact networks .
each panel shows the stationary value @xmath8 of the prevalence in the stationary state of the sis model , computed as described in methods , as a function of @xmath6 , for several values of @xmath7 .
simulations are performed in each case using either the complete data set ( continuous lines ) , resampled data ( dashed lines ) or reconstructed contact networks using the wst method ( pluses ) .
the fraction of excluded nodes in the resampling is @xmath98 for a , c , e , g and @xmath99 for b , d , f , h. ] * sir simulations on shuffled data .
* we compare of the outcome of sir epidemic simulations performed on resampled and reconstructed contact networks , for shuffled data .
we plot the distribution of epidemic sizes ( fraction of recovered individuals ) at the end of sir processes simulated on top of either resampled ( a , c , e ) or reconstructed ( b , d , f ) contact networks , for different values of the fraction @xmath4 of nodes removed .
we use here the wst reconstruction method , and the data set considered consists in a cm - shuffled version ( see methods ) of the original data , in which the shuffling procedure removes structural correlations of the contact network within each group .
the parameters of the sir models are @xmath93 and @xmath94 ( invs ) or @xmath95 ( thiers13 and sfhh ) .
the case @xmath96 corresponds to simulations using the whole data set , _
i.e. _ , the reference ( reshuffled data ) case . for each value of @xmath4 ,
@xmath56 independent simulations were performed.,scaledwidth=90.0% ] * sir simulations for very large numbers of missing nodes .
* we simulate sir processes on reconstructed contact networks for large values of the fraction @xmath4 of removed nodes .
we plot the distributions of epidemic sizes for simulations on reconstructed networks and on the whole data set ( case @xmath96 ) , for large values of the fraction @xmath4 of removed nodes . here
@xmath93 and @xmath94 ( invs ) or @xmath95 ( thiers13 and sfhh ) and @xmath56 simulations were performed for each value of @xmath4 .
the distributions of epidemic sizes for simulations performed on resampled data sets are not shown since at these high values of @xmath4 , almost no epidemics occur . ]
as described in the main text , we consider temporally resolved networks of contacts @xmath100 in a population @xmath57 of @xmath27 individuals and we perform a resampling experiment by selecting a subpopulation @xmath58 of these individuals , of size @xmath59 .
we assume that only the contacts occurring among the subpopulation @xmath58 are known and we compare the properties of the corresponding resampled subnetwork @xmath101 with those of the original network .
supplementary figure [ fig : sampling_network ] shows how population sampling affects several statistical properties of the contact networks .
on the one hand , the degree distribution of the aggregated network of contacts systematically shifts towards smaller degree value .
this is expected as each remaining node has in the resampled network a degree which is at most its degree in the original network , and is strictly smaller if some of its neighbours are not part of the resampled population . on the other hand , the statistical distributions of several quantities of interest
are not affected by sampling : this is the case of the quantities attached either to single contacts or to single links , namely contact and inter - contact durations , number of contacts per link and link weights ( the weight of a link is given by the total duration of the contacts between the two corresponding nodes ) . moreover , as shown in supplementary figure [ fig : sampling ] , the density of the aggregated network , i.e. the ratio between the number of links and the number of possible links , is on average conserved by the random resampling procedure .
it varies however for different realisations of the resampling , and the corresponding variance increases with the fraction @xmath4 of excluded nodes .
supplementary figure [ fig : clust_trans ] shows how the average clustering coefficient of the aggregated network varies with the resampling : notably , it remains high and close to its original value until large values of @xmath4 are reached .
the transitivity of the network , defined as three times the number of triangles divided by the number of connected triplets ( connected subgraphs of @xmath102 nodes and @xmath103 edges ) , is even less affected than the clustering coefficient by the resampling procedure . in the case of structured populations , supplementary figures [ fig : cml_sample_invs ] & [ fig : cml_sample_thiers13 ]
show that the stability of the resampled network s density holds at the more detailed level of the contact matrices of link densities . in such matrices , the element @xmath104 is given by the number of links between individuals of groups @xmath1 and @xmath2 , normalised by the total number of possible links between these two groups ( if @xmath105 denotes the number of individuals in group @xmath1 , the number of possible links is equal to @xmath106 for @xmath107 and to @xmath108 for @xmath109 ) .
these figures clearly illustrate how the diagonal and block - diagonal structures are preserved , and supplementary figure [ fig : sampling ] gives a quantitative assessment of this stability by showing that the cosine similarity between contact matrices between the resampled and original aggregated contact networks remains high even for when a large fraction of the nodes are excluded .
we moreover illustrate in supplementary figures [ fig : int_ext_invs ] and [ fig : int_ext_thiers ] the difference in statistical properties of contacts and links within and between groups , still for structured populations : * the distributions of contact durations are indistinguishable ; * the distribution of link weights ( aggregated contact durations ) is broader for links between individuals belonging to the same group than for links joining individuals of different groups ; * this is due to the difference in the distributions of numbers of contacts per link , which is broader for links within groups than for links between groups ; * the distributions of inter - contact durations differ also slightly , with smaller averages for within - group links . most importantly ,
all these properties and distributions remain stable under resampling , showing that reliable information on the distributions of contact and inter - contact durations , aggregated contact durations , numbers of contacts per link , can be obtained in the resampled data , including the statistical differences between links joining members of different groups and links between two individuals of the same group .
as described in the main text and in particular in the methods section , we construct a surrogate set of contacts concerning the @xmath110 individuals excluded by the resampling .
we compare here the properties of the resulting contact networks ( obtained by merging the resampled contact network @xmath101 and the surrogate set of contacts ) and of the original contact network , @xmath100 .
supplementary figure [ fig : infer_network ] shows that the degree distribution , which is not constrained by the reconstruction procedure , deviates from the original distribution . on the other hand , the distributions of contact durations , inter - contact durations , number of contacts per link and link weights are preserved .
moreover , the link density contact matrices of the reconstructed networks ( supplementary figure [ fig : cml_infer_invs ] & [ fig : cml_infer_thiers13 ] ) share a high similarity with the original contact matrices , even for high fractions of nodes excluded ( supplementary figure [ fig : simcm ] ) . for completeness
, we also compute the contact matrices in contact time density ( cmt ) , in which each element @xmath104 is given by the total time in contact between individuals of groups @xmath1 and @xmath2 , normalised by the total number of possible links between these two groups : it gives the average time spent in contact by two random individuals of groups @xmath1 and @xmath2 .
supplementary figures [ fig : simcm ] , [ fig : cmt_infer_invs ] and [ fig : cmt_infer_thiers13 ] show that the structure of these matrices is well recovered by the reconstruction methods , with high similarity with the original matrices .
we observe for the high school and the conference the same effect on the phase diagram of the sis model as in the workplace : sampling leads to a shift of the epidemic threshold to higher values and thus to an underestimation of the epidemic risk .
the phase diagram and the epidemic threshold are estimated more accurately by using reconstructed networks , thus giving a better evaluation of the epidemic risk ( supplementary figures [ fig : sis_thiers ] & [ fig : sis_sfhh ] ) . in the main text ,
we have considered values of the sir model parameters leading to a non - negligible epidemic risk and a value of @xmath6 corresponding to slow processes .
we consider here several other values of the parameters , corresponding either to faster processes ( supplementary figures [ fig : beta.004 ] - [ fig : beta.04_large ] ) or to smaller epidemic risk ( supplementary figure [ fig : beta.004_small ] ) .
in all cases , simulations performed on the resampled contact networks lead to a strong underestimation of the epidemic sizes , with distributions shifting to smaller values as @xmath4 increases , while the use of reconstructed data sets leads to a better estimation and generally speaking a slight overestimation of the epidemic risk .
we give here details on the alternative reconstruction methods mentioned in the main text , which use less information than the * wst * method . in each case
we consider the same setup as the complete method : a population @xmath57 of @xmath27 individuals ( the nodes of the contact network ) , potentially organised in groups , for which we know all the contacts taking place among a subpopulation @xmath58 of size @xmath59 . for the remaining @xmath60 individuals ,
no contact information is available , but we know to which group they belong .
we also have access to the overall activity timeline , _
i.e. _ , to the successive intervals during which contacts can happen ( daytimes ) , and are excluded ( nights and weekends ) .
the alternative reconstruction methods are the following : 0 : : we perform the reconstruction using only the network density and the average link weight , both measured in the resampled network @xmath101 .
the algorithm goes as follows : + 1 .
we measure in the resampled data : * the density @xmath63 of links in the time - aggregated network ; * the average link weight @xmath29 ( the weight of a link is defined as the total contact time between the two linked nodes ) ; 2 .
we compute the number of links @xmath72 that must be added to keep the network density constant when we add the @xmath73 excluded nodes ; 3 .
we construct @xmath72 links according to the following procedure : * a node @xmath1 is randomly chosen from the set @xmath61 of excluded nodes ; * a node @xmath2 is randomly chosen from the set @xmath111 of all other nodes ; * we compute @xmath112 , where @xmath113 is the temporal resolution of the data set , and we randomly choose @xmath114 time windows of length @xmath115 within the activity windows defined by the activity timeline as contact events between @xmath1 and @xmath2 .
w : : we perform the reconstruction using only the network density and the distribution of link weights , both measured in the resampled network @xmath101 .
the algorithm goes as follows : + 1 .
we measure in the resampled data : * the density @xmath63 of links in the time - aggregated network ; * the list @xmath31 of link weights ( the weight of a link is defined as the total contact time between the two linked nodes ) ; 2 .
we compute the number of links @xmath72 that must be added to keep the network density constant when we add the @xmath73 excluded nodes ; 3 .
we construct @xmath72 links according to the following procedure : * a node @xmath1 is randomly chosen from the set @xmath61 of excluded nodes ; * a node @xmath2 is randomly chosen from the set @xmath111 of all other nodes ; * from @xmath31 , we draw the weight @xmath116 of the link @xmath77 ; * we compute @xmath117 , where @xmath113 is the temporal resolution of the data set , and we randomly choose @xmath114 time windows of length @xmath115 within the activity windows defined by the activity timeline as contact events between @xmath1 and @xmath2 .
ws : : we perform the reconstruction using the network density , the distributions of link weights for internal ( within groups ) and external ( between groups ) links , and the structure of the aggregated network given by the link density contact matrix , all measured in the resampled network @xmath101 .
the algorithm goes as follows : + 1 .
we measure in the resampled data : * the density @xmath63 of links in the time - aggregated network ; * a _ row - normalised _
contact matrix @xmath64 , in which the element @xmath65 gives the probability for a node in group @xmath17 to have a link to a node of group @xmath18 ; * the lists @xmath32 and @xmath33 of link weights for respectively internal and external links
( internal links are links between nodes that belong to the same group , external links are links between nodes from different groups ) ; 2 .
we compute the number of links @xmath72 that must be added to keep the network density constant when we add the @xmath73 excluded nodes ; 3 .
we construct @xmath72 links according to the following procedure : * a node @xmath1 is randomly chosen from the set @xmath61 of excluded nodes ; * knowing the group @xmath17 that @xmath1 belongs to , we extract at random a target group @xmath18 with probability given by @xmath65 ; * we draw a target node @xmath2 at random from @xmath18 ( if @xmath74 , we check that @xmath118 ) ; * depending on whether nodes @xmath1 and @xmath2 belong to the same group or not , we draw from @xmath32 or @xmath33 the weight @xmath116 of the link @xmath77 ; * as for the @xmath119 method , we extract at random @xmath120 contact events of length @xmath113 within the activity timeline .
wt : : we perform the reconstruction using the network density , the distribution of link weights and the temporal structure of the contacts given by the distributions of contact durations , inter - contact durations , number of contacts per link and initial waiting times before the first contact , all measured in the resampled network @xmath101 .
the algorithm goes as follows : + 1 .
we compute from the activity timeline the time @xmath62 as the total duration of the periods during which contacts can occur . 2 .
we measure in the resampled data : * the density @xmath63 of links in the time - aggregated network ; * the list @xmath66 of contact durations ; * the list @xmath89 of inter - contact durations ; * the list @xmath90 of numbers of contacts per link ; * the list @xmath71 of initial waiting times before the first contact for each link ; 3 .
we compute the number of links @xmath72 that must be added to keep the network density constant when we add the @xmath73 excluded nodes ; 4 .
we construct @xmath72 links according to the following procedure : 1 .
a node @xmath1 is randomly chosen from the set @xmath61 of excluded nodes ; 2 .
a node @xmath2 is randomly chosen from the set @xmath111 of all other nodes ; 3 .
we draw from @xmath90 the number of contact events @xmath76 taking place over the link @xmath77 ; 4 . from @xmath71 ,
we draw the initial waiting time @xmath85 before the first contact ; 5 . from @xmath66 ,
we draw @xmath76 contact durations @xmath78 , @xmath79 ; 6 . from @xmath89 , we draw @xmath80 inter - contact durations @xmath81 , @xmath82 ; 7 . while @xmath83 , we repeat steps ( c ) to ( f ) ; 8 . from @xmath85,the @xmath78 and @xmath81
, we build the contact timeline of the link @xmath77 ; 9 . finally , we insert in the contact timeline the breaks defined by the activity timeline . 1
. we measure in the resampled data the transitivity @xmath121 of the time - aggregated network ; 2 . for the construction of each link of a node @xmath1 : * we calculate the current transitivity @xmath122 of the network ; * we list the potential targets @xmath2 in two lists @xmath123 and @xmath124 , depending on whether the creation of a link between @xmath1 and @xmath2 would close a triangle or not ; * * * if @xmath125 , we draw a target node @xmath2 at random from @xmath123 such that @xmath1 and @xmath2 are not linked ; * * else if @xmath126 , we draw a target node @xmath2 at random from @xmath124 such that @xmath1 and @xmath2 are not linked . | data describing human interactions often suffer from incomplete sampling of the underlying population . as a consequence ,
the study of contagion processes using data - driven models can lead to a severe underestimation of the epidemic risk . here
we present a systematic method to alleviate this issue and obtain a better estimation of the risk in the context of epidemic models informed by high - resolution time - resolved contact data .
we consider several such data sets collected in various contexts and perform controlled resampling experiments .
we show how the statistical information contained in the resampled data can be used to build a series of surrogate versions of the unknown contacts .
we simulate epidemic processes on the resulting reconstructed data sets and show that it is possible to obtain good estimates of the outcome of simulations performed using the complete data set .
we discuss limitations and potential improvements of our method .
human interactions play an important role in determining the potential transmission routes of infectious diseases and other contagion phenomena @xcite .
their measure and characterisation thus represent an invaluable contribution to the study of transmissible diseases @xcite .
while surveys and diaries in which volunteer participants record their encounters @xcite have provided crucial insights ( see however @xcite for recent investigations of the memory biases inherent in self - reporting procedures ) , new approaches have recently emerged to measure contact patterns between individuals with high resolution , using wearable sensors that can detect the proximity of other similar devices @xcite .
the resulting measuring infrastructures register contacts specifically within the closed population formed by the participants wearing sensors , with typically high spatial and temporal resolutions . in the recent years
, several data gathering efforts have used such methods to obtain , analyse and publish data sets describing the contact patterns between individuals in various contexts in the form of temporal networks @xcite : nodes represent individuals and , at each time step , a link is drawn between pairs of individuals who are in contact @xcite .
such data has been used to inform models of epidemic spreading phenomena used to evaluate epidemic risks and mitigation strategies in specific , size - limited contexts such as schools or hospitals @xcite , finding in particular outcomes consistent with observed outbreak data @xcite or providing evidence of links between specific contacts and transmission events @xcite .
despite the relevance and interest of such detailed data sets , as illustrated by these recent investigations , they suffer from the intrinsic limitation of the data gathering method : contacts are registered only between participants wearing sensors .
contacts with and between individuals who do not wear sensors are thus missed . in other words , as most often not all individuals accept to participate by wearing sensors , many data sets obtained by such techniques suffer from population sampling , despite efforts to maximise participation through e.g. scientific engagement of participants @xcite .
hence , the collected data only contains information on contacts occurring among a fraction of the population under study .
population sampling is well - known to affect the properties of static networks @xcite : various statistical properties and mixing patterns of the contact network of a fraction of the population of interest may differ from those of the whole population , even if the sampling is uniform @xcite , and several works have focused on inferring network statistics from the knowledge of incomplete network data @xcite .
both structural and temporal properties of time - varying networks might as well be affected by missing data effects @xcite .
in addition , a crucial though little studied consequence of such missing data is that simulations of dynamical processes in data - driven models can be affected if incomplete data are used @xcite .
for instance , in simulations of epidemic spreading , excluded nodes are by definition unreachable and thus equivalent to immunised nodes . due to herd vaccination effects , the outcome of simulations of epidemic models on sampled networks is thus expected to be underestimated with respect to simulations on the whole network .
( we note however , that in the different context of transportation networks , it was found in @xcite that the inclusion of the most important transportation nodes can be sufficient to describe the global worldwide spread of influenza - like illnesses , at least in terms of times of arrival of the spread in various cities . ) how to estimate the outcome of dynamical processes on contact networks using incomplete data remains an open question . here
we make progresses on this issue for incompletely sampled data describing networks of human face - to - face interactions , collected by infrastructures based on sensors , under the assumption that the population participating to the data collection is a uniform random sample of the whole population of interest .
( we do not therefore address here the issue of non - uniform sampling of contacts that may result from other measurement methods such as diaries or surveys . )
we proceed through resampling experiments on empirical data sets in which we exclude uniformly at random a fraction of the individuals ( nodes of the contact network ) .
we measure how relevant network statistics vary under such uniform resampling and confirm that , although some crucial properties are stable , numerical simulations of spreading processes performed using incomplete data lead to strong underestimations of the epidemic risk .
our goal and main contribution consists then in putting forward and comparing a hierarchy of systematic methods to provide better estimates of the outcome of models of epidemic spread in the whole population under study . to this aim
, we do not try to infer the true sequence of missing contacts .
instead , the methods we present consist in the construction of surrogate contact sequences for the excluded nodes , using only structural and temporal information available in the resampled contact data . we perform simulations of spreading processes on the reconstructed data sets , obtained by the union of the resampled and surrogate contacts , and
investigate how their outcomes vary depending on the amount of information included in the reconstruction method .
we show that it is possible to obtain outcomes close to the results obtained on the complete data set , while , as mentioned above , using only the incomplete data severely underestimates the epidemic risk .
we show the efficiency of our procedure using three data sets collected in widely different contexts and representative of very different population structures found in day - to - day life : a scientific conference , a high school and a workplace .
we finally discuss the limitations of our method in terms of sampling range , model parameters and population sizes . |
up to now , progress in physics has been marked by a progress in precision , experimentally as well as theoretically , often the one going hand in hand with the other .
physicists have extended the limits of their theories over and over again , included the principles of special relativity and quantum effects and formed the standard model of particle physics which has proven its predictive power to extreme precision and down to distances as small as @xmath0 m. the question whether there exists a fundamental limit to the possibly achievable resolution of our spacetime is a reoccurring issue in science .
it was in the 5th century b.c .
that demokrit postulated a smallest particle out of which matter is build .
he called it an ` atom ' . in greek
, the prefix ` a ' means ` not ' and the word ` tomos ' means ` cut ' .
thus , atomos or atom means uncuttable or undividable .
2500 years later , we know that not only the atom is dividable , but so is the atomic nucleus . the nucleus itself is a composite of neutrons and protons and further progress in science has revealed that even the neutrons and protons have a substructure .
is there an end to this or will the quarks and gluons turn out to be non - fundamental too ?
any answer to this question has to include not only the structure of matter in our spacetime but the structure of our spacetime itself
it has to include gravity .
one of the major problem within the standard model is that the naive attempt to formulate a quantized theory of gravity fails .
the length scale at which the effects of quantum gravity are expected to become as important as the effects of the other interactions is the so - called planck length @xmath1 .
since gravity is so weak , the planck length is extremely small @xmath2 m and effects could be neglected in the experimentally accessible range so far .
the planck length which is derived from the standard model of physics is far out of reach for future experiments .
but this depressing fact looks completely different if we work within the model of large extra dimensions . here
, the energy scale of quantum gravity can be considerably lowered which means , the minimal length will be raised and the quantum structure of spacetime might lie only one order of magnitude beyond the current experimental possibilities .
physics at planckian energies represents a challenge located between the interests of particle physics and those of general relativity . at those energies ,
large gravitational fields are both generated and felt by the interacting particles , which can then be used as sources and probes for quantum gravity effects .
planck scale physics is thus a key for our understanding of the fundamental laws of nature .
this brief review will focus on the phenomenological implications of planck scale physics and their examination in effective models .
the outline of this review is as follows : the next section serves as a motivation for the central subject of a minimal length .
section three will give an introduction into the model with large extra dimensions and its phenomenology . section four
will provide an extension of this model which includes the minimal length .
observables of minimal length effects within the extra dimensional scenario are discussed in section five , and we end with the conclusions in section six .
gravity itself is inconsistent with physics at very short scales .
the introduction of gravity into into quantum field theory appears to spoil their renormalizability and leads to incurable divergences .
it has therefore been suggested that gravity should lead to an effective cutoff in the ultraviolet , i.e. to a minimal observable length .
it is amazing enough that all attempts towards a fundamental theory of everything necessarily seem to imply the existence of such a minimal length scale .
the occurrence of the minimal length scale has to be expected from very general reasons in all theories at high energies which attempt to include effects of quantum gravity .
this can be understood by a phenomenological argument which has been examined in various ways .
test particles of a sufficiently high energy to resolve a distance as small as the planck length are predicted to gravitationally curve and thereby to significantly disturb the very spacetime structure which they are meant to probe .
thus , in addition to the expected quantum uncertainty , there is another uncertainty caused which arises from spacetime fluctuations at the planck scale .
consider a test particle , described by a wave packet with a mean compton - wavelength @xmath3 .
even in standard quantum mechanics , the particle suffers from an uncertainty in position @xmath4 and momentum @xmath5 , given by the standard heisenberg - uncertainty relation @xmath6 .
usually , every sample under investigation can be resolved by using beams of high enough momentum to focus the wave packet to a width below the size of the probe , @xmath4 , which we wish to resolve .
the smaller the sample , the higher the energy must become and thus , the bigger the collider .
general relativity tells us that a particle with an energy - momentum @xmath7 in a volume of spacetime @xmath8 causes a fluctuation of the metric @xmath9 .
einstein s field equations allow us to relate the second derivative of the metric to the energy density and so we can estimate the perturbation with which leads to @xmath10 and @xmath11 . ]
@xmath12 this leads to an additional fluctuation , @xmath13 , of the spacetime coordinate frame of order @xmath14 , which increases with momentum and becomes non - negligible at planckian energies . but
from this expression we see not only that the fluctuations can no longer be neglected at planckian energies and the uncertainty of the measurement is amplified .
in addition , we see that a focussation of energy of the amount necessary to resolve the planck scale leads to the formation of a black hole whose horizon is located at @xmath15 , should the matter be located inside . since the black hole radius has the property to expand linearly with the energy inside the horizon , both arguments lead to the same conclusion : a minimal uncertainty in measurement can not be erased by using test particles of higher energies .
it is always @xmath16 : at low energies because the compton - wavelengths are too big for a high resolution , at high energies because of strong curvature effects .
thus , to first order the uncertainty principle is generalized to @xmath17 without doubt , the planck scales marks a thresholds beyond which the old description of space - time breaks down and new phenomena have to appear .
more stringent motivations for the occurrence of a minimal length are manifold . a minimal length can be found in string theory , loop quantum gravity and non - commutative geometries .
it can be derived from various studies of thought - experiments , from black hole physics , the holographic principle and further more .
perhaps the most convincing argument , however , is that there seems to be no self - consistent way to avoid the occurrence of a minimal length scale . for reviews
see e.g. @xcite . to give a non - complete list : 1 .
string theory naturally includes a minimal length scale as its very success arises from the fact that interactions do no longer take place at one point in spacetime but are spread out on the world sheet . by analyzing a spacetime picture of string scattering at planckian energies , is has been shown@xcite that the extension of the string grows at high energies , leading to a generalized uncertainty relation of the form eq .
( [ gup]).[string ] 2 .
loop quantum gravity@xcite is a non - perturbative approach to quantum gravity . via the definition of so - called loop - states ,
the metric information is expressed in operators which in the classical limit yield the standard spacetime picture . in the quantum gravity regime
, it turns out that some of these operators , e.g. the area operator , have a discrete spectrum which gives rise to a smallest - distance structure.[qlg ] 3 .
non - commutative geometries@xcite modify the algebra of the generators of space - time translations such that position measurements fail to commute .
the commutation relation is replaced with @xmath18 = { \rm i } \theta^{ij } \quad,\end{aligned}\ ] ] where the tensor @xmath19 has a dimension of ( length)@xmath20 and measures the scale at which the non - local effects become important .
this modification can be pursued throughout the usual development of quantum field theories , leading to a non - local theory .
it can be shown@xcite that within this approach a gaussian distribution naturally exhibits a maximally possible localization of width @xmath21.[ngc ] 4 .
a minimal length can be found from the holographic principle , which states@xcite that the degrees of freedom of a spatial region are determined through the boundary of the region and that the number of degrees of freedom per planck area is no greater than unity .
this leads to a minimal possible uncertainty in length measurements@xcite.[hol ] 5 . in the approach by padmanabhan _
et al_@xcite , the path integral amplitude is made invariant under a duality transformation which replaces a length , @xmath22 , by @xmath23 .
the proper distance between two events in space - time , @xmath24 , then is replaced by @xmath25 , yielding a zero point length of spacetime . due to this , the divergences in quantum field theories are regularized by removing the small distance limit.[pad ] 6 .
several phenomenological examinations of possible precision measurements@xcite , thought experiments about black holes@xcite or the general structure of classical@xcite , semi classical@xcite and quantum - foamy space - time@xcite . all of them lead to the conclusion that there exists a fundamental limit to distance measurement
. the above listed points are cross - related in many ways .
point ( [ ngc ] ) arises as a limit of string theory and also the imposed duality in ( [ pad ] ) is motivated by string t - duality .
the holographic principle has its origin in black hole physics and black hole physics itself connects to almost every point mentioned before .
not listed are the topics of spacetime foams and spin foams as they evade a brief description .
the interested reader is referred to @xcite . instead of finding evidence for the minimal scale
as has been done in these numerous studies , on can use its existence as a postulate and derive extensions to quantum theories with the purpose to examine the arising properties .
such approaches have been undergone and the analytical properties of the resulting theories have been investigated closely@xcite . in the scenario without extra dimensions ,
the derived modifications are important mainly for structure formation and the early universe@xcite .
the importance to deal with a finite resolution of spacetime , however , is sensibly enhanced if we consider a spacetime with large extra dimensions .
during the last decade , several models using compactified large extra dimensions ( lxds ) as an additional assumption to the quantum field theories of the standard model ( sm ) have been proposed@xcite .
the setup of these effective models is motivated by string theory though the question whether our spacetime has additional dimensions is well - founded on its own and worth the effort of examination .
the models with lxds provide us with an useful description to predict first effects beyond the sm .
they do by no means claim to be a theory of first principles or a candidate for a grand unification .
instead , their simplified framework allows the derivation of testable results which can in turn help us to gain insights about the underlying theory .
there are different ways to build a model of extra dimensional space - time . here
, we want to mention only the most common ones : 1 . the add - model proposed by arkani - hamed , dimopoulos and dvali@xcite adds @xmath26 extra spacelike dimensions without curvature , in general each of them compactified to the same radius @xmath27 .
all sm particles are confined to our brane , while gravitons are allowed to propagate freely in the bulk .
[ add ] 2 .
the setting of the model from randall and sundrum@xcite is a 5-dimensional spacetime with an non - factorizable geometry .
the solution for the metric is found by analyzing the solution of einsteins field equations with an energy density on our brane , where the sm particles live . in the type
i model@xcite the extra dimension is compactified , in the type ii model@xcite it is infinite .
3 . within the model of universal extra dimensions@xcite all particles ( or in some extensions , only bosons ) can propagate in the whole multi - dimensional spacetime .
the extra dimensions are compactified on an orbifold to reproduce sm gauge degrees of freedom .
in the following we will focus on the model ( [ add ] ) which yields a beautiful and simple explanation of the hierarchy problem .
consider a particle of mass @xmath28 located in a spacetime of @xmath29 dimensions .
the general solution of poissons equation yields its potential as a function of the radial distance @xmath30 to the source @xmath31 where we have introduced a new fundamental mass - scale @xmath32 .
the hierarchy problem then is the question why , for @xmath33 , this mass - scale is the planck mass , @xmath34 , and by a factor @xmath35 smaller than the mass - scales in the sm , e.g. the weak scale .
the additional @xmath26 spacetime dimensions are compactified on radii @xmath27 , which are small enough to have been unobserved so far . then , at distances @xmath36 , the extra dimensions will freeze out and the potential eq .
( [ newtond ] ) will turn into the common @xmath37 potential , but with a fore - factor given by the volume of the extra dimensions @xmath38 in this limit , we will rediscover the usual gravitational law which yields the relation @xmath39 given that @xmath32 has the right order of magnitude to be compatible with the other observed scales , it can be seen from this argument that the volume of the extra dimensions suppresses the fundamental scale and thus , explains the huge value of the planck mass .
the radius @xmath27 of these extra dimensions , for @xmath40 tev , can be estimated with eq.([master ] ) and typically lies in the range from mm to @xmath41 fm for @xmath26 from @xmath42 to @xmath43 , or the inverse radius @xmath44 lies in energy range ev to mev , respectively .
the case @xmath45 is excluded .
it would result in an extra dimension about the size of the solar system .
due to the compactification , momenta in the direction of the lxds can only occur in quantized steps @xmath46 for every particle which is allowed to enter the bulk .
the fields can be expanded in fourier - series @xmath47 where @xmath22 are the coordinates on our brane and @xmath48 the coordinates of the lxds .
this yields an infinite number of equally spaced excitations , the so called kaluza - klein - tower . on our brane , these massless kk - excitations act like massive particles , since the momentum in the extra dimensions generates an apparent mass term @xmath49 \psi^{(n)}(x ) = 0\quad.\end{aligned}\ ] ] the most obvious experimental test for the existence of extra dimensions is a measurement of the newtonian potential at sub - mm distances .
cavendish like experiments which search for deviations from the @xmath37 potential have been performed during the last years with high precision@xcite . also the consequences for high energy experiments are intriguing . since the masses of the kk - modes are so low , they get excited easily but it is not until energies of order @xmath32 that their phase - space makes them give an important contribution in scattering processes . the number of excitations @xmath50 below a energy @xmath5114 tev .
] can , for an almost continuous spectrum , be estimated with the volume of the @xmath26- dimensional sphere of radius @xmath52 .
we can then estimate the total cross - section for a point interaction , e.g. @xmath53 ( @xmath54 denotes the graviton ) by @xmath55 where we have used eq.([master ] ) .
as can be seen , at energy scales close to the new fundamental scale the estimated cross - section becomes comparable to cross - sections of electroweak processes .
the necessary feynman rules for exact calculations of the graviton tree - level interactions have be derived@xcite and the cross - sections have been examined closely .
since the gravitons are not detected , their emission would lead to an energy loss in the collision and to a higher number of monojets .
modifications of sm predictions do also arise by virtual graviton exchange , which gives additional contributions in the calculation of cross - sections .
another exiting signature of lxds is the possibility of black hole production . in the standard @xmath56 dimensional space - time
, the production of black holes requires a concentration of energy - density which can not be reached in the laboratory . as we have seen , in the higher dimensional space - time , gravity becomes stronger at small distances and therefore the event horizon is located at a larger radius .
we can estimate the horizon radius , @xmath57 , of a mass @xmath28 by using the newtonian potential eq .
( [ newtond ] ) , where we will assume that the black hole is small enough to completely fit into the extra dimensions @xmath58 . in the newtonian limit , the radial entry of the metric tensor is approximately given by @xmath59 , and the horizon appears at the zero of @xmath60 which leads to @xmath61 the exact formula which can be derived from the higher dimensional schwarzschild - metric@xcite differs from the given one by some numerical coefficients .
it is not surprising to see that a black hole with a mass about the new fundamental mass @xmath62 , has a radius of about the new fundamental length scale @xmath63 ( which justifies the use of the limit @xmath58 ) . for @xmath40 1tev
this radius is @xmath64 fm .
thus , at the lhc it would be possible to bring particles closer together than their horizon .
a black hole could be created . to compute the production details ,
the cross - section of the black holes can be approximated by the classical geometric cross - section @xmath65 .
this cross section has been under debate@xcite , but further investigations justify the use of the classical limit at least up to energies of @xmath66@xcite .
however , the topic is still under discussion , see also the very recent contributions@xcite . setting @xmath67tev and @xmath68 one finds @xmath69 pb . with this , the total cross - section in hadronic processes can be computed and shows that @xmath70 of these black holes per year could be produced at the lhc@xcite .
once produced , the black holes will undergo an evaporation process whose thermal properties carry information about @xmath32 and @xmath26 .
furthermore , crossing the threshold for black hole production causes a sharp cut - off for high energetic jets as those jets now end up as black holes instead , and are re - distributed into thermal particles of lower energies .
thus , black holes will give a clear signal .
for recent review on tev - scale black holes see@xcite . as we have seen , the lxd - model predicts a rich phenomenology .
presently available data from collider physics as well as from astrophysics set constraints on the parameters of the model .
for a recent update see e.g.@xcite .
the lxd scenario predicts first effects of quantum gravity to appear close to the new fundamental scale @xmath32 .
signatures of gravitons and black holes would be observable in soon future at the lhc .
for self - consistence , the model further must consider the fact that a lowering of the fundamental scale leads to a raise of the minimal length . within the model of lxds ,
not only effects of quantum gravity occur at lowered energies but so do the effects of the minimal length scale . to incorporate the notion of a minimal length into ordinary quantum field theory
one can apply a simple model which has been worked out in detail in@xcite .
similar approaches have been used in @xcite .
we assume , no matter how much we increase the momentum @xmath71 of a particle , its wavelength can never be decreased below some minimal length @xmath72 or , equivalently , we can never increase its wave vector @xmath73 above @xmath74 .
thus , the relation between the momentum @xmath71 and the wave vector @xmath73 is no longer linear , @xmath75 , but a function .. ] this function @xmath76 has to fulfill the following properties : 1 .
for energies much smaller than the new scale we reproduce the linear relation : for @xmath77 we have @xmath78 [ limitsmallp ] .
it is an odd function ( because of parity ) and @xmath73 is collinear to @xmath71 .
the function asymptotically approaches the upper bound @xmath74 .
[ upperbound ] an example for a function fulfilling the above properties is @xmath79 , which is a very common choice in the literature also known as the unruh relation@xcite .
this function is depicted in fig .
[ fig1 ] , left , along with other possible choices . in the absence of a first principle approach ,
these relations are chosen in an _ ad hoc _ manner .
results from such a model thus have to be examined as to their dependence on the functional relation and their stability .
however , the exact choice of the function is important only in the regime @xmath80 as at higher energies the same asymptotic behavior must be fulfilled . in the regime of first effects , an expansion of the function can capture this ignorance in a free parameter . in the following , we will use the tanh as a concrete example .
an expansion to first order gives in this case @xmath81 in the context of this model , it is further assumed that @xmath82 , such that the spacing of the kaluza - klein excitations compared to energy scales @xmath32 becomes almost continuous and we can use an integration instead of summing up the kk - tower .
the momentum @xmath71 is kept as the common lorentz - invariant quantity which is conserved in the standard ( special relativistic ) way . since @xmath73 is a function of @xmath71
only , the wave vector also is a constant of motion . in scattering processes with
more than one participating particle , the wave vector @xmath73 in general will not be conserved as its transformation is non - linear . in the presented approach
, the mass - shell condition @xmath83 leads to a modified dispersion relation @xmath84 which is implicitly defined by @xmath85 . inserting the assumed relation @xmath76
it can be put in the general form @xmath86 thus , the investigation of models with modified dispersion relation is another way to look at the minimal length proposal .
the quantization of the given ansatz is straightforward and follows the usual procedure .
the commutators between the corresponding operators @xmath87 and @xmath88 remain in the standard form . using the well known commutation relations and inserting the functional relation between the wave vector and the momentum then yields the modified commutator for the momentum
@xmath89={\mathrm i } \delta_{ij}\quad\rightarrow\quad [ \,\hat{x}_i,\hat{p}_j]&= & + { \rm i } \frac{\partial p_i}{\partial k_j } \quad.\end{aligned}\ ] ] this results in the generalized uncertainty principle ( gup ) @xmath90 which reflects the fact that by construction it is not possible to resolve space - time distances arbitrarily well .
since @xmath76 gets asymptotically constant its derivative @xmath91 drops to zero and the uncertainty in ( [ gu ] ) increases for high energies .
the behavior of our particles thus agrees with our expectations . applying the approximation for the tanh one finds for
the above relations the first order alteration @xmath92 & \approx & { \rm i } \hbar\left ( 1 + \frac { \hat{p}^2}{m_{\rm f}^2 } \right ) \quad , \quad \delta p \delta x \geq \frac{1}{2 } \hbar \left ( 1 + \frac{\langle \hat{p}^2 \rangle } { m_{\rm f}^2 } \right ) \quad.\end{aligned}\ ] ] since @xmath93 we have for the eigenvectors @xmath94 and so @xmath95 . we could now add that both sets of eigenvectors have to be a complete orthonormal system and therefore @xmath96 , @xmath97 .
this seems to be a reasonable choice at first sight , since @xmath98 is known from the low energy regime .
unfortunately , now the normalization of the states is different because @xmath73 is restricted to the brillouin zone @xmath99 to @xmath100 . to avoid the need to recalculate normalization factors , we choose the @xmath101 to be identical to the @xmath98 . following the proposal of @xcite this yields then a modification of the measure in momentum space . in the presence of @xmath26 lxds with radii @xmath27 , the eigenfunctions are
then normalized to@xcite @xmath102 where the expression in the vertical bar denotes the functional determinant of the relation @xmath76 and is responsible for an extra factor accompanying the @xmath103-functions .
the completeness relation of the modes takes the form @xmath104 where @xmath105 denotes the volume of the @xmath26-dimensional momentum space . to avoid a new normalization @xmath106 of the eigenfunctions we take the factors into the integral by a redefinition of the measure in momentum space @xmath107 this redefinition has a physical interpretation because we expect the momentum space to be squeezed at high momentum values and weighted less .
with use of an expansion of the tanh for high energies we have @xmath108 and so we can draw the important conclusion that the momentum measure is exponentially squeezed at high energies .
one can now retrace the usual steps and derive equations of motion in quantum mechanics and quantum field theory .
it can be shown@xcite that the general replacement @xmath109 and the new measure eq .
( [ rescalevolume3 ] ) turn out to be a very simple recipe to rewrite the usual equations and feynman rules .
in quantum mechanics , the momentum operator in position representation can be derived and with this , the modified schrdinger equation : @xmath110 ^ 2 + v(r ) \quad . \label{schroed}\end{aligned}\ ] ] further , the the quantization of the energy - momentum relation yields the modified klein - gordon equation and the dirac equation @xmath111 where @xmath71 is now a function of @xmath73 . the above given requirements for @xmath76 do not list lorentz - covariance .
it is easy to see that the proposed model can not conserve this symmetry .
consider an observer who is , in the standard way , boosted relative to the minimal length .
he then would observe a contracted minimal length which would be even smaller than the minimal length . to resolve this problem
it is inevitable to modify the lorentz - transformation .
several attempts to construct such transformations have been made@xcite , most notably the so - called doubly special relativity by amelino - camelia . in this approach ,
the modified lorentz - transformations have two invariants .
not only is the speed of light , @xmath112 , a quantity independent of the reference frame , but so is the minimal length @xmath113 .
lorentz - invariance in this setting is not broken but deformed at high energies .
there is no preferred reference frame .
the minimal length is a fundamental limit for all observers .
note , that the problem of finding the modified transformation between reference frames is identical to finding the relation between @xmath73 and @xmath71 .
since we assume that @xmath71 is a standard lorentz vector , we will know how @xmath73 transforms if @xmath76 is fixed ( and the other way round ) .
thus , the modified lorenz - transformations which have gained an enormous amount of interest during the last years are another way to look at the minimal length proposal .
the above derived properties of a quantum theory with a minimal length lead to modifications of the familiar equations in quantum mechanics and can be used to predict effects arising from the existence of a fundamental limit to space - time resolution . with the modified version of the feynman rules
, it can be examined how the minimal length influences sm cross - sections . a calculation of cross - sections of qed tree level processes
@xmath114 shows that the result can be expressed in a particularly simple form .
the relation between the modified , @xmath115 , and the unmodified , @xmath116 , differential cross - section is given by @xmath117 where the index @xmath118 labels the participating particles and @xmath119 ( @xmath120 ) is is the three - momentum of the initial ( final ) particles .
the dominant contribution to the cross - section is the squeezing - factor from the measure of momentum space which lowers the phase space of the outgoing particles .
this yields a cross - section which is below the sm value . applying this result to fermionic pair production processes @xmath121 , or @xmath122 , respectively
, one finds that the modified cross - sections are in agreement with the data for a minimal length in the range @xmath123 fm , see fig [ fig1 ] , right .
the effect of the squeezed momentum space does also modify predictions from the lxd - scenario .
the dilepton production under inclusion of virtual graviton exchange in hadron collisions has been examined in the minimal length scenario in ref.@xcite .
it was found that the bounds on the new fundamental scale @xmath32 are less stringent than those derived without a minimal length scale .
since the modifications get important at energies close to the new scale , the predicted black hole production is strongly influenced .
black hole production is less probable @xcite since the approach of the partons to distances below the schwarzschild - radius requires higher energies within the minimal length scenario .
figure [ fig2 ] , left , shows the total cross - section of the black hole production with and without a generalized uncertainty principle . for the expected lhc energies the rate is lowered by a factor @xmath124 5 .
the existence of a minimal length scale also influences the evaporation spectra of the produced black holes .
this topic is especially interesting with regard to the unsolved information loss problem which arises if one applies the semi - classically derived result all the way down to the planck scale . in this case ,
the black hole emits thermal radiation whose sole property is its temperature whatever the initial state of the collapsing matter has been .
so , if the black hole first captures all information behind its horizon and then completely vanishes into thermally distributed particles , the basic principle of unitarity can be violated .
this happens when the initial state was a pure quantum state and then evolves into a mixed one@xcite .
when one tries to escape the information loss problem there are two possibilities left : the information is released back by a modification of the evaporation properties or a stable black hole remnant is left which keeps the information . once again , we see that black holes exist in a regime where quantum physics and gravity overlap and the interplay between such completely different fields of physics as general relativity , thermodynamics and quantum theory shows us the limits of our current knowledge . using the generalized uncertainty principle ,
the properties@xcite of the evaporation process are modified when the black hole radius approaches the minimal length which decreases the number of decay particles and leads to the formation of a stable remnant .
the investigation of ultra high energetic cosmic rays ( uhecrs ) is a promising area for physics beyond the sm as the observed spectrum extends to energies above @xmath125 ev@xcite .
theoretically , the uhecrs , mostly protons , are expected to interact with the photons of the cosmic background radiation ( cbr ) to produce pions via the reactions @xmath126 , @xmath127 .
the cross - section for this photopion production at the given center of mass energies is known from experiments in the laboratory with high precision . in the center of mass frame of the interacting particles ,
the threshold for the photopion process is @xmath128 gev@xcite . for the uhecrs
this causes the so - called greisen - zatsepin - kuzmin ( gzk ) cut - off@xcite at @xmath129 ev , above which the universe becomes non - transparent for the traveling proton .
the highest energetic protons have most likely an extragalactic origin . given the known properties of the cbr , the mean free path of the proton traveling towards the earth is much too low to allow protons from such distant sources to cause the observed uhecrs .
to solve this gzk - puzzle , it has been proposed that the absence of the cut - off is a consequence of a modified lorentz - transformation coming along with the observer - independent minimal length scale@xcite . to see this ,
consider in which regard the traveling ultra high energetic proton differs from the one in the laboratory .
it travels through a soup of photons which looks pretty cold , @xmath130 k , in the earth rest frame and has a density of @xmath131 @xmath132 . under an ordinary lorentz - transformation , however , this soup will look very hot and dense for the high speed proton . the threshold for pion production is crossed and the mean free path will be lowered due to the increased density .
now , the minimal length imposes a constraint on the number density of photons that the proton can experience .
no matter how fast it travels and how much it is boosted , the distance from one cbr photon to the next will never be smaller than the minimal length .
thus , compared to the standard case , its mean free path can be considerably raised , relaxing the gzk cut - off . note again , that lorentz - invariance in this scenario is not broken but deformed .
the apparent violation of observer - independence is caused by a non - equivalence of reference frames .
the reference frame in which the cbr is isotropic is exceptional and not identical with the rest frame of the uhecr proton .
high precision measurements confirm the predictions of sm calculations with impressing accuracy .
besides high energy physics , the detailed analysis at the high precision frontier is likely to report first observed deviations from the sm .
current calculations of sm processes include contributions of quantum field theory at higher loop order .
fortunately , such effort is not necessary for an examination of the modifications arising from a minimal length scale .
since the generalized uncertainty already influences quantum mechanical processes , an analysis of the quantum mechanical equations of motion is sufficient to derive constraints on the minimal length . the derived modified schrdinger equation eq .
( [ schroed ] ) will cause a shift in the energy levels of atoms@xcite .
the most precise data are available for the hydrogen s1-s2 transition . for the electromagnetic potential one inserts @xmath133 and find a shift of the energy levels from the old values @xmath134 to the new values @xmath135 @xmath136 with the current accuracy of experiments for the hydrogen s1-s2 transition , a very weak constraint on the new scale can be found @xmath137 .
the relative shift as a function of @xmath138 is also depicted in figure [ fig2 ] , right . to derive more stringent predictions ,
an examination of the magnetic moment of the muon @xmath139 @xcite can be done .
the value of @xmath140 is modified by self energy corrections in quantum field theory .
modifications from the minimal length are important even at the classical level and occur in the qed - range .
a particles energy in a magnetic field @xmath141 depends on its spin .
energy levels which are degenerated for free particles split up in the presence of a field . in a static , homogeneous magnetic field
, the expectation value of the spin vector will perform a precession around the direction of the field .
the rotation frequency is proportional to the strength of the field and the magnetic moment of the particle and so can be used to measure the magnetic moment .
the modifications arising from the minimal length can be derived by using eq .
( [ schroed ] ) and coupling the particle to the electromagnetic field @xmath142 in the usual way @xmath143 analyzing this equation , one finds a constraint on the new scale in the range of the constraints from lxds : @xmath144 tev .
one of the main motivations for the inclusion of a minimal length scale into the framework of quantum field theory in extra dimensions , is the necessity of an ultraviolet regulator for quantum gravity . in this regard ,
it is not surprising that in the above discussed model with a minimal length scale , @xmath113 acts as a natural regulator@xcite .
in contrast to the @xmath145-dimensional cases , in a higher dimensional space - time , the emerging divergences can not be removed by applying an appropriate renormalization scheme ( e.g. dimensional regularization ) as every regularized result will depend on the regulator and must carefully be investigated as to its interpretation .
the minimal length therefore reliably removes an inherent arbitrariness of choice of regulators in higher dimensional quantum field theories . to see this ,
consider a typical kk - loop amplitude with an external momenta restricted to the brane @xmath146 in @xmath147 dimensions : @xmath148\left[(p - q)_x^2+m_n^2\right ] } \quad , \label{loop}\end{aligned}\ ] ] where @xmath149 are the apparent masses of the kk - modes and @xmath150 is a polynom of order two whose precise ( tensorial ) structure is determined by the polarization tensor of the gravitons@xcite . by taking the continuum limit @xmath151 , eq .
( [ loop ] ) can be written as a @xmath147 dimensional integral which is badly divergent @xmath152\left[(p - q)_x^2+p_y^2\right ] } \quad . \label{loop2}\end{aligned}\ ] ] within the minimal length scenario , the one loop amplitude reads @xmath153\left[(p - q)_x^2+p_y^2\right ] } \quad , \label{loop3}\end{aligned}\ ] ] and is finite due to the exponentially squeezed phase space which appears via the jacobian determinant . naturally , the result of this integration depends on the regulator @xmath113 , but this quantity is no longer arbitrary and instead signals physically motivated corrections by the minimal length .
the so regularized loop - corrections have been used in ref.@xcite for the calculation of the virtual graviton contributions .
the occurrence of a minimal observable length scale arises from the attempt to include quantum gravity into the standard model of particle physics .
should our spacetime turn out to have additional large compactified dimensions , the effects of the minimal length would become testable soon . proposing a model to include a minimal length ,
we have shown that observable effects have to be expected in high precision and high energy experiments .
the predictions from the extra dimensional scenario are modified , virtual contributions are naturally regularized .
furthermore , the existence of a minimal scale necessarily leads to a modification of the lorentz - transformations , which might explain the observed ultra - high energetic cosmic ray events .
the central issue of the minimal length , however , remains the fundamental finiteness of space - time resolution , which sets a limit to what future experiments possibly can achieve .
this limit might be reached soon .
i thank stefan hofmann for clarifying the issue of loop quantum gravity .
i would also like to thank k. k. phua for the possibility to write this brief review .
l. j. garay , int .
j. mod .
phys . a * 10 * , 145 ( 1995 ) .
y. j. ng , mod .
a * 18 * , 1073 ( 2003 ) .
d. j. gross and p. f. mende , nucl .
b * 303 * , 407 ( 1988 ) ; k. konishi , g. paffuti and p. provero , phys .
b * 234 * , 276 ( 1990 ) ; d. amati , m. ciafaloni and g. veneziano , phys .
b * 216 * ( 1989 ) 41 ; t. yoneya , mod .
lett . a * 4 * , 1587 ( 1989 ) . c. rovelli , living rev .
* 1 * , 1 ( 1998 ) ; a. perez , class .
grav . * 20 * , r43 ( 2003 ) ; t. thiemann , lect .
notes phys .
* 631 * , 41 ( 2003 ) ; a. ashtekar and j. lewandowski , class .
* 21 * , r53 ( 2004 ) .
r. j. szabo , phys .
rept . * 378 * , 207 ( 2003 ) .
m. r. douglas and n. a. nekrasov , rev .
phys . * 73 * , 977 ( 2001 ) .
g. t hooft , [ arxiv : gr - qc/9310026 ] , l. susskind , j. math .
phys . * 36 * , 6377 ( 1995 ) .
y. j. ng and h. van dam , [ arxiv : gr - qc/0403057 ] .
t. padmanabhan , phys .
lett . * 78 * , 1854 ( 1997 ) ; a. smailagic , e. spallucci and t. padmanabhan , [ arxiv : hep - th/0308122 ] .
e. p. wigner , rev .
* 29 * , 255 ( 1957 ) , h. salecker and e. p. wigner , phys .
rev . * 109 * , 571 ( 1958 ) , j. d. barrow , phys .
d * 54 * , 6563 ( 1996 ) , n. sasakura , prog .
* 102 * , 169 ( 1999 ) . f. scardigli , phys .
b * 452 * , 39 ( 1999 ) ; t. padmanabhan , t. r. seshadri and t. p. singh , int .
j. mod .
phys . a * 1 * , 491 ( 1986 ) .
x. calmet , m. graesser and s. d. h. hsu , [ arxiv : hep - th/0405033 ] ; v. daftardar and n. dadhich , int .
. phys . a * 1 * ( 1986 ) 731 ; y. j. ng and h. van dam , mod .
lett . a * 9 * , 335 ( 1994 ) . c. i. kuo and l. h. ford , phys .
d * 47 * , 4510 ( 1993 ) . l. j. garay , int .
j. mod .
a * 14 * , 4079 ( 1999 ) ; a. perez , class .
grav . * 20 * , r43 ( 2003 ) ; j. c. baez , class . quant
. grav . *
15 * , 1827 ( 1998 ) ; y. j. ng , int .
j. mod .
d * 11 * , 1585 ( 2002 ) .
a. kempf , g. mangano and r. b. mann , phys .
d * 52 * , 1108 ( 1995 ) .
m. maggiore , phys .
b * 319 * , 83 ( 1993 ) ; m. maggiore , phys .
d * 49 * , 5182 ( 1994 ) ; a. camacho , int .
j. mod .
d * 12 * , 1687 ( 2003 ) ; i. dadic , l. jonke and s. meljanac , phys .
d * 67 * ( 2003 ) 087701 ; f. brau , j. phys .
a * 32 * ( 1999 ) 7691 ; r. akhoury and y. p.
yao , phys .
b * 572 * , 37 ( 2003 ) .
u. h. danielsson , phys .
d * 66 * ( 2002 ) 023511 ; s. shankaranarayanan , class . quant
. grav .
* 20 * ( 2003 ) 75 ; l. mersini , m. bastero - gil and p. kanti , phys .
d * 64 * ( 2001 ) 043508 ; a. kempf , phys .
d * 63 * ( 2001 ) 083514 ; a. kempf and j. c. niemeyer , phys .
d * 64 * ( 2001 ) 103501 ; j. martin and r. h. brandenberger , phys .
d * 63 * ( 2001 ) 123501 ; r. easther , b. r. greene , w. h. kinney and g. shiu , r. h. brandenberger and j. martin , mod .
phys . lett .
a * 16 * ( 2001 ) 999 ; r. casadio and l. mersini , int .
j. mod .
a * 19 * , 1395 ( 2004 ) ; g. l. alberghi , r. casadio and a. tronconi , phys . lett .
b * 579 * , 1 ( 2004 ) .
i. antoniadis , phys .
b * 246 * , 377 ( 1990 ) . n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 429 * , 263 ( 1998 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 436 * , 257 ( 1998 ) ; n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
d * 59 * , 086004 ( 1999 ) .
l. randall and r. sundrum , phys .
* 83 * , 4690 ( 1999 ) .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) .
t. appelquist , h. c. cheng and b. a. dobrescu , phys .
d * 64 * , 035002 ( 2001 ) ; c. macesanu , c. d. mcmullen and s. nandi , phys . rev .
d * 66 * , 015009 ( 2002 ) ; t. g. rizzo , phys .
d * 64 * , 095010 ( 2001 ) .
j. c. long and j. c. price , comptes rendus physique * 4 * , 337 ( 2003 ) ; c. d. hoyle _ et al _ , phys . rev .
d * 70 * , 042004 ( 2004 ) ; c. d. hoyle _ et al _ , phys .
lett . * 86 * , 1418 ( 2001 ) ; j. chiaverini _ et al _ , phys .
* 90 * , 151101 ( 2003 ) .
j. hewett and m. spiropulu , ann .
nucl . part .
* 52 * , 397 ( 2002 ) ; t. g. rizzo , phys .
d * 64 * , 095010 ( 2001 ) ; t. han , j. d. lykken and r. j. zhang , phys .
d * 59 * ( 1999 ) 105006 ; s. cullen , m. perelstein and m. e. peskin , phys . rev .
d * 62 * , 055012 ( 2000 ) .
m. b. voloshin , phys .
b * 518 * , 137 ( 2001 ) ; phys . lett .
b * 524 * , 376 ( 2002 ) ; s. b. giddings , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , econf * c010630 * , p328 ( 2001 ) .
s. n. solodukhin , phys .
b * 533 * , 153 ( 2002 ) ; a. jevicki and j. thaler , phys . rev .
d * 66 * , 024041 ( 2002 ) ; t. g. rizzo , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , econf * c010630 * , p339 ( 2001 ) ; d. m. eardley and s. b. giddings , phys .
rev . * d 66 * , 044011 ( 2002 ) .
v. s. rychkov , phys .
d * 70 * , 044003 ( 2004 ) ; k. kang and h. nastase , [ arxiv : hep - th/0409099 ]
. s. dimopoulos and g. landsberg phys .
lett . * 87 * , 161602 ( 2001 ) ; p.c .
argyres , s. dimopoulos , and j. march - russell , phys .
b441 * , 96 ( 1998 ) .
p. kanti , [ arxiv : hep - ph/0402168 ] ; g. landsberg , [ arxiv : hep - ph/0211043 ] ; m. cavaglia , int .
j. mod .
phys . a * 18 * , 1843 ( 2003 ) .
k. cheung , [ arxiv : hep - ph/0409028 ] .
s. hossenfelder , m. bleicher , s. hofmann , j. ruppert , s. scherer and h. stocker , phys .
b * 575 * , 85 ( 2003 ) .
w. g. unruh , phys .
d * 51 * , 2827 ( 1995 ) .
g. amelino - camelia , nature * 418 * , 34 ( 2002 ) ; j. magueijo and l. smolin , phys .
lett . * 88 * , 190403 ( 2002 ) ; m. toller , mod .
a * 18 * , 2019 ( 2003 ) ; c. rovelli and s. speziale , phys . rev .
d * 67 * , 064019 ( 2003 ) .
s. hawking , commun . math .
87*,395 ( 1982 ) ; j. preskill , arxiv : hep - th/9209058 . m. cavaglia , s. das and r. maartens , class .
grav . * 20 * , l205 ( 2003 ) ; m. cavaglia and s. das , [ arxiv : hep - th/0404050 ] .
f. w. stecker , j. phys .
g * 29 * , r47 ( 2003 ) .
a. mucke , r. engel , j. p. rachen , r. j. protheroe and t. stanev , comput .
commun .
* 124 * , 290 ( 2000 ) .
k. greisen , phys .
* 16 * , 748 ( 1966 ) ; g. t. zatsepin and v. a. kuzmin , jetp lett .
* 4 * , 78 ( 1966 ) [ pisma zh .
teor . fiz .
* 4 * , 114 ( 1966 ) ] .
t. jacobson , s. liberati and d. mattingly , phys .
d * 67 * , 124011 ( 2003 ) ; g. amelino - camelia , int .
j. mod .
d * 12 * , 1211 ( 2003 ) ; s. sarkar , mod .
a * 17 * , 1025 ( 2002 ) ; j. r. chisholm and e. w. kolb , phys .
d * 69 * , 085001 ( 2004 ) ; j. w. moffat , int .
j. mod .
d * 12 * , 1279 ( 2003 ) ; l. gonzalez - mestres , aip conf .
proc . *
558 * , 878 ( 2001 ) ; o. bertolami and c. s. carvalho , phys .
d * 61 * , 103002 ( 2000 ) ; g. amelino - camelia and t. piran , phys .
d * 64 * , 036005 ( 2001 ) .
u. harbach , s. hossenfelder , m. bleicher and h. stocker , phys .
b * 584 * , 109 ( 2004 ) . | planck scale physics represents a future challenge , located between particle physics and general relativity .
the planck scale marks a threshold beyond which the old description of spacetime breaks down and conceptually new phenomena must appear .
little is known about the fundamental theory valid at planckian energies , except that it necessarily seems to imply the occurrence of a minimal length scale , providing a natural ultraviolet cutoff and a limit to the possible resolution of spacetime .
motivated by string theory , the models of large extra dimensions lower the planck scale to values soon accessible .
these models predict a vast number of quantum gravity effects at the lowered planck scale , among them the production of tev - mass black holes and gravitons . within the extra dimensional scenario ,
also the minimal length comes into the reach of experiment and sets a fundamental limit to short distance physics .
we review the status of planck scale physics in these effective models . |
during the past decade the use of entanglement in the study of complex quantum systems has developed at an increasingly fast pace @xcite .
extended analyses have established the monotonic scaling of the von neumann entropy in the ground state of spin models @xcite , the so called _ area - law _ , and its profound relations both with conformal field theory ( cft ) @xcite and with the so - called `` majorization '' of the entanglement along renormalization group flows in quantum spin chains @xcite .
the von neumann entanglement entropy is in general extremely hard to compute and its measurement requires complete state tomography @xcite . on the other hand , it has been argued that the rnyi entropies of higher order contain substantial information about the universal properties of a quantum many - body system @xcite . in particular , the rnyi entropy of order @xmath0 is directly related to the purity of the ground - state block reduced density matrix , is significantly easier to compute compared to the von neumann entropy ( i.e. the rnyi entropy of order @xmath1 ) , and can in principle be measured directly @xcite . moreover , characterizing the behavior of the rnyi block entanglement entropies is equivalent to characterizing the entanglement spectrum , i.e. the entire set of eigenvalues of the ground - state reduced block density matrix @xcite . indeed ,
these quantities play a relevant role in determining the scaling properties of numerical algorithms based on matrix product states @xcite .
the rnyi entropies are also a useful tool to determine the continuous or discontinuous nature of a phase transition @xcite and to estimate quasi - long - range order in low - dimensional systems @xcite .
furthermore , the concept of topological entanglement entropy @xcite can be extended to the rnyi entropies for which it has been shown to coincide with the total quantum dimension @xcite .
concerning the relation of the entanglement spectrum to quantum phase transitions , it has been recently shown that a particular subset of such spectrum , that is the difference between the two largest eigenvalues of the reduced density matrix , the so - called schmidt gap , plays the role of an effective order parameter in one - dimensional quantum spin-@xmath1 models endowed with haldane topological phases @xcite . above all
, there is a growing awareness that the entanglement properties of quantum ground states provide the most fundamental characterization of quantum phases of matter beyond the traditional approach based on symmetry breaking , especially when considering quantum phases of matter that are established in the absence of symmetry breaking and local order parameters , and are due , e.g. , to the presence of hidden topological order associated to a non - vanishing topological component of the ground - state entanglement entropies @xcite . in this perspective
, the entanglement spectrum and the topological components of the rnyi entropies are being actively investigated in various problems at the cutting edge of condensed matter physics , including bose - hubbard spin liquids @xcite , frustrated models on nontrivial lattice geometries @xcite , non - abelian fractional hall systems @xcite , and low - dimensional gapless models @xcite . when the rnyi entropies are computed for blocks @xmath2 of @xmath3 spins in the ground state of critical models at zero field ,
it is observed that , for order @xmath4 , they are characterized by large sub - leading corrections that violate monotonicity in the size of the block ( @xmath3 ) , while no effect appears in the ising chain at any value of the transverse field @xcite .
the presence of this oscillatory behavior has been explained thus far as a peculiar characteristics of critical models , due to parity effects , i.e. depending on whether the number of spins in a given partition is even or odd . in the present work
we show that the oscillatory scaling is a universal feature of the entanglement spectra and of the rnyi entropies in the ground state of critical and noncritical one - dimensional models and that this behavior is related to the existence of a factorizing field , located always below or coinciding with the critical field @xcite , at which the system hamiltonian admits at least one fully separable ground state . using exact analytical methods for integrable models and numerical methods for non - integrable ones ,
we show that all one - dimensional models admitting ground - state factorization exhibit two different scaling regimes , separated by the factorizing field @xmath5 . in the region @xmath6 the schmidt gap and the rnyi block entropies exhibit a non monotonic scaling in the size @xmath3 of the block , while monotonicity is restored for @xmath7 .
these oscillations are associated to a series of crossovers between the two entangled eigenvectors of the reduced block density matrix associated to the two largest eigenvalues . increasing the size of the block the schmidt gap closes exponentially in the symmetry broken phase but with a ratio that increases as @xmath8 decrease for @xmath9 . on the contrary
, it becomes independent of @xmath8 as soon as @xmath10 .
we will relate this general scaling structure in one dimension to a finite - range , entanglement - driven order that is captured by introducing a global order parameter of entanglement .
the paper is organized as follows . in section
ii we introduce the class of @xmath11 and @xmath12 models in transverse field and discuss , using exact analytic methods , the scaling of the rnyi block entropies in the ground state of such models , either gapped or gapless , in the thermodynamics limit . in all cases
, we show that the factorizing field @xmath5 separates two different scaling regimes in the behavior of the rnyi block entropies , either as functions of the transverse field @xmath8 or of the size of the block @xmath3 . in section
iii we analyze the behavior of the entanglement spectrum as a function of @xmath3 and we show how the oscillatory behavior is associated to crossovers between the eigenvectors of the block reduced density matrix corresponding to the two largest eigenvalues . as a consequence ,
we show how the schmidt gap , i.e. the difference between the two largest eigenvalues , captures most of the significant aspects of the behavior of the full entanglement spectrum . in section
iv we analyze the detailed features of the oscillatory scaling , showing that it is characterized by a set of frequencies that depend only on the normalized transverse field @xmath13 . as a consequence , a series of ordered entangled structures in terms of quasi - dimers , quasi - trimers , and quasi - polymers is identified for all fields below the factorization point .
this form of quantum order is characterized by a global parameter of entanglement defined as the integral , over all block lengths , of the rnyi entropy of infinite order . in section
v we extend our investigation to non - integrable models of the @xmath14 class . at variance with the @xmath11 , @xmath12 , and ising classes ,
non - integrable models may not always admit ground - state factorization points . using numerical methods , in particular the density matrix renormalization group ( dmrg ) algorithm on systems of large but finite size ( up to @xmath15 spins ) ,
we discuss analogies and differences between the scaling of the entanglement spectrum in models that admit and models that do not admit factorized ground states .
to set the stage let us first consider the class of translationally invariant , one - dimensional @xmath11 spin-1/2 hamiltonians @xmath16 where @xmath17 @xmath18 stands for the spin-@xmath19 pauli operator on site @xmath20 , @xmath8 is the external transverse field , and @xmath21 is the anisotropy , taking values in the interval @xmath22 $ ] , whose extremes correspond , respectively , to the fully isotropic , gapless @xmath12 model and to the maximally anisotropic , gapped ising chain . for every @xmath21 ,
when @xmath8 takes the value @xmath23 the system admits two degenerate , fully factorized ground states that are products of single - site states @xcite . for anisotropy @xmath24 such local states
are not , in general , eigenstates of @xmath25 .
hence , their tensor product is not an eigenstate of the parity operator along @xmath26 . on the contrary , their coherent linear symmetric superpositions ( with positive or negative relative sign )
define two degenerate ground states of definite parity ( respectively even or odd ) whose entanglement , in general , does not vanish . for @xmath27
the two factorized ground states collapse into a single state that preserve all the symmetries of the hamiltonian . in the thermodynamic limit , regardless of the value of @xmath21 , at @xmath28 the system undergoes a quantum phase transition in the @xmath11-plane . for @xmath29 and @xmath30
the system is characterized by a two - fold degenerate ground space and a gapped energy spectrum with a magnetic order along the @xmath31 axis , while for @xmath32 and @xmath30 the ground state is unique and the energy spectrum is gapless .
the presence of possible degeneracies in the ground space and the fact that different ground states may be characterized by different entanglement properties require , before proceeding further , that we fix the class of ground states on which the investigation shall be focused .
indeed , in order to capture the possible universal aspects in the scaling behavior of the entanglement spectrum and of the rnyi block entropies , one needs to treat the gapped case @xmath24 consistently with the isotropic case @xmath33 and with the paramagnetic phase @xmath34 , since in these two instances the system hamiltonian admits only a single , non - degenerate ground state that preserves all the symmetries .
therefore , from now on we will consider , also in the magnetically ordered phases of gapped models ( @xmath24 and @xmath35 ) , only the class of ground states that preserve all the symmetries of the hamiltonian , including parity .
that is , we will consider only states of definite fixed parity , either even or odd
. of course , also the case of states with no definite parity is physically very relevant . on the other hand ,
it is technically much more complex , since in such instance one needs to resort to challenging numerical investigations even for the simplest , exactly solvable hamiltonians like the one - dimensional @xmath11 model
. therefore the case of parity non - conserving ground states will be attacked and investigated carefully in a separate work .
models of the @xmath11 and @xmath12 class play a relevant role in the field of quantum statistical mechanics because they are exactly solved by resorting to the jordan - wigner transformations of the spin operators into pseudo - fermion operators @xcite . indeed , given a bipartition of our one - dimensional lattice of total size @xmath36 into a block of @xmath3 spins and a remainder of size @xmath37 , it is always possible to obtain , for ground states of fixed parity , an exact analytical expression of the reduced block density matrix @xmath38 , i.e. of the block reduced state obtained by tracing the ground state over the degrees of freedom of the remainder .
knowledge of @xmath38 allows to compute exactly the associated rnyi block entropies , defined as : @xmath39 \ ; . \label{renyi}\ ] ] in fig .
[ figxyfunh ] we report the behavior of the rnyi entropies @xmath40 , both in the gapless and gapped cases , as functions of the external field @xmath8 for blocks of spins of different size @xmath3 .
all calculations are exact and are performed for the one - dimensional @xmath11 , @xmath12 , and ising models in the thermodynamic limit . as functions of the transverse field @xmath8 for blocks of different sizes @xmath3 , respectively for gapped ( @xmath41 left column ) and gapless ( @xmath32 right column ) models .
dotted black line : @xmath42 ; dot - dashed red line : @xmath43 ; solid green line : @xmath44 ; dashed blue line : @xmath45 . from top to bottom : plot of the rnyi entropies @xmath46 , @xmath47 , and @xmath48 .
] we observe that both in the critical and noncritical cases the entropy @xmath49 , i.e. the block von neumann entropy , exhibits a monotonic behavior both in @xmath8 and in @xmath3 .
however , as soon as @xmath50 increase an oscillatory behavior is observed in the region @xmath51 .
specifically , for @xmath52 and @xmath53 the rnyi entropies violate monotonicity in @xmath8 while remaining monotonically non - decreasing in @xmath3 : @xmath54 .
the amplitude of the oscillations increases as @xmath3 and @xmath55 increase . for @xmath56
the scaling becomes non - monotonic also in the block size : @xmath57 .
as already discussed previously , notice that at the factorizing field @xmath5 , the entanglement does not vanish , in general , in the parity preserving ground states .
however , unambiguous signatures of factorization are readily identified also in the symmetry protected sectors . in particular , for every value of @xmath3 , at factorization the block entanglement in parity - symmetric ground states becomes independent on the relative distance between the different spins in the block .
viceversa , all entanglement - related quantities vanish , correctly , in the non - symmetric fully separable ground states at the factorization point . as functions of the block size @xmath3 for different values of the index @xmath55 and of the external field @xmath8 .
left column : gapped case with @xmath41 .
right column : gapless case ( @xmath58 ) .
black dotted line : @xmath59 ; red solid line : @xmath60 ; blue dashed line : @xmath61 . left column , uppermost and central panels : non - monotonic behavior of @xmath62 below factorization ( @xmath63 ) .
lowermost panel : restored monotonic behavior above factorization ( ) , yet below criticality .
panels on the right column : oscillatory scaling in the gapless case , for all values of @xmath8 up to factorization that coincides with criticality : @xmath64 . ]
the central role of the factorizing field for the scaling of the rnyi block entropies is illustrated also in fig .
[ figxyfunm ] , where they are analyzed , for different value of the external field , as functions of the block size @xmath3 both in gapped and gapless models . for @xmath53
the entropies of order @xmath65 exhibit damped oscillations that violate the area law behavior . as @xmath8 is increased , the frequency and the amplitude of the oscillations decrease ; the area - law monotonic scaling is restored exactly at factorization .
oscillations are exponentially damped in noncritical ( anisotropic ) models and only weakly algebraically damped in critical ( isotropic ) systems as the monotonic area - law behavior is restored at the factorization point , which , for gapped models , means well before the quantum critical point , the oscillatory scaling behavior must be directly related to the patterns of entanglement in the ground state , independently of the onset of the long - range magnetic order at the critical point .
in order to gain a full understanding of the anomalous scaling in relation to the patterns of ground - state entanglement , we turn now to the analysis of the entanglement spectrum , that is , the entire set @xmath66 of the eigenvalues of the reduced density matrix @xmath38 . the oscillatory behavior of the rnyi entropies becomes ever more pronounced as the order @xmath55 grows .
correspondingly , it follows from the definition of the rnyi entropies , eq .
( [ renyi ] ) , that the relative weight of the largest eigenvalues of the reduced density matrix @xmath38 increases . since the characterizations either in terms of the full entanglement spectrum or in terms of the entire hierarchy of the rnyi entropies must be equivalent
, we may expect that only the largest eigenvalues of @xmath38 will contribute significantly to the scaling of the rnyi entropies .
indeed , this is verified both in the gapped and gapless cases , as shown in fig .
[ entanglementspectrum ] . as a function of the external field @xmath8 for different values of the block size @xmath3 .
upper panel : anisotropic gapped models ( @xmath41 ) .
lower panel : isotropic gapless models ( @xmath32 ) . in both cases
the relevant contributions are mainly due to the two largest eigenvalues ( red line and blue line ) compared to the sum of the remaining eigenvalues ( green line ) , the more so as factorization and criticality are approached .
oscillations increase with increasing size of the block , and terminate asymptotically at the factorization point .
criticality is signaled by the opening of the schmidt gap between the two largest eigenvalues.,title="fig:",width=302 ] as a function of the external field @xmath8 for different values of the block size @xmath3 .
upper panel : anisotropic gapped models ( @xmath41 ) .
lower panel : isotropic gapless models ( @xmath32 ) . in both cases the relevant contributions are mainly due to the two largest eigenvalues ( red line and blue line ) compared to the sum of the remaining eigenvalues ( green line ) , the more so as factorization and criticality are approached .
oscillations increase with increasing size of the block , and terminate asymptotically at the factorization point .
criticality is signaled by the opening of the schmidt gap between the two largest eigenvalues.,title="fig:",width=302 ] from fig .
[ entanglementspectrum ] we see that the two largest eigenvalues in the entanglement spectrum exhibit an oscillating scaling behavior analogous to that of the rnyi entropies , with growing frequency of the oscillations as the size of the partition increases .
moreover , for partitions of increasing size @xmath3 the ending point of the oscillations converges , asymptotically , to the factorization point @xmath5 , where @xmath67 in gapped systems and coincides with it in critical model .
the approach to quantum criticality is signaled by the opening of the schmidt gap , i.e. the difference between the two largest eigenvalues @xcite .
classical saturation to a paramagnetic product ground state at high fields @xmath8 occurs asymptotically for @xmath34 in the gapped case , and instantaneously at @xmath68 in the gapless case .
given these results , besides considering the entire entanglement spectrum , it is sensible to analyze in more detail the scaling of the schmidt gap @xmath69 , whose importance has been recently discussed in the analysis of the quantum phase diagram of some one - dimensional spin-1 models @xcite .
as function of the block size @xmath3 for different values of @xmath8 and @xmath21 .
main panel : gapped case ( @xmath41 ) .
black squares : ; red circles : @xmath70 ; green up - triangles : @xmath71 ; blue down - triangles : @xmath72 .
inset : gapless case ( @xmath32 ) .
black squares : ; red circles : .,width=302 ] in fig .
[ schmidtgap ] we report the behavior of the schmidt gap @xmath69 as function of the block size @xmath3 for different value of the external field @xmath8 both in critical and non - critical models . as @xmath3 increases , the schmidt gap exhibits an overall envelope that is exponentially decreasing in the ordered phase , @xmath73 , in full agreement with the presence of a gap in the energy spectrum . on the other hand , besides the overall exponential envelope , fig .
[ schmidtgap ] shows two clearly different behaviors : one for @xmath74 and one for @xmath75 . in the first region
the schmidt gap undergoes a net exponential decay as @xmath8 decreases .
when the system enters the region @xmath76 an oscillating behavior is superimposed on the exponential envelope , due to a series of crossovers between the eigenvectors corresponding to the two largest eigenvalues of the reduced block density matrix .
contrary to what happens in the first region , the argument of the exponential decay becomes independent of the external field .
being independent of @xmath8 , the argument of the exponential can then be evaluated analytically at factorization .
we obtain that @xmath77 , for obeys the relation @xmath78 , where @xmath79 d \phi \right|$ ] , and @xmath80 contains slowly , non - exponentially decaying oscillating terms and tends to unity as @xmath8 tends to @xmath5 . in the gapless case , see inset of fig .
[ schmidtgap ] , for all @xmath81 the schmidt gap exhibits a power law decay modulated by an oscillating dependence on @xmath3 , in agreement with the fact that the system is in a critical region and that , being @xmath82 , the external field lies always below its factorizing value .
these observations indicate that there are universal aspects in the scaling of the entanglement spectrum in one dimension that are independent of the critical or non - critical nature of the systems considered .
these universal effects can then be captured , as reported in fig .
[ residualschmidtgap ] , by analyzing the reduced schmidt gap @xmath83 , i.e. the ratio of the schmidt gap parametrically dependent on the external field @xmath8 and on the anisotropy @xmath21 to the schmidt gap evaluated at the factorizing field @xmath5 , as a function of the block size @xmath3 .
[ residualschmidtgap ] shows that the oscillation frequency in the scaling of the reduced entanglement spectrum is a universal aspect common to critical and noncritical models , the former differing from the latter only in the overall exponentially decaying envelope whose presence depends on the existence of a gap in the energy spectrum .
the non - perfect superposition of the different curves is partly due to the finite numerical precision and likely also due to the fact that this is an effective , emerging form of universality . as a function of the block size @xmath3 for different values of the external field @xmath8 and the anisotropy @xmath21 .
uppermost panel : @xmath84 .
central panel : @xmath85 .
lowermost panel:@xmath86 . in all panels : red solid line : gapless case ( @xmath87 ; black dotted line : gapped case with @xmath88 ; green dashed line : gapped case with @xmath89 .
the equal oscillation frequency is a universal effect common to critical and noncritical models .
, width=302 ] the scaling of the schmidt gap with the external field @xmath8 is reported in fig .
[ schmidtgapfuncofh ] .
it reports the schmidt gap as a function of @xmath90 for different values of the block size in the thermodynamic limit .
the frequency of the oscillations increases for growing @xmath3 .
the oscillatory scaling terminates at the value of @xmath91 corresponding to the factorizing field , and @xmath69 opens , independently of @xmath3 , at the critical point @xmath92 . as a function of @xmath90 for the @xmath11 model with anisotropy @xmath93 in the thermodynamic limit , for different values of the block size @xmath3 .
blue diamonds : @xmath94 ; black squares : @xmath95 ; red dots : @xmath96.,width=302 ]
the above results for the scaling of the entanglement spectrum and of the rnyi entropies imply that the factorizing field @xmath5 is itself to be considered a quantum critical point , of a different nature with respect to the standard one @xmath92 and coinciding with it for gapless models .
exactly at @xmath97 one - dimensional systems undergo a further phase change beyond the quantum phase transition occurring at the critical point @xmath92 . in the region
@xmath98 a finite - range , entanglement - driven order is established that is superimposed to the global magnetic one driven by spontaneous symmetry breaking .
the implications of the anomalous scaling of the entanglement spectrum for a transitions between different patterns of entanglement in the ground state can be understood by recalling that the rnyi entropy of infinite order coincides with the single - copy bipartite entanglement @xcite and is monotonic in the bipartite geometric entanglement @xcite , where @xmath99 is the largest eigenvalue of the ground - state block reduced density matrix @xmath38 . given a system of total size @xmath36 , @xmath100 is the minimum distance between the ground state and the set of pure bi - separable states @xmath101 , where @xmath102 is a state of a block @xmath3 , and @xmath103 is a state of the remainder of the chain . from the last panel of fig .
[ figxyfunm ] one sees that above factorization @xmath48 and hence also @xmath100 increase monotonically in @xmath3 . on the contrary ,
when @xmath10 , fig .
[ figxyfunm ] show that @xmath100 and hence the distance of the ground state from the set of biseparable states oscillate as a function of @xmath3 , with the amplitude of the oscillations increasing for decreasing @xmath8 and @xmath21 . in particular ,
recalling the definition of @xmath104 , we see from fig .
[ entanglementspectrum ] that at sufficiently small fields the ground state is closest to the bi - separable state @xmath105 .
this implies that well below factorization ( @xmath106 ) the ground state tends to order in dimerized domains , i.e. entangled structures that involve only two spins in a maximally entangled bell state .
a similar oscillating behavior in the eigenstates of the block reduced density matrices is observed in models that possess exactly dimerized ground states , either due to the presence of competing interactions of different spatial range or due to complex spatial patterns of the nearest - neighbor interaction couplings , as is the case , respectively , of the @xmath107 model at the majumdar - ghosh point @xcite and of models admitting long - distance end - to - end entanglement in the ground state @xcite .
however , at variance with such models , in the class of the @xmath11 models with nearest - neighbor interactions and spatially constant coupling amplitudes , the oscillatory behavior holds only for the rnyi entropies of order @xmath108 and therefore dimerization is not exact .
the fact that dimerization is only partial is due to the fact that the relative weight of all the remaining eigenstates of the reduced density matrix with respect to the one of largest amplitude never becomes so small to be put to zero , contrary to what happens , e.g. , in the model for which it vanishes at each fixed value of @xmath3 . from fig .
[ residualschmidtgap ] we see that increasing the strength of the transverse field from values @xmath106 to values progressively closer to the factorization point , the frequency of the oscillations reduces progressively .
this change in the periodicity of the oscillations signals the growth in the spatial dimension of the ordered entangled structures .
the latter gradually evolve from strongly entangled domains of quasi - dimers , leaving place to larger structures ranging from quasi - trimers up to quasi - polymers , until their size diverges exactly at the factoring field . in other words , as one moves away from vanishingly small values of the external field and approaches the factorization point , the associated entangled structures become spatially more extended . at the same time , as these extended structures develop close to factorization and involve ever larger number of sites , the associated entanglement becomes smaller and smaller in each pair of sites , in full agreement with the behavior of the pairwise concurrence ( entanglement of formation ) close to a factorization point @xcite .
as the oscillatory behavior is maximized by the single - copy entanglement @xmath109 , the finite - range , entanglement - driven order below factorization is naturally characterized by an order parameter of entanglement defined as the integral of @xmath109 over partitions of all sizes : @xmath110 \ ; . \label{op}\ ] ] in fig .
[ opfig ] we report the contour plot of @xmath111 for the class of one - dimensional @xmath11 models .
it vanishes identically for @xmath112 .
approaching the ising limit it is strongly reduced , corresponding to strongly damped oscillations . approaching the @xmath12 limit , as the oscillations become long - ranged , @xmath111 increases and diverges exactly at @xmath113 , consistently with the fact that in gapless models below the critical point the oscillations persist , with weak algebraic decay , also for large @xmath3 . as a function of @xmath8 and @xmath21 .
left inset : one - dimensional projection as a function of @xmath8 for different values of @xmath21 .
red dotted line : @xmath114 ; green solid line : @xmath93 ; black dashed line : @xmath115 .
right inset : one - dimensional projection as a a function of @xmath21 for different values of @xmath8 .
red dotted line : @xmath116 ; green solid line : @xmath117 ; black dashed line : @xmath118 . in both insets
the vertical grid lines correspond to the factorization points . ]
( [ op ] ) and fig .
[ opfig ] show that the quasi - dimerized order below factorization is exclusively due to the entanglement properties of each individual ground state and therefore , unlike magnetic order , it is not a consequence of ground - state degeneracy and symmetry breaking .
we will now show that the results of the above analysis extend as well to generic non - integrable models of the @xmath14 and heisenberg type : @xmath119 here @xmath120 are the spin - spin couplings along the @xmath121 directions and , without loss of generality , we set @xmath122 , while , contrary to the integrable cases discussed above , factorized ground states do not necessarily exist . at each finite value of the external field ,
the factorizing field is defined as @xmath123 , and therefore it exists if and only if @xmath124 @xcite .
the models in eq .
( [ hamiltonian2 ] ) are not exactly solvable . in order to determine the block reduced density matrices we diagonalized the system by means of the density matrix renormalization group ( dmrg ) @xcite applied to open chains of up to @xmath15 spins .
we kept up to @xmath125 states of the reduced density matrix , letting the truncation error stay well below @xmath126 at each step . for models admitting a factorization point ,
the results of the @xmath11 case carry over essentially unmodified to the general @xmath14 instance .
models with @xmath127 , @xmath128 , @xmath129 , and @xmath130 .
main plot : single - copy entanglement @xmath48 as a function of @xmath8 for different values of the block size @xmath3 .
black squares : @xmath131 ; red circles : @xmath132 ; green up - triangles : @xmath133 ; blue down - triangles : @xmath134.uppermost right inset : von neumann entropy @xmath46 as a function of @xmath8 for different values of @xmath3 .
lowermost left inset : schmidt gap @xmath69 as a function of @xmath3 at different values of @xmath8 .
black squares @xmath135 ; red circle @xmath136 ; green up - triangles @xmath137 . ] in fig .
[ figwith ] we report the scaling of the rnyi entropies in @xmath14 models that possess a factorizing field @xmath5 ( @xmath124 ) and the scaling of the schmidt gap as a function of the block size @xmath3 for different value of the external field .
the qualitative behavior of these two quantities is analogous to that of the @xmath11 models and confirms the existence of an entanglement - induced order of quasi - dimerized domains in the region @xmath6 .
models with @xmath127 , @xmath138 , @xmath139 , that thus do not admit factorization points .
main plot : single - copy entanglement @xmath48 as a function of the external field @xmath8 for different values of the block size @xmath3 .
black squares : @xmath131 ; red circles : @xmath140 ; green up - triangles : @xmath141 ; blue down - triangles : @xmath142 .
right inset : von neumann entropy @xmath46 as a function of @xmath8 for different values of @xmath3 .
left inset : schmidt gap @xmath69 as a function of @xmath3 at different values of @xmath8 .
black squares : @xmath135 ; red circles : @xmath143 ; green up - triangles : @xmath144 . ] for models that do not admit a factorized ground state at any finite value of the magnetic field , the scaling behavior of the rnyi entropies and of the entanglement spectrum is rather different , as reported in fig .
[ figwithout ] . at variance with models that do admit ground - state factorization
, the area - law scaling with the size of the block , @xmath145 , appears to be _ always _ violated for _ all _ values of @xmath8 , as one can see from fig .
[ figwithout ] .
however , all the rnyi entropies behave smoothly and do not acquire local maxima . on the other hand ,
when considered as functions of the block size @xmath3 ( at fixed external field @xmath8 ) , the block entropies exhibit the same oscillating behavior as in models with factorized ground states .
these features can be intuitively understood by considering that in these models factorization , so to speak , is shifted away towards infinitely large values of @xmath8 , indeed corresponding to a classical saturation .
the latter is the physical limiting situation in which an extremely strong transverse field orients all spins along the field direction , resulting in a classical paramagnetic state that is naturally a product of single - site paramagnets , with all quantum correlations washed away . indeed , as we can see from the left inset of fig .
[ figwithout ] , the true quantum oscillations in the schmidt gap appear to be confined up to a value of the external field @xmath146 , which , somehow , corresponds to an _ inverted , de - complexified _ factorization point .
therefore , a sort of would - be factorizing field seems to play a very important role also in models that do not admit a physical factorized ground state .
a complete , rigorous confirmation of these behaviors for all values of the transverse field @xmath8 in non - integrable models is beyond our current numerical possibilities for two reasons : firstly , the differences between rnyi entropies of the same order @xmath55 and different partition size @xmath3 fall off extremely rapidly with @xmath8 ; moreover , the limitations on current algorithms do not allow a reliable analysis for partitions of size @xmath147 .
this is due to the fact that the algorithm is based on the direct computation of the two - site density matrix of the two central sites of the dmrg ansatz .
the further density matrices , e.g. the four - site density matrix , are obtained by repeated merging of two physical sites in one computational site .
this protocol is very effective in obtaining all quantities of interest but , as the local dimension of the computational sites increases exponentially with the number of physical sites grouped together , it prevents at the moment going beyond blocks of size @xmath147 .
if further confirmed by future more powerful numerical analysis , the reported behavior would imply that the entanglement - induced order can exist also in disordered phases with vanishing local order parameters .
we have investigated the scaling behavior of the entanglement spectrum and of the rnyi entanglement entropies in the ground state of one - dimensional gapped and gapless quantum spin models .
we have showed that a violation of the area law scaling behavior occurs both in the critical and non - critical cases for all rnyi entropies of order @xmath148 and for external fields @xmath149 in models admitting ground - state factorization .
an analogous behavior is observed in the schmidt gap , that is the difference between the two largest eigenvalues of the block reduced density matrix .
the existence of a factorizing field and the associated anomalous scaling of the entanglement spectrum correspond to the existence of a quantum regime superimposed to and distinct from the ordered and disordered phases associated to the existence of ground states breaking some symmetry of the hamiltonian . unlike long - range magnetic order due to spontaneous symmetry breaking , this finite - range , entanglement - driven order does not rely on the onset of ground - state degeneracy , but is rather due to the entanglement properties of each ground state of fixed parity .
the factorizing field plays a key role in the understanding of the anomalous scaling of the entanglement spectrum by clarifying the origin of this phenomenon even in the gapless case in which the factorization point and the critical point coincide .
it also explains why the anomalous scaling is not observed in systems , like the ising model , for which @xmath150 . for non - integrable models that do not admit factorization , the oscillatory scaling appears to extend to all values of the external field , while the non - monotonic behavior of the schmidt gap appears to be confined in the region @xmath151 .
however , further numerical analyses will be required in order to reach a definite assessment for such models .
as already mentioned in the introduction @xcite , a new quantum revolution is taking place in condensed matter physics inasmuch as it is being recognized that there are kinds of quantum phases that are truly and uniquely characterized by their entanglement properties , the paramount example being the quantum spin liquid phase associated to a nonvanishing topological entanglement entropy for some classes of two - dimensional , geometrically frustrated quantum spin systems ( e.g. the heisenberg model on the kagom lattice ) .
in fact , this fundamental realization has motivated the characterization and classification of quantum phases according to equivalence classes of entanglement : for instance , ground states endowed with topological order are characterized by a constant , topological entanglement entropy of infinite spatial range .
therefore , they can not be adiabatically connected to ground states with short - range entanglement patterns without crossing a quantum phase transition . in particular , phases endowed with topological order are not compatible with factorized ground states , the latter being a particular element of the equivalence class of states with short - range entanglement @xcite .
the above holds true ( although a rigorous proof is still lacking ) for two- and higher - dimensional systems , where topological effects con occur independently of the presence or absence of symmetries . in one - dimensional systems
, it has been conjectured that topological phases can occur only if the associated ground states share some given symmetry and can not be adiabatically deformed into each other without crossing a quantum phase transition , if the deformation preserves that symmetry @xcite , leading to a classification of gapped symmetric phases in one - dimensional spin systems @xcite .
this is certainly not the case for the exchange quantum spin-1/2 chains that we have discussed in the present work . on the other hand ,
the conjecture puts the role of the factorized ground state at the center stage of the investigation , and for models that are thought of possessing true symmetry - protected topological phases in one dimension , our methods for the study of ground - state factorization could be applied in order to verify the conjecture in explicit model hamiltonians in one dimension .
frustration , a crucial ingredient for the realization of topological phases in two dimensions , appears to be an important factor in one - dimensional systems as well .
we have noticed that the single - copy entanglement @xmath48 is monotonic in the bipartite geometric entanglement @xmath100 , which in turn has been shown to be a universal lower bound to ground - state frustration , independently of the spatial dimension @xcite .
this correspondence suggests the existence of an intimate relation between the scaling of the entanglement spectrum , the tendency to form ordered entangled structures below the factorization point , and the frustration of purely quantum origin .
in such a perspective , the present investigation might be further fruitfully extended to geometrically frustrated models and systems with topological order and other classes of exotic quantum phases .
finally , the fact that the factorization point has been identified to constitute the boundary point for different types of scaling of the ground - state entanglement suggests that the presence of factorization might impose fundamental limits to trotterization and other classes of adiabatic methods @xcite for the simulatability of quantum many - body hamiltonians .
this crucial point of great practical importance will be the subject of forthcoming work .
one of us ( fi ) thanks a. hamma for inspiring exchanges .
one of us ( sm ) thanks m. rizzi for discussions .
sds , smg , and fi acknowledge support from the eu strep project iqit , grant agreement no .
smg also gratefully acknowledges the austrian science fund ( fwf - p23627-n16 ) .
sm acknowledges the pwp project ( www.dmrg.it ) for the dmrg code , the dfg ( sfb / trr 21 ) , the bwgrid , and the eu - siqs for support .
p. horodecki and a. ekert , phys .
lett . * 89 * , 127902 ( 2002 ) ; s. j. van enk and c.w . j. beenakker , phys .
lett . * 108 * , 110503 ( 2012 ) ; d. a. abanin and e. demler , phys .
lett . * 109 * , 020504 ( 2012 ) . | we investigate the scaling of the entanglement spectrum and of the rnyi block entropies and determine its universal aspects in the ground state of critical and noncritical one - dimensional quantum spin models . in all cases , the scaling exhibits an oscillatory behavior that terminates at the factorization point and whose frequency is universal .
parity effects in the scaling of the rnyi entropies for gapless models at zero field are thus shown to be a particular case of such universal behavior .
likewise , the absence of oscillations for the ising chain in transverse field is due to the vanishing value of the factorizing field for this particular model . in general
, the transition occurring at the factorizing field between two different scaling regimes of the entanglement spectrum corresponds to a quantum transition to the formation of finite - range , ordered structures of quasi - dimers , quasi - trimers , and quasi - polymers .
this entanglement - driven transition is superimposed to and independent of the long - range magnetic order in the broken symmetry phase .
therefore , it conforms to recent generalizations that identify and classify the quantum phases of matter according to the structure of ground - state entanglement patterns .
we characterize this form of quantum order by a global order parameter of entanglement defined as the integral , over blocks of all lengths , of the rnyi entropy of infinite order .
equivalently , it can be defined as the integral of the bipartite single - copy or geometric entanglement .
the global entanglement order parameter remains always finite at fields below the factorization point and vanishes identically above it . |
one of the minimalistic extensions of the standard model ( sm ) is that by one or more gauge singlet real ( or complex ) scalar field(s ) .
motivations to introduce a singlet scalar to the sm are , amongst others : ( i ) to provide a viable cold dark matter ( cdm ) candidate through higgs portal models @xcite , ( ii ) to make the electroweak phase transition a strong first - order one @xcite , and ( iii ) to address the naturalness problem of the sm higgs boson @xcite .
phenomenological aspects of such singlets have also been discussed in case of colliders @xcite , and in the context of electroweak precision constraints @xcite .
such a singlet with a mass around 750 gev may also be responsible for the recently observed excess in the diphoton channel @xcite if one adds vectorial fermions to the model .
one often imposes a @xmath1-symmetry on the scalar potential under which the sm particles are all even and the extra singlet(s ) is(are ) odd , which can make the lightest @xmath1-odd particle a cdm candidate . in an analogous way to what happens in r - parity conserving supersymmetry and universal extra dimension models , a @xmath1-symmetry on the scalar potential under which the singlet @xmath0 is odd , can in principle lead to the higgs portal dark matter models where @xmath0 constitutes the dark matter .
a necessary condition for this is zero ( or infinitesimally small ) vacuum expectation value ( vev ) for @xmath0 , so that it can not mix with the sm doublet scalar @xmath6 .
one must remember that this @xmath1 is rather ad hoc , introduced just for the sake of having a cdm candidate .
the nature of the tree - level scalar potential has also been discussed by several authors @xcite .
the potential is more complicated than the sm one because of one extra field and the possibility that the cp - even neutral component of the su(2 ) doublet @xmath6 , which will be denoted by @xmath7 , and the gauge singlet scalar @xmath0 can both have nonzero vevs .
we denote these vevs by @xmath8 and @xmath9 respectively .
if there is only one minimum for @xmath9 , one often uses the shift symmetry @xmath10 to ensure @xmath11 , which in turn ensures a cdm candidate if the model is @xmath1-symmetric .
however , if there are more than one minima for @xmath9 , there is no particular advantage in using the shift symmetry , except ensuring that @xmath11 is an extremum . in this paper
, we investigate the nature of the tree - level potential of the sm extended by one real singlet scalar , which we call sm+s , with and without the @xmath1-symmetry .
we find the allowed parameter space for three different sm+s models : ( 1 ) the higgs portal model _
i.e. _ , the @xmath1-symmetric model where @xmath0 can be a cdm candidate , ( 2 ) the @xmath1-symmetric model with no cdm candidate , and ( 3 ) the @xmath1-asymmetric model .
obviously , the allowed parameter space has to be consistent with all theoretical and current experimental constraints .
for the third case , @xmath1 breaking is soft , coming from operators with mass dimension less than 4 .
dimension-4 @xmath1 breaking operators are forbidden from gauge and/or lorentz symmetry .
as we consider only one real singlet scalar , all the couplings are real in these models .
data from the large hadron collider ( lhc ) essentially constrains the mixing between @xmath0 and @xmath7 .
the mixing angle @xmath12 is constrained to be so small that all sm+s models also satisfy the constraints coming from the oblique parameters .
we will discuss it in detail later .
the measurement of the @xmath13 boson mass also puts serious restrictions on the parameter space of sm+s @xcite .
apart from these experimental constraints , there are three other important constraints .
firstly , the potential has to be stable at all energy scales till one reaches the range of validity .
this range is , in general , way below the planck scale , apart from some exceptional choices of the parameters . above this limit , either at least one of the couplings blow up , or the potential becomes unbounded from below along some direction in the field space .
the second one is the existence of a minimum , either global or local , where @xmath14 gev , which will be referred to as the electroweak ( ew ) vacuum .
the third constraint comes from the stability of the ew vacuum ; if there is another minimum deeper than the ew vacuum , the tunnelling lifetime should not be less than the age of the universe .
we also study the effect of one - loop corrections to the potential . in general
, the one - loop corrections are expected to be small compared to the tree - level potential , unless one looks along a flat direction .
even when both @xmath0 and @xmath6 have non - zero mass terms , one can find a direction in the field space , at least for large values of the fields , following the prescriptions of gildener and weinberg @xcite .
one should also choose the regularization scale properly .
this choice , in principle , is arbitrary if one uses renormalization group ( rg ) improved couplings . on the other hand
, one can always tune the scale so that the one - loop corrected ew vacuum still has @xmath15 gev @xcite .
however , this makes the choice of the regularization scale dependent on the model parameters .
the paper is arranged as follows . in
, we give a brief outline of the singlet - enhanced sm , and discuss the constraints .
discusses the @xmath1-symmetric model where the cdm is allowed , and is on models with a non - zero singlet - doublet mixing .
while these are all tree - level results , we discuss the one - loop corrections to the potential in .
in , we summerize and conclude .
let us consider the most general potential for sm+s , the single real - scalar extended sm : v(,s ) & = & -^2 ^ - m^2 s^2 + ( ^)^2 + a_1 ^s + a_2 ^s^2 + b_1 s + b_3 s^3 + b_4 s^4 , [ pot ] where @xmath6 is the sm doublet and @xmath0 is a gauge singlet scalar field .
we denote the cp - even neutral component of @xmath6 by @xmath16 , and the vevs are given as @xmath17 , @xmath18 .
there are six new parameters in over and above those in the sm . of them
@xmath19 , @xmath20 , and @xmath21 respect the @xmath1 symmetry of @xmath22 , while @xmath23 , @xmath24 and @xmath25 break it softly .
note that @xmath26 stand for wrong - sign mass terms in the potential .
all the couplings are , of course , real , because we have only one real singlet @xmath0 in this model .
the stability conditions are obtained from the requirement that the potential should not become negative along any direction of the field space , which gives > 0 , b_4 > 0 , a_2 + 2 > 0 , [ stability ] along the directions @xmath27 , @xmath28 , and @xmath29 directions respectively . in case
there is mixing between @xmath7 and @xmath0 , the mass eigenstates @xmath30 are defined as h = + s , s = -+ s , [ mass - eigen ] where @xmath12 is the mixing angle . in terms of vevs @xmath8 and @xmath9
, the potential in becomes v(v , v_s ) = -12 ^2 v^2 - m^2 v_s^2 + 14v^4 + 12 a_1 v^2 v_s + 12 a_2 v^2 v_s^2 + b_1 v_s + b_3 v_s^3 + b_4 v_s^4 , [ pot - v - vs-1 ] and the extremization conditions are -^2 v + v^3 + a_1 vv_s + a_2 vv_s^2 & = & 0 , [ pot - min1-v ] + 12 a_1 v^2 + a_2 v^2 v_s + b_1 - 2m^2 v_s + 3b_3 v_s^2 + 4b_4 v_s^3 & = & 0 .
[ pot - min1-vs ] one can always apply the shift symmetry @xmath10 to , where @xmath31 is some constant , since this shift in @xmath0 does not change the physics .
we can use this freedom - symmetric .
however , for a @xmath1-asymmetric potential , there can be two minima with different values of @xmath8 , and removal of the tadpole at one minimum does not ensure its removal at the other . ] to remove one of the independent terms of the potential .
let us choose b_1 = - a_1 v^2 .
[ b1 ] use of in removes the linear terms in @xmath0 and simplifies the potential to v(v , v_s ) = -12 ^2 v^2 - m^2 v_s^2 + 14v^4 + 12 a_2 v^2 v_s^2 + b_3 v_s^3 + b_4 v_s^4 .
[ pot - v - vs ] the extremization conditions now guarantee one extremum line along @xmath32 and another along @xmath11 ( because of the shift symmetry ) : v ( -^2 + v^2 + a_2 v_s^2 ) & = & 0 , [ pot - min2-v ] + v_s ( a_2 v^2 - 2m^2 + 3b_3 v_s + 4b_4 v_s^2 ) & = & 0 .
[ pot - min2-vs ] from eqs .
and one finds that there are three extrema for @xmath9 , and for each @xmath9 there are three extrema for @xmath8 , depending on the existence of real solutions , there is one extremum at @xmath32 .
if has three real solutions , and has three real solutions for each of the solutions for @xmath9 , there can be nine such extrema . out of these nine extrema , three have @xmath32 and thus can not be the ew vacuum . the two solutions for nonzero @xmath8 for a given solution of @xmath9 are symmetrically placed about @xmath32 .
] . for @xmath32
, can be simplified to v_s ( - 2m^2 + 3b_3 v_s + 4b_4 v_s^2 ) = 0 .
[ pot - min3-vs - v0 ] for the other two nonzero extrema in the @xmath6-direction , one uses @xmath33 from and get v_s ( v_s^2 + 3 b_3 v_s + ) = 0 .
[ pot - min3-vs ] given the parameters of the potential , eq .
( [ pot - min3-vs ] ) gives the condition for real non - zero solutions for @xmath9 along @xmath34 .
note that for a @xmath1-symmetric potential ( @xmath35 ) , the stability criteria ensure that @xmath36 .
the condition that any extremum of @xmath37 is a minimum , and not a maximum or saddle point , is > 0 , > 0 .
[ cond - minima ] to get the mixing between the cp - even component of the sm higgs doublet @xmath7 and the real scalar @xmath0 , we expand the potential in about the vevs of the fields .
the terms quadratic in fields are given by v(,s ) & & ^2 + s^2 + 2s + & & & s m s , where = 12 ( -^2 + 3 v^2 + a_1 v_s + a_2v_s^2 ) & + a_2 vv_s + a_2 vv_s & -m^2 + + 3 b_3 v_s + 6 b_4 v_s^2 , [ massmat ] in its most general form .
it should be noted that the mixing between @xmath7 and @xmath0 also depends on the parameter @xmath23 , which does not appear in the minimization conditions in and , because of the choice in .
if @xmath38 are the eigenvalues of @xmath39 , the masses of the physical states are given by m_h = , m_s = , [ mass ] and the mixing angle @xmath12 is the angle which parametrize the @xmath40 rotation matrix that diagonalises @xmath39 .
one instructive case is when one of the minima has either @xmath32 or @xmath11 or both .
the point @xmath41 is an extremum where = -^2 , = -2 m^2 , = 0 , [ extreme0 ] and will be a minimum if @xmath42 i.e. the mass terms are right - sign . from one
can see for this case that @xmath8 will also have two non - zero real roots if @xmath43 , which requires a large negative @xmath19 , resulting in a large singlet - doublet mixing .
if there is a minimum for @xmath32 , @xmath45 , and the condition for minimum translates to > 0 a_2 v_s^2 -^2 > 0 , > 0 8 b_4 v_s^2 + 3 b_3 v_s > 0 .
the first condition shows , from , that there is only one real solution for @xmath8 at @xmath32 .
as @xmath46 from stability criterion , the second condition yields interesting bounds . if @xmath47 and @xmath48 or @xmath49 and @xmath50 , @xmath51 is definitely positive . on the other hand , if @xmath52 but @xmath47 , there is a lower bound on @xmath9 , v_s > . similarly ,
if @xmath53 and @xmath48 , there is an upper bound on @xmath9 : v_s < - . for @xmath54 and @xmath11 ,
the minima along @xmath8 are symmetric , at @xmath55 .
this means @xmath56 for @xmath8 to be real and hence @xmath57 .
so the condition for minimum at @xmath34 , @xmath11 reduces to @xmath58 . the constraints for @xmath34 , @xmath44 are obtained using the concavity condition in , along with the expression for the second derivatives , to be = 2(^2 - a_2 v_s^2 ) = 2 v^2 , = 3 b_3 v_s + 8 b_4 v_s^2 , = 2 a_2 v v_s .
for the @xmath1-symmetric potential , @xmath35 , and hence the condition for a minimum at @xmath34,@xmath61 simplifies to @xmath62 , which is nothing but one of the stability criteria for the potential as shown in .
further simplifications occur if we have only one minimum of the potential . in this case
, it has to be at @xmath63 gev , and hence solutions for @xmath64 can be obtained in a straightforward way from : v_s = .
[ vs - sol2 ] thus , for the @xmath1-symmetric case , the condition for a minimum at @xmath61 is @xmath65 , as @xmath46 from stability criterion . for nonzero @xmath25
, the condition is a_2 v^2 - 2 m^2 < 9 b_3 ^ 2 / 16 b_4 .
one can easily have more than one minima with @xmath54 and @xmath66 .
however , if one minimum is at the origin , the second minimum at nonzero @xmath8 and @xmath9 requires @xmath19 to be large and negative , as discussed in section [ v0-vs0 ] .
such large values of @xmath19 are under severe kosh from the lhc data .
a detailed discussion on the nature of the scalar potential related to the electroweak phase transition can be found in ref .
. if there is no mixing between @xmath7 and @xmath0 , there are no constraints on @xmath0 coming from the lhc data , except that it can not be so light ( @xmath67 ) that @xmath68 is allowed and the branching ratio is more than @xmath69 at 95% cl @xcite .
similarly , there is no constraint from electroweak precision observables as @xmath0 does not have any gauge coupling .
if there is a mixing between @xmath7 and @xmath0 parametrized by an angle @xmath12 , the number of events , which is just the production cross - section times the branching ratio , goes down by @xmath70 . denoting the production cross - section times the decay width scaled to that in the sm by @xmath71 ,
the atlas and cms combined result shows = 1.09^+0.11_-0.10 , from which we can put a limit of @xmath72 at @xmath73 , which we will use for our subsequent discussion .
this helps us to avoid both lhc data and precision constraints at one stroke
. only the @xmath74 parameter may be significant in the small mixing case .
in this model , the @xmath74 parameter is given by @xcite t^sm+s & = & - ( ) \ { ^2 .
+ ^2 } , where @xmath75 gev ) and @xmath76 are the two mass eigenstates , and @xmath12 is the mixing angle .
the sm expression for @xmath74 can be found by putting @xmath77 and @xmath78 .
the quantity constrained by the electroweak fit , @xmath79 , is given by @xcite t = t^sm+s - t^sm = 0.01 0.12 .
@xmath74 is related with the @xmath80-parameter by @xmath81 . for small
mixing ( @xmath72 ) , the constraints coming from the oblique parameters are not significant .
we would also like to see how the couplings evolve with energy .
the one - loop @xmath82-functions are @xcite 16 ^ 2 _ & = & 12 ^ 2 + 6g_t^2 + a_2 ^ 2 -32(g_1 ^ 2 + 3g_2 ^ 2 ) - 3 g_t^4 + ( g_1 ^ 4 + 2 g_1 ^ 2 g_2 ^ 2 + 3 g_2 ^ 4 ) , + 16 ^ 2 _ b_4 & = & 36 b_4 ^ 2 + a_2 ^ 2 , + 16 ^ 2 _ a_2 & = & a_2 , + 16 ^ 2 _ g_t & = & g_t [ all - rge ] where @xmath83 , and @xmath84 .
the @xmath82-functions for all gauge couplings are identical to that of the sm . for simplicity ,
we have put all the sm yukawa couplings equal to zero except for that of the top quark .
this hardly changes our conclusions . for the @xmath1-asymmetric case
, the trilinear couplings @xmath23 and @xmath25 also evolve : 16 ^ 2 _ a_1 & = & a_1(9 + 4 a_2 ) + 6 a_2 b_3 , + 16 ^ 2 _ b_3 & = & 2 a_1 a_2 + 36 b_3 b_4 .
[ z2-odd - rge ] note that @xmath85 is always positive and hence can only increase , starting from a positive value .
@xmath86 also gets a positive contribution on top of the sm ones .
these make the couplings blow up at a much lower scale than the planck scale ( @xmath87 gev ) , unless one starts with very small values of @xmath21 .
similarly , @xmath88 is proportional to @xmath19 itself and can lead to a blow - up for large @xmath19 . for our analysis
, we have taken the initial values of the couplings at the electroweak scale in such a way that the higgs boson mass is correctly reproduced as @xmath89 $ ] gev .
the threshold effects are taken at the singlet mass scale , however it has been seen that the final results are not very sensitive on the exact choice of this scale , and moreover , uncertainties coming from possible higher - loop contributions are larger compared to the uncertainties coming from the threshold corrections .
the constraints on the parameter space are all obtained with the one - loop improved values of the couplings .
if there are two minima of the potential and the ew vacuum is shallower , the universe can tunnel down to the deeper vacuum .
in such cases , the parameters must be chosen such that the lifetime of the shallower vacuum should be at least as large as the lifetime of the universe , which is about 13.7 billion years . and @xmath90 control the tunnelling lifetime .
if there are more than one fields , one gets a multi - dimensional contour .
( b ) right : example of a two - dimensional contour for sm+s .
the line shows the shortest path joining the two minima along which the tunnelling probability should be calculated .
notice that the path does not pass through the local maximum .
contour values denote the potential at that point in the unit @xmath91 gev@xmath92.,title="fig:",width=264,height=219 ] and @xmath90 control the tunnelling lifetime .
if there are more than one fields , one gets a multi - dimensional contour .
( b ) right : example of a two - dimensional contour for sm+s .
the line shows the shortest path joining the two minima along which the tunnelling probability should be calculated .
notice that the path does not pass through the local maximum .
contour values denote the potential at that point in the unit @xmath91 gev@xmath92.,title="fig:",width=264,height=211 ] to calculate the lifetime of the metastable state @xcite , let us assume that the potential at the ew vacuum is zero ( this can always be achieved by a constant shift ) , the true vacuum has a depth @xmath93 ( @xmath94 ) , and the height of the barrier with respect to the sm vacuum is @xmath90 , as shown in .
the decay width density of the universe , @xmath95 , is given by = a ( -b ) , where @xmath96 is a small pre - factor , and @xmath97 is estimated in the thin - wall approximation as @xcite b = ( ) ^3 .
[ b - def ] this expression is true for a single field whose self - quartic coupling is given by @xmath98 . with more than one fields ,
there are several dimensionless couplings and the tunnelling path may not even go through under the hill .
however , for sm+s , what we have done numerically is to find the two minima and then calculate the tunnelling along the straight line joining them , as shown in .
this is effectively a single - field approximation , where the field is a combination of @xmath99 and @xmath0 .
the height @xmath90 is taken to be the maximum height above zero - level _ along this path _ and not the local maximum of the field space .
the ratio @xmath100 should be greater than @xmath101 to make the shallower vacuum stable with respect to the lifetime of the universe .
keeping in mind of the necessary simplifications , we have chosen only those models for which this ratio is more than unity , and therefore the stability is assured .
the one - loop effective potential in the sm is given by v_1(_c ) = _ i n_i m_i(_c)^4 ( - c_i ) , where @xmath102 runs over @xmath103 and @xmath104 , and n_i = 1,3,6,3,-12 , c_i = 32,32,56,56,32 , for @xmath105 and @xmath104 respectively .
the masses are field - dependent and depend on the classical minimum @xmath106 .
@xmath107 is the arbitrary regularization scale . to be precise , the radiative corrections , being suppressed by the loop factor , are significant only along a flat or near - flat direction in the field space .
if all the dimensionful parameters are zero , the theory becomes scale invariant .
the minimization conditions in this case , namely , @xmath108 and @xmath109 , yield @xmath110 as the consistency condition if neither @xmath8 nor @xmath9 vanishes .
the last condition makes the determinant of the mass matrix equal to zero , ensuring a massless mode and hence a flat direction in the field space .
if dimensionful couplings are present , there is in general no flat direction in the @xmath7-@xmath0 plane , and one expects the radiative corrections to the potential to have a small effect .
note that @xmath111 is the limiting case of the stability condition .
however , if there is a strong hierarchy between the two vevs , the direction along the smaller - vev field is almost flat .
the choice of the regularization scale may change @xmath9 significantly , and if there is a significant singlet - doublet mixing , both the physical scalar masses may get affected .
we will see the numerical estimates later .
this also shows that putting @xmath112 in does not lead to a cdm candidate , because such a fine - tuned relationship is not stable under radiative corrections . with two scalar fields ,
the form of the one - loop corrected potential is v(_c,_c ) = v_0(_c,_c ) + v_1(_c,_c ) , where @xmath106 and @xmath113 are the classical minima along @xmath7 and @xmath0 respectively , and @xmath114 is the tree - level potential written in terms of the classical minima .
@xmath115 is the one - loop correction , given by v_1(_c,_c ) = v_1(_c ) + m_s^4(_c,_c ) .
one should note that in case of nonzero mixing , the field - dependent mass of @xmath99 in @xmath116 is a function of @xmath113 too .
thus , @xmath115 includes all the sm fields and the singlet .
the regularization scale @xmath107 is arbitrary , but one can choose it in such a way that the one - loop corrected minimum for @xmath6 remains unchanged , in other words , @xmath106 falls at @xmath8 .
this keeps all the sm fermion and gauge boson masses invariant , and also keeps the goldstone bosons massless ; thus , we do not need to consider the goldstone boson contributions for the effective potential . a detailed discussion of the procedure is given in ref .
@xcite for the two - higgs doublet model potential .
let us first go through the well - studied case of the @xmath1-symmetric potential ( @xmath119 in ) , for the sake of completeness of this study .
one can always get a minimum at @xmath11 by applying the shift symmetry @xmath120
. this minimum should better be the only one , or at least the global one ( a local one with lifetime more than the age of the universe will also do ) if we want to have a cdm candidate in @xmath0 , with mass of @xmath121 .
the points with @xmath122 need a very large and negative @xmath19 and are ruled out by the cdm spin - independent scattering cross - section limits .
thus , all such higgs portal dark matter models must have a right - sign mass term ( @xmath123 ) for @xmath0 .
this is true even for the narrow region of higgs resonance , at about @xmath124 .
( left ) and @xmath21 ( right ) , in the @xmath1-symmetric case with the minimum at @xmath11.,title="fig:",width=321 ] ( left ) and @xmath21 ( right ) , in the @xmath1-symmetric case with the minimum at @xmath11.,title="fig:",width=321 ] the only way for @xmath0 to interact with the sm sector is through the term @xmath125 in , since @xmath126 in this case .
the spin - independent cdm - nucleon scattering cross - section is given , in this scenario , by @xcite = , where @xmath127 and @xmath128 are the masses of the nucleon , @xmath0 , and higgs respectively .
the matrix element for scattering is given by @xmath129 , whose value is approximately @xmath130 .
if @xmath19 is very small , the cdm detection cross - section becomes small , but the annihilation rate goes down too , leaving more dark matter in the universe than is allowed , and thus leading to overclosure . if @xmath19 is large , the scattering cross - section of cdm with nucleons is also large and hence will be severely constrained by direct detection experiments , in particular lux , which gives the best limits now @xcite .
thus , apart from the narrow higgs resonance region , only a small wedge for the dark matter mass @xmath131 gev is still allowed , and we focus only on those models that provide @xmath132 in this range . in , we show the allowed regions as a function of @xmath19 , every dot corresponds to a particular choice of parameters .
for all these models @xmath123 , which , by our definition of the potential , means a right - sign mass term for the singlet and no symmetry breaking in the @xmath0-direction . technically , @xmath122 can also lead to a local minimum at @xmath133 , but if we take @xmath19 to be in the perturbative region , such models lead to low @xmath132 and are hence ruled out by the direct detection data . depending on the values of @xmath19 and @xmath21
, one can also check how far in the energy scale the singlet dm model remains valid . for this , we use the one - loop renormalization group ( rg ) equations and see where the couplings become non - perturbative and ultimately hit the landau pole , or the potential becomes unstable .
our results are shown in for the two relevant parameters @xmath19 and @xmath21 .
note that there is an upper limit on @xmath134 above which the model ceases to be valid even before 50 tev . while there is no such limit for @xmath19 ,
low-@xmath19 models have a smaller range of validity compared to medium-@xmath19 ( @xmath135 ) models , where the validity can be as high as @xmath136 gev .
while the allowed region for each model depends on the exact values of the parameters chosen , some intuitive insights can be put forward . as there is no mixing
, @xmath98 must start from its sm value @xmath137 .
the only modification to its @xmath82-function comes from the @xmath138 term , so the range of validity increases with increasing @xmath19 , provided @xmath21 is sufficiently small to start with and does not hit its landau pole earlier ( @xmath21 starts from a positive value , and always increases ) .
so , for the low-@xmath19 regions , it is the vacuum stability that mostly controls the allowed range .
after @xmath19 passes a certain value , and/or @xmath21 becomes large , the range is controlled by the blowing up of one or more of the couplings . as we have just shown , @xmath139 is already ruled out for this model .
a similar study on the parameter space of higgs portal dark matter models was performed recently in ref .
there is always a chance that with improved measurements , the wedge region may go away .
in the higgs resonance region , the allowed values of @xmath19 are much smaller , and from , we see that such models cease to be valid at about @xmath140 gev .
this model , with addition of vectorlike fermions , may explain the recently observed resonance at 750 gev .
however , introduction of such fermions spoils the possibility of a higgs portal dark matter , as the scalar decays through the fermion loops . on the other hand
, there is a new constraint on the singlet mass , which narrows down the parameter space even further .
the renormalization group equations also change , with new yukawa couplings introduced , and may affect the stability of the potential .
we will not discuss this extension any further here .
-symmetric case with @xmath66.,title="fig:",width=321 ] -symmetric case with @xmath66.,title="fig:",width=321 ] and @xmath98 in the @xmath1-symmetric sm+s models with @xmath61 .
the plot for @xmath21 is similar to that for @xmath98 .
large values of either @xmath21 or @xmath98 limit the range of validity of the models , because of the nature of the rg equations .
@xmath19 plays a subdominant role here.,title="fig:",width=321 ] and @xmath98 in the @xmath1-symmetric sm+s models with @xmath61 .
the plot for @xmath21 is similar to that for @xmath98 .
large values of either @xmath21 or @xmath98 limit the range of validity of the models , because of the nature of the rg equations .
@xmath19 plays a subdominant role here.,title="fig:",width=321 ] if the singlet field @xmath0 develops a nonzero vev , _
i.e. _ @xmath44 , the physical fields @xmath99 and @xmath141 become orthogonal combinations of @xmath7 and @xmath0 , with the mixing angle constrained by the lhc data to be @xmath142 . hence there is no cdm candidate in this model .
the allowed parameter space is shown in for two distinct cases : @xmath123 ( right - sign mass term for the singlet ) and @xmath122 ( wrong - sign mass term ) .
we note the following characteristics : * for @xmath123 , only negative values of @xmath19 are allowed .
this follows from the condition @xmath143 .
both positive and negative values of @xmath19 are allowed for @xmath122 .
* there is a correlation between @xmath19 and @xmath98 for @xmath123 .
this follows from the @xmath144 dependence of @xmath145 in .
* only small positive values of @xmath146 are allowed for @xmath123 .
this can be understood from the constraint @xmath147 coming from the stability criteria , as mentioned in sec .
[ model ] .
it can be shown easily that for @xmath123 , @xmath19 and @xmath146 have to be of opposite signs to have real @xmath9 , and the only possibility to have a real @xmath8 as well is to have @xmath148 and @xmath149 .
the magnitudes of @xmath19 , @xmath146 and @xmath20 are , however , restricted by the higgs mass and the mixing angle @xmath12 . for @xmath123 , there is no such constraint and @xmath150 can be large .
the range of validity for the models is shown in .
one can see that large values of @xmath19 or @xmath98 necessarily mean a smaller range of validity , which follows from the nature of the rg equations .
the coupling @xmath21 always increases , so one has to start with a sufficiently small value of @xmath21 not to hit the landau pole .
the other two couplings , @xmath98 and @xmath19 , can be controlled by the negative contributions coming from yukawa or gauge couplings if they are sufficiently small to start with . too small a value means instability setting in at a low scale , and too large a value means a quick blowing up of the couplings .
thus , an intermediate range , @xmath151 is the region for maximum validity .
we have explicitly checked that the mixing angle is always well within the lhc limit .
one - loop corrections do not change the nature of the potential qualitatively , except changing the depth of the potential .
however , the composition of the cp - even neutral scalars change with the one - loop corrections , because the ratio of @xmath106 and @xmath113 , and hence the mixing angle @xmath12 of the mass matrix changes from its tree - level value . and @xmath19 , where the ew vacuum is the `` global '' ( red ) or the `` local '' ( blue ) minimum in @xmath1-asymmetric sm+s models , to have the validity at least upto 50 tev.,title="fig:",width=321 ] and @xmath19 , where the ew vacuum is the `` global '' ( red ) or the `` local '' ( blue ) minimum in @xmath1-asymmetric sm+s models , to have the validity at least upto 50 tev.,title="fig:",width=321 ] in this section , we focus on only those models that allow two non - zero minima for the potential , as discussed in section 2.2 . from we can see that for @xmath152 the two minima have unequal depths .
we demand one of them to be the ew vacuum with @xmath63 gev .
if this is the deeper minimum , the universe is stable ; if this is the shallower one , the universe can decay to the true vacuum and in that case the lifetime must at least be equal to the age of the universe . first , let us focus on the tree - level potential . in
, we show the range of validity of these models for various choices of @xmath19 and @xmath98 .
the trend is similar to what we have seen before in section 4.1 : large values of @xmath98 , @xmath19 , or even @xmath21 make the couplings blow up at a relatively low scale .
note that @xmath139 ( @xmath153 ) models tend to have a local ( global ) minimum at @xmath154 gev , but there are exceptions .
the range of validity of these models is almost the same as that of the @xmath1-symmetric case , because the rg evolutions of the couplings are controlled by the dimensionless couplings .
the allowed region for @xmath19 , however , is bunched more towards @xmath155 .
( left ) and @xmath98 ( right).,title="fig:",width=321 ] ( left ) and @xmath98 ( right).,title="fig:",width=321 ] for models with a shallower minimum at the ew vacuum , one may also estimate the lifetime of the universe .
this is bound to be a rough estimate as the path between the two minima need not pass through a local maximum or even a saddle point .
approximately , @xmath156 ( see ) leads to a metastable vacuum @xcite while @xmath157 tends to make the universe unstable .
assuming the maximum of the quartic couplings to be of the order of unity , this results in an approximate bound of @xmath158 .
in we show how the tunnelling lifetime depends on the parameter @xmath25 as a function of the ratio @xmath100 and @xmath98 as a function of @xmath97 .
the dependence of @xmath21 is similar to that of @xmath98 .
shows that the controlling parameter is @xmath25 because that creates the depth difference between the two minima .
the smaller @xmath25 is , the larger is the lifetime of the metastable minimum . as a conservative estimate , we have taken the stability limit to be @xmath159 , which excludes the low @xmath100 and hence low @xmath97 values .
such excluded models are also shown in the right hand side plot of .
the distribution of models does not depend much on the quartic couplings @xmath21 or @xmath98 .
( the classical minimum for the doublet - dominated field ) and @xmath160 ( the classical minimum for the singlet - dominated field ) for the two models described in the text .
the upper panel is for model 1 and the lower panel is for model 2 .
the profiles are drawn along the line joining the two minima .
note that the second minimum can change significantly and the shift depends on the choice of the regularization scale @xmath107 .
, title="fig:",width=321 ] ( the classical minimum for the doublet - dominated field ) and @xmath160 ( the classical minimum for the singlet - dominated field ) for the two models described in the text .
the upper panel is for model 1 and the lower panel is for model 2 .
the profiles are drawn along the line joining the two minima .
note that the second minimum can change significantly and the shift depends on the choice of the regularization scale @xmath107 .
, title="fig:",width=321 ] + ( the classical minimum for the doublet - dominated field ) and @xmath160 ( the classical minimum for the singlet - dominated field ) for the two models described in the text .
the upper panel is for model 1 and the lower panel is for model 2 .
the profiles are drawn along the line joining the two minima .
note that the second minimum can change significantly and the shift depends on the choice of the regularization scale @xmath107 .
, title="fig:",width=321 ] ( the classical minimum for the doublet - dominated field ) and @xmath160 ( the classical minimum for the singlet - dominated field ) for the two models described in the text .
the upper panel is for model 1 and the lower panel is for model 2 .
the profiles are drawn along the line joining the two minima .
note that the second minimum can change significantly and the shift depends on the choice of the regularization scale @xmath107 .
, title="fig:",width=321 ] the effect of one - loop corrections on the tree - level potential is not very drastic .
this is expected because both vevs are nonzero and neither @xmath6 nor @xmath0 direction is a flat one .
the one - loop corrections , being perturbative in nature , are suppressed by the standard loop factor of @xmath161 and are expected to be significant only if we look at a flat direction .
we show the one - loop corrections for two models , the parameters at the electroweak scale are given in .
.parameter values for the models for which the 1-loop corrections are shown in the .
[ cols="<,^,^",options="header " , ] we choose the regularization scale @xmath107 in such a way that even after the one - loop corrections , the ew vacuum stays at @xmath162 gev ( so that all sm particles have their masses unaffected ) .
this choice of regularization scale was motivated in ref .
we show the effect of one - loop corrections in for these two models .
while they both have the global minimum at @xmath163 gev , the required regularization scales differ by more than one order of magnitude .
the nature of change is similar for all models , and thus we do not expect an unstable vacuum model to become metastable ( or vice versa ) because of the one - loop corrections .
however , one may note how much the global minimum has been lowered by the one - loop corrections for the second model .
in fact , for all the models scanned , we have never found a switch from global to local minimum induced by the radiative corrections .
the potential of sm+s , a real singlet enhanced sm , shows several interesting features . in this paper , we have investigated the parameter space for several types of sm+s : the higgs portal dark matter models , the potential with an explicit @xmath1-symmetry and having singlet - doublet mixing , the @xmath1-asymmetric potential with a stable ew minimum , or the same with an unstable or metastable ew minimum .
the general features can be summarized as follows .
* adding one more real singlet makes the potential less stable in general at a high energy .
this happens because the renormalization group equations for the couplings tend to hit the landau pole much below the planck scale , more so if the starting values at the ew scale is large .
all the quartic couplings , namely , @xmath98 , @xmath21 , and @xmath19 , have to be small at the ew scale to keep the model valid up to a high scale , as the @xmath82-functions are coupled .
( however , such new scalar couplings are helpful to avoid the vacuum stability bound , coming from the negative pull caused by the large top yukawa coupling . again , such conclusions are not valid if there are more degrees of freedom , like vectorlike fermions . )
our numerical results are shown at one - loop but inclusion of higher - order corrections do not change the result qualitatively . in particular , if we start with a large value of either @xmath98 or @xmath21 , the model hits the landau pole at a relatively low scale .
this can be taken as a possible indication of some new physics taking over and one should consider the effect of higher - dimensional operators on the low - scale physics . *
while there are some minor variations for the allowed range of parameters among different class of models ( _ e.g. _ @xmath19 ) , the overlap is significant , and so one has to determine all the couplings experimentally to know what class of sm+s it really is .
this is , of course , an extremely challenging task , if not outright impossible ; the determination of @xmath21 is apparently beyond the reach of any present or upcoming colliders unless there is a significant mixing between the singlet and the doublet and the higgs self - couplings are determined with sufficient accuracy . for the prospect of determination of the singlet - doublet mixing angle in future colliders , we refer the reader to ref . @xcite . *
the tree - level results are robust enough as far as the metastability issue is concerned .
this is expected as we are not looking along any flat direction .
however , the one - loop corrections can change the position of the minima .
we thank filippo sala for pointing out a mistake in the first version of the paper . s.g .
acknowledges the university grants commission , government of india , for a research fellowship . a.k . acknowledges department of science and technology , government of india , and council for scientific and industrial research , government of india , for extramural projects .
s.r . likes to thank the department of science and technology ( dst ) , government of india for the insa - inspire faculty fellowship .
s.r . would also like to thank prof .
yuval grossman , lepp , cornell university for hosting the visit as a part of insa - inspire faculty fellowship , where a part of this work has been done .
v. silveira and a. zee , phys .
b * 161 * , 136 ( 1985 ) ; + c. p. burgess , m. pospelov and t. ter veldhuis , nucl .
b * 619 * , 709 ( 2001 ) [ arxiv : hep - ph/0011335 ] ; + j. mcdonald , phys .
d * 50 * , 3637 ( 1994 ) [ arxiv : hep - ph/0702143 ] ; + j. m. cline , k. kainulainen , p. scott and c. weniger , phys .
d * 88 * , 055025 ( 2013 ) [ phys .
d * 92 * , 039906 ( 2015 ) ] [ arxiv:1306.4710 [ hep - ph ] ] . j. r. espinosa , t. konstandin and f. riva , nucl .
b * 854 * , 592 ( 2012 ) [ arxiv:1107.5441 [ hep - ph ] ] . s. j. huber and m. g. schmidt , nucl .
b * 606 * , 183 ( 2001 ) [ hep - ph/0003122 ] ; + s. w. ham , y. s. jeong and s. k. oh , j. phys .
g * 31 * , 857 ( 2005 ) [ hep - ph/0411352 ] .
a. kundu and s. raychaudhuri , phys .
d * 53 * , 4042 ( 1996 ) [ arxiv : hep - ph/9410291 ] ; + a. drozd , b. grzadkowski and j. wudka , jhep * 1204 * , 006 ( 2012 ) [ arxiv:1112.2582 [ hep - ph ] ] ; + a. drozd , arxiv:1202.0195 [ hep - ph ] ; + i. chakraborty and a. kundu , phys .
d * 87 * , 055015 ( 2013 ) [ arxiv:1212.0394 [ hep - ph ] ] ; + f. bazzocchi and m. fabbrichesi , phys .
d * 87 * , 036001 ( 2013 ) [ arxiv:1212.5065 [ hep - ph ] ] . v. barger , p. langacker , m. mccaskey , m. j. ramsey - musolf and g. shaughnessy , phys .
d * 77 * , 035005 ( 2008 ) [ arxiv:0706.4311 [ hep - ph ] ] . v. barger , p. langacker , m. mccaskey , m. ramsey - musolf and g. shaughnessy , phys . rev .
d * 79 * , 015018 ( 2009 ) [ arxiv:0811.0393 [ hep - ph ] ] .
g. m. pruna and t. robens , phys .
d * 88 * , 115012 ( 2013 ) [ arxiv:1303.1150 [ hep - ph ] ] ; + t. robens and t. stefaniak , eur .
j. c * 75 * , 104 ( 2015 ) [ arxiv:1501.02234 [ hep - ph ] ] . c. y. chen , s. dawson and i. m. lewis , phys .
d * 91 * , 035015 ( 2015 ) [ arxiv:1410.5488 [ hep - ph ] ] .
a. datta , a. raychaudhuri , s. raychaudhuri and s. chakrabarti , z. phys .
c * 72 * , 449 ( 1996 ) [ hep - ph/9510432 ] ; + a. datta and a. raychaudhuri , phys . rev .
d * 57 * , 2940 ( 1998 ) [ hep - ph/9708444 ] ; + r. schabinger and j. d. wells , phys .
d * 72 * , 093007 ( 2005 ) [ hep - ph/0509209 ] ; + d. oconnell , m. j. ramsey - musolf and m. b. wise , phys .
d * 75 * , 037701 ( 2007 ) [ hep - ph/0611014 ] ; + v. barger , p. langacker and g. shaughnessy , phys .
d * 75 * , 055013 ( 2007 ) [ hep - ph/0611239 ] ; + o. bahat - treidel , y. grossman and y. rozen , jhep * 0705 * , 022 ( 2007 ) [ hep - ph/0611162 ] .
+ d. buttazzo , f. sala and a. tesi , jhep * 1511 * , 158 ( 2015 ) [ arxiv:1505.05488 [ hep - ph ] ] .
+ f. bojarski , g. chalons , d. lopez - val and t. robens , arxiv:1511.08120 [ hep - ph ] .
s. i. godunov , a. n. rozanov , m. i. vysotsky and e. v. zhemchugov , arxiv:1503.01618 [ hep - ph ] .
atlas collaboration , atlas - conf-2015 - 081 .
cms collaboration , cms - pas - exo-15 - 004 .
s. khan , s. goswami and s. roy , phys .
d * 89 * , 073021 ( 2014 ) [ arxiv:1212.3694 [ hep - ph ] ] ; + n. khan and s. rakshit , phys . rev .
d * 90 * , 113008 ( 2014 ) [ arxiv:1407.6015 [ hep - ph ] ] ; + g. chalons and f. domingo , phys .
d * 86 * , 115024 ( 2012 ) [ arxiv:1209.6235 [ hep - ph ] ] ; + n. haba , h. ishida , k. kaneta and r. takahashi , phys .
d * 90 * , 036006 ( 2014 ) [ arxiv:1406.0158 [ hep - ph ] ] .
d. lpez - val and t. robens , phys . rev .
d * 90 * , 114018 ( 2014 ) [ arxiv:1406.1043 [ hep - ph ] ] .
see the review on electroweak model in k. a. olive _ et al . _
[ particle data group collaboration ] , chin .
c * 38 * , 090001 ( 2014 ) . i. chakraborty and a. kundu , phys .
d * 87 * , no . 5 , 055015 ( 2013 ) [ arxiv:1212.0394 [ hep - ph ] ] . s. r. coleman , phys .
d * 15 * , 2929 ( 1977 ) [ phys .
d * 16 * , 1248 ( 1977 ) ] ; + v. a. rubakov , classical theory of gauge fields , princeton , usa : univ .
( 2002 ) . l. feng , s. profumo and l. ubaldi , jhep * 1503 * , 045 ( 2015 ) [ arxiv:1412.1105 [ hep - ph ] ] . m. duerr , p. fileviez perez and j. smirnov , arxiv:1508.04418 [ hep - ph ] .
m. ackermann _ et al .
_ [ fermi - lat collaboration ] , phys . rev .
d * 88 * , 082002 ( 2013 ) [ arxiv:1305.5597 [ astro-ph.he ] ] .
d. s. akerib _ et al . _
[ lux collaboration ] , phys .
lett . * 112 * , 091303 ( 2014 ) [ arxiv:1310.8214 [ astro-ph.co ] ] .
h. han and s. zheng , jhep * 1512 * , 044 ( 2015 ) [ arxiv:1509.01765 [ hep - ph ] ] .
s. profumo , m. j. ramsey - musolf , c. l. wainwright and p. winslow , phys .
d * 91 * , no . 3 , 035018 ( 2015 ) [ arxiv:1407.5342 [ hep - ph ] ] . | we investigate the parameter space of the standard model enhanced by a gauge singlet real scalar @xmath0 .
taking into account all the theoretical and experimental constraints , we show the allowed parameter space for two different types of such singlet - enhanced standard model . for the first case , the scalar potential has an explicit @xmath1-symmetry , and may lead to a dark matter candidate under certain conditions . for the second case , the scalar potential does not respect any @xmath1 .
this is again divided into two subcategories : one where the standard model vacuum is stable , and one where it is unstable and can decay into a deeper minimum .
we show how the parameters in the scalar potential control the range of validity of all these models .
finally , we show the effect of one - loop correction on the positions and depths of the minima of the potential .
* potential of a singlet scalar enhanced standard model * * swagata ghosh*,@xmath2 * anirban kundu*,@xmath2 and * shamayita ray * @xmath3 @xmath4 _ department of physics , university of calcutta , + 92 acharya prafulla chandra road , kolkata 700009 , india _
@xmath5 _ laboratory for elementary particle physics , cornell university , ithaca , ny 14853 , usa _ |
the advancement of fabrication and lithography techniques of semi - conductors have made it possible to investigate bi - layer systems made of two electronic two - dimensional layers at very close proximity ( several hundred angstroms ) in a very controlled manner . among the controllable parameters are the tunneling of electrons between the layers , the density of electrons at each layer , and external parameters such as the magnetic field and temperature .
moreover , it is also possible to make separate electrical contacts to the two layers .
experiments carried out on such systems have revealed a wealth of interesting phenomena , and have also stimulated an active theoretical study @xcite .
one of the measurements performed on systems of two electronic layers at close proximity is that of the transresistivity .
for that measurement a current @xmath10 is driven in one layer , while the current in the other layer , @xmath11 , is kept zero . due to the interactions between electrons at different layers ,
momentum is transfered from the current carrying layer to the second one . as a consequence of that momentum transfer
, one needs to apply a voltage on the second layer , @xmath12 , to keep @xmath11 zero .
the ratio @xmath13 is defined as the transresistivity , @xmath2 .
the transresistivity was measured experimentally in @xcite and studied theoretically at @xcite .
transresistivity in a magnetic field was studied in @xcite .
it was found that although @xmath2 is a @xmath14 measurement , it reflects the response of the two layers to driving forces of finite wave vector @xmath15 and finite frequency @xmath16 . in the limit of weak coupling between the layers ,
the resistivity was found to be mostly due to the coulomb interaction between electrons at the two layers , and to be given by the following expression , _ valid both in the absence and presence of a magnetic field _
@xcite , ( 3.375,0 ) ( 0,0)(1,0)3.375 ( 3.375,0)(0,1)0.08 @xmath17 ( 3.375,0 ) ( 0,0)(1,0)3.375 ( 0,0)(0,-1)0.08 2 where @xmath18 is the screened inter - layer coulomb interaction , @xmath19 is the single - layer density - density response function of the @xmath20th layer , @xmath21 is the average density of electrons at layer @xmath20 and @xmath6 is the temperature .
both @xmath18 and @xmath22 are defined precisely in the next section , where eq .
( [ donald ] ) is studied in detail . there
we also comment on the validity of eq .
( [ donald ] ) in the presence of a magnetic field . in this paper
we consider the transresistivity , in the limit of weak inter - layer coupling , of two systems of two dimensional electron gases ( 2deg ) in a strong magnetic field , where the landau level filling fractions of the two layers , @xmath23 and @xmath24 , are smaller than one and the electrons form a compressible state .
the most prominent example is that of @xmath23 and @xmath24 being equal to , or close to , @xmath25 . in section [ threehalf ]
we consider the @xmath26 case .
our work is motivated by several experimental and theoretical works that discovered a unique response of the @xmath1 state to finite @xmath27 driving forces .
a preliminary experimental study of coulomb drag in the regime of a partially filled landau level was recently carried out by eisenstein _
_ @xcite .
most theoretical studies of the single layer @xmath1 state have used the composite fermions approach , as introduced in @xcite , to calculate electronic quantities , such as the electronic linear response functions . in our study of the transresistivity
we use the composite fermion approach to obtain information regarding the electronic single layer density - density response function , _ but discuss the coupling between the layers in purely electronic terms_. this approach makes the physical picture underlying the transresistivity quite transparent . in particular , it allows us to point out that the unique features of the transresistivity in the case we study are closely related to the unique @xmath28-dependence of the electronic longitudinal conductivity , as observed in surface acoustic waves measurements .
both phenomena reflect the slow relaxation of density fluctuations in the @xmath1 state .
( 1,0.618 ) ( 0,0 ) ( 1.02,0.02)(0,0)[lb]@xmath29 $ ] ( 0,0.648)(0,0)[bl ] @xmath30 $ ] ( -0.005 , 0.206)(0,0)[r]@xmath31 ( -0.005 , 0.412)(0,0)[r]@xmath32 ( -0.005 , 0.618)(0,0)[r]@xmath33 ( 0.5 , -0.01)(0,0)[t]@xmath34 ( 1 , -0.01)(0,0)[t]@xmath35 as discussed in detail below , based on the use of eq .
( [ donald ] ) , with the @xmath22 and @xmath18 appropriate for the case at hand , we find , 1 . for @xmath36 , the leading order temperature dependence of the transresistivity is @xmath37 with @xmath38 in eq .
( [ 4:t0 ] ) @xmath39 for @xmath1 and @xmath40 where @xmath41 is the thermodynamical compressibility of the @xmath1 state and @xmath42 is the dielectric constant .
essentially , @xmath43 is the ratio of the thomas - fermi screening length of the @xmath1 state to the separation between the layers , @xmath44 . for realistic numbers , @xmath43 is small compared to one . in other words , the screening length is smaller than @xmath44 .
the temperature scale @xmath45 is of the order of @xmath46 . a plot of the calculated @xmath2 as a function of temperature , at @xmath1 is given in fig .
[ plot1 ] .
+ the effect of disorder is neglected in the derivation of eq .
( [ onehalf ] ) , and this neglect is justified as long as @xmath47 , where @xmath48 is the single layer diagonal resistivity , and as long as @xmath49 . + the transresistivity at @xmath1 , as given by eq .
( [ onehalf ] ) , is _ much larger _ than at @xmath50 .
the latter is given by @xcite @xmath51 where @xmath52 is the fermi wavevector .
interestingly , in the limit of strong screening , both @xmath2 at @xmath50 and @xmath53 at @xmath1 are determined by a combination of geometrical factors ( interlayer distance and inter - particle distance ) and the strength of the coulomb interaction .
both are independent of the mass , whether bare or effective .
( [ onehalf ] ) holds also for the @xmath54 and @xmath55 cases , with one modification , namely @xmath56 .
thus , at low @xmath6 , if the filling fraction is varied from @xmath1 to @xmath54 or @xmath55 , keeping the electron density fixed , the transresistivity is expected to increase by a factor smaller , but close to , @xmath57 .
+ while this paper was in preparation , two other related works were submitted for publication @xcite .
both have considered the case @xmath36 and have found a temperature dependence of @xmath3 .
our analysis agrees with this result .
both works have used a composite fermions approach to analyze the transresistivity , and in one the unique temperature dependence was interpreted as reflecting the non - fermi liquid properties of the composite fermions . as explained below ,
our understanding of this temperature dependence is different .
2 . for @xmath36
we study the temperature dependence of @xmath2 for electron - electron interaction of the form @xmath58 .
although this study is mostly of theoretical value , it was proven to be of interest in the subject of the composite fermion fermi liquid theory @xcite .
we find the leading term to be , @xmath59 3 . for the case of @xmath60 and both near @xmath25 we find the leading temperature dependence of @xmath2 to be @xmath9 .
based on a semiclassical approximation , we find that this statement holds as long as the composite fermion cyclotron radius in each layer , @xmath61 , is larger than the distance a composite fermion traverses without being scattered by impurities .
the effect of fermion - fermion scattering on @xmath2 is presently under study .
4 . for two layers of densities @xmath62 and @xmath63 , with the average density corresponding to @xmath1
, we find that for low temperature and small @xmath64 , @xmath65 \ , .
\label{dn}\ ] ] 5 . finally , we study the case of @xmath26 .
the spin polarization of the single layer @xmath8 state is , at the moment , not well understood @xcite .
we calculate @xmath2 for the cases of complete spin polarization and zero spin polarization , and find that the transresistivity for a completely unpolarized @xmath8 state is larger by a factor close to @xmath66 compared to the transresistivity of a fully polarized @xmath8 state .
+ thus , if the picture emerging from ref .
@xcite is correct , and the @xmath8 is not spin - polarized , but may become polarized by an application of a magnetic field @xmath67 _ parallel _ to the layer , a measurement of @xmath2 as a function of @xmath67 should be sensitive to that change of polarization . + by passing , we also note that a change of polarization should be reflected also in surface acoustic waves measurements similar to the ones reported by willet @xcite . before turning to describe the physical picture emerging from our study , we comment on the approximations made
( [ donald ] ) is derived under the assumption of weak inter - layer coupling .
all throughout this paper we neglect the finite thickness of the two layers . since our main findings result from small-@xmath28 behavior , we expect the finite thickness not to affect any of the qualitative features outlined above , but rather to modify prefactors only .
both of these approximations are conventionally taken in coulomb drag studies @xcite .
approximations special to the case at hand are taken when the single layer @xmath68 is calculated .
as explained below , we carry our calculations in the modified rpa ( mrpa ) approach to the composite fermions fermi liquid theory @xcite .
we find , however , that our main findings turn out to be independent of the composite fermion effective mass @xmath69 and landau parameter @xmath70 introduced within mrpa , and thus can be expected to be insensitive to the details of the approximation .
the structure of the paper is as follows : in section [ physicalp ] we describe the physical picture behind our results . in section [ calculations ]
we give the detailed calculations .
section [ summary ] summarizes the paper . in the appendix
we discuss the relation of our approach to the composite fermion approach to the problem , as employed in @xcite and @xcite .
in this section we describe the physical picture behind the results summarized above .
we try to point out clearly what features of our analysis are independent of the magnetic field , and , in contrast , what is unique to @xmath1 and @xmath0 close to @xmath25 . to this end , we start by reviewing the general theoretical considerations leading to eq .
( [ donald ] ) and the precise definition of each term in this equation . after doing that
, we focus on the @xmath1 , and show that the unique features we find in the transresistivity result from the slow relaxation of density fluctuations in that state . the physical picture behind eq .
( [ donald ] ) was discussed by various authors @xcite . here , we elaborate on some points that are essential to our discussion .
drag resistivity stems from scattering events between electrons of different layers .
this scattering results from the screened inter - layer coulomb interaction @xmath18 .
scattering events transfer momentum @xmath71 and energy @xmath72 between the layers .
phase space availability effectively constrains the energy transfer to be smaller than the temperature , as implied by the @xmath73 factor .
the function @xmath68 is the single layer density - density response function , irreducible with respect to the coulomb interaction .
it is defined in the following manner : suppose that one applies a weak scalar potential @xmath74 on a single layer ( in the absence of a second one ) . in linear response
this potential leads to a particle density response @xmath75 , which induces a coulomb potential @xmath76 .
the response function @xmath22 is the ratio of the induced density to _ total _ electric potential @xmath77 , @xmath78 since current conservation implies @xmath79 and the electric field is related to the potential by a gradient operator , the longitudinal conductivity @xmath80 , relating charge current to the electric field , is related to @xmath81 through @xmath82 the response function @xmath68 also determines the screening of the coulomb potential , both inter - layer and intra - layer .
we denote the bare intra - layer coulomb interaction by @xmath83 and the bare inter - layer one by @xmath84 .
it is convenient to describe the interaction by @xmath85 matrices , where the two entries correspond to the two layers .
the interlayer screened coulomb interaction is the off - diagonal element of the matrix @xmath86 .
the latter satisfies the equation @xmath87 with @xmath88 note the notation : @xmath89 are matrices , @xmath90 , @xmath91 and @xmath92 are scalar functions .
we now limit ourselves to the case of two identical layers , which is our main focus in this paper . for that case @xmath93 ,
such that the subscript may be omitted .
then , the screened inter - layer potential can be written as @xmath94 eq .
( [ interscg ] ) can be understood in the following way .
the screened potential is the potential induced in one layer by a test charge of magnitude one in the other layer , together with the screening cloud it creates .
one can view a test charge in one layer as a sum of two charge distributions a symmetric distribution , with the same test charge in both layers , and an antisymmetric distribution , with opposite test charges in the two layers .
the bare coulomb potential induced by a symmetric distribution is @xmath95 , while the bare potential induced by an anti - symmetric distribution is @xmath96 .
the two terms in ( [ interscg ] ) are the screened coulomb potentials induced by the symmetric and anti - symmetric distributions . for the case
we consider here , of two coupled two - dimensional systems , @xmath97 and @xmath98 , such that , for @xmath99 , @xmath100 and @xmath101 .
a symmetric distribution induces a screened coulomb potential , while an anti - symmetric one induces a screened coulomb potential of _ electric dipoles _ , whose dipole moment is @xmath102 .
we now focus our attention on the product @xmath103 , a product that appears in eq .
( [ donald ] ) . using eq .
( [ interscg ] ) , and suppressing , for brevity , the @xmath27 arguments of @xmath104 and @xmath22 , this product may be written as @xmath105 by eq .
( [ donald ] ) , the transresistivity will be determined by @xmath27 for which the product @xmath103 is large .
thus , it is interesting to examine its poles .
as we see below , @xmath106 does not have any poles , and therefore the only poles are the solutions to the equations , @xmath107 @xmath108 to illuminate the physical significance of the poles , we expressed @xmath22 in terms of the conductivity @xmath109 .
the solutions to ( [ sympole ] ) and ( [ asympole ] ) are the dispersion relations for charge density modes : suppose that at time @xmath110 one sets up a symmetric charge modulation , @xmath111 in each of the two layers , and observes its time evolution .
this charge modulation induces , in each of the layers , an electric field given by @xmath112 .
the electric field generates , in turn , , a current @xmath113 in each of the layers .
since @xmath114 , the current leads to a decay of the charge modulation .
the dispersion relation corresponding to that decay is the one given in eq .
( [ sympole ] ) .
the second pole results from an anti - symmetric charge distribution , i.e. , a distribution in which at @xmath110 the charge modulation in the two layers is @xmath115 . in that case ,
the electric field induced in the two layers is @xmath116 .
again , this electric field leads to a decay of the charge modulation , following the dispersion relation given in eq .
( [ asympole ] ) . in both cases , when @xmath109 is real , the density modes are damped , and density modulations decay in time .
when @xmath109 is purely imaginary , the solutions to eqs .
( [ sympole ] ) and ( [ asympole ] ) are the plasma modes of the double layer system . as pointed out by zheng and macdonald @xcite , eq .
( [ donald ] ) is valid both in the absence and in the presence of a magnetic field . a simple way of understanding that point is by following the derivation of ( [ donald ] ) , as given by zheng and macdonald . upon doing that
, one finds that the derivation is unchanged in the presence of a magnetic field .
thus , the study of drag resistivity for the system at hand is reduced to a study of the density - density response function @xmath22 for electrons at @xmath1 . and it is there where the behavior of electrons at @xmath1 differs strongly from that of electrons at @xmath50 .
the calculation of @xmath68 is carried out , following a similar calculation by halperin , lee and read ( hlr)@xcite , in the next section , and is summarized by eq .
( [ pizz ] ) . for the description of the physical picture
it is enough to quote the result for @xmath117 and @xmath118 ( with @xmath119 being the fermi velocity ) : @xmath120 where @xmath121 is , by definition , the thermodynamic electronic compressibility of the @xmath1 state , and @xmath122 .
this form of @xmath22 is unique , and deserves a few comments : firstly , it is similar to the form of @xmath22 for a diffusive system at @xmath50 , namely , @xmath123 , but with the effective `` diffusion constant '' being linear in @xmath28 .
( [ pieq ] ) describes , then , a very slow diffusion , where the diffusion constant vanishes for @xmath124 .
this slow diffusion affects the dispersion relation of the two density relaxation poles of the screened coulomb potential , defined in eqs .
( [ sympole ] ) and ( [ asympole ] ) .
those become @xmath125 $ ] and @xmath126 $ ] , and describe a slow relaxation of density fluctuations , that eventually leads to a large drag resistivity . secondly , as revealed by surface acoustic waves experiments , explained by the composite fermion theory , and manifested in eq .
( [ pieq ] ) , the @xmath28-dependence of the longitudinal conductivity of electrons at @xmath1 is very different from that of electrons at zero magnetic field . at @xmath50
the electronic conductivity is usually at its maximum at @xmath127 , and decreases with increasing @xmath28 . at @xmath1 , in the absence of disorder , the longitudinal conductivity _ vanishes _ at @xmath127 , and increases linearly with increasing @xmath28 , as long as @xmath28 is not too large .
again , this small longitudinal conductivity leads to a slow decay of density fluctuations , and thus to a large transresistivity .
thirdly , eq . ( [ pieq ] ) describes a strong suppression of @xmath128 for large values of @xmath28 .
consequently , the @xmath28 integral in ( [ donald ] ) is _ not _ dominated by the upper cut - off @xmath129 , as is the case at @xmath50 , but rather by the solution @xmath130 of eq .
( [ asympole ] ) . since the @xmath16 integral
is dominated by @xmath131 , the most important contribution to the transresistivity comes from the region @xmath132 substituting eq .
( [ pieq ] ) into eqs .
( [ impiv ] ) and ( [ donald ] ) we get the leading order of eq .
( [ onehalf ] ) . the detailed calculation ,
including the next two terms of eq .
( [ onehalf ] ) , is given in the next section .
( [ qnot ] ) puts a limit on the validity of our results . our use of eq .
( [ pieq ] ) is valid only for @xmath133 , where @xmath134 is the single layer composite fermion elastic mean free path @xcite . since the electronic @xmath135
, this condition amounts to @xmath136 with the above discussion in mind , the origin of the dependence of @xmath2 on the range of the coulomb interaction [ eq .
( [ range ] ) ] can be understood .
interactions decaying slower than @xmath137 lead to a faster relaxation of long wavelength density fluctuations , and thus to a less singular temperature dependence of @xmath2 ( although still more singular than the @xmath138-dependence of non - interacting electrons ) . the case of short range interactions is discussed below .
we now comment on the temperature dependence of the transresistivity when the electron density in the two layers is varied such that the filling fraction is slightly away from @xmath1 .
the transresistivity is again determined by eq .
( [ donald ] ) , with the crucial ingredient being the density - density response function @xmath22 .
for the analysis of @xmath68 we resort to the composite fermion approach . for @xmath0 slightly away from @xmath25 , a new length , the composite fermion cyclotron radius , @xmath139 , and a new frequency scale , the composite fermion cyclotron frequency , @xmath140 are introduced .
as long as @xmath139 is larger than the transport mean free path @xmath141 , introduced by disorder , the composite fermion dynamics is , to a large extent , unaffected by the effective magnetic field they are subject to .
the qualitative @xmath27 dependence is then similar to that of the @xmath1 case , and we expect @xmath142 .
moreover , even when @xmath139 is smaller than the disorder - induced mean free path , scattering events between composite fermions might suppress the sensitivity of the composite fermions to the effective magnetic field they are subject to .
we thus expect that the @xmath9 temperature dependence may even extend to the region @xmath143 .
the response function @xmath22 for a single layer at @xmath1 was calculated by hlr within the random phase approximation ( rpa ) .
improvements to that approximations were developed in refs .
we now review the rpa calculation and its improvements .
the calculation of the electronic density - density response function necessitates a notation for single layer response functions .
the single layer response functions are @xmath85 matrices , whose entries correspond to density and transverse current .
we urge the reader to distinguish single layer response functions from the @xmath85 matrix @xmath144 ( defined in eq .
[ matrixdef ] ) . in the latter
the two entries correspond to densities in the two layers . to help this distinction
, we denote single layer matrices by tildes , while double layer matrices are denoted by hats .
the single layer electronic density - density response function is the density - density element of the electronic single layer response function @xmath145 .
the following discussion , concluded by eq . ( [ pizz ] ) , is devoted to a calculation of that element .
consequently , the two matrices we are about to define now , @xmath146 and @xmath147 , are _ single layer
_ matrices , with entries corresponding to density and transverse current . defining the chern - simons interaction matrix , @xmath148
we write @xmath149 where @xmath150 is the composite fermion density - density response function , describing the response of the composite fermions to the total scalar and vector potentials they are subject to , including external , coulomb and chern - simons contributions .
diagrammatically , @xmath151 is the sum of all diagrams irreducible with respect to a single chern - simons or coulomb interaction line . within rpa ,
the composite fermion response function @xmath151 is that of non - interacting fermions at zero magnetic field .
it is therefore a diagonal matrix .
to lowest order in @xmath152 and @xmath153 ( with @xmath119 being the fermi velocity ) its components are , @xmath154 where @xmath155 is the bare mass . the finite value of @xmath156 in that limit reflects the compressibility of the system , while the limit of @xmath157 reflects landau diamagnetism ( the real part ) and landau damping ( the imaginary part ) .
improvements to rpa include the modified rpa ( mrpa ) , in which the bare mass is replaced by a renormalized mass @xmath69 , and a landau interaction parameter @xmath70 is introduced @xcite , and the magnetized modified rpa ( mmrpa ) , in which an orbital magnetization is attached to each fermion @xcite . within mrpa , eq .
( [ picfrpaa ] ) is replaced by @xmath158 the attachment of magnetization does not affect the component @xmath159 of the matrix @xmath160 , which is the only one relevant for coulomb drag @xcite .
thus , we use eqs .
( [ picf ] ) and ( [ picfrpab ] ) to calculate the electronic density - density response function @xmath161 obtaining , @xmath162 which is eq .
( [ pieq ] ) .
note that the thermodynamic compressibility is @xmath163 .
substituting eq .
( [ pizz ] ) in eqs .
( [ impiv ] ) and ( [ donald ] ) we find , @xmath164 the origin of the @xmath165 term is in the second order expansion of the @xmath166 factor in the bare inter - layer coulomb interaction .
a numerical evaluation of eq . ( [ donald ] ) using ( [ pizz ] ) yields the temperature dependence plotted in fig .
( [ plot1 ] ) .
the parameters used in the numerical calculation are @xmath167@xmath168 , @xmath169 , and @xmath1 .
the numerical calculation uses eq .
( [ pizz ] ) for @xmath22 , but uses the exact value of @xmath92 , rather than a small @xmath170 expansion . the transresistivity at @xmath54 is obtained directly from eqs .
( [ onehalfd ] ) and ( [ 4:t0 ] ) .
the transresistivity at @xmath55 is obtained by regarding the @xmath55 state as a superposition of a @xmath54 state of holes and a @xmath171 state of electrons . in the study of the chern - simons fermi liquid theory formed by composite fermions in the @xmath1 state
it is useful to consider electron - electron interactions of the form @xmath172 with @xmath173 @xcite . for @xmath174 ,
the composite fermions form a conventional fermi liquid
the effective mass and quasi - particle residues are finite . at @xmath175 , coulomb interaction ,
the fermi liquid is `` marginal '' , i.e. , both quantities are logarithmically singular . and for @xmath176 , the effective mass strongly diverges and the quasi - particle residue vanishes at the fermi level .
in all cases , however , the compressibility of the @xmath1 state is not singular @xcite .
the small @xmath28 behavior of the potential ( [ eta ] ) is @xmath177 , and consequently the bare potential induced by anti - symmetric density modulations is @xmath178 .
the form of @xmath68 given in eqs .
( [ pieq ] ) and ( [ pizz ] ) does not depend on the interaction potential .
thus , in the calculation of the drag resistivity one has to distinguish between two cases : 1 .
long range interactions where @xmath179 : in that case the range of the interaction determines the dispersion relation of the anti - symmetric density relaxation modes ( [ asympole ] ) to be @xmath180 .
the leading temperature dependence of the transresistivity is then @xmath181 .
although the composite fermions form a fermi liquid , @xmath2 does not exhibit the @xmath138 temperature dependence characteristic of non - interacting electrons at zero magnetic field .
the difference stems from the slow decay of density modulations .
short range interactions where @xmath182 : in that case the dispersion relation of anti - symmetric density relaxation modes ( [ asympole ] ) is not determined by the interaction , but rather by the compressibility a density modulation induces a spatial dependence in the chemical potential , which creates a current , which relaxes the modulation of the density .
the dispersion relation is @xmath183 , independent of @xmath184 .
consequently , the temperature dependence of @xmath2 is not determined by @xmath184 , and is @xmath185 . when the two layers are at identical filling fractions close to @xmath1 , the density - density response functions are identical @xmath186 .
within the composite fermion approach , both can be calculated by eq .
( [ picf ] ) . the function @xmath187 is , in that case , the response function of composite fermions at an effective magnetic field of @xmath188 .
we limit ourselves here to relatively high temperatures , @xmath189 . in that regime
a semiclassical random phase approximation may be employed for a calculation of @xmath81 , as was done , e.g. , by hlr and by simon and halperin@xcite .
the semiclassical composite fermion response function @xmath147 is @xcite , @xmath190'}{1 - \frac{i}{\omega_c^*\tau } \frac{\pi } { \sin{\pi w } } j_w(qr_c^*)j_{-w}(qr_c^ * ) } \nonumber \\
\pi^{\mbox { \tiny cf}}_{11 } & = & \frac { \omega^2}{q^2 } \pi^{\mbox { \tiny cf}}_{00 } \nonumber \\ \lefteqn { - v_{\scriptscriptstyle f}^2 \frac{m}{2 \pi \hbar^2 } \frac { \frac { \omega}{\omega_c^ * } \frac { \pi}{\sin \pi w } j_{1 + w}(qr_c^*)j_{1 - w}(qr_c^*)}{1 - \frac{i}{\omega_c^*\tau } \frac{\pi } { \sin{\pi w } } j_w(qr_c^*)j_{-w}(qr_c^ * ) } \ , , } \label{scpicf}\end{aligned}\ ] ] where @xmath191 , @xmath192 is the bessel function of order @xmath193 , and a prime denotes differentiation with respect to the argument . in the limit of @xmath194 ( or ,
equivalently @xmath195 ) we find , using asymptotic expansion of the bessel functions , that the response functions ( [ scpicf ] ) reduce , at the relevant range of @xmath196 to the zero effective magnetic field response functions ( [ picfrpaa ] ) .
thus , in that regime the leading temperature dependence of the transresistivity is expected to be approximately equal to that of the @xmath1 case .
there are several contributions to the composite fermion scattering rate .
the first is impurity scattering , which is independent of energy .
the second is mutual composite fermion scattering , which does depend linearly on energy ( at least at the @xmath1 case @xcite ) .
the precise effect of the latter on the response at finite @xmath28 is a delicate subject , which is presently under investigation @xcite .
we now turn to consider two layers of densities @xmath197 and @xmath198 , where the density @xmath199 corresponds to a filling fraction @xmath1 .
measurements of the dependence of the transresistivity on a density bias between the two layers were done in the past as a way of separating out coulomb drag and phonon drag mechanisms .
a calculation of the dependence of @xmath2 on such a density bias could therefore be useful for such measurements at @xmath1 .
since the two layers are not identical , the screened inter - layer coulomb potential can not be put in the form ( [ interscg ] ) , and is rather given by @xmath200 [ { \bar\pi}(v_b - u_b)]-\delta\pi^2(v_b^2-u_b^2 ) } \ , , \label{interscgne}\ ] ] where @xmath201 . the density bias @xmath64 changes the density - density response function @xmath22 in two ways .
the first , trivial , way , is through the dependence of @xmath22 on the electron density .
the second way is by changing the filling fraction .
however , as shown above , at least within the semiclassical approximation , if @xmath139 is larger than the mean free path the composite fermion response functions at the relevant range of @xmath196 are approximately insensitive to the effective magnetic field @xmath202 .
thus , the main effect of @xmath203 comes through the dependence of @xmath22 on the density . using ( [ donald ] ) and ( [ interscgne ] ) with two different response functions for the two layers
, we find that for small @xmath64 , and to leading order in the temperature , the transresistivity is @xmath204 \ , .
\label{deltan}\ ] ] we now analyze the transresistivity for two layers of @xmath8 each .
our motivation for studying this case stems from the ambiguous experimental picture of spin polarization at the @xmath8 state @xcite .
while transport measurements at tilted magnetic fields indicate only a partial spin polarization of the @xmath8 state , surface acoustic waves measurements do not seem to support such a picture @xcite .
we therefore attempt to find out whether a measurement of transresistivity can shed light on the spin polarization of the @xmath8 state .
we regard the @xmath8 state as a superposition of a @xmath205 state of electrons and a @xmath1 state of holes , a picture which is valid when the electronic landau level separation @xmath206 is much larger than the temperature and the coulomb energy scales , such that landau level mixing can be neglected . under these conditions
the contribution of the electronic @xmath205 state to the single - layer density - density response function @xmath22 is exponentially small for the relevant frequency range @xmath207 , due to the energy gap between the @xmath208 and @xmath209 electronic landau levels .
the transresistivity is therefore almost solely due to the @xmath1 state of the holes
. the spin polarization of the @xmath1 hole state may be anywhere between zero and full polarization .
so far we considered only fully polarized states .
when considering a partially polarized state , one expects all relevant functions ( @xmath210 of eqs .
( [ screened ] ) and ( [ matrixdef ] ) ) to acquire spin indices .
as a first step , then , a convenient choice of a basis may be helpful .
this choice emerges naturally when considering the bare coulomb interaction , even within one layer .
we write the intra - layer bare interaction as a @xmath85 matrix , where the two entries are for densities of electrons with two spin states . in the basis of eigenvectors of @xmath211 ,
@xmath212 since the coulomb interaction does not depend on the spin direction .
the convenient basis is the basis of charge density @xmath213 and spin density @xmath214 ( where @xmath215 ( @xmath216 ) denote the density of electrons whose spin is parallel ( anti - parallel ) to the magnetic field ) . in that basis ,
@xmath217 the screened coulomb interaction has a structure similar to ( [ barespinb ] ) , since it , too , couples only to charge densities , and does not depend on spin densities .
consequently , the only component of the response function @xmath22 that affects the transresistivity is the one coupling charge density to charge density . for the fully polarized case
, it is given by eq .
( [ pieq ] ) .
we now turn to discuss this component in the other extreme case , that of zero spin polarization . in principle , one may envision several chern - simons transformations that might be applicable to the fully unpolarized @xmath1 state . at @xmath1
there are two flux quanta per a hole .
one possible transformation attaches four flux quanta to each hole , and allows an interaction only between same - spin composite fermions .
this transformation creates a correlated wave function of same - spin holes , and reduces the probability of finding two same spin holes very close to one another @xcite .
it does not , however , do the same for holes of different spin directions .
thus , while the coulomb energy cost associated with holes of identical spins is reduced , that of holes of different spin directions is unaffected .
an energetically favorable transformation , that reduces the coulomb energy of both types of interactions , is one that attaches two flux quanta to each hole , and allows an interaction that is independent of the spin direction .
the correlations induced by this transformation are , at least on the rpa level , spin - independent .
we use the second transformation .
the calculation of @xmath22 is similar to that leading to eq .
( [ pizz ] ) .
( [ picf ] ) still holds , although @xmath218 and @xmath219 become single layer @xmath220 matrices , with the entries corresponding to densities and transverse current of two spin states . within mrpa , in the basis of eigenstates of @xmath211 ,
the response function @xmath221 is a block - diagonal matrix with two pairs of identical blocks , denoted by @xmath222 ( note that @xmath223 are @xmath85 matrices with entries corresponding to density and transverse current ) . since in our choice of the chern - simons transformation
the chern - simons interaction is independent of the spin direction , it is again more convenient to transform the spin states from eigenstates of @xmath211 to @xmath224 ( and similarly for transverse currents ) .
the components relevant for our calculation are the ones coupling charge density and currents ( i.e. , spin density and currents are irrelevant ) , and they are given by the @xmath85 matrix @xmath225 .
using eq .
( [ picfrpaa ] ) and ( [ picfrpab ] ) , we then find the following expression for the electronic density - density response function , to be compared with eq .
( [ pizz ] ) , @xmath226 qualitatively , eq .
( [ pieq_th ] ) is very similar to eq .
( [ pizz ] ) , the corresponding response function for the polarized state .
quantitatively , there are three differences : a factor of @xmath227 in the numerator resulting from the presence of two spin directions , a factor of @xmath228 originating from the interaction of each composite fermion with flux tubes carried by fermions of @xmath229 spin directions , and a factor @xmath230 difference between @xmath231 and @xmath232 , originating from the difference between the fermi wave - vector at the polarized and unpolarized states .
we note , by passing , that this quantitative difference between @xmath22 of the fully unpolarized state and @xmath22 of the fully polarized state should be observable in measurements of the @xmath28-dependent conductivity using surface acoustic waves .
for example , one might imagine a measurement of surface acoustic waves attenuation in a @xmath8 state as a function of a magnetic field applied parallel to the two dimensional electronic system .
if the parallel magnetic field modifies the spin polarization of the @xmath8 state , a significant change in @xmath233 should be observable .
coming back to the transresistivity , an examination of eq .
( [ pieq_th ] ) , eq .
( [ impiv ] ) and eq .
( [ donald ] ) reveals that the drag transresistivity in the unpolarized case can be obtained from that of the polarized case by three modifications : firstly , @xmath234 should be multiplied by @xmath227 , to account for the factor @xmath228 in the denominator of ( [ pieq_th ] ) .
secondly , the bare interaction should be multiplied by @xmath227 , to account for the factor @xmath227 in the numerator of ( [ pieq_th ] ) . and thirdly , the fermi momentum @xmath52 appearing in the electronic density - density response function ( [ pizz ] )
should be divided by @xmath235 . neglecting @xmath43
compared to @xmath35 , we then find that the transresistivity in the unpolarized case is larger , by a factor of @xmath66 compared to that of the polarized case .
in this paper we considered the coulomb drag between two layers of two dimensional electron gases in a strong magnetic field , when the filling fraction in each layer is at , or close , to @xmath1 , @xmath54 and @xmath55 , and the electrons in each layer form a compressible state . using the composite fermion approach to analyze the properties of each layer , and an electronic approach to analyze the coupling between the layers , we find that the unique linear @xmath28-dependence of the longitudinal conductivity , @xmath236 , at and close to even denominator filling fractions , lead to a unique temperature dependence of the transresistivity between two layers , @xmath3 .
in contrast to previous works , we do not associate this temperature dependence to the nature of the liquid formed by composite fermions , but rather interpret it as a result of the slow relaxation of density fluctuations in a compressible partially filled landau level .
we examine the transresistivity at @xmath8 and find that it depends significantly on the spin polarization of the @xmath8 state .
we therefore propose that measurements of @xmath2 at @xmath8 might shed light on that polarization , which so far is not well understood .
our results rely on the assumption of weak coupling between the layers and neglect the finite thickness of each layer .
the properties of each layer are analyzed using approximation schemes for the composite fermion problem .
we find that @xmath2 depends mostly on the strength of the coulomb interaction and on geometrical factors , such as the inter - layer and inter - electron distance .
its dependence on masses , whether bare or effective , is rather weak .
we believe that these observations make the results we obtain mostly independent of the fine details of the approximation we use .
we are indebted to j. p. eisenstein and t. j. gramila for discussing unpublished data with us , and to b. i. halperin , a. kamenev , y. oreg , d. orgad , and s. h. simon for instructive discussions .
this research was supported by a grant from the us - israel binational science foundation and by a grant from the minerva foundation ( munich , germany ) .
a. s. is supported by the v. ehrlich career development chair .
in the main body of the paper we discussed the transresistivity from a purely electronic point of view , showing that it originates from the coulomb drag between electrons at the two layers . in this appendix
we relate the electronic calculation to a calculation in terms of composite fermions , which is the approach taken in refs .
@xcite and @xcite .
first , we note that as long as the inter - layer coupling is weak , composite fermions in one layer do not interact with the chern - simons field of the other layer ( an inter - layer chern - simons interaction gives rise to @xmath237-type states ) . under these conditions , the electronic drag resistivity is identical to the composite fermion drag resistivity , @xmath238 in a single layer it is well known @xcite that the composite fermion resistivity matrix @xmath239 and the electronic resistivity matrix @xmath240 are related by @xmath241 eq .
( [ rer ] ) results from an extension of eq .
( [ rescfre ] ) to the @xmath220 double layer resistivities .
similar to the single layer case , electronic longitudinal resistivities are equal to those of the composite fermions .
the electronic ( measurable ) drag conductivity is _ not _ identical to the composite fermion drag conductivity .
thus , in a composite fermion approach , we calculate @xmath242 .
the following derivation is limited to the case of two identical layers at @xmath243 , and assumes that @xmath244 is diagonal .
extensions will be discussed elsewhere @xcite .
similarly to the usual case , the transresistivity is given by @xcite , 2 where @xmath246 get the values @xmath247 or @xmath35 , corresponding to density and transverse currents .
the summation over both density and transverse current components is needed since the screened interaction couples composite fermions both through their densities and their transverse currents . in eq .
( [ rdcf ] ) the composite fermion transresistivity is given in terms of composite fermion response functions .
we now turn to express it in electronic terms .
we first relate the screened interlayer interaction of composite fermions @xmath248 to the electronic screened interlayer interaction @xmath249 ( the former is a @xmath250 matrix , with density and transverse current indices , while the latter is not , since electrons interact only through the coulomb interaction ) . by comparing the two , one finds @xmath251 where the chern - simons interaction matrix @xmath146 is defined in eq .
( [ cs ] ) . this relation may be observed diagrammatically as in fig .
[ plot2 ] , and may be derived from the expression for @xmath249 [ eq . ( [ interscg ] ) ] , @xmath252 where @xmath91 and @xmath92 are the intralayer and interlayer bare coulomb interactions . inserting eq . ( [ a : ue ] ) in the right hand side of eq .
( [ 3:uurelation ] ) , and using [ see eq .
( [ picf ] ) ] @xmath253 we obtain for the right hand side of eq .
( [ 3:uurelation ] ) , @xmath254 this is exactly the expression for @xmath248 @xcite . 2 where @xmath256 we now show that @xmath257 is , in fact , the imaginary part of the electronic density - density response function , @xmath258 .
from eq .
( [ a : picf ] ) we have @xmath259 . using @xmath260 , we obtain @xmath261 on comparing ( [ 3:xi ] ) with ( [ 3:impi00 ] ) , it is left to show that @xmath262 both relations are verified using ( [ a : picf ] ) . 2 the only response function which enters this formula is the electron density - density response function @xmath265 , and this is the only place where properties of the half - filled landau level affect the drag resistivity .
p. j. price , physica b+c * 117/118 * , 750 ( 1983 ) ; m. b. pogrebinskii , fiz .
. poluprovodn . * 11 * , 637 ( 1977 ) [ sov .
. semicond . * 11 * , 372 ( 1977 ) ] ; p. m. solomon and b. laikhtman , superlatt . microstruc . * 10 * , 89 ( 1991 ) ; a .-
jauho and h. smith , phys . rev .
b * 47 * , 4420 ( 1993 ) ; k. flensberg , b. y .- k .
hu , a .- p .
jauho , and j. m. kinaret , phys .
b * 52 * , 14761 ( 1995 ) .
m. c. bnsager , k. flensberg , b. y .- k .
hu , and a .-
jauho , phys .
lett . * 77 * , 1366 ( 1996 ) ; m. w. wu , h. l. cui , and n. j. m. horing , preprint cond - mat/9604004 ( unpublished ) ; h. c. tso and p. vasilopoulous phys .
b * 45 * , 1333 ( 1992 ) ; e. shimshoni and s. sondhi , phys .
b * 49 * , 11484 ( 1994 ) .
a. stern and b. i. halperin , phys .
b * 52 * , 5890 ( 1995 ) . y. b. kim , a. furusaki , x .-
wen , and p. a. lee , phys .
b * 50 * 17917 ( 1994 ) .
b. l. altshuler , l. b. ioffe and a. j. millis , phys .
b * 50 * , 14048 ( 1994 ) . c. nayak and f. wilczek , nuc .
* b417 * , 359 ( 1994 ) | we consider the coulomb drag between two layers of two - dimensional electronic gases subject to a strong magnetic field .
we first focus on the case in which the electronic density is such that the landau level filling fraction @xmath0 in each layer is at , or close to , @xmath1 . discussing the coupling between the layers in purely electronic terms ,
we show that the unique dependence of the longitudinal conductivity on wave - vector , observed in surface acoustic waves experiments , leads to a very slow decay of density fluctuations .
consequently , it has a crucial effect on the coulomb drag , as manifested in the transresistivity @xmath2 .
we find that the transresistivity is _ very large _ compared to its typical values at zero magnetic field , and that its temperature dependence is unique @xmath3 .
for filling factors at or close to @xmath4 and @xmath5 the transresistivity has the same @xmath6-dependence , and is larger than at @xmath7 .
we calculate @xmath2 for the @xmath8 case and propose that it might shed light on the spin polarization of electrons at @xmath8 .
we compare our results to recent calculations of @xmath2 at @xmath1 where a composite fermion approach was used and a @xmath9-dependence was obtained .
we conclude that what appears in the composite fermion language to be drag induced by chern - simons interaction is , physically , electronic coulomb drag . 2 |
high - precision measurement of hyperfine structure in the excited states of alkali atoms provides a stringent testing ground for state - of - the - art atomic calculations based on the best wavefunctions @xcite , because it is sensitive to effects such as core polarization and electron correlation .
the @xmath2 line of alkali atoms is routinely used in laser cooling and bose - einstein condensation experiments .
these ultracold atoms are important in experiments ranging from fundamental physics issues such as high - resolution spectroscopy , measurement of fundamental constants , and the study of cold collisions , to applications in atomic clocks and inertial sensors .
heavy alkali atoms are also important in studies of atomic parity violation and the search for a permanent atomic electric dipole moment . in these experiments ,
accurate calculations are needed to properly interpret the experimental results .
in addition , hyperfine structure provides important information about the structure of the nucleus , i.e. the multipole moments of its charge and current distribution @xcite .
hyperfine structure has been measured with high accuracy in the ground state of alkali atoms using atomic beam magnetic resonance techniques . thus the hyperfine constants are known with a relative precision of better than @xmath3 .
indeed , this high precision of the measurements has led to the definition of the si unit of time in terms of the ground hyperfine interval in cs . however , measurement of hyperfine structure in the excited state is often limited by the linewidth of the transition , which is about 510 mhz for the @xmath1 lines of alkali atoms .
the advent of narrow linewidth tunable lasers ( ring dye lasers and frequency - stabilized diode lasers ) and high - resolution doppler - free spectroscopy techniques ( such as saturated - absorption spectroscopy ) has allowed individual hyperfine transitions to be resolved .
but there is still an issue of frequency calibration and linearity of the laser scan axis .
we have solved this problem by using an acousto - optic modulator ( aom ) , driven by a easily - measured radio frequency source , to span the difference between hyperfine transitions .
we are thus able to measure hyperfine intervals in the @xmath1 lines of alkali atoms with a typical precision of 12 parts in 1000 of the linewidth . in earlier work ,
we had reported measurement of hyperfine structure in a few alkali atoms . in the current work ,
we present comprehensive results on all the naturally - occurring alkali atoms .
the results , in most cases , are consistent with previously published values but represent a significant improvement in precision .
let us first consider the definition of the hyperfine constants .
the basis states used for defining hyperfine structure are the @xmath4 states , where @xmath5 , @xmath6 , and @xmath7 are the quantum numbers associated with , respectively , the electronic angular momentum @xmath8 , the nuclear spin @xmath9 , and the total angular momentum @xmath10 . in the approximation that @xmath5 and @xmath6 are good quantum numbers , the hyperfine interaction energy @xmath11 is given by , @xmath12 where @xmath13 , @xmath14 is the magnetic - dipole coupling constant , and @xmath15 is the electric - quadrupole coupling constant .
the first excited @xmath0 state of alkali atoms splits into two states , @xmath16 and @xmath17 , due to spin - orbit interaction ( fine - structure splitting ) . of these , the @xmath16 state ( @xmath18 line ) further splits into two hyperfine levels with energy shifts determined by the magnetic - dipole interaction . the electric - quadrupole interaction is present only for @xmath19 , and hence does not affect this state . on the other hand ,
the @xmath17 state ( @xmath2 line ) splits into four hyperfine levels with energy shifts determined by both interactions . in the case of @xmath20cs
, we will see later that there is evidence of a small additional contribution from a magnetic - octupole hyperfine interaction .
the experimental schematic to measure hyperfine intervals is shown in fig . 1 , and is essentially the same for all atoms .
the output from a frequency - stabilized tunable laser is split into two parts .
the first part goes into an atomic spectrometer , the signal from which is used to lock the laser to a particular hyperfine transition .
the second part is frequency shifted through an aom and then sent into a second atomic spectrometer .
the frequency of the aom is adjusted so that the shifted beam is on a neighboring hyperfine transition .
the error signal from this peak is fed back to the aom driver to lock its frequency .
thus the aom frequency gives a direct measurement of the hyperfine interval . for intervals that are too small or too large to be measured with a single aom ,
we use an additional aom with a fixed frequency offset .
the @xmath1 lines in sodium are in the visible ( at 589 nm ) , hence the tunable laser is a ring dye laser ( coherent 699 - 21 ) .
the laser is frequency stabilized to a reference cavity that gives it an instantaneous linewidth of 1 mhz .
the @xmath1 lines in all the other atoms are in the near infrared which can be accessed using diode lasers .
the diode laser is frequency stabilized using optical feedback from a piezo - mounted grating so that the resulting linewidth is about 500 khz .
the linewidth is reduced considerably when the laser is locked . for locking ,
the laser is frequency modulated at @xmath21 khz , and the signal from the spectrometers is demodulated at @xmath22 to generate the error signals .
such third - harmonic detection provides narrow dispersive signals that are insensitive to intensity fluctuations , to peak pulling from neighboring transitions , or to any residual doppler profile in the spectrum @xcite . in the case of li
, the spectroscopy is done using a collimated atomic beam .
the atoms are excited by a perpendicular laser beam , and the resulting doppler - free fluorescence signal is measured with a photomultiplier tube ( pmt ) . in all the other atoms , we use absorption spectroscopy of a probe beam through a vapor cell .
the cell is either at room temperature ( rb and cs ) or heated to @xmath23c ( na and k ) , to have an atomic density of about @xmath24 @xmath25 . in order to get doppler - free absorption spectra
, we use one of two techniques .
the first technique is the normal saturated - absorption spectroscopy ( sas ) @xcite , where a strong counter - propagating pump beam decreases the absorption of the weak probe beam on resonance . as is well known , the sas technique produces additional crossover resonances , which occur exactly midway between two hyperfine peaks . for closely - spaced levels ,
the large crossover resonances often swamp the true peaks .
this is a problem especially in the @xmath2 lines of na and k , where the level spacing is of the order of one or two natural linewidths . to overcome this problem ,
we have recently developed an alternate technique of high - resolution spectroscopy , which we call coherent - control spectroscopy ( ccs ) @xcite .
this technique uses co - propagating control and probe beams , and does not produce crossover resonances . the probe beam is locked on to a hyperfine transition , while the control beam is scanned across a neighboring transition .
when the control comes into resonance with a transition coupling to the same ground level , it `` coherently '' reduces probe absorption through a process similar to electromagnetically induced transparency .
since the probe is locked on a transition , it addresses only the zero - velocity atoms , and the resultant lineshape remains doppler - free even in hot vapor .
we have also done a density - matrix analysis of the effective three - level atom in the presence of the probe and control beams .
the calculated lineshape , after accounting for thermal averaging in hot vapor , describes the measured spectrum quite well @xcite .
apart from the ability to resolve closely - spaced levels , another important advantage of the ccs technique for measuring hyperfine intervals is that the measured value is completely insensitive to detuning of the probe beam .
any detuning of the probe from resonance would imply that it is resonant with a non - zero velocity group . since the control beam co - propagates with the probe beam ,
the spectrum will show peaks only when the control beam comes into resonance with the same non - zero velocity group , which happens only when the aom shift matches the hyperfine interval .
thus , as the probe beam is detuned from resonance , the manifold of peaks ( corresponding to the different hyperfine levels ) will shift within the doppler profile , but their relative separation will remain the same .
the different sources of error in the measurement have been discussed extensively in our earlier publications , and are reviewed here for completeness .
the primary sources of statistical error are the fluctuations in the lock point of the laser and the aom . to minimize these errors
, we use an integration time of 10 s in the frequency counter during each measurement of the aom frequency .
then we take an average of 3540 measurements for a given transition , and repeat the set several times . the resulting standard deviation ( scatter ) in the data set is about 10 khz , from which we take the statistical error in the mean to be @xmath26 khz .
the timebase in the frequency counter used for measuring the aom frequency has a stability of better than @xmath27 , which translates to a negligible error of @xmath28 khz in the frequency measurement .
systematic errors can occur if there are systematic shifts in the lock points of the laser and the aom .
this can arise due to one of the following reasons
_ radiation - pressure effects .
_ radiation pressure causes velocity redistribution of the atoms in the vapor cell . in the sas technique , the opposite doppler shifts for the counter - propagating beams can result in asymmetry of the observed lineshape @xcite .
we minimize these effects by using beam intensities that are much smaller than the saturation intensity . in the ccs technique ,
radiation pressure effects are less important because the doppler shift will be the same for both beams and will not affect the hyperfine interval , similar to how the interval is insensitive to any detuning of the probe from resonance .
effect of stray magnetic fields .
_ the primary effect of a magnetic field is to split the zeeman sublevels and broaden the line without affecting the line center .
however , line shifts can occur if there is asymmetric optical pumping into zeeman sublevels . for a transition @xmath29 , the systematic shift of the line center is
@xmath30 , where @xmath31 mhz / g is the bohr magneton , @xmath32 s denote the land @xmath32 factors of the two levels , and @xmath15 is the magnetic field .
the selection rule for dipole transitions is @xmath33 , depending on the direction of the magnetic field and the polarization of the light .
thus , if the beams are linearly polarized , there will be no asymmetric driving and the line center will not be shifted .
we therefore minimize this error in two ways .
first , we use polarizing cubes to ensure that the beams have near - perfect linear polarization .
second , we use a special magnetic shield @xcite around the cells to minimize the field . the residual field ( measured with a 3-axis fluxgate magnetometer )
is below 5 mg with one layer of shielding , and further reduces to 1 mg with two layers .
phase shifts in the feedback loop .
_ we check for this error by replacing the aom with two identical aoms , and adjusting them so that they produce opposite frequency offsets .
with the laser locked to a given hyperfine transition , the first aom then produces a fixed frequency offset which is compensated by the second aom . thus the same hyperfine transition is used for locking in both spectrometers . under these conditions
, the second aom should lock to the fixed frequency of the first aom , with any error arising solely due to phase - shift errors .
we find that the second aom tracks the frequency of the first aom to within 1 khz .
4 . _ shifts due to collisions . _
to first order , collisional shifts are the same for different hyperfine levels , and hence do not affect the interval . small differential shifts of the interval have been studied carefully in the ground state of cs , due to its importance in atomic clocks . however
, the size of the shift is in the mhz range @xcite .
we have also studied shifts in the excited state of cs by repeating the measurements with a heated vapour cell .
an increase from 25@xmath34c to 45@xmath34c increases the cs density by a factor of 5 , and we find that the measured values shift by less than 5 khz .
we therefore estimate that the maximum shift due to collisions is 5 khz .
the sizes of the various sources of error are listed in table [ t1 ] .
the size of the zeeman shift is calculated in each atom by considering all allowed combinations of @xmath35 and @xmath36 and taking the one with the largest shift , assuming atoms get completely optically pumped into these sublevels . as mentioned in point ( i ) above
, the peak center can be shifted due to radiation - pressure effects .
similarly , from point ( ii ) , the center can be shifted if there is _ asymmetric _ pumping into the zeeman sublevels in the presence of a residual magnetic field .
asymmetric optical pumping can occur if the beam polarizations are not perfectly linear , for example due to imperfections in the cubes or birefringence at the cell windows .
finally , there can be a small ac stark shift of the transition due to nearby hyperfine levels from which the laser is detuned .
from an experimental point of view , all these effects will change with laser power .
therefore , we can check for these errors by repeating the measurements at different values of power , and extrapolate to zero power if necessary .
we now consider the hyperfine measurements in each atom in turn .
we present a detailed description of the measurement in li for several reasons .
first , li is the only atom in which we use an atomic beam for spectroscopy .
therefore , it is slightly different from the vapor - cell technique used in other atoms , which has been described in detail in our earlier publications .
second , we give a complete list of the hyperfine intervals measured at different powers .
as mentioned above , this method is used to check for intensity - dependent errors , and we just present a summary of the results in the other atoms . finally , the small ground hyperfine interval in li offers a unique possibility of checking the reliability of our technique , which is not easy in the other atoms .
lithium has two stable isotopes : @xmath37li with @xmath38 and natural abundance of 7.3% , and @xmath39li with @xmath40 and natural abundance of 92.7% .
li spectroscopy was done using a collimated atomic beam generated by heating a stainless steel oven containing li metal to 300@xmath34c .
the oven was placed inside a vacuum chamber maintained at a pressure below @xmath41 torr with a 20 l / s ion pump .
the atomic beam was excited with a perpendicular laser beam generated from a home - built diode laser system operating at 670 nm .
the laser beam was linearly polarized and the interaction region was shielded with a two - layer magnetic shield .
the resulting doppler - free fluorescence signal was detected by a pmt ( hamamatsu r928 ) .
we obtained spectra from the @xmath1 lines of both isotopes . in the @xmath18 line ,
the individual hyperfine transitions are well resolved and we can lock to each peak separately .
however , in the @xmath2 line , the different transitions are only partially resolved because the level spacing is less than the natural linewidth . in recent work where we measured the absolute frequencies of these transitions
@xcite , we showed that a multipeak fit to the partially - resolved spectrum yields the hyperfine constants in the @xmath42 state with a precision of about 50 khz . this is less than the precision with which they are known from earlier experiments .
therefore , in this work we report hyperfine structure only in the @xmath43 state .
typical spectra for the @xmath18 line in the two isotopes of li are shown in figs .
[ f2a](a ) and ( b ) .
the four hyperfine transitions in each isotope are clearly resolved .
close - up scans ( normalized ) of a representative peak in each isotope are shown in figs .
[ f2a](c ) and ( d ) , respectively .
the solid lines are lorentzian fits showing the excellent fit with featureless residuals .
the size of the residuals gives an idea of the overall signal to noise ratio .
the fit linewidth is @xmath44 mhz ( compared to the natural linewidth of 5.87 mhz ) , indicating that the atomic beam is well collimated because the spread in transverse velocity increases the total linewidth by only about 50% .
the laser linewidth does not contribute significantly to this broadening .
an important consideration in atomic beam experiments is the systematic doppler shift of line center that can occur if the laser beam is not perfectly orthogonal .
in the recent work on absolute frequency measurements of the @xmath1 lines @xcite , we used the same spectrometer to measure the frequencies with laser beams traversing the atomic beam in opposite directions .
since the doppler shift is opposite for the two cases , their difference gives a measure of the misalignment from perpendicularity .
our measurements show that the doppler shift of the transition frequency is only 150 khz , corresponding to a misalignment angle of 0.1 mrad . for the hyperfine interval , this implies a negligible error of less than 1 hz . in order to measure the hyperfine intervals , both the unshifted laser beam and the aom - shifted beam
were sent across the atomic beam .
the diode laser and the aom were modulated at different frequencies , and the signal from the pmt was independently demodulated at the two frequencies .
the first error signal was used to lock the laser on a given hyperfine transition , while the second error signal was use to lock the aom to the hyperfine interval .
a stringent check on our error estimate in table [ t1 ] is to use our technique to measure the ground hyperfine interval .
as mentioned in the introduction , this interval is already known in all alkali atoms with @xmath45 relative precision .
however , our technique requires the interval to be accessible with an aom , and this is possible only in @xmath46li , where the intervals are about 200 mhz and 800 mhz respectively . among the different sources of error in this measurement considered before , the effect of collisions is negligible in an atomic beam .
we can also check for the intensity - dependent errors by varying the laser power .
the results of ground hyperfine measurements at three powers are shown in table [ t2 ] .
the beam size is such that the peak intensity ( at beam center ) for the highest power of 42 @xmath47w is 0.84 mw/@xmath48 , which is still smaller than the saturation intensity of 2.51 mw/@xmath48 .
the important thing to note from the table is that all the measurements lie within a few khz of each other , even when the laser power is increased by a factor of three .
this means that shifts due to radiation - pressure and optical - pumping effects are less than 5 khz , consistent with our earlier estimate .
note that increasing the laser power also increases the height of the peaks in the spectrum , their linewidth , and the overall signal - to - noise ratio .
the consistency of the values shows that there are no unknown systematic errors related to these parameters . to see
if there is any intensity - dependent trend in the data , we have plotted the measured values versus power in fig .
the linear fit has a small slope , and the extrapolated values of the intervals are @xmath37li , @xmath49 : = @xmath50 mhz .
@xmath39li , @xmath49 : = @xmath51 mhz . the quoted error of 6 khz is the total error obtained by adding in quadrature the different sources of error in table [ t1 ] .
all the values in the table lie within 6 khz of the extrapolated value for @xmath37li , and 10 khz for @xmath39li .
these values can be compared with the accurately known values @xcite : 228.2052611 mhz in @xmath37li and 803.5040866 mhz in @xmath39li .
our values are consistent within @xmath52 , confirming that our error estimate is reasonable .
we now turn to the measurement of hyperfine intervals in the @xmath43 state .
since there are two ground hyperfine levels , the same interval can be measured with transitions starting from either level .
as before , the measurements were done at three laser powers as listed in table [ t2 ] .
the measurements at the different powers again lie within a few khz of each other .
the measurements are plotted against power to yield extrapolated values of the intervals as @xmath37li , @xmath43 : = @xmath53 = @xmath54 mhz .
@xmath39li , @xmath43 : = @xmath55 = @xmath56 mhz .
the maximum deviation of the values in the table from the extrapolated values is only 8 khz ( @xmath57 ) . to the extent that different transitions are susceptible to different degrees of systematic error
, this again confirms that our error estimate is reasonable .
the value of the magnetic - dipole coupling constant @xmath14 in the two isotopes is compared to previous values in table [ t4 ] .
also listed is the recommended value from the review work of arimondo , inguscio , and violino @xcite .
the result in @xmath37li is completely consistent with previous measurements , as seen from the deviation plot in fig .
[ f2c ] , but the accuracy is improved by a factor of 5 . in @xmath39li ,
our value is slightly non - overlapping with the recommended value .
however , it is consistent with a recent measurement @xcite , which has the same accuracy as the recommended value , and an earlier less precise measurement @xcite .
our accuracy is almost an order of magnitude better .
the last column in the table lists the result of a recent calculation of @xmath14 in @xmath39li @xcite , which agrees with our value at the 1.7% level .
sodium has one stable isotope , @xmath58na , with @xmath40 . both the @xmath1 lines are near 589 nm , and are accessed using a cw ring - dye laser ( coherent 699 - 21 ) operating with rhodamine-6 g dye .
na spectroscopy is done in a 100-mm long vapor cell .
the cell is heated to a temperature of @xmath23c and has a magnetic shield around it .
the hyperfine levels in the @xmath59 state are far apart and the individual transitions in the @xmath18 line are well resolved .
hence we have used the standard sas technique for spectroscopy to lock the laser and the aom . as in the case of li ,
intensity - dependent errors are checked by repeating the measurements at three laser powers .
the average value of the interval is @xmath58na , @xmath59 : = @xmath55 = @xmath60 mhz .
the value of the hyperfine constant is compared to the recommended value and another published result @xcite in fig .
our value is consistent with these results but the accuracy is improved by a factor of 20 .
as mentioned earlier , the hyperfine levels in the @xmath61 state of @xmath58na are too close for the individual transitions in the @xmath2 line to be resolved .
hence we have used the technique of ccs to resolve these transitions .
the results of our measurements have been published previously @xcite .
the average values of the intervals are : @xmath58na , @xmath61 : = @xmath62 = @xmath63 mhz , + @xmath64 @xmath65 mhz . in fig .
[ f3b ] , we compare our values of the hyperfine constants to the recommended value and two subsequent measurements , one using time - resolved hyperfine quantum - beat spectroscopy @xcite and the second using polarization quantum - beat spectroscopy @xcite .
our values are consistent with the earlier values but the accuracy is improved by a factor of 5 .
potassium has two stable isotopes : @xmath66k with @xmath40 and natural abundance of 93.3% , and @xmath67k with @xmath40 and natural abundance of 6.7% .
the @xmath18 line is at 770 nm and the @xmath2 line is at 767 nm .
it is difficult to obtain reliable laser diodes at these wavelengths , hence our experiments were done sometimes with a frequency - stabilized diode and sometimes with a ring - cavity ti : sapphire laser ( coherent 899 - 21 ) similar to the ring dye laser used for the sodium experiments .
potassium spectroscopy was done in a 50-mm long vapor cell .
the cell was heated to a temperature of @xmath23c and had a magnetic shield around it . due to the low natural abundance of @xmath67k , it is difficult to get a good absorption signal for this isotope , hence we have measured hyperfine structure only in @xmath66k .
the hyperfine levels in the @xmath68 state are separated by about 58 mhz compared to the natural linewidth of 6 mhz .
thus the individual transitions in the @xmath18 line are fairly well resolved . in a first set of experiments
, we measured the interval using normal sas to lock the laser and the aom .
note that the crossover resonance is only 29 mhz away , which is not very far given that the experimentally obtained linewidth in sas is around 10 mhz .
the result for the interval from these measurements was @xmath69 mhz .
the slightly large error is due to the nearby crossover resonance which causes some peak pulling . in an earlier measurement from our laboratory @xcite
, we had measured the absolute frequencies of individual transitions in the @xmath18 line and hence determined the hyperfine interval indirectly .
the direct measurement of the interval was consistent with this earlier result .
in addition , it agreed with the recommended value in ref .
@xcite ( from a measurement by buck and rabi @xcite using the atomic beam magnetic resonance technique ) . however , there has since been a report of absolute frequency measurements on the same line using a femtosecond comb @xcite .
the value of the interval disagrees from our value ( by @xmath70 combined ) and the recommended value ( by @xmath71 combined ) , too large to be accounted for by any known sources of error .
we have therefore repeated the measurement of the interval but this time using the ccs technique . as mentioned earlier
, this technique does not produce crossover resonances and the measured interval is insensitive to detuning of the control laser . in fig .
[ f4a ] , we show a typical probe transmission spectrum in @xmath66k obtained with the probe locked to the @xmath72 transition and the control scanning across the @xmath73 transition
. the scanning is done by varying the frequency of the aom in the path of the control , so that the scan axis is properly calibrated and its linearity is guaranteed by the linearity of the voltage - controlled oscillator ( vco ) driving the aom .
note that direct scanning of the laser using intracavity elements usually results in a nonlinear scan .
the aom is double passed to ensure that the direction of the control beam remains unchanged while the aom is scanned , and the output beam intensity is stabilized by feedback control of the rf power exciting the aom .
the open circles in the figure represent the measured spectrum while the solid curves are fits .
we did two kinds of fitting to the spectrum .
the first fit is to a lorentzian lineshape with the peak center and linewidth as fit parameters .
the second fit is to a density - matrix calculation of the effective three - level system taking into account the thermal velocity distribution in hot vapor , as described in our earlier work @xcite .
the hyperfine interval ( which determines the control detuning at the peak ) and the control rabi frequency are free fit parameters .
the residuals from the two fits are also shown in the figure , with the lorentzian fit yielding slightly smaller residuals . however , the location of the peak center from both fits is the same . after fitting to several independent spectra ( to take care of drifts in the vco driving the aom ) ,
we obtain a value of 57.713(42 ) mhz for the interval , close to our earlier measurement and the one by buck and rabi .
the dashed curve shows what the spectrum would look like if the interval were 55.500 mhz as measured by falke _
et al . _
although the spectrum obtained by scanning the aom yields the interval , the locking technique has higher accuracy because the aom frequency is measured while it is locked , and is not susceptible to drifts of the vco . the new result for the interval by locking the aom is @xmath66k , @xmath68 : = @xmath55 = @xmath74 mhz , which is again consistent with the above measurements and our earlier published value . in fig
. [ f4b ] , we compare this value of @xmath14 with other published values . as can be seen
, there are two sets of values that are in complete disagreement with each other .
the recommended value and our two results are clustered on one side , while the recent result of falke _ et al_. @xcite and two older measurements ( which are both from the same group @xcite ) are clustered together .
the experiment of falke _
et al_.@xcite has also been done with great care and takes into account many potential sources of error , hence their measurement seems to be reliable within the stated accuracy .
we do not have a satisfactory explanation for such a large discrepancy .
as mentioned before , the hyperfine levels in the @xmath75 state of @xmath66k are too close to be resolved completely .
hence we have used our technique of ccs to get a partially - resolved spectrum . however , the peaks are still so close that the lock point will get pulled by the nearby transition . therefore , as described above for the @xmath18 spectrum , we measure the intervals by scanning the aom in the path of the control beam .
the aom is again double - passed and the output beam is intensity stabilized . a typical probe transmission spectrum of the @xmath2 line with the probe locked on the @xmath76 transition is shown in fig .
we saw above in the case of the @xmath18 line that the lineshape is described quite well by a lorentzian curve , hence we extract the locations of the individual hyperfine transitions by doing a multipeak fit to the spectrum . as seen from the figure ,
the featureless residuals are less than 1% of the peak .
the fit linewidth is 8.4 mhz , or only 45% larger than the natural linewidth of 5.8 mhz . after fitting to 25 spectra
we obtain the average values of the intervals as @xmath66k , @xmath75 : = @xmath62 = @xmath77 mhz , + @xmath64 @xmath78 mhz .
the slightly larger errors are due to the drift of the vco from shot to shot .
the difficulty in measuring hyperfine structure in this state is evidenced by the fact that there has only been one other measurement in recent times @xcite .
our values of the hyperfine constants @xmath14 and @xmath15 are compared to this measurement and the recommended values in fig .
the three sets are in complete agreement with each other , with our values having the smallest uncertainty .
note that the recent values in ref .
@xcite are from the same work which differed so significantly for the @xmath18 line .
rubidium has two stable isotopes : @xmath79rb with @xmath80 and natural abundance of 72.2% , and @xmath81rb with @xmath40 and natural abundance of 27.8% . the @xmath18 line at 795 nm and the @xmath2 line at 780 nm are both accessed using frequency - stabilized diode lasers .
rb spectroscopy is done in a 50-mm long vapor cell containing both isotopes .
the cell is at room temperature and has a magnetic shield around it .
the hyperfine levels in the @xmath82 state are far apart and the individual transitions in the @xmath18 line are well resolved .
therefore we have used the standard sas technique for spectroscopy .
the results of our measurements have been published before @xcite , with the average values in the two isotopes as : @xmath79rb , @xmath82 : = @xmath83 = @xmath84 mhz .
@xmath81rb , @xmath82 : = @xmath55 = @xmath85 mhz .
these values are compared to earlier values in fig.[f5a ] .
one of the reasons for our interest in rb was that the values from ref .
@xcite differed significantly from an earlier measurement from our laboratory @xcite . in both these cases ,
the absolute frequencies of the individual transitions were measured and the hyperfine interval determined from their difference .
by contrast , in the present work we measure the interval directly .
as can be seen from the figure , the present value agrees with our earlier indirect measurement , but disagrees with the one from ref .
note that the value for @xmath81rb from ref .
@xcite also disagrees with the recommended value .
the hyperfine levels in the @xmath86 state are far apart and the individual transitions in the @xmath2 line are well resolved .
therefore we have used the standard sas technique for spectroscopy . as usual , we check for intensity - dependent errors by repeating the measurements at several powers .
we have earlier published results in @xmath79rb using a similar technique to measure the intervals @xcite .
although those results were the most accurate then , our current work improves on that precision significantly .
in addition , we report results in @xmath81rb for the first time .
the average values for the intervals are : @xmath79rb , @xmath86 : = @xmath62 = @xmath87 = @xmath88 mhz .
@xmath79rb , @xmath86 : @xmath89 @xmath90 @xmath91 mhz , + @xmath62 @xmath87 @xmath88 mhz , + @xmath64 @xmath92 @xmath93 mhz .
@xmath81rb , @xmath86 : = @xmath62 = @xmath94 = @xmath95 mhz , + @xmath64 @xmath96 @xmath97 mhz , + @xmath98 @xmath99 @xmath100 mhz .
the hyperfine constants are compared to previous values in fig .
[ f5b ] . in @xmath79rb ,
these values are consistent with our earlier measurement and the recommended values .
for @xmath15 , there is a slight discrepancy from the value in ref .
@xcite , but this is the same work where there was a large discrepancy in the @xmath18 line as well . for @xmath81rb ,
the three recent high - precision measurements are in good agreement with each other .
the values in ref.@xcite have roughly the same accuracy as ours , and are obtained from absolute frequency measurements of various transitions in the @xmath2 line .
indeed , their measurement of the @xmath2 frequency with a precision of @xmath101 has enabled us to use this line as a frequency reference along with a ring - cavity resonator for absolute optical frequency measurements of other transitions @xcite .
cs has just one stable isotope , @xmath20cs with @xmath102 . the @xmath18 line at 895 nm and the @xmath2 line at 852 nm are both accessed using frequency - stabilized diode lasers .
cs spectroscopy is done in a 50-mm long vapor cell .
the cell is at room temperature and has a magnetic shield around it .
the hyperfine levels in the @xmath103 state are more than 1 ghz apart and the individual transitions in the @xmath18 line are well resolved .
therefore we have used the standard sas technique to lock the laser and the aom .
the hyperfine interval is actually larger than the doppler width , hence the spectrum has no crossover resonances .
the results have been published previously @xcite , with the average value of the hyperfine interval as @xmath20cs , @xmath103 : = @xmath104 = @xmath105 mhz . since the publication of our result
, there has been a report of another high - precision measurement of the interval in this state by gerginov _
et al_.@xcite . in that work
, the absolute frequencies of the transitions were measured using a femtosecond frequency comb with a stated precision of 4 khz .
our value is compared to this value and other published values in fig.[f6ab ] .
two of the previous values are also from measurements of the absolute frequencies , one using a frequency comb by hnsch and coworkers @xcite , and the other from our laboratory using a rb - stabilized ring - cavity resonator @xcite .
our value is consistent with both these measurements and an earlier less precise measurement in ref .
@xcite , and appears inconsistent only with the most recent measurement by gerginov _
et al_. @xcite .
the hyperfine levels in the @xmath106 state are far apart and the individual transitions in the @xmath2 line are well resolved .
however , the measurements on the @xmath2 line were done using our technique of ccs and not sas .
the advantages of the ccs technique for hyperfine measurements have been highlighted earlier . in cs , there is an additional advantage because , in normal sas , the hyperfine peaks can get distorted and even change sign due to the effects of optical pumping and radiation pressure @xcite . on the other hand , the lineshape in ccs remains symmetric even at high powers .
the measurements in cs have been published previously @xcite .
the large nuclear spin in @xmath20cs implies that , at sufficiently high precision , the intervals will show the effect of a small magnetic - octupole interaction ( characterized by the hyperfine constant @xmath107 ) .
the average values of the hyperfine intervals in terms of these constants are : @xmath20cs , @xmath106 : = @xmath108 = @xmath109 mhz , + @xmath110 = @xmath111 mhz , + @xmath112 = @xmath113 mhz .
one of the motivations for our work on the @xmath106 state of cs was that the published values of the hyperfine interval @xmath114 had changed from 251.000(20 ) mhz in the work of tanner and wieman @xcite to 251.092(2 ) mhz in the recent work of gerginov , derevianko , and tanner @xcite , a change of 92 khz ( @xmath115 ) .
this discrepancy is seen clearly in the deviation plot of the hyperfine constants in fig .
[ f6ab ] , comparing our values to these previous results .
our results are close to the earlier values @xcite , but disagree with the recent measurement by gerginov _
et al_. @xcite .
it is also from a measurement by gerginov _
et al_.@xcite that our interval in the @xmath18 line differs by 70 khz .
in conclusion , we have recently developed a technique for measuring hyperfine intervals using an aom whose frequency is directly locked to the frequency difference between two transitions .
we use techniques of high - resolution laser spectroscopy to resolve hyperfine transitions in the @xmath1 lines of all the alkali atoms . in lithium
, we study the fluorescence signal from a collimated atomic beam excited by a laser beam at right angles . in all other atoms ,
we study the absorption of a probe beam through a vapor cell , in the presence of either a counter - propagating pump beam ( saturated - absorption spectroscopy ) or a co - propagating control beam ( coherent - control spectroscopy ) .
we are thus able to measure hyperfine intervals in the first - excited @xmath16 and @xmath17 states of these atoms .
this yields the magnetic - dipole coupling constant @xmath14 and the electric - quadrupole coupling constant @xmath15 in these states . in the @xmath106 state of @xmath20cs
, there is evidence of a small magnetic - octupole coupling constant @xmath107 .
the measured hyperfine constants in the various atoms are summarized in table [ t5 ] . as discussed in earlier sections , in most cases our values
are consistent with other measurements but have considerably improved precision .
improved knowledge of the hyperfine constants should prove useful to both theorists and experimentalists working with alkali atoms .
the only state in which we have not measured hyperfine structure with higher precision is the @xmath42 state of li , where the level spacing is smaller than the natural linewidth .
in addition , we have previously used our rb - stabilized ring - cavity resonator to measure the absolute frequencies of the @xmath1 lines in all atoms except @xmath58na .
the hyperfine intervals determined from the differences in the transition frequencies match the intervals measured with the direct aom technique , though the precision is typically five times less for the absolute - frequency method .
the two techniques are very different in the way the frequencies are determined .
hence the consistency indicates that both techniques are reliable within their error bars . of the 17 constants listed in the table ,
there are only 2 values that are discrepant from recent measurements , namely in the @xmath16 states of @xmath66k and @xmath20cs . in both cases ,
the recent measurements use the frequency - comb technique to measure the absolute frequencies of the transitions .
however , in both cases there are other independent measurements ( albeit with slightly lower precision ) that agree with our results . in the case of @xmath66k
, there is an earlier atomic beam magnetic resonance measurement by buck and rabi @xcite , and in the case of @xmath20cs , there is a previous measurement by hnsch and coworkers also using the frequency - comb technique @xcite .
another common feature of the two frequency - comb measurements is that they use fluorescence spectroscopy from an atomic beam , while we use absorption spectroscopy in a vapor cell .
cell experiments have two advantages over the use of an atomic beam . 1 .
the line center is completely insensitive to the first - order doppler effect .
even if there were a small misalignment angle between the pump / control and probe beams , this would result in a broadening of the line and not a shift . in a beam experiment ,
any deviation from perpendicularity of the atomic beam from the laser beam would result in a systematic doppler shift .
the cell can be magnetically shielded much better than a beam .
our residual field with the two - layer shield is 1 mg , while the residual field in beam experiments is typically 10 times larger .
this could lead to correspondingly larger zeeman shifts .
this is why we have done all our experiments with vapor cells except in the case of lithium , for which it is difficult to make vapor cells because hot lithium vapor is highly reactive and attacks most glasses .
the frequency - comb technique is known to be highly reliable .
the aom technique is also inherently free of systematic errors at this level of precision since the rf frequency can be measured with @xmath116 khz accuracy .
this suggests that any discrepancy with frequency - comb measurements arises due to errors in the atomic spectrometers used , especially when the precision is being pushed to @xmath117 of the spectral linewidth . in other words , the spectral linewidth and the ability to split
the line are going to limit the precision and not the measurement technique _ per se_. the discrepancy in the interval for the @xmath103 state of @xmath20cs between our work and that of gerginov _
et al_. @xcite is about 70 khz , or 1% of the linewidth .
it is conceivable that one of the two spectrometers is systematically off by this fraction of the linewidth .
however , the discrepancy in the interval for the @xmath68 state of @xmath66k from the work of falke _
et al_. @xcite is 2 mhz , or 33% of the linewidth .
this is too large to be accounted for by any known problems with the spectrometer .
we have also done the spectroscopy with both the sas and the ccs techniques , and get consistent results .
further high - precision measurements with alternate techniques might shed light on these two cases .
we thank e. arimondo for useful discussions related to collisional shifts in hyperfine measurements .
we are grateful to w. a. van wijngaarden for help with the design of the lithium oven and to hema ramachandran for loan of a potassium vapor cell .
this work was supported by the department of science and technology and the board of research in nuclear sciences ( dae ) , government of india .
one of us ( d.d . ) acknowledges a graduate fellowship from the council of scientific and industrial research , india .
10 m. s. safronova , w. r. johnson , and a. derevianko .
relativistic many - body calculations of energy levels , hyperfine constants , electric - dipole matrix elements , and static polarizabilities for alkali - metal atoms . , 60(6):44764487 , dec 1999 .
s. falke , e. tiemann , c. lisdat , h. schnatz , and g. grosche .
transition frequencies of the d lines of @xmath66k , @xmath124k , and @xmath67k measured with a femtosecond laser frequency comb . , 74(3):032503 , 2006 .
f. touchard , p. guimbal , s. bttgenbach , r. klapisch , m. de saint simon , j. m. serre , c. thibault , h. t. duong , p. juncar , s. liberman , j. pinard , and j. l. vialle .
isotope shifts and hyperfine structure of @xmath127k by laser spectroscopy . , 108b:169171 , 1982 .
d. das , a. banerjee , s. barthwal , and v. natarajan .
a rubidium - stabilized ring - cavity resonator for optical frequency metrology : precise measurement of the d@xmath121 line in @xmath20cs .
, 38:545 , 2006 .
v. gerginov , k. calkins , c. e. tanner , j. j. mcferran , s. diddams , a. bartels , and l. hollberg .
optical frequency measurements of @xmath130 @xmath131 @xmath132 ( d@xmath121 ) transitions in @xmath20cs and their impact on the fine - structure constant . , 73(3):032504 , 2006 .
cccccc isotope & laser & aom & + & & & 15 @xmath47w & 28 @xmath47w & 42 @xmath47w + @xmath37li & @xmath137 & @xmath138 & 228.204 & 228.208 & 228.210 + & @xmath139 & @xmath140 & 228.207 & 228.211 & 228.205 + + @xmath39li & @xmath141 & @xmath142 & 803.506 & 803.514 & 803.511 + & @xmath143 & @xmath144 & 803.507 & 803.508 & 803.513 + cccccc isotope & laser & aom & + & & & 15 @xmath47w & 28 @xmath47w & 42 @xmath47w + @xmath37li & @xmath139 & @xmath137 & 26.091 & 26.094 & 26.099 + & @xmath140 & @xmath138 & 26.096 & 26.098 & 26.098 + + @xmath39li & @xmath143 & @xmath145 & 92.051 & 92.055 & 92.055 + & @xmath146 & @xmath142 & 92.048 & 92.052 & 92.054 + llc & reference + & & + 17.394(4 ) & 46.024(3 ) & this work + 17.386(31 ) & 46.010(25 ) & @xcite + 17.48(15 ) & 46.17(35 ) & @xcite + 17.375(18 ) & 45.914(25 ) & recommended @xcite + & 45.945 & theory @xcite + lccccc atom & state & @xmath14 ( mhz ) & @xmath15 ( mhz ) & & technique + @xmath37li & @xmath43 & 17.3940(40 ) & & & beam ( fs ) + + @xmath39li & @xmath43 & 46.0235(30 ) & & & beam ( fs ) + + @xmath58na & @xmath59 & 94.3485(50 ) & & & cell ( sas ) + & @xmath61 & 18.530(3 ) & 2.721(8 ) & & cell ( ccs ) + + @xmath66k & @xmath68 & 28.848(5 ) & & & cell ( ccs ) + & @xmath75 & 6.077(23 ) & 2.875(55 ) & & cell ( ccs ) + + @xmath79rb & @xmath82 & 120.645(5 ) & & & cell ( sas ) + & @xmath86 & 25.0403(11 ) & 26.0084(49 ) & & cell ( sas ) + + @xmath81rb & @xmath82 & 406.119(7 ) & & & cell ( sas ) + & @xmath86 & 84.7200(16 ) & 12.4970(35 ) & & cell ( sas ) + + @xmath20cs & @xmath103 & 291.9135(15 ) & & & cell ( sas ) + & @xmath106 & 50.28163(86 ) & @xmath147 & & cell ( ccs ) + | we have measured hyperfine structure in the first - excited @xmath0 state ( @xmath1 lines ) of all the naturally - occurring alkali atoms .
we use high - resolution laser spectroscopy to resolve hyperfine transitions , and measure intervals by locking the frequency shift produced by an acousto - optic modulator to the difference between two transitions . in most cases ,
the hyperfine coupling constants derived from our measurements improve previous values significantly . |
self - induced transparency is a dramatic manifestation of nonlinear effects , leading to coherent localized pulse propagation in an atomic medium @xcite .
the controlled , localized excitation and de - excitation of an ensemble of two level systems is effectively captured by the hyperbolic secant soliton pulse profile , which propagates as a stable soliton .
the maxwell - bloch equations aptly describe the pulse and averaged atomic population dynamics , which have been extensively investigated for multi - level systems @xcite . localized and nonlinear cnoidal wave solutions have been identified , as exact solutions @xcite , which has found experimental verification @xcite .
the subject of pulse propagation in nonlinear fibers has also been well - studied , since its prediction by hasegawa in 1974 @xcite . the non - linear schrdinger equation ( nlse ) , where the cubic nonlinearity arises due to kerr effect
, describes the pulse dynamics .
this integrable system in one dimension possesses soliton solutions @xcite , which can be dark @xcite , bright @xcite or grey @xcite , depending on the nature of nonlinearity and dispersion .
some of these solitons were experimentally observed , long after their prediction @xcite .
gross - pitaevskii equation in one - dimension , which captures the mean - field dynamics of bose einstein condensates ( bec ) , being identical to nlse , has also naturally evoked strong interest in solitons and their dynamics in cigar - shaped bec @xcite .
dark solitons in the repulsive regime and bright soliton and soliton trains in the attractive sector have been experimentally produced @xcite .
recently , the grey soliton , having complex profile has been produced through soliton collision in bec @xcite .
grey solitons have been the subject of some recent studies @xcite .
these are analogs of complex envelope bloch solitons in condensed matter systems , which can connect differently ordered domains , without passing through the normal phase .
this naturally makes it energetically favorable , as compared to the nel soliton , wherein the soliton profile , being of the hyperbolic tangent type , passes through the normal phase , at the vanishing point of the above profile .
it is interesting to note that the speed of the soliton can vary from zero to a maximum , depending on the depth of the grey soliton .
furthermore , the energy shows a maximum value as a function of momentum .
the doping of nonlinear fibers with suitable multilevel systems has opened up the exciting possibility of naturally combining the the atomic system with nlse , which can possibly lead to effective soliton control and manipulation .
recently , the localized solitons of a nonlinear fiber with a three level dopant has been investigated , wherein exact solitons of @xmath0 type have been identified @xcite .
the constraints of this coupled nonlinear system has led to strong restrictions on the pulse dynamics .
it was found that , the frequencies have to be perfectly matched for these solutions to exist @xcite . in this paper , we present exact grey soliton profiles in this system , which removes the above restriction on the pulse propagation . the complex profile s phase degree of freedom plays a crucial role in relaxing the constraint of frequency matching .
a three level @xmath1 atomic system whose transitions @xmath2 and @xmath3 are in near resonance with the two modes of the optical fiber .
the rabi frequencies of the two modes are @xmath4 and @xmath5 .
@xmath6 denotes the single photon detuning.,width=226 ] stable propagation of ( a ) bright and ( b ) dark solitons in a three level medium in presence of nonresonant nonlinearity and fiber dispersion .
this is in the regime of opposite signs of the group velocity dispersion at the two mode frequencies.,width=226 ] growth of instability for ( a ) bright and ( b ) dark solitons in a three level medium in presence of nonresonant nonlinearity and fiber dispersion .
this is in the regime of same signs of the group velocity dispersion at the two mode frequencies.,width=226 ] the paper is organized as follows : in the following , grey solitons , with complex envelope , are obtained as exact solutions of the coupled nonlinear system describing doped optical nonlinear fibers .
it is explicitly shown how the phase degree of freedom removes the restrictive equal frequency conditions on the @xmath0 paired pulses @xcite .
the cross - phase modulation is found to play a crucial role , in controlling the speed of the grey soliton . for the two level system , it is observed that grey soliton is strictly forbidden and only the sit soliton can propagate in this dynamical system .
subsequently , we present the numerical techniques and the results of numerical propagation of our solution , which clearly shows stability of the obtained solutions in a wide parameter range .
we then conclude with directions for future work .
we investigate the possibility of generalization of the solution obtained in @xcite for shape - preserving pulses in a kerr nonlinear two - mode optical fiber doped with 3-level @xmath1 atoms as shown in fig([fig1 ] ) . for the earlier obtained solutions ,
the frequency shift was assumed identical for the two possible fiber modes .
the modes of the fiber are near resonant with the transitions of the atomic system .
the two modes of the optical fiber may be described by the profile , @xmath7 @xmath8 being the slowly varying envelope , @xmath9 and @xmath10 , the carrier frequency and wave number respectively of the @xmath11 mode .
the time variation of the population of atomic levels are governed by optical bloch equations .
if @xmath12 denotes the probability amplitude of the @xmath11 atomic level ( fig 1 ) , then the following equations are obeyed within the rotating wave approximation @xmath13 the rabi frequencies 2 g and 2 g for the two field modes are related to the slowly varying amplitudes @xmath14 , according to the relations @xmath15 where @xmath16 is the transition dipole moment matrix element .
we take into account the effects of kerr nonlinearity ( proportional to the square of the electric field ) and dispersion , in order to truly describe the spatio - temporal variation of the light pulses through the fiber . using the approximation of slowly varying envelope
one can cast the nonlinear schrdinger equation , in rabi frequencies , as @xmath17 where the terms with @xmath18 are due to group velocity dispersion ( gvd ) , those with @xmath19 and @xmath20 represent the effect of kerr nonlinearity , and the coupling of the @xmath11 mode with the atomic system , respectively .
a dark - bright coupled soliton pair was reported in @xcite in this system .
there stable propagation of bright and dark solitons was reported , in the regime of opposite signs of the group velocity dispersion at the two mode frequencies .
correspondingly , unstable propagation was reported , when the gvds were of the same sign .
figs.([fig2 ] ) and ( [ fig3 ] ) show the solutions obtained by them in stable and unstable regimes of gvd .
motivated by the observed grey solitons in bec @xcite , we assume the following ansatz for the complex envelope solitons @xmath21e^{i(p_{1}z-\omega_{1}t)}\\ \label{field_amplitude2}g&=b\left[\cos(\theta)\tanh\left[\frac{(t - uz)}{\sigma}\cos(\theta)\right]+i\sin(\theta)\right]e^{i(p_{2}z-\omega_{2}t)}\\ \label{atom_amplitude1}\zeta_{1}&=\frac{i\alpha\cos(\theta)}{a\sigma}\mathrm{sech}\left[\frac{(t - uz)}{\sigma}\cos(\theta)\right]e^{i(p_{1}z-\omega_{1}t)}\\ \label{atom_amplitude2}\zeta_{2}&=-\frac{\alpha b\cos(\theta)}{a}\mathrm{sech}\left[\frac{(t - uz)}{\sigma}\cos(\theta)\right]e^{i(p_{1}-p_{2})z - i(\omega_{1}-\omega_{2})t}\\ \label{atom_amplitude3}\zeta_{3}&=\alpha\cos(\theta)\tanh\left[\frac{(t - uz)}{\sigma}\cos(\theta)\right]+(1-\alpha)\gamma\end{aligned}\ ] ] @xmath22 gives the envelope velocity in the moving frame . @xmath23
appearing on the rhs of ( [ field_amplitude1 ] ) gives the temporal width of the pulse .
we have removed the restriction @xmath24 .
the subsequent sections show that we have been able to derive analogous expressions as in @xcite for the unknown variables introduced in the ansatz .
the ansatz is put into the bloch and the nonlinear schrdinger equations ( 2 - 6 ) and the coefficients of @xmath25 , @xmath26 , @xmath25@xmath26 , etc .
are collected on both sides to yield the following relations .
the bloch equations ( 2 - 4 ) give @xmath27 the probabilities of occupation of the three levels should add up to one . imposing this condition
we get , @xmath28 the nonlinear schrdinger equations ( 5 - 6 ) yield @xmath29 equation ( [ conu2 ] ) can be alternately cast as , @xmath30 for stable propagation of pulses @xmath31 @xcite .
the following constraint can be obtained from equations ( [ conu1 ] ) and ( [ conu2 ] ) @xmath32 equations ( [ consistent equation_a ] ) , ( [ con1 ] ) and ( [ con2 ] ) lead to ( [ constraint2 ] ) , which is an important result that needs to be satisfied by the nonlinearity and dispersion in the medium at the two mode frequencies : @xmath33 another important point to note is that , for @xmath34 , in order to have a solution , @xmath35 is required .
this is because of the consistency relation ( [ consistent equation_b ] ) viz . , @xmath36 .
if @xmath37 , then @xmath38 and the arguments of all the sech and tanh functions in the ansatz ( [ field_amplitude1]-[atom_amplitude3 ] ) blow up to give constant solutions . therefore @xmath39 is required when @xmath34 .
thus in @xcite , while looking for pure dark soliton solution for g , the condition @xmath39 was automatically enforced . in comparison , in our solution ,
we fix @xmath40 to any value and find the rest of the parameters in terms of @xmath41 ( @xmath42 , @xmath37 ) . therefore our ansatz ( [ field_amplitude1]-[atom_amplitude3 ] ) represents valid solutions for all @xmath41 , given the consistency conditions are satisfied , showing that grey solitons of varying depths can be solutions of the system .
the velocity of the solitons ( u ) has a nontrivial dependence on @xmath41 .
this physically means that the velocity of the solitons is dictated by the depth of the grey soliton in the system .
fig ( [ fig4 ] ) shows that the solitons are velocity restricted with maximum value attained at @xmath43 .
grey solitons are usually velocity restricted .
often the relationship is as simple as @xmath44 . in our case
it is given by ( [ group velocity ] ) .
the parameters used are @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 and @xmath51 . ]
linear stability analysis is done by perturbing around the known solutions . perturbing around 5 complex waveforms results in 10
coupled differential equations in 10 variables with complicated potentials ( composed from the solutions themselves ) , which are difficult to handle either analytically or numerically .
hence , we have carried out numerical evolution of the pulse for the purpose of stability check . here
, we state the equations of motion of the perturbations , as derived from the equations of motion for @xmath52 , @xmath53 and @xmath54 ( 2 - 6 ) by substituting @xmath55 , @xmath56 , @xmath57 , @xmath58 and @xmath59 . the quantities @xmath60 , @xmath61 and @xmath62
are given by the rhs of equations ( [ field_amplitude1]-[atom_amplitude3 ] ) respectively , without the exponential terms .
the perturbations satisfy , @xmath63\left[\begin{array}{c } \delta \tilde{g}\\ \delta \tilde{g}\\ \delta\tilde{\zeta_1}\\ \delta\tilde{\zeta_2}\\ \delta\tilde{\zeta_3}\end{array}\right]=0 $ ] where the operators @xmath64 stand for : @xmath65 here , @xmath66 is the function that takes the complex conjugate of the term it acts on eg . , @xmath67 .
the above matrix equation is thus a condensed form of an actual set of 10 coupled partial differential equations in 10 variables .
spatio - temporal evolution of the grey - bright soliton pair in medium plotted against the spatial length of the medium at different propagation times .
the other parameters used are a=0.33361 , b=0.19261 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath47 , @xmath48 , @xmath51 .
@xmath75 , @xmath76 , @xmath77 , @xmath78 , and @xmath79.,width=302 ] amplitude and phase of the dark and grey solitons are plotted as a function of time at the exit of the medium .
all other parametrs are as in fig .
[ fig5].,width=264 ] spatio - temporal evolution of the grey - bright solitons using parameters such that the group velocity dispersion have the opposite sign .
we used a=0.29551 , b=0.25592,@xmath80 , @xmath81,@xmath82 , @xmath71 , @xmath83 , @xmath84 , @xmath46 , @xmath47 , @xmath48 , @xmath51 .
@xmath85 , @xmath86 , @xmath87 , @xmath78 , and @xmath79.,width=226 ] in this section , we investigate the propagation dynamics of the grey and bright solitons by integrating the full set of coupled nonlinear schdinger and bloch equations .
we use runge - kutta and split step operator methods to investigate the spatio - temporal evolution of the optical pulses in fiber doped with three level @xmath1 system .
the initial shapes of the input pulse and the initial atomic populations can be obtained in the limit of @xmath88 and @xmath89 from the analytical expressions ( [ field_amplitude1]-[field_amplitude2 ] ) and ( [ atom_amplitude1]-[atom_amplitude3 ] ) respectively .
we use grey and bright solitons with width @xmath90 psec in the present investigation .
parameters used in the numerical simulation obey self consistent relations ( [ constraint1 ] ) and ( [ constraint2 ] ) .
we first explore the stable propagation of the grey and bright solitons in presence of gvd with same character .
figure ( [ fig5 ] ) shows that the amplitudes of the output pulses are retaining their initial shape .
both solitons propagate through the medium with same group velocity . a small oscillation though has been noticed at both the ends of the grey soliton due to the modulation instability .
our analytic and numerical solution of grey - bright soliton pairs are presented in the inset of the figures ( [ fig5 ] ) at the propagation length @xmath91 m .
the numerical results obtained from the runge - kutta and split step operator methods match well with those obtained analytically .
next we study how the blackness parameter @xmath41 leads to stable propagation of soliton through the medium even though gvd has same sign
. a comparative study of the dark and grey solitons through the medium is shown in the figure [ fig6 ] .
the dark soliton can be obtained with @xmath92 whereas @xmath93 represents the grey soliton . for dark soliton
, the intensity of the dip always vanishes but the same for the grey soliton always remains .
the variation of phase with time with propagation length for both dark and grey solitons is seen to be antisymmetric in nature .
interestingly , dark soliton shows a sudden change in phase across the line center while grey soliton displays no change in shape of phase across the line center during the length of propagation . in the course of propagation , the antisymmetric phase has a temporal shift due to the presence of self and cross phase modulations of the medium in both the cases .
the dark soliton develops instability , both in intensity as well as in the phase while grey soliton propagates through the medium without loss of generality upto sufficient length of propagation .
hence the instability due to the same gvd sign can be compensated with the proper choice of the blackness parameter @xmath41 .
figures ( [ fig7]a)-([fig7]b ) show the results for the spatio - temporal evolution of grey and bright solitons in presence of the opposite group velocity dispersion .
both the grey and bright solitons maintain their initial shape .
in conclusion , grey solitons , analog of bloch solitons in condensed matter systems , have been identified as exact solutions of the coupled nonlinear schrdinger and bloch equations .
the complex nature of the profile played a crucial role in removing the restrictive frequency matching condition of @xmath0 paired solutions @xcite .
we numerically evolved our solution and found it to be stable in both regimes of group velocity dispersion .
we note that , for the case of coupling with a two - level system ( @xmath94 , @xmath95 ) instead of the @xmath1 system considered here , the pure @xmath26p and grey solitonic profiles are prohibited for g , while a pure sech profile is allowed .
the fact that our solutions are stable for some parameters in both regimes of group velocity dispersion signs , is very interesting and needs further investigation .
t. n. dey , s. dutta gupta and g. s. agarwal optics express * 16*,17441 ( 2008 ) .
s. l. mccall and e. l. hahn , phys . rev . * 183 * , 457 ( 1969 ) .
l. allen and j. h. eberly , `` optical resonance and two - level atoms '' ( dover publications , mineola , ny , 1987 ) . s. l. mccall and e. l. hahn , phys .
lett . * 18 * , 908 ( 1967 ) a. i. maimistov , a. m. bhasrov , s. o. elyutin , and m. y. sklyarov , phys . rep .
* 191 * , 1 ( 1990 )
. t. w. barnard , phys .
rev . a * 7 * , 373 ( 1973 ) .
f. t. arecchi , v. degiorgio , and s. g. someda , phys .
a * 27 * , 588 ( 1968 ) .
m. d. crisp , phys .
* 22 * , 820 ( 1969 ) .
p. k. panigrahi and g. s. agarwal , phys .
a * 67 * , 033817 ( 2003 ) f. t. hioe and r. grobe , phys .
lett . * 73 * , 2559 ( 1994 ) .
m. a. newbold and g. j. salamo , phys .
lett . * 42 * , 887 ( 1979 ) . j. l. shultz and g. j. salamo ,
lett . * 78 * , 855 ( 1997 ) .
t. s. raju and p. k. panigrahi , phys .
a * 81 * , 043820 ( 2010 ) . v.
e. zakharov and a. b. shabat , zh .
teoret . fiz . *
61 * , 118 ( 1971 ) = sov .
34 * , 62 ( 1972 ) . s. burger , k. bongs , s. dettmer , w. ertmer , k. sengstock , a. sanpera , g.v .
shlyapnikov and m. lewenstein , phys .
lett . * 83 * , 5198 ( 1999 ) .
j. denschlag , j.e .
simsarian , d.l .
feder , c.w .
clark , l.a .
collins , j. cubizolles , l. deng , e.w .
hagley , k. helmerson , w.p .
reinhardt , s.l .
rolston , b.i .
schneider and w.d .
phillips , science * 287 * , 97 ( 2000 ) .
l. khaykovich , f. schreck , g. ferrari , t. bourdel , j. cubizolles , l.d .
carr , y. castin and c. salomon , science * 296 * , 1290 ( 2002 ) .
k. e. strecker , g. b. partridge , a.g .
truscott and r.g .
hulet nature * 417 * , 150 ( 2002 ) .
u. al khawaja , h.t.c .
stoof , r.g .
hulet , k.e . strecker and g.b .
partridge , phys .
89 * , 200404 ( 2002 ) .
cornish , s.t .
thompson and c.e .
wieman , phys .
lett . * 96 * , 170401 ( 2006 ) .
i. shomroni , e. lahoud , s. levy and j. steinhauer , nature phys . * 5 * , 193 ( 2009 ) .
l. f. mollenauer , r. h. stolen , and j. p. gordon , phys .
* 45 * , 1095 ( 1980 ) .
a. d. jackson and g. m. kavoulakis , phys .
lett . * 89 * 070403 - 1 ( 2002 ) . c. j. pethick , h. smith , `` bose - einstein condensation in dilute gases '' , cambridge university press , cambridge , uk , ( 2002 ) .
a. khan , r. atre , p. k. panigrahi , arxiv:0903.4859v2 .
p. das , s. gangopadhyay and p. k. panigrahi , arxiv:1003.5745v1 .
j. h. eberly , quantum semiclass . opt .
* 7 * , 373 ( 1995 ) .
a. rahman and j. h. eberly , phys .
a * 58 * , r805 ( 1998 ) .
j. h. eberly and v. v. kozlov , phys .
* 88 * , 243604 - 1 ( 2002 ) .
e. lieb , phys . rev . * 130 * , 1616 , ( 1963 ) .
_ , theor .
* 28 * , 615 ( 1976 ) .
g. p. agarwal , nonlinear fiber optics " , 2nd ed .
( academic press , san diego ca 1995 ) . | grey solitons , with complex envelope , are obtained as exact solutions of the coupled system describing doped optical nonlinear fibers .
the phase degree of freedom plays a crucial role in removing the restrictive equal frequency conditions on the tanh - sech paired pulses , identified earlier @xcite .
the cross - phase modulation is found to control the speed of the grey soliton , which has an upper bound .
the coupled soliton solution obtained here was found to be stable in a wide parameter range . |
we first prove a lemma which may be of independent interest ; we refer to it as a ramsey stability result , since it states that if a graph @xmath3 is ramsey for the clique @xmath1 then we can remove any small number of vertices and the remaining graph will still possess a ramsey property almost as strong as the original .
[ lem : big ] let @xmath0 and @xmath32 .
let @xmath33 with @xmath34 and @xmath35 be any set with @xmath36 .
finally , let @xmath37 and @xmath38 be any vertices from @xmath39 .
then , in any colouring of the edges of @xmath3 , there exists a monochromatic copy of @xmath40 in @xmath41 , say with vertex set @xmath42 , such that either @xmath43 , or @xmath44 , or @xmath45 .
[ h ] 0.8 [ fig : lem1_red ] for lemma [ lem : big ] .
a black ring around a vertex class indicates that we colour edges between this class and @xmath46 blue .
a white ring around a vertex class indicates that we colour edges between this class and @xmath46 red .
no ring indicates that such edges retain their original colour .
edges inside @xmath47 are red , edges inside @xmath48 are blue , and edges inside @xmath49 retain their original colour.,title="fig:",scaledwidth=100.0% ] 0.8 for lemma [ lem : big ] .
a black ring around a vertex class indicates that we colour edges between this class and @xmath46 blue .
a white ring around a vertex class indicates that we colour edges between this class and @xmath46 red .
no ring indicates that such edges retain their original colour .
edges inside @xmath47 are red , edges inside @xmath48 are blue , and edges inside @xmath49 retain their original colour.,title="fig:",scaledwidth=100.0% ] without loss of generality , we may suppose that @xmath50 and @xmath51 .
we arbitrarily divide @xmath52 into two sets of @xmath53 vertices each , say @xmath48 and @xmath47 , and four single vertices , @xmath54 . for brevity , we let @xmath55 . to define a recolouring of the edges incident to @xmath56 ,
let us define an auxiliary graph @xmath57 with vertex set @xmath58 instead of giving an incomprehensible list of edges , we refer the reader to figure [ fig : lem : big ] ( a ) for the definition of @xmath57 .
let @xmath59 be the complement of @xmath57 , depicted in figure [ fig : lem : big ] ( b ) .
we now recolour the edges incident to @xmath56 as follows . if @xmath60 and @xmath61 such that @xmath62 , then colour the edge @xmath63 red if @xmath64 , and colour the edge @xmath63 blue otherwise .
furthermore , colour all edges in @xmath65 red , and all edges in @xmath66 blue .
the edges in @xmath67 retain their original colouring .
for all @xmath68 and all @xmath69 , colour the edge @xmath70 blue . for all @xmath71 and all @xmath69 , colour the edge @xmath70 red
. it will be convenient to call the vertices in @xmath72 _ blue vertices _ , and to call the vertices in @xmath73 _ red vertices_. the recolouring is indicated in figure [ fig : lem : big ] .
the crucial properties of this recolouring are the following , which are easy to verify from examining figure [ fig : lem : big ] : 1 .
both @xmath57 and @xmath59 are @xmath74-free .
every triangle in @xmath57 contains at least one of @xmath75 , and every triangle in @xmath59 contains at least one of @xmath76 .
the blue vertices @xmath77 are connected by only red edges in @xmath3 , and the red vertices @xmath78 are connected by only blue edges in @xmath3 . since @xmath3 is ramsey for @xmath1 there must be a monochromatic copy of @xmath1 present in @xmath3 after this recolouring .
we claim that , thanks to the fortuitous properties of our recolouring , this forces a monochromatic @xmath40 in the original colouring with the required properties .
let @xmath79 be the vertex set of the monochromatic @xmath1 present in @xmath3 after this recolouring .
if @xmath80 then the lemma follows immediately , since discarding at most one vertex would leave a monochromatic @xmath40 in the original colouring ( as the only edges which are recoloured are incident with @xmath81 ) , completely disjoint from @xmath81 as required .
we may suppose , therefore , that @xmath82 . the first case to consider is when @xmath83 . by property @xmath84 and since @xmath0
, @xmath79 must contain at least two vertices from one of the classes @xmath48 , @xmath47 , or @xmath49 .
suppose first that @xmath85 .
then @xmath79 must form a red @xmath1 , and hence can contain at most one vertex from @xmath48 , and no vertex from @xmath49 , since all edges between @xmath47 and @xmath49 are blue . by similar reasoning ,
if @xmath86 , then @xmath79 must form a blue @xmath1 , and hence can not use any vertex from @xmath87 . therefore
, there exists at most one class @xmath88 such that @xmath89 .
since each such class contains at most @xmath53 vertices , discarding all but one vertex of @xmath90 would force a monochromatic copy of @xmath74 within @xmath56 , using at most one vertex from each of @xmath48 , @xmath47 , and @xmath49 .
this would force a copy of @xmath74 in either @xmath57 or @xmath59 , which contradicts property @xmath84 .
assume now that @xmath91 , and suppose that @xmath79 hosts a red copy of @xmath1 .
since all blue vertices are connected to @xmath46 by blue edges , @xmath79 can not contain any blue vertices .
therefore , by property @xmath92 , @xmath79 uses vertices of at most two nodes in @xmath57 .
furthermore , since the copy is red , @xmath93 .
if @xmath94 then @xmath79 can use at most one vertex from @xmath95 , and discarding this vertex leaves a monochromatic @xmath40 in the original colouring , completely disjoint from @xmath81 , as required . if @xmath96 then , by property @xmath92 again , it must use exactly two vertices from @xmath97 . since there are only blue edges between vertices in @xmath98 , by property @xmath99 , at least one of these two vertices in @xmath100 must be @xmath37 or @xmath38 . discarding the other vertex in @xmath100 leaves a monochromatic copy of @xmath40 in the original colouring which intersects @xmath56 in either @xmath37 or @xmath38 , but no other vertices , as required .
the case when @xmath79 hosts a blue copy of @xmath1 is handled similarly , and the proof is complete .
the following corollary is immediate when @xmath0 ; in this weakened form it also holds for @xmath101
. we will not need corollary [ cor : small ] in the rest of this paper , but it is a pleasingly simple way to demonstrate the ramsey stability ethos of lemma [ lem : big ] .
[ cor : small ] let @xmath102 and @xmath32 .
if @xmath33 has @xmath103 then @xmath104 . for @xmath0
this follows immediately from lemma [ lem : big ] , after expanding @xmath56 by two arbitrary vertices from @xmath46 . for @xmath101
, it suffices to give an explicit colouring of @xmath74 in a similar fashion , as we do in figure [ ] .
[ h ] thus , if we recolour the edges adjacent to @xmath56 as indicated in figure [ ] , then any monochromatic @xmath20 in @xmath3 must have at least two vertices from @xmath46 , and hence @xmath105 as required .
we recall our goal : to show that @xmath1 is ramsey equivalent to @xmath2 for @xmath0 .
it is , of course , trivial that if @xmath106 then @xmath32 .
it remains to show that if @xmath32 then @xmath106 .
our strategy will be to accumulate more and more information about the monochromatic structures present in a colouring of a graph , ramsey for @xmath1 , without a monochromatic @xmath2 , until we are eventually able to obtain a contradiction .
[ lem - redandblue ] let @xmath107 .
if @xmath32 then , in every colouring of @xmath3 , there is either a monochromatic @xmath2 , or a red @xmath1 and a blue @xmath1 .
suppose , without loss of generality , that the edges of @xmath3 are coloured so that there is a red copy of @xmath1 .
let @xmath48 be the vertex set of this red @xmath1 .
as in the proof of lemma [ lem : big ] , we will recolour some edges of @xmath3 and use the assumption that @xmath32 to prove the claim .
suppose first that there is an edge @xmath108 of @xmath48 which has the property that every red @xmath1 intersects @xmath48 in at least one vertex besides @xmath109 and @xmath110 . in this case , we recolour every other edge of @xmath48 blue , and colour the edges between @xmath111 and @xmath112 red .
since @xmath32 there must be a monochromatic @xmath1 in this recoloured @xmath3 .
suppose first that there is a red @xmath1 .
if it uses at least @xmath113 vertices from @xmath112 then there is a red @xmath40 present in @xmath112 in the original colouring , and hence a red @xmath2 .
otherwise , it must use a red edge from @xmath48 .
but the only red edge remaining in @xmath48 is @xmath108 , and the edges from @xmath114 to @xmath112 retained their original colouring .
therefore , we must have a red @xmath1 in the original colouring that intersects @xmath48 in exactly @xmath114 , which contradicts our choice of @xmath108 .
secondly , suppose that there is a blue @xmath1 in the recoloured @xmath3 .
if it uses any of the new blue edges inside @xmath48 , then it must be contained entirely inside @xmath48 , since the edges from @xmath111 to @xmath112 are all red . however , this is impossible , since @xmath48 has @xmath108 still coloured red
. therefore we must have a blue @xmath1 that uses only edges which were originally blue , and so we have a red @xmath1 and a blue @xmath1 , as required .
[ htbp ] .,title="fig:",scaledwidth=90.0% ] we may now assume that , for every pair @xmath115 , there is another red @xmath1 intersecting @xmath48 in only the edge @xmath108 .
let @xmath116 be the vertex set of another red @xmath1 such that @xmath117 , say @xmath118 , and let @xmath119 be any two vertices in @xmath120 .
we recolour ( some of ) the edges incident to @xmath116 in the following way .
an illustration of this colouring can be found in figure [ fig : lemma3 ] .
* for all @xmath121 , all @xmath122 ( @xmath123 ) , and all @xmath124 , we colour the edge @xmath125 blue and the edge @xmath126 red ( if present in @xmath3 ) .
* for all @xmath127 , we recolour the edges @xmath128 and @xmath129 blue , and the edges @xmath130 and @xmath131 red ( if present in @xmath3 ) . * for all @xmath132 , we colour the edges @xmath133 , @xmath134 and @xmath135 in red ( the edge @xmath136 retains its original colour ) . * every edge in @xmath137 is recoloured blue , except for @xmath138 which remains red . again , since @xmath32 there must be a monochromatic @xmath1 in this recoloured @xmath3 .
suppose first that there is a red @xmath1 , say on vertex set @xmath42 .
if it uses at least @xmath113 vertices from @xmath139 then there is a red @xmath40 present in @xmath139 in the original colouring , and hence a red @xmath2 .
otherwise , it must use a red edge from @xmath116 .
but the only red edge remaining in @xmath116 is @xmath138
. then @xmath42 must be disjoint from @xmath140 , since each such @xmath133 is blue .
hence , @xmath141 . but none of the edges inside @xmath142 were recoloured , and hence @xmath142 hosts a red @xmath40 in the original colouring that is vertex disjoint from @xmath48 . secondly , suppose that there is a blue @xmath1 in the recoloured @xmath3 , say on vertex set @xmath42 .
if it uses any of the new blue edges inside @xmath143 , then it must be contained entirely inside @xmath143 , since the edges from @xmath144 to @xmath145 are all red .
however , @xmath143 does not host a blue @xmath1 in this recolouring
. therefore we must have a blue @xmath1 that uses only edges which were originally blue , and so we have a red @xmath1 and a blue @xmath1 , as required .
[ lem - nand1 ] let @xmath0 .
if @xmath32 then , in any colouring of @xmath3 , if there is a monochromatic @xmath146 then there is a monochromatic @xmath2 .
suppose that @xmath3 has , say , a red @xmath146 , on vertex set @xmath48 .
by lemma [ lem - redandblue ] , we may assume that there exists a blue @xmath1 , say on vertex set @xmath47 .
let @xmath147 , so that @xmath148 .
we now apply lemma [ lem : big ] , with @xmath35 being any set of @xmath149 vertices containing @xmath47 .
this yields a monochromatic @xmath40 which intersects the red @xmath146 in at most one vertex , and the blue @xmath1 not at all , and hence we must have a monochromatic @xmath2 . [ lem-2intersection ]
let @xmath0 , and let @xmath3 be a graph such that @xmath32 .
assume that there is a colouring of the edges of @xmath3 with no monochromatic copy of @xmath2 .
then , in this colouring , no two monochromatic copies of @xmath1 intersect in exactly two vertices .
suppose otherwise ; without loss of generality , we have two red copies of @xmath1 , say on vertex sets @xmath48 and @xmath150 , such that @xmath151 . by lemma [ lem - redandblue ]
we may further assume that there is a blue @xmath1 , say on vertex set @xmath47 .
assume first that @xmath152 .
let @xmath153 and @xmath154 ( which exist since @xmath0 and since @xmath47 intersects with @xmath48 and @xmath150 with at most one vertex each ) .
further , set @xmath155 and @xmath156 . by assumption , @xmath157 and @xmath36 .
therefore , by lemma [ lem : big ] , there is a monochromatic copy of @xmath40 , say on set @xmath42 , such that either @xmath158 , or @xmath159 . in the first case ,
when @xmath158 , then @xmath42 is disjoint from both @xmath47 and @xmath150 , and hence there is a monochromatic copy of @xmath2 , a contradiction .
otherwise , @xmath42 is disjoint from both @xmath47 and @xmath48 , and again , we find a monochromatic copy of @xmath160 , a contradiction .
we argue similarly if @xmath161 , and therefore assume from now on that @xmath162 .
let @xmath163 and @xmath164 be some arbitrarily chosen vertices .
we again apply lemma [ lem : big ] , with @xmath165 , where @xmath166 , and @xmath167 .
it is clear that @xmath157 and @xmath51 , as required .
suppose that there is a monochromatic copy of @xmath40 which intersects @xmath56 in only vertices of @xmath42 .
in particular , it is vertex - disjoint from @xmath168 , and hence it creates a monochromatic @xmath2 , which is a contradiction .
it follows that there exists a monochromatic copy of @xmath40 which intersects @xmath56 in either @xmath37 or @xmath38 , but no other vertices .
since it is disjoint from @xmath47 , we may assume that it is red .
if this red @xmath40 does not use @xmath169 , however , then together with @xmath150 we have a red @xmath160 , which is a contradiction .
therefore , either @xmath170 or @xmath171 is red .
since @xmath37 and @xmath38 were an arbitrary choice of two vertices from @xmath172 , it follows that all but at most one vertex of @xmath48 is connected to @xmath169 by a red edge .
that is , @xmath173 hosts two red copies of @xmath1 that intersect in @xmath113 vertices .
note that if @xmath173 forms in fact a red copy of @xmath146 , then we are done by lemma [ lem - nand1 ] .
therefore , to finish the argument , let @xmath174 such that the edge @xmath170 is blue or not present in @xmath3 . as noted
, there is at most one such @xmath37 .
we apply lemma [ lem : big ] yet again to reach a contradiction .
let @xmath175 , set @xmath167 and @xmath176
. then @xmath50 and @xmath177 . by lemma [ lem : big ]
, there exists a monochromatic copy of @xmath40 , say on vertex set @xmath42 , such that either @xmath178 , @xmath179 , or @xmath180 .
if @xmath178 , then @xmath42 is disjoint from @xmath181 and hence forms a monochromatic copy of @xmath2 in the original colouring , a contradiction . if @xmath179 , then @xmath42 is disjoint from @xmath47 , and hence we may assume that it is red .
but then , @xmath42 is either disjoint from @xmath48 and forms a red copy of @xmath2 , or @xmath182 , and hence the edge @xmath183 is red , a contradiction . finally ,
if @xmath180 , then @xmath42 together with @xmath184 forms a monochromatic copy of @xmath2 .
we will now conclude the proof of the main result .
let @xmath0 , and let @xmath3 be a graph such that @xmath32 .
assume that there exists a colouring of the edges of @xmath3 without a monochromatic copy of @xmath2 .
by lemma [ lem - redandblue ] , we can assume that there are two ( not necessarily disjoint ) sets @xmath48 and @xmath47 of vertices such that @xmath185 $ ] and @xmath186 $ ] form a red and a blue copy of @xmath1 , respectively . by assumption , any other red ( blue ) copy of @xmath1 intersects @xmath48 ( @xmath47 ) in at least two vertices ; in fact , by lemma [ lem-2intersection ] , any other red ( blue ) copy of @xmath1 intersects @xmath48 ( @xmath47 ) in at least three vertices .
that is , every set @xmath187 of size @xmath188 meets every red copy of @xmath1 in at least one vertex , and every set @xmath189 of size @xmath190 meets every blue copy of @xmath1 in at least one vertex .
if @xmath191 , fix two arbitrary subsets @xmath187 and @xmath189 , both of size @xmath192 . if @xmath193 , let @xmath194 be a set of size @xmath195 such that @xmath196 , and let @xmath197 be a subset of size @xmath53 such that @xmath198 ( note that @xmath199 ) . in both cases , the sets @xmath187 and @xmath189 are disjoint and , by the above discussion , any monochromatic copy of @xmath1 meets @xmath200 in at least one vertex
. we now recolour the graph and show that the resulting colouring does not contain a monochromatic copy of @xmath1 .
we may assume , without loss of generality , that all edges in @xmath201 are present , since losing edges will only help prevent a monochromatic @xmath1 occurring .
let @xmath202 and @xmath203 . * if @xmath204 and @xmath193 ( i.e. @xmath205 ) , colour one edge between @xmath116 and @xmath206 red , and the other one blue
otherwise , colour the edges between @xmath116 and @xmath206 so that for every @xmath207 there are @xmath208 such that @xmath209 is red and @xmath210 is blue , and for every @xmath211 there are @xmath212 such that @xmath209 is red and @xmath210 is blue .
, i.e. if @xmath191 or @xmath213 . ] * for all @xmath214 , @xmath215 , and @xmath216 , colour the edge @xmath217 blue and colour the edge @xmath170 red .
* for all @xmath218 , @xmath219 , and @xmath220 , colour the edge @xmath217 red and colour the edge @xmath170 blue .
this recolouring is illustrated in figure [ fig : proof ] ( where we label as black those edges which retain their original colouring ) .
[ htbp ] .,title="fig:",scaledwidth=50.0% ] note that we only recolour edges incident to @xmath200 . therefore , by our choice of @xmath200 , any monochromatic copy of @xmath1 ( after recolouring the edges ) must meet @xmath200 in at least one vertex .
suppose now that a red @xmath1 exists and uses vertices from @xmath116 but not @xmath206 .
then it must use just one vertex from @xmath116 and @xmath113 from @xmath221 , and hence we have a red @xmath2 in the original colouring .
if a blue @xmath1 exists and uses vertices from @xmath116 but not @xmath206 , then it can not use any vertices from @xmath222 or @xmath223 , and can only use at most one vertex from @xmath224 ( since the edge @xmath225 remains red as in the original colouring ) .
but this contradicts the fact that @xmath226 . similarly , we can rule out the case that a monochromatic copy of @xmath1 uses vertices from @xmath206 but not @xmath116 . therefore ,
if there is a monochromatic copy of @xmath1 after recolouring the edges , then it must use vertices from both @xmath116 and @xmath206 . assume first that this copy is red .
since all vertices in @xmath206 are connected to @xmath223 via blue edges , the red copy of @xmath1 must lie entirely inside @xmath201 .
but then , it can use at most one vertex from @xmath116 and at most one of @xmath222 .
the remaining @xmath195 vertices must come from @xmath206 , so we must use all vertices from @xmath206 .
however , in the case @xmath191 or @xmath213 , every vertex in @xmath116 sees at least one vertex of @xmath206 in blue . in the case
@xmath204 , @xmath227 and @xmath228 , the two edges between @xmath116 and @xmath206 are of opposite colour , and hence , at most one vertex of @xmath206 can contribute to a red @xmath74 .
a similar argument shows that we do not find a blue copy of @xmath1 using vertices from both @xmath116 and @xmath206 .
we have therefore constructed a colouring of @xmath3 which has no monochromatic @xmath1 , contradicting the original ramsey property of @xmath3 and concluding the proof .
in this section we give a proof of theorem [ carly+tibor ] , a result of szab and bodkin @xcite . we need to show that , if @xmath28 and @xmath229 , then @xmath230 .
let @xmath3 be a graph which is ramsey for @xmath20 and not ramsey for @xmath21 , and fix some colouring of @xmath3 with no monochromatic @xmath21 .
we first show that @xmath3 must possess both a red @xmath20 and a blue @xmath20 . without loss of generality
, there is a red @xmath20 , say on vertex set @xmath231 .
we now recolour the edges @xmath225 and @xmath232 blue , and colour all the edges from @xmath233 to @xmath112 red .
it is now straightforward that a blue copy of @xmath20 must be a blue copy in the original colouring , and that a red copy of @xmath20 forces either a monochromatic copy of @xmath21 in the original colouring , or it uses the edge @xmath234 and single new vertex , say @xmath235 . in this case , we recolour once again in the following way , as indicated in figure [ fig : k4 ] .
[ htbp ] we colour the three - edge path @xmath236 red , and the complement in @xmath237 blue .
furthermore , we colour all edges between @xmath238 and @xmath239 red . as before , if there is now a blue @xmath20 , then it can not use either of the vertices @xmath240 or @xmath241 , and hence it must have been already present in the original colouring of @xmath3 .
otherwise , a red @xmath20 must use exactly two vertices from @xmath242 .
in particular , we have a red @xmath243 that is either disjoint from @xmath244 or @xmath245 , and hence a red @xmath21 in the original colouring .
we have shown that there must be , in our coloured graph @xmath3 , a red @xmath20 , say on @xmath48 , and a blue @xmath20 , say on @xmath47 .
we now show that we can assume that @xmath48 and @xmath47 are disjoint .
suppose that our original choices are not , so that @xmath246 .
suppose @xmath247 and @xmath248 and @xmath249 . clearly , any edges between @xmath250 and @xmath223 must be red .
if their neighbourhoods intersect in @xmath223 we have found another red @xmath20 , entirely disjoint from @xmath47 , and we may proceed .
otherwise , we may assume that the neighbourhoods of @xmath240 and @xmath241 in @xmath223 are disjoint .
similarly , we can assume that the neighbourhoods of @xmath251 and @xmath252 in @xmath223 are disjoint .
we now colour the edges incident to @xmath201 as indicated in figure [ fig:5v ] .
the edges between @xmath250 and @xmath253 are red , the edges between @xmath254 and @xmath253 are blue.,title="fig:",scaledwidth=50.0% ] since @xmath28 , there must be a monochromatic copy of @xmath20 after recolouring .
furthermore , it must intersect @xmath201 in exactly two vertices , since the original colouring would contain a monochromatic @xmath21 otherwise .
if it is a red @xmath20 , say , then it must therefore use @xmath240 , @xmath241 , and a single vertex from @xmath223 , which contradicts the fact that their neighbourhoods are disjoint as discussed above , and we argue similarly if we have found a blue @xmath20 .
we may therefore assume that we have produced two disjoint sets , @xmath48 and @xmath47 , each of which spans a red and blue @xmath20 respectively .
suppose first that there are two vertex - disjoint edges missing from @xmath201 .
we then recolour the edges incident to @xmath201 as in figure [ fig : k62missing ] ( where , as usual , a red ( blue ) vertex represents the fact that the edges between that vertex and @xmath223 are coloured red ( blue ) ) .
[ htbp ] it is easy to check that this colouring of @xmath201 contains no monochromatic @xmath20 .
moreover , there are no blue edges between blue vertices , and , vice versa , no red edges between red vertices .
it follows that a monochromatic copy of @xmath20 in this recoloured @xmath3 must use at least two vertices from @xmath223 , which would create a monochromatic @xmath21 in the original colouring of @xmath3 , a contradiction .
we may suppose , therefore , that there is a vertex , without loss of generality say @xmath255 , such that every missing edge in @xmath201 is adjacent to @xmath233 .
furthermore , we may suppose that at least one edge is missing , or else we have a @xmath256 in @xmath3 as required .
let @xmath257 be some missing edge , where @xmath258 .
assume first that there is a vertex , say @xmath259 , in @xmath223 that has at least five neighbours in @xmath201 .
if it is adjacent to every vertex of @xmath260 then this creates a @xmath256 , as required .
hence , we can assume that @xmath261 is an edge in @xmath3 .
furthermore , all edges between @xmath259 and @xmath48 ( if present in @xmath3 ) must be red , and all edges between @xmath259 and @xmath47 must be blue ( as otherwise they create a monochromatic copy of @xmath21 in the original colouring ) .
suppose that @xmath259 is adjacent to every vertex of @xmath48 and to two vertices of @xmath47 , say @xmath109 and @xmath110 , and that the edge @xmath262 is missing .
if either of the edges @xmath263 or @xmath264 is missing , then by considering @xmath265 we have a similar situation as above namely , disjoint vertex sets of a red and a blue copy of @xmath20 with two vertex disjoint edges missing , and we are done .
otherwise , we have a @xmath256 in @xmath266 .
suppose now that @xmath259 is adjacent to every vertex of @xmath47 and @xmath233 and some other vertex of @xmath48 , say @xmath109 , and the edge @xmath267 is missing , where @xmath268 .
as above , we are now done by considering @xmath269 , since @xmath267 and @xmath270 are two independent edges missing . for the remainder of the argument
, we may therefore assume that every vertex of @xmath223 has at most four neighbours in @xmath201 .
we now describe a recolouring of the edges incident to @xmath201 such that there is no monochromatic @xmath20 that uses at least two vertices from @xmath201 .
recolour the interior edges of @xmath201 as in figure [ fig : k6missing ] .
[ htbp ] let now @xmath271 and let @xmath272 be any set of four vertices containing @xmath273 .
then , either @xmath84 , @xmath274 and we see a red copy of the three - edge path @xmath275 and a blue copy of @xmath275 in @xmath276 , or @xmath92 , @xmath277 , say @xmath278 , for some @xmath279 , and we see a monochromatic copy of @xmath280 or a monochromatic star @xmath281 with @xmath109 being the centre of the star . in case @xmath84 ,
say @xmath282 forms the red @xmath275 in @xmath276 , then we colour the edges @xmath267 and @xmath262 blue , and the edges @xmath283 and @xmath284 red ( if present in @xmath3 ) . in case @xmath92 , we colour the edge @xmath283 , @xmath261 and @xmath285 the opposite colour of @xmath286 and @xmath267 the same colour as @xmath286 .
note first that the colouring of @xmath201 does not contain a monochromatic triangle .
furthermore , it is evident that we do not create a monochromatic triangle on vertices @xmath287 with @xmath288 and @xmath289 , since no such @xmath259 sees both vertices of a red edge in red nor both vertices of a blue edge in blue .
however , since @xmath28 , there must be an edge @xmath290 with @xmath291 , which creates a monochromatic @xmath21 in the original colouring , a contradiction .
we have shown that @xmath1 and @xmath2 are ramsey equivalent for @xmath0 .
furthermore , we have seen that @xmath256 is the only obstruction to the ramsey equivalence of @xmath20 and @xmath21 , i.e. any graph @xmath3 that satisfies @xmath28 and @xmath292 must contain @xmath256 as a subgraph .
the only pairs of graphs @xmath293 known to be ramsey equivalent are of the form @xmath294 and @xmath295 , where @xmath296 is a graph of clique number less than @xmath25 .
furthermore , it is known ( @xcite and @xcite ) that the only connected graph that is ramsey equivalent to @xmath1 is the clique @xmath1 itself .
it is an open question , first posed in @xcite , whether there are two connected non - isomorphic graphs @xmath6 and @xmath7 that are ramsey equivalent .
it follows from @xcite that , if such a pair exist , they must have the same clique number .
in @xcite it is shown that they must also have the same chromatic number , under the assumption that one of the two graphs satisfies an additional property , called _ clique - splittability_. to tackle problems on ramsey equivalence , a weaker concept was proposed by szab @xcite .
we will first introduce some necessary notation .
we say that @xmath3 is ramsey minimal for @xmath4 if @xmath3 is ramsey for @xmath4 and no proper subgraph of @xmath3 is ramsey for @xmath4 .
denote by @xmath297 the set of all graphs which are ramsey minimal for @xmath4 , and by @xmath298 the set all graphs which are ramsey for @xmath4 .
finally , let @xmath299 be the class of graphs @xmath3 that are ramsey minimal for @xmath6 , but which are not ramsey for @xmath7 , or vice versa .
equivalently , @xmath300 is the set of minimal obstructions to the ramsey equivalence of @xmath6 and @xmath7 .
in particular , @xmath6 and @xmath7 are ramsey equivalent if and only if @xmath301 .
we say that @xmath6 and @xmath7 are ramsey close , denoted by @xmath302 , if @xmath300 is finite .
we stress that this is not an equivalence relation : reflexivity and symmetry are trivial , but transitivity does not hold , since every graph containing at least one edge is close to @xmath243 .
two graphs may be ramsey close in a rather trivial sense if @xmath303 and @xmath304 are both finite , or if @xmath305 and @xmath304 is finite .
graphs such that @xmath306 is finite are known as ramsey - finite graphs .
the class of ramsey - finite graphs has been studied quite intensively ; see , for example , @xcite for some results and further references .
in particular , it has been shown that the only ramsey - finite graphs are disjoint unions of stars . if one wishes to prove that two graphs are ramsey equivalent , a possible first step is to show that the two graphs are ramsey close .
szab @xcite has posed the following weaker version of the open problem mentioned earlier .
neetil and rdl @xcite proved that if @xmath307 then there exist infinitely many ramsey - minimal graphs @xmath308 such that @xmath309 . in particular , it follows that if @xmath310 and @xmath311 , and @xmath312 , then @xmath313 .
theorem [ carly+tibor ] states that , although @xmath20 and @xmath21 are not ramsey equivalent , they are ramsey close .
indeed , the only graph @xmath3 that is ramsey minimal for @xmath20 and satisfies @xmath314 is @xmath256 itself .
this is the only example of a pair of ramsey - infinite graphs which are ramsey close but not ramsey equivalent that we know of . in this case , @xmath315 .
we pose the following . | we prove that , for @xmath0 , the graphs @xmath1 and @xmath2 are ramsey equivalent .
that is , if @xmath3 is such that any red - blue colouring of its edges creates a monochromatic @xmath1 then it must also possess a monochromatic @xmath2 .
this resolves a conjecture of szab , zumstein , and zrcher @xcite .
a finite graph @xmath3 is ramsey for another finite graph @xmath4 , written @xmath5 , if there is a monochromatic copy of @xmath4 in every two - colouring of the edges of @xmath3 .
we say that @xmath6 and @xmath7 are ramsey equivalent , written @xmath8 , if , for any graph @xmath3 , we have @xmath9 if and only if @xmath10 .
the concept of ramsey equivalence was first introduced by szab , zumstein , and zrcher @xcite .
a fundamental question to ask is which graphs are ramsey equivalent to the complete graph @xmath1 .
it follows from a theorem of folkman @xcite that if a graph @xmath4 is ramsey equivalent to @xmath1 , then @xmath11 , where @xmath12 denotes the size of the largest complete subgraph of @xmath4 . in a recent paper , fox , grinshpun ,
person , szab , and the second author @xcite showed that @xmath1 is _ not _ ramsey equivalent to any connected graph containing @xmath1 .
furthermore , it is easily seen that @xmath1 is not ramsey equivalent to @xmath13 the graph composed of two vertex - disjoint copies of @xmath6 and @xmath7 . more generally , we denote by @xmath14 the graph that consists of a copy of @xmath6 and @xmath15 pairwise vertex - disjoint copies of @xmath7 . ] , see e.g. @xcite .
it follows that if @xmath16 then @xmath4 is of the form @xmath17 where @xmath18 .
it was shown in @xcite that @xmath19 . however , the question of whether @xmath1 is ramsey equivalent to @xmath2 was left open .
it is easily checked that @xmath20 is not ramsey equivalent to @xmath21 , since @xmath22 but @xmath23 . in @xcite
it is conjectured that this is an aberration , and that @xmath24 for large enough @xmath25 , a conjecture repeated in @xcite . in this paper , we prove this conjecture .
[ maintheorem ] for any @xmath0 @xmath26 it is shown in @xcite that this is best possible , in the sense that @xmath1 is not ramsey equivalent to @xmath27 .
our methods are combinatorial and explicit , and the idea is the following : suppose we have a graph @xmath3 which is ramsey for @xmath1 , and yet has been coloured so as to avoid a monochromatic @xmath2 .
we will then attempt , by giving an explicit recolouring of some edges , to give a colouring which no longer possesses a monochromatic @xmath1 , which contradicts the ramsey property of @xmath3 .
sadly , this is not quite possible directly , and instead we will build up our proof in stages : in each lemma , we will show that either a colouring of @xmath3 must have a monochromatic @xmath2 , or if not we can deduce some further structural information about the colouring of @xmath3 , which will help us in the following lemmas .
eventually , we will have accumulated enough information about our supposed counterexample so that it collapses under the weight of contradiction into non - existence , which proves theorem [ maintheorem ] . as mentioned above
, the clique on six vertices is an unfortunate obstruction which prevents the ramsey equivalence of @xmath20 and @xmath21 .
interestingly , bodkin and szab @xcite have shown that , essentially , this is the _ only _ such obstruction
. [ carly+tibor ] if @xmath28 and @xmath29 then @xmath30 . in section [ sec - triangle ]
, we give an alternative proof of this theorem , using similar techniques to those developed for the proof of theorem [ maintheorem ] .
* notation . *
all graphs are simple and finite . as a convenient abuse of notation
, we write @xmath3 both for a graph and for its set of vertices .
we write @xmath31 for the set of edges of @xmath3 .
* structure of the paper * in section [ sec - stability ] we prove a ramsey stability lemma , crucial for the proof of theorem [ maintheorem ] , but which is also of independent interest . in section [ sec - proof ]
we give the proof of theorem [ maintheorem ] . in section [ sec - triangle ]
we give the proof of theorem [ carly+tibor ] .
finally , we conclude by giving further discussion of ramsey equivalence , including a discussion of some still - open conjectures in this field , and adding some more . |
the discovery of intense gamma - ray emission from blazars had a strong impact on our understanding of relativistic jets .
independent evidence of relativistic beaming was provided by the condition that the source should be transparent to high energy gamma - rays ( mattox et al .
1993 , von montigny et al .
1995 , dondi & ghisellini 1995 ) .
the gamma - ray fluxes and spectra , together with data at lower frequency allowed detailed estimates of the physical parameters of different models ( maraschi , ghisellini and celotti 1992 ; blandford 1993 ; sikora , begelman and rees 1994 ; dermer and schlickeiser 1993 ) .
more recently the discovery of tev emission from some objects ( mrk 421 , mrk 501 and 1es 2344 + 514 ) and the detection of very fast variability at these high energies motivated new theoretical and observational studies ( e.g. celotti , fabian and rees 1998 ) the objects detected at tev energies have similar spectral energy distributions ( sed ) from radio to x - rays .
they emit synchrotron radiation with power peaking in the x - ray band . for mrk 421 and mrk 501
the x - ray and tev radiation vary in a correlated fashion ( macomb et al .
1995 , buckley et al .
1996 , catanese et al .
1997 , schubnell 1997 ) suggesting that the tev emission derives from the same high energy electrons responsible for the synchrotron x - rays . for mrk 421
the tev flux has been observed to vary with a 20 min timescale ( gaidos et al .
1996 ) , which strongly limits the size of the emitting region .
furthermore a lag has been measured between soft and medium energy x - ray photons .
a lag of the same order between soft and medium x - ray photons has been measured also in pks 2155 - 304 , whose spectral energy distribution closely resembles that of mrk 421 up to the x - ray range , but has not yet been observed at tev energies due to its southern declination .
the lags strongly suggest a radiative origin yielding a new important constraint on the physical parameters of the emitting region . for mrk 501 ,
similar in spectral shape to the previous two , an extraordinary flaring behaviour has been observed recently ( pian et al .
1998 ) .
the simplest model proposed for tev blazars involves a single zone and a single population of relativistic electrons emitting synchrotron radiation from radio to x - rays and inverse compton radiation from x - rays to gamma - rays . for this class of sources
the seed photons for the inverse compton process are likely to be the synchrotron photons themselves ( ssc ) ( see e.g. ghisellini & maraschi 1996 , mastichiadis & kirk 1997 ) , although other models have also been considered ( for a review see sikora 1997 ) .
we will limit ourselves to the ssc model in the following .
our scope is to provide general analytic expressions for all the constraints that can be envisaged within this simple model and discuss the consequent restrictions in the parameter space for the above mentioned sources individually and as a group .
a similar approach was initiated by ghisellini , maraschi and dondi ( 1996 ) ( hereafter gmd ) , however it did not include all the conditions and was limited to regimes well below the klein nishina limit , while for tev blazars the klein nishina corrections are important .
work along these lines has been developed by dermer , sturner and schlickeiser ( 1997 ) and by bednarek and protheroe ( 1997 ) .
the latter authors chose to use precise formulas requiring however numerical computations , with loss of generality .
the paper is organized as follows . in section 2
we discuss the constraints that can be put on the homogeneous ssc model using all the available observations . in section 3 the relevant formulae
are derived which link model parameters to observative quantities .
the effects of the klein nishina limit are discussed in detail and revised analytic formulae are given valid also in the latter regime . in section 4
we apply our diagnostics to the case of three well known blazars and finally ( section 5 ) we discuss the results .
the homogeneous ssc model assumes that radiation is produced in a single zone of the jet ( approximated here for simplicity as a sphere with radius @xmath5 ) , relativistically moving at small angle @xmath6 to the observer s line of sight .
photons up to the x - ray range are produced by relativistic electrons through the synchrotron process and are subsequently inverse compton scattered by the same electrons to energies in the @xmath7-ray range .
the observed radiation will be strongly affected by relativistic effects .
the key parameter is the doppler factor @xmath8^{-1}$ ] , where @xmath9 is the bulk lorentz factor and @xmath10 .
the observed sed of tev emitting blazars ( in a @xmath11 plot ) is characterized by two broad peaks , the first approximately in the x - ray band , the second in the 100 gev-1 tev band . below and above the peaks
the spectrum is quite smooth and can be approximated with power - law profiles with indices @xmath12 respectively where @xmath13 and @xmath14 ( see e.g. gmd ) .
note that within the present uncertainties the `` left '' side of both peaks can be described by the same spectral index .
note also that the spectrum in the tev range is poorly known and may be affected by intergalactic absorption ( see e.g. stecker and de jager 1997 and reference therein ) . in figure ( [ prima ] ) we show the observed sed of mrk 421 at two different epochs ( macomb et al .
1995 ) .
the observed spectral shape requires that the the relativistic electron spectrum steepens with increasing energy .
we approximate this behaviour with a broken power law ( gmd ) with indices @xmath15 , respectively below and above the `` break''energy @xmath16 : @xmath17 with these approximations we can completely specify the model using 7 parameters : the magnetic field intensity @xmath18 , @xmath5 , @xmath19 , the slopes @xmath20 and @xmath21 , the lorentz factor of the electrons at the `` break '' , @xmath22 , and the electron density parameter @xmath23 .
the peak synchrotron power is emitted by electrons with the break energy @xmath24 .
the maximum energy @xmath25 attained by electrons as well as a possible lower limit @xmath26 are unimportant here provided that @xmath27 and @xmath28 spectra computed with the above assumptions are compared with the mkn 421 data in fig .
the theoretical curves show that the double power - law approximation can reproduce reasonably well the shape of the observed sed .
the available data on the sed can be used to derive 6 `` observable '' quantities of particular relevance for the above model , namely the indices @xmath29 and @xmath30 , the frequency of the synchrotron and inverse compton peaks , @xmath31 and @xmath32 , and the peak luminosities @xmath33 and @xmath34 .
these quantities are rather uncertain being ideally the result of a fitting procedure on simultaneous multifrequency data , but in practice often inferred from observations obtained at different epochs and with incomplete frequency coverage .
the best sampled spectral region is usually the optical to x - ray range , therefore @xmath29 , @xmath30 , @xmath31 and @xmath35 are better determined than @xmath36 and @xmath34 .
1 can serve as an illustration of the uncertainties in the determination of the `` observables '' for mrk 421 , perhaps the object with the best overall data .
we recall that the one - zone homogeneous model is selfabsorbed at radio frequencies and can not explain the radio emission which implies further contributions from the outer regions of the jet .
it is important that , at least in principle , one can derive from the broad band spectral observations 6 quantities that have to be reproduced with the 7 parameter model .
one more observable quantity is sufficient to close the system .
this can be provided by the minimum timescale of variation , @xmath37 , which can be directly connected to the source dimension , @xmath5 , through the causality relation @xmath38 .
this relation gives only an upper limit on the dimension of the source , but a small dimension implies a high doppler factor if the transparency condition is taken into account ( see below ) . in order not to force @xmath39 to be too large we will adopt limits to the radius of the emission region in the range @xmath40- @xmath41 . at this point
the system is closed and any additional constraint that can be satisfied should be regarded as supporting the model .
these are : * the transparency of the source to @xmath7-rays .
+ the high energy photons may interact with the low energy ones producing pairs .
the fact that we observe the tev emission poses a strong limit to the optical depth of the source to photon photon interaction and therefore to the energy density of the soft radiation . from this limit
one can obtain a lower limit on the value of the doppler factor ( see e.g. dondi & ghisellini 1995 ) . *
the radiative interpretation of the time lags in the light curves at different frequencies .
+ recently , observations of mrk 421 and pks 2155 - 304 with asca ( 0.3 - 10 kev ) have revealed that when pronounced flares occur , the soft photons lag the harder ones ( takahashi et al .
1996 , urry et al .
the simplest way to explain the time lags in a homogeneous model is through the hypothesis that they are associated to the time necessary for freshly injected high energy electrons to cool . with this assumption the measured time lags determine the cooling time setting a well defined condition on the model .
* the consistency of the break energy of the electrons with the condition of balance between cooling and escape .
+ in our discussion the electron energy distribution is not calculated self - consistently ( for an example of this alternative approach see e.g. inoue and takahara 1996 , mastichiadis and kirk 1997 , ghisellini et al .
the two spectral indices and the break energy are determined from the observed spectra . having determined at the same time the dimension of the emission region and the magnetic and radiative energy densities we can verify a posteriori whether the cooling time of the electrons at the break is of the order of the escape ( expansion ) time from the emission region .
we will assume that consistency is achieved if the required escape / expansion velocity is in the range @xmath42 .
we recall however that in the standard hypothesis of a power law injection and subsequent cooling and escape in a homogeneous region , a spectral change @xmath43 is expected , while in general for the sources discussed below the observed steepening is larger , suggesting a more complex situation .
+ we approximate the cooling time as the minimum between the synchrotron and the compton cooling times . in the @xmath18@xmath19 parameter space
the regions of validity of these two approximations are separated by the line expressing equipartition between the magnetic and radiation energy density . in each of these regions
a value of the cooling time defines only one line .
bednarek & protheroe ( 1997 ) discuss a similar constraint on the cooling time of the electrons , but they incorrectly apply the condition to both the synchrotron and the compton cooling times _ simultaneously _ obtaining two independent constraints .
summarizing , the ssc homogeneous model needs 7 independent parameters : the spectral observations give 6 independent quantities , and the seventh observational information used to close the system is the minimum variability timescale .
other constraints are provided by the optical depth for @xmath7-@xmath7 interaction , by the observed time lags between soft and hard x - ray photons and by the consistency of the value of @xmath22 .
these last conditions are additional independent constraints : if verified , these constraints argue for the physical consistency of a homogeneous ssc model . +
in this section we derive the fundamental relations between the model parameters and the observable quantities discussed above . we chose to do this analytically at the cost of some inevitable approximation in order to maintain generality .
most of these formulae are also discussed in gmd and dermer et al .
( 1997 ) .
we assume that the two spectral indices can be determined directly from the data , independently of other quantities .
the structure of the relations is then such that we can eliminate 3 of the remaining 5 parameters ( i.e. @xmath44 , k and r ) and express the constraints as allowed regions in the b-@xmath39 parameter space . in the scheme that we consider the dominant synchrotron power
is emitted by electrons at the break ( those with lorentz factor @xmath22 ) .
primed quantities refer to the blob frame so that the observed frequency is given by @xmath45 .
the photons at @xmath31 dominate the energy density of the target photons and the inverse compton power in the thomson limit will be dominated by photons at frequency @xmath46 , produced in the scattering between the electrons at the break and the photons with frequency @xmath47 ( see also gmd ) . inverting the last equation
one obtains : @xmath48 therefore the ratio between the peak frequencies gives directly the value of @xmath22 .
+ the synchrotron frequency averaged over the spectral shape for an electron of lorentz factor @xmath44 is : @xmath49 using the value of @xmath22 given by eq .
( [ gb ] ) , gives : @xmath50 the last formula shows that for fixed @xmath31 and @xmath32 , @xmath18 and @xmath19 are inversely proportional .
we will see in the next section that in the kn limit this situation is reversed .
another relation can be obtained from the ratio of the total luminosity of the synchrotron peak and the total luminosity of the self - compton peak , which is directly related to the ratio between the radiation and the magnetic field energy densities inside the source ( in the comoving frame ) :
@xmath51 for our aims it is useful to express the total luminosity through the peak luminosity , which is more accessible to observations .
a simple calculation shows that the total luminosities ( @xmath52 and @xmath53 ) and the peak luminosities [ @xmath33 and @xmath54 ) ] are related by : @xmath55 where @xmath56 is given by @xmath57 expressing @xmath58 as : @xmath59 we can rewrite eq .
( [ natan ] ) as : @xmath60 if we express the source dimension @xmath5 with the variability timescale @xmath61 we can finally write : @xmath62^{1/2 } \label{dodici}\ ] ] this formula relates the magnetic field intensity and the doppler factor to observable quantities .
( [ dodici ] ) is an inequality because of the use of the inequality eq .
( [ undici ] ) ; taking @xmath63 as a lower limit on the radius of the source , as discussed in section 2 , eq .
( [ dodici ] ) will be represented in the @xmath18 , @xmath19 space by a strip of finite width .
thus the intersection of eq .
( [ dodici ] ) with eq .
( [ cinque ] ) allows to estimate @xmath18 and @xmath39 .
the adopted spectral shape for the electrons ensures that the synchrotron spectrum has a well defined peak ( break ) ( as observed ) .
thus both , electrons and photons , have a low energy branch and a high energy branch ( below and above the break ) . with the adopted approximations one can compute the inverse compton spectrum as the sum of 4 separate integrals : @xmath64 and @xmath65 , @xmath66 , representing the contributions from electrons of the low ( 1 ) and high ( 2 ) energy branches scattering photons of the low and high energy branches respectively .
full expressions are given in the appendix .
this subdivision is very useful in the discussion of the klein nishina limit .
the different contributions , computed using the well known @xmath19-function approximation for the single electron emission and a step function for the energy dependence of the cross section , @xmath67 for @xmath68 and @xmath69 otherwise ( see coppi and blandford 1990 for a critical discussion ) , are shown in fig .
all the spectra are computed in the source frame for fixed values of @xmath70 hz and varying @xmath22 from @xmath71 to @xmath72 , that is for increasing importance of the kn limit .
since the assumed particle spectra have no high energy cut offs , at high enough energy , the kn limit is always relevant and steepens the spectrum .
the limiting frequency can be estimated considering that the largest contribution to the radiation energy density is due to photons of frequency @xmath73 .
therefore the kn limit will set in at a frequency @xmath74 .
this is seen in fig .
2a and 2b , where three ( 12 , 21 , 22 ) of the four components in the total spectrum cut off at high energies , while the peak of the inverse compton spectrum occurs as expected in the thomson limit
. however the effect will be severe and affect both the value of the peak frequency and the bulk of the emitted power if @xmath75 is close to @xmath76 , that is if the kn regime sets in already for the scattering of electrons of energy @xmath22 with photons at the peak of the synchrotron emission .
therefore the condition for the kn regime to affect severely the luminosity and the peak frequency of the inverse compton emission , requiring a revision of the formulae given in section 3 is : @xmath77 using eq .
( [ gb ] ) for @xmath22 to express the above condition with observed quantities , eq .
( [ condizione ] ) becomes a condition on @xmath19 : @xmath78^{1/2 } \label{limite}\ ] ] since eq .
( [ gb ] ) ( valid in the thomson limit ) gives the lowest possible value of @xmath22 , eq .
( [ limite ] ) gives the limit above which the full kn regime sets in .
however even for @xmath19 somewhat above the kn limit corrections are non negligible ( see section 4.2 ) the contribution involving both electrons and photons on the high energy branches ( @xmath66 ) is already strongly suppressed in fig .
2b and disappears completely when the kn condition is met ( fig .
2c , 2d ) .
the other contributions are reduced according to the availability of photons below the kn threshold . in fig . 2c and
2d @xmath79 is negligible and the low energy side is dominated by by the @xmath80 branch .
it is interesting to note that for the assumed spectral shape the scattering of highest energy electrons with very low energy photons gives a significant power law contribution to the inverse compton spectrum at the highest energies .
the latter component is not important energetically ( at the peak ) however it provides a high energy tail which extends the range of the inverse compton emission .
the spectral index of the latter contribution can be easily derived from the expression for @xmath81 given in the appendix and is @xmath82 .
this component could explain the power law spectra measured in the tev range for mrk 421 and mrk 501 ( zweerink et al . 1997 ) . in the kn regime both the expression of the compton peak frequency and luminosity must be modified . as shown in fig .
2c , 2d , with the assumed electron spectrum the peak frequency in the kn limit is determined by the scattering between all the electrons and photons with @xmath83 .
maximizing the sum of @xmath80 and @xmath81 ( see appendix ) leads to the following relation between @xmath32 and @xmath22 : @xmath84 where : @xmath85\ ] ] is a factor smaller than one . substituting @xmath22 derived from eq .
( [ intef ] ) in eq .
( [ nus ] ) we obtain : @xmath86 which is the analogous of the eq .
( [ cinque ] ) for the kn regime .
note that in this equation @xmath18 and @xmath19 are directly proportional .
having obtained a revised expression of eq .
( [ gb ] ) we can use the latter ( eq . [ [ intef ] ] ) to express ( [ condizione ] ) through observables .
we therefore arrive at a revised version of the threshold for the kn regime : @xmath78^{1/2 } g(\alpha _ 1,\alpha _ 2)^{-1/2 } \label{limitedue}\ ] ] since @xmath87 is less than 1 , eq .
( [ limitedue ] ) is more stringent than eq .
( [ limite ] ) : therefore we find two _ different _ limits dividing the regions of applicability of the two treatments . between this two limits
there is a transition region , where both the approaches are approximations . in the kn regime
the power in the compton peak is substantially diminished .
equation ( [ natan ] ) should be replaced by : @xmath88 where @xmath89 includes only photons below the kn limit for electrons with the break energy , that is with @xmath90 .
we can write : @xmath91 where @xmath92 is the energy density of the synchrotron radiation for a given frequency .
a simple calculation shows that the integral is given by : @xmath93 where the factor in parentheses represents the ratio between the available energy density and the total energy density . from the last relation
we can obtain the following expression for the energy density ratio in the comoving frame in the kn regime : @xmath94^{\alpha _ 1 - 1}\ ] ] using the last formula and expressing the total luminosities through eq .
( [ royan ] ) and the radius with eq .
( [ undici ] ) we can finally obtain the analogous of eq .
( [ dodici ] ) for the kn limit : @xmath95 ^{(1-\alpha _ 1)/2 } \left [ \frac { 2 ( \nu _ sl_s(\nu _ s))^2 f(\alpha _
1,\alpha _ 2 ) } { c^3 t_{var}^2 \nu _ c l_c(\nu _ c ) } \right]^{1/2 } \left ( \frac{3mc^2}{4h } \right)^{1-\alpha _ 1}\ ] ]
another strong constraint is obtained from the condition of transparency of @xmath96-rays to pair - production absorption .
following dondi & ghisellini ( 1995 ) we can write this condition in the form : @xmath97^{1/(4 + 2\beta ) } \label{diciotto}\ ] ] where @xmath98 is the frequency of target photons and @xmath99 is the spectral index of the target photons ( @xmath29 for @xmath100 and @xmath30 for @xmath101 ) .
the transparency condition does not depend on the specific emission mechanisms : therefore it provides a strong independent constraint on the minimum value of the doppler factor in the ssc model .
the radiative cooling time for a synchrotron self compton emitting electron is given by : @xmath102^{-1}\ ] ] clearly the cooling time is determined by the fastest of the two cooling processes that is synchrotron cooling for @xmath103 or inverse compton cooling for @xmath104 .
let us first consider for simplicity the thomson regime . in the case in which the magnetic field energy density dominate we can approximate the cooling time as ( after some manipulations ) : @xmath105 while in the opposite case ( using eq . [ [ density ] ] for the energy density and expressing @xmath7 with eq . [
[ nus ] ] ): @xmath106 where @xmath107 and @xmath108 are constants . in the observer frame
: @xmath109 the equipartition condition @xmath110 divides the regions of validity of the two approximations in the @xmath111 plane .
the compton cooling approximation will be valid in the lower region ( where @xmath112 ) , while the synchrotron cooling approximation will be valid in the upper region . in both cases
a given value of @xmath113 defines an inverse relation between b and @xmath19 , that is lines with negative slopes , intersecting each other at the equipartition line ( see figures in the application sections ) .
thus a condition on the cooling time corresponds to a minimum value of @xmath39 above which the condition is represented by a double valued function with two branches .
lower values of @xmath39 correspond to shorter cooling times .
the measurement of time delays between variations in different energy bands provides an important restriction on the possible values of the physical parameters of the model .
the most natural way to explain the lags in a homogeneous model is to interpret them as due to the cooling time of the emitting electrons ( takahashi et al .
1996 , urry et al .
1997 ) . in this picture
the quiescent emission is produced by the `` stationary '' population of electrons , while the flare is due to the injection of a monoenergetic population of high energy electrons : while these electrons cool their emission drifts to lower frequencies as observed . adopting this hypothesis
we can relate the observed time lags to the cooling time of the electrons and therefore to the physical parameter of the source .
if the cooling is dominated by the synchrotron emission ( i.e. @xmath114 , where @xmath115 is @xmath116 in the kn limit ) , we obtain the following relation ( see also takahashi et al .
1996 ) : @xmath117^{2/3 } g \label{diciassette}\ ] ] where @xmath118 and @xmath119 are the x - ray frequencies ( @xmath120 ) and @xmath121 is the measured lag ( in seconds ) .
in the compton cooling dominated region ( i.e. @xmath122 ) we can write the equivalent relation : @xmath123 ^ 2 \label{diciassettebis}\ ] ] in our approach the value of @xmath22 enters as a free parameter in the model and is determined from the observations .
however in more sophisticated models @xmath22 could be determined selfconsistently by the equilibrium between injection , cooling and escape of electrons from the source . here
we include a posteriori the condition that the cooling time at @xmath22 is equal to the escape time .
the latter constraint can be written as : @xmath124^{-1}= \frac{5\cdot 10^{8}}{\gamma _ bb^2(1+u^{\prime } _ { syn , avail}/u^ { \prime } _ b ) } \label{bond}\ ] ] where @xmath125 is the electron escape velocity in unit of @xmath126 and @xmath127 and @xmath89 are , respectively , the magnetic field energy density and the radiation energy density . + if cooling is dominated by the synchrotron process , expressing ( [ bond ] ) in terms of the observational quantities @xmath31 and @xmath32 gives : @xmath128^{1/2 } \beta _ { esc } ^{1/2 } \label{sedici}\ ] ] for the thomson limit and : @xmath129^{1/2 } \beta _ { esc}^{1/2 } \label{sedicikn}\ ] ] for the kn limit .
even when compton cooling is less important than synchrotron cooling , if it occurs in the kn regime , the value of @xmath22 must be taken from eq .
( [ intef ] ) instead of eq .
( [ gb ] ) , which causes the difference between ( [ sedici ] ) and ( [ sedicikn ] ) .
in the compton - cooling dominated region the equations are : @xmath130^{2.1 } \left [ \beta _ { esc}(\nu _ s/ \nu _
c)^{1/2}(1+z ) \right]^{-8/5 } t_{var}^{-13/5}\ ] ] for the thomson regime and : @xmath131^{1/(1-\alpha _ 1 ) } \left [ \nu _ c ( 1+z)\right]^{(3\alpha _ 1 - 2)/(1-\alpha _ 1)}\ ] ] [ @xmath132 is a constant given in the appendix ] for the kn regime .
the electron escape time is not a well known parameter and depends on the transport processes in the emitting region : in the applications we check the consistency of the observed value of @xmath22 for values of @xmath125 in the range @xmath133 .
in the previous section we have shown that the constraints considered lead to equations that can be expressed in terms of two parameters : the magnetic field and the doppler factor .
each constraint then corresponds to an allowed region or to a line in the log @xmath18 - log @xmath19 plane .
three equations do not apper explicitly : one determines @xmath44 from the peak frequencies , the second determines the radius from the variability timescale , the third the normalization constant of the electron spectrum from the observed synchrotron luminosity .
we examine here what consequences can be derived comparing the constraints available for the three brightest and best observed bl lacs with synchrotron peaks near or within the x - ray band .
we start from the following values for the observable quantities , derived from the simultaneous observations of macomb et al .
( 1995 ) ( the corresponding data are shown in fig .
1 ) : @xmath134 hz , @xmath135 hz , @xmath136 , @xmath137 @xmath138 erg s@xmath139 and @xmath140 erg s@xmath139 ( see table 1 ) . these refer to the low state ( for the high state we find similar results for the values of @xmath18 and @xmath39 ) .
the tev observations show extremely short variability timescales ( gaidos et al .
1996 report a flare with @xmath141 min ) , while in the x - ray band variability timescales of several hours are observed ( see takahashi et al .
1996 ) : in fig .
( [ 2 ] ) we have used @xmath142 h. for mrk 421 the soft photon lag within the asca band was measured by takahashi et al ( 1996 ) yielding @xmath143 s between @xmath144 kev and @xmath145 kev . in fig . ( [ 2 ] ) we show the parameter space for mrk 421 .
the transparency condition ( [ diciotto ] ) calculated for photons with @xmath146 hz yields a lower limit @xmath147 represented by the vertical line . the kn condition [ eq . ( [ limite ] ) ]
yields an upper limit @xmath148 , while the more stringent version [ eq . ( [ limitedue ] ) ] yields @xmath149 .
for @xmath150 we are therefore in an intermediate region where both limits are inaccurate .
we chose to use the thompson limit here since the resulting value of @xmath19 will be 25 , close to the upper end .
we recall however that even if the thomson limit is adequate for the peak of the compton component , kn effects can not be neglected in the computation of spectra at higher energies ( see fig . 1 , fig . 2a ) . the observed ratio between the two peak frequencies [ eq . ( [ cinque ] ) ] imposes an inverse proportionality relation between b and @xmath19 .
this is shown as a shaded area allowing for a factor 3 uncertainty in the actual value of the compton peak frequency .
the ratio of the peak luminosities ( [ dodici ] ) yields a steeper relation between b and @xmath39 , @xmath151 .
again it is represented by a shaded area to allow for uncertainties in the observational input data .
the intersection between the two regions above defines the allowed region , where the actual values of the parameters should fall .
this region could shrink if the observational data improve , but given the present gap between the energy ranges of egret and the ground based cherenkov telescopes it is unlikely that it could be substantially reduced .
the time lag constraint has a shallow dependence on @xmath19 ( @xmath152 ) and is well defined observationally .
therefore it is represented as single thick line ( labelled _ l _ ) in fig .
it is extremely interesting that this line , which is completely independent of the spectral constraints , is consistent with the parameter region allowed by those constraints .
it is also interesting that the region defined by the condition that the break energy of the electrons is determined by a balance between cooling and escape eq .
( [ sedici ] ) ( region c ) encompasses the lag constraint in the relevant parameter interval .
we conclude that for mrk 421 the spectral constraints are consistent both with the radiative interpretation of the soft photon lag and with the break energy as due to cooling .
these are persuasive arguments in support of the simple general model .
the estimated parameter values are @xmath153 and @xmath154 ( see table 2 ) note that the value of @xmath19 is rather high compared to other estimates ( ghisellini et al .
1998 ) but similar to that found from direct model fits by mastichiadis & kirk ( 1997 ) and ghisellini & maraschi ( 1996 ) .
this bl lac object has been studied for a long time with numerous multifrequency campaigns ( urry et al .
1997 ; edelson et al .
it has been detected as a @xmath7-ray source by egret on one occasion ( vestrand et al .
1995 ) and more recently in a flaring state ( iau circ .
no simultaneous x - ray and gamma - ray observations are available yet .
it has not been detected in the tev band .
this may however be due to its southern declination , since in the southern hemisphere cherenkov telescopes have started observations only recently .
the parameters of the synchrotron component are inferred from the spectrum measured during the 1994 multifrequency campaign ( @xmath155 hz and @xmath156 erg s@xmath139 , urry et al .
comparison with previous observations ( see sambruna et al .
1994 ) shows that during these observations pks 2155 - 304 was in an intermediate state , a factor 2 less luminous than the brightest state recorded by exosat .
the gamma - ray spectrum measured with egret is hard , implying that the compton peak is beyond the egret range .
however the absence of tev observations does not allow to determine the position of the peak .
we will therefore use the other constraints to estimate the peak frequency of the compton emission and the associated tev luminosity , assuming that all of them should be satisfied as is the case for mkn 421 . in fig .
( [ 3 ] ) we show the parameter obtained for @xmath157 hz and for an `` observed '' luminosity of @xmath158 erg s@xmath139 extrapolated from the gev range .
we assume a minumum variability time of @xmath159 s and a delay of @xmath160s between the variations observed in the 0.75 kev and 6 kev energy bands ( urry et al .
the vertical line representing the transparency condition is plotted for photons with frequency @xmath161 hz for consistency with the following estimates of the tev flux .
note , however , that the lower limit on the doppler factor given by the transparency condition would be weaker in the absence of tev emission . in this case ,
the largest uncertainty is on the position of the compton peak , while the frequency of the synchrotron peak is reasonably well known .
we therefore plot the uncertainty band for fixed @xmath31 and for @xmath32 ranging from @xmath162 hz to @xmath163 hz . for these values of the peak frequencies
the kn limit yields @xmath164 , smaller than the transparency limit : we can make use of the thomson formulation .
we find that all the constraints are satisfied for values of @xmath18 and @xmath19 very similar to those found in mrk 421 ( @xmath153 and @xmath154 ) ; the cooling - escape consistency for @xmath22 is satisfied also in this case .
we can estimate the value of the tev flux of pks 2155 - 304 with the hypothesis that beyond the break the spectral indices of mrk 421 and pks 2155 - 304 are similar . from the spectrum of mrk 421
we can obtain the value @xmath165 ; from the inferred value of the @xmath7 peak flux of pks 2155 - 304 we can evaluate the tev flux as : @xmath166 that gives an integrated flux @xmath167 ph s@xmath139
@xmath168 , about @xmath169 that of mrk 421 in the in the non - flaring state . hence we expect that a detection of pks 2155 - 304 should be possible but requires sufficient sensitivity , due to the low flux .
tev observations of pks 2155 - 304 would be extremely important also because they may reveal or constrain absorption of high energy photons by the intergalactic ir background ( stecker and de jager 1997 ) , better than in the case of mrk 421 because of the larger distance .
the bl lac object mrk 501 ( z=0.034 ) underwent a dramatic tev and x - ray outburst in april 1997 ( see e.g. pian et al .
1998 , catanese et al . 1997 ) .
previous observations showed a spectral distribution very similar to mrk 421 , with a synchrotron peak in the soft x - ray region ; in contrast , during the april observation the peak was found at about 100 kev , a shift in energy of at least a factor 100 .
the broad band spectral shape leads to interpret the hard x - ray emission as synchrotron radiation and the flare as due to injection of electrons with very high energy ( pian et al . 1998 ) . in fig .
( [ 4 ] ) we show the parameter space for mrk 501 in the quiescent state . in this state
we can apply the relation for the thomson case , given that the value of @xmath170 required by the transparency condition is higher than @xmath171 .
we have used the following values for the observable quantities : @xmath155 hz , @xmath172 hz , @xmath173 erg / s , @xmath174 erg / s , @xmath175 h , @xmath136 and @xmath137 . + for the quiescent state we have large uncertainties on the position of the compton peak : the boundaries of region a refer to values of @xmath176 hz ( upper line ) to @xmath177 hz .
since a measurement of a time lag between variations in different frequencies is lacking the associated constraint can not be used .
the longer variability timescales and the lower luminosity with respect to the two previous sources allow in this case a value of @xmath19 around 10 , lower than those inferred in mrk 421 and pks 2155 - 304 , as can be seen from fig .
( [ 4 ] ) , while the value found for the magnetic field intensity , @xmath178 is similar .
for the high state we must use the formulation in the kn limit : in fact the thomson condition ( [ limite ] ) yields a value of @xmath179 , highly implausible . for the observables we have used the following values : @xmath180 , @xmath181 erg / s , @xmath182 erg / s and the variability time and the spectral indices as in the quiescent state .
the compton peak frequency is not well determined since the spectra measured in the tev range suffer from various uncertainties ( samuelson et al . 1998 ) .
however all the measurements give slopes steeper than 1 indicating that the peak is at lower energies .
we therefore consider values between @xmath183 hz . in the kn regime the parameter region allowed by the peak frequency ratio ( region a ) has @xmath18 increasing with @xmath19 [ see fig .
( [ 5 ] ) ] .
the luminosity ratio ( region b ) has almost no intersection with region a for values of @xmath19 allowed by the transparency condition .
on the other hand the emitting region could be smaller than imposed by the variability time scale by a factor larger than 3 . in that case region
b would extend upwards defining a permitted region at @xmath184 , @xmath185 however , the condition of consistency for the value of @xmath186 , given in the kn limit by eq .
( [ sedicikn ] ) ( region c ) , is definitely outside the parameter region allowed by the other constraints , implying that the shape of the electron spectrum can not be determined balancing radiation losses and escape , unless again the size of the region is much smaller than estimated from the observed variability .
since the flaring state lasts much longer than the cooling time , efficient continuous reacceleration processes are needed .
a comparison of the parameter space for the low and high states of mkn 501 indicates that not only @xmath22 but also @xmath18 increased in the flaring state .
we have extended the analytic treatment of the spectral constraints on the ssc homogeneous model for blazars initiated by gmd including kn corrections and adding the discussion of other constraints now available .
the constraints can be summarized in the magnetic field - doppler factor plane . in principle
the knowledge of @xmath187 plus @xmath37 allows the determination of b , @xmath39 , @xmath44 , r and k. the transparency condition , the time lag condition , the cooling - escape balance are further restrictions that overconstrain the model . in practice
large uncertainties on the observational quantities still exist , but at least for the best observed tev blazar , mrk 421 , we can significantly constrain the parameter space .
we find that the radiative interpretation of the lags is consistent with the spectral constraints .
moreover , the break in the distribution is consistent with the equilibrium between the energy losses and electron escape timescale .
the derived value of the the doppler factor is moderately large ( @xmath154 ) , in agreement with the values found by specific spectral fits ( see e.g. ghisellini & maraschi 1996 ; mastichiadis & kirk 1997 ) .
we have applied the analysis also to pks 2155 - 304 , a well observed object in the low energy bands , detected in two occasions with egret in @xmath7-rays . assuming consistency between all the constraints ( as found in the case of mrk 421 ) we estimate that the peak of the compton component should fall at about @xmath188 hz : for higher values we do not find consistency between the spectral constraints and the time lags as measured in the 1994 observational campaign ( urry et al .
we estimate the tev flux , which could be measured by southern cherenkov telescopes . the last object analized is mrk 501 , a tev emitter that has shown a dramatic flare in april 1997 . for the quiescent state we can apply the classical formulation , while for the flaring state we must use the formulae in the kn limit .
the absence of information about time lags prevents to completely constrain the model : however we can estimate that the value of the doppler factor is about 10 , smaller than in the other two sources .
the high state is consistent with a costant doppler factor , but impllies an increase in @xmath44 and in the magnetic field .
the required value of @xmath44 is inconsistent with cooling - escape equilibrium , indicating the need of an efficient continuous reacceleration mechanism . in conclusion
the analytic approach offers a solid basis for a general discussion of the model and its applicability to specific sources .
it allows to derive parameter ranges taking into account uncertainties in the `` observable '' quantities and a better comprehension of the interrelations between various parameters .
we stress however that table 2 has been derived applying a number of restrictive choices and hypothesis for illustrative purposes and should not be taken at face value .
individual fits remain essential to better understand the detailed shape of the particle spectra and their origin .
_ note : _ after the paper was accepted we learned about the detection of pks 2155 - 304 by the durham mark 6 cerenkov telescope ( chadwick et al . 1998 ) with a flux of @xmath189 ph @xmath168 s@xmath139 , consistent with our estimate .
in this appendix we give the expression of the compton emissivity for the electron distribution assumed in this work and we prove eq .
( [ intef ] ) for the frequency of the compton peak in the kn limit .
all the quantities are referred to the blob reference frame . +
as previously assumed , we consider a population of relativistic electrons with a double power - law distribution with indices @xmath20 and @xmath21 and break lorentz factor @xmath22 : @xmath190 where @xmath27 .
+ the coefficients @xmath191 and @xmath192 are related by the condition @xmath193 which gives @xmath194 . + the seed photons are those emitted by the synchrotron mechanism and in the blob reference frame the radiation energy density can be written as ( hereafter we use the parameter @xmath195
@xmath196 where @xmath197 . + using a @xmath19-approximation ( see e.g. coppi & blandford 1990 ) the compton emissivity in the blob reference frame can be written as : @xmath198 \left ( \frac { 3x^{\prime}}{4x^{\prime}_0 } \right)^{1/2 } \frac{\epsilon ( x^{\prime}_0)}{x^{\prime}_0 } dx^{\prime}_0\ ] ] where the limits of the integral are analyzed in the following .
+ for @xmath199 and @xmath200 given above we can split the integral in four similar integrals : @xmath201 the first term is : @xmath202 with @xmath203 .
the limits are : @xmath204 ; \
x_2=\min \left[x^{\prime}_s , \frac{3x^{\prime}}{4 } , \frac{3}{4x^{\prime } } \right]\ ] ] the term @xmath205 in the upper limit express the kn suppression of the high energy emission .
it originate from the step approximation of the kn cross - section , that can be rewritten as the condition that the scattering are in the thomson regime up to : @xmath206 this condition , with the use of @xmath207 gives : @xmath208 that is the condition used in the integrals .
the second integral is : @xmath209\ ] ] where : @xmath210 ; \
x_2=\min \left[\frac{3x^{\prime}}{4 } , \frac{3}{4x^{\prime } } \right]\ ] ] the third term can be written as : @xmath211\ ] ] where , in this case @xmath212 and : @xmath213 ; \
x_2=\min \left [ x^{\prime}_s , \frac { 3x^{\prime}}{4\gamma _ b^2 } , \frac{3 } { 4x^{\prime } } \right]\ ] ] the last term is : @xmath214 with : @xmath215 ; \
x_2=\min \left [ x^{\prime}_{max},\frac { 3x^{\prime}}{4\gamma _ b^2},\frac{3}{4x^{\prime } } \right]\ ] ] the compton emissivity around the peak in the kn limit is dominated by the contribution of @xmath80 and @xmath81 ( see fig .
one can easily see that the emissivity can be written as : @xmath216 + c(\alpha _
2)k_1 \epsilon _
0x^{\prime -\alpha _ 1 } x_s^{\prime \alpha _ 1 } \left ( \frac { 3}{4 } \right)^{\alpha _ 2-\alpha _ 1 } \frac{1}{\alpha _ 2-\alpha _ 1}\ ] ] > from this expression we can find the frequency of the compton peak , that is the maximum of the function @xmath217 . taking the derivative we can write the maximum condition as : @xmath218 + \frac{1}{\alpha _ 2-\alpha _ 1 } - \frac { 2}{1-\alpha _ 1}=0\ ] ] which gives the value of the frequency of the compton peak : @xmath219\ ] ]
the constant @xmath220 has the expression : @xmath31 ( hz ) & & @xmath222 & @xmath223 & @xmath223 & @xmath224 + @xmath36 ( hz ) & & @xmath225 & @xmath226 & @xmath227 & @xmath228 + @xmath229 & & @xmath230 & @xmath231 & @xmath232 & @xmath233 + @xmath234 & & @xmath235 & @xmath231 & @xmath236&@xmath237 + @xmath37 ( h ) & & @xmath238 & @xmath239 & @xmath240 & @xmath240 + @xmath241 & & @xmath242 & 4700 & - & - + @xmath19 & @xmath243 & @xmath244 & @xmath245 & @xmath246 + @xmath18 ( g ) & @xmath247 & @xmath247 & @xmath248 & @xmath249 + @xmath250 & @xmath251 & @xmath252 & @xmath253 & @xmath254 + @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 + | we consider the constraints that can be derived from the spectral shape and variability of tev blazars on the homogeneous synchrotron self
compton ( ssc ) model . assuming that the relativistic electron spectrum is a broken power law , where the break energy is a free parameter
, we write analytical formulae that allow to connect observable quantities to the physical parameters of the model.we give approximate analytic formulae also for the case of compton scattering occurring in the klein nishina regime , which is particularly relevant for tev blazars .
in particular we find that even in the latter regime a power law component can be present at the highest energies . further restrictions in the parameter space are set assuming that the break energy results from a balance between cooling and escape and that the soft photon lags measured in some sources derive from radiative cooling of high energy particles .
the constraints can be summarized as allowed regions in the doppler factor magnetic field parameter space and are in principle sufficient to univocally determine the model parameters and their uncertainties .
we apply the method to three well studied sources mkr 421 , pks 2155304 and mkr 501 . for mrk 421
the available data are sufficient to fully constrain the model .
the additional restrictions are found to be consistent with the model parameters , supporting the proposed interpretation . in the case of pks 2155304 ,
not yet detected in the tev band , we estimate the peak frequency of the compton component ( @xmath0 gev ) and the expected tev flux .
the derived physical parameters are similar for the two sources , with a relatively large value of the doppler factor ( @xmath1 ) and a magnetic field of @xmath2 g. for mrk 501 we consider both the historical low state and the flaring state observed in april 1997 . in the first case consistency between the various assumptions
is reached for @xmath3 and @xmath4 g. for the high state a similar value of the doppler factor and a somewhat larger value of the magnetic field are indicated , while the extremely large frequency of the observed synchrotron peak requires continuous injection / reacceleration of relativistic particles . |
consider the set of all radial eigenfunctions of the hyperbolic laplacian assuming the value 1 at the origin , i.e. the set of all solutions for the following system @xmath0 written in the geodesic polar coordinates of the hyperbolic space of constant sectional curvature @xmath1 .
recall that for every @xmath2 there exists a unique solution @xmath3 such that @xmath4 , see @xcite ( p. 272 ) . if we choose two radial eigenfunctions @xmath5 and @xmath6 such that @xmath7 , then , how many values of @xmath8 such that @xmath9 are sufficient to ensure that @xmath10 and @xmath11 ?
we shall see that we need only one such a value of @xmath12 if the value is small enough .
in other words , if @xmath13 , then there exists an interval @xmath14 $ ] , such that @xmath15 for all @xmath16 $ ] . in particular
, we prove that if @xmath17 are real and @xmath18 then @xmath15 for all @xmath19 .
so , if @xmath18 and @xmath9 just for one arbitrary @xmath8 , then @xmath10 and @xmath11 .
+ on the other hand , according to corollaries [ many_eigefcns_cor ] and [ many_eigefcns_cor2 ] from p. , for every fixed point @xmath20 , and for a fixed radial eigenfunction @xmath6 , assuming the value one at the origin , there are infinitely many other radial eigenfunctions assuming the value one at the origin and the value @xmath21 at the point @xmath22 .
for the future reference , let us introduce the following notation .
let @xmath23 and @xmath24 be the half - space model and the ball model , respectively , of a hyperbolic @xmath25dimensional space with a constant sectional curvature @xmath26 .
recall that the half - space model @xmath23 consists of the open half - space of points @xmath27 in @xmath28 for all @xmath29 and the metric is given by @xmath30 , where @xmath31 is the euclidean distance element . the ball model @xmath24
consists of the open unit ball @xmath32 in @xmath28 , and the metric for this model is given by @xmath33 .
recall also that in a hyperbolic space a laplacian can be represent as @xmath34 \\ & = \frac{\partial^2}{\partial r^2}+ \frac{n-1}{\rho}\coth\left(\frac{r}{\rho}\right)\frac{\partial}{\partial r}+\triangle_{s(0;r)}\ , , \end{split}\ ] ] where the first expression is the hyperbolic laplacian in @xmath35 expressed by using euclidean rectangular coordinates and the second expression represents the hyperbolic laplacian in @xmath24 expressed in the geodesic hyperbolic polar coordinates , where @xmath36 is the laplacian on the geodesic sphere of a hyperbolic radius @xmath12 about the origin .
the next step is to introduce the function @xmath37 and show that @xmath38 as an eigenfunction of the hyperbolic laplacian in @xmath39 .
[ eigenfunction - omega - proposition ] let @xmath40 be any point of @xmath41 and @xmath42 , where @xmath43 and @xmath44 are the euclidean sphere and ball , respectively , both of the same radius @xmath45 centered at the origin @xmath46 .
let @xmath47 be the euclidean distance between the origin @xmath46 and a point @xmath48 , @xmath49 and @xmath50 then @xmath51 notice that @xmath52 and the relationship between @xmath35 and @xmath24 is given by cayley transform + @xmath53 expressed by the following formulae @xmath54 where @xmath55 . since cayley transform is an isometry , an orthogonal system of geodesics defining the laplacian in @xmath24 is mapped isometrically to an orthogonal system of geodesics in @xmath35 and then , @xmath56 therefore , equation is precisely equation written in @xmath24 , which completes the proof of proposition [ eigenfunction - omega - proposition ] . in this subsection we shall see that every radial eigenfunction in @xmath57 depending only on the distance from the origin has an explicit integral representation .
[ sharp - definition ] recall that @xmath58 is the ball model of the hyperbolic space @xmath59 with the sectional curvature @xmath1 and let @xmath60 be the euclidean ball of radius @xmath45 centered at the origin and represents the ball model @xmath58 .
suppose that @xmath61 is a function on @xmath58 .
we define its radialization about the origin @xmath46 , written @xmath62 , by setting @xmath63 where the integration is considered with respect to the measure on @xmath64 induced by the euclidean metric of @xmath65 ; recall also that @xmath66 is the euclidean distance between the origin @xmath46 and a point @xmath67 ; @xmath68 is the euclidean volume of @xmath64 .
[ laplacian - commutes - lemma ] @xmath69 let us introduce two measures on @xmath70 .
one is the lebesque measure of the sphere , i.e. the riemannian volume of the flat metric of the ambient space restricted to @xmath70 and the other one is the haar measure of @xmath71 pushed forward to the sphere by the map @xmath72 , where @xmath73 is a fixed vector on the sphere and @xmath74 .
these two measures agree up to constant multiple factor so that these two averaging procedures must be the same .
it is clear that laplacian commutes with averaging over @xmath71 , since any individual isometry of @xmath75 commutes with laplacian .
therefore , laplacian must commute with radialization defined in .
this completes the proof of lemma [ laplacian - commutes - lemma ] .
the next step is to obtain the explicit representation for radial eigenfunctions .
let @xmath76 be a fixed point and @xmath77 .
then let us define @xmath78 where @xmath79 is a complex number .
thus , @xmath80 is the radialization of @xmath81 about the origin .
[ radial - func - repre - thoerem ]
let @xmath12 be the hyperbolic distance between the origin @xmath46 and @xmath82 .
then , the following function @xmath83 where @xmath84 , is the unique radial eigenfunction assuming the value @xmath85 at the origin and corresponding to an eigenvalue @xmath86 , i.e. , @xmath87 recall that @xmath47 is the euclidean distance between @xmath42 and the origin , while @xmath88 is the hyperbolic distance between the origin and @xmath48 .
therefore , the relationship between @xmath12 and @xmath89 is @xmath90 which justifies the last expression in .
recall also that according to , @xmath38 is the eigenfunction of the hyperbolic laplacian with the eigenvalue @xmath91 .
therefore , according to lemma [ laplacian - commutes - lemma ] , p. , the radialization of @xmath38 defined in is also an eigenfunction with the same eigenvalue .
the last step is to show that the eigenfunction @xmath5 assuming the value @xmath85 at the origin is unique .
we obtain the uniqueness by the procedure described in @xcite , pp .
148 - 153 or in @xcite , p. 272
resulting that the general radial eigenfunction is on the form @xmath92 , where @xmath93 is an entire function of @xmath12 with @xmath94 , and @xmath95 is defined for all @xmath8 , with a singularity at @xmath96 .
therefore , for a radial eigenfunction to have the value @xmath85 at the origin , we must set @xmath97 and then , we can see that the radial eigenfunction is defined uniquely by its value at the origin and by the eigenvalue .
observing that @xmath98 completes the proof of theorem [ radial - func - repre - thoerem ] . recall that one radius theorem for the function @xmath37 introduced in and considered as a function of two arbitrary non - equal points in euclidean space
was obtained in @xcite . in this subsection
we are going to reproduce the statement of this theorem and use it later as the basic result in the proof of one radius theorem for radial eigenfunctions of the hyperbolic laplacian . [
basic - theorem ] let @xmath70 , @xmath99 , be a @xmath100-dimensional sphere of radius @xmath101 centered at the origin @xmath46 , @xmath102 such that @xmath103 and @xmath104 until further notice we assume that @xmath101 and a point @xmath105 are fixed .
then , the following statements hold .
( a ) : : [ sphe - pro - direct - statement ] if @xmath106 and @xmath107 , then @xmath108 ( b ) : : [ sphe - pro - inverse - real - statement ] if @xmath109 are real , then @xmath110 ( c ) : : [ sphe - pro - liuville - statement ] for every @xmath111 there are infinitely many numbers @xmath112 such that @xmath113 ( d ) : : [ sphe - pro - inverse - complex - statement ] suppose that for the fixed point @xmath105 @xmath114 then @xmath115 statement ( c ) shows that the implication introduced in fails if we just let @xmath79 and @xmath116 be complex .
however , a certain additional restriction formulated in statement ( d ) for the complex numbers @xmath109 allows us to obtain precisely the same implication as in .
observe that statement ( b ) is a special case of statement ( d ) . indeed ,
if @xmath109 are real , then holds for every @xmath117 .
in this section we shall see that if @xmath9 for @xmath96 and for some @xmath118 chosen sufficiently close to the origin , then @xmath10 and @xmath9 for all @xmath119 .
in other words , we shall introduce a condition for an eigenvalue under which the eigenfunction is uniquely determined just by its value at @xmath120 point sufficiently close to the origin .
moreover , we shall see that with an additional condition for the eigenvalue , a radial eigenfunction is uniquely determined by its value at an arbitrary point .
[ parabola - strip - equivalence - lemma ] let @xmath112 and @xmath121 , where @xmath122 is given by @xmath123 , where @xmath124 are constants .
then @xmath125\iff \mu\in\{\text{shaded area inside the parabola bellow}\}\ ] ] let @xmath126 and @xmath127
. then @xmath128 .
the direct computation shows that @xmath129 , which means that the image of the following strip @xmath130\}\ ] ] must be the same as the image of the upper half of the strip above .
i.e. , @xmath131 where @xmath132\}\,.\ ] ] to figure out what is the image of @xmath133 , it is enough to look at the image of a horizontal line from @xmath133 .
let us choose the boundary line : @xmath134 , where @xmath135 is fixed and @xmath136 serves as a parameter .
then , @xmath137 where , clearly , the imaginary part depends on @xmath136 linearly , while the real part depends quadratically .
therefore , the horizontal line chosen is mapped to a parabola .
any parabola is uniquely determined by three arbitrary points on it .
thus , to determine this parabola , we find all points of intersections with the coordinate axis .
if @xmath138 , then @xmath139 gives the point of intersection the parabola with the horizontal coordinate axis .
if @xmath140 , then @xmath141 what is denoted on figure [ parabola - strip - equiv ] above .
therefore , both of these two parallel lines @xmath142 are mapped to the parabola described above .
notice also that this parabola is symmetric with respect to the horizontal axis . in addition , the smaller the parameter @xmath135 , the narrower the parabola and its tip is closer to the origin .
so , the proof of proposition [ parabola - strip - equivalence - lemma ] is complete .
let @xmath13 and @xmath6 , @xmath5 be two radial eigenfunctions for the hyperbolic laplacian given by @xmath144 in a ball model of a hyperbolic @xmath145 dimensional space of a constant sectional curvature @xmath146 .
let @xmath147 .
then @xmath148 for every @xmath149 $ ] , where @xmath150 and @xmath109 are complex numbers related to @xmath151 by the following quadratic equations @xmath152 recall that the first quadratic equation above together with implies that @xmath153 \iff \re\mu\leq-\kappa\left(p^2+\frac{k^2}{4}\right)+\frac{1}{4\kappa p^2}(\im\mu)^2\,,\ ] ] i.e. , @xmath86 belongs to the inner part of the parabola depending on parameter @xmath135 and pictured on figure [ parabola - strip - equiv ] , p. .
if @xmath154 , we arrive at the case when @xmath109 are real or , equivalently , @xmath155 , which , according to the one radius theorem , means that the condition @xmath13 implies that @xmath148 for every @xmath156_{p=0}=(0,\infty)$ ] .
let us assume that @xmath9 for some @xmath149 $ ] . recall that according to theorem [ radial - func - repre - thoerem ] from p. , a radial eigenfunction in @xmath58 assuming the value @xmath85 at the origin is uniquely defined by its eigenvalue and this eigenfunction
can be expressed by using the euclidean coordinates as @xmath157 where @xmath158 $ ] and @xmath159 .
let us set @xmath160 and @xmath161 in statement ( d ) of theorem [ basic - theorem ] , p. .
using we can observe that @xmath162\ ] ] is equivalent to say that there exists @xmath163 such that @xmath164 it is clear that , according to statement ( d ) of main theorem [ basic - theorem ] , p. yields @xmath107 or @xmath165 , which implies that @xmath10 .
the last equality contradicts to the assumption of the theorem , which completes the proof of theorem [ one - radius - theorem ] .
[ many_eigefcns_cor ] using , we observe that @xmath166 is the set of all radial eigenfunctions such that @xmath167 and with eigenvalues within the shaded area inside the parabola depending on @xmath135 .
the parabola was pictured on figure [ parabola - strip - equiv ] , p. .
then the value of @xmath168 at any point @xmath169 $ ] determines the radial eigenfunction @xmath170 uniquely .
meanwhile , according to statement ( c ) of theorem [ basic - theorem ] form p. , there are infinitely many eigenfunctions @xmath171 such that @xmath172 .
[ many_eigefcns_cor2 ] if @xmath154 or , equivalently , if @xmath173 is defined as @xmath174 then the value @xmath168 defines @xmath175 uniquely for any @xmath176 . and
again , by statement ( c ) of theorem [ basic - theorem ] form p. , there are infinitely many radial eigenfunctions @xmath177 such that @xmath172 .
isaac chavel , * eigenvalues in riemannian geometry * , academic press , 1984
. w.j.olver , * asymptotics and special functions * , academic press , inc . , new york and london , 1974 .
s.artamoshin , * the spherical ratio of two points and its integral properties * , http://arxiv.org/pdf/1411.1915.pdf , 2014 . | let us fix two different radial eigenfunctions of a hyperbolic laplacian and assume that both of them have the same value at the origin
. both eigenvalues can be complex numbers .
the main goal of this paper is to estimate the lower bound for the interval ( 0,t ] , where these two eigenfunctions must assume different values at every point .
we shall see that t is a function of two different eigenvalues corresponding to the given pair of radial eigenfunctions . on the other hand
, we shall see that at every fixed point and for the value already assumed by a radial eigenfunction at the fixed point , there are infinitely many other radial eigenfunctions , assuming the same value at the fixed point and satisfying the same initial condition . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.